EP4688852A2 - Anti-cd33 antibodies - Google Patents

Abstract

Isolated antibodies which specifically binds to a CD33 protein are described. Related nucleic acids, cells and therapeutics uses are also described.

Description

Anti-CD33 antibodies

Field of the Invention

The present invention relates to antibodies capable of binding to CD33 and particularly, although not exclusively, to novel therapeutic antibodies. Methods for using anti-CD33 antibodies in the treatment of neurodegenerative diseases and cancer are also described.

Background

CD33 (Siglec-3) is an inhibitory immune receptor and type I transmembrane protein belonging to the family of sialic acid-binding immunoglobulin-like lectins (Siglecs), which is expressed on the cell surface of myeloid cells, monocytes, macrophages, and microglia in the brain. CD33 was one of the top-ranked genes associated with risk of developing Alzheimer’s disease (AD) in genome-wide association studies. CD33 signalling in microglia has been implicated in AD pathology, and CD33 expression is elevated in AD patients where it is thought to modulate microglial activation and inhibit clearance of amyloid p (Zhao et al., 2019).

Alzheimer’s disease is the most common cause of dementia in older adults (Zhao et al., 2019). New therapies to Alzheimer’s disease are being actively sought to modify the course of the disease. Current candidates targeting beta-amyloid, Tau, and innate immunity in the brain have in some cases shown pharmacodynamic effects on pathological mechanisms in clinical trials but have yet to demonstrate convincing disease modification in late-stage clinical trials to date (Golde et al., 2022).

Siglecs expressed on tumour-infiltrating immune cells have been suggested to influence anti-tumour immunity, and to represent a potential target for cancer immunotherapy (Stanczak & Laubli, 2023).

Antibodies that are capable of binding CD33 have been developed. For example, US 2022/0162309 describes humanised antibodies that bind human CD33 and use of these antibodies in the treatment of Alzheimer’s disease, dementia, frontotemporal dementia, vascular dementia, mixed dementia, taupathy disease, infection and cancer. However, clinical trials of the lead antibody AL003 described in US 2022/0162309 have been terminated. A humanised anti-CD33 antibody (lintuzumab) has been shown to have modest activity in the treatment of acute myeloid leukemia (AML) (Jurcic, 2012).

There remains a need for further antibodies that are capable of binding and modulating CD33 signalling in an effective manner.

Summary of the Invention

The invention includes the combination of the aspects and preferred features described except where such a combination is clearly impermissible or expressly avoided. According to a first aspect, the invention provides an isolated antibody which specifically binds to a CD33 protein, wherein the antibody increases phagocytosis of a cell expressing CD33 compared to a comparative antibody and/or wherein the antibody binding to a human CD33 protein comprising mutations at positions 20, 21 , 22 and 24 is increased compared to the antibody binding to a human CD33 protein without said mutations.

The antibody according to the present aspect may comprise a heavy chain variable domain (VH) with the following CDRs: CDRH1 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810; CDRH2 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810; CDRH3 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810; or a set of CDRs which contains zero, one or two amino acid substitutions in each CDR compared with the above set of CDRs.

According to a second aspect, the invention provides an isolated antibody which specifically binds to a CD33 protein, wherein the antibody comprises a heavy chain variable domain (VH) with the following CDRs: CDRH1 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810; CDRH2 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810; CDRH3 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810; or a set of CDRs which contains zero, one or two amino acid substitutions in each CDR compared with the above set of CDRs.

Antibodies with these CDRs were identified as CD33 binders in centenarians with exceptional cognitive ability, indicating a likely protective role in neurodegenerative diseases as well as likely good therapeutic tolerability. These antibodies were identified as suffering less from peripheral degradation than comparative antibodies, while being able to induce phagocytosis in myeloid and microglial cells to the same or a higher extent than comparative antibodies. These antibodies were identified as binding to a different epitope than comparative antibodies.

The antibody according to the present aspect may increase phagocytosis of a cell expressing CD33 compared to a comparative antibody, and/or the antibody binding to a human CD33 protein comprising mutations at positions 20, 21 , 22 and 24 may be increased compared to the antibody binding to a human CD33 protein without said mutations.

The antibody according to the first or second aspect may have any one or more of the following optional features.

The antibody binding to a human CD33 protein comprising mutations at positions 20, 21 , 22, 24 and 132 may be increased compared to the antibody binding to a human CD33 protein without said mutations. The antibody binding to a human CD33 protein comprising mutations at positions 47, 50, 51 and 52 may be reduced compared to the antibody binding to a human CD33 protein without said mutations. The antibody may not bind to a human CD33 protein comprising mutations at positions 47, 50, 51 and 52. The antibody binding to a human CD33 protein comprising mutations at positions 47, 50, 51 , 52 and 122 may be reduced compared to the antibody binding to a human CD33 protein without said mutations. The antibody may not bind to a human CD33 protein comprising mutations at positions 47, 50, 51 , 52 and 122. The antibody binding to a human CD33 protein comprising mutations at positions 83 may be reduced compared to the antibody binding to a human CD33 protein without said mutations.

The mutations may be selected from: at position 20: N20R, at position 21 : F21V, at position 22: W22R, at position 24: Q24E, at position 47: 147V, at position 50: Y50H, at position 51 : D51T, at position 52: K52R, at position 83: Q83R, at position 122: R122K, and at position 132: P132T.

The binding may be as measured using a single point ELISA. The human CD33 protein may comprise residues 18-232 of human CD33. The human CD33 protein without mutations may be CD33M2_ECD_18- 232_WT.

The antibody may have one or more, or all of: (i) increased binding to a protein comprising the sequence of CD33M2_ECD_18-232_MutPos1 [P1] compared to a protein comprising the sequence of CD33M2_ECD_18-232_WT, (ii) increased binding to a protein comprising the sequence of CD33M2_ECD_18-232_MutPos1_MutPos6 [P6+1] compared to a protein comprising the sequence of CD33M2_ECD_18-232_WT, (iii) reduced binding to a protein comprising the sequence of

CD33M2_ECD_18-232_MutPos2 [P2] compared to a protein comprising the sequence of

CD33M2_ECD_18-232_WT, (iv) reduced binding to a protein comprising the sequence of

CD33M2_ECD_18-232_MutPos2_MutPos5 [P2+5] compared to a protein comprising the sequence of CD33M2_ECD_18-232_WT, and (v) reduced binding to a protein comprising the sequence of

CD33M2_ECD_18-232_MutPos4 [P4] compared to a protein comprising the sequence of

CD33M2_ECD_18-232_WT.

An antibody of the disclosure may have lower peripheral clearance when administered to a subject, compared to a comparative anti-CD33 antibody.

An antibody of the disclosure may increase phagocytosis of Ap plaques by microglial cells in vivo compared to a control. An antibody of the disclosure may increase the phagocytosis of tau aggregates by microglia with inflammatory phenotype (e.g. LPS treated iPSC microglia) compared to a control. An antibody of the disclosure may increase the phagocytosis of tau aggregates by microglia with inflammatory phenotype to a larger extent than a comparative anti-CD33 antibody. An antibody of the disclosure may not induce release of one or more cytokines including IL-6 and/or MCP-1 by microglia in vitro and/or in vivo. An antibody of the disclosure may reduce the levels of IL-6 and/or MCP-1 released by microglia with an inflammatory phenotype in vitro (e.g. LPS treated human iPSC derived microglia) and/or in vivo. An antibody of the disclosure may reduce inflammation induced release of one or more markers of inflammation in a human neural cell culture assay and/or in the central nervous system of a subject. The one or more markers of inflammation may be selected from: MCP-1 , IP-10, GFAP and IL-6.

A comparative antibody may be selected from: an isotype control antibody, another CD33-binding antibody, and an antibody with the heavy chain variable sequence of ATL_5909 and the light chain variable sequence of ATL_5909. The phagocytosis may be assessed by measuring a fluorescence signal associated with uptake of a labelled particle by imaging or flow cytometry. The cell may be a monocyte or a microglial cell. The cell may be a human cell. The cell may be an iPSC derived microglial cell. The cell may be a cell that has been stimulated with an inflammatory signal (e.g. LPS) prior to exposure to the antibody.

The antibody may bind to a CD33 protein that comprises the V domain of CD33. The antibody may not bind to a CD33 protein that does not comprise the V domain of CD33. The antibody may not bind to a CD33 protein that has the sequence of protein CD33_HUMAN_ECD_Cdomain_His_007.

The antibody may comprise a heavy chain variable domain (VH) with the following CDRs: a CDRH1 comprising an amino acid sequence selected from: HCDR1_ATL_0005802; HCDR1_ATL_0005807; HCDR1_ATL_0005808: GYSFTSYW; HCDR1_ATL_0005853: GYKFNNNW; HCDR1_ATL_0005854: GYKFSNNW; or an amino acid sequence with 1 or 2 mutations compared to the above sequences; a CDRH2 comprising an amino acid sequence selected from: HCDR2_ATL_0005802; HCDR2_ATL_0005807; HCDR2_ATL_0005808; HCDR2_ATL_0005853; HCDR2_ATL_0005854:

IYPGDSDT; or an amino acid sequence with 1 or 2 mutations compared to the above sequences; and a CDRH3 comprising an amino acid sequence selected from: HCDR3_ATL_0005802; ARPRGFGEYYFDY HCDR3_ATL_0005853 ARHSGGLDGYTAAALDY; HCDR3_ATL_0005854 ATWGGSNWFVD; or an amino acid sequence with 1 or 2 mutations compared to the above sequences.

The antibody may comprise a heavy chain variable domain (VH) with the following CDRs: a CDRH1 comprising the sequence of HCDR1_ATL_0005802, a CRH2 comprising the sequence of HCDR2_ATL_0005802, and a CDRH3 comprising the sequence of HCDR3_ATL_0005802, or a set of CDRs comprising 1 or 2 mutations in the CDRH1 and CDRH2 compared to these sequences and/or 1 , 2 or 3 mutations in the CDRH3 compared to these sequences.

The antibody may comprise a heavy chain variable domain (VH) with the following CDRs: a CDRH1 comprising the sequence of HCDR1_ATL_0005853, a CRH2 comprising the sequence of HCDR2_ATL_0005853, and a CDRH3 comprising the sequence of HCDR3_ATL_0005853, or a set of CDRs comprising 1 or 2 mutations in the CDRH1 and CDRH2 compared to these sequences and/or 1 , 2 or 3 mutations in the CDRH3 compared to these sequences.

The antibody may comprise a heavy chain variable domain (VH) with the following CDRs: a CDRH1 comprising the sequence of HCDR1_ATL_0005854, a CRH2 comprising the sequence of HCDR2_ATL_0005854, and a CDRH3 comprising the sequence of HCDR3_ATL_0005854, or a set of CDRs comprising 1 or 2 mutations in the CDRH1 and CDRH2 compared to these sequences and/or 1 , 2 or 3 mutations in the CDRH3 compared to these sequences.

The antibody may have a heavy chain variable domain (VH) with the following framework sequences: HFWR1 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810,

HFWR2 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, HFWR3 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, and

HFWR4 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, or framework sequences with one to six substitutions compared to the framework sequences above.

The antibody may have a heavy chain variable domain (VH) with a framework sequence HFWR2 of ATL_0005802. The antibody may have a heavy chain variable domain (VH) with a framework sequence comprising a ‘A’ at position 40 in standard IMGT numbering. The substitutions in the framework sequences of the heavy chain variable domain may be located at any position other than position 40 in standard IMGT numbering. The antibody may have a heavy chain variable domain (VH) comprising CDRH1 , CDRH2, and CDRH3 within a germline framework, provided that position 40 in standard IMGT numbering is A.

The antibody may have a heavy chain variable domain (VH) with the following framework sequences: HFWR1 of ATL_0005802, ATL_0005853 or ATL_0005854; HFWR2 of ATL_0005802, ATL_0005853 or ATL_0005854; HFWR3 of ATL_0005802, ATL_0005853 or ATL_0005854; and HFWR4 of ATL_0005802, ATL_0005853 or ATL_0005854.

The antibody may have a heavy chain variable domain (VH) comprising a sequence that has a least 95% sequence identity with a sequence selected from the VH sequence of antibodies: ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810,

ATL_0006040, ATL_0006041 , ATL_0006042, ATL_0006043, ATL_0006044, ATL_0006045,

ATL_0006046, ATL_0006047, and ATL_0006048. The antibody may have a heavy chain variable domain (VH) comprising a sequence that has at most 2 mutations in each HCDR and at most 3 mutations in each framework region compared with a sequence selected from the VH sequences above. The antibody may have a heavy chain variable domain (VH) comprising a sequence that has a least 95% sequence identity with a sequence selected from the VH sequence of antibodies ATL_0005802, ATL_0005853, or ATL_0005854 or a sequence that has at most 2 mutations in each HCDR and at most 3 mutations in each framework region compared with a sequence selected from said VH sequences.

The antibody may bind human CD33. The antibody may bind human CD33 with an EC50 of at most 2e-08 M, or at most 3e-09 M, as assessed by ELISA (such as binding of plated rhCD33).

The antibody may deplete CD33 on the cell surface of human monocytes by less than 50%, or less than 80% after 5 hours of incubation with the antibody. The antibody may deplete CD33 on the cell surface of human monocytes after 5 hours of incubation with the antibody to a lower extent than a comparative antibody at the same concentration.

The antibody may selectively bind to CD33 over other one or more siglecs. The antibody may selectively bind to CD33 over one or more (or all of): siglec-6, siglec-7, siglec-8, and siglec-9. The antibody may selectively bind to human CD33 over other one or more homologs. The antibody may selectively bind to human CD33 over mouse CD33 and cyno CD33. The antibody may be an scFv antibody molecule, a nanobody, or a whole antibody. The antibody may comprise an antibody constant region. The antibody may be a whole antibody. The antibody may be an lgG1 or variant thereof. The antibody may be an lgG1 variant L234A/L235A (LALA).

The antibody may comprise a light chain variable domain (VL) with the following CDRs:

CDRL1 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810,

CDRL2 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, and

CDRL3 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, or a set of CDRs which contains zero, one, or two amino acid substitutions in each CDR compared with the above set of CDRs.

The CDRL1 , CDRL2 and CDRL3 of the VL domain may be within a germline framework.

The antibody may have a light chain variable domain (VL) with the following framework sequences: LFWR1 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810;

LFWR2 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810;

LFWR3 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810; and

LFWR4 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810; or a set of FWRs which contains one to six amino acid substitutions compared with the above set of FWR.

The antibody may have a light chain variable domain (VL) comprising a sequence selected that has a least 95% sequence identity with a sequence selected from the VL sequence of antibodies: ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810,

ATL_0006040, ATL_0006041 , ATL_0006042, ATL_0006043, ATL_0006044, ATL_0006045,

ATL_0006046, ATL_0006047, and ATL_0006048. The antibody may have a light chain variable domain (VL) comprising a sequence that has at most 2 mutations in each LCDR and at most 3 mutations in each framework region compared with a sequence selected from said VL sequences. The antibody may have a light chain variable domain (VL) comprising a sequence that has a least 95% sequence identity with a sequence selected from the VL sequence of antibodies ATL_0005802, ATL_0005853, or ATL_0005854 or a sequence that has at most 2 mutations in each LCDR and at most 3 mutations in each framework region compared with a sequence selected from said VL sequences.

According to a third aspect, there is provided an isolated VH domain of an antibody according to any embodiment of the first or second aspect. According to a fourth aspect, there is provided an isolated nucleic acid which comprises a nucleotide sequence encoding an antibody, including a VH or VL domain, according to any embodiment of the first or second aspects, or a fragment thereof.

Also described herein is a vector or set of vectors comprising the nucleic acid according to the fourth aspect, a host cell in vitro transformed with said nucleic acid or a host cell comprising said vector or set of vectors.

Also described herein is a composition comprising an antibody, including an antibody VH domain or antibody VL domain, according to any embodiment of the first or second aspect, and at least one additional component, optionally comprising a pharmaceutically acceptable excipient, vehicle or carrier.

Also described herein is an antibody according to any embodiment of the first or second aspect, for use in the treatment of a disease or disorder. Also described herein is a method of treating a disease or disorder in a subject in need thereof, the method comprising administering a therapeutically effective amount of an antibody according to any embodiment of the first or second aspect. Also described herein is an antibody according to any embodiment of the first or second aspect, for use in the manufacture of a medicament. The disease or disorder may be a disease associated with dysfunction of microglial cells. The disease or disorder may be a neurodegenerative disease or disorder. The disease or disorder may be a tauopathy. The neurodegenerative disorder may be selected from: frontotemporal dementia (FTD), Alzheimer’s disease (AD), Huntington’s Disease (HD), Parkinson’s disease (PD), amyotrophic lateral sclerosis (ALS), human immunodeficiency virus (HlV)-induced encephalitis, Chronic traumatic encephalopathy (CTE), vascular dementia, prion diseases, Lewy body disease, Spinal muscular atrophy (SMA), Motor Neuron Disease (MND), such as amyotrophic lateral sclerosis (ALS), progressive supranuclear palsy (PSP), spinocerebellar ataxias (SCA) types 1 , 2, 6, 7 and 17, Machado-Joseph disease (MJD/SCA3), dentatorubral pallidoluysian atrophy (DRPLA), spinal bulbar muscular atrophy X-linked type 1 (SMAX1/SBMA), Anderson-Fabry (X-linked Fabry Disease), and DNAJB6 Myopathies, optionally wherein the neurodegenerative disease is selected from FTD, AD, HD, and PD. The disease or disorder may be a cancer. The cancer may be selected from AML or a cancer associated with hypersyalilation of tumour cells and/or overexpression of CD33 by tumour cells. The disease or disorder may be a disease characterised by insufficient macrophage phagocytosis and/or macrophage dysfunction. For example, the disease may be COPD or I PF. The medicament may be for the treatment of any of the above disease or disorders.

Summary of the Figures

Embodiments and experiments illustrating the principles of the disclosure will now be discussed with reference to the accompanying figures in which:

Figure 1 illustrates schematically the results of a process of identifying convergent BCR sequences in two individuals who were positive for amyloid but negative for tau and cognitively normal. The numbers indicated are numbers of clonotypes. The two individuals shared 64 clonotypes and a representative sequence from one of the 64 shared clonotypes was shown to bind to CD33 (see below). Figure 2 shows the results of an ELISA for binding of ATL_0005082 (also referred to herein as ATL_5082) to CD33. Isotype control is a commercially available human lgG1 isotype control (Absolute Antibody; Ab00102-10.0 Anti-Fluorescein) and does not show binding to CD33. Lysozyme= negative control antigen. CD33=binding to recombinant human CD33

Figure 3 shows the results of serum screening by ELISA across a broad panel of individual donors. The graph shows the level of autoreactivity to CD33 in multiple individuals by age. Z-score is calculated as [raw signal - mean signal]/Standard deviation (SD).

Figure 4 shows the results of an ELISA for identification of CD33 reactive plasma from a cohort of supercentenarians. Twelve plasma samples were tested by ELISA for binding to Recombinant human CD33 (A) and Lysozyme control (B). A subject with the highest signal for CD33 (arrow, SU_0000877) was chosen for phage display library generation.

Figure 5 shows the results of a phage ELISA for identification of single-chain variable fragment (scFv) clones binding to CD33 antigen. Phage-displayed scFvs derived from phage display selections were tested for binding to recombinant CD33 (pink circles) and lysozyme control (black circles).

Figure 6 shows the result of ELISA for binding of nine anti-CD33 antibodies identified by phage display, and a comparative antibody (ATL_5503) to the full extracellular region of CD33 (clear bars, “CD33 HUMAN_M17-H259_002_004”) or the C2 domain (dark blue bars, “CD33_HUMAN_ECD_Cdomain_His_007”). Results are shown as signal versus isotype.

Figure 7 shows quantification of CD33 levels on human monocytes 5 hours after incubation with the indicated antibodies at a concentration of between 1 nM and 40nM, measured by flow cytometry. The graph in (A) shows MFI (Median fluorescence intensity on CD14+CD33+ cells plotted against antibody concentration. The tested (ATL) antibodies exclusively bind the V domain of CD33 receptor while the CD33 antibody used for detection by flow cytometry is specific for the C domain. Reduction in median fluorescence intensity can be attributed to loss of the CD33 receptor from the cell membrane surface since there is no interference between the detection antibody binding and the test antibody binding. ATL5909 shows a complete depletion of CD33 at all concentrations tested whereas ATL5802 shows a complete lack of CD33 depletion, comparable to the isotype control, at all concentrations. ATL5810 has an intermediate profile, with a dose dependent decrease in CD33 levels. (B) shows MFI ofCD33 domain V on CD14+CD16+ (bottom) or CD14+CD16- (top) of the indicated antibodies. (C) shows MFI of CD33 domain C on CD14+CD16+ (bottom) or CD14+CD16- (top) of the indicated antibodies. It is worth noting that, since the test antibodies all bind in the V domain, changes to MFI of domain V may be attributable to both competition with the detection antibody and depletion of CD33. In contrast, the assay using detection of MFI in domain C will be a true representation of CD33 depletion. (D) Area under the curve (AUC) of binding to CD33 C domain and V domain as a function of concentration ranging from 40 to 1 nM for CD14+CD16- cells and CD14+CD16+ cells (curves on bottom panels in B and C). Data are normalised to the AUC for the isotype control antibody, which was set to 100. (E) Area under the curve of binding to CD33 C domain and V domain as a function of concentration ranging from 40 to 1 nM (curves in top panels in B and C). Figure 8 shows the set-up and results of an in vitro phagocytosis assay in inflammatory human iPSC- derived microglia. (A) schematically shows how the phagocytosis assay is carried out. Microglia are incubated with LPS for 24 hours, followed by incubation with an anti-CD33 antibody (or isotype control) for 24 hours, followed by measuring uptake of pHrodo red-labelled amyloid beta. pHrodo-red only fluoresces red in low pH environment of the lysosome of the cell after phagocytosis of labelled amyloid beta has occurred. (B) shows the results of a phagocytosis assay testing the effect of ATL5802 and ATL5810 on phagocytosis in microglia, compared with the prior art antibody ATL5909, an isotype control and a negative control (Ap only), in terms of surface area of red cells per well over time with incubation of the antibody. (C) shows the results of the phagocytosis assay of panel B represented as the area under the curve of surface area (pm²) of red cells per well over time, which represents uptake of pHrodo red-amyloid beta into those cells by phagocytosis. (D) shows the results of a phagocytosis assay testing the effect of ATL5802, ATL_5853, ATL_5854, and ATL_6044, on phagocytosis in microglia, compared with the prior art antibody ATL5909 and isotype control. (E) shows the quantification of the results of the phagocytosis assay on panels D and F represented as area under the curve of the surface area (pm²) of red cells per well over time, after 18 hours. The surface area of red cells represents uptake of pHrodo red-amyloid beta into those cells, and is therefore indicative of the level of phagocytosis. Statistical analysis was carried out using a one-way Anova versus isotype control. * =p<0.05; ** = p< ... ; ***= p<.., ns= not significant. At least ATL_5854 showed a statistically significant difference when tested directly vs ATL_5909 (one-way Anova) (F) shows the results of experiments comparing phagocytosis of labelled amyloid beta in iPSC-derived microglia cells treated with anti-TREM2 antibodies (ATL6166; ATL 6167; and ATL6170) or with ATL5802 (top and bottom panels show repeats of the same experiments).

Figure 9 shows the results of an ex vivo phagocytosis assay using human whole blood. (A) shows a flow cytometry plot of pHrodo red-labelled S. Aureus myeloid cells (pHrodo⁺CD14 cells), which indicates uptake of the pHrodo in CD14+ cells. PHrodo-red only fluoresces red in low pH environment of the lysosome of the cell after phagocytosis of labelled S.Aureus has occurred. (B) shows pHrodo in CD14⁺ cells (which indicates phagocytosis) measured as MFI of pHrodo within CD14+ myeloid cells incubated with the indicated antibodies (ATL5802; ATL5810; ATL5909) or isotype control. Negative controls included wells without pHrodo-labelled S. Aureus and untreated cells. (C) shows phagocytosis of pHrodo-labelled S. Aureus on freshly isolated CD14+ monocyte cells (red object count per well, corresponding to cells having pHrodo, normalised to maximum value observed) treated with ATL5802 from 0.029nM to 30 nM (shades of green), ATL5909 benchmarkl (red), ATL4828 Benchmark2 (blue) and isotype control (yellow). Latrunculin A prevents actin polymerization in cells and was used as control which would be expected to fully block phagocytosis.

Figure 10 shows CD33 levels on peripheral myeloid cells in CD34+ NSG mice 24 hours after intraperitoneal (ip) injection of antibody (ATL5802; 5810; 5909 or isotype control) at the indicated concentrations (10 mg/kg or 40 mg/kg). The level of CD33 on the surface of human engrafted myeloid cells (defined as huCD45+huCD14+ cells) was measured with an antibody specific to the C-Domain of CD33. Values reported are the fluorescence intensity (MFI) of CD33 on human CD45⁺CD14⁺cells. Figure 11 shows a schematic illustration of the mutations introduced to the CD33 V-domain, with sialoglycan ligand shown centrally in gold, for epitope mapping of antibodies described herein (A) and corresponding results (B). Variants of CD33 extracellular domain were prepared as individual recombinant proteins as indicated in A. Nine IgG 1 antibodies were tested in a single point ELISA against the wild-type CD33 and all 7 CD33 variant proteins and the binding relative to binding to the wt CD33 extracellular domain was quantified. The binding profile of each antibody to the different CD33 variants was used to group the antibodies into 4 different epitope bins as shown in B.

Figure 12 shows CD33 levels on human CD11 b+CD45+ cells in the brain of CD34⁺ NSG mice 24 hours after intraperitoneal (ip) injection of antibody (ATL5802; 5810; 5909 or isotype control). Cells were isolated from brain homogenate using anti-CD11 b magnetic beads and then stained for human and mouse anti- CD45. CD33 levels were assessed on human cells and the median fluorescence intensity of CD33 staining is reported.

Figure 13 shows the results of a flow cytometry assay detecting median fluorescence intensity of pHrodo- labelled S. aureus in human CD45+ CD14⁺CD11 b⁺ peripheral myeloid cells in CD34⁺NSG mice 24 hours after intraperitoneal injection of antibody (ATL5802; 5810; 5909 or isotype control) - 3 hours ex vivo incubation with labelled S. aureus.

Figure 14 shows SEC-HPLC chromatograms for ATL5802 at the indicated forced degradation conditions (2 weeks at -80°C, 2 weeks at 40°C, shaking overnight, 3 freeze-thaw cycles (3x FT); and low pH hold).

Figure 15 shows the results of an ELISA testing the binding to ATL5802 to recombinant human CD33 (rhCD33) at the indicated forced degradation conditions (2 weeks at -80°C, 2 weeks at 40°C, shaking overnight, 3 freeze-thaw cycles (3x FT); and low pH hold).

Figure 16 shows an alignment of variable chain sequences for selected anti-CD33 antibodies. Dashed lines indicate comparative antibodies. A. VH FWR1 . B. VH CDR1 . C. VH FWR2. D. VH CDR2. E. VH FWR3. F. VH CDR3. G. VH FWR4. H. VL FWR1. I. VL CDR1. J. VL FWR2. K. VL CDR2. L. VL FWR3. M. VL CDR3. N. CL FWR4.

Figure 17 shows the results of an in vivo pK study in C57BL/6-Cd33tm1 (CD33) mice. Groups of mice (n=3/group) were administered 1 mg/kg of ATL-5802 or ATL-5909 intraperitoneally and serum levels of hlgG1 were measured at 4 hours, 24 hours and 7-days following a single administration of antibodies. (A) shows a graph plotting the serum levels of ATL_5802 or ATL_5909 at the indicated time points. (B) Analysis of total antibody concentration over time (area under the curve; AUC). **p<0.01 , unpaired t-test.

Figure 18 shows the results of an in vivo study in an animal model of Alzheimer’s disease testing the ability of ATL_5802 to alter in vivo phagocytosis of amyloid-p in APP^{NL G F} mice xenotransplanted with human microglia (hMG). Following 12 weeks of weekly treatment with 40mg/kg ATL_5802 (n=6) or isotype control, ATL_5338 (n=7), the uptake of amyloid-p by CD45+ human microglia was measured by flow cytometry using the florescent amyloid-p label, Methoxy-X04. There was a significant increase in the percent of human microglial cells positive for Methoxy-X04 in mice treated with ATL_5802 compared with mice treated with the control, ATL_5338, indicating that ATL_5802 increased phagocytosis of amyloid-p by human microglia in vivo. *p<0.05, unpaired t-test. Figure 19 shows the results of a real-time live cell imaging study using LPS-primed microglia and pHrodo- la belled Tau P301 S aggregates. (A) Representative plot of total red (pHrodo.red) area per well versus time elapsed for antibodies ATL_5338 (Isotype control); ATL_5802; ATL_5909; ATL_6170; and LPS alone. (B) Quantification of Area Under the Curve (AUC) of total red per well are shown in (A).

Figure 20 shows the results of a multiplex immunoassay in iPSC microglia pre-treated with LPS for 6 hours followed by treatment with ATL_5802 or ATL_5909. (A) Shows the concentration (pg/ml) of IL-6 in the supernatants of iPSC microglia treated with LPS and ATL5802 or an isotype control antibody. The data show that ATL_5802 decreases microglial IL-6 release in response to LPS. (B) Shows the concentration of MCP-1 (pg/ml) in the supernatants of iPSC microglia treated with LPS and ATL5802 or an isotype control antibody. The data show that ATL_5802 decreases microglial MCP-1 release in response to LPS. Statistical analysis was carried out using a Student's t test, * =p<0.05; ** = p<0.01 ; ***= p<0.001 , ns= not significant.

Figure 21 shows the results of a western blot for phosphorylated SYK (pSYK) in cultured microglia cells stimulated with LPS and treated with ATL_5802. The figure shows a Western blot for pSYK (Tyr525/526) and total SYK in microglia cells 20 minutes after treatment with isotype control antibody or ATL_5802.

Figure 22 shows the results of a western blot for P2RY12 in cultured microglia cells stimulated with LPS, 6 hours after treatment with ATL_5802 or isotype control antibody. The housekeeper molecule GAPDH was used as a control.

Figure 23 shows the results of a gene expression analysis following RNA sequencing on iPSC-derived microglia stimulated with LPS/interferon gamma (LI) or vehicle and treated with ATL_5802; ATL_5854, or ATL_5909. (A) shows a principal component analysis (PCA) plot of RNAseq data showing principal component 1 (PC) versus PC2, from PCA calculated for 25% most variable genes. (B-C) shows volcano plots of gene expression changes in iPSC-derived microglia treated with ATL_5802 versus isotype control in vehicle stimulated (B) or LPS/interferon gamma stimulated conditions (C). The volcano plot shows fold changes (x-axis) versus statistical significance, expressed as the -Iog10 of adjusted p value (y-axis). Coloured points (points left of the leftmost vertical line, right of the rightmost vertical line and above the horizontal line) indicated differentially expressed with statistical significance as per the selected thresholds; red (positive log2FoldChange): upregulated upon treatment with ATL_5802, blue (negative log2FoldChange): downregulated. Horizontal line marks threshold for significance, vertical lines marks thresholds for log2FoldChange. (D) shows a bar plot with the number of differentially expressed genes as shown in (B) and (C) and its direction of change (i.e. upregulated or downregulated), for all stimuli and antibody treatments. (E) Bar plot showing the results of a Gene Set Enrichment analysis (GSEA) for oxidative phosphorylation and respiratory electron transport gene sets/pathways with adjusted p-values (x- axis) and normalised enrichment scores (NES, colour intensity) per comparison (y-axis) and pathway (vertical boxes 1 to 4). Red (top bar in each set of 3 horizontal bars): upregulated, blue (middle and lower bar in each set of 3 horizontal bars): downregulated. NES is represented by the intensity of colour. Vertical line marks threshold for significance. (F) Representative GSEA plot for the REACTOME pathway ’’Respiratory electron transport” for all anti-CD33 antibodies compared against isotype control when stimulated with LPS/interferon gamma. These plots show that the pathway is much more significantly enriched upon treatment with ATL_5802 (and to a lower extent ATL_5854) compared to comparative antibody ATL_5909.

Figure 24 shows the results of analysis of effect of antibodies of the disclosure in a human central nervous system (CNS) co-culture experiment. (A) Shows a cartoon illustrating the human CNS quad cell cultures comprising human iPSC-derived Glutamatergic/GABAergic neurons, astrocytes and microglia and a fluorescent image showing cells in such a coculture. (B - E) levels of interleukin-6 (IL-6) (B); GFAP (C), IP- 10 (D); and MCP-1 (E). Two-way ANOVA with Tukey’s multiple comparisons test (n= 2-3 wells), *p<0.05, **p<0.01 , ***p<0.001 ****p<0.0001 compared to isotype treated LPS/IFNy group. The data in B-C show that ATL_5802 protects against inflammation in human iPSC-derived CNS Quad-cultures, as evidenced by the reduction in interleukin-6 (IL-6) and a decrease in GFAP, a marker of astrogliosis. The data in D-E show a significant decrease in IP-10 (Interferon gamma-induced protein 10) and MCP-1 levels with ATL_5802 treatment under LPS/IFNy conditions.

Detailed Description of the Invention

Aspects and embodiments of the present invention will now be discussed with reference to the accompanying figures. Further aspects and embodiments will be apparent to those skilled in the art. All documents mentioned in this text are incorporated herein by reference.

Disclosed herein are antibodies and fragments thereof that are capable of specifically binding to CD33 protein or a fragment thereof. As used herein, an antibody capable of “specific binding” or “specifically binding” a target is one able to bind through the association of the epitope recognition site with an epitope within the target. It is distinct from non-specific binding, for example Fc-mediated binding, ionic and/or hydrophobic interactions. In other words, an antibody which specifically binds a target recognises and binds to a specific protein structure within it rather than to proteins generally.

The present disclosure refers to antibodies described herein using references specified as “ATL_000xxxx”, “ATL_xxxx” or “xxxx”, where “xxxx” is a four digits reference number specific to an antibody described herein. All of the above notations are used interchangeably to refer to the same antibody or a portion thereof (e.g. a VH, VL or part thereof, of the antibody). For example, antibody ATL_0005802 is interchangeably referred to herein as ATL_5802 and 5802. Further the reference ATLX-1088 refers to ATL_5802.

CD33 is an inhibitory immune receptor belonging to the family of sialic acid binding immunoglobulin-like lectins. CD33 is also known as of sialic acid binding immunoglobulin-like lectin 3 (Siglec-3). The human CD33 gene (gene ID: 945) is located on chromosome 19 (19q13.33) in humans and consists of seven exons. The full sequence of human CD33 is a 364 amino acids long sequence available under Uniprot identifier P20138. It consists of (i) an N-terminal signal peptide that targets CD33 to the secretory pathway (amino acids 1-17 of P20138); (ii) two extracellular domains, consisting of the N-terminal Ig-like V-set domain for recognition of carbohydrate ligands, and a C2-set domain (amino acids 18-259 of P20138); (iii) a transmembrane domain (amino acids 260-282 of P20138); and (iv) a cytoplasmic domain with immunoreceptor tyrosine-based inhibitory motifs (ITIM) that mediate immune cell signalling (amino acids 283-364 of P20138) (Eskandari-Sedighi et al., 2023). CD33 undergoes alternative splicing to generate a long isoform, denoted as hCD33M (M=’major’), and a short isoform denoted as hCD33m (m = ‘minor’), which is generated through exon 2 exclusion (amino acids 13-139 of hCD33M) and lacks a functional V domain (Hernandez-Caselles, et al., 2006). The major isoform of human CD33 (hCD33) is a 67 kDa transmembrane glycoprotein available under Uniprot identifier P20138-1 (canonical sequence, 364 amino acids), provided as SEQ ID NO: 176. The minor isoform is a 25 kDa protein available under Uniprot identifier P20138-3 (237 amino acids), provided as SEQ ID NO: 177. Antibodies described herein may bind to hCD33M and not to hCD33m. CD33 is expressed on the cell surface of myeloid cells, monocytes, macrophages, and microglia in the brain. CD33 participates in adhesion processes of immune cells and mediates cell-cell interaction (Varki et al., 2006).

Genome-wide association studies have identified CD33 as a genetic modulator of Late-Onset Alzheimer’s disease susceptibility and pathology (Hollingworth et al. 2011). Upregulation of CD33 expression on microglia of AD patients is associated with more advanced cognitive decline or disease status (Siddiqui et al, 2017). It is thought that this is due to altered microglial activation and inhibition of microglial-mediated phagocytosis of the toxic insoluble amyloid beta 42 (Ap42) species. On the other hand, an AD protective allele of the CD33 SNP rs3865444 results in lower CD33 expression and a higher proportion of a truncated, non-signalling form, along with reduced levels of insoluble Ap42 in AD brain (Griciuc et al., 2013). In addition, CD33 has been suggested to interact with other microglial AD risk genes, such as TREM2, to influence AD onset and pathogenesis. In particular, it has been suggested that CD33 negatively regulates TREM2/DAP12- mediated microglial activation, resulting in reduced cellular function such as microglial phagocytosis which in turn leads to reduced clearance of toxic Ap42 species (Chan et al., 2015). Accordingly, the antibodies described herein are capable of increasing phagocytosis of cells expressing CD33, such as microglial phagocytosis, compared with a comparative antibody.

In addition to its role in inhibiting cellular processes such as phagocytosis, CD33 is involved in several other processes including cell adhesion and the modulation of immune responses (Cao et al., 2010). CD33 is also highly expressed on leukemic blasts in acute myeloid leukaemia, as well as myeloid leukaemia initiating cells (Bonnet et al., 1997; Vercauteren et al., 2007) . Accordingly, in some instances, the antibodies described herein may be useful in the treatment of AML by binding to CD33 in leukemic blast cells.

CD33 has endocytic properties (receptor internalisation), a property which has been exploited for targeting by antibody-drug conjugates (Laszlo et al., 2014). Although this property may result in a decrease of CD33 at the cell surface, it could also reduce the efficacy of CD33-directed therapies due to internalisation of the anti-CD33 antibodies by CD33-expressing cells that are not the primary therapeutic targets. For example, when targeting CD33-expression microglial cells, CD33-expressing monocytes may represent a peripheral antibody sink that may prevent the antibody from reaching therapeutic doses in the brain with acceptable toxicity. In some instances, the antibodies described herein induce reduced CD33 endocytosis compared with a comparative antibody. CD33 internalisation may be measured, for example, by determining cell surface levels of CD33 using flow cytometry in the presence of a candidate antibody. The cell may be a human cell such as a microglial cell, myeloid cells, a monocyte, a tumour cell, such as a leukemic blast cell, and/or an iPSC derived microglial cell and/or a cell that has been stimulated with an inflammatory signal (e.g. LPS) prior to exposure to the antibody. A comparative antibody may be selected from: an isotype control antibody, and/or an antibody with the heavy chain variable sequence of ATL_5909 and the light chain variable sequence of ATL_5909.

Therapeutic strategies that target TREM2, an anti-inflammatory receptor expressed in myeloid cells have been explored for the treatment of AD, with the aim of restoring normal microglial function. Indeed, loss of TREM2 function is known to reduce the response of microglia to amyloid beta plaque. For example, multiple TREM2 agonists are currently under clinical investigation for the treatment of AD, including Alector/Abbvie’s AL002 (referred to herein as ATL6166, in phase 2 clinical trial), Denali Therapeutic Inc’s DNL919 (referred to herein as ATL6167, in phase 1 clinical trial), and Vigil’s VGL101 (referred to herein as ATL6170; in phase 2 clinical trial). Although CD33 is known to modulate TREM2 activity, no direct CD33 antagonist antibodies are currently in clinical development to the best of the inventors’ knowledge.

As used herein, the term “CD33” encompasses truncations, derivatives, and variants of the sequence of CD33 provided herein as SEQ ID NO: 176 or a homolog thereof, and may refer to any protein with at least 80%, at least 90%, or at least 95% sequence identity with said sequence. The CD33 sequence may be a human CD33 sequence. The CD33 sequence may be a sequence that comprises at least part of the V domain (also referred to as “Ig-like V-type” domain) of CD33 (amino acids 19-135 of Uniprot ID P20138-1 or a homolog thereof). The CD33 sequence may further comprise one or more of: at least part of the C domain (also referred to as “Ig-like C2-type” domain) of CD33 (amino acids 145-228 of Uniprot ID P20138- 1 or a homolog thereof), at least part of the transmembrane domain of CD33 (amino acids 260-282 of P20138-1 or a homolog thereof), and at least part of the cytoplasmic domain of CD33 (amino acids 283- 364 of P20138-1 or a homolog thereof). Antibodies according to the present disclosure may bind to CD33 proteins that comprise the V domain of human CD33 (amino acids 19-135 of Uniprot ID P20138-1 provided as SEQ ID NO: 176) or a sequence that has at least 90%, 95%, 98% or 99% sequence identity with said sequence.

The present disclosure relates primarily to antibody molecules, whether whole antibody (e.g. IgG, such as lgG1) or antibody fragments (e.g. single-chain variable fragment (scFv), Antibody fragments (Fab) or bivalent antibody fragments (F(ab’)2), single-domain antibody (sdAb). Antibody antigen binding regions (also referred to as “antigen binding portions”) are provided, as are antibody heavy chain variable (VH) and light chain variable (VL) domains. Within VH and VL domains are provided complementarity determining regions (CDRs), which may be provided within different framework regions (FRs), to form VH or VL domains, as the case may be. An antigen binding site may consist of an antibody VH domain and/or a VL domain.

Antibodies according to the present disclosure may be provided in isolated form. The term “antibody” encompasses a fragment or derivative thereof, or a synthetic antibody or antibody fragment. An antibody or fragment thereof may be a monoclonal antibody (mAb). mAbs are homogenous populations of antibodies specifically targeting a single epitope on an antigen. Antibodies and methods for their construction and use are well-known in the art and are described in, for example, Holliger & Hudson, Nature Biotechnology 23(9): 1126-1 136 (2005). In view of today's techniques in relation to monoclonal antibody technology, antibodies can be prepared to most targets. It is possible to take monoclonal and other antibody molecules and use techniques of recombinant DNA technology to produce other antibody or chimeric molecules which retain the specificity of the original antibody. Such techniques may involve introducing CDRs or variable regions of one antibody into a different antibody molecule (see e.g. GB 2188638A and EP0239400).

An “antigen binding domain” describes the part of a molecule that binds to all or part of the target antigen. An antibody generally comprises six complementarity-determining regions (CDRs); three in the VH region: HCDR1 , HCDR2 and HCDR3, and three in the VL region: LCDR1 , LCDR2, and LCDR3. The six CDRs together define the paratope of the antigen binding domain, which is the part of the antigen binding domain which binds to the target antigen. A monoclonal monospecific IgG antibody molecule contains two antigen binding domains, each of which are able to bind the same target (i.e. it is bivalent for a single target). A Fab fragment generally comprises a VH domain, a CH1 domain, a VL domain and a CL domain. A full antibody may comprise a pair of Fab fragments and an Fc fragment comprising a pair of chains each comprising a CH2 domain and a CH3 domain. An Fv fragment comprises a VH domain and a VL domain. The VH region and VL region comprise framework regions (FRs) either side of each CDR, which provide a scaffold for the CDRs. From N-terminus to C-terminus, VH regions comprise the following structure: N term-[HFR1]- [HCDR1]-[HFR2]-[HCDR2]-[HFR3]-[HCDR3]-[HFR4]-C term; and VL regions comprise the following structure: N term-[LFR1]-[LCDR1]-[LFR2]-[LCDR2]-[LFR3]-[LCDR3]-[LFR4]-C term.

The term "ScFv molecules" refers to molecules wherein the VH and VL partner domains are covalently linked, e.g. by a flexible oligopeptide. Fab, Fv, ScFv and sdAb antibody fragments can all be expressed in and secreted from E. coli, thus allowing the facile production of large amounts of the said fragments. Whole antibodies, and F(ab')2 fragments are "bivalent". The term "bivalent" means that the said antibodies and F(ab')2 fragments have two antigen combining sites. In contrast, Fab, Fv, ScFv and dAb fragments are monovalent, having only one antigen combining site.

Antibodies according to the present disclosure may be detectably labelled or, at least, capable of detection. For example, the antibody may be labelled with a radioactive atom or a coloured molecule or a fluorescent molecule or a molecule which can be readily detected in any other way. Suitable detectable molecules include fluorescent proteins, luciferase, enzyme substrates, and radiolabels. The binding moiety (antibody or fragment thereof) may be directly labelled with a detectable label or it may be indirectly labelled. For example, the binding moiety may be an unlabelled antibody which can be detected by another antibody which is itself labelled. Alternatively, the second antibody may have bound to it biotin and binding of labelled streptavidin to the biotin is used to indirectly label the first antibody.

A ’’fragment” of an antibody may comprise any number of residues of a “parental” antibody, whilst retaining target binding ability. A fragment may lack effector function, for example may be entirely unable to bind or show diminished binding to the Fc receptor, relative to the parent. A fragment is typically smaller than the parental antibody. A fragment may comprise 50%, 60%, 70%, 80%, 90%, 95% or more of the contiguous or non-contiguous amino acids of the parental antibody. A fragment may comprise 50, 100, 150, 200, 250, 300 or more contiguous or non-contiguous amino acids of the parental antibody. A fragment may comprise deletions in the Fc region, or of the Fc region. A fragment may retain the CDRs and/or the variable domains of the parental antibody, unaltered. In some embodiments, a fragment is a Fab fragment or an F(ab’)2 fragment. CDR sequences are described herein using the IMGT numbering (Lefranc, M.-P., Immunology Today, 18, 509 (1997)).

Antibodies according to the present disclosure may have a VH with the VH-CDRs of the following antibodies: ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, or a set of CDRs which contains zero, one or two amino acid substitutions in each CDR compared with the above set of CDRs. Thus, the isolated antibody may comprise a heavy chain variable domain with the following CDRs: a CDRH1 comprising an amino acid sequence selected from: (i) HCDR1_ATL_0005802; HCDR1_ATL_0005807; HCDR1_ATL_0005808: GYSFTSYW (SEQ ID NO:44); (ii) HCDR1_ATL_0005809: GYSFNTYW (SEQ ID NO:45); (iii) HCDR1_ATL_0005810: GYTFTSYY (SEQ ID NO:46); (iv) HCDR1_ATL_0005853: GYKFNNNW (SEQ ID NO:47); (v) HCDR1_ATL_0005854: GYKFSNNW (SEQ ID NO:48); or (vi) an amino acid sequence with 1 or 2 mutations compared to the above sequences; a CDRH2 comprising an amino acid sequence selected from: (i) HCDR2_ATL_0005802;

HCDR2_ATL_0005807; HCDR2_ATL_0005808; HCDR2_ATL_0005853; HCDR2_ATL_0005854: IYPGDSDT (SEQ ID NO: 52); (ii) HCDR2_ATL_0005809: IYPGDSET (SEQ ID NO:53); (iii) HCDR2_ATL_0005810: INPSGGST (SEQ ID NO:54); or (iv) an amino acid sequence with 1 or 2 mutations compared to the above sequences; and a CDRH3 comprising an amino acid sequence selected from: (i) HCDR3_ATL_0005802: ARPRGFGEYYFDY (SEQ ID NO: 59); (ii) HCDR3_ATL_0005807: ARQGAGPGGFDI (SEQ ID NO:60); (iii) HCDR3_ATL_0005808: ARGQGGGAGAFDI (SEQ ID NO:61); (iv) HCDR3_ATL_0005809: ARHRADAPSDAFDI (SEQ ID NO:62); (v) HCDR3_ATL_0005810: ARDWHNTPQFDTDYYYYGMDV (SEQ ID NO:63); (vi) HCDR3_ATL_0005853: ARHSGGLDGYTAAALDY (SEQ ID NO:64); (vii) HCDR3_ATL_0005854: ATWGGSNWFVD (SEQ ID NO:65); or (viii) an amino acid sequence with 1 or 2 mutations compared to the above sequences.

An antibody according to the present disclosure may comprise a heavy chain variable domain (VH) with the following CDRs: a CDRH1 comprising the sequence of HCDR1_ATL_0005802, a CRH2 comprising the sequence of HCDR2_ATL_0005802, and a CDRH3 comprising the sequence of HCDR3_ATL_0005802, or a set of CDRs comprising 1 or 2 mutations in the CDRH1 and CDRH2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRH3 compared to these sequences.

An antibody according to the present disclosure may comprise a heavy chain variable domain (VH) with the following CDRs: a CDRH1 comprising the sequence of HCDR1_ATL_0005853, a CRH2 comprising the sequence of HCDR2_ATL_0005853, and a CDRH3 comprising the sequence of HCDR3_ATL_0005853, or a set of CDRs comprising 1 or 2 mutations in the CDRH1 and CDRH2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRH3 compared to these sequences.

An antibody according to the present disclosure may comprise a heavy chain variable domain (VH) with the following CDRs: a CDRH1 comprising the sequence of HCDR1_ATL_0005854, a CRH2 comprising the sequence of HCDR2_ATL_0005854, and a CDRH3 comprising the sequence of

HCDR3_ATL_0005854, or a set of CDRs comprising 1 or 2 mutations in the CDRH1 and CDRH2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRH3 compared to these sequences.

An antibody according to the present disclosure may comprise a heavy chain variable domain (VH) with the following CDRs: a CDRH1 comprising the sequence of HCDR1_ATL_0005810, a CRH2 comprising the sequence of HCDR2_ATL_0005810, and a CDRH3 comprising the sequence of HCDR3_ATL_0005810, or a set of CDRs comprising 1 or 2 mutations in the CDRH1 and CDRH2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRH3 compared to these sequences.

An antibody according to the present disclosure may comprise a heavy chain variable domain (VH) with the following CDRs: a CDRH1 comprising the sequence of HCDR1_ATL_0005807, a CRH2 comprising the sequence of HCDR2_ATL_0005807, and a CDRH3 comprising the sequence of HCDR3_ATL_0005807, or a set of CDRs comprising 1 or 2 mutations in the CDRH1 and CDRH2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRH3 compared to these sequences.

An antibody according to the present disclosure may comprise a heavy chain variable domain (VH) with the following CDRs: a CDRH1 comprising the sequence of HCDR1_ATL_0005808, a CRH2 comprising the sequence of HCDR2_ATL_0005808, and a CDRH3 comprising the sequence of HCDR3_ATL_0005808, or a set of CDRs comprising 1 or 2 mutations in the CDRH1 and CDRH2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRH3 compared to these sequences.

An antibody according to the present disclosure may comprise a heavy chain variable domain (VH) with the following CDRs: a CDRH1 comprising the sequence of HCDR1_ATL_0005809, a CRH2 comprising the sequence of HCDR2_ATL_0005809, and a CDRH3 comprising the sequence of HCDR3_ATL_0005809, or a set of CDRs comprising 1 or 2 mutations in the CDRH1 and CDRH2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRH3 compared to these sequences.

In an antibody according to the present disclosure at least one of the VH CDR 1-3 sequences may vary. A variant may have one or two amino acid substitutions compared with the set of VH CDR1-3 described above. In embodiments, an antibody according to the present disclosure comprises CDRs with sequences that have one or two substitutions compared with VH CDR sequences of any antibody described herein. For example, an antibody according to the disclosure may comprise VH CDRs with the sequences of any antibody above, except that one or two of the CDRHs comprise a substitution, where the total number of substitutions across CDRHs does not exceed two. In embodiments, a variant may have one, two or three substitutions, preferably at most one or two substitutions in each of one or more of the VH CDR1-3 described above. CDRH1 regions of any antibodies or fragments described herein may have a length of 8 amino acids. CDRH2 regions of any antibodies or fragments described herein may have a length of 8 amino acids. CDRH3 regions of any antibodies or fragments described herein may have a length of 13 to 21 amino acids. In embodiments, a variant may have VH CDRs that have at least 70%, at least 80% or at least 90% sequence identity with any set of VH CDRs described herein.

Antibodies according to the present disclosure may have a light chain variable domain (VL) with the CDRs of the light chain variable domain (VL) of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854,

ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, or a set of CDRs which contains zero, one, or two amino acid substitutions in each CDR compared with the above set of CDRs. Thus, the isolated antibody may comprise a light chain variable domain with the following CDRs: a CDRL1 comprising an amino acid sequence selected from: (i) LCDR1_ATL_0005802;

LCDR1_ATL_0006040; LCDR1_ATL_0006041 ; LCDR1_ATL_0006042; LCDR1_ATL_0006043;

LCDR1_ATL_0006044; LCDR1_ATL_0006045; LCDR1_ATL_0006046; LCDR1_ATL_0006047;

LCDR1_ATL_0006048: SSNIGAGYD (SEQ ID NO: 110); (ii) LCDR1_ATL_0005807: ALARQY (SEQ

ID NO: 1 11); (iii) LCDR1_ATL_0005808; LCDR1_ATL_0005854: SSDVGGYNY (SEQ ID NO: 112); (iv) LCDR1_ATL_0005809: SLRNYY (SEQ ID NO: 113); (v) LCDR1_ATL_0005810: SGSVSTSYY (SEQ ID NO: 114); (vi) LCDR1_ATL_0005853: QSLLHSDGYNY (SEQ ID NO: 115); or (vii) an amino acid sequence with 1 or 2 mutations compared to the above sequences; a CDRL2 comprising an amino acid sequence selected from: (i) LCDR2_ATL_0005802; LCDR2_ATL_0006040; LCDR2_ATL_0006041 ; LCDR2_ATL_0006042; LCDR2_ATL_0006043;

LCDR2_ATL_0006044; LCDR2_ATL_0006045; LCDR2_ATL_0006046; LCDR2_ATL_0006047;

LCDR2_ATL_0006048: GNS (SEQ ID NO:119); (ii) LCDR2_ATL_0005854: DVS (SEQ ID NO:120);

(iii) LCDR2_ATL_0005807: KDS (SEQ ID NO:121); (iv) LCDR2_ATL_0005808: DVT (SEQ ID

NO:123); (v) LCDR2_ATL_0005809: GKN (SEQ ID NO:124); (vi) LCDR2_ATL_0005810: STN (SEQ ID NO:125); (vii) LCDR2_ATL_0005853: VGS (SEQ ID NO:126) ; or (viii) an amino acid sequence with 1 or 2 mutations compared to the above sequences; and a CDRL3 comprising an amino acid sequence selected from: (i) LCDR3_ATL_0005802; LCDR3_ATL_0006040; LCDR3_ATL_0006041 ; LCDR3_ATL_0006042; LCDR3_ATL_0006043; LCDR3_ATL_0006044; LCDR3_ATL_0006045; LCDR3_ATL_0006046; LCDR3_ATL_0006047; LCDR3_ATL_0006048: QSYDSSLSGDV (SEQ ID NO: 129); (ii) LCDR3_ATL_0005807: QSPDSSGTYPV (SEQ ID NO: 130); (iii) LCDR3_ATL_0005808: SSYTSSSTLEV (SEQ ID NO: 131); (iv) LCDR3_ATL_0005809: NSRDSSGYHLGL (SEQ ID NO: 132); (v) LCDR3_ATL_0005810: LLYMGSGIWM (SEQ ID NO: 133); (vi) LCDR3_ATL_0005853: MQALQTPIT (SEQ ID NO: 134); (vii) LCDR3_ATL_0005854: SSYTNSSTLEV (SEQ ID NO: 236); or (viii) an amino acid sequence with 1 or 2 mutations compared to the above sequences.

An antibody according to the present disclosure may comprise a light chain variable domain with the following CDRs: a CDRL1 comprising an amino acid sequence selected from: LCDR1_ATL_0005802: SSNIGAGYD (SEQ ID NO: 110); LCDR1_ATL_0005854: SSDVGGYNY (SEQ ID NO: 112); LCDR1_ATL_0005853

QSLLHSDGYNY (SEQ ID NO: 115); or an amino acid sequence with 1 or 2 mutations compared to the above sequences; a CDRL2 comprising an amino acid sequence selected from: LCDR2_ATL_0005802: GNS (SEQ ID NO:119); LCDR2_ATL_0005854: DVS (SEQ ID NO:120); LCDR2_ATL_0005853: VGS (SEQ ID

NO:126); or an amino acid sequence with 1 or 2 mutations compared to the above sequences; and a CDRL3 comprising an amino acid sequence selected from: LCDR3_ATL_0005802: QSYDSSLSGDV (SEQ ID NO: 129); LCDR3_ATL_0005853: MQALQTPIT (SEQ ID NO: 134); LCDR3_ATL_0005854: SSYTNSSTLEV (SEQ ID NO: 236); or an amino acid sequence with 1 or 2 mutations compared to the above sequences.

An antibody according to the present disclosure may comprise a light chain variable domain (VL) with the following CDRs: a CDRL1 comprising the sequence of LCDR1_ATL_0005802, a CRL2 comprising the sequence of LCDR2_ATL_0005802, and a CDRH3 comprising the sequence of LCDR3_ATL_0005802, or a set of CDRs comprising 1 or 2 mutations in the CDRL1 and CDRL2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRL3 compared to these sequences.

An antibody according to the present disclosure may comprise a light chain variable domain (VL) with the following CDRs: a CDRL1 comprising the sequence of LCDR1_ATL_0005853, a CRL2 comprising the sequence of LCDR2_ATL_0005853, and a CDRH3 comprising the sequence of LCDR3_ATL_0005853, or a set of CDRs comprising 1 or 2 mutations in the CDRL1 and CDRL2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRL3 compared to these sequences.

An antibody according to the present disclosure may comprise a light chain variable domain (VL) with the following CDRs: a CDRL1 comprising the sequence of LCDR1_ATL_0005854, a CRL2 comprising the sequence of LCDR2_ATL_0005854, and a CDRH3 comprising the sequence of LCDR3_ATL_0005854, or a set of CDRs comprising 1 or 2 mutations in the CDRL1 and CDRL2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRL3 compared to these sequences.

An antibody according to the present disclosure may comprise a light chain variable domain (VL) with the following CDRs: a CDRL1 comprising the sequence of LCDR1_ATL_0005810, a CRL2 comprising the sequence of LCDR2_ATL_0005810, and a CDRH3 comprising the sequence of LCDR3_ATL_0005810, or a set of CDRs comprising 1 or 2 mutations in the CDRL1 and CDRL2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRL3 compared to these sequences.

An antibody according to the present disclosure may comprise a light chain variable domain (VL) with the following CDRs: a CDRL1 comprising the sequence of LCDR1_ATL_0005807, a CRL2 comprising the sequence of LCDR2_ATL_0005807, and a CDRH3 comprising the sequence of LCDR3_ATL_0005807, or a set of CDRs comprising 1 or 2 mutations in the CDRL1 and CDRL2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRL3 compared to these sequences. An antibody according to the present disclosure may comprise a light chain variable domain (VL) with the following CDRs: a CDRL1 comprising the sequence of LCDR1_ATL_0005808, a CRL2 comprising the sequence of LCDR2_ATL_0005808, and a CDRH3 comprising the sequence of LCDR3_ATL_0005808, or a set of CDRs comprising 1 or 2 mutations in the CDRL1 and CDRL2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRL3 compared to these sequences.

An antibody according to the present disclosure may comprise a light chain variable domain (VL) with the following CDRs: a CDRL1 comprising the sequence of LCDR1_ATL_0005809, a CRL2 comprising the sequence of LCDR2_ATL_0005809, and a CDRH3 comprising the sequence of LCDR3_ATL_0005809, or a set of CDRs comprising 1 or 2 mutations in the CDRL1 and CDRL2 (1 or 2 mutation in each CDR or 1 or 2 mutations over both CDRs) compared to these sequences and/or 1 , 2 or 3 mutations in the CDRL3 compared to these sequences.

In an antibody according to the present disclosure at least one of the VL CDR1-3 sequences may vary. A variant may have 1 , 2, or 3 amino acid substitutions compared with the set of VL CDR1-3 described above. In embodiments, an antibody according to the disclosure comprises CDRs with sequences that have between 1 and 3 substitutions compared with the VL CDR sequences of an antibody described herein. For example, an antibody according to the disclosure may comprise a substitution, where the total number of substitutions does not exceed 3. In embodiments, a variant may have one, two or three, preferably at most one or two substitutions in each of one or more of the VL CDR1-3 described above.

CDRL1 regions of any antibodies or fragments described herein may have a length of between 6 and 11 amino acids. CDRL2 regions of any antibodies or fragments described herein may have a length of 3 amino acids. CDRL3 regions of any antibodies or fragments described herein may have a length of 9 to 11 amino acids. In embodiments, a variant may have VL CDRs that have at least 70%, at least 80% or at least 90% sequence identity with any set of VL CDRs described herein.

The VH CDRs 1-3 and optionally VL CDRs 1-3 of any of the antibodies described above may also be particularly useful in conjunction with a number of different framework regions. Accordingly, light and/or heavy chains having CDRs 1-3 as described above may possess an alternative framework region. Suitable framework regions are known in the art and are described for example in M. Lefranc & G. Le Franc (2001) "The Immunoglobulin Facts Book", Academic Press.

An antibody of the disclosure may have a heavy chain variable domain (VH) with the following framework sequences: HFWR1 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, HFWR2 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, HFWR3 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809,

ATL_0005810, and HFWR4 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, or framework sequences with one to six substitutions compared to the framework sequences above. An antibody according to the disclosure may have a heavy chain variable domain (VH) with the following framework sequences: a HFWR1 comprising an amino acid sequence selected from:

HFWR1_ATL_0005807; HFWR1_ATL_0006041 QVQLVQSGAEVKKPGESLKISCKGS (SEQ ID NO: 71)

HFWR1_ATL_0005808 QVQLQQSGAEVKKPGESLKISCKGS (SEQ ID NO:72) HFWR1_ATL_0005809 QVQLQQSGGEVKKPGESLKISCKGS (SEQ ID NO:73) HFWR1_ATL_0005810 QVQLVQSGAEVKKPGASVKVSCKAS (SEQ ID NO:74) HFWR1_ATL_0005853 QVQLVQSGAEVKKTGESLRISCKAS (SEQ ID NO:235) HFWR1_ATL_0005854 QVQLVQSGAEVKKTGEYLRISCKAS (SEQ ID NO:75) HFWR1_ATL_0006040; HFWR1_ATL_0006044; HFWR1_ATL_0006045; HFWR1_ATL_0006046; HFWR1_ATL_0006047; HFWR1_ATL_0006048: EVQLVQSGAEVKKPGESLKISCKGS (SEQ ID NO:77) HFWR1_ATL_0006042: EVQLVESGAEVKKPGESLKISCKGS (SEQ ID NO:78)

HFWR1_ATL_0006043: EVQLVQSGAEVKKPGESLRISCKGS (SEQ ID NO:79) a HFWR2 comprising an amino acid sequence selected from:

HFWR2_ATL_0005802; HFWR2_ATL_0005853; HFWR2_ATL_0005854; HFWR2_ATL_0006044: IAVWRQMPGKGLEWMGI (SEQ ID NO: 83)

HFWR2_ATL_0005807; HFWR2_ATL_0005808; HFWR2_ATL_0006040; HFWR2_ATL_0006041 ; HFWR2_ATL_0006042; HFWR2_ATL_0006043; HFWR2_ATL_0006045; HFWR2_ATL_0006046; HFWR2_ATL_0006047; HFWR2_ATL_0006048: IGVWRQMPGKGLEWMGI (SEQ ID NO: 84) HFWR2_ATL_0005809: IAWVRHAPGKGLEWMGI (SEQ ID NO: 85)

HFWR2_ATL_0005810: MHWVRQAPGQGLEWMGI (SEQ ID NO: 86) a HFWR3 comprising an amino acid sequence selected from:

HFWR3_ATL_0005802: RYSPSFEGQVTISADKSIGTAYLQWSSLKASDTAMYYC (SEQ ID NO:91) HFWR3_ATL_0005807: RYSPSFQGQVSISVDKSISTAFLQWSSLKSSDSAMYYC (SEQ ID NO:92) HFWR3_ATL_0005808; HFWR3_ATL_0006040; HFWR3_ATL_0006041 ; HFWR3_ATL_0006042; HFWR3_ATL_0006043; HFWR3_ATL_0006044; HFWR3_ATL_0006047; HFWR3_ATL_0006048: RYSPSFQGQVTISADKSISTAYLQWSSLKASDTAMYYC (SEQ ID NO:93)

HFWR3_ATL_0005809: RYSPSFQSQVTISADKSIDTAYLEWNTLEASDTAMYYC (SEQ ID NO: 94) HFWR3_ATL_0005810: SYAQKFQGRVTMTRDTSTSTVYMELSSLRSEDTAVYYC (SEQ ID NO: 95) HFWR3_ATL_0005853: RYSPSFEGQVTISADKSSGIVYLQWTSLKASDTAIYYC (SEQ ID NO: 96) HFWR3_ATL_0005854: RYSPSFQGQVTISADKSISTAYLQWSSLKASDTAIYYC (SEQ ID NO: 97) or framework sequences with one to six, 1 to 5, 1 to 4, 1 to 3, or 1 to 2 substitutions compared to the framework sequences above.

An antibody of the disclosure may have a heavy chain variable domain (VH) with the framework sequences HFWR1 , HFWR2, HFWR3 and HFWR4 of antibody ATL_0005802, or framework sequences with one to six substitutions compared to these framework sequences. An antibody of the disclosure may have a heavy chain variable domain (VH) with the framework sequences HFWR1 , HFWR2, HFWR3 and HFWR4 of antibody ATL_0005853, or framework sequences with one to six substitutions compared to these framework sequences. An antibody of the disclosure may have a heavy chain variable domain (VH) with the framework sequences HFWR1 , HFWR2, HFWR3 and HFWR4 of antibody ATL_0005854, or framework sequences with one to six substitutions compared to these framework sequences. An antibody of the disclosure may have a heavy chain variable domain (VH) with the framework sequences HFWR1 , HFWR2, HFWR3 and HFWR4 of antibody ATL_0005807, or framework sequences with one to six substitutions compared to these framework sequences. An antibody of the disclosure may have a heavy chain variable domain (VH) with the framework sequences HFWR1 , HFWR2, HFWR3 and HFWR4 of antibody ATL_0005808, or framework sequences with one to six substitutions compared to these framework sequences. An antibody of the disclosure may have a heavy chain variable domain (VH) with the framework sequences HFWR1 , HFWR2, HFWR3 and HFWR4 of antibody ATL_0005809, or framework sequences with one to six substitutions compared to these framework sequences. An antibody of the disclosure may have a heavy chain variable domain (VH) with the framework sequences HFWR1 , HFWR2, HFWR3 and HFWR4 of antibody ATL_0005810, or framework sequences with one to six substitutions compared to these framework sequences.

In embodiments, substitutions in the framework sequences of the heavy chain variable domain may be located at any position other than position 40 in standard IMGT numbering. In embodiments, the substitutions in the framework sequences do not include a substitution at position 40 in standard IMGT numbering. Substitutions in the heavy chain variable domain framework sequences may be located at any position in HFWR1 and/or HFRW3. Substitutions in the heavy chain variable domain framework sequences may be located at any position that is not in HFRW2. Substitutions in the heavy chain variable domain framework sequences may be selected from the following positions in standard IMGT numbering: HFWR1 : positions 1 , 6, 20; HFWR3: positions 72, 85. Substitutions in the heavy chain variable domain framework sequences may be selected from the following positions in standard IMGT numbering: HFWR1 : at position 1 : Q1 E, at position 6: E6Q, at position 20: R20K; HFWR3: at position 72: E72Q, at position 85: G85S.

An antibody of the disclosure may have a heavy chain variable domain (VH) with a framework sequence HFWR2 of ATL_0005802 (HFWR2_ATL_0005802; HFWR2_ATL_0005853; HFWR2_ATL_0005854). The isolated antibody may have a heavy chain variable domain (VH) with a framework sequence HFWR2 comprising the sequence: IAWVRQMPGKGLEWMGI (SEQ ID NO: 83). An antibody of the disclosure may have a heavy chain variable domain (VH) with a framework sequence HFWR2 of ATL_0005802. An antibody of the disclosure may have a heavy chain variable domain (VH) with a framework sequence comprising a A at position 40 in standard IMGT numbering.

An antibody according to the disclosure may have a heavy chain variable domain (VH) with the following framework sequences:

HFWR1 of SEQ ID NOs: 70, 71 , 72, 73, 74, 235, 75, 77, 78 or 79 (HFWR1_ATL_0005807; HFWR1_ATL_0006041 ; HFWR1_ATL_0005808; HFWR1_ATL_0005809; HFWR1_ATL_0005810; HFWR1_ATL_0005853; HFWR1_ATL_0005854; HFWR1_ATL_0006040; HFWR1_ATL_0006044; HFWR1_ATL_0006045; HFWR1_ATL_0006046;HFWR1_ATL_0006047; HFWR1_ATL_0006048;

HFWR1_ATL_0006042; HFWR1_ATL_0006043), preferably SEQ ID NOs: 70, 235, 75;

HFWR2 of SEQ ID NOs; 83, 84, 85, 86 (HFWR2_ATL_0005802; HFWR2_ATL_0005853; HFWR2_ATL_0005854; HFWR2_ATL_0006044; HFWR2_ATL_0005807; HFWR2_ATL_0005808; HFWR2_ATL_0006040; HFWR2_ATL_0006041 ; HFWR2_ATL_0006042; HFWR2_ATL_0006043; HFWR2_ATL_0006045; HFWR2_ATL_0006046; HFWR2_ATL_0006047; HFWR2_ATL_0006048;

HFWR2_ATL_0005809; HFWR2_ATL_0005810), preferably SEQ ID NO:83;

HFWR3 of SEQ ID NOs: 91 , 92, 93, 94, 95, 96, 97 (HFWR3_ATL_0005802 HFWR3_ATL_0005807;

HFWR3_ATL_0005808; HFWR3_ATL_0006040; HFWR3_ATL_0006041 ; HFWR3_ATL_0006042;

HFWR3_ATL_0006043; HFWR3_ATL_0006044; HFWR3_ATL_0006047;

HFWR3_ATL_0006048; HFWR3_ATL_0005809; HFWR3_ATL_0005810 HFWR3_ATL_0005853;

HFWR3_ATL_0005854), preferably SEQ ID NOs: 91 , 96, 97, and

HFWR4 of SEQ ID NOs: 101 , 104, 105 (HFWR4_ATL_0004828; HFWR4_ATL_0005802;

HFWR4_ATL_0005853; HFWR4_ATL_0005854; HFWR4_ATL_0006040; HFWR4_ATL_0006041 ;

HFWR4_ATL_0006042; HFWR4_ATL_0006043; HFWR4_ATL_0006044; HFWR4_ATL_0006045;

HFWR4_ATL_0006046; HFWR4_ATL_0006047; HFWR4_ATL_0006048; HFWR4_ATL_0005807;

HFWR4_ATL_0005808; HFWR4_ATL_0005809; HFWR4_ATL_0005810), preferably SEQ ID NO: 101.

An antibody of the disclosure may have the CDRL1 , CDRL2 and CDRL3 of the VL domain within a germline framework. An antibody of the disclosure may have a heavy chain variable domain (VH) comprising CDRH1 , CDRH2, and CDRH3 within a germline framework, provided that position 40 in standard IMGT numbering is A.

An antibody according to the disclosure may have a light chain variable domain (VL) with the following framework sequences: a LFWR1 comprising an amino acid sequence selected from:

LFWR1_ATL_0005802: QAVLTQPSSVSGAPGQRVTISCTGS (SEQ ID NO:138)

LFWR1_ATL_0005807: QSVLTQPPSVSVSPGQTARITCSGD (SEQ ID NO:139)

LFWR1_ATL_0005808; LFWR1_ATL_0005854: QSALTQPASVSGSPGQSITISCTGT (SEQ ID NQ:140)

LFWR1_ATL_0005809: SSELTQDPAVSVAVGQTVRITCQGD (SEQ ID NO:141)

LFWR1_ATL_0005810: QTVVTQEPSFSVSPGGTVTLTCGLS (SEQ ID NO:142) LFWR1_ATL_0005853: DVVMTQSPLSLPVNPGEPASISCRSS (SEQ ID NO:143) LFWR1_ATL_0006040; LFWR1_ATL_0006041 ; LFWR1_ATL_0006042; LFWR1_ATL_0006043;

LFWR1_ATL_0006044; LFWR1_ATL_0006045; LFWR1_ATL_0006046: QSVLTQPPSVSGAPGQRVTISCTGS (SEQ ID NO: 145)

LFWR1_ATL_0006047 QAVLTQPPSVSGAPGQRVTISCTGS (SEQ ID NO:146)

LFWR1_ATL_0006048 QSVLTQPSSVSGAPGQRVTISCTGS (SEQ ID NO:147) a LFWR2 comprising an amino acid sequence selected from:

LFWR2_ATL_0005802: VHWYQQLPGTAPKLLIY (SEQ ID NO:151)

LFWR2_ATL_0005807: AFWYQQKPGQAPVLVIY (SEQ ID NO: 152)

LFWR2_ATL_0005808; LFWR2_ATL_0005854: VSWYQQHPGKAPKLMIY (SEQ ID NO:153)

LFWR2_ATL_0005809: ANWYQQKPGQAPVLVIY (SEQ ID NO:154)

LFWR2_ATL_0005810: PSWYQQTPGQPPRTLIY (SEQ ID NO:155)

LFWR2_ATL_0005853: LHWYLQKPGQSPQLLIY (SEQ ID NO:156) a LFWR3 comprising an amino acid sequence selected from:

LFWR3_ATL_0005802: NRPSGVPDRFSGSKSGTSASLAITGLQAEDEADYYC (SEQ ID NO:160) LFWR3_ATL_0005807: ERPSGIPERFSGSSSGTTVTLTISGVQAEDEADYYC (SEQ ID NO:161) LFWR3_ATL_0005808: NRPSGVSSRFSASKSGNTASLTISGLQAEDEADYYC (SEQ ID NO:162) LFWR3_ATL_0005809: NRPSGIPDRFSGSSSGNTASLTLTGAQAEDEADYYC (SEQ ID NO:163) LFWR3_ATL_0005810: TRSSGVPDRFSGSILGNKAALTITGAQADDESDYYC (SEQ ID NO:164) LFWR3_ATL_0005853: DRAPGVPDRFSGSGSGTDFTLKINRVEAEDVGVYYC (SEQ ID NO:165) LFWR3_ATL_0005854: SRPSGVSYRFSGSKSGNTASLTISGLQAEDEADYYC (SEQ ID NO:166); and a LFWR4 comprising an amino acid sequence selected from:

LFWR4_ATL_0005802: FGGGTKLTVL (SEQ ID NO:172)

LFWR4_ATL_0005807; LFWR4_ATL_0005808; LFWR4_ATL_0005854: FGGGTKVTVL (SEQ ID NO: 173)

LFWR4_ATL_0005853: FGQGTRLEIK (SEQ ID NO:174) or framework sequences with one to six, 1 to 5, 1 to 4, 1 to 3, or 1 to 2 substitutions compared to the framework sequences above.

An antibody of the disclosure may have a light chain variable domain (VL) with the framework sequences LFWR1 , LFWR2, LFWR3 and LFWR4 of antibody ATL_0005802, or framework sequences with one to six substitutions compared to these framework sequences. An antibody of the disclosure may have a light chain variable domain (VL) with the framework sequences LFWR1 , LFWR2, LFWR3 and LFWR4 of antibody ATL_0005853, or framework sequences with one to six substitutions compared to these framework sequences. An antibody of the disclosure may have a light chain variable domain (VL) with the framework sequences LFWR1 , LFWR2, LFWR3 and LFWR4of antibody ATL_0005854, or framework sequences with one to six substitutions compared to these framework sequences. An antibody of the disclosure may have a light chain variable domain (VL) with the framework sequences LFWR1 , LFWR2, LFWR3 and LFWR4 of antibody ATL_0005807, or framework sequences with one to six substitutions compared to these framework sequences. An antibody of the disclosure may have a light chain variable domain (VL) with the framework sequences LFWR1 , LFWR2, LFWR3 and LFWR4of antibody ATL_0005808, or framework sequences with one to six substitutions compared to these framework sequences. An antibody of the disclosure may have a light chain variable domain (VL) with the framework sequences LFWR1 , LFWR2, LFWR3 and LFWR4of antibody ATL_0005809, or framework sequences with one to six substitutions compared to these framework sequences. An antibody of the disclosure may have a light chain variable domain (VL) with the framework sequences LFWR1 , LFWR2, LFWR3 and LFWR4of antibody ATL_0005810, or framework sequences with one to six substitutions compared to these framework sequences.

Substitutions in the light chain variable domain framework sequences may be selected from the following positions in standard IMGT numbering: LFWR1 : positions 2, 8. Substitutions in the light chain variable domain framework sequences may be selected from the following positions in standard IMGT numbering: LFWR1 : at position 2: A2S, at position 8: S8P.

The antibody may have a light chain variable domain (VL) with the following framework sequences: LFWR1 of ATL_0005802, ATL_0005853 or ATL_0005854; LFWR2 of ATL_0005802, ATL_0005853 or ATL_0005854; LFWR3 of ATL_0005802, ATL_0005853 or ATL_0005854; and LFWR4 of ATL_0005802, ATL_0005853 or ATL_0005854.

In this specification, antibodies may have VH (and optionally VL) regions comprising an amino acid sequence that has a high percentage sequence identity to the VH and/or VL amino acid sequences described above. For example, antibodies according to the present invention include antibodies that bind CD33 and have a VH regions that comprises an amino acid sequence having at least 70%, more preferably one of at least 75%, 80%, 85%, 90%, 95%, 96%, 97%, 98%, 99% or 100% sequence identity to the VH region amino acid sequence of any antibody described herein. An antibody according to the disclosure may have a heavy chain variable domain (VH) comprising a sequence that has a least 95% sequence identity with a sequence selected from SEQ ID NOs:4 to 10 and 12 to 20. An antibody according to the disclosure may have a heavy chain variable domain (VH) comprising a sequence that has at most 2 mutations in each HCDR and at most 3 mutations in each framework region compared with a sequence selected from SEQ ID NOs:4 to 10 and 12 to 20.

An antibody according to the disclosure may have a heavy chain variable domain (VH) comprising a sequence that has a least 95% sequence identity with a sequence selected from SEQ ID NOs:4, 9 and 10 (antibodies ATL_0005802, ATL_0005853, ATL_0005854) An antibody according to the disclosure may have a heavy chain variable domain (VH) comprising a sequence that has at most 2 mutations in each HCDR and at most 3 mutations in each framework region compared with a sequence selected from SEQ ID NOs:4, 9 and 10 (antibodies ATL_0005802, ATL_0005853, ATL_0005854).

Alternatively or additionally, antibodies of the disclosure may have a VL region that comprises an amino acid sequence having at least 70%, more preferably one of at least 75%, 80%, 85%, 90%, 95%, 96%, 97%, 98%, 99% or 100% sequence identity to the VL region amino acid sequence of any antibody described herein. For example, antibodies of the disclosure may have a VL region that comprises an amino acid sequence having at least 70%, more preferably one of at least 75%, 80%, 85%, 90%, 95%, 96%, 97%, 98%, 99% or 100% sequence identity to the VL region amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809,

ATL_0005810

An antibody according to the disclosure may have a light chain variable domain (VL) comprising a sequence that has a least 95% sequence identity with a sequence selected from SEQ ID NOs:24 to 30 and 32 to 40. An antibody according to the disclosure may have a light chain variable domain (VL) comprising a sequence that has at most 2 mutations in each LCDR and at most 3 mutations in each framework region compared with a sequence selected from SEQ ID NOs: 24 to 30 and 32 to 40.

An antibody according to the disclosure may have a light chain variable domain (VL) comprising a sequence that has a least 95% sequence identity with a sequence selected from SEQ ID NOs: 24, 29, 30 (antibodies ATL_0005802, ATL_0005853, ATL_0005854) An antibody according to the disclosure may have a light chain variable domain (VL) comprising a sequence that has at most 2 mutations in each LCDR and at most 3 mutations in each framework region compared with a sequence selected from SEQ ID NOs:24, 29, 30 (antibodies ATL_0005802, ATL_0005853, ATL_0005854).

Antibodies of the disclosure may have a lambda (A) or kappa (K) light chain.

Overall percentage identity of a variable region or full-length heavy light/chain sequence may be combined with specific CDR sequences from the same antibody.

Percentage (%) sequence identity is defined as the percentage of amino acid residues in a candidate sequence that are identical with residues in a comparative sequence after aligning the sequences and introducing gaps if necessary, to achieve the maximum sequence identity, and not considering any conservative substitutions as part of the sequence identity. Sequence identity is preferably calculated over the entire length of the respective sequences. Where the aligned sequences are of different length, sequence identity of the shorter comparison sequence may be determined over the entire length of the longer given sequence or, where the comparison sequence is longer than the given sequence, sequence identity of the comparison sequence may be determined over the entire length of the shorter given sequence. Sequence identity may be defined with reference to the algorithm GAP (Wisconsin GCG package, Accelerys Inc, San Diego USA). GAP uses the Needleman and Wunsch algorithm to align two complete sequences that maximizes the number of matches and minimizes the number of gaps. Generally, default parameters may be used, with a gap creation penalty = 12 and gap extension penalty = 4. Use of GAP may be preferred but other algorithms may be used, e.g. BLAST (which uses the method of Altschul et al. (1990) J. Mol. Biol. 215: 405-410), FASTA (which uses the method of Pearson and Lipman (1988) PNAS USA 85: 2444-2448), SSEARCH (Smith and Waterman (1981) J. Mol Biol. 147: 195-197; ), HMMER3 (Johnson LS et al BMC Bioinformatics. 2010 Aug 18; 11 :431) or the TBLASTN program, of Altschul et al. (1990) supra, generally employing default parameters (see for example Pearson Curr Prot Bioinformatics (2013) Chapt 3 Uniy 3.1 doi:10.1002/0471250953. bi0301s42). In particular, the psi-Blast algorithm may be used (Altschul et al. Nucl. Acids Res. (1997) 25 3389-3402). Sequence identity and similarity may also be determined using GenomequestTM software (Gene-IT, Worcester MA USA). Sequence comparisons are preferably made over the full-length of the relevant sequences to be compared.

The antibodies of the present disclosure may comprise one or more substitutions within the framework of the VH and/or VL region. As used herein, “substitution” refers to the exchange of one amino acid for another at a specific position, relative to the same position in a baseline molecule. In some embodiments, the baseline molecules are exemplified antibodies herein, for example antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810.

In some embodiments, antibodies or fragments thereof according to the present disclosure are able to cross the blood-brain barrier. In some embodiments, the antibodies of the present disclosure may have one scFV chain, such as the transferrin receptor (Yu et al., 2014), as well as a scFv chain that binds CD33 as described herein. In some embodiments, antibodies according to the disclosure comprise an antibody or fragment thereof that binds CD33 as described herein (such as e.g. an antibody, scFv, sdAb, etc.), and further binding moiety that binds to another target. The other target may be a receptor in the brain, such as the transferrin receptor. The further binding moiety may be an antibody, a scFv, a nanobody, or an aptamer.

The two binding moieties of such a bispecific molecule may form a fusion protein.

Also described herein are single domain antibodies (sdAbs), otherwise known as nanobodies, comprising the heavy chain CDRs and/or the VH sequence of any antibody described herein. Thus, also described herein are antibodies or fusion molecules comprising a nanobody that binds CD33 as described herein, and a nanobody that binds a receptor in the brain. Also described herein are antibodies of fusion molecules comprising a scFV chain or nanobody that binds CD33 as described herein, and an aptamer that binds a receptor in the brain.

Antibodies described herein may increase phagocytosis of a cell expressing CD33. Increase in phagocytosis may be measured compared to a comparative antibody. A comparative antibody may be an isotype control antibody. A comparative antibody may be another CD33-binding antibody. A comparative antibody may be an antibody with the heavy chain variable sequence of ATL_5909 and the light chain variable sequence of ATL_5909. A comparative antibody may be another CD33-binding antibody with a different epitope. Phagocytosis may be assessed by measuring a fluorescence signal associated with uptake of a labelled particle by imaging or flow cytometry. Phagocytosis may be measured by detecting a fluorescence signal indicative of phagocytosis of a substrate that has pH dependent fluorescence. The substrate may be selected from amyloid beta and S. aureus. The labelled substrate may be a substrate labelled with pHrodo red. The cell expressing CD33 may be an inflammatory human iPSC-derived microglial cell. An inflammatory cell may be a cell that has been stimulated with an inflammatory signal. For example, the cell may have been exposed to an inflammatory signal (e.g. LPS, IFN-y) for at least 6, 12 or 24 hours, or about 6, 12 or 24 hours prior to exposure to the antibody. The cell expressing CD33 may be a PBMC. The cell expressing CD33 may be a cell isolated from human blood. The cell expressing CD33 may be a monocyte or a microglial cell. The cell expressing CD33 may be a human cell. The cell expressing CD33 may be iPSC derived microglial cell. The cell expressing CD33 may be a cell that has been stimulated with an inflammatory signal (e.g. LPS) prior to exposure to the antibody.

Antibodies described herein may bind human CD33. Antibodies described herein may bind human CD33 with an EC50 of at most 2e-08 M, or at most 3e-09 M, as assessed by ELISA (such as binding of plated rhCD33). An antibody of the disclosure may bind to human CD33 with an EC50 of at most 15 pg/ml, at most 12 pg/ml, at most 5 pg/ml, at most 3 pg/ml or at most 0.5 pg/ml. An antibody of the disclosure may bind to human CD33 with an EC50 of at most 1 e-7M, at most 8e-8 M, at most 3e-8 M, at most 2e-08 M or at most 3e-09 M. Binding to human CD33 may be assessed by ELISA using plated rhCD33, such as e.g. CD33_HUMAN_ECD, His_004_001 from Sino Biological, 12238-H08H.

An antibody according to the disclosure may have a Koff for binding to human CD33 of at least 1 e-3 s ¹. An antibody according to the disclosure may have a Koff for binding to human CD33 of between 1 e-3 s ¹ and 5e-2 S'¹ . An antibody according to the disclosure may have a KD of binding to human CD33 of at most 5e- 7 M. An antibody according to the disclosure may have a KD of binding to human CD33 of between 2e-8 M and 5e-7 M. A human CD33 may be a rhCD33-his (e.g. R&D Systems, 10375-SL-050). KD and Koff may be as measured by biolayer interferometry (BLI) on the octet-l. An antibody according to the disclosure may have a Koff for binding to human CD33 that is lower than that of a comparative antibody. Suitable comparative antibodies are as described above.

An antibody according to the disclosure may have a melting temperature Tm1 of at least 57, or between 57 and 70. An antibody according to the present disclosure may have a purity SEC-HPLC % monomer of at least 95%.

Antibodies described herein may result in lower internalisation of CD33 on C33-expressing cells compared to a comparative antibody. An antibody described herein may deplete CD33 on the cell surface of human monocytes by less than 50%, or less than 80% after 5 hours of incubation with the antibody. An antibody of the disclosure may deplete CD33 on the cell surface of human monocytes after 5 hours of incubation with the antibody to a lower extent than a comparative antibody at the same concentration. Suitable comparative antibodies are as described above.

Antibodies described herein may have an increased binding to a human CD33 protein comprising mutations at positions 20, 21 , 22 and 24 compared to the antibody binding to a human CD33 protein without said mutations. Antibodies described herein may have an increased binding to a human CD33 protein comprising mutations at positions 20, 21 , 22, 24 and 132 compared to the antibody binding to a human CD33 protein without said mutations. Antibodies described herein may have a reduced binding to a human CD33 protein comprising mutations at positions 47, 50, 51 and 52 compared to the antibody binding to a human CD33 protein without said mutations. Antibodies described herein may not bind to a human CD33 protein comprising mutations at positions 47, 50, 51 and 52. Antibodies described herein may have reduced binding to a human CD33 protein comprising mutations at positions 47, 50, 51 , 52 and 122 compared to the antibody binding to a human CD33 protein without said mutations. Antibodies described herein may not bind to a human CD33 protein comprising mutations at positions 47, 50, 51 , 52 and 122. Antibodies described herein may have reduced binding to a human CD33 protein comprising mutations at positions 83 is reduced compared to the antibody binding to a human CD33 protein without said mutations. The mutations may be selected from: at position 20: N20R, at position 21 : F21V, at position 22: W22R, at position 24: Q24E, at position 47: 147V, at position 50: Y50H, at position 51 : D51T, at position 52: K52R, at position 83: Q83R, at position 122: R122K, and at position 132: P132T. The positions may refer to the position in the full length human CD33 sequence (CD33M (Uniprot ID: P20138-1)). Binding may be as measured using a single point ELISA. The human CD33 protein may comprise residues 18-232 of human CD33. The human CD33 protein without said mutations may be CD33M2_ECD_18-232_WT (SEQ ID NO: 226). An antibody as described herein may have increased binding to a protein comprising the sequence of CD33M2_ECD_18-232_MutPos1 [P1] (SEQ ID NO: 227) compared to a protein comprising the sequence of CD33M2_ECD_18-232_WT (SEQ ID NO: 226). An antibody as described herein may have increased binding to a protein comprising the sequence of CD33M2_ECD_18-232_MutPos1_MutPos6 [P6+1] (SEQ ID NO:228) compared to a protein comprising the sequence of CD33M2_ECD_18-232_WT (SEQ ID NO: 226). An antibody as described herein may have reduced binding to a protein comprising the sequence of CD33M2_ECD_18-232_MutPos2 [P2] (SEQ ID NO:229) compared to a protein comprising the sequence of CD33M2_ECD_18-232_WT (SEQ ID NO: 226). An antibody as described herein may have reduced binding to a protein comprising the sequence of CD33M2_ECD_18-232_MutPos2_MutPos5 [P2+5] (SEQ ID NO:230)compared to a protein comprising the sequence of CD33M2_ECD_18-232_WT (SEQ ID NO: 226). An antibody as described herein may have reduced binding to a protein comprising the sequence of CD33M2_ECD_18-232_MutPos4 [P4] (SEQ ID NO:232) compared to a protein comprising the sequence of CD33M2_ECD_18-232_WT (SEQ ID NO: 226). Antibodies described herein may interfere with binding of sialic acid. Antibodies described herein may bind to an epitope such that the binding of the antibody interferes with binding of sialic acid to CD33.

An antibody of the disclosure may bind to a CD33 protein that comprises the V domain of CD33. An antibody of the disclosure may not bind to a CD33 protein that does not comprise the V domain of CD33. An antibody of the disclosure may not bind to a CD33 protein that has the sequence of protein CD33_HUMAN_ECD_Cdomain_His_007 (SEQ ID NO: 175). An antibody of the disclosure may selectively bind to CD33 over other one or more siglecs, optionally wherein the antibody selectively binds to CD33 over one or more (or all of): siglec-6, siglec-7, siglec-8, and siglec-9, and/or wherein the antibody selectively binds to human CD33 over other one or more homologs, optionally wherein the antibody selectively binds to human CD33 over mouse CD33 and cyno CD33.

An antibody of the disclosure may reduce neuroinflammation (i.e. inflammation in the brain), for example compared with a comparative antibody (e.g. an isotype control antibody, or ATL_5909). The level of neuroinflammation may be determined by measuring the levels of pro-inflammatory cytokines in neural cells, such as microglia (e.g. human Ipsc-derived microglia) as described herein (e.g. using an immunoassay, such as an ELISA). The pro-inflammatory cytokines may be one or more of: monocyte chemoattractant protein-1 (MCP-1), IL-6, Interferon gamma-induced protein 10 (IP10), Glial fibrillary acidic protein (GFAP).

An antibody described herein can have lower peripheral clearance when administered to a subject, compared to a comparative anti-CD33 antibody. The comparative anti-CD33 antibody can be ATL_5909. The subject can be a subject (e.g. a mouse) expressing a CD33 protein comprising human CD33 exons 1- 3. Administration can be parenteral. Administration can be intravenous or intraperitoneal.

An antibody described herein can increase phagocytosis of Ap by microglial cells in vivo. This can be assessed in vivo, for example as described herein (see Example 10 and accompanying methods). Effect of the antibody on phagocytosis of Ap can be assessed on microglia from iPSC-derived human microglia progenitors transplanted into the brains of Alzheimer’s disease (AD) mice (e.g. APP^{NL G F} knock-in mice).

An antibody described herein can increase clearance of Ap plaques in a subject in need thereof. The antibody can increase clearance of Ap plaques in a subject in need thereof through microglial phagocytosis.

An antibody described herein can increase the phagocytosis of tau aggregates by microglia with inflammatory phenotype (e.g. LPS treated iPSC microglia). An antibody described herein can increase the phagocytosis of tau aggregates by microglia with inflammatory phenotype to a larger extent than a comparative anti-CD33 antibody, such as e.g. ATL_5909. This can be assessed in vitro as described herein, such as e.g. in Example 11. An antibody described herein can increase the phagocytosis of tau aggregates by microglia in a subject in need thereof.

An antibody described herein may not induce significant release of any one or more cytokines by human iPSC derived microglia, human PBMCs and/or human isolated monocytes exposed to the antibody in vitro, compared to a control (e.g. isotype control antibody). The one or more cytokines can be selected from: CCL2 (MCP-1); CXCL8 (IL-8); IFN-y; IL-10; |31 ; TNF-a ; CCL17 (TARC) ; CCL2 (MIP-3a (Gro-a); CXCL5 (ENA-78); CXCL9 (MIG); CCL1 1 (Eotaxin); CX3CL1 (Fractalkine); IL-18; sRAGE; sTREM- 1 ; sTREM-2; VEGF; VILIP-1 ; p-NGF; BDNF; CCL2 (MCP-1). The antibody may not induce release of any of the above cytokines. The antibody may induce lower levels of release of any one or more or all of the above cytokines by human iPSC derived microglia, human PBMCs and/or human isolated monocytes exposed to the antibody in vitro, compared to a comparative anti-CD33 antibody (e.g. ATL_5909). An antibody as described herein may reduce the levels of IL-6 and/or MCP-1 released by microglia with an inflammatory phenotype in vitro (e.g. LPS treated human iPSC derived microglia) and/or in vivo (e.g. when administered to a subject in need thereof). An antibody as described herein may reduce the levels of IL-6 and/or MCP-1 released by microglia with an inflammatory phenotype in vitro (e.g. LPS treated human iPSC derived microglia) more than a comparative anti-CD33 antibody (E.g. ATL_5909).

An antibody described herein may increase the levels of phosphorylated SYK protein in microglial cells treated with the antibody in vitro compared to a control (e.g. isotype control antibody). The microglial cells can be microglial cells with an inflammatory phenotypes (e.g. microglial cells treated with LPS). SYK phosphorylation may increase without significant increase in overall protein levels.

An antibody described herein may increase the levels of P2RY12 in microglial cells treated with the antibody in vitro compared to a control (e.g. isotype control antibody). The microglial cells can be microglial cells with an inflammatory phenotypes (e.g. microglial cells treated with LPS).

An antibody described herein may enhance TREM2 signalling in microglial cells treated with the antibody in vitro compared to a control (e.g. isotype control antibody). The microglial cells can be microglial cells with an inflammatory phenotypes (e.g. microglial cells treated with LPS).

An antibody described herein may alter the activity of the oxidative phosphorylation pathway in microglia, compared to a control (e.g. treatment with an isotype control antibody). Activity of the oxidative phosphorylation pathway can be assessed by RNA sequencing optionally followed by gene set enrichment analysis, e.g. as described herein (see Example 14).

An antibody described herein may reduce inflammation induced release of one or more markers of inflammation in a human neural cell culture assay, compared to a control (e.g. exposure to an isotype control antibody). The human neural cell culture assay can be a co-culture assay comprising glutamatergic and GABAergic neurons, microglia and astrocytes. Inflammation induced release of one or more cytokines can be assessed by exposing cells to one or more pro-inflammatory signals, such as e.g. lipopolysaccharide (LPS) and/or Interferon-gamma (INFy). The one or more markers of inflammation can be selected from: MCP-1 , IP-10, GFAP and IL-6. Thus, antibodies ofthe disclosure can reduce the expression of one or more markers of inflammation in the CNS of a subject in need thereof. The one or more markers of inflammation can include one or more of MCP-1 , IP-10, GFAP and IL-6.

An antibody described herein may bind selectively to CD33. The antibody may not bind to any other human protein. Isolated nucleic acids encoding an antibody, antigen binding fragment, or polypeptide as described herein are provided. Also provided is a vector comprising a nucleic acid described herein, and a host cell comprising the vector. For example, the host cell may be a eukaryotic, or mammalian, e.g. Chinese Hamster Ovary (CHO), cell or may be a prokaryotic cell, e.g. E. coli. In some embodiments, the vector is a viral vector, for example a bacteriophage.

Further provided are methods for making an antibody, or antibody fragment as described herein, the method comprising culturing a host cell as described herein under conditions suitable for the expression of a vector encoding the antibody, or antibody fragment, and isolating and/or purifying the antibody, or antibody fragment. The method further comprises formulating the antibody or antibody fragment into a composition including at least an additional component.

The antibodies and fragments thereof may find use in therapy.

A subject to be treated or diagnosed may be any animal or human. The subject is preferably mammalian, more preferably human. The subject may be male or female. The subject may be a patient. Therapeutic uses may be in human or animals (veterinary use). Unless, specified otherwise, the subject is a human.

Medicaments and pharmaceutical compositions according to aspects of the present invention may be formulated for administration by a number of routes, including but not limited to, parenteral, intravenous, intra-arterial, intramuscular, oral and nasal. The medicaments and compositions may be formulated for injection.

Pharmaceutical compositions may be prepared using a pharmaceutically acceptable “carrier” composed of materials that are considered safe and effective. "Pharmaceutically acceptable" refers to molecular entities and compositions that are "generally regarded as safe", e.g., that are physiologically tolerable and do not typically produce an allergic or similar untoward reaction, such as gastric upset and the like, when administered to a human. In some embodiments, this term refers to molecular entities and compositions approved by a regulatory agency of the US federal or a state government, as the GRAS list under section 5 204(s) and 409 of the Federal Food, Drug and Cosmetic Act, that is subject to premarket review and approval by the FDA or similar lists, the U.S. Pharmacopeia or another generally recognised pharmacopeia for use in animals, and more particularly in humans. The term “carrier” refers to diluents, binders, lubricants and disintegrants. Those with skill in the art are familiar with such pharmaceutical carriers and methods of compounding pharmaceutical compositions using such carriers. The pharmaceutical compositions provided herein may include one or more excipients, e.g., solvents, solubility enhancers, suspending agents, buffering agents, isotonicity agents, antioxidants or antimicrobial preservatives. When used, the excipients of the compositions will not adversely affect the stability, bioavailability, safety, and/or efficacy of the active ingredients, i.e. the anti-CFH antibodies used in the composition. Thus, the skilled person will appreciate that compositions are provided wherein there is no incompatibility between any of the components of the dosage form. Excipients may be selected from the group consisting of buffering agents, solubilizing agents, tonicity agents, chelating agents, antioxidants, antimicrobial agents, and preservatives.

Administration is preferably in a "therapeutically effective amount", this being sufficient to show benefit to the individual. The actual amount administered, and rate and time-course of administration, will depend on the nature and severity of the disease being treated. Prescription of treatment, e.g. decisions on dosage etc., is within the responsibility of general practitioners and other medical doctors, and typically takes account of the disorder to be treated, the condition of the individual patient, the site of delivery, the method of administration and other factors known to practitioners. Examples of the techniques and protocols mentioned above can be found in Remington’s Pharmaceutical Sciences, 20^th Edition, 2000, pub. Lippincott, Williams & Wilkins.

Conditions treatable in accordance with the present invention include any in which CD33 plays a role, including neurodegenerative disorders, and in particular those characterised by increased toxic protein species, such as Ap42 characteristic of Alzheimer’s Disease (AD), where phagocytosis of these toxic species is useful. The antibodies of the present disclosure may be used in the treatment of any neurological disease or disorder associated with the accumulation of toxic protein species. Indeed, the antibodies of the present disclosure have been shown to increase phagocytosis in microglia and are therefore likely to be useful in the treatment of any neurological disease or disorder where phagocytosis of toxic protein species is impaired or insufficient. In addition, respiratory disorders characterised by defective phagocytosis, including chronic obstructive pulmonary disease (COPD) and idiopathic pulmonary fibrosis (IPD) (Donelly et al., 2012) may also be treated using antibodies of the present disclosure. Other conditions treatable in accordance with the present invention include cancers such as AML and solid tumours where sialic acid may be causing immune suppression, such as cancers where tumours show hypersialylation of their cell surface so as to avoid immune cell recognition (Stanczak & Laubli, 2023), including many solid tumours, as well as cancers characterised by aberrant CD33 expression such as acute myeloid leukaemia (AML). The antibodies of the present disclosure were originally identified through analysis of resilient AD individuals. These antibodies were subsequently found in resilient centenarians, indicating their relevance beyond AD. Thus, also described herein are antibodies of the disclosure for use as a medicament. Also described are antibodies for use in the treatment or prevention of a neurodegenerative disease, and/or respiratory disease, and/or cancer. Also described herein are antibodies of the disclosure for use in the manufacture of a medicament, such as a medicament for the treatment or prevention of a neurodegenerative disease, or respiratory disease, or cancer. Also described herein a method of treating a subject who has been diagnosed as having or being at risk of having a neurodegenerative disease, or respiratory disease, or cancer, the method comprising administering an antibody as described herein in a therapeutically effective amount to the subject.

A neurodegenerative disease or disorder can be selected from: frontotemporal dementia (FTD), Alzheimer’s disease (AD), Huntington’s Disease (HD), Parkinson’s disease (PD), amyotrophic lateral sclerosis (ALS), human immunodeficiency virus (HIV)-induced encephalitis, Chronic traumatic encephalopathy (CTE), vascular dementia, prion diseases, Lewy body disease, Spinal muscular atrophy (SMA), Motor Neuron Disease (MND), such as amyotrophic lateral sclerosis (ALS), progressive supranuclear palsy (PSP), spinocerebellar ataxias (SCA) types 1 , 2, 6, 7 and 17, Machado-Joseph disease (MJD/SCA3), dentatorubral pallidoluysian atrophy (DRPLA), spinal bulbar muscular atrophy X-linked type 1 (SMAX1/SBMA), Anderson-Fabry (X-linked Fabry Disease), and DNAJB6 myopathies, multiple sclerosis (MS - a single nucleotide polymorphism in CD33 is known to be associated with an increased risk of multiple sclerosis), and microgliopathies, such as adult-onset leukoencephalopathy with axonal spheroids and pigmented glia (ALSP). For example, the neurodegenerative disease may be selected from FTD, AD, HD, and PD. The neurogenerative disease or disorder can be a tauopathy. Thus, the neurodegenerative disorder can be AD (Alzheimer’s disease), CTE (chronic traumatic encephalopathy), PiD (Pick's disease), PSP (Progressive supranuclear palsy), CBD (corticobasal degeneration), AGD (Argyrophilic grain disease).

A respiratory disease can comprise one or more of the following: COPD, cystic fibrosis, asthma, and idiopathic pulmonary fibrosis (IPD).

A cancer can be selected from: cervical cancer, breast cancer, brain cancer bladder cancer, colon adenocarcinoma, cervical cancer, fibrocarcinoma, head and neck cancer, hepatocellular carcinoma, kidney cancer, ovarian cancer, pancreatic cancer, prostate cancer, renal cell cancer, non-small cell lung cancer, non-hodgkin lymphoma, and haematological malignancies such as acute myeloid leukaemia (AML), acute lymphoblastic leukaemia (ALL), chronic lymphocytic leukaemia (CLL), and multiple myeloma. In some embodiments, the cancer is AML. In embodiments, the cancer is selected from: melanoma, hepatocellular carcinoma, pancreatic cancer, colon adenocarcinoma, cervical cancer, breast cancer, non-small cell lung cancer, head and neck cancer and hematological malignancies. All of these cancers are known to have tumour cells that hypersialylate their cell surfaces to suppress immune cells and the antibodies of the present disclosure could therefore be used to relieve this suppression.

The antibodies of the disclosure may be used in therapy with one or more further therapeutic agents. As used herein, a “further therapeutic agent” is an additional compound, protein, vector, antibody, cell or entity with a therapeutic effect. The antibodies described herein may be co-administered with a further therapeutic agent. The antibodies may be co-formulated with a further therapeutic agent. The antibodies may be sequentially administered, before or after a further therapeutic agent.

The antibodies described herein may be used as biomarkers indicating that a subject is likely to respond to therapy using an antibody or antibody fragment as described herein. Also described herein is a method of determining whether a subject is likely to respond to treatment with an antibody or fragment thereof as described herein, the method comprising obtaining BCR sequence data from the subject and determining, using said sequence data, whether the subject’s BCR repertoire comprises one or more antibodies that are likely to bind to CD33 (e.g. antibodies as described herein, such as e.g. antibodies having at least 70%, at least 80%, at least 85%, at least 90%, at least 95% or at least 99% homology to any specific antibody or antibody fragment described herein), wherein a subject whose BCR repertoire does not comprise one or more antibodies that are likely to bind to CD33 as described herein is likely to respond to treatment with an antibody or antibody fragment thereof as described herein. Thus, also described herein is a method of treating a subject who has been diagnosed as having or likely to have a disease in which CD33 plays a role (e.g. any disease where CD33 expressing cells have impaired or insufficient phagocytosis, such as e.g. a neurodegenerative disorder or cancer), the method comprising: obtaining BCR sequence data from the subject; determining, using said sequence data, whether the subject’s BCR repertoire comprises one or more antibodies that are likely to bind to CD33; and administering to a subject whose BCR repertoire does not comprise one or more antibodies that are likely to bind to CD33 a therapeutically effective amount of an antibody or antibody fragment thereof as described herein.

Some methods of the present disclosure involve a sample containing cells. The sample may be a culture of cells grown in vitro. For example, the culture may comprise a suspension of cells or cells cultured in a culture plate or dish. Methods according to the present disclosure may be performed, or products may be present, in vitro, ex vivo, or in vivo.

According to some aspects of the present disclosure a kit of parts is provided comprising an antibody according to the present invention. In some embodiments, the kit comprises an antibody according to the present invention and one or more of: reagents for use in immunochemistry; the antibodies immobilised to a solid support; means for labelling the antibodies; means for linking the antibodies to a cytotoxic moiety; and a further therapeutic agent.

***

The features disclosed in the foregoing description, or in the following claims, or in the accompanying drawings, expressed in their specific forms or in terms of a means for performing the disclosed function, or a method or process for obtaining the disclosed results, as appropriate, may, separately, or in any combination of such features, be utilised for realising the invention in diverse forms thereof.

While the invention has been described in conjunction with the exemplary embodiments described above, many equivalent modifications and variations will be apparent to those skilled in the art when given this disclosure. Accordingly, the exemplary embodiments of the invention set forth above are considered to be illustrative and not limiting. Various changes to the described embodiments may be made without departing from the spirit and scope of the invention.

For the avoidance of any doubt, any theoretical explanations provided herein are provided for the purposes of improving the understanding of a reader. The inventors do not wish to be bound by any of these theoretical explanations.

Any section headings used herein are for organizational purposes only and are not to be construed as limiting the subject matter described.

Throughout this specification, including the claims which follow, unless the context requires otherwise, the word “comprise” and “include”, and variations such as “comprises”, “comprising”, and “including” will be understood to imply the inclusion of a stated integer or step or group of integers or steps but not the exclusion of any other integer or step or group of integers or steps. It must be noted that, as used in the specification and the appended claims, the singular forms “a,” “an,” and “the” include plural referents unless the context clearly dictates otherwise. Ranges may be expressed herein as from “about” one particular value, and/or to “about” another particular value. When such a range is expressed, another embodiment includes from the one particular value and/or to the other particular value. Similarly, when values are expressed as approximations, by the use of the antecedent “about,” it will be understood that the particular value forms another embodiment. The term “about” in relation to a numerical value is optional and means for example +/- 10%.

Sequences

Table 0. Sequences used in this specification.

In Table 0 above and throughout this specification, antibodies may be referred to by their reference numbers as indicated in the table above (e.g. ATL_0005802), or the same reference number without the initial string of ‘0’ (e.g. ATL_5802), or the same reference with only the number (e.g. 5802). For example, antibody ATL_0005802 may be referred to interchangeably as “ATL_0005802”, “ATL_5802” and “5802”. The notations “VH” and “VL” when appended to an antibody reference (whether as a prefix or a suffix) refer to the heavy chain variable domain of the referenced antibody. The notations HCDR1 , HCDR2 and HCDR3, when appended to an antibody reference (whether as a prefix or a suffix) refer to the heavy chain CDRs (CDR1 , CDR2 and CDR3, respectively) of the referenced antibody. The notations LCDR1 , LCDR2 and LCDR3, when appended to an antibody reference (whether as a prefix or a suffix) refer to the light chain CDRs (CDR1 , CDR2 and CDR3, respectively) of the referenced antibody. The notations HFWR1 , HFWR2, HFWR3 and HFWR4, when appended to an antibody reference (whether as a prefix or a suffix) refer to the heavy chain framework regions (FWR1 , FWR2, FWR3 and FWR4, respectively) of the referenced antibody. The notations LFWR1 , LFWR2, LFWR3 and LFWR4, when appended to an antibody reference (whether as a prefix or a suffix) refer to the light chain framework regions (FWR1 , FWR2, FWR3 and FWR4, respectively) of the referenced antibody.

Examples

The examples below demonstrate the identification of new antibodies with therapeutic potential from a cohort of subjects that have or are considered at risk of developing Alzheimer’s disease, as well as cognitively healthy centenarians seropositive for CD33 antibodies (Example 1), as well as phage display analysis of repertoires from CD33-reactive subjects (Example 2). Examples 3 to 15 show results of functional screening and further characterisation of some of the newly discovered antibodies. Materials and methods

Serum and plasma ELISA to identify subjects with CD33 reactivity

CD33 recombinant human antigen (R&D Systems, cat: 10375-SL-050) or negative control Lysozyme (MP Biomedicals #195303) were directly absorbed to ELISA plates at 3pg/ml (50pl per well) and incubated overnight at 4°C. The plates were washed with phosphate-buffered saline (PBS). Plates were blocked with 200 pl/well of blocking solution (1 % Bovine Serum Albumin (BSA) w/v in PBS) for 1 hour at room temperature. Following this, the blocking solution was removed and the serum and plasma samples to be assessed were diluted 1/100 in blocking solution (1 % BSA w/v in PBS) and applied to the plates. Plates were incubated at room temperature for 1 hour. The plates were washed with PBS/0.1 % Tween. Next, Anti-human (Fab) ’2- horseradish peroxidase (HRP) antibody (Jackson Immunoresearch #109-035-097; Lot: 148466) was added to each well and incubated for 1 hour at room temperature to detect antibody binding. The plates were washed with PBS/0.1 % Tween and 3,3’, 5,5;-tetramethylbenzidine (TMB) solution (Life Technology; #002023) was added to each well. Plates were incubated for 5 minutes at room temperature prior to the addition of stopping solution (0.5M sulphuric acid). Absorbance read on Molecular Devices FilterMaxF5 plate reader at 450nm.

Phaqe ELISA and sequence analysis on scFv derived from phaqe display

Phages were prepared by culturing each phagemid-carrying TG1 clone from a glycerol stock in 10OpI of 2TYAG (2TY media supplemented with 10Opg/ml ampicillin and 2% glucose) at 37°C with aeration to optical density GD600=0.6 followed by rescue with helper phage added at MOI 10 (Invitrogen, cat: 18311-019) for 1 hour and media change to 2TYAK (2TY media supplemented with 100 pg/ml ampicillin and 50 mg/ml kanamycin). The cultures were then incubated overnight at 25°C with good aeration and next day the phages were separated form bacteria by centrifugation for 10 minutes at 3200rpm. The phage-containing supernatant was transferred to a new plate and blocked with 3% milk w/v in PBS.

CD33 recombinant human antigen (R&D Systems, cat: 10375-SL-050) or negative control Lysozyme (MP Biomedicals #195303) were directly absorbed to ELISA plate at 3 pg/ml (50 pl per well) and incubated overnight at 4°C. Each antigen-coated plate was washed with PBS and blocked with 200ul/well of blocking solution (3% milk w/v in PBS) for 1 hour at room temperature. Following this, the blocking solution was removed and blocked phage samples to be assessed were applied to the plates. Plates were incubated at room temperature for 1 hour. Each plate was washed with PBS/0.1 % Tween and incubated with Anti-M13 HRP (Sino Biological, #11973-MM05T-H) for 1 hour at room temperature to detect phage binding. The plates were washed with PBS/0.1 % Tween and TMB solution (Life Technology; #002023) added. Plates were incubated for 5 minutes at room temperature prior to the addition of stopping solution (0.5M sulphuric acid). Absorbance read on Molecular Devices FilterMaxF5 plate reader at 450nm.

Sinqle point ELISA forbindinq to full CD33 and CD33 C2 domain and to related CD33 proteins

Seven lgG1 converted antibodies were tested in a single point ELISA against a panel of antigens: human recombinant CD33 (Sino Biological cat: 12238-H08H), Cynomologus/ Rhesus CD33 (Sino Biological, cat: 90303-C08H), human recombinant CD33 C2 domain (in-house production), mouse CD33 (Sino Biological, cat: 50712-M08H), Recombinant human Siglec-6/CD327 Fc Chimera (R&D Systems, 2859-SL), Recombinant Human Siglec-7/CD328 Fc chimera (R&D Systems cat: 1138-SL-050), Recombinant Human Siglec-8 Fc Chimera (R&D Systems, cat: 9045-SL), Recombinant Human Siglec-9 Fc Chimera (R&D Systems, cat: 1 139-SL). After the single point ELISA, for each antibody and antigen pair for which a positive signal was detected, titration ELISA was performed to obtain the EC50 values.

Each recombinant protein antigen or negative control Lysozyme (MP Biomedicals #195303) were directly absorbed to ELISA plate at 3 pg/ml (50 pl per well) and incubated overnight at 4 °C. Each plate was washed with PBS. Plates were blocked with 200ul/well of blocking solution (1 % BSA w/v in PBS) for 1 hour at room temperature. Following this, the blocking solution was removed and antibodies to be assessed were diluted to 100ug/ml in blocking solution (1 % BSA w/v in PBS) and applied to the plate. Plates were incubated at room temperature for 1 hour. Each plate was washed with PBS/0.1 % Tween and with Anti-human (Fab)’2- HRP (Jackson Immunoresearch #109-035-097; Lot: 148466) was added to plate and incubated for 1 hour at room temperature to detect antibody binding. Each plate was washed with PBS/0.1 % Tween and TMB solution (Life Technology; #002023) added. Plates were incubated for 5 minutes at room temperature prior to the addition of stopping solution (0.5M sulphuric acid). Absorbance read on Molecular Devices FilterMaxF5 plate reader at 450nm.

Bio-layer interferometry for characterisation of CD33 binding kinetics

Eleven lgG1 converted antibodies were tested in a bio-layer interferometry (BLI) experiment for binding to human recombinant CD33 (R&D Systems, 10375-SL-050).

Each antibody was loaded on a row of eight anti-human IgG Fc capture biosensor tips (AHC tips, Sartorius Stedim part #18-5060), and the row was then equilibrated to kinetic buffer (Sartorius Stedim part #18-1 105) before immersion in a dilution series of CD33 analyte in kinetic buffer at 25°C. The binding response was measured. Tips were then immersed in kinetic buffer and the response was measured. Association and dissociation curves were fitted to response measurements using the manufacturer’s analysis software (Octet Data Analysis HT12.0) and kinetic parameters were calculated.

Phagocytosis in inflammatory-stimulated human iPSC derived microglia

Induced pluripotent stem cell (iPSC)-derived microglia (FujiFilm (CDI) # R1131) were seeded at 20000 cells/well and rested for 3 days with a 50% media change on day 3 in culture. On day 4, cells were treated with 50ng/ml LPS (tlrl-eklps Invivogen) for 24 hours and then with 50pg/ml test antibody for a further 24 hours. Cells were then treated with 0.5pg/well of pHrodo red-labelled Amyloid p and red fluorescence monitored using the Incucyte S3. (Beta-Amyloid(1-42) Aggregation Kit, rPeptide #A-1170-025 labelled with pHrodo Red, Succinimidyl Ester (ThermoFisher Scientific #P36600).

Human whole blood phagocytosis assay

Blood was sourced from an NHS Blood and Transplant Apheresis Cone and diluted 1 :10 in RPMI media (Invitrogen A4192301). 25pl of diluted blood and 2ml of 1 x lysis buffer (ebioscience 00-4300-54 10 x diluted in water to 1 x just before use) was added to each well and mixed well by pipetting. Samples were incubated at room temperature for 15 minutes and then spun at 400g for 5 minutes to pellet remaining white blood cells. Supernatant was discarded to waste and cells resuspended in 100 pl of RPMI media containing 10 pg/ml S.Aureus pHrodo-red and 10 pg/ml antibody. Samples were incubated at 37°C for 3 hours and then washed by adding 1 ml PBS/ 2% FBS, spinning at 400g for 5 mins to pellet and then discarding supernatant to waste. Samples were resuspended in 50ul of 1 in 50 human Fc block diluted in PBS/2% FBS (BD Biosciences 564219) and incubated for 15 minutes at 2-8°C. 50 pl of 1/100 diluted anti-human CD14 APC in PBS/ 2% FBS was added on top of the Fc block and incubated for a further 30 minutes at 2-8°C. Samples were washed as above with 1 ml PBS/2% FBS and then resuspended in 200ul PBS/2% FBS/1 in 10,000 DAPI for analysis on a BD FACSymphony flow cytometer. Live monocytes were gated (DAPI negative, CD14 positive) and the 561-586/15 channel used to assess pHrodo-red fluorescence within the live monocyte population.

CD33 depletion on human monocytes

Previously frozen PBMC-isolated human monocyte cells were defrosted, washed and rested O/N in RPMI media (Invitrogen A4192301) + 10% Fetal Bovine Serum (FBS) at 37 °C, 5% CO2 (standard cell culture condition). The next day, rested monocyte cells were collected and washed once in RPMI+10% FBS. Then, cells were resuspended at a final concentration of 2x10⁶ cells/ml and 50pl were plated in each well of a flat bottom 96 well plate for 1x10⁵ cells/well (Corning, Cat. 3595). Cells were rested in the incubator while reagents were prepared. Reagents (ATL_5802, ATL_5810, ATL5338, ATL_5909) were prepared at double of the experiment final concentration in RPMI+10%FBS, at a single concentration (80nM) or a range (80, 60, 40, 30, 20, 2 nM) at 60 pl/well in a round bottom 96 well plate (Corning, cat. 3799). ATL5338 is an isotype control produced in house based on the FITC binding antibody 4-4-20 described in Jung et al JMB 294, 163 (1999). Then 50pl/well from the reagents plate were transferred onto the 50pl/well 96-well plate containing the resting monocytes to a final volume of 10OpI and then to a final concentration of 40, 30, 20, 10, I nM.The plate was subsequently incubated for 5 hours at standard cell culture conditions. After incubation the plate was maintained on ice so as to avoid any further antibody internalisation. All subsequent steps were performed at 4 °C in the dark. Cells were collected by centrifugation (5 minutes at 300g) and supernatant removed. Then, cells were resuspended in 50pl Live/Dead fixable violet dead cell stain (Cat. L34955) and incubated for 7 minutes. A washing step was performed by adding 150pl of PBS/2% FBS per well on top of each sample and the plate was centrifuged at 300g for 5 minutes. Supernatants were removed and cells resuspended in staining antibodies: anti-human CD33 domain V specific (Biolegend, cat. 303428), anti-human CD33 domain C specific (Biolegend, cat. 366620), anti-human CD45 (Biolegend, cat. 304027), anti-human CD14 (Biolegend, cat. 301806), anti-human CD16 (Biolegend, cat. 302016), anti-human CD3 (Biolegend, cat. 300412) plus 1 :50 human Fc block (BD Biosciences 564219) diluted in PBS-2% FBS. Staining was performed at 4°C for 40 minutes. Then cells were washed twice as above and resuspended in a final volume of 200°pl/well of PBS-2% FBS for analysis on a BD FACSymphony flow cytometer. Live monocytes were gated (dead stain negative, CD45 positive, CD3 negative, CD14 positive, CD16 positive/negative) and the MFI emitted by the two CD33 antibodies recorded.

Phagocytosis modulation via CD33 engagement of freshly isolated human monocyte cells

The antibodies were prepared at double the final assay concentration as a single concentration (60nM) and as a 1 :4 serial dilution starting from 60nM in RPMI+10%FBS and distributed in a round bottom 96-well plate (Corning, cat. 3799), 60ul/well. Latrunculin A was diluted 1 :5000 for the wells used as negative control. The S.Aureus pHrodo bait (Thermo Fisher, cat. A10010) for phagocytosis was prepared in RPMI+10%FBS media at double the final concentration of 10pg/ml and distributed in a flat bottom 96 well plate (Corning, Cat. 3595). Then, human monocyte cells were isolated from fresh PBMC following the Miltenyi protocol (Cat. 130-096-537). The obtained monocyte cells were resuspended at 1x10⁶ cells/ml and 10OpI distributed in each well of a round bottom 96 well plate (Corning, cat. 3799). Cells were centrifuged at 300g for 5 minutes, the supernatant removed and cells resuspended in the prepared antibody solutions before transfer to the plate containing pHrodo reagent for a final volume of 100ul. The emitted fluorescence was recorded live using the Incucyte system (Sartorius).

There were 17 treatment groups. Each treatment group was given a letter so that the observer was blind to the identity of the treatment groups. Mice were randomly allocated to treatment groups.

This experiment consisted of 17 conditions (65 animals in total):

1 . Control 10 mg/kg Pre-dose & 24 hours

2. Tested antibodies (ATL5802, ATL5810, and ATL5909) 10, 40 mg/kg; Pre-dose & 24 hours

3. Control 10 mg/kg Pre-dose, 24 hrs, 7 days

4. Tested antibodies (ATL5802, ATL5810, and ATL5909) 10, 40 mg/kg; Pre-dose, 24 hrs, 7 days

5. Tested antibodies (ATL5802, ATL5810, and ATL5909) 1 mg/kg; Pre-dose, 4 hrs, 24 hrs, 7 days Groups 1 and 2 consisted of 3 animals per group (n=3 biological replicates, 21 animals total), groups 3 and 4 consisted of 5 animals per group (n=5 biological replicates, 35 animals total), group 5 consisted of 3 animals per group (n=3 biological replicates, 9 animals total). Volumes of blood was guided by UK Home Office guidelines (maximum of 200 pl per mouse in total), so groups 1 and 2 were 200 pl pre-dose, groups 3 and 4 were 100 pl at both pre-dose and 24 hours, and group 5 will be 66 pl at pre-dose, 4 and 24 hours. The final timepoint for each group involved a terminal blood sample. Mice were humanised from two separate donors (n=30 from one and n=35 from the second). Groups 1 ,2 and 5 will use animals from donor 1 , with groups 3 and 4 from donor 2.

Samples from each group are used as follows:

• Group 1 and 2 -Pre-dose (ELISA), 24 hours (ELISA & Flow cytometry)

• Group 3 and 4 -Pre-dose (ELISA), 24 hours (Flow cytometry), 7 days (ELISA & Flow cytometry)

• Group 5 -All timepoints (ELISA) The dose volume for all treatments was 10 ml/kg.

Phagocytosis in peripheral monocytes from CD34+ mice

Blood from CD34+ NSG mice was processed as follows. 25 pl of mouse blood and 2ml of 1 x lysis buffer (ebioscience 00-4300-54 10 x diluted in water to 1 x just before use) was added to each well and mixed well by pipetting. Samples were incubated at room temperature (18-22°C) for 15 minutes and then spun at 400g for 5 minutes to pellet remaining white blood cells. Supernatant was discarded and cells resuspended in 100 pl of RPMI media containing 10 pg/ml S. aureus pHrodo. Samples were incubated at 37°C for 3 hours and then washed by adding 1 ml PBS/ 2% FBS, spinning at 400g for 5 mins to pellet and then discarding supernatant to waste. Samples were resuspended in 50ul of PBS/2% FBS containing 1 in 50 human Fc block (BD Biosciences 564219) and 1/100 mouse Fc block (Biolegend 156603) and incubated for 15 minutes at 2-8°C. An antibody cocktail was prepared containing the following antibodies at 1 in 50 dilution in PBS/2% FBS: anti-mouse CD45 AF488 (Biolegend 157608) anti-human CD45 APC (Biolegend 304012) anti human/mouse CD1 1 b BV785 (Biolegend 101243). 50 pl of antibody cocktail was added to each sample, on top of the Fc block already in the well. Samples were incubated for a further 30 minutes at room temperature in the dark. Samples were washed as above with 1 ml PBS/2% FBS and then resuspended in 200 pl PBS/2% FBS/1 in 10,000 DAPI for analysis on a BD FACSymphony flow cytometer. Live human monocytes were gated (DAPI negative, human CD45+, CD11 b+ positive) and the 561-586/15 channel used to assess pHrodo fluorescence within the live human myeloid population. Size-exclusion chromatography (SEC-HPLC)

Antibody samples were diluted to 1 mg/ml in 20 mM histidine acetate, 150 mM NaCI pH 5.5 to run on a Zorbax GF-250 SEC-HPLC column (Agilent) on a Vanquish Flex (Thermo Scientific). Samples were separated by size in mobile phase 20mM Sodium phosphate, 300mM Sodium sulfate and 100mM Arginine at a flow rate of 0.75ml/minute at +25C for 25 minutes per sample. Data acquisition at 280nm, Chromeleon software (Thermo Scientific) was used to integrate the chromatograms.

Charge variant analysis was performed by preparing a master mix to dilute antibody samples to run on a clEF cartridge on a Maurice instrument (Protein Simple). The master mix had a final concentration of methyl cellulose 0.35% (Protein Simple, 101876), pharmalyte pH 3-10 4% (Protein Simple, 17-0456-01), 10mM arginine (Protein Simple, 042-691), pl markers 4.05 and 9.99 0.01 % (Protein Simple, 046-029 and 046- 034). Samples were diluted at 0.15-0.25mg/ml in the master mix and were run for 1 minute at 1500 volts followed by 4.5 minutes at 3000 volts. A system suitability standard (Protein Simple, 046-044) was also run at the beginning of the run. Stability data generated at 2 and 4 weeks was overlaid to compare the charge species profile.

Thermal shift assay

Protein thermal shift measurements were performed on an Uncle (Unchained labs). Antibodies were diluted to 1 mg/ml or 5mg/ml in 20 mM Histidine Acetate, 150 mM NaCI, pH 5.5 buffer and run through a temperature ramp of 25-95°C increasing at a rate of 0.5°C/minute. Samples were run in triplicate, loading 8.8pl in 3 different wells of a uni (Unchained Labs). Laser settings were set to achieve an initial fluorescence in the 300-350 nm range of 10000 - 50000 counts. Melting temperature (Tm1/Tm2) and aggregation temperature (Tagg/Tonset) were analysed using Uncle Analysis software v6 (Unchained Labs). Tm measurement was calculated using the 350/330nm ratio, while Tonset and Tagg were obtained from SLS read at 266nm.

Variants of the CD33 extracellular domain were prepared as individual recombinant proteins. The variations were based on differences between human and cynomolgus monkey CD33, since the lead panel of mAbs were known to bind to human, but not cynomolgus, CD33.

The DNA sequence encoding human CD33 residues 21-232 was synthesised and cloned into pcDNA3.1 (+) by Genscript. Insertion of native residues 18-20, single and multiple residue cynomolgus mutations were conducted using the Q5 Site-Directed Mutagenesis kit (NEB). The wild-type plasmid (residues 18-232) and seven mutant plasmids were individually transfected into 15 mL cultures of Expi293F cells (Thermofisher) according to the manufacturer’s instructions. The seven mutant plasmids were: N20R F21V W22R Q24E [P1]; I47V Y50H D51T K52R [P2]; I67V R69L [P3]; Q83R [P4]; R122K [P5]; P132T [P6]; N20R F21V W22R Q24E P132T [P6+1]; I47V Y50H D51T K52R R122K [P2+5], These mutations are illustrated on Figure 11A. The cells were removed from the cultures by centrifugation 5 days after transfection, and the proteins were purified from the supernatant on an AKTA pure (Cytiva) by Ni-affinity using a HisTrap excel column (Cytiva) followed by size exclusion chromatography using a Superdex Increase 10/300 GL column (Cytiva). Eleven lgG1 antibodies were tested in a single point ELISA against the wild-type CD33 and all 7 CD33 variant proteins. Each recombinant protein antigen or negative control Lysozyme (MP Biomedicals #195303) were directly absorbed to an ELISA plate at 3 pg/ml (50 pl per well) and incubated overnight at 4 °C. Each plate was washed with PBS. Plates were blocked with 200ul/well of blocking solution (1 % BSA w/v in PBS) for 1 hour at room temperature. Following this, the blocking solution was removed and antibodies to be assessed were diluted to 266 nM in blocking solution (1 % BSA w/v in PBS) and applied to the plate. Plates were incubated at room temperature for 1 hour. Each plate was washed with PBS/0.1 % Tween and anti-human (Fab)’2- HRP (Jackson Immunoresearch #109-035- 097; Lot: 148466) was added to plate and incubated for 1 hour at room temperature to detect antibody binding. Each plate was washed with PBS/0.1% Tween and TMB solution (Life Technology; #002023) added. Plates were incubated for 5 minutes at room temperature prior to the addition of stopping solution (0.5M sulphuric acid). Absorbance was read on Molecular Devices FilterMaxF5 plate reader at 450nm. Absorbance values were converted to % of WT CD33 binding, where WT=100%.

Antigen constructs

Table 1 below summarises the antigen constructs used in these examples.

Table. 1. Summary of antigen constructs used in examples.

Mouse pK study

Male homozygous C57BL/6-Cd33tm1 (CD33)/Bcgen (common name: hCD33) were obtained from Biocytogen Technology Co., Ltd. and shipped to Pharmaron (Ningbo, China) for this study. At 6-8 weeks of age, mice (n=3/treatment group) were weighed and administered a single intraperitoneal (i.p.) dose of freshly prepared antibody solutions at 1 mg/kg and/or l Omg/kg at a dosing volume of 5ml/kg. All mice had free access to food and water. Cage side observations were performed daily and clinical observations were made prior to dosing and at each time point of sample collection.

Blood samples were collected from the orbital vein at 4 hours, 24 hours, and 7 days (144 hours) after a single i.p. administration of antibody. Briefly, whole blood samples were left at room temperature for 30min and then centrifuged at 3,500 x g for 15 minutes at 4°C to obtain the serum fraction. Serum samples were immediately transferred to cryogenic vials and stored at -75°C until analysis. Samples were processed and analysed as follows. Serum samples were diluted in assay buffer (0.1 % bovine serum albumin [BSA]-0.05% Tween 20-PBS) and vortexed for 30 seconds prior to loading into microwell plates (Corning Incorporated 96-well cell culture plates). Serum concentrations of test articles were determined using a custom developed ELISA method (Pharmaron, Ningbo, China) using a goat antihuman IgG Fc polyclonal primary antibody (5ug/ml) and a goat anti-human IgG monoclonal antibody, HRP (1 :5000). Readouts were obtained on a Molecular Devices SpectraMax ID3 and PK calculations performed using WinNonlin (PhoenixTM, version 8.3).

In vivo Alzheimer’s Disease model

Male and female Rag2-/- Il2ry-/- hCSFIKI App ^NL-^G-^F mice (thereafter called APP^{NL G F} mice) were transplanted with iPSC-derived human microglia progenitors at postnatal day 4 (P4) as described in Mancuso et al., 2022). Beginning at ~4 months of age, mice were treated once weekly for 12 weeks with 40mg/kg ATL_5802 (n=6) or the isotype control antibody, ATL-5338 (n=7). To assess in vivo phagocytosis, the florescent amyloid-p label, Methoxy-X04 (MX04) was used and the extent of amyloid-p uptake assessed by measuring the percentage of MX04 positive microglia using flow cytometry. At ~16 weeks of age, mice were injected with freshly prepared Methoxy-X04 (ab142818, prepared in a 1 :1 ratio with DMSO:HBSS) at 10mg/kg, i.p.. Twenty-four hours following injection with MX04, mice were injected with the euthanasia agent, Dolethal, and perfused with ice-cold PBS. The harvested brain tissue was immediately placed in FACS buffer (1X PBS with 2% FBS/FCS and 2mM EDTA) for downstream processing by flow cytometry. For flow cytometric analysis, brain tissue was dissociated using the Miltenyi neural tissue dissociation kit as per manufacturer’s instructions. Samples were run on the MACSQUANT Analyzer 10 and gated for viability and human vs. mouse microglia using the following antibodies: Fixable Viability Dye eFluor780 (1 :2000, Invitrogen eBioscience Cat#65-0865-14), APC mouse anti-human monoclonal CD45 antibody (1 :50, BD Bioscience clone HI30 Cat#555485) and the PE-conjugated recombinant human anti-mouse CD1 1 b antibody (1 :100, Miltenyi Biotec Cat#130-1 13- 806). Following exclusion of non-viable cells, cells were gated using hCD45 and the median florescent intensity (MFI) of the human microglia population was measured and set as a threshold for each mouse. The percentage of MX04 positive microglia was calculated by quantifying the proportion of MX04+ microglial cells that was higher than this set threshold.

Tau phagocytosis in human iPSC-derived microglia iPSC derived microglia from FujiFilm (CDI) (Catalogue Number R1131) were seeded at 20,000 cells/well and rested for 3 days with 50% media change on day 3 in culture. On day 4 cells were treated with 50 ng/ml LPS (tlrl-eklps, Invivogen) for 6 hours and then with 50 pg/ml test antibody for 24 hours. Cells were then treated with 0.5 pg/well of pHrodo.Red (ThermoFisher Scientific #P36600)-labelled TauP301 S (Abeam #ab246003) sonicated once for 5 min directly prior to treatment. Red fluorescence was monitored using the Incucyte S3. iCell Microglia (Cellular Dynamics C1110) were seeded at 25,000 cells per well in iCell Microglia Media into a flat bottom Poly-D-Lysine (PDL) coated 96 well plate. PBMCs were isolated from a leukapheresis cone (NHS BT) using Lymphopure, (Biolegend 426201) and Leucosep tubes (Greiner 227290). After 24 hours, media was harvested and cytokine levels measured using a flow cytometry-based multiplex immunoassay (LEGENDplex™ Biolegend 741081 , 740930, 740502, 740796, or 741795).

PBMC were isolated from a leukapheresis cone (NHS BT) using Lymphopure, (Biolegend 426201) and Leucosep tubes (Greiner 227290).

Monocytes were isolated from PBMC with a Pan Monocyte Isolation kit, (Miltenyi 130-096-537) and MS columns (Miltenyi #130-042-201).

100,000 PBMC or monocytes per well were seeded in RPMI media (Gibco A4192301) + 10% FBS in a U- bottom plate (PBMC) or flat bottom plate (monocytes). Cells were treated with 50 pg/ml of isotype control antibody (anti-fluorescein) or ATL_5802 antibody. Some assays involved treating the cells with 50ng/ml LPS (Invivogen, tlrl-eklps) for 6 hours, before adding the antibody. After 24 hours, media was harvested and cytokine levels measured using a flow cytometry-based multiplex immunoassay (LEGENDplex™ Biolegend 741081 , 740930, 740502, 740796, or 741795).

Cytokines tested in these immunoassays included: CCL2 (MCP-1); CXCL8 (IL-8); IFN-y; IL-10; IL-12p40; IL12p70; IL-17A; IL-1 p; IL-2; IL-23 ; IL-4 ; IL-6 ; TGF-p1 ; TNF-a ; CCL17 (TARC) ; CCL2 (MIP-3a) ; CCL3 (MIP-1a) ; CCL4 (MIP-1 ) ; CCL5 (RANTES) ; CXCL1 (Gro-a); CXCL5 (ENA-78); CXCL9 (MIG); CCL1 1 (Eotaxin); CX3CL1 (Fractalkine); IL-18; sRAGE; sTREM-1 ; sTREM-2; VEGF; VILIP-1 ; p-NGF; BDNF; CCL2 (MCP-1). Details of the cytokines tested are provided below:

Monocytes: CXCL8 (IL-8), CCL11 (Eotaxin), CCL17 (TARC), CCL2 (MCP-1), CCL5 (RANTES), CCL3 (MIP-1a), CXCL9 (MIG), CXCL5 (ENA-78), CCL20 (MIP-3a), CXCL1 (GROa), CCL4 (MIP-1 ) PBMC: IL-12p70, TNF-a (TNFSF2), IL-6, IL-4, IL-10, IL-1 p, CCL17 (TARC), IL-12p40, IL-23, IFN-y PBMC - 6 donors: TNF-a, IL-6, IL-1 beta, IFN-y iPSC Microglia: CXCL8 (IL-8), CCL11 (Eotaxin), CCL17 (TARC), CCL2 (MCP-1), CCL5 (RANTES), CCL3 (MIP-1a), CXCL9 (MIG), CXCL5 (ENA-78), CCL20 (MIP-3a), CXCL1 (GROa), CCL4 (MIP-1 p) VILIP-1 , sTREM-2, BDNF, TGF-p1 , VEGF, IL-6, sTREM-1 , p-NGF, IL-18, TNF-a, sRAGE, CX3CL1 (Fractalkine) PBMC and Monocytes: IL-4, IL-2, IL-1 p, TNF-a (TNFSF2), CCL2 (MCP-1), IL-17A, IL-6, IL-10, IFN-y, IL- 12p70, TGF-p1 (Free Active), CXCL8 (IL-8).

PBMC - 4 donors: MCP-1 , IL-1 p, TNF-a, IL-6, IFN- IL12p70, IL-4, IL-10, CCL17, IL12p40, IL-23, TARC Microglia: IL-4, IL-2, IL-1 p, TNF-a (TNFSF2), CCL2 (MCP-1), IL-17A, IL-6, IL-10, IFN-y, IL-12p70, TGF-p1 (Free Active), CXCL8 (IL-8).

Microglia Cell Culture

Microglial cells (iCell Microglia, FUJIFILM Cellular Dynamics, Inc., Catalog #: R1131) were cultured in accordance with the manufacturer's recommended protocols. Prior to cell seeding, culture surfaces were prepared by coating with 0.1 mg/mL Poly-D-Lysine solution (Gibco) and incubated overnight to enhance cell adhesion. Subsequently, cells were plated in 24-well plates at a density of 120,000 cells per well. After a four-day resting period, microglial cells underwent a stimulation protocol. This involved pre-treating the cells with 50 ng/mL LPS (Sigma Aldrich, Catalog #: L2637-10Mg) for 24 hours, followed by the addition of specific antibodies for 20 minutes, 6 hours and 24 hours. Measuring phosphorylated proteins at the 20- minute mark is generally regarded as ideal, due to the transient nature of phosphorylation. Consequently, later time points, such as 6 and 24 hours, are used to observe alterations in protein levels. Western blotting

For protein extraction, we employed RIPA cell lysis buffer, consisting of 50 mM Tris (pH 8.0), 150 mM NaCI, 5 mM EDTA, 1 % NP-40, 0.5% sodium deoxycholate, and 1% SDS, supplemented with protease and phosphatase inhibitors from Sigma. The Auto Western Testing Service was provided by RayBiotech, Inc. (Peachtree Corners, GA USA). 0.1 mg/mL sample concentration was loaded into the automated capillary electrophoresis machine. The following antibodies were employed for the protein analysis: Phospho SYK (Tyr525/526) (Cell Signaling; Catalog No. 12710T) and total SYK (Cell Signaling; Catalog No. 113198S). As a loading control, we used the Gluceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody, provided by RayBiotech from their service library.

Microglia Cell Culture for RNA Seguencing

Microglial cells (iCell Microglia, FUJIFILM Cellular Dynamics, Inc., Catalog #: R1131) were cultured in accordance with the manufacturer's recommended protocols. Prior to cell seeding, culture surfaces were prepared by coating with 0.1 mg/mL Poly-D-Lysine solution (Gibco) and incubated overnight to enhance cell adhesion. Subsequently, cells were plated in 24-well plates at a density of 120,000 cells per well. After a four-day resting period, microglial cells underwent a stimulation protocol to mimick inflamed conditions. This involved pre-treating the cells (i.e. prior to antibody treatment) with 10 ng/mL lipopolysaccharide (LPS, Sigma Aldrich, Catalog #: L2637-10Mg) and 20 ng/mL interferon-gamma (IFN-gamma, PeproTech, Inc., Catalog #: 300-02-20ug) for 24 hours, followed by the addition of ATL_0005802, ATL_0005854, ATL_0005909 or ATL_0005338 (isotype), for a further 24 hours, resulting in a total exposure period of 48 hours to LPS/IFNy (LI). This is compared with a condition mimicking healthy conditions in which the cells are pre-treated with vehicle only. Post-treatment, the cell culture medium was aspirated, and cells were lysed directly in the wells by adding 400 pL of TRIzol Reagent (Invitrogen, Catalog Number: 15596026). It was ensured that the TRIzol Reagent uniformly covered the surface of each well. Plates were gently agitated to facilitate thorough mixing and effective cell lysis. The cell lysate, comprising both TRIzol Reagent and cellular contents, was then carefully transferred into individually labelled tubes using a pipette. These tubes were immediately placed on dry ice for rapid cooling, a critical step to preserve RNA integrity. Subsequently, the samples were stored at -80°C for long-term preservation until RNA extraction was conducted.

RNA extraction, library preparation and seguencing

Collected samples were processed for next-generation sequencing as follows. Total RNA was extracted from TRIzol-cryopreserverd samples following manufacturer’s recommendations. RNA integrity was assessed by Tapestation (Agilent Technologies, Palo Alto, CA, USA) and concentration by Qubit 2.0 Fluorometer (ThermoFisher Scientific, Waltham, MA, USA). Samples with low quality were excluded. Poly(A) mRNA was enriched with Oligod(T) beads following manufacturer’s recommendations.

Libraries were prepared from RNA with ERCC spike-in using Twist RNA Library Prep Kit and the Twist UMI Adapter System (Twist Bioscience) following manufacturer’s recommendations. Library size was assessed on the Tapestation, concentration by Qubit 2.0 Fluorometer, and final quantitation by quantitative PCR (KAPA Biosystems, Wilmington, MA, USA). Multiplexed samples were sequenced with single index using 2x150bp reads on an llumina HiSeq4000 sequencer. Bioinformatic analysis

Quality Control, sequence alignment and quantification was performed as follows. Briefly, raw data quality was evaluated with FastQC (Andrews, 2010). Sequence reads were trimmed using fastp v.0.23.1 (Chen et al., 2018). UMI-based de-duplication was performed using fastp v.0.23.1 (Chen et al., 2018). Trimmed and de-duplicated reads were mapped to the Homo sapiens GRCh38.p7 with ERCC genes using STAR aligner v.2.5.2b (Dobin et al., 2013). Unique gene hit counts were calculated by using ‘featurecounts’ from Subread v.1.5.2 (Liao et al., 2014).

Differential expression analysis was performed with R using DESeq2 v1 .38.3 (Love et al. 2014). The Wald test was used to generate p-values and Iog2 fold changes. Genes with a BH adjusted p-value < 0.05 and absolute Iog2 fold change > 1 were considered as differentially expressed genes for each comparison of interest (namely, vehicle + anti-CD33 antibodies compared to vehicle + isotype and LI + anti-CD33 antibodies compared to LI+ isotype). Principal component analysis (PCA) was calculated using ‘plotPCA’ function from DESeq2 on variance stabilised counts (DESEq2 ‘varianceStabilizingTransformation’ function) and using the top 25% most variable genes. Volcano and PCA plots were plotted using the R library ggplot2 v.3.4.3 (cran.r-project.org/web/packages/ggplot2/index.html).

Pathway analysis: The R package msigdbr v.7.5.1 (Dolgalev, 2022) was used to retrieve gene sets from the Molecular Signature Database, MSigDb (Broad Institute, www.gsea-msigdb.org), including the curated HALLMARK collection and REACTOME pathways. Gene set enrichment analysis (Subramanian et al. 2005) was performed with fgsea v1 .24.0 (Korotkevich, 2019) for each comparison of interest, using the corresponding gene list ranked using Wald statistics. Gene sets/pathways with a -Iog10(adjusted p-value) > 10 were considered enriched in one condition compared to another. The directionality of the change was assessed using the normalised enrichment score (NES). Plots were generated with ‘plotEnrichment’ function from fgsea, ggplot2 and ggpatern v.1.1.0-0 (cran.r- project.org/web/packages/ggpattern/readme/README.html).

Establishment of the Quad-culture system

Induced pluripotent stem cell (iPSC)-derived glutamatergic neurons (Fujifilm; Catalog #: R1061), GABAergic neurons (Fujifilm; Catalog #: R1013), astrocytes (Fujifilm; Catalog #: R1092), and microglia (Fujifilm; Catalog #: R1131) were utilized to establish a quad-culture system to recapitulate physiological conditions of the central nervous system. These cells were cultured under standard conditions until they reached the appropriate confluence for experimentation. On Day 6, post-establishment of the quad-culture system, cells were pre-treated with 10 ng/mL lipopolysaccharide (LPS, Sigma Aldrich, Catalog #: L2637- 10Mg) and 20 ng/mL interferon-gamma (IFNy, PeproTech, Inc., Catalog #: 300-02-20ug). The next day, the quad-culture system was exposed to CD33 antibody, ATL_5802, or ATL_5338 (isotype control antibody). Each antibody was used at a concentration of 50 pg/mL. The cultures were incubated with these antibodies for an additional 24 hours. On day 8 of the culture, supernatant was collected. Quantification of lnterleukin-6 (IL-6), GFAP, IP-10 and MCP-1 in the collected samples was performed using an enzyme- linked immunosorbent assay (ELISA) service provided by RayBiotech, Inc. (Peachtree Corners, GA, USA). All ELISA assays were conducted according to the manufacturer's protocols. In vitro

The screening was performed by Charles River (UK) using their proprietary Retrogenix Cell Microarray Technology (see www.criver.com/products-services/discovery-services/screening-and-profiling- assays/retrogenix-cell-microarray-technology?region=3696).

Convergent sequence clusters derived from the antibody repertoire of resilient individuals can be used to identify disease-specific antibody sequences. The present inventors sought to identify candidate protective antibodies from Alzheimer’s Disease (AD)-resilient individuals in a prospective cohort (the European Prevention of Alzheimer’s Dementia Consortium, ep-ad.org).

Pathophysiological changes of Alzheimer’s Disease can be detected in individuals decades before the onset of cognitive symptoms of dementia, i.e. in the preclinical phase of disease. For example, the concentration of Ap42 in cerebrospinal fluid (CSF) is reduced in early disease progression relative to healthy controls, while the concentration of CSF phospho tau is increased. The ‘A/T/N’ classification scheme (Jack et al., 2016) defines convenient binary labels to describe such biomarkers, where “A” refers to the value of a p-amyloid biomarker (amyloid Positron Emission Tomography (PET) or CSF Ap42); “T,” the value of a tau biomarker (CSF phospho tau, or tau PET); and “N,” biomarkers of neurodegeneration or neuronal injury ([18F]-fluorodeoxyglucose-PET, structural MRI, or CSF total tau) (Jack et al. 2016). Below 1 ,000 pg/mL is considered amyloid positive, and above 27 pg/mL is considered pTau positive (Amft et al., 2022; Blennow et al., 2019).

From the total cohort of 2,096 participants, 127 were selected for convergence analysis based on data and sample availability (plasma and PBMC, and pTau and Ap42 values from CSF and plasma) including 37 AD-resilient subjects, where resilience is defined by A+T- CSF biomarkers and no cognitive impairment using the Azheimer’s disease Assessment scale — Cognitive subscale (ADAS-Cog), the Mini-Mental State Examination (MMSE), and Clinical Dementia Rating Scale Sum of Boxes (CDR-SOB) (Duff et al., 2008; Balsis et al., 2015). Sequencing of the antibody repertoires of these resilient individuals revealed a convergent heavy chain variable (VH) sequence among two resilient individuals, one of whom additionally carries a heterozygous Apo E3/E4 genotype, further predisposing them to AD (Michaelson et al., 2014). Figure 1 shows the number of clonotypes, i.e. clusters of BCRs with high sequence similarity, identified in these two resilient individuals, showing that there were 64 shared clonotypes between these 2 resilient individuals. A representative sequence from one of the 64 shared clonotypes was shown to bind CD33.

The VH sequence identified in resilient individuals was subsequently paired with a VL using a transformerbased model comprising an encoder-decoder model trained on a corpus of paired VH-VL sequences. Further details on how such a model may be trained and used is provided in WO2022/223451 (the entirety of which is incorporated herein by reference). The trained model takes a VH sequence as input and generates a single complementary VL sequence as output. The resulting antibody, ATL_5082 (also referred to herein as ATL_0005082) was included in an ELISA screen for binding against a panel of neurodegeneration linked targets. The panel consisted of 14 targets, including BA-1-14 biotinylated peptide, BA-1-42 biotinylated peptide, Tau-352, Tau-441 , NF-L (Neurofilament light chain), Trem2, Galectin-3, CD33 (Siglec 3), ApoE4, LRP8 (ApoE4 Receptor), HTT exonl 48Q (mutant HTT exon 1), Transthyretin, Baculovirus particles, Lysozyme. The antibody was shown to bind CD33 (Figure 2). This indicated a link between CD33 reactivity and resilience to AD in the Alzheimer’s Disease (AD)-resilient individuals in the EP-AD cohort.

Interestingly, serum and plasma autoreactivity to CD33 (defined as a z-score cutoff of 0.5 in sera ELISA) was observed in 21 subjects (25 samples), including one “super-centenarian”, that is an individual over the age of 100 with a self-reported good cognitive function (score of at least 28 out of 30 in a mini-mental state examination, cognitively healthy individuals (Holstege et al., 2018)) (Figure 3).

In addition, many cancers are characterised by high levels of CD33 expression, in particular myeloid malignancies such as acute myeloid leukaemia, and lymphoma. Further, it has been shown that expression of CD33-related siglecs on tumour-associated macrophages supports cancer progression, and that inhibitory siglecs can inhibit immune cell activation, thereby suggesting siglec targeting as a possible immune checkpoint inhibition therapeutic strategy (Stanczak et al., 2022). Therefore, antibodies identified and described herein may not only be useful in the treatment and/or prevention of neurodegenerative diseases, but also in cancer.

Example 2 -deep mining bv phage display of repertoires from CD33-reactive subjects

Based on the results in Example 1 , the inventors set out to screen serum and plasma libraries from “supercentenarians” to identify possible protective anti-CD33 antibodies. An ELISA against CD33 was performed, using plasma from a cohort of “super-centenarians” (www.100plus.nl), a subset of the samples on Figure 3. All twelve plasma samples available from the cohort were analysed and the sample with the highest ELISA signal for CD33 (Figure 4A), and low signal for Lysozyme control antigen (Figure 4B), was chosen for phage library generation (Subject SU_0000877 in Figure 4).

A phage Library of scFv molecules displayed on an M13 phage was generated by cloning the heavy chain variable region (VH) repertoire of subject SU_0000877 into a sub-library of phagemid vectors with light chain variable region (VL) sequences from healthy donors. A phage display library of a size 1.4x10⁸ clones was generated and phages were produced for phage display selections.

In order to isolate antibodies binding to CD33 protein, two to three rounds of phage display panning selections were performed on CD33 protein. More than 100 clones were analyzed by Sanger sequencing and full-length sequences were shortlisted for phage ELISA.

Full-length clones derived from phage display selections were subsequently analyzed by phage ELISA using CD33 antigen and lysozyme control (Figure 5). Nine clones with positive ELISA signal and unique sequence were chosen for IgG conversion and further tests.

The nine antibodies derived from the phage display panning selections were shortlisted for IgG conversion and made as lgG1 with LALA mutation in Fc region IgG 1 variants comprising L234A/L235A substitutions reducing binding to the IgG Fc receptors FcyRI, FcyRII and FcyRIII as well as to complement component C1 q, reducing Fc-mediated toxicity (Lund et al. 1991). Next, these, along with comparative antibodies ATL_5909 (also known as “AL003” from Alector), ATL_4828 (also known as Gemtuzumab); and ATL_5503 (an AL003 precursor), were tested in single point ELISA against a panel of antigens comprising CD33 and related proteins (other sialic-acid-binding immunoglobin-like lectins, siglecs): human recombinant CD33, Cynomologus/ Rhesus CD33, human recombinant CD33 C domain, mouse CD33, Recombinant human siglec-6/CD327 Fc Chimera, Recombinant Human Siglec-7/CD328 Fc chimera, Recombinant Human Siglec-8 Fc Chimera, Recombinant Human Siglec-9 Fc Chimera. After the single point ELISA, for each antibody and antigen pair for which a positive signal was detected, titration ELISA was performed to obtain the EC50 value for binding to CD33 (see Table 1).

The nine tested antibodies (see data for ATL_5802; 5810; 5803; 5808; 5854; 5807; 5809; 5853 in Table 2A) were found to bind to full recombinant human CD33 (rhCD33-His), but not the C2 domain (see also Figure 6, where the single point ELISA signal is shown normalised to isotype control=100 and absorbance was measured at 450 nm). The antibodies further showed minimal species cross-reactivity in mice (mCD33-His) and cynomolgus monkey (cCD33-His). Importantly, while the tested antibodies show good binding to CD33, they did not bind to the related Siglec-family proteins (Siglec-6-F; Siglec-7-F; Siglec-8-F; Siglec-9-F), indicating selectivity for CD33. Kinetic data is shown in Table 2C. These data show that although the comparative antibodies have higher affinity for CD33, this is primarily driven by lower Koff (e.g. the on-rate for ATL5909 vs ATL5802 is about two-fold higher, but the off-rate is two orders of magnitude lower). This indicates that the antibodies of the present disclosure dissociate faster from the target, which could underlie their lower rate of peripheral degradation (see Example 3 below). As demonstrated in Examples 4 and 5, the presently described antibodies show very significant effects on phagocytosis (even higher than comparative antibodies), indicating a complex relationship between the antibody’s mode of action and kinetics of target engagement. Table 2A shows binding signal, measured by ELISA, of each antibody for binding to human recombinant CD33 (rhCD33-His); human recombinant CD33 C domain (hCD33(C domain)-His); mouse CD33 (mCD33- His); Cynomologus/ Rhesus CD33 (cCD33-His); Recombinant human siglec-6/CD327 Fc Chimera (Siglec- 6 Fc), Recombinant Human Siglec-7/ CD328 Fc chimera (Siglec-7 Fc), Recombinant Human Siglec-8 Fc Chimera (Siglec-8 Fc), Recombinant Human Siglec-9 Fc Chimera (Siglec-9 Fc) and negative control (Lysozyme). Data shown are raw absorbance values measured at 450nm, normalised to an isotype control set at value 100. ATL_0005909 is a comparative antibody (Alector AL003). ATL_0004828 is a comparative antibody (Gemtuzumab). ATL_0005503 is a comparative antibody (Alector’s 2F5, a murine precursor of AL003 described in US11136390B2)

In conclusion, deep mining by phage display of repertoires from CD33-reactive subjects resulted in a panel of 8 selective CD33-binding antibodies that all bind full CD33 but not the C2 domain.

Amongst these 8 antibodies, two were selected for further investigation on the basis of CDR3H diversity, strong binding, epitope mapping (see Example 8), and different internalisation behaviours (see Example 3), including the antibody ATL5802 and the antibody ATL5810. These were then investigated further in Examples 3-7.

As explained in Example 8, epitope mapping was performed for these antibodies and comparative antibodies. This revealed that 7 out of these 8 antibodies (5810, 5853, 5807, 5809, 5802, 5854 and 5805, see sequence Table 2B) showed similar binding profiles indicating similar epitopes, and these respective binding profiles were different from those of comparative antibodies 4828 and 5909, and from antibody 5803. Therefore, antibody 5803 was not investigated further.

Table 1 B shows the CDR3-H and CDR3-L sequences of eight CD33 antibodies identified by phage display as well as comparative antibodies ATL_0005909, ATL_0004828; ATL_0005503 (indicated by *)

Table 2C shows kinetic data for sequences of anti-CD33 antibodies identified by phage display as well as comparative antibodies (*).

Example 3 - Cell binding and internalisation

CD33 is known to be expressed at high levels in the plasma cell membrane of peripheral blood cells, such as monocytes. In addition to binding to recombinant CD33 as shown in Example 2, it was desirable to select a CD33 antibody that binds to endogenously expressed CD33 but results in minimal internalisation of the antibody-CD33 complex by peripheral blood cells, and therefore avoids accelerated clearance of administered antibodies via this peripheral clearance mechanism. Internalisation of ATL5802 and ATL5810 by monocytes was therefore tested.

Figure 7A (see also Table 3) shows the CD33 depletion on human monocytes 5 hours after adding with ATL_5802 or ATL5810 compared with comparative antibodies ATL5909 and ATL4828. ATL_5802 showed minimal CD33 depletion, and therefore minimal internalisation, on monocytes, similar to the isotype control antibody. ATL_5810 showed some CD33 depletion but less than prior art antibodies ATL5909 or ATL4828. Figures 7B and 7C show the levels of depletion of the CD33 V domain and C domain (respectively for Figures 7B and 7C) on human CD14+CD16+ or CD14+CD16- cells (bottom and top panels, respectively, in each figure) after adding the specified antibodies at a concentration ranging from 1 nM to 40nM. Figure 7D and Figure 7E the area under the curves of Figs. 7B, C for the CD14+CD16- monocytes (see also Table 3 below) and CD14+CD16+ monocytes, respectively. Notably, all tested antibodies of the disclosure showed less CD33 internalisation than comparative antibody 5909. ATL_5802 shows very little CD33 internalisation (see Fig.7A), , and ATL_5854 shows medium CD33 internalisation (see Fig. 7B). It is worth noting that, since the test antibodies all bind in the V domain, changes to MFI of domain V may be attributable to both competition with the detection antibody and depletion of CD33. In contrast, the assay using detection of MFI in domain C will be a true representation of CD33 depletion. Separate data are shown for both classical (CD14+CD16-) monocytes and nonclassical/intermediate (CD14+CD16+) monocytes to show that the CD33 depletion behaviour for the antibodies is similar in both monocyte subsets.

Table 3. Results of flow cytometry experiments measuring cell binding and internalisation. * denotes comparative antibodies.

Comparative antibody ATL5909 has a non-linear PK profile (Ward et al., 2021) and this is likely caused by internalisation into CD33-positive peripheral monocytes. In other words, ATL5909 internalises rapidly in complex with CD33 and depletes CD33 on the cell surface. While this may be desirable at the disease location (e.g. brain tissue), this property causes an antibody sink in the periphery, where the antibody is internalised and degraded inside monocytes. The fact that no or reduced CD33 depletion in the periphery is observed with ATL5802 and ATL5810 (and all other tested antibodies of the disclosure) is an indication that these antibodies are not internalised by peripheral blood cells to the same extent as ATL5909, and therefore an improved PK profile compared with ATL5909 can be expected, and improved availability of the antibody in desired target areas such as the brain.

Example 4 - Phagocytosis assay

Microglia are brain-resident innate immune cells resident in the central nervous system and are essential for CNS health. Phagocytosis of toxic proteins, such as amyloid p, is an important function of microglia. The phagocytosis pathway is suppressed by CD33 signalling, and notably, CD33 is upregulated in microglia of AD patients and associated with cognitive decline. It was therefore of interest to test whether the identified anti-CD33 antibodies are able to relieve this CD33-mediated suppression of phagocytosis.

To investigate the effect of anti-CD33 antibodies ATL5802 and 5810, an in vitro assay using induced pluripotent stem cell (iPSC)-derived microglia was devised to measure phagocytosis of amyloid beta (see methods). No effect of anti-CD33 antibodies was observed in iPSC microglia in a basal resting state (data not shown). However, this was not considered representative of in vivo microglia, and especially microglia in a neurodegenerative context. Induced pluripotent stem cell (iPSC)-derived microglia were therefore first stimulated with LPS, to induce an inflammatory response, representative of the in vivo pathological state of these cells in neurodegeneration. This step was followed by incubation with the panel of anti-CD33 antibodies. Cells were subsequently treated with a pH-sensitive pHrodo red-labelled amyloid p and phagocytosis was measured by monitoring the red fluorescence signal in each well, with increased intracellular levels of red dye compared with the isotype control indicating increased phagocytosis (Figure 8A).

Figure 8B shows the results of the assay with anti-CD33 antibodies ATL5802 and ATL5810. The graph in Figure 8B shows that incubation with ATL5802 results in increased phagocytosis in iPSC microglia compared with prior art antibody ATL5909, and an isotype control.The results on Figure 8C confirm these results on a repeat experiment for ATL_5802 indicating a replicable increase in phagocytosis in inflammatory iPSC microglia in the presence of the antibody of the disclosure (ATL_5802) but not a comparative antibody ATL_5909.

Next, the experiment was repeated with three other anti-CD33 antibodies, ATL_5853; ATL_5854; and ATL_6044 (germlined ATL5802, see Example 7 below). The results of this experiment are shown in Figure 8D. The graph in Figure 8D shows that all of the tested antibodies of the disclosure, i.e. antibodies ATL_5802, ATL_5854 and ATL_5853 increase phagocytosis compared with prior art antibody ATL_5909 or an isotype control antibody.

Figures 8E and F further show that ATL_5802 induces increased phagocytosis to a larger extent compared with agonist TREM2 antibodies currently in clinical trial for the treatment of AD (ATL6166=Alector AL002; ATL6167=Denali DNL919; ATL6170=Vigil VGL101). TREM2 is a microglial activating signalling receptor and supports microglial cell survival by promoting phagocytosis of Ap plaques (McQuade et al., 2020).

To test the effect of anti-CD33 antibodies on phagocytic activity in human peripheral myeloid cells, which are known to express high levels of CD33 (e.g. on monocytes), the phagocytosis assay described above was carried out ex vivo using human peripheral blood mononuclear cells (PBMCs) from healthy donors. The ex vivo myeloid assay uses pH-dependent dye (pHrodo)-labelled S. aureus to measure the level of phagocytosis in myeloid cells (CD14+) by flow cytometry (Figure 9A).

Figures 9B and 9C show that ATL5802 significantly increases phagocytosis in human myeloid cell compared with comparative antibodies ATL5909 (Alector), and ATL4828 (Gemtuzumab). The results for ATL5810 did indicate that the antibody likely induced phagocytosis but the results did not reach significance (vs isotype control) due to high variability in the cellular assay.

In conclusion, these data show that that the antibodies of the disclosure likely increase phagocytosis in a range of human cells expressing CD33, including PBMCs and inflammatory iPSC-derived microglia (as shown in particular by antibodies ATL5802, ATL_5853, ATL_5854, and ATL_5810). The data further show that the antibodies of the disclosure (as shown in particular by antibodies ATL5802, ATL_5853, ATL_5854) increase phagocytosis in inflammatory human iPSC-derived microglia in vitro as well as human myeloid cells ex vivo.

Example 5 - in vivo testing of antibodies in CD34+ humanised mice

The pharmacodynamic properties of ATL5802 and ATL5810 were further tested in vivo in a CD34⁺ NSG™ mouse model, a humanised mouse model modified by the addition of human haematopoietic stem cells to produce a humanised immune system (comprising a mixture of mouse and human monocytes) which can be targeted by anti-CD33 antibodies. Testing these antibodies in a humanised mouse model was necessary because substantial species differences exist between mice and humans in CD33 expression patterns and ligand recognition due to evolutionary divergence between human CD33 and non-primate CD33, and mCD33 biology is therefore not functionally relevant to hCD33 biology (Brinkman-Van der Linden, et al., 2003; Cao et al, 2009). In any case, the anti-CD33 antibodies described herein do not show species crossreactivity to mouse CD33, and therefore in vivo testing is only possible in humanised animal models. First, levels of CD33 on humanised myeloid cells were tested as a way to assess in vivo internalisation of CD33 (Figure 10). Figure 10 shows that injection of ATL5802 and ATL5810 results in minimal CD33 depletion on myeloid cells 24 hours after injection compared with prior art antibody ATL5909. Figure 10 shows the mean fluorescence intensity (MFI) of CD45+CD14+CD33+ cells by flow cytometry. This indicates there is minimal peripheral drug loss of ATL5802 compared with ATL5909.

Figure 12 shows that there is a dose-dependent decrease in levels of CD33 expression in human CD45+ cells in the brain 24 hours after ip injection of ATL5802, supporting target engagement of ATL5802 in the brain.

In addition, Figure 13 demonstrates that intraperitoneal injection of ATL5802 in CD34+ mice results in increased phagocytosis of pHrodo labelled S. Aureus by human peripheral myeloid cells (CD45⁺CD14⁺ CD11 b⁺ cells) compared with injection of ATL5909, and this effect appears to be dose-dependent. The relatively large variability is due to the small number of human cells in the experiment. Nevertheless, the trends observed in vivo confirm the in vitro data shown above.

Together, these data show that ATL5802 shows minimal peripheral internalisation in CD34⁺ humanised mice, while showing target engagement in the brain and increased phagocytosis. ATL5802 was therefore further characterised in terms of antibody expression and developability.

Example 6 - Antibody expression and developability

To ensure that the antibodies identified are stable and therefore suitable for further optimisation, a developability study was carried out.

Initial baseline values were calculated for all 10 antibodies selected in Example 2, which demonstrated high levels of monomer and good thermostability and clEF values (Table 4).

Protein aggregation during antibody storage must be kept to a minimum as it can cause immunogenic reactions. Size exclusion chromatography (SEC-HPLC) was used to evaluate the purity and aggregation of the antibodies after storage at a range of conditions including 2 weeks at -80°C, 2 weeks at 40°C, shaking overnight, 3 freeze-thaw cycles (3x FT); and low pH hold (Figure 14). Figure 14 shows that ATL5802 shows very little change in monomer profile under these forced degradation conditions, indicating excellent stability, and this is further reflected by the 99% monomer value in Table 4. Similar results were obtained for the other candidate antibodies identified in Example 2, all of which showed acceptable stability in terms of % monomer, and 5802, additionally showed particularly good thermostability (Tm1).

Table 4 shows the results of a thermostability assay (melting temperature (Tm1)/ aggregation (Tagg); clEF SEC-HPLC (% monomer).

In addition, target binding of ATL5802 was tested after storage at the forced degradation conditions using ELISA (Figure 15). ATL5802 showed consistent binding to CD33 under all conditions indicating that binding potency is unaffected by temperature changes and freeze thaw cycles.

In conclusion, all antibodies but in particular ATL5802, showed excellent stability and ATL5802 was therefore retained for further development.

Example 7 - Germlininq in framework regions

The lead antibody ATL5802 has 8 framework mutations from the corresponding germline sequence: 6 located in the VH (IMGT positions 1 , 6, 20, 40, 72, 85) and 2 in the VL region (IMGT positions 2 and 8). In order to test which of them can be reversed to the germline sequence without loss of binding, 9 variants of the original antibody were made and tested for binding to CD33 by ELISA (Table 5). The following variants were tested: ATL6040 which had all the mutations reversed to germline sequence and ATL6041-ATL6048 which had all but one mutation reversed. The variants were produced as IgG 1 (LALA Fc mutation).

Table 5 shows the germline mutations introduced in each variant of ATL_5802 as well as their EC50 values obtained by ELISA.

All the antibodies were produced and tested in titration ELISA with CD33 antigen and negative control antigen. Only one antibody, ATL6044, showed binding comparable to the parent suggesting that removal of liabilities or changes to the framework position G40A impact binding. This showed that the A at position 40 in ATL5802’s VH contributed positively to binding to the antigen, but all other positions could be reverted back to germline. Note that this does not indicate that strict adherence to any framework sequence is required, as all antibodies still bound the target irrespective of the germlining mutation introduced. Instead, this indicates that maintaining the A at position 40 of ATL5802 is particularly advantageous.

Example 8 -Epitope mapping

Figure 11 B show the results of the epitope mapping process illustrated schematically on Figure 11 A. The binding profile of each antibody to the different CD33 variants was used to group the antibodies into different epitope bins. Epitope bin 1 contained ATL4828 (comparative antibody Gemtuzumab) and was characterised by a complete loss of binding for variant P1 and P6+1 and a roughly 50% loss of binding for variant P4. In contrast epitope bin 2 contained ATL5909 (Comparative antibody AL003) and was characterised by loss of binding for variant P2 and P2+5, but no effect of any of the other variants. Bin 3 and 4 were highly related groups and contained seven antibodies (5810, 5853, 5807, 5809, 5802, 5854, 5808), which had in common a loss of binding for variant P2 and P2+5, a 50% loss of binding for P4 and an increase in binding for P1 and P6+1. Bin 5 comprised antibody ATL5803 (which was not characterised further herein). This data suggests that other antibodies with the same epitopes as variants with confirmed phagocytosis enhancing activity (ATL5802, ATL5854, and ATL5853) are likely to show therapeutic effects. The data further suggests that the engagement kinetics shown in Example 2 that differ for the antibodies of the disclosure compared to comparative antibodies may be underlined by different epitopes, potentially supporting improved therapeutic effects for these antibodies by reduced peripheral depletion.

Finally, the epitope mapping data also supports a different mode of action of the antibodies of the present disclosure compared to prior art antibodies, and is further consistent with the reduced internalisation behaviour observed in Example 3 compared to comparative antibodies. In particular, the epitope mapping data indicates that the antibodies described herein may interfere with the binding of sialic acid to CD33.

Together, these data show that anti-CD33 antibodies as described herein, and in particular ATL5802, ATL5854, and ATL5853, are likely to show therapeutic effects in the treatment and/or prevention of neurodegenerative diseases such as AD, as well as CD33-expressing cancers such as myeloid leukaemia. This was further validated through extensive in vitro and in vivo studies, described in the subsequent examples.

Figure 16 summarises the sequences of antibodies described herein, aligned using the IMGT antibody numbering convention. The sequences displayed on Figure 16 are also summarised in Table 5 (VH) and Table 6 (VL). In Tables 5 and 6, antibodies of the disclosure are in bold. “Full” refers to the full sequence of the VH (Table 5) or VL (Table 6).

Table 5. Summary of VH sequences. All numbers are SEQ ID Nos referring to Table 0.

Table 6. Summary of VL sequences. All numbers are SEQ ID Nos referring to Table 0.

Example 9 - Mouse pK study

The antibodies described herein are specific to human CD33 and do not significantly cross-react with mouse CD33. Therefore, in order to assess the in vivo pharmacokinetics (PK) of antibody ATL_5802, a transgenic mouse line was used wherein exons 1~3 of the mouse Cd33 gene, which encode the extracellular domain, were replaced by human CD33 exons 1~3.. Expression of human CD33 was previously validated and characterized in this mouse model. Antibodies were intraperitoneally administered at 1 mg/kg and serum human lgG1 levels were assessed at varying time points following a single administration of the human lgG1 antibodies, ATL_5802 or ATL_5909.

The results of this pK study are shown in Figure 17. Figure 17A shows serum levels of ATL_5802 and ATL_5909 at 4 hours, 24 hours and 144 hours (7 days). Figure 17B shows the quantification of total antibody concentration over time (measured by area under the curve; AUC) and reveals significantly increased serum levels of ATL_5802 compared with ATL-5909.

These data demonstrate superior peripheral exposure of ATL_5802 compared with ATL_5909, suggesting that peripheral clearance of ATL_5802 is lower than that of the comparative antibody.

Together these data demonstrate that anti-CD33 antibodies described herein show excellent pharmacokinetic properties thus further supporting their potential use in therapy.

Example 10 -Alzheimer’s disease model

To test whether the anti-CD33 antibodies are capable of attenuating Alzheimer’s disease pathology, the effect of ATL_5802 was tested in a xenotransplantation mouse model of Alzheimer’s disease. Here, amyloid precursor protein (APP) mice carrying the human NL-G-F mutations (named APP ^{NL G F} mice, Saito et al., 2014) were xenotransplanted with iPSC-derived human microglial progenitors, as described in Fattorelli et al., 2021 , Mancuso et al., 2019 and Mancuso et al., 2022. Here, the transplanted cells mature to transcriptionally resemble human primary microglia ex vivo and, after three months, mice show signs of Alzheimer’s disease pathology. Thus, this mouse model can be used to investigate human microglial biology within the context of Alzheimer’s disease pathology.

From 4 months of age mice were treated with 40 mg/kg ATL_5802 or isotype control antibody (ATL_5338) i.p. once/week for three months. This was followed by i.p. injection of methoxy-X04, a fluorescent dye that crosses the blood brain barrier and binds to Ap. Afterwards, microglia were isolated using flow cytometry methods (cells were sorted using FACs and gated on hCD45+) and their phagocytic capacity assessed by quantifying the percentage of methoxy-X04-positive human microglia. The results of this experiment are shown in Figure 18. Mice treated with ATL_5802 showed a significant increase in the percentage of methoxy-X04+ microglia compared with isotype-treated control mice as measured by flow cytometry. These results indicate that ATL_5802 increases in vivo phagocytosis of amyloid-p by human microglia in AppNL-G-F mice xenotransplanted with iPSC-derived human microglia progenitors.

Together, these data demonstrate that ATL_5802 alleviates AD pathology in vivo at least by increasing clearance of neurotoxic Ap through microglial phagocytosis.

Aggregation of pathological (mutant) tau protein into abnormal filaments, or so-called neurofibrillary tangles, is a hallmark of multiple neurodegenerative conditions. In humans and transgenic mice, mutations of P301 (e.g. P301 S and P301 L) in tau leads to tau pathology with spreading of aggregated tau (Strang et al. 2018). Effective clearance of these aggregates from neurons is therefore of high importance to prevent and/or reduce tau pathology. As described in Example 4, microglia are essential for the clearance of toxic extracellular proteins and maintenance of cellular health in the central nervous system.

To test whether the antibodies described herein are able to reduce the spreading of mutant (P301S) aggregated tau, the uptake of these aggregates by microglia with inflammatory phenotype was measured. Briefly, LPS pre-treated iPSC microglia were incubated with the anti-CD33 antibody ATL5802, the comparative anti-CD33 antibody ATL_5909; anti-TREM2 antibody ATL_6170; or an isotype control antibody (ATL_5338) and the uptake of pHrodo-labelled mutant Tau aggregates was subsequently measured using live cell imaging.

The results of this experiment are shown in Figure 19. Figure 19A shows an increase in phagocytosis of labelled tau aggregates, measured as the total red area per well over time, following treatment with either anti-CD33 antibody ATL_5802 or anti-TREM2 antibody ATL_6170 relative to an isotype control antibody using live cell imaging. In contrast, comparative anti-CD33 antibody ATL_5909 tended to decrease the uptake of tau. Figure 19B shows the quantification of AUC of the total area per well for each condition.

Together, these results indicate that ATL_5802 increases phagocytosis of toxic tau aggregates by inflammatory microglia, thus contributing to effective clearance of these aggregates. This is expected to reduce tau-associated pathology and reduce and/or prevent (further) neurodegeneration.

Example 12 - Cytokine release from different cells

CD33 is expressed on microglia but also on peripheral myeloid cells such as monocytes. Modulation of these immune cells can cause systemic release of cytokines with potentially undesirable consequences and safety concerns.

To test whether the anti-CD33 antibodies described herein cause undesirable release of cytokines, iPSC microglia, PBMCs and isolated monocytes from multiple donors were treated with isotype control antibody or ATL5802 and cell media was assayed for multiple different cytokines, chemokines, and inflammatory mediators, as set out in Table 7.

Table 7. Levels of cytokines were measured in the supernatants of iPSC microglia, monocytes and PBMC treated with isotype control antibody or ATL_5802. NS= no significant difference between antifluorescein isotype control antibody and ATL5802 treated cells (t-test). NA= Not applicable because not tested for that cell type. These results indicate that treatment with ATL5802 did not cause immune-mediated cytokine release. In particular, ATL_5802 did not cause release of MCP-1 by untreated microglia after 24 hours of treatment with 100microg/ml or 10microg/ml (in fact showing levels slightly below the isotype control), a benefit that was not shared by the comparative anti-CD33 antibody tested.

LPS treatment of iPSC microglia is known to lead to the production of inflammatory cytokines MCP-1 and IL-6. To test whether the anti-CD33 antibody ATL-5802 was able to reduce the levels of these pro- inflammatory cytokines, microglia were treated with LPS, followed by treatment with ATL-5802 or an isotype control antibody. After 24 hours media was harvested and cytokine levels were measured using a multiplex immunoassay.

The results are shown in Figure 20 and indicate that treatment with ATL_5802 significantly reduces LPS induced production of IL-6 (Figure 20A) and MCP-1 (Figure 20B) by iPSC microglia in response to LPS. This suggests that anti-CD33 antibodies described herein are capable of having an anti-inflammatory effect in human iPSC microglia. This as significantly different to the isotype control antibody and comparative anti- CD33 antibody ATL_5909.

MCP-1 recruits immune cells to the site of inflammation and is implicated in the pathogenesis of neuroinflammation and AD, autoimmune disease (e.g. arthritis) and artherosclerosis. It is therefore very beneficial that the antibodies of the disclosure do not lead to cytokine release including MCP-1 , a benefit that was not present in the comparative anti-CD33 antibody ATL_5909.

Together, these data show that the anti-CD33 antibodies described herein do not cause any undesirable cytokine release, suggesting they are safe for human use, and have potent anti-inflammatory effects in vitro.

Example 13 - Intracellular signalling changes in microglial cells following ATL 5802 addition

Spleen tyrosine kinase (SYK) is a key intracellular regulator in microglial activation and phagocytosis. It is particularly significant in neurodegenerative diseases where it assists in the clearance of toxic protein aggregates (Ennerfelt et al. 2022). In this example, the inventors investigated the alterations in intracellular signalling following CD33 engagement, with a specific focus on the phosphorylation state of Syk (pSYK) under inflamed conditions.

Activation of ITAM receptors expressed on microglia cells, such as TREM2, results in recruitment and phosphorylation of ITAM-containing adapter molecules, which in turn recruit SYK. Upon activation, phosphorylation of SYK results in upregulation of cytokine production, phagocytosis and ROS production in microglia (Linnartz and Neumann, 2013). TREM2 is antagonised by inhibitory signalling from ITIM receptors such as CD33. CD33 signalling through ITIM domains recruits phosphatases that counteract ITAM signalling by dephosphorylation of the ITAM domains as well as ITAM-associated kinases such as SYK (Huang et al., 2003).

Here, the inventors sought to investigate whether the anti-CD33 antibodies described herein are capable of modulating intracellular signalling following CD33 engagement with ITIM, and in particular whether the antibodies described herein are able to relieve the CD33-mediated repression of ITAM receptor signalling. To test this, the inventors measured the phosphorylation state of Syk (pSYK) under inflamed conditions. Increased phosphorylation would indicate increased activation of the ITAM-signalling pathway and thus inhibition of the suppressive effect of CD33.

Figure 21 shows the results of a western blot of microglia cells treated with LPS and ATL_5802 or an isotype control antibody. This revealed a notable increase in SYK phosphorylation following ATL_5802 treatment, while the overall levels of SYK protein remained stable. These findings imply that ATL_5802 modulates signalling pathways possibly resulting in de-repression of the TREM2 signalling axis and leading to enhanced TREM2 signalling, as evidenced by the increased levels of pSYK. This underscores the potential therapeutic value of targeting these specific pathways.

Next, the inventors investigated whether anti-CD33 antibodies described herein, such as ATL_5802, have an effect on the purinergic receptor P2RY12. P2RY12 is a member of the P2 purinergic family of receptors, a seven transmembrane-spanning G protein-coupled receptor that responds to ADP/ATP, and is associated with chemotaxis and phagocytosis. P2RY12 promotes microglial chemotaxis towards sites characterised by necrotic or apoptotic cells, which is important for maintaining brain health and responses to injury or inflammation (Walker et al., 2020). The chemotactic response mediated by P2RY12 is a fundamental aspect of the microglial surveillance mechanism, allowing for the early detection and response to brain damage. Activation of P2RY12 enhances the phagocytic capability of microglia, promoting the clearance of apoptotic cells, amyloid-beta plaques, and other debris associated with neurodegenerative diseases. Through its involvement in phagocytosis, P2RY12 contributes to the resolution of inflammation and the prevention of further tissue damage. By modulating P2RY12 activity, it may be possible to enhance beneficial microglial functions while mitigating harmful inflammation, thus protecting against neurodegeneration and supporting brain health.

To test whether ATL_5802 is able to enhance P2RY12 expression in microglia a western blot was carried out using LPS-stimulated iPSC-induced microglia treated with ATL_5802 or isotype control antibody.

Figure 22 shows the results of a western blot for P2RY12 in microglia 6 hours after the addition of ATL_5802 or isotype control antibody. ATL_5802 enhances the protein expression of the P2RY12 receptor, suggesting the antibody is able to modulate intracellular signalling pathways in microglia important for responding to injury and inflammation. In particular, the data on Figure 22 shows that non-inflamed control conditions have high P2RY12 expression whereas inflamed control conditions have low P2RY12 expression. By contrast, after treatment with ATL_5802, non-inflamed conditions have higher P2RY12 expression compared to isotype control (indicative of the microglia having returned to a surveillance state, characterized by high P2RY12 expression. In this state, microglia are not actively engaged in inflammatory responses but are ready to respond to new damage or threats), and inflamed conditions show higher P2RY12 expression compared to isotype control.

This demonstrates that ATL_5802 has the ability to enhance the protein expression of the P2RY12 receptor, which is indicative of non-activated microglia. This effect guides microglia towards specific destinations, playing a vital role in upholding brain health and orchestrating responses to injury or inflammation. This change in signaling dynamics could result in a more effective immune response from microglia, a crucial factor in slowing the progression of neurodegenerative diseases.

Together, these data indicate that the anti-CD33 antibodies described herein are able to drive intracellular signalling pathways in microglia important for their effective function, such as phagocytosis, cytokine release and chemotaxis, while inhibiting signalling pathways involved in suppressing microglial activation and function. This change in signalling dynamics may contribute to a more effective immune response, a crucial factor in slowing the progression of neurodegenerative diseases.

To understand the transcriptional changes resulting from treatment with anti-CD33 antibodies both in healthy and inflammation-mimicking conditions, a genome-wide RNA sequencing study was carried out on iPSC-derived microglia. Briefly, iPSC-derived microglia were stimulated with vehicle, or with LPS/interferon gamma (LI), to resemble inflammatory conditions, followed by treatment with ATL_5802, ATL_5909, ATL_5854, or isotype control antibody followed by RNA sequencing as described in the “Materials and Methods” section set out above.

The results of the analysis following RNA sequencing are shown in Figure 23. The PCA plot shown in Figure 23A indicates that treatment with ATL_5802 leads to greater changes in gene expression than those driven by ATL_5854, ATL_5909 or ATL_5338 (isotype).

The comparison of the expression profiles of samples treated with any anti-CD33 antibody and the isotype control in either LI- or vehicle-stimulated cells shows a number of differentially expressed genes (DEG) (Figure 23 B-C) The greatest number of differentially expressed genes were detected in iPSC-derived microglia treated with vehicle + ATL_5802 versus those treated with vehicle + isotype control, as shown in Figures 23B and C, indicating an effect of ATL_5802 in healthy cells. A total of 194 genes were upregulated in vehicle + ATL_5802 versus vehicle + isotype control (see Figure 23D).

Interestingly, pathway analysis revealed downregulation with high statistical significance of several gene sets involved in oxidative phosphorylation (OXPHOS) upon treatment with ATL_5802 in inflammatory microglia in comparison to LI and isotype treated samples (Figure 23E-F). This result reveals a differentiating mechanism of action for ATL_0005802, whereby transcription of oxidative phosphorylation and respiratory electron transport components is dampened.

Oxidative phosphorylation (OXPHOS) in microglia enables efficient ATP production under homeostatic conditions, whereas activation under inflammatory conditions is reliant on PI3K/mTOR/HIF1 a-dependent aerobic glycolysis to produce ATP faster, akin to the Warburg effect seen in tumours (Laura et al., 2020). This in turn results in the microglia adopting a phagocytic phenotype. Microglia continuously monitor the cerebral parenchyma to detect neuronal damage and alteration of homeostatic processes, and defects in cellular metabolism involving both glycolysis and OXPHOS have been implicated in neurodegenerative diseases such as Alzheimer’s disease (Baik et al., 2019). Boosting of the anaerobic glycolytic metabolism has been shown to restore phagocytic activity of microglia and improves cognitive impairment in a mouse model of AD (Baik et al., 2019). While these experiments only show the decrease in OXPHOS (but not the corresponding glycolytic boost), inhibition of CD33 would inhibit SHI P112, negatively regulate PI3K, activate mTOR and hence boots glycolysis, which would be expected to have beneficial cognitive improvement.

The result that ATL_5802 reduces transcription of OXPHOS and respiratory electron transport components therefore suggest that ATL_5802 may boost anaerobic glycolysis and thereby dampen the OXPHOS pathway. With decreased OXPHOS, microglia cells undergoing CD33 antagonism by ATL_5802 will have a beneficial decrease in the release of reactive oxygen species, a by-product of OXPHOS.

Together, these data suggest that ATL_5802 is able to modulate cellular metabolic pathways, ultimately leading to improved cognitive function in neurodegenerative diseases.

Example 15 - Quad-culture with Cortical Neurons, Astrocytes and Microglia

Next, the inventors developed a human neural cell culture referred to as “CNS quad-culture platform” (Figure 24A). This platform integrates a diverse set of cell types - specifically, glutamatergic and GABAergic neurons, along with microglia and astrocytes - co-cultured to accurately replicate the complex intercellular dynamics of the human brain in a controlled in vitro environment. This in vitro approach allows a deeper understanding of the interactions between various brain cell types and acts as a critical experimental intermediary for a human-relevant system to assess the effectiveness of ATL_5802 on markers of inflammation and astrogliosis. To assess this, lipopolysaccharide (LPS) and Interferon-gamma (INFy) were introduced into the system to induce an inflammatory response, simulating conditions of neuroinflammation, followed by analysis of inflammation, astrogliosis, Glia dysfunction, and microglia activation.

ATL_5802 significantly reduced LPS/INFy-induced interleukin-6 (IL-6) levels relative to isotype control (Figure 24B). Additionally, ATL_5802 led to a significant reduction in the astrogliosis marker, GFAP (Figure 24C), further suggesting an overall anti-inflammatory effect of ATL_5802.

Interferon gamma-induced protein 10 (IP-10) is a marker of microglia activation and mediates the initiation of neuroinflammatory processes. MCP-1 is a marker of glia dysfunction and high CSF MCP-1 levels are linked to the brain atrophy and cognitive impairment in AD. These are therefore key chemokines known to be elevated in AD. ATL_5802 reduced both IP-10 (Figure 24D) and MCP-1 levels (Figure 24E) significantly relative to the isotype control.

These results provide further evidence supporting the therapeutic potential of ATL_5802 in reducing neuroinflammation, thereby protecting against brain atrophy and cognitive impairment in neurodegenerative diseases.

Example 16 - Selectivity of ATL 5802

To assess selectivity of ATL_5802, a screen was performed for binding to fixed HEK293 cells expressing 6105 individual human proteins, as well as a further 400 human heterodimers. The screening library includes plasma membrane, secreted and cell surface-tethered human proteins. Screening was performed with the operators blinded to the target of ATL_5802, CD33. The test antibody ATL_5802 showed a single significant specific interaction with CD33 on both fixed and live cell microarrays (see Table 8). These data confirm that ATL_5802 is selective for CD33 and therefore unlikely to cross-react with other targets.

Table 8. Results of a microarray binding screen for ATL5802 confirming selective binding to CD33.

References

A number of publications are cited above in order to more fully describe and disclose the invention and the state of the art to which the invention pertains. Full citations for these references are provided below. The entirety of each of these references is incorporated herein. For standard molecular biology techniques, see Sambrook, J., Russel, D.W. Molecular Cloning, A Laboratory Manual. 3 ed. 2001 , Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press.

Amft, M., Ortner, M., Eichenlaub, U. Et al. The cerebrospinal fluid biomarker ratio A042/4O identifies amyloid positron emission tomography positivity better than Ap42 alone in a heterogeneous memory clinic cohort. Alz Res Therapy 14, 60 (2022).

Andrews, S. (2010). FastQC: A Quality Control Tool for High Throughput Sequence Data: www.bioinformatics.babraham.ac.uk/projects/fastqc/

Baik, Sung Hoon et al. “A Breakdown in Metabolic Reprogramming Causes Microglia Dysfunction in Alzheimer's Disease.” Cell metabolism vol. 30,3 (2019): 493-507.

Balsis, Steve et al. “How Do Scores on the ADAS-Cog, MMSE, and CDR-SOB Correspond?.” The Clinical neuropsychologist vol. 29,7 (2015): 1002-9.

Blennow, K., Shaw, L.M., Stomrud, E. et al. Predicting clinical decline and conversion to Alzheimer’s disease or dementia using novel Elecsys Ap(1-42), pTau and tTau CSF immunoassays. Sci Rep 9, 19024 (2019).

Bonnet D, Dick JE. Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nat Med. (1997) 3:730.

Brinkman-Van der Linden EC, Angata T, Reynolds SA, Powell LD, Hedrick SM, Varki A. CD33/Siglec-3 binding specificity, expression pattern, and consequences of gene deletion in mice. Mol Cell Biol. 2003 Jun;23(12):4199-206.

Cao, H., de Bono, B., Belov, K. et al. Comparative genomics indicates the mammalian CD33rSiglec locus evolved by an ancient large-scale inverse duplication and suggests all Siglecs share a common ancestral region. Immunogenetics 61 , 401-417 (2009).

Cao, Huan, and Paul R Crocker. “Evolution of CD33-related siglecs: regulating host immune functions and escaping pathogen exploitation?.” Immunology vol. 132,1 (2011): 18-26.

Chan G, et al.: CD33 modulates TREM2: convergence of Alzheimer loci. Nat Neurosci 2015; 18: 1556- 1558 Shifu Chen, Yanqing Zhou, Yaru Chen, Jia Gu; fastp: an ultra-fast all-in-one FASTQ preprocessor, Bioinformatics, Volume 34, Issue 17, 1 September 2018, Pages i884— i890,

Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, Gingeras TR. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013 Jan 1 ;29(1 ): 15-21 .

Dolgalev I (2022). _msigdbr: MSigDB Gene Sets for Multiple Organisms in a Tidy Data Format_. R package version 7.5.1 , <https://igordot.github.io/msigdbr/>.

Donnelly, Louise E, and Peter J Barnes. “Defective phagocytosis in airways disease.” Chest vol. 141 ,4 (2012): 1055-1062.

Duff, Kevin et al. “Utility of the RBANS in detecting cognitive impairment associated with Alzheimer’s disease: sensitivity, specificity, and positive and negative predictive powers.” Archives of clinical neuropsychology : the official journal of the National Academy of Neuropsychologists vol. 23,5 (2008): 603-12.

Ennerfelt H, et al. SYK coordinates neuroprotective microglial responses in neurodegenerative disease. Cell. 2022 Oct 27;185(22):4135-4152.e22.

Eskandari-Sedighi, Ghazaleh et al. “CD33 isoforms in microglia and Alzheimer’s disease: Friend and foe.” Molecular aspects of medicine vol. 90 (2023): 101111.

Fattorelli N, Martinez-Muriana A, Wolfs L, Geric I, De Strooper B, Mancuso R. Stem-cell-derived human microglia transplanted into mouse brain to study human disease. Nat Protoc. 2021 Feb;16(2):1013-1033.

Golde, Todd E. “Disease-Modifying Therapies for Alzheimer’s Disease: More Questions than Answers.” Neurotherapeutics : the journal of the American Society for Experimental NeuroTherapeutics vol. 19,1 (2022): 209-227.

Griciuc, Ana et al. “Alzheimer’s disease risk gene CD33 inhibits microglial uptake of amyloid beta.” Neuron vol. 78,4 (2013): 631-43.

Holstege H, Beker N, Dijkstra T, et al. The 100-plus Study of cognitively healthy centenarians: rationale, design and cohort description. Eur J Epidemiol. 2018;33(12):1229-1249.

Hollingworth, Paul et al. “Common variants at ABCA7, MS4A6A/MS4A4E, EPHA1 , CD33 and CD2AP are associated with Alzheimer’s disease.” Nature genetics vol. 43,5 (2011): 429-35.

Huang, Zhen-Yu et al. “The effect of phosphatases SHP-1 and SHIP-1 on signaling by the ITIM- and ITAM-containing Fcgamma receptors FcgammaRIIB and FcgammaRIIA.” Journal of leukocyte biology vol. 73,6 (2003): 823-9.

Jack CR Jr, Bennett DA, Blennow K, et al. A/T/N: An unbiased descriptive classification scheme for Alzheimer disease biomarkers. Neurology. 2016;87(5):539-547.

Jung et al. Selection for improved protein stability by phage display. Journal Molecular Biology 294, 163- 180 (1999).

Jurcic, 2012. What Happened to Anti-CD33 Therapy for Acute Myeloid Leukemia? Curr Hematol Malig Rep (2012) 7:65-73.

Knopman, D.S., Amieva, H., Petersen, R.C. et al. Alzheimer disease. Nat Rev Dis Primers 7, 33 (2021).

G. Korotkevich, V. Sukhov, A. Sergushichev. Fast gene set enrichment analysis. bioRxiv (2019), doi:10.1101/060012

Lauro, Clotilde, and Cristina Limatola. “Metabolic Reprograming of Microglia in the Regulation of the Innate Inflammatory Response.” Frontiers in immunology vol. 11 493. 20 Mar. 2020.

Laszlo et al. The past and future of CD33 as therapeutic target in acute myeloid leukemia. Blood Rev. 2014 Jul.

Liao Y, Smyth GK and Shi W (2014). featurecounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics, 30(7):923-30. Linnartz, Bettina, and Harald Neumann. “Microglial activatory (immunoreceptor tyrosine-based activation motif)- and inhibitory (immunoreceptor tyrosine-based inhibition motif)-signaling receptors for recognition of the neuronal glycocalyx.” Glia vol. 61 ,1 (2013): 37-46.

Love, M.I., Huber, W., Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2 Genome Biology 15(12):550 (2014)

Lund J, Winter G, Jones PT, Pound JD, Tanaka T, Walker MR et al. Human Fc gamma Rl and Fc gamma RII interact with distinct but overlapping sites on human IgG. J Immunol. 1991 ; 147:2657-2662

Mancuso R, et al. Stem-cell-derived human microglia transplanted in mouse brain to study human disease. Nat Neurosci. 2019 Dec;22(12):2111-2116.

Manscuso R., et al. A multi-pronged human microglia response to Alzheimer’s disease Ab pathology.

Preprint at bioRxiv. 2022. doi: 10.1101/2022.07.07.499139.

Michaelson DM. APOE e4: the most prevalent yet understudied risk factor for Alzheimer’s disease.

Alzheimers Dement. 2014;10(6):861-868.

McQuade, A., Kang, Y.J., Hasselmann, J. Et al. Gene expression and functional deficits underlie TREM2- knockout microglia responses in human models of Alzheimer’s disease. Nat Commun 11 , 5370 (2020).

Mancuso, Renzo et al. “Stem-cell-derived human microglia transplanted in mouse brain to study human disease.” Nature neuroscience vol. 22,12 (2019)

Siddiqui SS, Springer SA, Verhagen A, Sundaramurthy V, Alisson-Silva F, Jiang W, Ghosh P, Varki A: The Alzheimer’s disease-protective CD33 splice variant mediates adaptive loss of function via diversion to an intracellular pool. J Biol Chem 2017; 292: 15312-15320.

Stanczak MA et al. Targeting cancer glycosylation repolarizes tumor-associated macrophages allowing effective immune checkpoint blockade. Science Translational Medicine 14, eabj1270 (2022) 2 November 2022.

Stanczak, Michal A, and Heinz Laubli. “Siglec receptors as new immune checkpoints in cancer.” Molecular aspects of medicine vol. 90 (2023): 101 112.

Strang KH, et al. Distinct differences in prion-like seeding and aggregation between Tau protein variants provide mechanistic insights into tauopathies. J Biol Chem. 2018 Feb 16;293(7):2408-2421 . doi: 10.1074/jbc.M117.815357. Epub 2017 Dec 19. Erratum in: J Biol Chem. 2018 Mar 23;293(12 ):4579.

Trinidad Hernandez-Caselles, et al., A study of CD33 (SIGLEC-3) antigen expression and function on activated human T and NK cells: two isoforms of CD33 are generated by alternative splicing, Journal of Leukocyte Biology, Volume 79, Issue 1 , Jan 2006, Pages 46-58

Walker, D.G.; et al. Patterns of Expression of Purinergic Receptor P2RY12, a Putative Marker for Non- Activated Microglia, in Aged and Alzheimer’s Disease Brains. Int. J. Mol. Sci. 2020, 21 , 678.

Ward et al., 2021. A phase I study of AL003 in Healthy Volunteers and Participants with Alzheimer’s Disease. Abstract P45 in “Abstract: Symposia, Conferences, Oral communications: 14^th Clinical Trials on Alzheimer’s Disease (CTAD) November 9-12, 2021.” The journal of prevention of Alzheimer’s disease vol. 8,S1 (2021): S2-S72.

Varki A, Angata T: Siglecs - the major subfamily of l-type lectins. Glycobiology 2006; 16: 1 R-27R.

Vercauteren S, Zapf R, Sutherland H. Primitive AML progenitors from most CD34+ patients lack CD33 expression but progenitors from many CD34- AML patients express CD33. Cytotherapy (2007) 9:194- 204.

Yu YJ, Atwal JK, Zhang Y, et al. Therapeutic bispecific antibodies cross the blood-brain barrier in nonhuman primates. Sci Transl Med. 2014;6(261):261 ra154.

Zhao, Lingzhi. “CD33 in Alzheimer’s Disease - Biology, Pathogenesis, and Therapeutics: A Mini-Review. Gerontology vol. 65,4 (2019): 323-331.

Claims

Claims:

1 . An isolated antibody which specifically binds to a CD33 protein, wherein the antibody increases phagocytosis of a cell expressing CD33 compared to a comparative antibody and/or wherein the antibody binding to a human CD33 protein comprising mutations at positions 20, 21 , 22 and 24 is increased compared to the antibody binding to a human CD33 protein without said mutations.

2. The isolated antibody of claim 1 , wherein the antibody binding to a human CD33 protein comprising mutations at positions 20, 21 , 22, 24 and 132 is increased compared to the antibody binding to a human CD33 protein without said mutations, and/or wherein the antibody binding to a human CD33 protein comprising mutations at positions 47, 50, 51 and 52 is reduced compared to the antibody binding to a human CD33 protein without said mutations, and/or wherein the antibody does not bind to a human CD33 protein comprising mutations at positions 47, 50, 51 and 52, and/or wherein the antibody binding to a human CD33 protein comprising mutations at positions 47, 50, 51 , 52 and 122 is reduced compared to the antibody binding to a human CD33 protein without said mutations, and/or wherein the antibody does not bind to a human CD33 protein comprising mutations at positions 47, 50, 51 , 52 and 122, and/or wherein the antibody binding to a human CD33 protein comprising mutations at positions 83 is reduced compared to the antibody binding to a human CD33 protein without said mutations.

3. The isolated antibody of claim 1 or claim 2, wherein the mutations are selected from: a. at position 20: N20R, b. at position 21 : F21V, c. at position 22: W22R, d. at position 24: Q24E, e. at position 47: I47V, f. at position 50: Y50H, g. at position 51 : D51T, h. at position 52: K52R, i. at position 83: Q83R, j. at position 122: R122K, and k. at position 132: P132T.

4. The isolated antibody of any preceding claim, wherein the binding is as measured using a single point ELISA and/or the human CD33 protein comprises residues 18-232 of human CD33, and/or the human CD33 protein is CD33M2_ECD_18-232_WT.

5. The isolated antibody of any preceding claim, wherein the antibody has one or more, or all of: a. increased binding to a protein comprising the sequence of CD33M2_ECD_18- 232_MutPos1 [P1] compared to a protein comprising the sequence of CD33M2_ECD_18- 232_WT, b. increased binding to a protein comprising the sequence of CD33M2_ECD_18- 232_MutPos1_MutPos6 [P6+1] compared to a protein comprising the sequence of CD33M2_ECD_18-232_WT, c. reduced binding to a protein comprising the sequence of CD33M2_ECD_18- 232_MutPos2 [P2] compared to a protein comprising the sequence of CD33M2_ECD_18- 232JA/T, d. reduced binding to a protein comprising the sequence of CD33M2_ECD_18- 232_MutPos2_MutPos5 [P2+5] compared to a protein comprising the sequence of CD33M2_ECD_18-232_WT, and e. reduced binding to a protein comprising the sequence of CD33M2_ECD_18- 232_MutPos4 [P4] compared to a protein comprising the sequence of CD33M2_ECD_18- 232JA/T.

6. The isolated antibody of any preceding claim, wherein the comparative antibody is selected from: an isotype control antibody, another CD33-binding antibody, and an antibody with the heavy chain variable sequence of ATL_5909 and the light chain variable sequence of ATL_5909, and/or wherein the phagocytosis is assessed by measuring a fluorescence signal associated with uptake of a labelled particle by imaging or flow cytometry.

7. The isolated antibody of any preceding claim, wherein the cell is a monocyte or a microglial cell, optionally wherein the cell is a human cell and/or an iPSC derived microglial cell and/or a cell that has been stimulated with an inflammatory signal (e.g. LPS) prior to exposure to the antibody.

8. The isolated antibody of any preceding claim, wherein the antibody binds to a CD33 protein that comprises the V domain of CD33, and/or wherein the antibody does not bind to a CD33 protein that does not comprise the V domain of CD33, and/or wherein the antibody does not bind to a CD33 protein that has the sequence of protein CD33_HUMAN_ECD_Cdomain_His_007.

9. The isolated antibody of any preceding claim, wherein the antibody comprises a heavy chain variable domain (VH) with the following CDRs:

CDRH1 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810;

CDRH2 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810;

CDRH3 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810; or a set of CDRs which contains zero, one or two amino acid substitutions in each CDR compared with the above set of CDRs.

10. The isolated antibody according to claim 9, wherein the antibody comprises a heavy chain variable domain (VH) with the following CDRs: a CDRH1 comprising an amino acid sequence selected from:

HCDR1_ATL_0005802; HCDR1_ATL_0005807; HCDR1_ATL_0005808: GYSFTSYW (SEQ ID NO:44)

HCDR1_ATL_0005853: GYKFNNNW (SEQ ID NO:47)

HCDR1_ATL_0005854: GYKFSNNW (SEQ ID NO:48) or an amino acid sequence with 1 or 2 mutations compared to the above sequences; a CDRH2 comprising an amino acid sequence selected from:

HCDR2_ATL_0005802; HCDR2_ATL_0005807; HCDR2_ATL_0005808; HCDR2_ATL_0005853;

HCDR2_ATL_0005854; IYPGDSDT (SEQ ID NO: 52) or an amino acid sequence with 1 or 2 mutations compared to the above sequences; and a CDRH3 comprising an amino acid sequence selected from:

HCDR3_ATL_0005802; ARPRGFGEYYFDY (SEQ ID NO: 59)

HCDR3_ATL_0005853 ARHSGGLDGYTAAALDY (SEQ ID NO:64)

HCDR3_ATL_0005854 ATWGGSNWFVD (SEQ ID NO:65) or an amino acid sequence with 1 or 2 mutations compared to the above sequences.

11. The isolated antibody according to claim 9 or claim 10, wherein the antibody comprises a heavy chain variable domain (VH) with the following CDRs: a CDRH1 comprising the sequence of HCDR1_ATL_0005802, a CRH2 comprising the sequence of HCDR2_ATL_0005802, and a CDRH3 comprising the sequence of HCDR3_ATL_0005802, or a set of CDRs comprising 1 or 2 mutations in the CDRH1 and CDRH2 compared to these sequences and/or 1 , 2 or 3 mutations in the CDRH3 compared to these sequences; or a CDRH1 comprising the sequence of HCDR1_ATL_0005853, a CRH2 comprising the sequence of HCDR2_ATL_0005853, and a CDRH3 comprising the sequence of HCDR3_ATL_0005853, or a set of CDRs comprising 1 or 2 mutations in the CDRH1 and CDRH2 compared to these sequences and/or 1 , 2 or 3 mutations in the CDRH3 compared to these sequences; or a CDRH1 comprising the sequence of HCDR1_ATL_0005854, a CRH2 comprising the sequence of HCDR2_ATL_0005854, and a CDRH3 comprising the sequence of HCDR3_ATL_0005854, or a set of CDRs comprising 1 or 2 mutations in the CDRH1 and CDRH2 compared to these sequences and/or 1 , 2 or 3 mutations in the CDRH3 compared to these sequences.

12. The isolated antibody according to any preceding claim, wherein the antibody has a heavy chain variable domain (VH) with the following framework sequences:

HFWR1 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807,

ATL_0005808, ATL_0005809, ATL_0005810,

HFWR2 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807,

ATL_0005808, ATL_0005809, ATL_0005810,

HFWR3 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807,

ATL_0005808, ATL_0005809, ATL_0005810, and ATL_0005808, ATL_0005809, ATL_0005810, or framework sequences with one to six substitutions compared to the framework sequences above.

13. The isolated antibody according to any preceding claim, wherein the antibody has a heavy chain variable domain (VH) with a framework sequence HFWR2 of ATL_0005802 and/or wherein the antibody has a heavy chain variable domain (VH) with a framework sequence comprising a A at position 40 in standard IMGT numbering.

14. The isolated antibody according to claim 12 or claim 13, wherein the substitutions in the framework sequences of the heavy chain variable domain are located at any position other than position 40 in standard IMGT numbering.

15. The isolated antibody according to any preceding claim, wherein the antibody has a heavy chain variable domain (VH) with the following framework sequences: HFWR1 of ATL_0005802, ATL_0005853 or ATL_0005854; HFWR2 of ATL_0005802, ATL_0005853 or ATL_0005854; HFWR3 of ATL_0005802, ATL_0005853 or ATL_0005854; and HFWR4 of ATL_0005802, ATL_0005853 or ATL_0005854.

16. The isolated antibody according to any preceding claim, wherein the antibody has a heavy chain variable domain (VH) comprising a sequence that has a least 95% sequence identity with a sequence selected from the VH sequence of antibodies: ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, ATL_0006040

ATL_0006041 , ATL_0006042, ATL_0006043, ATL_0006044, ATL_0006045, ATL_0006046, ATL_0006047, and ATL_0006048, or wherein the antibody has a heavy chain variable domain (VH) comprising a sequence that has at most 2 mutations in each HCDR and at most 3 mutations in each framework region compared with a sequence selected from said VH sequences; optionally wherein the antibody has a heavy chain variable domain (VH) comprising a sequence that has a least 95% sequence identity with a sequence selected from the VH sequence of antibodies ATL_0005802, ATL_0005853, or ATL_0005854 or a sequence that has at most 2 mutations in each HCDR and at most 3 mutations in each framework region compared with a sequence selected from said VH sequences.

17. The isolated antibody of any preceding claim, wherein the antibody: (i) has a heavy chain variable domain (VH) comprising CDRH1 , CDRH2, and CDRH3 within a germline framework, provided that position 40 in standard IMGT numbering is A, and/or (ii) is an scFv antibody molecule, a nanobody, or a whole antibody, and/or wherein the antibody comprises an antibody constant region, and/or wherein the antibody is a whole antibody and/or wherein the antibody is an IgG 1 or variant thereof, optionally wherein the antibody is an lgG1 variant L234A/L235A (LALA).

18. The isolated antibody of any preceding claim, wherein the antibody binds human CD33, optionally wherein the antibody binds human CD33 with an EC50 of at most 2e-08 M, or at most 3e-09 M, as assessed by ELISA (such as binding of plated rhCD33), and/or wherein the antibody selectively binds to CD33 over other one or more siglecs, optionally wherein the antibody selectively binds to CD33 over one or more (or all of): siglec-6, siglec-7, siglec-8, and siglec-9, and/or wherein the antibody selectively binds to human CD33 over other one or more homologs, optionally wherein the antibody selectively binds to human CD33 over mouse CD33 and cyno CD33.

19. The isolated antibody of any preceding claim, wherein the antibody depletes CD33 on the cell surface of human monocytes by less than 50%, or less than 80% after 5 hours of incubation with the antibody, and/or wherein the antibody depletes CD33 on the cell surface of human monocytes after 5 hours of incubation with the antibody to a lower extent than a comparative antibody at the same concentration.

20. The isolated antibody according to any of the preceding claims, wherein the antibody comprises a light chain variable domain (VL) with the following CDRs:

CDRL1 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810,

CDRL2 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, and

CDRL3 comprising the amino acid sequence of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, or a set of CDRs which contains zero, one, or two amino acid substitutions in each CDR compared with the above set of CDRs.

21. The isolated antibody according to claim 20, wherein the CDRL1 , CDRL2 and CDRL3 of the VL domain are within a germline framework, and/or wherein the antibody has a light chain variable domain (VL) with the following framework sequences:

LFWR1 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810;

LFWR2 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810;

LFWR3 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810; and

LFWR4 of any of antibodies ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810; or a set of FWRs which contains one to six amino acid substitutions compared with the above set of FWR.

22. The isolated antibody according to any preceding claim, wherein the antibody has a light chain variable domain (VL) comprising a sequence selected that has a least 95% sequence identity with a sequence selected from the VL sequence of antibodies: ATL_0005802, ATL_0005853, ATL_0005854, ATL_0005807, ATL_0005808, ATL_0005809, ATL_0005810, ATL_0006040, ATL_0006041 , ATL_0006042, ATL_0006043, ATL_0006044, ATL_0006045, ATL_0006046, ATL_0006047, and ATL_0006048, or wherein the antibody has a light chain variable domain (VL) comprising a sequence that has at most 2 mutations in each LCDR and at most 3 mutations in each framework region compared with a sequence selected from said VL sequences; optionally wherein the antibody has a light chain variable domain (VL) comprising a sequence that has a least 95% sequence identity with a sequence selected from the VL sequence of antibodies ATL_0005802, ATL_0005853, or ATL_0005854 or a sequence that has at most 2 mutations in each LCDR and at most 3 mutations in each framework region compared with a sequence selected from said VL sequences.

23. The isolated antibody according to any preceding claim, wherein the antibody has lower peripheral clearance when administered to a subject, compared to a comparative anti-CD33 antibody, and/or wherein the antibody increases phagocytosis of Ap by microglial cells in vivo compared to a control, and/or increases the phagocytosis of tau aggregates by microglia with inflammatory phenotype (e.g. LPS treated iPSC microglia) compared to a control, and/or increases the phagocytosis of tau aggregates by microglia with inflammatory phenotype to a larger extent than a comparative anti-CD33 antibody, and/or does not induce release of one or more cytokines including IL-6 and/or MCP-1 by microglia in vitro and/or in vivo, and/or reduce the levels of IL-6 and/or MCP-1 released by microglia with an inflammatory phenotype in vitro (e.g. LPS treated human iPSC derived microglia) and/or in vivo, and/or reduces inflammation induced release of one or more markers of inflammation in a human neural cell culture assay and/or in the central nervous system of a subject, optionally wherein the one or more markers of inflammation are selected from: MCP-1 , IP-10, GFAP and IL-6.

24. An isolated nucleic acid, vector or set of vectors which comprises a nucleotide sequence encoding an antibody, including a VH or VL domain, according to any preceding claim, or a fragment thereof.

25. A host cell comprising the vector(s) of claim 24, or a host cell in vitro transformed with a nucleic acid of claim 24.

26. An antibody according to any of claims 1 to 23, for use in the treatment of a disease or disorder selected from: (i) a disease associated with dysfunction of microglial cells; (ii) a neurodegenerative disease or disorder, optionally wherein the neurodegenerative disease or disorder is a tauopathy or a disease or disorder selected from: frontotemporal dementia (FTD), Alzheimer’s disease (AD), Huntington’s Disease (HD), Parkinson’s disease (PD), amyotrophic lateral sclerosis (ALS), human immunodeficiency virus (HlV)-induced encephalitis, Chronic traumatic encephalopathy (CTE), vascular dementia, prion diseases, Lewy body disease, Spinal muscular atrophy (SMA), Motor Neuron Disease (MND), such as amyotrophic lateral sclerosis (ALS), progressive supranuclear palsy (PSP), spinocerebellar ataxias (SCA) types 1 , 2, 6, 7 and 17, Machado-Joseph disease (MJD/SCA3), dentatorubral pallidoluysian atrophy (DRPLA), spinal bulbar muscular atrophy X-linked type 1 (SMAX1/SBMA), Anderson-Fabry (X-linked Fabry Disease), and DNAJB6 Myopathies, optionally wherein the neurodegenerative disease is selected from FTD, AD, HD, and PD; (iii) a cancer, optional wherein the cancer or AML or a cancer associated with hypersyalilation of tumour cells and/or overexpression of CD33 by tumour cells; and (iv) a disease characterised by insufficient macrophage phagocytosis and/or macrophage dysfunction, optionally wherein the disease is COPD or IPF.