WO2022167571A1

WO2022167571A1 - Treatment and method of identifying coronavirus therapeutics

Info

Publication number: WO2022167571A1
Application number: PCT/EP2022/052690
Authority: WO
Inventors: David Hoffmann; Stefan MEREITER; Josef Penninger; Gerald WIRNSBERGER
Original assignee: Imba - Institut Für Molekulare Biotechnologie Gmbh
Priority date: 2021-02-04
Filing date: 2022-02-04
Publication date: 2022-08-11

Abstract

The invention provides a polypeptide or nucleic acid for use in a method of treating a coronavirus infection comprising administering a polypeptide comprising a carbohydrate recognition domain of CLEC4G or a nucleic acid encoding said polypeptide to a patient suffering from an infection with a coronavirus expressing a SARS-CoV-2 Spike protein.

Description

Treatment and method of identifying coronavirus therapeutics

The present invention relates to the field of coronavirus therapeutics and the identification of therapeutic agents.

Background of the invention

COVID-19 caused by SARS-CoV-2 infections has triggered a pandemic massively disrupting health care, social and economic life. SARS-CoV-2 entry into target cells is mediated by the vi- ral Spike protein, which binds to angiotensin converting enzyme 2 (ACE2) expressed on host cells (Monteil et al. (2020) Cell 181, 905-913 e907). The Spike protein is divided into two subu- nits, SI and S2. The SI subunits comprises the receptor binding domain (RBD) which confers ACE2 binding activity. The S2 subunit mediates virus fusion with the cell wall following proteolytic cleavage (Hoffmann et al. (2020) Cell 181, 271-280 e278; Shang et al. (2020) Proc Natl Acad Sci USA 117, 11727-11734; Walls et al. (2020) Cell 181, 281-292 e286). Cryo-electron microscopy studies have shown that the Spike protein forms a highly flexi- ble homotrimer containing 22 N-glycosylation sites each (Walls et al., 2020, supra).

Glycosylation of viral proteins ensures proper folding and shields antigenic viral epitopes from immune recognition (Watanabe et al. (2019) Biochim Biophys Acta Gen Subj 1863, 1480-1497; Watanabe et al. (2020) Nat Commun 11, 2688). To cre- ate this glycan shield, the virus hijacks the host glycosylation machinery and thereby ensures the presentation of self-associ- ated glycan epitopes. Apart from shielding epitopes from anti- body recognition, glycans can be ligands for lectin receptors. For instance, mannose-specific mammalian lectins, like DC-SIGN (CD209) or its homolog L-SIGN (CD299), bind to viruses like HIV- 1 (Van Breedam et al. (2014) FEMS Microbiol Rev 38, 598-632). Lectin receptors are often expressed on immune and endothelial cells and serve as pattern recognition receptors involved in vi- rus internalization and transmission (Osorio et al. (2011) Im- munity 34, 651-664). Recent studies have characterized the recognition of the SARS-CoV-2 Spike by previously known virus- binding lectins, such as DC-SIGN, L-SIGN, MGL and MR (Gao et al. (2020) bioRxiv, doi.org/10.1101/2020.07.29.227462). Given that SARS-CoV-2 relies less on oligo-mannose-type glycosylation, as compared to for instance HIV-1, and displays more complex-type glycosylation, it is unknown if additional lectin receptors are capable of binding the Spike protein and whether such interac- tions might have functional relevance in SARS-CoV-2 infections.

There remains a need to find effective therapies for corona- virus infections, especially SARS-CoV-2 infections. It is there- fore a goal of the invention to find new therapies for corona- virus infections and in particular for COVID-19.

Summary of the invention

The present invention provides a polypeptide or nucleic acid for use in a method of treating a coronavirus infection compris- ing administering a polypeptide comprising a carbohydrate recog- nition domain of CLEC4G or a nucleic acid encoding said polypep- tide to a patient suffering from an infection with a coronavirus expressing a SARS-CoV-2 Spike protein.

Related thereto, the invention provides a method of treating a coronavirus infection comprising administering a polypeptide comprising a carbohydrate recognition domain of CLEC4G or a nu- cleic acid encoding said polypeptide to a patient suffering from an infection with a coronavirus expressing a SARS-CoV-2 Spike protein.

Also provided is a use of a polypeptide or nucleic acid for manufacturing a medicament for a treatment of a patient suffer- ing from an infection with a coronavirus expressing a SARS-CoV-2 Spike protein, the treatment comprising administering a polypep- tide comprising a carbohydrate recognition domain of CLEC4G or a nucleic acid encoding said polypeptide to a patient suffering from an infection with a corona-virus expressing a SARS-CoV-2 Spike protein.

Further provided is a library of at least 10 different poly- peptides, each polypeptide comprising a different carbohydrate recognition domain of a lectin and an immunoglobulin domain.

The invention also provides a set of nucleic acids encoding the polypeptides of the library.

Also provided is a method of identifying or screening a lec- tin candidate that binds to a carbohydrate-containing target of interest comprising contacting the target with the polypeptides of the library, and detecting polypeptides bound to said target.

All aspects of the invention are related and any disclosure of specific embodiments for one aspect also relate to other as- pects. E.g. a disclosure of polypeptides of the library equally relates to nucleic acids encoding said polypeptides. Such poly- peptides and nucleic acids can be used in the inventive methods.

Detailed description of the invention

Most viruses use glycosylation for pathogenesis and immune evasion with critical implications for therapies or vaccina- tions. Here we used a comprehensive library of mammalian carbo- hydrate-binding proteins (lectins) to probe critical sugar resi- dues on the full-length trimeric Spike and the receptor binding domain (RBD) of SARS-CoV-2. Two lectins, Clec4g and CD209, were identified to strongly bind to Spike. Clec4g and CD209 binding to Spike was dissected and visualized in real time and at single molecule resolution using atomic force microscopy. 3D modelling showed that Clec4g can bind to a glycan within the RBD-ACE2 in- terface and thus interferes with Spike binding to cell surfaces. Importantly, Clec4g and CD209 significantly reduced SARS-CoV-2 infections. The invention provides the first extensive map and 3D structural modelling of lectin-Spike interactions and uncover lectins critically involved in Spike binding and SARS-CoV-2 in- fections.

As Spike and RBD glycosylation affect ACE2 binding and SARS- CoV-2 infections, targeting virus specific glycosylation is a novel means for therapeutic intervention.

CLEC4G (C-Type Lectin Domain Family 4 Member G) is a lectin with its sequence being deposited at the UniProtKB database en- try Q6UXB4 (Entry version 129 of 2 December 2020, provided as SEQ ID NO: 2). As used herein, it refers to any form, variant or origin of the protein. CLEC4G is preferably a mammalian protein, preferably a human protein. Other CLEC4G proteins are from mouse, rat, hamster, cow, cat, non-human primate, monkey, pig, horse, dog, etc. Other names of CLEC4G according to gene- cards,org are: C-Type Lectin Domain Family 4 Member G; LSECtin; Liver And Lymph Node Sinusoidal Endothelial Cell; C-Type Lectin Superfamily 4, Member G; UNQ431; C-Type Lectin Domain Family 4, Member G.

A preferred CLEC4G protein is human CLEC4G according to all embodiments of the invention and is used as reference (Liu et al., JBC 279(18), 2004: 18748-18758). Human CLEC4G (SEQ ID NO: 2) comprises at amino acids (aa) 1-31 a cytoplasmic domain, a transmembrane domain at aa 32-52, and at aa 53-293 an extracel- lular domain. The carbohydrate recognition domain of human CLEC4G is within the extracellular domain and may comprise amino acids 165-288 of SEQ ID NO: 2.

According to the invention, a polypeptide comprising a car- bohydrate recognition domain of CLEC4G or a nucleic acid encod- ing said polypeptide is sufficient. Of course, the invention also includes variants thereof that are capable of binding SARS- CoV-2 Spike protein, such as the Spike protein of SEQ ID NO: 1. A preferred carbohydrate recognition domain of CLEC4G as used in the invention has at least 60% amino acid identity to the amino acids 165-288 of SEQ ID NO: 2 (human CLEC4G). In even more pre- ferred embodiments, the carbohydrate recognition domain of CLEC4G has at least 65%, more preferred at least 70%, at least 75%, at least 80%, at least 85%, at least 90%, at least 95% or at least 98%, or at least 99% amino acid identity to the amino acids 165-288 of SEQ ID NO: 2. As shown in example 20, mouse CLEC4G, also termed Clec4g, is effective in preventing SARS-CoV- 2 of human cells. Mouse CLEC4G is e.g. available in the UniProt database entry Q8BNX1 (entry version 128 of 2 December 2020, in- corporated herein by reference). The CRD of human CLEC4G has a sequence identity of 70.5% to the CRD of mouse CLEC4G, clearly illustrating that sequence variation while maintaining binding to SARS-CoV-2 Spike protein and protecting against SARS-CoV-2 is possible.

Sequence identities are calculated with regard to a refer- ence sequence, here in the above paragraph to the CRD of human CLEC4G of SEQ ID NO: 2. The compared test sequence (in the above example mouse CLEC4G) by aligning the compared test sequence to the refence sequence. Sequence identity is most preferably as- sessed by alignment with the BLAST version 2.1 program advanced search (parameters as above). BLAST is a series of programs that are available online at blast.ncbi.nlm.nih.gov/. The BLAST search may be set to de-fault parameters (i.e. Matrix BLOSUM62; Gap existence cost 11; Per residue gap cost 1; Lambda ratio 0.85 default). References to BLAST searches are: Altschul et al., J. Mol. Biol. (1990) 215: 403-410; Gish & States, Nature Genet.

(1993) 3: 266-272; Madden et al., Meth. Enzymol. (1996) 266: 131-141; Altschul et al. Nucleic Acids Res. (1997) 25: 3389- 3402. Identical amino acids in the alignment are counted and the number of identical amino acids over the length of the reference sequence give the ratio of identical amino acids, preferably ex- pressed in %.

Preferably, CLEC4G or the CRD of CLEC4G have a similar length as human CLEC4G (SEQ ID NO: 2) or the CRD of human CLEC4G (aa 165-288 of SEQ ID NO: 2), respectively. Preferably, there are at most 80, preferably at most 50, amino acid deletions or additions in the sequence of CLEC4G or the CRD of CLEC4G in com- parison to human CLEC4G or the CRD of CLEC4G. Amino acid addi- tions at the end, e.g. as in a fusion protein are irrespectively possible. Such fusion proteins include preferred embodiments as described further below, such as fusion with immunoglobulin do- mains. As such, this number refers to amino acids within the reference sequence while additions at the end of the reference sequence are not counted. Some deletions may be possible. Pref- erably, in the inventive CRD, about 1, 2, 3, 4, 5, 6 to 10, 11 to 25, amino acids of the CRD of amino acids 165-288 of SEQ ID NO: 2 are modified or deleted. Also here, this number refers to amino acids within the reference sequence while additions at the end of the reference sequence are not counted.

Non-identities in the amino acids are preferably conserva- tive changes or substantially conservative changes. Accordingly, the invention includes polypeptides having conservative changes or substitutions in amino acid sequences. Conservative amino acid substitutions insert one or more amino acids, which have similar chemical properties as the replaced amino acids. The in- vention includes sequences where conservative amino acid substi- tutions are made that do not abrogate binding to Spike glycoprotein of SARS-CoV-2.

"Conservative amino acid substitutions" are those substitu- tions that are predicted to interfere least with the properties of the reference polypeptide. In other words, conservative amino acid substitutions substantially conserve the structure and the function of the reference protein. The following Table provides a list of exemplary conservative amino acid substitutions:

Conservative amino acid substitutions generally maintain one or more of: (a) the structure of the polypeptide backbone in the area of the substitution, for example, as a beta sheet or alpha helical conformation, (b) the charge or hydrophobicity of the molecule at the site of the substitution, and/or (c) the bulk of the side chain. Preferably, at least 50% of the non-identical amino acids in the CRD to the CRD of human CLEC4G are conserva- tive amino acid substitutions.

Preferably, the polypeptide comprises CLEC4G or only an ex- tracellular domain of CLEC4G. Preferably, the polypeptide lacks the cytoplasmic domain of CLEC4G and/or the transmembrane domain of CLEC4G, in particular preferred lacks any cytoplasmic domain and/or any transmembrane domain. In further preferred embodiments, as alternative or combinable with the lack of a cy- toplasmic domain and/or of a transmembrane domain, only a part of a CLEC4G protein is used. Accordingly, the polypeptide may comprise 50 to 250 amino acids in length of CLEC4G. In preferred embodiments, the polypeptide may comprise 70 to 200, even more preferred 80 to 180 or 100 to 150, amino acids in length of CLEC4G.

Preferably, the polypeptide comprises a homolog of the CRD of human CLEG4G from any mammal, such as mouse, rat, hamster, cow, cat, non-human primate, monkey, pig, horse, dog, etc.

In preferred embodiments the polypeptide comprises an immu- noglobulin domain. More than one immunoglobulin domain is possi- ble. One or more immunoglobulin domains can be added e.g. as in a fusion protein. A fusion protein may comprise a linker, such as short peptide connecting different parts of the fusion pro- tein. E.g. the linker may be located between a part comprising the carbohydrate recognition domain of CLEC4G, and a part com- prising the immunoglobulin domain (s). Preferably, the immuno- globulin domain comprises or consists of antibody CHI, CH2 or CH3 domain, or combinations thereof.

The immunoglobulin domain may be of a human immunoglobulin domain. It may be of an IgG, such as an IgGl, IgG2, IgG3, IgG4, preferably IgGl, IgA, such as IgAl or IgA2, IgGD, IgE, IgM or combinations thereof. Preferably, the polypeptide comprises an antibody Fc fragment. An Fc fusion is preferably to an Fc of IgG, IgM, IgD, or IgA or a part thereof, such as a CHI, CH2 or CH3 domain, or FcRn. A CH3 domain is preferred. It may or may not include the C-terminus of the Fc part. IG is preferably hu- man IgGl, IgG2, and IgG4.

In further preferred embodiments the polypeptide is a dimer. Preferably, the CLEC4G is human CLEC4G and the patient is a human. Any part of human CLEC4G, in particular the CRD thereof, as described above may be used.

As used herein, the term "subject" may be used interchangea- bly with the term "patient" or "individual" and may include an "animal" and in particular a "mammal", that can be treated ac- cording to the invention. Mammalian subjects may include humans and non-human primates, domestic animals, farm animals, and com- panion animals such as dogs, cats, guinea pigs, rabbits, rats, mice, horses, cattle, cows, and the like. Preferably, the pa- tient to be treated by the inventive method and uses of the in- vention is a human, preferably a human adult of the age of 18 years or more.

The inventive polypeptide is used to treat an infection with a coronavirus expressing a SARS-CoV-2 Spike protein. This in- cludes treatment of COVID-19 and any infections with various strains and variants of SARS-CoV-2 or with any coronavirus that expresses a SARS-CoV-2 Spike protein, including variations thereof.

A main variant of the SARS-CoV-2 Spike protein is provided in UniProtKB database entry P0DTC2, entry version of 2 December 2020, provided as SEQ ID NO: 1 herein.

Preferably, the SARS-CoV-2 Spike protein has a sequence identity of at least 85%, preferably at least 90%, more pre- ferred at least 95%, to SEQ ID NO: 1. Sequence identity is de- termined as discussed above.

Preferably, the infection is an infection with SARS-CoV-2. The infection may be of any SARS-CoV-2 variant or strain, in- cluding 20A.EU1, 20A.EU2, B.l.1.7 (also known as 20I/501Y.V1 or Alpha), B.1.351 (also known as 20H/501Y.V2 or Beta), P.l (also known as 20J/501Y.V3 or Gamma), B.1.617.2 and AY lineages (also known as Delta), B.1.525 (also known as Eta), B.1.526 (also known as Iota), B.1.617.1 (also known as Kappa), B.1.617.3, P.2 (also known as Zeta), B.1.621 (also known as Mu), B.1.1.529 lin- eages, including BA.l, BA.1.1, BA.2, and BA.3 (collectively also known as Omicron), 20B/S.484K, as well as variants or strains with mutations or deletions in the Spike protein, including S:L18F, S:T19I, S:L24S, S:A67V, S:D80Y, T95I, S:S98, especially S:S98F and S:S98Y, S:S13I, S:D138Y, S:G142D, S:Y145H, S:W152C, S:L189F, S:P209H,S:N211I, S:V213G, S:A222V, S:P272L, S:G339D, S:R346K, S:S371, especially S371F and S371L, S:S373P, S:S375F, S:T376A, S:D405N, S:R408S, S:K417, especially S:K417N, S:K417T,, S:N439K, S:N440K, S:K444N, S:G446S, S:L452, especially S:L452R and S:L452Q, S:N453F, S:S477N, S:T478K, S:E484, especially S:E484K and S:E484A, S:F490S, S:Q493R, S:Q496S, S:Q498R, S:N501, especially S:N501Y, S:N501T, S:N501S,, S:Y505H, S:A522S, S:T547K, S:A570D, S:E583D, S:D614G, S:A626S, S:H655Y, S:Q675R, S:Q677H, S:N679K, S:P681, especially S:P681H, S:P681R, S:P681L, S:I692V, S:A701V, S:T716I, S:N764K, S:V772I, S:D796Y, S:N856K, S:Q954H, S:N969K, S:L981F, S:S982A, S:D1118H, S:V1122L, S:M1229I, S:N501, and deletion variant with deletion(s) S:25-27- , S:H69-, S:V70-, S:69/70-, S:144-, S:143-145-, S:212-, and in- sertion variants S:214EPE+, or variants with mutations at other loci, such as E:T9I, M:D3G, M:Q19E, M:A63T, N:P13L, N:31-33-, N:S186Y, N:P199L, N:R203K, N:G204R, N:A220V, N:M234I, N:A376T, N:D377Y, N:S413R, ORFla:S135R, ORFla:I2501T, ORFla:I4205V, ORFla:T842I, ORFla:K856R, ORFla:T945I, ORFla:G1307S, ORFla:T1567I, ORFla:S2083I, ORFla:A2710T, ORFla:L3027F, ORFla:T3090I, ORFla:L3201F, ORFla:T3255I, ORFla:Q3346K, ORFla:P3395H, ORFla:V3475F, ORFla:I3758V, ORFla:M38621, ORFlb:A176S, ORFlb:P255T, ORFlb:P314L, ORFlb:V767L, ORFlb:KI141R, ORFlb:DI183Y, ORFlb:E1184D, ORFlb:R1315C, ORFlb:I1566V, ORFlb:T21631, ORF3a:Q38R, ORF3a:G172R, ORF3a:V202L, ORF3a:T223I, ORF6:D61L, ORF7a:R80I, ORF8:S84L, ORF10:V30L, or combinations thereof, such as in B.l.1.7, e.g. a combination of S:deletion 69-70-, S:deletion 144-, S:N501Y, S:A570D, S:D614G, S:P681H, S:T716I, S:S982A, S:D1118H. Deletion variants are marked by a sign, after the amino acid indica- tion, such as in S:25-27-, S:H69-, S:V70-, S:69/70-, S:144-, S:143-145-, S:212-. Insertions are marked by a "+" sign, such as in S:214EPE+.

Example variants of the spike protein of the coronavirus may comprise any one of the following mutations to the original SARS-CoV-2, e.g. spike protein of SEQ ID NO: 1: any one or more mutations selected from E484Q E484G K417N F456V T478I E484A S494Q N439K F490S S477R S477I S477N N501T K417T T478R L455F N501Y G446S E484K Y449N T478K S494P (in GH clade (B.l.*); any one or more mutations selected from E484Q T478I N439K Y449H G446V F490S Y495H F490L S477N Y489H N501T G476S K417T G496S L455F N501Y V445I E484K T478K G485R S494P (in GR & GRY clade, e.g. B.l.1.1 & B.l.1.7); any one or more mutations selected from E484Q T478I A475T S494T F490I G504S S477N G476S S494A N501Y

G446L G447V V503A P499R E484K E484V Q493L K458N T500S Y453H

Y505W E484G G504D V503F Y453F T500N E484A Q493E K417E V503I

G485V Q506K P499L K417T G485D R403K G504N G485S G446S G502K

N501S K458R V445F V445A P499H G485F S494L Y449H G446V F456L

N439D Y495H F490L F486L N501T T478R G496S G446A T478K G485R S494P Q498R Q493R K417N N501I A475S N439K S477G A475V F490S

S477R G502N S477I Y489L F490V L455F V445I E484D G446R G446D

N501K (in G, GK & GV clade (B.l, B.1.617.2, AY.* & B.1.177); any one or more mutations selected from K417N E484Q Y453F S477G

S494L G446V F490L S477I K417T G496S N501Y L455F E484K P499R

T478K K458N S494P (in nonG clade (A, B & B.2).

A particular advantage of the present invention is that var- ious variants and mutations of the coronavirus can be treated or bound, including the recent Omicron variant of SARS-CoV-2. Such a coronavirus may have one or more of the following mutations in the spike protein compared to unmutated SARS-CoV-2 spike pro- tein, e.g. of SEQ ID NO: 1: A67V, deletion of amino acids 69-70, T95I, G142D, deletion of amino acids 143-145, N211I or deletion of amino acid 211 (N211del), deletion of amino acid 212 or L212I, ins214EPE+, G339D, S371L, S373P, S375F, K417N, N440K, G446S, S477N, T478K, E484A, Q493R, G496S, Q498R, N501Y, Y505H, T547K, D614G, H655Y, N679K, P681H, N764K, D796Y, N856K, Q954H, N969K, L981F; or any combination thereof, such as 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32 or more of these mutations. Other mutations in addition to these may exist or also may not exist.

In accordance with the invention, a pharmaceutical composi- tion or medicine comprising the polypeptide can be provided. The pharmaceutical composition may be in a container, such as a vial, flask, syringe, bag, and/or in kit. Such compositions may be pharmaceutically acceptable salts themselves, with additional buffers, tonicity components or pharmaceutically acceptable car- riers. Pharmaceutical carrier substances serve to improve the compatibility of the composition and provide better solubility as well as better bioavailability of the active ingredients. Ex- amples are emulsifiers, thickeners, redox components, starches, alcoholic solutions, polyethylene glycol and lipids. Selection of a suitable pharmaceutical carrier is highly dependent on the administration route. For oral administration, liquid or solid carriers may be used; for injections, liquid final compositions are required.

Preferably, the polypeptide is provided in a composition comprising buffers or tonic substances. The buffer can adjust the pH of the medicine to the physiological conditions and fur- ther, can reduce or buffer variations in pH. An example is a phosphate buffer. Tonic substances can adjust the osmolarity and may include ionic substances, such as inorganic salts, for exam- ple NaCl or KC1, or non-ionic substances such as glycerin or carbohydrates .

Preferably, the composition for use in accordance with the invention is suitably prepared for systemic, topical, oral or intranasal administration or as an inhaled preparation. Such ad- ministration routes are preferred embodiments of the inventive methods. These forms of administration for the composition of the present invention allow fast, uncomplicated take-up. When the polypeptide is intended for oral administration, it is pref- erably provided in a formulation which is resistant to stomach acid or it is encapsulated. For oral administration, solid or liquid medicines can be taken directly or dissolved or diluted, for example. The pharmaceutical composition or kit for use in accordance with the invention is preferably produced for intra- venous, intra-arterial, intramuscular, intravascular, intraperi- toneal or subcutaneous administration. Injections or transfu- sions, for example, are suitable for this purpose. Administra- tion directly into the bloodstream has the advantage that the active ingredient of the medicine can be distributed through the entire body and the target tissue, such as lungs, heart, kidney, intestine or liver, is reached quickly.

The composition may be pharmaceutically acceptable. The term "pharmaceutically acceptable" indicates that the designated car- rier, vehicle, diluent, excipient (s), and/or salt is generally chemically and/or physically compatible with the other ingredi- ents comprising the formulation, and physiologically compatible with the recipient thereof. In some embodiments, compounds, ma- terials, carriers, compositions, and/or dosage forms that are pharmaceutically acceptable refer to those approved by a regula- tory agency (such as U.S. Food and Drug Administration, National Medicine or European Medicines Agency) or listed in generally recognized pharmacopoeia (such as U.S. Pharmacopoeia, China Pharmacopoeia or European Pharmacopoeia) for use in animals, and more particularly in humans.

Pharmaceutical acceptable carriers for use in the pharmaceutical compositions disclosed herein may include, for example, pharmaceutically acceptable liquid, gel, or solid car- riers, aqueous vehicles, non-aqueous vehicles, antimicrobial agents, isotonic agents, buffers, tonicity-adjusting agents, an- tioxidants, anesthetics, suspending/dispending agents, seques- tering or chelating agents, diluents, adjuvants, excipients, or non-toxic auxiliary substances, other components known in the art, or various combinations thereof. Suitable carriers and aux- iliary components may include, for example, fillers, binders, disintegrants, buffers, preservatives, lubricants, flavorings, thickeners, coloring agents, emulsifiers or stabilizers such as sugars and cyclodextrins. In some embodiments, the suitable buffers may include, for example, a phosphate buffer or a MES (2-(N-morpholino)ethane sulfonic acid) buffer.

To further illustrate, pharmaceutical acceptable carriers may include, for example, aqueous vehicles such as sodium chlo- ride injection, Ringer's injection, isotonic dextrose injection, sterile water injection, or dextrose and lactated Ringer's in- jection, nonaqueous vehicles such as fixed oils of vegetable origin, cottonseed oil, corn oil, sesame oil, or peanut oil, an- timicrobial agents at bacteriostatic or fungistatic concentra- tions, isotonic agents such as sodium chloride or dextrose, buffers such as phosphate or citrate buffers or MES (2- (N-mor- pholino)ethane sulfonic acid) buffers, antioxidants such as so- dium bisulfate, local anesthetics such as procaine hydrochlo- ride, suspending and dispersing agents such as sodium carbox- ymethylcelluose, hydroxypropyl methylcellulose, or polyvinylpyr- rolidone, emulsifying agents such as Polysorbate 80 (TWEEN-80), sequestering or chelating agents such as EDTA (ethylenedia- minetetraacetic acid) or EGTA (ethylene glycol tetraacetic acid), ethyl alcohol, polyethylene glycol, propylene glycol, so- dium hydroxide, hydrochloric acid, citric acid, or lactic acid. Antimicrobial agents utilized as carriers may be added to phar- maceutical compositions in multiple-dose containers that include phenols or cresols, mercurials, benzyl alcohol, chlorobutanol, methyl and propyl p-hydroxybenzoic acid esters, thimerosal, ben- zalkonium chloride and benzethonium chloride. Suitable excipi- ents may include, for example, water, saline, dextrose, glyc- erol, or ethanol. Suitable non-toxic auxiliary substances may include, for example, wetting or emulsifying agents, pH buffer- ing agents, stabilizers, solubility enhancers, or agents such as sodium acetate, sorbitan monolaurate, triethanolamine oleate, or cyclodextrin.

The pharmaceutical compositions with the polypeptide can be a liquid solution, suspension, emulsion, pill, capsule, tablet, sustained release formulation, or powder. Oral formulations can include standard carriers such as pharmaceutical grades of man- nitol, lactose, starch, magnesium stearate, polyvinyl pyrol- lidone, sodium saccharine, cellulose, magnesium carbonate, etc.

The form of pharmaceutical compositions depends on a number of criteria, including, but not limited to, route of administra- tion, extent of disease, or dose to be administered. The pharma- ceutical compositions can be formulated for intravenous, oral, nasal, rectal, percutaneous, or intramuscular administration. For example, dosage forms for intravenous administration, may be formulated as lyophilized powder or fluid formulation; dosage forms for nasal administration may conveniently be formulated as aerosols, solutions, drops, gels or dry powders. In accordance to the desired route of administration, the pharmaceutical com- positions can be formulated in the form of tablets, capsule, pill, dragee, powder, granule, sachets, cachets, lozenges, sus- pensions, emulsions, solutions, syrups, aerosols (as a solid or in a liquid medium), spray, inhalant, or suppository.

In some embodiments, the pharmaceutical compositions are formulated into an injectable composition. The injectable phar- maceutical compositions may be prepared in any conventional form, such as for example liquid solution, suspension, emulsion, or solid forms suitable for generating liquid solution, suspen- sion, or emulsion. Preparations for injection may include ster- ile and/or non-pyretic solutions ready for injection, sterile dry soluble products, such as lyophilized powders, ready to be combined with a solvent just prior to use, including hypodermic tablets, sterile suspensions ready for injection, sterile dry insoluble products ready to be combined with a vehicle just prior to use, and sterile and/or non-pyretic emulsions. The so- lutions may be either aqueous or nonaqueous. Aqueous is pre- ferred.

In some embodiments, unit-dose i.v. or parenteral preparations are packaged in an ampoule, a vial, bag or a sy- ringe with a needle. All preparations for parenteral administra- tion should be sterile and not pyretic, as is known and prac- ticed in the art.

Depending on the route of administration, the polypeptide can be coated with or disposed in a selected material to protect it from natural conditions which may detrimentally affect its ability to perform its intended function. It may be administered alone, or in conjunction with a pharmaceutically acceptable car- rier.

Preferably, the polypeptide or nucleic acid is administered by inhalation, intravenous, intraarterial, intramuscular, intra- vascular, intraperitoneal, sub-cutaneous or oral administration.

Also disclosed is a polypeptide or nucleic acid for use in a method of treating a coronavirus infection comprising adminis- tering a polypeptide comprising a carbohydrate recognition do- main of CD209c or a nucleic acid encoding said polypeptide to a patient suffering from an infection with a coronavirus express- ing a SARS-CoV-2 Spike protein. The entire disclosure herein as disclosed for CLEC4G also equivalently applies to CD209c.

The invention further provides a library of at least 10 different polypeptides, each polypeptide comprising a different carbohydrate recognition domain of a lectin and an immunoglobu- lin domain.

A library is a collection of polypeptides of the invention. The polypeptides of the library may be provided in separate con- tainers, such as flasks, wells, cavities, vials and the like. The individual containers may be combined in a kit-in-parts, wherein the kit contains these containers. A kit may comprise a packaging, such as further containers including the individual containers of the polypeptides. In other embodiments, the poly- peptides may be provided in a mixture. In such a mixture, the polypeptides may remain identifiable through their amino acid sequence or through the use of different labels, such as nucleic acid labels.

Carbohydrate recognition domains (CRDs) of lectins are do- mains for binding carbohydrates as known in the art. An example CRD of CLEC4G has been described above. They are usually found on an extracellular part of the lectins. Preferably, such extracellular parts of the domains themselves of the lectins may be provided. Preferably, the polypeptides lack a lectin membrane domain and or lack a lectin cytosolic domain. The carbohydrate recognition domains (CRDs) of lectins may be homologous to the CRD of CLEC4G and/or may correspond to amino acids 165-288 of SEQ ID NO: 2. The CRDs are usually short sequences. Preferably, the polypeptide comprises 100 to 250 amino acids of a CRD or a lectin or of CLEC4G. In preferred embodiments, the polypeptide may comprise 70 to 200, even more preferred 80 to 180 or 100 to 150, amino acids of a CRD or a lectin or of CLEC4G. Further parts are possible, especially connected to an end of the CRD containing part of the polypeptide, such as in a fusion protein. In particular, the polypeptides contain an immunoglobulin domain, as described above, such as an Fc part of IgG, IgM, IgD, or IgA or a part thereof, such as a CHI, CH2 or CH3 domain, or FcRn.

Preferably, the lectins are selected from C-type lectins, galectins and siglecs. These types of lectins are a preferred selection of with potential for the inventive uses, most promi- nently screening for therapeutic uses.

In particular preferred embodiments, at least 10, preferably at least 20, at least 30, at least 40, at least 50, CRDs are se- lected from the different lectins of table 1. The lectins of ta- ble 1 (lectins given in the right-most column) are particular suitable for the inventive purposes, such as screening for ther- apeutic potential.

Also provided is a set of nucleic acids encoding the poly- peptides of a library of the invention. The nucleic acids can be used to express the inventive polypeptides.

Further provided is a method of identifying or screening a lectin candidate that binds to a carbohydrate-containing target of interest comprising contacting the target with the polypep- tides of a library of the invention, and detecting polypeptides bound to said target.

A carbohydrate-containing target of interest may be a carbo- hydrate itself or a carbohydrate fused to a protein, fatty acid or lipid, such as a glycoprotein or a carbohydrate on a cell membrane. The carbohydrate-containing target may be a bacterial protein, such as virulence factor, or a cell binding protein, e.g. utilised by a pathogenic cell or virion to bind to cells.

In preferred embodiments, the carbohydrate-containing target is a viral protein, a bacterial protein, a fungal protein or a cancer-associated protein. A cancer-associated protein may be a protein that is expressed or overexpressed or expressed in al- tered form and contributes to a cancer cells amplification, im- mortality, tissue invasion or immune evasion.

Preferably, the method further comprises immobilization of the target and detecting polypeptides that are immobilized through binding the immobilized target; and/or wherein the poly- peptides are labelled by binding a labelled immunoglobulin-bind- ing moiety to the polypeptides' immunoglobulin domain. An exam- ple of such a method is an immune assay, such as an ELISA.

Throughout the present disclosure, the articles "a", "an", and "the" are used herein to refer to one or to more than one (i.e., to at least one) of the grammatical object of the arti- cle.

As used herein, words of approximation such as, without lim- itation, "about", "substantial" or "substantially" refer to a condition that when so modified is understood to not necessarily be absolute or perfect but would be considered close enough to those of ordinary skill in the art to warrant designating the condition as being present. The extent to which the description may vary will depend on how great a change can be instituted and still have one of ordinary skill in the art recognize the modi- fied feature as still having the required characteristics and capabilities of the unmodified feature. In general, but subject to the preceding discussion, a numerical value herein that is modified by a word of approximation such as "about" may vary from the stated value by e.g. ±10%.

As used herein, the words "comprising" (and any form of com- prising, such as "comprise" and "comprises"), "having" (and any form of having, such as "have" and "has"), "including" (and any form of including, such as "includes" and "include") or "con- taining" (and any form of containing, such as "contains" and "contain") are inclusive or open-ended and do not exclude addi- tional, unrecited elements or method steps. The "comprising" ex- pressions when used on an element in combination with a numerical range of a certain value of that element means that the element is limited to that range while "comprising" still relates to the optional presence of other elements. E.g. the el- ement with a range may be subject to an implicit proviso exclud- ing the presence of that element in an amount outside of that range. As used herein, the phrase "consisting essentially of" requires the specified integer (s) or steps as well as those that do not materially affect the character or function of the claimed invention. As used herein, the closed term "consisting" is used to indicate the presence of the recited elements only.

The present invention is further described by the following figures and examples, without necessarily being limited to these embodiments of the invention.

Figures

Figure 1. Lectin library and SARS-CoV-2 Spike and RBD glyco- sylation. (A) Schematic overview of cloning, expression and pu- rification of 143 carbohydrate recognition domain (CRD) - mouse IgG2a-Fc fusion proteins, from 168 annotated murine CRD contain- ing proteins. The constructs were expressed in HEK293F cells and secreted Fc-fusion proteins were purified using protein A col- umns. See table 1 for full list of expressed CRDs. (B) Exempli- fied SDS-PAGE of purified Clec7a and Mgl2 stained with Coomassie blue. (C) Glycosylation map of the SARS-CoV-2 Spike and RBD. The most prominent glycan structures are represented for each site, with at least 15% relative abundance. * marks highly variable glycosylation sites in which no single glycan structure ac- counted for >15% relative abundance. The different monosaccha- rides are indicated using standardized nomenclature. NTD, n-ter- minal domain; RBD, receptor binding domain; S1/S2 and S2', pro- teolytic cleavage sites; HR1 and HR2, a-helical heptad repeat domains 1 and 2; GlcNAc, N-acetylglucosamine.

Figure 2. Identification of lectins that bind to Spike and RBD of SARS-CoV-2. (A) and (B) ELISA screen of the lectin-Fc li- brary against full-length trimeric SARS-CoV-2 Spike (A) or mono- meric RBD (B). Results are shown as mean OD values of 2 repli- cates normalized against a BSA control and ranked by value. Lec- tin-Fc fusion proteins with a normalized OD > 0.5 in either (A) or (B) are indicated in both panels. See table 2 for primary ELISA data. (C) Lectin-Fc and human ACE2-mIgGl Fc-fusion protein (hACE2) binding to untreated or heat-denatured full-length SARS- CoV-2 Spike by ELISA. hACE2-m!gGl was used as control for com- plete denaturation of Spike protein. Results are shown as mean OD values ± SD normalized to the BSA control (N=3). (D) Lectin-Fc binding to full-length SARS-CoV-2 Spike with or without de-N- glycosylation by PNGase F. "PNGase F only" denotes wells that were not coated with the Spike protein. Results are shown as mean OD values ± SD normalized to BSA controls (N=3). (E) and (F) Surface plasmon resonance (SPR) analysis with immobilized full- length trimeric Spike, probed with various concentrations of Clec4g-Fc (E) and CD209c-Fc (F). See Table 3 for kinetics val- ues. (C) t-test with Holm-Sidak correction for multiple compari- sons. (D) One-way ANOVA with Tukey's multiple comparisons; *P<0.05; **P<0.01; ***P<0.001; ns: not significant.

Figure 3. Single molecule, real time imaging of lectin-Spike binding. (A) Schematic overview of single molecule force spec- troscopy (SMFS) experiments using full-length trimeric Spike coupled to an atomic force microscopy (AFM) cantilever tip and surface coated murine Clec4g-Fc or CD209c-Fc. Arrow indicates pulling of cantilever. (B) Representative force traces showing sequential bond ruptures in the SMFS experiments. Measured forces are shown in pico-Newtons (pN). (C) Experimental proba- bility density function (PDF) of unbinding forces (in pN) deter- mined by SMFS (black line, measured data). The three distinct maxima fitted by a multi-Gaussian function reveal rupture of a single bond (blue dotted line), or simultaneous rupture of 2 (red dotted line) and 3 (green dotted line) bonds, respectively. (D) SMFS-determined binding probability for the binding of tri- meric Spike to Clec4g and CD209c. Data are shown as mean binding probability ± SD of single, double, triple or quadruple bonds (N=2). (E) High speed AFM of single trimeric Spike visualizing the real-time interaction dynamics with lectins. Top panel shows 5 frames of trimeric Spike alone imaged on mica. Middle and bot- tom panels show 5 sequential frames of trimeric Spike/Clec4g and trimeric Spike/CD209c complexes, acquired at a rate of 153.6 and 303 ms/frame, respectively. Association and dissociation events between lectin and Spike are indicated by white and red arrows, respectively. The blue dotted ellipses display the core of the complexes showing low conformational mobility. Color schemes in- dicate height of the molecules in nanometers (nm). Volumes of single trimeric Spike, trimeric Spike/Clec4g and trimeric Spike/CD209c complexes are indicated, as well as numbers of lec- tins bound to trimeric Spike, averaged over the experimental re- cording period.

Figure 4. Characterization of human lectin-Spike interac- tions. (A) ELISA analyses of human lectin-hlgGl Fc-fusion pro- tein (hCLEC4g, hCD209, hCD299) binding to untreated or heat-de- natured full-length SARS-CoV-2 Spike. A human ACE2-hIgGl-Fc fu- sion protein (hACE2) was used to control for the complete dena- turation of Spike. Results are shown as mean OD values ± SD nor- malized to a BSA control (N=3). (B) hCLEC4g, hCD209, and hCD299 binding to full-length SARS-CoV-2 Spike with or without de-N- glycosylation by PNGase F. "PNGase F only" denotes wells that were not coated with the Spike protein. Results are shown as mean OD values ± SD normalized against the BSA control (N=3). (C) and (D) SPR analysis with immobilized full-length trimeric Spike, probed with various concentrations of hCLEC4G (C) and hCD209 (D). See Table 3 for kinetics values. (E) High speed AFM of single trimeric Spike visualizing the real-time interaction dynamics with lectins. Top and bottom panels show 5 sequential frames of trimeric Spike/hCLEC4G and trimeric Spike/hCD209 com- plexes, acquired at a rate of 303 ms/frame. Association and dis- sociation events between hCLEC4G or hCD209 and Spike are indi- cated by white and red arrows, respectively. The blue dotted el- lipses display the core of the complexes showing low conforma- tional mobility. Color schemes indicate height of the molecules in nanometers (nm). Volumes of single trimeric Spike and tri- meric Spike/Clec4g and trimeric Spike/CD209c complexes are indi- cated, as well as numbers of lectins bound to trimeric Spike, averaged over the experimental recording period. (A) t-test with Holm-Sidak correction for multiple comparisons. (B) One-way ANOVA with Tukey's multiple comparisons; *P<0.05; ***P<0.001; ns: not significant.

Figure 5. Structural modelling and functional determination of lectin-Spike binding in SARS-CoV-2 infections. (A) 3D struc- tural modelling of glycosylated trimeric Spike (green with gly- cans in yellow) interacting with glycosylated human ACE2 (purple with glycans in salmon). The CRD of hCLEC4G (cyan with Ca²⁺ in orange) was modelled onto Spike monomer 3 glycan site N343 (com- plex type glycan with terminal GlcNAc in purple-blue) and the CRD of hCD209 (dark blue with Ca²⁺ in orange) was modelled onto the Spike monomer 3 glycan site N234 (Oligomannose structure Man9 in red). Structural superposition of CLEC4G and ACE2 high- lights sterical incompatibility. (B) and (C) Binding activity of the full-length trimeric Spike coupled to an AFM cantilever tip to the surface of Vero E6 cells in the presence of (B) hCLEC4G and hCD209 or (C) mouse Clec4g or murine CD209c, probed at the indicated concentrations in SMES experiments. A definite de- crease in binding was observed at concentrations of 37 - 75 nM for Clec4g and 33 - 71 nM for hCLEC4G. Data are normalized to the untreated control and shown as mean + SD (N=4). (D) Infectiv- ity of mouse Clec4g or CD209c and (E) hCLEC4G pre-treated SARS- CoV-2 virus in Vero E6 cells. Viral RNA was measured with qRT- PCR 15 hours after infection with 0.02 MOI (10³ PFUs) of SARS- CoV-2. Data represent two pooled experiments for mouse lectins (N=5-6) and one experiment for hCLEC4g (N=3) and are presented as fold changes of viral loads over mock-treated controls (mean loglO values ± SD). (B) - (E) ANOVA with Dunett's multiple com- parisons test with the mock-treated group; *P<0.05; **P<0.01; ***P<0.001.

Figure 6. ELISA assays to detect lectin binding. (A) Sche- matic representation of the ELISA protocol, consisting of coat- ing with trimeric full-length Spike or the monomeric receptor binding domain (RBD) followed by sequential incubation with lec- tin-Fc fusion proteins and secondary anti-IgG-HRP antibodies. The binding of lectin-Fc fusion proteins was quantified by pe- roxidase-dependent substrate conversion, measured by optical density (OD) at 490nm and normalized against a BSA control. (B) ELISA screen of the lectin-Fc library against human recombinant soluble ACE2 (hrsACE2). Results are shown as mean OD values of 2 replicates normalized against a BSA control and ranked by value. (C) SDS-Page of full-length Spike de-N-glycosylated with PNGase F and stained with Coomassie blue. A PNGase F control was added to display the size of the PNGase F protein.

Figure 7. Single molecule atomic force microscopy of a sin- gle trimeric Spike binding to murine Clec4g or CD209c. (A) and (B) Unbinding forces versus loading rates for trimeric Spike dissociating from (A) Clec4g-Fc or (B) CD209c-Fc. Unbinding forces were determined from the magnitude of the vertical jumps measured during pulling of the cantilever (Fig. 3B) and individ- ually plotted versus the respective force loading rates (equal to the pulling speed times effective spring constant) to deci- pher the dissociation dynamics (Table 3). A well-defined single- bond behavior of a unique monovalent bond was found (red dots) that, in line with Evans's single energy barrier model, yielded a linear rise of the unbinding force with respect to a logarith- mically increasing loading rates for both (A) Clec4g and (B) CD209c. Double (green) and triple (blue) bond behaviors were calculated according to the Markov binding model using parame- ters derived from the single barrier model. Unbinding force val- ues scattered between single and triple bond strengths, indicat- ing that interactions with various glycosylation sites with dif- ferent binding strengths. pN=picoNewton, pN/s = picoNewton per second. (C)-(E) High speed AFM of a single trimeric Spike visu- alizing the real-time interaction dynamics with lectins. (C) 5 frames of Clec4g or CD209c alone imaged on mica. (D) Sequential movie frames of trimeric Spike/Clec4g complexes, acquired at a rate of 153.6 ms/frame, corresponding to Fig. 3C. (E) Sequential movie frames of trimeric Spike/CD209c complexes, acquired at 303 ms/frame, corresponding to Fig. 3C. White arrows point to lec- tins associating with the Spike trimer body. Red arrows indicate dissociation of lectins from the Spike trimer, highlighting po- sitions where the lectin was bound in the previous frame. Blue dotted ellipses display low mobility regions. Color schemes in- dicate height in nanometers (nm).

Figure 8. Single molecule atomic force microscopy of a sin- gle trimeric Spike binding to human CLEC4g or CD209. (A) Single molecule force spectroscopy (SMFS) to determine the binding probability for trimeric Spike to mica coated hCLEC4G and hCD209. Data are shown as mean binding probabilities ± SD of sin- gle, double, triple or quadruple bonds (N=2). (B) and (C) Un- binding forces versus loading rates for a single trimeric Spike dissociating from (B) hCLEC4G or (C) hCD209. Unbinding forces were determined from the magnitude of the vertical jumps meas- ured during pulling (Fig. 3B) and individually plotted vs. their force loading rates (equal to the pulling speed times effective spring constant) to assess the dissociation dynamics (Table 3). Single bond interactions (red dots) were fitted using the Bell- Evans single barrier model (red line). A well-defined single- bond behavior of a unique monovalent bond was found (red dots) that, in line with Evans's single energy barrier model, yielded a linear rise of the unbinding force with respect to a logarith- mically increasing loading rate for both (B) hCLEC4g and (C) hCD209. Double (green) and triple (blue) bond behaviors were calculated according to the Markov binding model using parame- ters derived from the single barrier model. Unbinding force val- ues scattered between single and triple bond strengths, indicat- ing that they arise from multiple interactions with various gly- cosylation sites. pN=picoNewton, pN/s = picoNewton per second. (D)-(F) High speed AEM of a single trimeric Spike visualizing the real-time interaction dynamics with lectins. (D) 5 frames of hCLEC4g or hCD209 alone imaged on mica. (E) Sequential movie frames of trimeric Spike/hCLEC4g complexes, acquired at a rate of 303 ms/frame, corresponding to Fig. 4E. (F) Sequential movie frames of trimeric Spike/hCD209 complexes, acquired at 153.6 ms/frame, corresponding to Fig. 4E. White arrows point to lec- tins associating with the Spike trimer body. Red arrows indicate dissociation of lectins from the Spike trimer, highlighting po- sitions where the lectin was bound in the previous frame. The blue dotted ellipses display low mobility regions. Color schemes indicate height in nanometers (nm).

Figure 9. Structural modelling of lectin-Spike interactions. (A) and (B) 3D structural modelling of glycosylated trimeric Spike (green with glycans in yellow) interacting with the CRD of hCD209 (dark blue with Ca²⁺ in orange). The model shows the (A) Spike monomer 1 and (B) Spike monomer 2 glycan site N234 (Oligo- mannose structure Man9 in red) bound to hCD209. (C) and (D) 3D structural modelling of glycosylated trimeric Spike (green with glycans in yellow) interacting with the CRD of hCLEC4g (cyan with Ca²⁺ in orange). The model shows Spike (A) monomer 1 and (B) monomer 2 glycan site N343 (complex type glycan with terminal GlcNAc in purple-blue) bound to hCLEC4g. (E) 3D structural mod- elling of glycosylated trimeric Spike (green with glycans in yellow) interacting with glycosylated human ACE2 (purple with glycans in salmon). The CRD of mClec4g (cyan with Ca²⁺ in orange) was modelled onto Spike monomer 3 glycan site N343 (complex type glycan with terminal GlcNAc in purple-blue). Structural superpo- sition of mClec4g and ACE2 highlights steric incompatibility.

Figure 10. Glycosylation of SARS-CoV-2 Spike and RBD . Rela- tive abundance of all measured glycans in % of all glycans pre- sent at each position. Glycans are grouped in families consist- ing of designated glycan features.

Figure 11. Binding comparison between SARS-CoV-2 Wuhan and Omicron variants. CLEC4G-hIgGl binding to Spike protein deter- mined by ELISA. Results are shown as mean OD values of 3 tech- nical replicates normalized against a BSA control.

Examples

Material and Methods:

Example 1: Identification of proteins containing carbohydrate recognition domains (CRDs).

Mouse lectin sequences were obtained using a domain-based approach. Briefly, proteins with a C-type lectin-like/IPR001304 domain were downloaded from InterPro 66.0 and supplemented with proteins obtained in jackhmmer searches using the PF00059.20 lectin C-type domain definition versus the mouse-specific Uni- Prot and Ensembl databases. The collected set of candidate mouse lectins was made non-redundant using nrdb 3.0. The C-type lec- tin-like regions were extracted from the full-length proteins using the SMART CLECT domain definition with hmmersearch v3.Ib2 and extended by 5 amino acids on both sides. To reduce redun- dancy, principal isoforms were selected using appris 2016_10.v24. In addition, the CRD domains for Galectins and Sig- lecs were added. In case a gene contained more than 1 CRD, all CRDs were cloned separately and differentiated by _1, _2, etc.

Example 2: Cloning of C-type lectin expression vectors.

We used the pCAGG_00_ccb plasmid and removed the toxic ccb element by cleaving the plasmid with Bsal. Thereafter, we in- serted a Fc-fusion construct, consisting of the IL2 secretion signal, followed by an EcoRV restriction site, a (GGGS)a linker domain and the mouse IgG2a-Fc domain. Subsequently, each identified CRD, was cloned in-frame into the EcoRV site.

Example 3: Transfection and purification of the lectin-m!gG2a fusion proteins.

The CRD containing plasmids were transfected into Free- style™ 293-F cells. Briefly, the day before transfection, 293-F cells were diluted to 0.7xl0⁶ cells/ml in 30 ml Freestyle™ 293-F medium and grown at 120 rpm at 37°C with 8% CO2. The next morn- ing, 2 μl polyethylenimine (PEI) 25K (Img/ml; Polysciences, 23966-1) per pg of plasmid DNA were mixed with pre-warmed Opti- MEM media (ThermoFisher Scientific, 31985-062) to a final volume of 950 μl in tube A. In tube B, 1 pg of DNA per ml of media was mixed with pre-warmed Opti-MEM to a final volume of 950 pl. Then, the contents of tube A and B were mixed, vortexed for 1 min and incubated at room temperature for 15 min. Thereafter, the transfection mixture was added to the cell suspension. 24h after transfection, EX-CELL 293 Serum-Free Medium (Sigma Al- drich, 14571C) was added to a final concentration of 20%. The transfected cells were grown for 120h and the supernatants, con- taining the secreted lectin-m!gG2a fusion proteins, harvested by centrifugation at 250g for 10 min.

Purification of the mouse lectin-m!gG2a fusion proteins was performed using Protein A agarose resin (Gold Biotechnology, P- 400-5). The protein A beads were pelleted at 150g for 5 min and washed once with lx binding buffer (0.02 M Sodium Phosphate, 0.02% sodium azide, pH=7.0), before resuspending in lx binding buffer. Immediately preceding purification, aggregates were pel- leted from the cell culture supernatant by centrifugation for 10 min at 3000g. lOx binding buffer was added to the cell culture supernatant to a final concentration of lx as well as 4 μl of protein A beads per ml of cell culture supernatant. The bead/ supernatant mixture was incubated overnight at 4 °C. The next morning, beads were collected by centrifugation for 5 min at 150g, washed twice with 20 and 10 bead volumes of lx binding buffer, the bead pellets transferred to a 1ml spin column (G-Bi- osciences, 786-811) and washed once more with 1 bead volume of lx binding buffer. Excess buffer was removed by centrifugation at 100g for 5 sec. Lectin-mIgG2a fusion proteins were eluted from the protein A beads by resuspending the beads in 1 bead volume of Elution buffer (lOOmM Glycine-HCl, pH=2-3, 0.02% so- dium azide). After 30 sec of incubation the elution buffer was collected into a 2 ml Eppendorf tube, containing Neutralization buffer (IM Tris, pH=9.0, 0.02% sodium azide) by centrifugation at 100g for 15 sec. Elution was performed for a total of 3 times. The 3 eluted fractions were pooled and the protein con- centrations measured with the Pierce™ BCA Protein Assay Kit (ThermoFisher Scientific, 23225) using the Pierce™ Bovine Gamma Globulin Standard (ThermoFisher Scientific, 23212). To confirm the purity of the eluted lectin-IgG fusion proteins, we per- formed an SDS-PAGE, followed by a Coomassie staining. Briefly, 1 pg of eluted lectin-IgG fusion protein was mixed with Sample Buffer, Laemmli 2x Concentrate (Sigma-Aldrich, S3401) and heated to 95°C for 5 minutes. Thereafter, the samples were loaded onto NuPage™ 4-12% Bis-Tris gels (ThermoFisher Scientific, NP0321BOX) and run in lx MOPS buffer at 140V for 45 minutes. The gel was subsequently stained with InstantBlue™ Safe Coomassie Stain (Sigma-Aldrich, ISB1L) for Ih and de-stained with distilled wa- ter. The gel picture was acquired with the ChemiDoc MP Imaging System (BioRad) in the Coomassie setting.

Example 4: Recombinant expression of SARS-CoV-2 Spike protein and the receptor binding domain (RED).

Recombinant protein expression was performed by transient transfection of HEK293-6E cells, licensed from National Research Council (NRC) of Canada, as previously described (Durocher et al. (2002) Nucleic Acids Res 30, E9). Briefly, HEK293-6E cells were cultivated in Freestyle F17 expression medium (Thermo Fisher Scientific, A1383502) supplemented with 0.1% (v/v) Plu- ronic F-68 (Thermo Fisher Scientific, 24040032) and 4 mM L-glu- tamine (Thermo Fisher Scientific, 25030081) in shaking flasks at 37°C, 8% CO2, 80% humidity and 130 rpm in a Climo-Shaker ISF1-XC (Adolf Kuhner AG). The pCAGGS vector constructs, containing ei- ther the sequence of the SARS-CoV-2 RBD (residues R319-F541) or the complete luminal domain of the Spike protein (modified by removing all arginine (R) residues from the polybasic furin cleavage site RRAR and introduction of the stabilizing point mu- tations K986P and V987P) were kindly provided by Florian Krammer, Icahn School of Medicine at Mount Sinai (NY, United States) (Amanat et al. (2020) Nature Medicine 26, 1033-1036;

Stadlbauer et al. (2020) Curr Protoc Microbiol 57, elOO). Tran- sient transfection of the cells was performed at a cell density of approximately 1.7><10⁶ cells/mL culture volume using a total of 1 pg of plasmid DNA and 2 pg of linear 40-kDa PEI (Polysciences, 24765-1) per mL culture volume. 48 h and 96 h after transfec- tion, cells were supplemented with 0.5% (w/v) tryptone N1 (Or- ganotechnie 19553) and 0.25% (w/v) D(+)-glucose (Carl Roth X997.1). Soluble proteins were harvested after 120-144 h by cen- trifugation (10000 g, 15 min, 4°C).

Example 5: Purification of recombinant trimeric Spike protein and monomeric RBD of SARS-CoV-2.

For purification, the supernatants were filtered through 0.45 pm membrane filters (Merck Millipore HAWP04700), concen- trated and diafiltrated against 20 mM phosphate buffer contain- ing 500 mM NaCl and 20 mM imidazole (pH 7.4) using a Labscale TFF system equipped with a 5 kDa cut-off Pellicon™ XL device (Merck Millipore, PXC005C50). His-tagged trimer Spike and mono- meric RBD were captured using a 5 mL HisTrap FF crude column (Cytiva, 17528601) connected to an AKTA pure chromatography sys- tem (Cytiva). Bound proteins were eluted by applying a linear gradient of 20 to 500 mM imidazole over 20 column volumes. Frac- tions containing the protein of interest were pooled, concen- trated using Vivaspin 20 Ultrafiltration Units (Sartorius, VS2011) and dialyzed against PBS (pH 7.4) at 4°C overnight using a SnakeSkin Dialysis Tubing (Thermo Fisher Scientific, 68100). The RBD was further polished by size exclusion chromatography (SEC) using a HiLoad 16/600 Superdex 200 pg column (Cytiva, 28- 9893-35) equilibrated with PBS (pH 7.4). Both purified proteins were stored at -80°C until further use.

Example 6: Glycoproteomic analysis of Spike and RBD.

Peptide mapping and glycoproteomic analysis of all samples were performed on in-solution proteolytic digests of the respec- tive proteins by LC-ESI-MS (/MS). In brief, the pH of the samples was adjusted to pH 7.8 by the addition of 1 M HEPES, pH 7.8 to a final concentration of 100 mM HEPES, pH 7.8. The samples were then chemically reduced and S-alkylated using a final concentration of 10 mM dithiothreitol for 30 min at 56°C and a final concentration of 20 mM iodoacetamide for 30 min at room- temperature in the dark, respectively. To maximize protein se- quence-coverage of the analysis, proteins were digested with ei- ther Trypsin (Promega), a combination of Trypsin and GluC (Promega) or Chymotrypsin (Roche). Eventually, all proteolytic digests were acidified by addition of 10% formic acid to pH 2 and directly analyzed by LC-ESI-MS (/MS) using an a capillary Bi- oBasic C18 reversed-phase column (BioBasic-18, 150 x 0.32 mm, 5 pm, Thermo Scientific), installed in an Ultimate U3000 HPLC sys- tem (Dionex), developing a linear gradient from 95% eluent A (80 mM ammonium formate, pH 3.0, in HPLC-grade water) to 65% eluent B (80% acetonitrile in 80 mM ammonium formate, pH 3.0) over 50 min, followed by a linear gradient from 65% to 99% eluent B over 15 min, at a constant flow rate of 6 pL/min, coupled to a maXis 4G Q-TOF instrument (Bruker Daltonics; equipped with the stand- ard ESI source). For (glyco)peptide detection and identifica- tion, the mass-spectrometer was operated in positive ion DDA mode (i.e. switching to MS/MS mode for eluting peaks), recording MS-scans in the m/z range from 150 to 2200 Th, with the 6 high- est signals selected for MS/MS fragmentation. Instrument cali- bration was performed using a commercial ESI calibration mixture (Agilent). Site-specific profiling of protein glycosylation was performed using the dedicated Q-TOF data-analysis software pack- ages Data Analyst (Bruker Daltonics) and Protein Scape (Bruker Daltonics), in conjunction with the MS/MS search engine MASCOT (Matrix Sciences Ltd.) for automated peptide identification.

Example 7: ELISA assays to detect lectin binding to Spike and RBD.

Briefly, 50 μl of full-length Spike-H6 (4 pg/ml, purified from HEK, diluted in PBS), RBD-H6 (2 pg/ml, purified from HEK, diluted in PBS) or human recombinant soluble ACE2 (hrsACE2; 4 pg/ml, purified from CHO, diluted in PBS, see Monteil et al. (Cell 181, 2020: 905-913 e907), per well were used to coat a clear flat-bottom MaxiSorp 96-well plate (Thermo Fisher Scien- tific, 442404) for 2h at 37°C. Thereafter, the coating solution was discarded and the plate was washed 3 times with 300 μl of wash buffer (IxTBS, ImM CaC12, 2mM MgC12, 0.25% Triton X-100 (Sigma-Aldrich, T8787)). Unspecific binding was blocked with 300 μl of blocking buffer (IxTBS, 1% BSA Fraction V (Applichem, A1391,0100), ImM CaC12, 2mM MgC12 and 0.1% Tween-20 (Sigma-Al- drich, P1379)) for 30 min at 37°C. After removal of the blocking solution, 50 μl of either the mouse lectin-m!gG2a (10 pg/ml, di- luted in blocking buffer), human CD209-hIgGl (R&D Systems, 161- DC-050), human CD299-hIgGl (R&D Systems, 162-D2-050), human CLEC4G-hIgGl (Aero Biosystems, CLG-H5250-50ug) (10 pg/ml, di- luted in blocking buffer), recombinant human ACE2-mIgGl (2 pg/ml, diluted in blocking buffer, Sino Biological, 10108-H05H) or recombinant human ACE2-hIgGl (2 pg/ml, diluted in blocking buffer, Sino Biological, 10108-H02H) were added for Ih at room temperature. After washing for 3 times, 100 μl of 0.2 pg/ml HRP- conjugated goat anti-Mouse IgG (H+L) (Thermo Fisher Scientific, 31430) or goat anti-Human IgG (H+L) (Promega, W4031) antibodies were added for 30 min at room temperature. Subsequently, plates were washed as described above. To detect binding, 1 tablet of OPD substrate (Thermo Fisher Scientific, 34006) was dissolved in 9 ml of deionized water and 1ml of lOx Pierce™ Stable Peroxide Substrate Buffer (Thermo Fisher, Scientific, 34062). 100 μl of OPD substrate solution were added per well and incubated for 15 min at room temperature. The reaction was stopped by adding 75 μl of 2.5M sulfuric acid and absorption was read at 490 nm. Ab- sorption was measured for each lectin-Fc fusion protein tested against full-length Spike-H6, RBD-H6 or hrsACE2 and normalized against bovine serum albumin coated control wells.

Example 8: Protein denaturation and removal of N-glycans.

To denature the full-length Spike-H6, 10 mM DTT was added to 40 pg/ml of protein. The samples were incubated at 85°C for 10 min. Thereafter, the denatured proteins were diluted to 4 pg/ml with PBS and a clear flat-bottom Maxisorp 96-well plate was coated with 50 μl per well for 2h at 37°C. To remove the N-gly- cans from the full-length Spike-H6, 0.2 pg protein were dena- tured as above and adjusted to a final concentration of lx Gly- cobuffer 2 containing 125U PNGase F per pg (NEB, P0704S) in 50 pl. After incubation for 2h at 37°C, the reaction was stopped by heat-inactivation for 10 min at 75°C. Spike proteins were then diluted to 4 pg/ml with PBS and a clear flat-bottom Maxisorp 96- well plate was coated with 50 μl per well for 2h at 37°C. ELISA protocols were performed as described above. To confirm the de- glycosylation of Spike proteins, 0.5 pg were loaded on an SDS- PAGE gel followed by a Coomassie staining.

Example 9: Surface plasmon resonance (SPR) measurements.

A commercial SPR (BIAcore X, GE Healthcare, USA) was used to study the kinetics of binding and dissociation of lectin-Fc di- mers to the trimeric full-length Spike in real time. Spike-H6 was immobilized on a Sensor Chip NTA (Cytiva, BR100034) via its His6-tag after washing the chip for at least 3 minutes with 350 mM EDTA and activation with a 1 min injection of 0.5 mM N1C12. 50 nM Spike were injected multiple times to generate a stable surface. For the determination of kinetic and equilibrium con- stants, the lectin samples (murine Clec4g, murine CD209c, human CLEC4G, human CD209) were injected at different concentrations (10 to 500 nM). As the binding of the lectins is Ca²⁺ dependent, lectins were removed from the surface by washing with degassed calcium free buffer (TBS, 0.1% Tween-20, pH = 7.4). The reso- nance angle was recorded at a 1 Hz sampling rate in both flow cells and expressed in resonance units (1 RU = 0.0001°). The re- sulting experimental binding curves were fitted to the "bivalent analyte model", assuming a two-step binding of the lectins to immobilized Spike. All evaluations were done using the BIAevalu- ation 3.2 software (BIAcore, GE Healthcare, USA).

Example 10: Single molecule force spectroscopy (SMFS) measure- ments.

For single molecule force spectroscopy a maleimide-Poly (eth- ylene glycol) (PEG) linker was attached to 3-aminopropyltrieth- oxysilane (APTES)-coated atomic force microscopy (AFM) cantile- vers by incubating the cantilevers for 2h in 500 pL of chloro- form containing 1 mg of maleimide-PEG-N-hydroxysuccinimide (NHS) (Polypure, 21138-2790) and 30 μl of triethylamine. After 3 times washing with chloroform and drying with nitrogen gas, the canti- levers were immersed for 2h in a mixture of 100 pL of 2 mM thiol-trisNTA, 2 pL of 100 mM EDTA (pH 7.5), 5 pL of 1 M HEPES (pH 7.5), 2 μl of 100 mM tris(carboxyethyl)phosphine (TCEP) hy- drochloride, and 2.5 pL of 1 M HEPES (pH 9.6) buffer, and subsequently washed with HEPES-buffered saline (HBS). Thereaf- ter, the cantilevers were incubated for 4h in a mixture of 4 pL of 5 mM N1C12 and 100 pL of 0.2 pM His-tagged Spike trimers. Af- ter washing with HBS, the cantilevers were stored in HBS at 4°C (Oh et al., 2016). For the coupling of lectins to surfaces, a maleimide-PEG linker was attached to an APTES-coated silicon ni- tride surface. First, 2 μl of 100 mM TCEP, 2 μl of IM HEPES (pH 9.6), 5 μl of IM HEPES (pH 7.5), and 2 μl of 100 mM EDTA were added to 100 μl of 200 pg/ml Protein A-Cys (pro-1992-b, Prospec, NJ, USA) in PBS. The surfaces were incubated in this solution for 2h and subsequently washed with PBS and 0.02 M sodium phos- phate containing 0.02% sodium azide, pH=7.0. Finally, 100 μl of 200 pg/ml lectin-Fc fusion proteins were added to the surfaces overnight.

Force distance measurements were performed at room tempera- ture (~25 °C) with 0.01 N/m nominal spring constants (MSCT, Bruker) in TBS buffer containing 1 mM CaC12 and 0.1 % TWEEN-20. Spring constants of AFM cantilevers were determined by measuring the thermally-driven mean-square bending of the cantilever using the equipartition theorem in an ambient environment. The deflec- tion sensitivity was calculated from the slope of the force-dis- tance curves recorded on a bare silicon substrate. Determined spring constants ranged from 0.008 to 0.015 N/m. Force-distance curves were acquired by recording at least 1000 curves with ver- tical sweep rates between 0.5 and 10 Hz at a z-range of typi- cally 500 - 1000 nm (resulting in loading rates from 10 to 10,000 pN/s), using a commercial AFM (5500, Agilent Technolo- gies, USA). The relationship between experimentally measured un- binding forces and parameters from the interaction potential were described by the kinetic models of Bell (Bell, 1978) and Evans and Ritchie (Evans and Ritchie, 1997). In addition, multi- ple parallel bond formation was calculated by the Williams model (Williams, 2003) from the parameters derived from single bond analysis. The binding probability was calculated from the number of force experiments displaying unbinding events over the total number of force experiments.

The probability density function (PDF) of unbinding force was constructed from unbinding events at the same pulling speed. For each unbinding force value, a Gaussian unitary area was computed with its center representing the unbinding force and the width (standard deviation) reflecting its measuring uncer- tainty (square root of the variance of the noise in the force curve). All Gaussian areas from one experimental setting were accordingly summed up and normalized with its binding activity to yield the experimental PDF of unbinding force. PDFs are equivalents of continuous histograms as shown in Fig. 3C.

Example 11: High-speed AFM (hsAFM) and data analysis.

Purified SARS-CoV-2 trimeric Spike glycoproteins, murine Clec4g and CD209c and hCLEC4g and hCD209 were diluted to 20 pg/ml with imaging buffer (20mM HEPES, ImM CaC12, pH 7.4) and 1.5 μl of the protein solution was applied onto freshly cleaved mica discs with diameters of 1.5 mm. After 3 minutes, the surface was rinsed with ~15pL imaging buffer (without drying) and the sample was mounted into the imaging chamber of the hsAFM (custom-built, RIBM, Japan). Pictures for movies were captured in imaging buffer containing 3pg/ml of either Clec4g, CD209c, hCLEC4G or hCD209. An ultra-short cantilever (USC-F1.2-kO.15 nominal spring constant 0.15 N/m, Nanoworld, Switzerland) was used and areas of lOOxlOOnm containing single molecules were selected to capture the hsAFM movies at a scan rate of ~150-300 ms per frame. During the acquisition of the movies, the amplitude was kept constant and set to 90-85% of the free amplitude (typically ~3 nm). Data analysis was performed using the Gwyddion 2.55 software. Images were processed to remove background and transient noise. For volume measurements, a height threshold mask was applied over the protein structures with a minimum height of 0.25 - 0.35 nm to avoid excessive background noise in the masked area. The num- bers of lectin molecules bound to the Spike trimers was calcu- lated based on the measured mean volumes of the full-length Spike, the lectins, and the Spike-lectin complexes, averaged over the recorded time-periods.

Example 12: AFM measured Spike binding to Vero E6 cells.

Vero E6 cells were grown on culture dishes using DMEM con- taining 10% FBS, 500 units/mL penicillin and 100 pg/mL strepto- mycin, at 37°C with 5% CO2. For AFM measurements, the cell den- sity was adjusted to about 10-30% confluency. Before the measurements, the growth medium was exchanged to a physiological HEPES buffer containing 140 mM NaCl, 5 mM KC1, 1 mM MgC12, 1 mM CaC12, and 10 mM HEPES (pH 7.4). Lectins were added at the indi- cated concentrations. Using a full-length Spike trimer anchored to an AEM cantilever (described above), force-distance curves were recorded at room temperature on living cells with the as- sistance of a CCD camera for localization of the cantilever tip on selected cells. The sweep range was fixed at 3000 nm and the sweep rate was set at 1 Hz. For each cell, at least 100 force- distance cycles with 2000 data points per cycle and a typical force limit of about 30 pN were recorded.

Example 13: Structural modelling.

Structural models of the SARS-CoV-2 Spike protein were based on the model of the fully glycosylated Spike-hACE2 complex.

Experimental structures deposited in the protein databank (PDB) were used to model the complex that is formed by the bind- ing of SARS-CoV-2 Spike and ACE2 (Walls et al. (2020) Cell 181, 281-292 e286; Yan et al. (2020) Science 367, 1444-1448). RBD do- main in complex with ACE2 was superimposed with Spike with one open RBD domain (PDB: 6VYB) and SWISS-MODEL (Waterhouse et al. (2018) Nucleic Acids Res 46, W296-W303) was used to model miss- ing residues in Spike (GenBank QHD43416.1). Glycan structures were added using the methodology outlined by Turupcu et al. (FEMS Microbiol Rev 38, 2014: 598-632). For hCLEC4G a homology model of residues 118 - 293 was constructed using Swiss-Model using residues 4 - 180 of chain A of the crystal structure of the carbohydrate recognition domain of DC-SIGNR (CD299) (PDB en- try lsl6). This fragment shows a sequence identity of 36% with hCLEC4G, and the resulting model showed an overall QMEAN value of -2.68. For mClec4g, the model consisted of residues 118 - 294, with a sequence identity of 38 % to the same template model. The resulting QMEAN value was -2.88. A calcium ion and the bound Lewis x oligosaccharide of the template were taken over into the model, indicating the location of the carbohydrate binding site. For hCD209, the crystal structure of the carbohy- drate recognition domain of CD209 (DC-SIGN) complexed with Man4 (PDB entry lsl4) was used. To identify binding sites of hCLEC4G and hCD209 to the Spike-hACE2 complex, a superposition of the bound carbohydrates with the glycans on Spike was performed. For hCLEC4G, we used the complex glycans at N343 of the third mono- mer of Spike, with the receptor binding domain in an 'up' posi- tion, while N343 glycans on monomer 1 and 2 were modelled with the receptor binding domain in a 'down' position. For hCD209 we used the high-mannose glycan at position N234 in monomer 1-3 of Spike, respectively. These glycan structures were chosen in ac- cordance with the full-length Spike glycoproteome.

Example 14: SARS-CoV-2 infections.

Vero E6 cells were seeded in 48-well plates (5xl0⁴ cells per well) in DMEM containing 10% FBS. 24 hours post-seeding, differ- ent concentrations of lectins were mixed with 10³ PFU of virus (1:1) to a final volume of lOOμl per well in DMEM (resulting in a final concentration of 5% FBS). After incubation for 30 min at 37°C, Vero E6 were infected either with mixes containing lec- tins/SARS-CoV-2, SARS-CoV-2 alone, or mock infected. 15 hours post-infection, supernatants were removed, cells were washed 3 times with PBS and then lysed using Trizol Reagent (Thermo Fisher Scientific, 15596026). The qRT-PCR for the detection of viral RNA was performed as previously described (Monteil et al., 2020, supra). Briefly, RNA was extracted using the Direct-zol RNA MiniPrep kit (Zymo Research, R2051). The qRT-PCR was per- formed for the SARS-CoV-2 E gene and RNase P was used as an en- dogenous gene control to normalize viral RNA levels to the cell number. Lectins were independently tested for cellular toxicity in an ATP-dependent assay (Cell-Titer Gio, Promega) and cells found to exhibit >80% viability up to a concentration of 200 pg/ml.

The following PCR Primers were used:

Results :

Example 15: Preparation of the first near genome-wide lectin li- brary to screen for novel binders of Spike glycosylation.

To systematically identify lectins that bind to the trimeric Spike protein and RBD of SARS-CoV-2, we searched for all anno- tated carbohydrate recognition domains (CRDs) of mouse C-type lectins, Galectins and Siglecs. Of 168 annotated CRDs, we were able to clone, express and purify 143 lectin-CRDs as IgG2a-Fc fusion proteins from human HEK293F cells (Fig. 1A, table 1). The resulting dimeric lectin-Fc fusion proteins (hereafter referred to as lectins) showed a high degree of purity (Fig. IB). This collection of lectins is, to our knowledge, the first comprehen- sive library of mammalian CRDs.

Table 1. Overview of carbohydrate recognition domains (CRDs) used for the lectin library

This table presents a list of CRDs expressed and purified as Fc- fusion proteins for the lectin library. Information displayed are the lectin name, the family and in the case of C-type lec- tins, the group and group name the CRD belongs to. CRDs from lectin that contain several CRDs are distinguished by suffix numbers. CTL = C-type lectin

We next recombinantly expressed monomeric RBD and full- length trimeric Spike protein (hereafter referred to as Spike protein) in human HEK293-6E cells. Using mass spectrometry, we characterized all 22 N-glycosylation sites on the full-length Spike protein and 2 N-glycosylation sites on the RBD (Fig. 1C and 10). Most of the identified structures were in accordance with previous studies using full-length Spike (Watanabe et al. (2020) Science 369, 330-333) with the exception of N331, N603 and N1194, which presented a higher structural variability of the glycan branches (Fig. 1C and 10). The detected N-glycan spe- cies ranged from poorly processed oligo-mannose structures to highly processed multi-antennary complex N-glycans in a site-de- pendent manner. This entailed also a large variety of terminal glycan epitopes, which could act as ligands for lectins. Nota- bly, the two glycosylation sites N331 and N343 located in the RBD carried more extended glycans, including sialylated and di- fucosylated structures, when expressed as an independent con- struct as opposed to the full-length Spike protein (Fig. 1C and 10). These data underline the complex glycosylation of Spike and reveal that N-glycosylation of the RBD within the 3D context of full-length trimeric Spike is different from N-glycosylation of the RBD expressed as minimal ACE2 binding domain.

Example 16: CD209c and Clec4g are novel high affinity binders of SARS-CoV-2 Spike.

We evaluated the reactivity of our murine lectin library against the trimeric Spike and monomeric RBD of SARS-CoV-2 using an ELISA assay (fig. 6A). This screen revealed that CD209c (SIGNR2), Clec4g (LSECtin), and Regl exhibited pronounced bind- ing to Spike, whereas Mgl2 and Asgrl displayed elevated binding to the RBD (Fig. 2A, B and table 2). Further, we investigated the reactivity of the lectin library against human recombinant soluble ACE2 (hrsACE2); none of the lectins bound to hrsACE2 (fig. 6B). We excluded Regl from further studies due to incon- sistent ELISA results, likely due to protein instability. Asgrl was excluded because it bound only to RBD but not to the Spike trimer, in accordance to the differences in glycosylation of glyco-sites N331 and N343 between Spike and RBD (Fig. 1C). This highlights the importance of using a full-length trimeric Spike protein for functional studies. To confirm that the observed in- teractions were independent of protein conformation, Spike was denatured prior to the ELISA assay; binding of CD209c and Clec4g to the unfolded Spike remained unaltered (Fig. 2C). Importantly, enzymatic removal of N-glycans by PNGase F treatment reduced the binding of CD209c, Clec4g, and Mgl2 towards Spike (Fig. 2D and 6C), confirming N-glycans as ligands. Binding of ACE2, which re- lies on protein-protein interactions, was completely abrogated when Spike was denatured (Fig. 2C). These data identify lectins that have the potential to bind to the RBD and trimeric Spike of

SARS-CoV-2.

Table 2. ELISA screen of the lectin-Fc library against SARS-CoV- 2 Spike, RED and hrsACE2

Results from the ELISA screens of the lectin-Fc library against indicated targets. Data displayed per lectin is mean and stand- ard deviation (SD) (n=2). SD=NA indicates that mean was calcu- lated from a single replicate only.

Based on the robust N-glycan dependent Spike binding, we fo- cused our further studies on CD209c and Clec4g. We first used surface plasmon resonance (SPR) to determine the kinetic and equilibrium binding constants of these lectins to the trimeric Spike. The resulting experimental binding curves were fitted to the "bivalent analyte model" (Traxler et al. (2017) Angew Chem Int Ed Engl 56, 15755-15759) which assumes two-step binding of the lectin dimers to adjacent immobilized Spike trimer binding sites (Fig. 2E, F). From these fits, we computed the kinetic as- sociation (k_a,i), kinetic dissociation (kd,i), and equilibrium dis- sociation (Kd,i) binding constants of single lectin bonds (Table 3). The equilibrium dissociations (Kd,i) values were 1.6 pM and 1.0 pM for Clec4g and CD209c, respectively.

Table 3. Values computed for surface plasmon resonance (SPR) and atomic force microscopy (AFM). SPR. Kinetic association (k_a,i), kinetic dissociation (kd,i), and equilibrium dissociation (Kd,i) constants of the first binding step fitted from the bivalent an- alyte model, assuming two-step binding and dissociation of the lectins to adjacent immobilized Spike trimer binding sites under spontaneous thermodynamic energy barriers; no reasonable fit was obtained with the simple 1:1 binding model. AFM. Kinetic off- rate constants (k_orr) and lengths of dissociation paths (xp) of single lectin bonds, originating from force-induced unbinding in single molecule force spectroscopy (SMFS) experiments and com- puted using Evans's model (Bell (1978) Science 200, 618-627; Ev- ans and Ritchie (1997) Biophys J 72, 1541-1555), assuming a sharp single dissociation energy barrier.

Example 17: Multiple CD209c and Clec4g molecules bind simultane- ously to SARS-CoV-2 Spike and form compact complexes.

To study Spike binding of these two lectins at the single- molecule level, we used atomic force microscopy (AFM) and per- formed single molecule force spectroscopy (SMFS) experiments. To this end, we coupled trimeric Spike to the tip of the AFM canti- lever and performed single-molecule force measurements (Hinter- dorfer et al. (1996) Proc Natl Acad Sci USA 93, 3477-3481), by moving the Spike trimer-coupled tip towards the surface-bound lectins to allow for bond formations (Fig. 3A). Unbinding was accomplished by pulling on the bonds, which resulted in charac- teristic downward deflection signals of the cantilever, whenever a bond was ruptured (Fig. 3B). The magnitude of these vertical jumps reflects the unbinding forces, which were of typical strengths for specific molecular interactions (Rankl et al. (2008) Proc Natl Acad Sci USA 105, 17778-17783). Using this method (Rankl et al., supra; Zhu et al. (2010) Nat Nanotechnol 5, 788-791), we quantified unbinding forces (Fig. 3C) and calcu- lated the binding probability and the number of bond ruptures between CD209c or Clec4g and trimeric Spike (Fig. 3D). Both lec- tins showed a very high binding probability and could establish up to 3 strong bonds with accumulating interaction force strengths reaching 150 pN in total with trimeric Spike (Fig. 3C), with the preference of single and dual bonds (Fig. 3C, 3D, fig. 7A, B, Table 3). Of note, multi-bond formation leads to stable complex formation, in which the number of formed bonds enhances the overall interaction strength and dynamic stability of the complexes. To assess dynamic interactions between single molecules of trimeric Spike and the lectins in real time we used high-speed AFM (Kodera et al. (2010) Nature 468, 72-76). Addi- tion of Clec4g and CD209c led to a volume increase of the lec- tin/Spike complex in comparison to the trimeric Spike alone; based on the volumes we could calculate that on average 3.2 mol- ecules of Clec4g and 5.2 molecules of CD209c were bound to one Spike trimer (Fig. 3E, fig. 7C-E). These data show, in real- time, at single molecule resolution, that mouse Clec4g and CD209c can directly associate with trimeric Spike.

Example 18: The human lectins CD209 and CLEC4G are high affinity receptors for SARS-CoV-2 Spike.

Having characterized binding of murine CD209 and Clec4g to Spike, we next assessed whether their closest human homologues, namely human CD209 (hCD209), and human CD299 (hCD299) for murine CD209c, and human CLEC4G (hCLEC4G) for mouse Clec4g, can also bind to full-length trimeric Spike of SARS-CoV-2. hCD209, hCD299, and hCLEC4G indeed exhibited binding to Spike (Fig. 4A), demonstrating conserved substrate specificities. The binding of these human lectins was again independent of Spike folding and abrogated by N-glycan removal (Fig. 4A, B). SPR measurements of hCLEC4G and hCD209 bond formation to Spike showed equilibrium dissociation (Kd,i) values of 0.3 pM and 2.4 pM, respectively, in which the high affinity of hCLEC4G is mainly contributed by its rapid kinetic association rate constant (Fig. 4G, D and Table 3). In addition, hCD209 and hCLEC4G showed a high binding proba- bility by AFM with the formation of up to 3 bonds per trimeric Spike (fig. 8A-C). When we monitored the dynamic interactions of the lectins with the Spike using high speed AFM, we observed binding of - on average - 3.5 hCLEC4G and 3.6 hCD209 molecules per trimeric Spike (Fig. 4E and fig. 8D-F). In summary, our data using ELISA, SMFS, surface plasmon resonance, and high-speed atomic force microscopy show that the human lectins CLEC4G and CD209 can bind to trimeric Spike of SARS-CoV-2, in which the overall interaction strength and dynamic stability leads to com- pact complex formation.

Example 19: CLEC4G sterically interferes with Spike/ACE2 inter- action.

We next 3D modelled binding of CLEC4G and CD209 to the can- didate glycosylation sites present on Spike and how such attach- ment might relate to the binding of the trimeric Spike protein to its receptor ACE2. hCD209 is known to bind with high affinity to oligo-mannose structures (Guo et al., supra). The N234 glyco- sylation site is the only site within the Spike that carries ex- clusively oligo-mannose glycans with up to 9 mannose residues (Fig. 1C, Fig. 10). 3D modelling revealed that the oligo-mannose glycans on N234 are accessible for CD209 binding on all 3 mono- mers comprising the trimeric Spike (Fig. 5A and fig. 9A, B). Su- perimposition of ACE2 interacting with the RBD and CD209 binding to N234, showed that the CD209 binding occurs at the lateral in- terface of Spike, distant from the RBD (Fig. 5A). Human CLEC4G and mouse Clec4g have a high affinity for complex N-glycans ter- minating with GlcNAc. The N-glycan at N343, located within the RBD, is the glycosylation site most abundantly decorated with terminal GlcNAc in Spike (Fig. 1C, Fig. 10). The terminal GlcNAc glycans on position N343 are accessible for hCLEC4G binding on all 3 Spike monomers, but in contrast to CD209, hCLEC4G binding interferes with the ACE2/RBD interaction (Fig. 5A, fig. 9C, D). Similarly, modelling murine Clec4g, having the same ligands as hCLEC4G (Pipirou et al. (2011) Glycobiology 21, 806-812), bind- ing to the N343 glycan site, also predicted interference with the ACE2/RBD interaction (fig. 9E). Thus, whereas hCD209 is not predicted to interfere with ACE2/RBD binding, murine Clec4g and human CLEC4G binding to the N343 glycan impedes Spike binding to ACE2.

Example 20: CD209c and CLEC4G block SARS-CoV-2 infection.

To test our structural models experimentally, we assessed whether these lectins could interfere with Spike binding to the surface of Vero E6 cells, a frequently used SARS-CoV-2 infection model (Monteil et al., supra). To determine this, we set-up an AFM based method, measuring spike binding activity on Vero E6 cells. Strikingly, as predicted by the structural modelling, we found that hCLEC4G, but not hCD209, significantly interfered with the binding of trimeric Spike to the Vero E6 cell surface (Fig. 5B). Similarly, mouse Clec4g, but not murine CD209c, in- terfered with Spike binding to Vero E6 cells (Fig. 5C). Finally, we tested the ability of these lectins to reduce the infectivity of SARS-CoV-2. Murine Clec4g significantly reduced SARS-CoV-2 infection of Vero E6 cells (Fig. 5D). SARS-CoV-2 infections of Vero E6 were also reduced by murine CD209c (Fig. 5D), indicating that glycosylation sites outside the ACE2/RBD interface are also important for SARS-CoV-2 infectivity. Finally, hCLEC4G also sig- nificantly reduced SARS-CoV-2 infection of Vero E6 cells (Fig. 5E). These data uncover that the lectins CLEC4G and CD209c can interfere with SARS-CoV-2 infections.

Example 21. ELISA assays to detect lectin binding to Spike pro- tein of SARS-CoV-2 Wuhan and Omicron variants.

50 μl of full-length Spike-H6 ("Wuhan", 4 pg/ml, purified from HEK, diluted in PBS, as described in Example 7, or full- length Omicron Spike-H6 ("Omicron", 4 pg/ml, purified from HEK, diluted in PBS, Aero Biosystems, PN-C52Hz) per well were used to coat a clear flat-bottom MaxiSorp 96-well plate (Thermo Fisher Scientific, 442404) for 2 h at 37°C. Thereafter, the coating so- lution was discarded and the plate was washed three times with 300 μl of wash buffer (lx TBS, 1 mM CaC12, 2 mM MgC12, 0.25% Tri- ton X-100 (Sigma-Aldrich, T8787)). Unspecific binding was blocked with 300 μl of blocking buffer (lx TBS, 1% BSA Fraction V (Ap- plichem, A1391,0100), 1 mM CaC12, 2 mM MgC12 and 0.1% Tween-20 (Sigma-Aldrich, P1379)) for 30 min at 37°C. After removal of the blocking solution, 50 μl of human CLEC4G-hIgGl (Aero Biosystems, CLG-H5250-50ug) (10 pg/ml, diluted in blocking buffer) were added for 1 h at room temperature. After washing for three times, 100 μl of goat anti-Human IgG (H + L) (Promega, W4031) antibody was added for 30 min at room temperature. Subsequently, plates were washed as described above. To detect binding, 100 μl of 1-Step Ultra TMB-ELISA (Thermo scientific, 34028) was added per well and incubated for 15 min at room temperature. The reaction was stopped by adding 75 μl of 2.5 M sulfuric acid, and absorption was read at 450 nm. Results are shown in Figure 11. In this ex- periment, the "Wuhan" spike is from a different manufacturer than the "Omicron" spike. Therefore the "Wuhan" spike should be considered as a positive control rather than for a quantitative comparison. The data shows that CLEC4G binds strongly to both the Wuhan and Omicron variants of SARS-CoV-2 spike protein, il- lustrating that CLEC4G binding to spike protein is little af- fected by spike protein mutations. Thus CLEC4G has broad use as a therapeutic against various SARS-CoV-2 variants and mutants.

Discussion: The examples show an unbiased screening of a compre- hensive mammalian lectin library for therapeutic agent identifi- cation. As an example, use of the library, potent SARS-CoV-2 Spike binding, identifying mouse CD209c and Clec4g, as well as their human homologs hCD209 and hCLEC4G, as N-glycan dependent Spike receptors were identified. hCD209 has been identified as candidate receptor for SARS-CoV-2 by other groups, and other lectins have been also implicated in cellular interactions with Spike (Gao et al. (2020) supra; Thepaut et al. (2020) bioRxiv doi: 10.1101/2020.08.09.242917). CLEC4G has been reported to as- sociate with SARS-CoV (Gramberg et al. (2005) Virology 340, 224- 236), but has not been implicated in SARS-CoV-2 infections and not as a therapeutic target.

High-speed atomic force microscopy allowed us to directly observe Spike/lectin interactions in real time. Our high-speed AFM data showed that Spike/CLEC4G formed more rigid clusters with lower conformational flexibility as compared to the Spike/CD209 complexes in equilibrium conditions. This is in ac- cordance with faster association rates and shorter dissociation paths of CLEC4G as compared to CD209. The experimentally ob- served association of 3-4 CLEC4G molecules with one molecule of the trimeric spike indicates the formation of high affinity bonds to 1-2 glycosylation sites per monomeric subunit of the trimeric Spike.

Since glycosylation is not a template driven process, but rather depends on the coordinated action of many glycosyltrans- ferases and glycosidases, each glycosylation site can - within some boundaries - carry a range of glycans. As a consequence, the 3 monomers of Spike can harbor different glycans on the same glycosylation site on different Spike proteins. We identified N343 as the one glycosylation site that is almost exclusively covered with GlcNAc terminated glycans, the ligands of CLEC4G. Given its localization in the RBD and its abundant decoration with potential CLEC4G ligands, we hypothesized that CLEC4G bind- ing interferes with the RBD-ACE2 interaction. Indeed, we found that murine Clec4g and human CLEC4G, acting as a multi-valent effective inhibitor (Ki ~ 35-70 nM), can functionally impede with Spike binding to host cell membranes, thereby providing a ra- tionale how this lectin can affect SARS-CoV-2 infections. In support of our data, it has recently been reported that a N343 glycosylation mutant exhibits reduced infectivity using pseudo- typed viruses (Li et al. (2020) Cell 182(5), 1284-1294.e9). While CLEC4G can interfere with RBD-ACE2 binding, CD209 does ap- parently not associate with glycans near the RBD and hence does not block Spike binding to cells. This is in agreement to its proposed high affinity oligo-mannosidic ligands, presented at N234, which is localized at a distance to the RBD-ACE2 inter- face. Whether CD209 interferes with proteolytic cleavage or al- pha-helical interactions of Spike can be explored. Our data identifies the key glycan ligands for CLEG4G and CD209. Binding to other glycosylation sites of lower affinities is not excluded.

Lectins play critical roles in multiple aspects of biology such as immune responses, vascular functions, or as endogenous receptors for various human pathogens. Hence, our library con- taining a selection of 143 lectins or more allows to comprehen- sively probe and map glycan structures on viruses, bacteria or fungi, as well as during development or on cancer cells, provid- ing novel insights on the role of lectin-glycosylation interac- tions in infections, basic biology, and disease. For instance, CD209 is expressed by antigen presenting dendritic cells, as well as inflammatory macrophages and is known to bind to a vari- ety of pathogens, like HIV and Ebola, but also Mycobacterium tu- berculosis or Candida albicans. CLEC4G is strongly expressed in liver and lymph node sinusoidal endothelial cells and can also be found on stimulated dendritic cells and macrophages. CD299, one of the two homologues of mouse CD209c, which we also identi- fied to possess Spike binding ability, is co-expressed with CLEC4G on liver and lymph node sinusoidal endothelial cells. Si- nusoidal endothelial cells are important in the innate immune response, by acting as scavengers for pathogens as well as anti- gen cross-presenting cells. Thus, lectin binding to Spike allows to couple SARS-CoV-2 infections to antiviral immunity. Since vi- ral protein glycosylation depends on the glycosylation machiner- ies of the infected cells which assemble viral particles, slight changes in glycosylation might explain differences in anti-viral immunity and possibly severity of the disease, with critical im- plications for vaccine designs. Moreover, Spike-binding lectins could enhance viral entry in tissues with low ACE2 expression, thus extending the organ tropism of SARS-CoV-2.

Claims

Claims:

1. A polypeptide or nucleic acid for use in a method of treat- ing a coronavirus infection comprising administering a polypep- tide comprising a carbohydrate recognition domain of CLEC4G or a nucleic acid encoding said polypeptide to a patient suffering from an infection with a coronavirus expressing a SARS-CoV-2 Spike protein.

2. The polypeptide or nucleic acid for use according to claim 1 wherein the polypeptide comprises CLEC4G.

3. The polypeptide or nucleic acid for use according to claim 1 or 2, wherein the polypeptide comprises an immunoglobulin do- main, preferably an antibody CHI, CH2 or CH3 domain.

4. The polypeptide or nucleic acid for use according to claim 3, wherein the polypeptide comprises an antibody Fc fragment.

5. The polypeptide or nucleic acid for use according to any one of claims 1 to 4, wherein the polypeptide is a dimer.

6. The polypeptide or nucleic acid for use according to any one of claims 1 to 5, wherein the CLEC4G is human CLEC4G and the pa- tient is a human.

7. The polypeptide or nucleic acid for use according to any one of claims 1 to 6, wherein the polypeptide lacks a membrane do- main and/or wherein the polypeptide comprises 50 to 250, prefer- ably 100 to 180, amino acids of CLEC4G.

8. The polypeptide or nucleic acid for use according to any one of claims 1 to 7, wherein the polypeptide or nucleic acid is ad- ministered by inhalation, intravenous, intraarterial, intramus- cular, intravascular, intraperitoneal, sub-cutaneous or oral ad- ministration.

9. The polypeptide or nucleic acid for use according to any one of claims 1 to 8, wherein the infection is an infection with SARS-CoV-2.

10. The polypeptide or nucleic acid for use according to any one of claims 1 to 8, wherein the SARS-CoV-2 Spike protein has a se- quence identity of at least 85%, preferably at least 90%, more preferred at least 95%, to SEQ ID NO: 1.

11. The polypeptide or nucleic acid for use according to any one of claims 1 to 10, wherein the SARS-CoV-2 Spike protein has a sequence identity to one or more of the following amino acid mu- tations in comparison to SEQ ID NO: 1: A67V, deletion of amino acids 69-70, T95I, G142D, deletion of amino acids 143-145, N211I or deletion of amino acid 211 (N211del), deletion of amino acid 212 or L212I, insertion 214EPE, G339D, S371L, S373P, S375F, K417N, N440K, G446S, S477N, T478K, E484A, Q493R, G496S, Q498R, N501Y, Y505H, T547K, D614G, H655Y, N679K, P681H, N764K, D796Y, N856K, Q954H, N969K, L981F, or any combination thereof; prefera- bly 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32 or more of these amino acid mutations.

12. A library of at least 10 different polypeptides, each poly- peptide comprising a different carbohydrate recognition domain of a lectin and an immunoglobulin domain.

13. The library according to claim 14, wherein the lectins are selected from C-type lectins, galectins and siglecs.

14. The library according to claim 12 or 13, wherein the poly- peptides lack a lectin membrane domain and/or wherein the poly- peptide comprises 50 to 250, preferably 100 to 180, amino acids in length of CLEC4G.

15. A set of nucleic acids encoding the polypeptides of a li- brary of any one of claims 14 to 16.

16. A method of identifying or screening a lectin candidate that binds to a carbohydrate-containing target of interest comprising contacting the target with the polypeptides of a library of any one of claims 12 to 14, and detecting polypeptides bound to said target.

17. The method of claim 16 comprising immobilization of the tar- get and detecting polypeptides that are immobilized through binding the immobilized target; and/or wherein the polypeptides are labelled by binding a labelled immunoglobulin-binding moiety to the polypeptides' immunoglobulin domain.