EP4058033A1 - Polypeptides related to hmgb1 useful for promoting tissue regeneration, compositions comprising same, and uses thereof - Google Patents

Abstract

This invention provides a polypeptide represented by the following formula: H2N-A-X-B-A-X-B-HOOC wherein A represents consecutive amino acids, the sequence of which (1) includes a sequence identical to the sequence of amino acids 90 – 93 of wildtype HMGB1, (2) has at its amino terminal end, between one and six consecutiveamino acids, for example, 1, 2, 3, 4, 5, or 6 amino acids, the sequence of which isidentical to the sequence of the corresponding one to six amino acids preceding aminoacid 90 in wild type HMGB1, and optionally (3) has a methionine at the aminoterminus;wherein X represents consecutive amino acids, the sequence of which is identical to the sequence of amino acids 94 – 162 of wild type HMGB1; and wherein B represents consecutive amino acids, the sequence of which (1) includes a sequence identical to the sequence of amino acids 163 – 168 of wild type HMGB1 and (2) has at its carboxy terminal end, between one and six consecutive amino acids, for example, 1, 2, 3, 4, 5, or 6 amino acids, the sequence of which is identical to the sequence of the corresponding one to six amino acids following amino acid 168 in wild type HMGB1; and wherein each - represents a peptide bond between each of A and X, X and B, B and A, A and X, and X and B.

Description

POLYPEPTIDES RELATED TO HMGB1 USEFUL FOR PROMOTING TISSUE

REGENERATION. COMPOSITIONS COMPRISING SAME. AND USES THEREOF

[0001] This application claims the benefit of U.S. Provisional Application No. 62/934,299 filed November 12, 2019, the contents of which is hereby incorporated by reference.

[0002] Throughout this application, various publications are referenced, including referenced in parenthesis. The disclosures of all publications mentioned in this application in their entireties are hereby incorporated by reference into this application in order to provide additional description of the art to which this invention pertains and of the features in the art which can be employed with this invention.

REFERENCE TO SEQUENCE LISTING

[0003] This application incorporate s-by-reference nucleotide sequences which are present in the file named “201112_91203-A-PCT_Sequence_Listing_AWG.txf’, which is 78 kilobytes in size, and which was created on November 12, 2020 in the IBM-PC machine format, having an operating system compatibility with MS-Windows, which is contained in the text file filed November 12, 2020 as part of this application.

TECHNICAL FIELD

[0004] This disclosure relates to engineered polypeptides related to HMGB1 that promote tissue regeneration without potential for deleterious inflammation and methods of treating an acute tissue injury by administering the engineered polypeptides to subjects in need thereof.

BACKGROUND OF THE INVENTION

[0005] Resident stem and progenitor cells play a key role in maintaining homeostasis and effecting repair of many tissues following injury [1]. However, most tissues in adults heal by scarring. Following the success of bone marrow transplantation [2] there has been considerable interest in exogenous stem cell therapies to promote regeneration of solid organs, with limited success in a few organs such as the eye [3] and skin [4]. The inflammatory environment following tissue injury is not conducive to stem cell engraftment and the subsequent scarring disrupts the stem cell niche [5]. Therefore, focus has shifted to promoting tissue regeneration by stimulating endogenous repair mechanisms [6]. Development of a successful therapeutic would depend on identification of soluble mediators to promote these pathways. We previously identified High Mobility Group Box 1 (HMGB1) as a key mediator of repair in multiple tissues, including bone, blood and skeletal muscle [7]. [0006] HMGB1 is a prototypical alarmin [8,9] and under physiological conditions has an essential role in transcription [10,11]. On cell injury it is passively released from the damaged and necrotic cells into the extracellular space and the circulation to act on stem and progenitor cells to transition them to G_Alert [7], a state intermediate between G₀ and G₁ [12]. On exposure to the appropriate activating factors, cells in G_Alert are able to rapidly enter Gi and effect tissue repair. If not required, stem cells in G_Alert revert back to Go after approximately 3 weeks [12], thereby ensuring that they are not exhausted and the niche is not depleted.

[0007] HMBG1 comprises two L-shaped Box domains, A and B, each containing 3 α-helices (I - III) connected by flexible regions that are involved in LPS (N-terminus of Box A and adjacent C-terminal linker region) [13] or RAGE (C-terminus of Box B) binding [14]. The C- terminus of the protein is intrinsically disordered and contains a high proportion of carboxylic acid residues (Glu/Asp) comprising the acidic tail. This binds to the HMG Boxes to regulate activities, including interactions with TLR-2 [10,15,16] and potentially also RAGE [15] (Figures 1A-1B). The oxidation status of HMGB1 cysteine residues (Cys 22, Cys 44 in Box A and Cys 105 in Box B) is a key determinant of the extracellular activities of HMGB1 and in turn is dependent on the mechanism of release. Three different redox forms have been described in vivo [17]. HMGB1 passively released from the nuclei following injury or cell necrosis is the fully-reduced form (FR-HMGB1). It binds to CXCL12 and the heterocomplex signals via the cell surface receptor CXCR4 to transition stem and progenitor cells to G_Alert [7] . Partial oxidation in the local inflammatory environment results in the formation of disulfide HMGB1 (DS-HMGB1) [18,19], which has a disulfide bond between Cys 22 and Cys 44. This is also the form that is actively secreted by immune cells following acetylation [20] and N- glycosylation [21]. DS-HMGB1 signaling via the receptor for advanced glycation end products (RAGE) activates platelets and is a key mediator of thrombosis [22,23] . DS-HMGB1 also acts via TLR-4 and TLR-2, leading to release of proinflammatory cytokines, including TNF and IL-6 [24]. Intracellular signaling via all three receptors converges to induce NF-kβ activity [25] in a MyD 88 -dependent manner [26,27]. Oxidation of all three cysteine residues through the action of extracellular reactive oxygen species results in sulfonyl-HMGB1 (SO₃), which is biologically inactive [17,28].

[0008] The disulfide bridge in Box A of DS-HMGB1 (Cys22-Cys44) is essential for TLR-4 signaling (Figures 1A-1B), initiating binding to TLR-4 but also has a relatively high dissociation rate. MD-2 then binds to Box B with low affinity but very low dissociation rates, stabilizing the interaction [29]; the Phe-Cys-Ser-Glu (FCSE, 104-107) peptide in Box B is essential for this interaction [30]. The capacity of DS-HMGB1 to signal via TLR-4 has been overcome by substituting cysteines at positions 22, 44 and 105 with serine, resulting in an engineered form described as 3S-HMGB1 [17]. Whilst the authors claimed that 3S-HMGB1 has enhanced regenerative properties compared to FR-HMGB1 [31], we found that in bone, blood and skeletal muscle injuries it was equivalent to FR-HMGB1. Interestingly, 3S-HMGB1 has been reported to be deleterious when administered locally following myocardial infraction whereas FR-HMGB1 resulted in a smaller infarct and enhanced cardiac function assessed over 4 weeks [32]. There are no published data on the effect of the 3S substitutions on TLR-2 or RAGE signaling.

[0009] The sites for TLR-2 interaction are not clearly defined but glycyrrhizin is known to inhibit this interaction [33]. This suggests that at least one, and potentially both HMG Box domains and the acidic tail [10] are involved. The acidic tail negatively modulates HMGB1 and it has been reported that co-ligands [34] are necessary to displace the acidic tail of HMGB1 from the Box domains to permit signaling via TLR-2 [35]. However, several publications have reported TLR-2 dependent proinflammatory signaling with HMGB1 alone [24,36] and it is possible that the requirement for a co-ligand is cell type and context dependent. The role of the redox state of HMGB1 with regards to TLR-2 signaling remains unclear as both the disulfide form [22] and the fully-reduced form [34] have been proposed to signal through TLR-2. RAGE interaction has been primarily mapped to a peptide in HMG Box B (residues 149-182) [14] (Figures 1A-1B ), and peptides derived from this sequence can effectively inhibit HMGB1- RAGE signaling [37]. In addition, there is also a second caspase-dependent site within Box A. More recently a second RAGE binding site was identified on HMG Box A [38], accessible only after proteolysis by Caspase 11 [38]. However, the relative contribution of each site to RAGE signaling remains unknown. It is also recognized that prothrombotic signaling mediated by RAGE requires the disulfide form of HMGB1, implying that Box A is involved [22]. The acidic tail of HMGB1 may negatively regulate RAGE signaling in a manner analogous to TLR- 2 as it binds residues within the RAGE binding peptide [10,15,39] .

[0010] Successful translation of the regenerative activities of FR-HMGB1 to the clinic is dependent on the elimination of all potential deleterious signaling via RAGE, TLR-2 and TLR- 4 whilst maintaining CXCL 12-binding and signaling via CXCR4. Here the residues in HMGB1 that are critical for binding to CXCL 12 are identified and an HMGB1 variant that retains regenerative activity whilst eliminating RAGE binding and TLR-2 and TLR-4 signaling is described. SUMMARY OF THE INVENTION

[0011] This invention provides a polypeptide represented by the following formula:

H2N-A-X-B-A-X-B-HOOC wherein A represents consecutive amino acids, the sequence of which (1) includes a sequence identical to the sequence of amino acids 90 - 93 of wild type HMGB1, (2) has at its amino terminal end, between one and six consecutive amino acids, for example, 1, 2, 3, 4, 5, or 6 amino acids, the sequence of which is identical to the sequence of the corresponding one to six amino acids preceding amino acid 90 in wild type HMGB1, and optionally (3) has a methionine at the amino terminus; wherein X represents consecutive amino acids, the sequence of which is identical to the sequence of amino acids 94 - 162 of wild type HMGB1; and wherein B represents consecutive amino acids, the sequence of which (1) includes a sequence identical to the sequence of amino acids 163 - 168 of wild type HMGB1 and (2) has at its carboxy terminal end, between one and six consecutive amino acids, for example, 1, 2, 3, 4, 5, or 6 amino acids, the sequence of which is identical to the sequence of the corresponding one to six amino acids following amino acid 168 in wild type HMGB1; and wherein each - represents a peptide bond between each of A and X, X and B,

B and A, A and X, and X and B.

[0012] This invention also provides a composition comprising the polypeptide in accordance with the invention and a carrier, and methods of treating a subject suffering from, or at risk for developing, a condition which would be alleviated by promoting regeneration of a tissue or cells that rely upon CXCR4⁺ cells for repair which comprise administering to the subject a polypeptide of the invention in an amount effective to promote regeneration of the tissue or cells and a therapeutic or prophylactic dose of a pharmaceutical composition of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figures 1A-1B show a schematic of the HMGB1 structure and locations of known immunogenic activities. Figure 1A: Structure of HMGB1 (PDB 2YRQ, conformer 1, colored in PyMol according to known interactions with LPS, TLR-4 or RAGE. The acidic tail, which is involved in transcriptional modulation and bactericidal activities, is not shown in the structure. Figure IB: Schematic representation of binding sites. Original figure coloring shows Box A - blue; Box B - green; Pink: residues involved in glycyrrhizin binding. Red: flexible N- terminal regions adjacent to Box A or Box B; Orange: cysteine residues; White: linker region between HMG Boxes; Bright green and yellow: RAGE binding region (incomplete as it extends into the acidic tail, yellow).

[0014] Figures 2A-2F show conserved residues in each HMG Box domain are critical for CXCL12 binding and include the N-terminal D-P-X-X tetramer. Figure 2A: Peptide array (11x10) of HMGB1 15-mers incubated with 1 μM CXCL12-His6 and detected with anti-His5- HRP antibody. Intensity of spots corresponds to amount of CXCL12 bound to the peptides; first two and last two spots in the array comprised 10-His positive controls. Figure 2B: Intensity quantification of spot intensity in (Figure 2A) (duplicate runs) normalized to 10-his control. Peptides used for alanine scanning experiments have been highlighted. Peptides in the acidic tail were not included, as due to its high negative charge it would non-specifically bind cationic molecules such as CXCL12. Peptides in graphs are represented in SEQ ID NOs: 8- 104, left to right. Figure 2C: Peptide array of alanine mutagenesis at single positions within peptides identified in (Figures 2A-2B). First spot in each row corresponds to the positive control; second spot to the unmodified peptide. Peptides shown are represented in SEQ ID NOs: 105-111. Figure 2D: Intensity quantification of the array in (Figure 2C, SEQ ID NOs: 105-111), normalized to the unmodified peptide. Residues which showed variation in CXCL12 signal greater than the variation observed seen in alanine residues (Ala -> Ala; synonymous mutation, in grey) are show in red with original figure coloring. Figure 2E: Michaelis-Menten saturation fits of biotinylated HMGB1 constructs [full-length FR red/3 S black; minimal box A 8-78 (violet) and box B 94-162 (brown), extended box A 1-88 (pink) and box B 89-174 (gray)] binding to CXCL12 with original figure coloring. Figure 2F: Summary of kinetic parameters derived from (Figure 2E). Affinity (Kd) constants from both fits follow the same relationship and are greatly decreased for HMGB1 94-162; analyzed by 1-way Brown-Forsythe ANOVA from the fitted data. Kd values were compared via a post-hoc 2-way ANOVA, averaging both values as no significant differences were found in the paired comparison (column factor). Raw interferograms can be found in Figure 9.

[0015] Figures 3A-3B show NMR validation of residues involved in CXCL12 binding. Figure 3A: Cumulative CSP of helical -only biotinylated Box B (94-162, HMGBlA-c028) or complete Box B (89-174, HMGBlA-c038) after titration with CXCL12 (0.42, 0.84 and 1.42 molar equivalents), calculated over several HSQC spectra including a parallel control with no CXCL12 measured after the last concentration point (CSP drift control). Intensity of the green color in the graph indicates relative CSP. The sequence of each HMGB1 construct has been overlaid with the residue number; an empty column (no number) represents residues which could not be mapped in the parallel 3D ¹H-¹⁵N HSQC/NOE/TOCSY experiments. The sequence of the residues corresponding to each HMG Box is shown as follows with original figure coloring: Light blue, residues previously reported in the literature as involved in CXCL12 binding; red, residues weakly involved in the peptide array; purple, residues involved according to both published literature and peptide array. Grey, alanine residues within CXCL12 binding peptides (underlined) that could not be assessed in the peptide arrays. Sequence shown is represented in SEQ ID NO: 5. Figure 3B: Cumulative peak height change of (Figure 3A); Heatmap of NMR changes. With original figure coloring, red indicates I/I0 change over 1 SEM of all residues, blue decrease over -1 SEM. Sequence shown is represented in SEQ ID NO: 5.

[0016] Figures 4A-4C show design of the dBB12L construct. Figure 4A: Alignment of Box A + linkers (1-88) (SEQ ID NO: 3) with Box B + linkers (89-174) (SEQ ID NO: 4); values correspond to NMR nomenclature (excluding N-terminal methionine). Vertical lines designate strictly conserved positions, and double dots similar substitutions. Underlined: peptide regions binding CXCL12 from the first peptide array. With original figure coloring, Red: residues flagged in the alanine scan as involved in CXCL12 binding which could not be verified by NMR; Orange: residues flagged in the alanine scan and also showing either CSP or peak volume change by NMR; Cyan: residues not flagged in our NMR or peptide array experiments but described in the NMR literature as contributing to CXCL12 binding [44]; Purple: residues flagged peptide array experiments and confirmed by the published or our NMR data; Green: residues flagged only on NMR experiments, which can either directly bind CXCL12 or be affected by binding to nearby residues; Pink: residues flagged by both our NMR experiments and the published data. Figure 4B: Structure of FR-HMGB1 1-166 (2YRQ), with the residues colored following the same color code as in (Figure 4A). Side chains of all colored residues have been shown. Dashed circles indicate the glycyrrhizin binding region in each HMG Box. Figure 4C: Schematic of dBB12L construct design. The initiation codon Met 1 is numbered as Met 0 herein, as it is partially lost in the cleaved peptide. Therefore, HMGB1 Met 1 -Gly 2...Glu 215 becomes Met 0- Gly 1...Glu 214. Domain organization and sequences of FR- HMGB1 (top - SEQ ID NO: 1) and dBB 12L construct (bottom - SEQ ID NO: 2). The dBB 12L construct is designed such that: 1. The acidic tail and part of the RAGE binding domain (175- 214) have been deleted; 2. Residues 1-88 (Box A) have been substituted by residues 89-174, resulting in two HMG Box B domains; and 3. Residues 163-174 C-terminal to Box B replace the native flexible linker (79-88) C-terminal to Box A in native HMGB1. CXCL12 binding peptides are shown as red letters. The repeat Box B units are separated in the diagram by a dashed black line.

[0017] Figures 5A-5D show dBB12L has similar stability and surface charge conformation to FR-HMGB1 1-214/1-164. Figure 5A: Calculated Tm₅₀ values (in °C) for full-length and 1- 164 FR-HMGB1, and dBB12L. Shading of individual Tm₅₀ values indicate highest (green) and lowest (red) values within the global dataset for all constructs. N/A: curve not fittable. Figure 5B: Native ESI/MS of HMGB1 constructs in either 50 mM or 0.2 M ammonium acetate, pH 6.5. All three HMGB1 constructs have similar native M/Z profiles, with dBB12L closely resembling a reduced HMGB1 construct with two HMG Boxes apart from each other. Continuous line; compact monomer. Dashed line; extended monomer (HMG Boxes distal to each other). Removal of the acidic tail (FR HMGB1 1-164, blue curves compared to FR- HMGB1, red curves with original figure coloring) and higher ionic strength (Comparison of the spectra for the same construct in either 50 mM or 200 mM ammonium acetate) increases the prevalence of higher M/Z states (partial unfolding). Figure 5C: Solvent accessible surface area (SASA) calculations for the average folded HMGB1 monomer, the extended and compact monomer states, and the unfolded monomer from (Figure 5D). Figure 5D: Denaturing ESI/MS deconvolution, SDS-PAGE and SEC profiles of HMGB1 constructs after storage at room temperature for 180 days (D0-D180), in 0.2 M ammonium acetate, pH 6.5.

[0018] Figure 6A-6F show dBB12L has reduced binding to RAGE and does not signal through TLR-2 or TLR-4. Figure 6A: Michaelis-Menten saturation fits from hybrid ELISA (n=4 per concentration, global fit) showing absence of RAGE binding by dBB12L construct, regardless of oxidation status. DS-HMGB1 binds RAGE more avidly compared to FR- HMGB1. Data normalized with respect to DS-HMGB1 control. Figure 6B: Michaelis-Menten saturation fits from biolayer interferometry (0-25 μM HMGB1; n = 6 sensor runs). Association and dissociation rates were calculated from the raw interferogram data only. Kinetics parameters and color legends (disulfide forms are indicated by dashed lines) are summarized in Figure 6C: Invalid fits represent R² < 0.6 (poor binding). Data from ELISA are proportionally more influenced by the dissociation constants than they are by association constants due to the nature of the experiment. For each kinetic parameter, green indicates the construct with the highest affinity, fastest association (k_on) or slowest dissociation (koff) midpoint in yellow, lowest in red in the originally colored figure. Figures 6D-6E show DS- HMGB1 promoted NF-kβ activity in reporter HEK-Dual cells expressing human TLR-2 and CD 14 (Figure 6D) or murine TLR-4, MD-2 and CD 14. (Figure 6E). DbB-HMGBl did not promote NF-kβ signaling, whereas FR-HMGB1 only induced minor NF-kβ activation in both cell lines, potentially due to partial oxidation during the assay. Values are shown as mean ± SEM fold change compared to control (media alone). Figure 6F: Disulfide HMGB1 (DS- HMGB1) increased TNF production in monocytes, which was further enhanced by the presence of suboptimal amounts of LTA, but not of LPS. Both FR-HMGB1 and DbB-HMGBl did not elicit TNF secretion, even when pre-incubated for 24h with LPS or LTA. Response to LPS pre-incubated with these constructs was also significantly reduced n=3 donors, each with three (3) technical replicates.

Figures 7A-7J show the regenerative effects of optimal doses of dBB-HMGBl and FR- HMGB1 are identical to those of an activating injury. Figure 7A: Volcano plot showing differentially expressed genes in muscle stem cells by fold change following injury orHMGBl induced G_Alert. Integration demonstrates conserved up- (brown dots in figures with color) and down- (blue dots in figures with color) regulation of core genes in G_Alert induced by contralateral lower limb injury or intravenous (iv) HMGB1. Figure 7B: Network map of gene ontology terms of differentially expressed cells during G_Alert induction in muscle stem cells. Figure 7C: Dose response of FR-HMGB1 in a BaCI₂ skeletal muscle injury model, with regeneration quantified by fiber cross-sectional area. The optimal dose was 0.75 mg/kg (28.75 nmol/kg) and was used in subsequent assays. Figure 7D: Animals were dosed with FR- HMGB1 (optimal dose) at the varying timepoints after injection of BaCI₂ to assess the interval where treatment with FR-HMGB1 is effective post-injury. Values in (Figure 7C), (Figure 7D) shown as mean ± SEM in nested ANOVA with Holm-Sidak correction (values shown for post- hoc tests). Figure 7E: Pharmacokinetics of iv HMGB1 in mice, fitted by nonlinear least squares to a two-phase exponential decay curve (circulating HMGB1 after intravenous injection of the optimal dose). Figure 7F: Survival following myocardial infarction (MI) at 5 wk FR-HMGB1 = 83 %, PBS = 52 %. Figure 7G: Ejection fraction. Dotted line indicates ejection fraction in normal/sham surgery mice. Figure 7H: Infarct size compared by 2-way ANOVA for treatment effect across times. Figure 71: Representative mid-ventricular short- axis cine-MRI images at end-diastolic and end-systolic phases of the cardiac cycle 1 and 5 wk after MI. Blood in the chambers appears bright. FR-HMGB1 group shows preservation of heart function and maintenance of wall thickness (yellow arrows in figures with color) with visible separation of right and left ventricles (red arrows in figures with color) during systole. In contrast, in the PBS treated group, there is significant left ventricle dilation (white arrows) and very limited contraction between diastole and systole, n = 10 per group. All MRI scans were performed and assessed by a blinded observer. Figure 7J: Plotted mean muscle cross-section area at given time points after BaCI₂ muscle injured animals treated with either PBS (black), 28.75 nM/kg of FR-HMGB1A-c001 (red in figures with color) or dBB12L (green in figures with color). N= 5 per group and timepoint, nested ANOVA (Holm-Sidak post-hoc correction). A representative image of each point is shown in Figure 12.

[0019] Figures 8A-8B show results of a peptide array of CXCL12 peptides interacting with HMGB1. Figure 8A: Peptide array of full-length CXCL12. “+” positions correspond to positive control 10-His peptides; the rest of the peptides comprise CXCL12 15-mers shifted two (2) residues in succession towards the C-terminus. The membrane was exposed to 1 uM HMGB1(FR or 3S)-His6 (1-214), BoxA-His6 (8-78) and BoxB-His6 (94-162) for 24 hours. Bound protein was detected by anti-His-HRP conjugate chemoluminescence. A peptide of CXCL12 interacting with full length HMGB1 cannot interact with either Box A or Box B alone, confirming the requirement of the N-terminal segment of each Box domain (particularly, D4 in box A/D90 in box B): intensity of the spots pertaining to the common CXCL12 peptide is also markedly decreased upon binding to the Box domains alone when compared to FL- HMGB1. Binding to 3S seems to be of higher intensity than that to FR; this is likely due to protein oxidation during the assay, although this was not quantified due to the low concentration of protein used being unsuitable for ESI/TOF MS. BLI data, however, do suggest a lower off rate of CXCL12 from 3S than from FR-HMGB1. Figure 8B: CXCL12 dimer (PDB 2J7Z) with the regions binding HMGB1 highlighted. In original figures with color, Red: shared binding region. Blue: non-shared binding region.

[0020] Figure 9 shows interferograms in BLI of CXCL12 binding to immobilized HMGB1 constructs. Biotinylated HMGB1 constructs were immobilized on streptavidin-coated Octet biosensors and dipped in rising concentration of CXCL12. Interferograms are colored according to CXCL12 concentration (key in top right). Each set of three replicates (cycles) for a given sensor is surrounded by a colored overlay according to construct. FR FL-HMGB1 (c011), black: 3S-FL HMGB1 (c022), purple: FR-HMGB1 Box A 8-78 (c027), brown: FR- HMGB1 Box B 94-162 (c028), pink: FR-HMGB1 Box A 1-88 (c037), grey: FR-HMGB1 Box B 90-162 (c038).

[0021] Figures 10A-10D show NMR validation of residues involved in CXCL12 binding (continuation). Figure 10A: Cumulative CSP of HMGB1 3S 1-184 (HMGBlA-c007) upon addition of 1:2 molar equivalents of CXCL12 in one step (1: 1 HMG Box to CXCL12 ratio). Box A and Box B residues have been considered separate molecules for the purposes of median CSP calculation. With the original figure coloring, intensity of the green color in the bar graph indicates higher relative CSP. The sequence of each HMGB1 construct has been overlaid with the residue number; an empty column (no number) represents residues which could not be mapped in the parallel 3D 1H-15N HSQC/NOE/TOCSY experiments. The sequence of the residues corresponding to each box is in the middle of each condition, colored originally as per: Light blue, residues previously reported in the literature as involved in CXCL12 binding; red, residues weakly involved in the peptide array; purple, residues involved according to both published literature and peptide array. The sequences shown are represented in SEQ ID NOs: 6 and 7. Figure 10B: 15N HSQC-HQMC peak spectra for (A) in 10 mM HEPES 150 mM NaCl pH 7.5 buffer. Protein concentrations are indicated in the spectra overlay. Figures IOC - 10D show 15N HSQC-HQMC peak spectra in 10 mM HEPES 150 mM NaCl pH 7.5 buffer. Protein concentrations are indicated in the spectra overlay. Figure IOC: HMGB1A 94-162 (2- day experiment); Figure 10D: HMGB1A 89-174 (6-day experiment; minor degradation occurs after day 4).

[0022] Figure 11 shows Interferograms in BLI of HMGB1 constructs binding to immobilized Fc-RAGE. RAGE-Fc was immobilized in the surface of AHC sensors and dipped in rising concentration of different HMGB1 constructs. Two experiments were run with different concentration ranges: the three columns of graphs in the left, 0 to 22.22 μM HMGB1 over 9 steps; on the right, 0 to 25 μM over 7 steps. Both are originally color-coded by concentration (top). Colors indicate the specific construct concentration. Each graph corresponds to a single sensor (replicate). Interferograms surrounded by a red rectangle had data points excluded due to poor quality (e.g. drift).

[0023] Figure 12 shows histological images of regenerating muscle in response to FR- HMGB1 (red) or dBB12L (green) compared to PBS control (black) in the originally colored figure. [0024] Figures 13A-13C shows plasmid vector maps. Vector maps with features and restriction sites. TEV: Tobacco etch virus protease recognition site. 6-His: 10/6-histidine residue affinity epitope. FLAG: FLAG affinity epitope. StrepTag: StreptactinXT affinity epitope. SacB: Levansucrase precursor (negative selection in the presence of sucrose). pLIC: Annealing sites for sequencing primers used in colony screening. All plasmids contain kanamycin resistance (50 μg/mL).

[0025] Figure 14 shows mutagenesis of the FR-HMGB1 sequence to generate 3S-HMGB1.

DETAILED DESCRIPTION

Terms

[0026] In order to facilitate an understanding of the invention, except as otherwise expressly provided herein, each of the following terms shall have the meaning set forth below.

[0027] As used herein, “engineered” means a non-naturally occurring compound that has been created based up changing a naturally occurring compound . An engineered compound, e.g. a polypeptide, may include portions of a naturally occurring compound that have been modified or rearranged. Such an engineered polypeptide may also be referred to as an “analogue” or “derivative” of the naturally occurring polypeptide.

[0028] As used herein, “stem cell” means any unspecialized cell that has the potential to develop into many different cell types in the body, including without limitation hemopoietic stem cells.

[0029] As used herein, the term "effective amount" means an amount of a compound that is capable of achieving a desired result, for example, alleviating a condition or the symptoms associated therewith, for example, an acute tissue injury as described herein. The specific dose of a compound administered according to this invention will, of course, be determined by the particular acts associated with the condition, for example, the route of administration, the physiological state of the subject, and the severity of the condition being treated. For example, an engineered HMGB1 protein administered to a subject is preferably in the form of a composition comprising a therapeutically effective amount of the engineered HMGB1 protein.

[0030] The phrase "pharmaceutically acceptable" refers to those compounds, materials, compositions, or dosage forms which are, within the scope of sound medical judgment, suitable for use in contact with the tissues of human beings and animals without excessive toxicity, irritation, allergic response, or other problem or complication, commensurate with a reasonable benefit/risk ratio. The phrase "pharmaceutically acceptable carrier" as used herein means a pharmaceutically acceptable material, composition or vehicle, such as a liquid or solid filler, diluent, excipient, or solvent encapsulating material. The choice of any specific pharmaceutically acceptable carriers is well within the knowledge of those skilled in the art. . Accordingly, there is a wide variety of suitable carries available and routinely used in pharmaceutical compositions. [0031] It should be understood that the terms “a” and “an” as used herein refer to “one or more” of the enumerated components.

[0032] As used herein, all numerical ranges provided are intended to expressly include the endpoints and consistent with context all numbers that fall between the endpoints of range.

Embodiments of the Invention

[0033] This invention provides a polypeptide represented by the following formula:

H2N-A-X-B-A-X-B-HOOC wherein A represents consecutive amino acids, the sequence of which (1) includes a sequence identical to the sequence of amino acids 90 - 93 of wild type HMGB1, (2) has at its amino terminal end, between one and six consecutive amino acids, the sequence of which is identical to the sequence of the corresponding one to six amino acids preceding amino acid 90 in wild type HMGB1, and optionally (3) has a methionine at the amino terminus; wherein X represents consecutive amino acids, the sequence of which is identical to the sequence of amino acids 94 - 162 of wild type HMGB1; and wherein B represents consecutive amino acids, the sequence of which (1) includes a sequence identical to the sequence of amino acids 163 - 168 of wild type HMGB1 and (2) has at its carboxy terminal end, between one and six consecutive amino acids, for example, 1, 2, 3, 4, 5, or 6 amino acids, the sequence of which is identical to the sequence of the corresponding one to six amino acids following amino acid 168 in wild type HMGB1; and wherein each - represents a peptide bond between each of A and X, X and B,

B and A, A and X, and X and B.

[0034] In some embodiments, the methionine is present at the amino terminus of the polypeptide.

[0035] In other embodiments, A has at its amino terminal end, one amino acid corresponding to amino acid 89 of wild type HMGB1. [0036] In some embodiments, B has at its carboxy terminal end, six amino acids the sequence of which corresponds to the sequence of amino acids 169-174 of wild type HMGB1.

[0037] This invention also provides a composition comprising the polypeptide of any one of the provided embodiments and a carrier.

[0038] In some embodiments, the polypeptide is present in a therapeutically or prophylactically effective amount and the carrier is a pharmaceutically acceptable carrier.

[0039] This invention also provides methods of treating a subject suffering from, or at risk for developing, a condition which would be alleviated by promoting regeneration of a tissue or cells that rely upon CXCR4⁺ cells for repair which comprise administering to the subject the polypeptide of any one of the provided embodiments in an amount effective to promote regeneration of the tissue or cells, that is, a therapeutically or prophylactically effective dose of the pharmaceutical composition of the invention.

[0040] In certain embodiments, the condition is myocardial infarction and the tissue is a cardiac tissue, particularly, myocardium.

[0041] In a currently preferred embodiment, the polypeptide is administered within 5 hours, preferably within 4 hours, more preferably within 3 hours, even more preferably within 2 hours and most preferably within 1 hour of the myocardial infarction.

[0042] In some embodiments, the condition is a fracture and the tissue is a bone.

[0043] In other embodiments, the condition involves liver damage and the tissue is liver tissue.

[0044] In yet further embodiments, the condition involves damage to the brain or nervous system and includes stroke, Parkinson’s disease and dementia.

[0045] In some embodiments, the condition involves damage to the lung.

[0046] In yet further embodiments, the condition involves the gut and includes surgery and inflammatory bowel disease.

[0047] In some embodiments, the condition involves damage to the skin and includes surgical procedures, bums and ulcers.

[0048] In additional embodiments, the condition involves the pancreas including type 1 diabetes and the cells are islet cells. [0049] In further embodiments, the condition is neutropenia, for example, neutropenia following chemotherapy and the tissue is bone marrow.

[0050] In some embodiments, the condition is kidney failure and the tissue is kidney tissue.

[0051] Further non-limiting details are described in the following Experimental Details section which is set forth to aid in an understanding of the invention but is not intended to, and should not be construed to, limit in any way the scope of the invention disclosed.

EXPERIMENTAL DETAILS

RESULTS

Identification of amino acids and motifs in HMGB1 involved in CXCL12 binding

[0052] Before considering which residues of FR-HMGB1 can be mutated to eliminate proinflammatory signaling, it is essential to map the amino acids and motifs involved in binding to CXCL12. Peptide SPOT arrays were used, [40] where overlapping peptides covering the sequence of one of the target proteins (HMGB1) are assessed by immunoblotting for their capacity to bind CXCL12 to identify key sequences involved in binding. Coarse peptide arrays with human HMGB1 peptides binding his-tagged CXCL12 highlighted a specific pattern of homologous sequences between Box A and Box B that bind CXCL12. Two main binding sites were identified by the clustering of CXCL12 binding peptides over the HMGB1 sequence (Figure 2A and Figure 2B). The first encompassed the initial one and a half α-helices of the HMG Box, overlapping the glycyrrhizin binding site [41]. The second was located at the C- terminal half of the third α-helix. In each HMG Box, the first CXCL12 binding peptide (helices I and II) appeared to be the most involved in CXCL12 binding as the intensity of CXCL12 binding to peptides from this segment was much higher. Binding of Box B peptides to CXCL12 appeared to be slightly weaker compared to Box A based on the intensity of the immunoblots.

[0053] To further delineate the importance of individual residues, a second peptide array was generated in which each amino acid within the CXCL12 binding peptides were substituted to alanine and their effect on CXCL12 binding was assessed. The aim was to pinpoint amino acid(s) on the peptides that directly contribute to CXCL12 interaction. This confirmed that several residues are critical for CXCL12 binding as their substitution to alanine altered the intensity of bound CXCL12 (Figure 2C and Figure 2D) in otherwise homologous peptides. Contrary to published data where the interaction with CXCL12 was shown to only involve the helical segments of the HMG Boxes [42-44], we found that part of the flexible N-terminal flanking region to each box (D-4-P-X-X-₁), as well as C-terminal residues (lie 78-Pro 80 for Box A and Ala 163-Asp168 for Box B), were involved in the interaction with CXCL12 (Figure 2E). This was confirmed with a reverse peptide array (Figure 9) of CXCL12 peptides. Full- length FR and 3S- HMGB1 bound to peptides containing sequences covering the whole of the b-sheet of CXCL12, whereas HMGB1 Box constructs alone (8-78 Box A) or (94-162 Box B), without the flanking flexible region [45], only interacted with the peptides covering the N- terminal strand of the b-sheet. [0054] We hypothesized that if these flexible regions adjacent to the HMG Boxes were involved in CXCL12 binding, their absence should significantly alter the interaction of HMGB1 with CXCL12. Using biolayer interferometry (BLI), we assessed binding of CXCL12 to HMGB1 constructs comprising each of the HMG Boxes with (full HMG Box constructs: HMGB1 1-88 for Box A, HMGB1 89-174 for Box B) [46] or without the flexible flanking residues (helical-only HMG Box constructs; HMGB1 9-78, HMGB1 94-162), full-length FR- HMGB1 and non-oxidizable (3S)-HMGB1, which shares the CXCL 12-binding properties of the wild-type protein [ 17] . As expected, binding of CXCL 12 to HMGB1 was greatly influenced by the presence of these flexible regions (Figure 2E and Figure 2F). Helical-only constructs had decreased CXCL 12 affinity and binding capacity compared to full HMG Box constructs with intact flanking regions as evidenced by the increased dissociation rates (k₀ff) of the helical- only constructs without the flanking regions. In contrast, affinities of CXCL 12 for full-length HMGB1 (FR or 3S) and full HMG Box constructs were comparable to each other and higher than for the helical -only constructs.

[0055] Our peptide arrays showed that the residues involved in binding CXCL12 were clustered in two peptides: one across α-helix I and the N-terminal half of II, and a second over the C-terminus of α-helix III. Both peptides included the flexible flanking regions at the N- and C-termini of each HMG Box. This pattern was mirrored across the HMG A and B Boxes (Figure 2B, Figure 2D, Figure 2E). BLI kinetic data confirmed that the absence of the flanking regions in single HMG Box domains resulted in destabilization of the recruited CXCL12, which did not remain bound to the helical only constructs (Figure 2F).

CXCL 12 binds to a concave pocket on the underside of each HMG Box as shown by peptide array and NMR

[0056] Next NMR was used to confirm from a structural perspective the amino acid residues in HMGB1 involved in CXCL12 binding. Residues involved in CXCL12 binding would result in NMR signal changes upon formation of the complex. We used FR-HMGB1 94- 162 and 89- 174 to represent HMG Boxes with and without the flanking regions, as well as 3S HMGB1 1- 184. Non-oxidizable 3S-HMGB1 was used as our full-length construct as the time required for obtaining a full set of 3D spectra at 750 MHz (¹⁵N HSQC-TOCSY/NOESY and associated ¹⁵N HSQC spectra) would result in oxidation of FR-HMGB1, altering both the peak resonances and the interaction with CXCL 12. [0057] CXCL12 titration ofHMGBl Box B 94-162 (Figure 3A) resulted in changes inNMR signal, either cumulative chemical shift perturbation (CSP) or peak height changes (I/Io), of several residues at both the C-terminal and N-terminal binding regions identified in the peptide arrays. However, several residues (A100, 1112, L119, A136, Y154, D157, 1158) identified in our peptide arrays and also documented in the literature [43,44] showed no CSP or volume changes. For the Box B construct that included the flanking regions, both the median CSP and mean volume change were significantly higher. Several residues (D90, G165, K166) in these flanking regions were found to be involved in CXCL 12 binding, in accordance with our peptide array data. Residues not affected by CXCL 12 binding in the helical -only (94-162) construct, such as Y154, D 157 and 1158, and which were identified as being involved in CXCL12 binding in the peptide arrays demonstrated significant CSP in the full HMG Box construct, indicative of improved binding when the flanking regions were present. In addition, we observed CSP changes for residues not identified in the peptide arrays (A147, M131, A169, K172, G173) in the construct with the flanking regions. Other residues that have not been previously identified but were flagged as being potentially important in the peptide arrays did not display either CSP or volume changes upon addition of CXCL12 (C105, E107, Y108). This group of residues was, therefore, reclassified as being not critical for CXCL 12 binding.

[0058] When the NMR experiment was repeated with 3 S-HMGB1 1-184, CXCL 12 addition resulted in CSP changes below the threshold for detection due to the signal to noise ratio. Surprisingly, in contrast to published data [47], residues A100, II 12, LI 19 and A136 in Box B of 3 S-HMGB1 1-184 failed to produce significant CSP or volume changes, although residues (S99, K113) very close to some of these were affected. Within this construct, residues corresponding to HMG Box B showed weaker CSP changes for residues involved in interaction with CXCL12 compared to those in Box A. This could represent a more fluid equilibrium of CXCL12 binding, as suggested by higher association and dissociation rates for Box B (Figure 2G). This allowed us to also denote residues such as H30, D32 or S34 as false positives from the peptide array (Figure 11). Whilst K89 in Box B 89-174 showed high CSP, this is the third residue from the N-terminus (after the TEV cleavage leftover N-terminal residues - Ser-Met ). In the experiments with 3S HMGB1 1-184, where it is in the middle of the flexible linker, it did not demonstrate increased CSP changes. Therefore, the changes associated with this residue in Box B alone are likely related to its position at the N-terminus potentially permitting high conformational flexibility. Design of a double Box B HMGB1 construct that retains CXCL12 binding whilst eliminating nroinflammatory signaling

[0059] We combined the data from the peptide arrays, alanine substitution, and NMR, and mapped them onto the NMR structure of HMGB1 (Figure 3A, PDB 2YRQ) to identify the residues involved in binding CXCL12. We found that the CXCL12 binding residues occupy the concave side of the HMG Box domains, forming a CXCL12 binding pocket on the underside of each HMG Box. These two pockets (one on each HMG box) also contain the described binding site for glycyrrhizin, a competitive inhibitor of HMGB1-CXCL12 binding [41], between Helix I and II on each Box. HMG Box A and B are each capable of binding one CXCL12 monomer with equivalent affinity. We also recognized that the distribution of peptides within each HMG Box that interact with CXCL12 is similar between both HMG Boxes, and form almost identical binding pockets (Figure 4A).

[0060] Having identified the sequence determinants for binding CXCL12, we next designed HMGB1 constructs which preserved these interaction surfaces but altered sequences to attenuate proinflammatory signaling. Based on the fact that each HMG Box can bind to a CXCL12 monomer independently and the requirement of both Box A (oxidized) and Box B for TLR-4 [48] and potentially for the prothrombotic RAGE [14,22] and TLR-2 [22,33] signaling activities, we hypothesized that an HMGB1 construct where Box A is substituted by another Box B (i.e. 1-88 replaced by 89-174) would not signal through TLR-2, TLR-4 or RAGE. In addition, we deleted part of the RAGE-binding sequence in Box B (175-184) to further decrease the affinity for this receptor. Replacing the Box A sequence for Box B also replaces the LPS glycan-binding peptide for a copy of the LPS lipid A binding peptide [13], which would further hamper the proinflammatory activities of HMGB1 [49] and LPS transfer to TLR-4/MD-2 in a manner analogous to LBP. This engineered construct dBB12L (Figure 4B) comprised of the following segments of the native HMGB1 protein: flexible N-terminal region (from HMGB1 89-93), first Box B (from HMGB1 94-162), 12-residue linker C-terminal of native Box B (from HMGB1 163-174), and second Box B (from HMGB1 94-162). The linker length of 12 residues in this dBB12L is similar to the 10 amino acids in the linker of native HMGB1. This small increase in linker length was due to our preserving residues 172 and 173 which showed changes in CSP on CXCL12 binding.

[0061] We compared the thermostability of dBB 12L and wild-type HMGB1 using Dynamic Scanning Fluorimetry (DSF), and solvent accessible surface area (SASA) by both native mass spectrometry (native ESI/MS) and size exclusion chromatography (SEC). dBB12L had similar stability and surface charge profiles as FR-HMGB1 1-164, which also contains two HMG Box domains and no C-terminal acid tail. Thermostability trends were similar for all constructs (Figure 5A), with lower Tm₅₀ at pH nearing the isoelectric point (9.9 for tail-less constructs as dBB12L or 1-164, and 6 for FL-HMGB1) and they were all equally stable in conditions relevant to clinical practice (PBS, purification buffers, and saline solution), with Tm₅₀ ≈50°C. However, we observed different optimal ionic strength ranges for FR-HMGB1/FR-HMGB1 1- 164 compared to dBB12L. In native ESI/MS, all three HMGB1 constructs had similar charge state distributions, with a compact monomer as the main species and a higher SASA extended monomer (Figure 5B and Figure 5C).The extended monomer was more prevalent for tail-less constructs or at higher ionic strength. Average monomer SASA values observed in native ESI/MS were in agreement with those derived from SEC or published NMR structures (PDB 2YRQ; HMGB1 1-164). Compact FL-HMGB1 monomer SASA matched computational models of FL-HMGB1 in water [44]. Storage up to 180 days did not affect SEC profiles (always monodisperse at equal RV), degradation, or aggregation (Figure 5D). This suggests that the conformations observed in ESI/MS are representative or the native folding ofHMGBl constructs, and that dBB 12L is not significantly different from an HMGB1 construct with two HMG Boxes alone. dBB12L construct has greatly reduced affinity for RAGE and cannot signal through TLR-2 or TLR-4

[0062] We then evaluated whether dBB12L had decreased TLR-2, TLR-4 signaling and RAGE binding, whilst preserving HMGB1 -mediated regeneration. Due to the lack of an established signaling assay for RAGE, we assessed the binding of RAGE to HMGB1 using real-time kinetics (BLI) and an endpoint assay (ELISA). ELISA-based affinity measurements (Figure 6A) showed that 3S-, FR- and DS-HMGB1 at equilibrium bound similar amounts of RAGE, with DS-HMGB1 and 3S-HMGB1 having significantly higher apparent affinity than FR-HMGB1. In contrast, dBB12L did not to bind RAGE in this assay. Three additional HMGB1 constructs were tested, including DS-HMGB1 1-184, which has an intact RAGE binding peptide and oxidized Box A but no acidic tail and therefore has all the requisites for RAGE binding; DS-HMGB1 1-164, which lacks a significant portion of the RAGE binding peptide but retains an oxidized Box A; and DS-Box A alone. We found that DS-HMGB1 1- 184 bound RAGE, but with reduced capacity and affinity compared to full-length DS-HMGB1. By comparison, DS-HMGB1 1-164 had greatly diminished RAGE binding capacity compared to full-length DS-HMGB1 but still higher than dBB12L, whilst DS Box A 1-88 (full HMG Box construct with flanking regions) was unable to bind RAGE.

[0063] Kinetics analysis using BLI confirmed the ELISA results (Figure 6B), with two exceptions. In BLI (Figure 6C) DS-HMGB1 1-184 had a much higher RAGE binding affinity than all other constructs, albeit with also slightly faster dissociation rate, compared to ELISA where it had lower affinity than DS-HMGB1. 3S-HMGB1, whilst binding equivalent amounts of RAGE to DS- or FR- HMGB1 , had similar affinity to FR-HMGB1 but much slower overall kinetic rates, whereas in ELISA it had affinity and binding capacity for RAGE equivalent to DS-HMGB1. Binding of all other HMGB1 constructs to RAGE in BLI reproduced the results from the ELISA. The higher RAGE affinity of DS-HMGB1 compared to FR in both assays was due to a much faster association rate (k_on), whereas dissociation rates ( k_off) were nearly identical for these two redox forms. In contrast, dBB12L, which had an association rate closer to DS-HMGB1, exhibited very unstable binding with a very high dissociation rate. DS- HMGB1 1 - 164 also had a faster RAGE binding equilibrium with overall lower binding affinity than full length DS-HMGB1, although with higher affinity than dBB12L-HMGBl, The deletion of both the final 10 residues in Box B (175-184) and the disulfide bridge in Box A (by substituting it with Box B) in dBB12L resulted in unstable binding of RAGE.

[0064] HMGB1 binds TLR-2, TLR-4, and RAGE and signaling from all receptors converges to the NF-kβ pathway [25]. Consequently, it is difficult to attribute downstream proinflammatory cytokine production to a given receptor. Therefore, we first evaluated TLR- specific signaling using NF-kβ reporter cell lines engineered to express either TLR-2 or TLR- 4 and their co-receptors. Disulfide HMGB1 promoted NF-kB signaling via TLR-2 (Figure 6D) and TLR-4 (Figure 6E). In contrast, dBB12L failed to signal in either cell type. Next, we confirmed the effects of the various HMGB1 constructs on primary human monocytes. It has been reported that DS-HMGB1 synergizes with TLR-2 ligands such as LTA to promote proinflammatory signaling [34]. We confirmed that DS-HMGB1 acted synergistically with LTA to promote higher TNF production than LTA or DS-HMGB1 alone. In contrast, dBB12L or FR-HMGB1 did not exhibit this synergistic effect and alone failed to elicit TNF secretion greater than media alone (Figure 6F). No synergistic response has been described with DS- HMGB1 and the TLR-4 ligand LPS. When combined with LPS, DS-HMGB1 promoted TNF expression by primary human monocytes to the same extent as LPS alone. In contrast, FR- HMGB1 or dBB12L alone did not promote TNF production. However, when combined with LPS, FR-HMGB1 or dBB12L reduced TNF expression compared to LPS alone. [0065] Taken together, these data show that dBB12L does not signal via TLR-2 or TLR-4, even in the presence of their cognate ligands and it has greatly decreased affinity for RAGE.

The dBB12L construct has pro-regenerative activity comparable to that of FR-HMGB1

[0066] Distant injury has been previously shown to transition stem cells to G_Alert [12]. Therefore, we first compared the transcriptomic response of skeletal muscle stem cells of FR- HMGB1 to injury to the contralateral limb. The genes up- and down-regulated by FR-HMGB1 or distant injury were remarkably similar (Figure 7A), with the major pathways upregulated being those associated with G_Alert [7,12], including mitochondrial metabolism, oxidative phosphorylation and cell cycle (Figure 7B). Interestingly, CXCR4 was one of the most highly upregulated genes.

[0067] Next, we determined the optimal in vivo treatment dose for FR-HMGB1 using a validated murine model of skeletal muscle injury [7,12]. We found that 0.75 mg/kg (29 nmol/kg) resulted in the maximal response, with no further improvement in regenerative activity with higher doses (Figure 7C). We also assessed the optimal time for administration in vivo after injury. FR-HMGB1 was found to be effective in promoting repair when injected up to 5 hours post-injury (Figure 7D). Thereafter, there was no improvement. We then looked at the half-life of FR-HMGB1 in the circulation following iv administration. We found that there was an initial rapid clearance (ti/2 ≈ 11 min) followed by subsequent slower clearance rate (ti/2 ≈ 120 min) (Figure 7E). This would be consistent with the half-life of 25 min in humans [50], with the protein being cleared by binding to haptoglobin [51,52].

[0068] We have previously shown that FR-HMGB1 accelerates regeneration of skeletal muscle, bone and blood following injury by promoting the transition of stem and progenitor cells to G_Alert [7]. There are no significant stem cells in the mammalian heart and the majority of new cardiomyocytes following injury are derived from existing cardiomyocytes [53]. Therefore, we assessed whether administration of FR-HMGB1 would promote cardiac regeneration. We found that iv injection at the time of myocardial infarction resulted in enhanced survival (83% in mice treated with FR-HMGB1 compared to 52% in PBS controls) (Figure 7F). FR-HMGB1 resulted in approximately 60% reduction in infarct size as assessed by serial MRI scans over 5 weeks (Figure 7H) and 16% improvement in overall left ventricular ejection fraction (Figure 7G).

[0069] Finally, we assessed the efficacy of dBB12L compared to FR-HMGB1 in promoting skeletal muscle regeneration in vivo. Mice injected with optimal doses (29 nM/kg) of FR- HMGB1 or dBB12L exhibited accelerated regeneration equally following injury (Figure 7J), as determined by an increase in the mean cross-sectional area of regenerating muscle fibers with centra nuclei [7,12]. This was most apparent at day 14, as previously described for FR- HMGB1 [7].

DISCUSSION HMGB1 needs to be modified in order to be used as a tissue repair therapeutic

[0070] Therapies based on administration of exogenous stem cells to promote repair of solid organs have failed to deliver on the initial promise [6,54], and killed cells are just as effective by triggering an immune response [55]. An alternative, potentially more rewarding approach, would be to target endogenous regenerative repair processes, including resident stem and progenitor cells [56,57]. Inhibition of prostaglandin dehydrogenase is a promising approach [58], although progress through to clinical translation has been slow [59]. Administration of growth factors has also been described [60,61] but is limited by in vivo proteolysis [62]. Currently, there is no approved therapeutic for promoting regeneration and accelerating repair of multiple tissues.

[0071] We previously showed that exogenous administration of FR-HMGB1 is effective in accelerating regeneration of bone, skeletal muscle and blood by transitioning resident stem and progenitor cells to G_Alert, when they are able to readily respond to the appropriate activating factors released on tissue injury to effect repair [7]. Whilst most of the systemic DS-HMGB1 detected in patients after trauma is secreted in a second release event [50], there is accumulating evidence supporting the conversion of FR-HMGB1 into DS-HMGB1 in vivo locally at the site of injury [18,19]. Therefore, the potential for deleterious proinflammatory signaling precludes the use of native FR-HMGB1 as a therapeutic.

[0072] Disulfide HMGB1 can signal through TLR-4 [24,30], TLR-2 [24,34,35] or RAGE [14,63] to induce the expression of proinflammatory cytokines. TLR-4 signaling by DS- HMGB1 results in production of several proinflammatory cytokines, including TNF [28], whilst TLR-2 signaling has been shown to be detrimental in multiple processes, including thrombosis and reperfusion injury [64], and autoimmune disorders [33]. DS-HMGB1 signaling via RAGE plays a key role in platelet activation and NET formation by neutrophils to promote thrombus formation [23,64-66]. All three receptors ultimately converge on NF-kβ [25,67], leading to synergistic expression of proinflammatory cytokines via all three receptors [68,69]. Therefore, development of HMGB1 as a therapeutic is crucially dependent on engineering the molecule to eliminate signaling via all three receptors.

Box A and Box B bind CXCL12 independently due to a shared peptide pattern

[0073] The regenerative activities of FR-HMGB1 are critically dependent on the formation of a heterocomplex with CXCL12 and signaling via CXCR4. Whilst it is known that CXCL12 binds to HMG Boxes [42,70], the structural motifs involved remain unknown, with few residues proposed [47,71]. It is also unknown whether this signaling involves homodimers of CXCL12 via the CXCR4 axis, including enforcing quiescence of hemopoietic stem cells, or CXCL12 monomers, which promote chemotaxis [72-74].

[0074] Peptide arrays allowed us to identify the residues [75] involved inHMGBl-CXCL12 interaction. We identified a common pattern of two peptide binding regions for CXCL12 present in both Box A and Box B. The first peptide region extends from the N-terminal flexible segment into half of Helix II, overlapping with the glycyrrhizin binding site [41], and the second on the C-terminal portion of helix III in each HMG Box. We confirmed the importance of these residues using alanine mutation and also the importance of the flanking flexible regions by BLI, their removal resulting in a dramatic increase in the dissociation rate of CXCL12. BLI also confirmed that each Box can bind a CXCL12 monomer independently with similar affinity, without cooperativity between the Boxes. Furthermore, our use of HEPES buffer prevented CXCL12 dimerization [76], allowing us to conclude that each Box is able to bind monomeric CXCL12, and dimerization of CXCL12 is not a requisite for complex formation. This model is consistent with the proposed mechanism of signaling of the HMGB1-CXCL12 heterocomplex via CXCR4 [72].

[0075] We then used NMR to confirm the critical role of the peptide regions identified by the peptide arrays, including the flexible flanking regions whose absence resulted in a decrease in CXCL12-induced changes upon binding to either full or helical only HMG Box B constructs. This is consistent with the faster dissociation rates observed in BLI for the helical only HMG Box B construct compared to the one containing flanking regions. We also observed weaker signals (peak broadening) for Box B compared to Box A, supporting the faster exchange binding equilibrium observed in BLI for full-length Box B compared to Box A and also the weaker signal for Box B CXCL 12-binding sequences in the peptide arrays. Some residues demonstrated shift changes in NMR despite not being flagged in the peptide arrays. These may represent sequence-independent contribution to binding such as main chain interactions or relayed effects of other residues binding CXCL12 and affecting nearby positions in turn.

[0076] When we superimposed the residues involved in CXCL12 binding determined by both NMR and peptide arrays on the structure of HMGB1 1-164 (PDB 2YRQ), we identified pockets within each Box domain where the side chains of the residues all lined the center of the pocket. The residues identified both with alanine scan and NMR form a concave surface on Box B which also includes the glycyrrhizin binding site (Figure 3B, dotted circles). Conversely, several of the residues only showing changes in NMR have side chains pointing outwards of this pocket, suggesting a sequence-independent binding to CXCL12 (main-chain mediated) or that they are affected by indirect chemical environment changes in other parts of the HMG Box upon CXCL12 binding.

[0077] With this detailed understanding of CXCL12 binding, we were able to design a construct to eliminate proinflammatory signaling via TLR-4, TLR-2 and RAGE. This construct consisted of two HMG Box B domains in tandem separated by a linker of similar length to that of wild-type HMGB1 (dBB12L), which is unable to oxidize as Box A has been replaced by Box B but still should bind CXCL12 as two HMG Boxes are present. dBB12L was as stable as 1-164 FR-HMGB1 or full length FR-HMGB1, being equally thermostable in clinically relevant buffer conditions (PBS, saline solution), with no aggregation or degradation on storage for prolonged periods of time. The surface area and charge profile of dBB12L also was like that of HMGB1 1-164, with monodisperse profiles in SEC and charge distribution in native ESI/MS. This similarity reflects a similar conformation in solution between dBB12L and a wild-type HMGB1 construct with two HMG Boxes and the linker region (FR-HMGB1 1-164) but without the acidic tail.

The engineered dBB12L construct does not signal via TLR-2 or TLR-4 or bind RAGE whilst retaining full pro-regenerative properties

[0078] Binding of HMGB1 to TLR-4/MD-2 is well-described [29] . Oxidized Box A initiates the binding of DS-HMGB1 by interacting with TLR-4 and Box B stabilizes this interaction by binding MD-2; the FCSE motif within Box B has been shown to be essential for signaling [30] . Whilst DS-HMGB1 can signal on its own via TLR-4/MD-2, it can also facilitate signaling via LPS by substituting for LPS-binding protein (LBP) which binds LPS and promotes transfer and recognition to TLR-4/MD-2 [13]. Deletion of Box A in our dBB12L construct effectively precluded signaling via TLR-4. Interestingly, we observed that both FR-HMGB1 and dBB 12L decrease TNF expression by monocyte in response to LPS. This could be due to these proteins binding LPS but are unable to transfer it to TLR-4/MD2 due to the lack of oxidized Box A, unlike DS-HMGB1, which can effectively substitute for LPS binding protein (LBP) [13] present in the serum of the culture medium [77] .

[0079] The interaction between HMGB1 and TLR-2 has been less well described and there is some controversy as to whether HMGB1 can induce TLR-2 signaling on its own [24,36] or requires a co-ligand to induce activity, and also whether this response is dependent on the redox state of the protein [34,35]. The published data where HMGB1 alone was unable to induce TLR-2 responses were performed either in absence of serum [35], or with low HMGB1 concentrations [34], suggesting that HMGB1 alone retains some signaling capacity through TLR-2 but a co-ligand is required to induce higher levels of response. It is known that binding must involve at least one HMG Box [33] and that the acidic tail negatively regulates binding to TLR-2 [35]. We found that DS-HMGB1 alone was able to signal via TLR-2 and the effect was enhanced by the presence of LTA in serum-containing media. However, there was no response to FR-HMGB1 or dBB12L, which did not synergize with LTA. This would suggest that, as with TLR-4, oxidized Box A with a disulfide bridge is necessary for TLR-2 mediated responses and that TLR-2 co-ligands synergize with DS-HMGB1, potentially by displacing the acidic tail to promote TLR-2 interaction.

[0080] The RAGE binding site within HMGB1 (residues 150-183) has been previously described [14]. A similar motif is also present in other RAGE ligands such as S100 proteins, and homologous peptides to these sequences are effective antagonists of HMGB1 -mediated RAGE signaling [37,78]. The acidic tail of HMGB1 shares residues with the RAGE binding peptide [15] and has been proposed as a regulator of RAGE interaction, analogous to its role in TLR-2 binding. However, only the disulfide form of HMGB1 has been linked specifically to prothrombotic activities via RAGE [22]. We found that constructs lacking the RAGE binding peptide, including dBB 12L, were unable to bind RAGE in the ELISA assay. In contrast DS-HMGB1, and interestingly also 3S-HMGB1, were able to bind RAGE better than FR- HMGB1. When we analyzed the kinetics of the interaction using BLI we found that, in accordance with our ELISA data, dBB12L had lower affinity for RAGE compared to FR- HMGB1 and DS-HMGB1. Whilst dBB12L had a three-fold higher association rate than FR- HMGB1, potentially due to the presence of two partial RAGE binding domains in this construct, the dissociation rate was five-fold greater, indicative of overall unstable, weaker binding. The extensive washes associated with ELISA magnify the effect of the dissociation rates, leading to the different behavior of the HMGB1 constructs in each assay, and represent the binding status at equilibrium. For instance, 3S-HMGB1, which has an affinity for RAGE similar to FR-HMGB1 in BLI, shows a much higher apparent affinity in ELISA due to its dissociation rate being much lower than that of FR-HMGB1 or DS-HMGB1, resulting in RAGE remaining bound to it. This increased RAGE binding may in part account for the increased fibrosis seen in mouse models of myocardial infarction compared to controls, whereas FR-HMGB1 promoted regeneration and improved function [32]. Using SPR others have also shown that 3S-HMGB1 likely binds RAGE with very high affinity, although comparison was not made with FR-HMGB1 [64] . We found that the affinity for RAGE for DS- HMGB1 using BLI (K_d = 0.2-1.3 μM) was similar to that previously reported using SPR (0.1 [79] - 0.65 μM [22]). Our BLI data also show that loss of the acidic tail increases the affinity of HMGB1 for RAGE, although it results in a less stable binding due to higher dissociation rates, whereas truncation of the RAGE binding peptide or reduction of Box A greatly increases dissociation rates, resulting in unstable binding. Interestingly, oxidized Box A alone is incapable of binding RAGE, suggesting that the interaction is likely initiated by the RAGE binding peptide. The absence of Box A together with truncation of the RAGE binding peptide in dBB 12L results in greatly reduced RAGE affinity, and reduced capacity to retain the RAGE once bound, due to the faster dissociation rates when compared to either FR- or DS-HMGB1.

[0081] We found that the transcriptomic changes induced by FR-HMGB1 in skeletal muscle stem cells are very similar to those induced by distant injury and consistent with upregulation of pathways of mitochondrial metabolism, oxidative phosphorylation and cell cycle described G_Alert [80,81]. Upregulation of CXCR4 expression by HMBG1 would potentially enhance its effects. Our data showing that FR-HMGB1 is only effective if administered up to 5 hours post injury would be consistent with it acting by transitioning stem cells to G_Alert. At later time points the stem cells will have been activated and hence cannot enter G_Alert. Muscle stem cells have been shown to migrate to the site 5-6 hours after injury and start to actively divide at around 12 hours [82]. We confirmed that dBB12L retains regenerative activity in vivo equivalent to FR-HMGB1. Importantly, we found that FR-HMGB1 administered intravenously at the time of myocardial infarction resulted in improved survival, reduction in infarct size and improved left ventricular ejection fraction. Based on these data we would predict that administration of dBB12L would also promote regeneration of tissues that rely on stem cells for repair such as bone, skeletal muscle and blood, as well as tissues where regeneration is predominantly reliant on mature cell populations such as cardiomyocytes in the heart. We also predict that dBB12L is likely to be effective if administered up to 5 hours after injury. This is important as the median time for admission to hospital following MI in the USA is 3 hours [83]. Approximately 800,000 people in the USA suffer from myocardial infarction every year [83] and approximately 20-30% go on to develop cardiac failure. Despite US healthcare expenditure for heart failure of >30 billion in 2012, projected to increase to $70Bn by 2030, 5-year survival is only ~60%, which is worse than most cancers [83]. Based on our data we predict that administration of dBB12L within 5 hours of the event will enhance survival of patients experiencing a myocardial infarction, especially ST-elevation myocardial infarction, and through reduction in infarct size and preservation of ejection fraction, reduce the incidence and severity of cardiac failure. Others have shown in mouse [32,84,85] and sheep [86] models that direct injection of FR-HMGB1 into the myocardium in the peri -infarct area 4 hours after infraction is effective in promoting cardiac repair. Our data demonstrating the efficacy of iv administration are important as this route is readily applicable for clinical use.

[0082] In conclusion, the sites of HMGB1 critical for CXCL12 binding were mapped and a construct (dBB12L) that does not signal via TLR-2 or TLR-4 and fails to effectively bind RAGE was designed. FR-HMGB1 transitions stem cells to G_Alert in a manner similar to distant injury despite a short half-life and is effective if administered up to 5 hours after injury. Furthermore, dBB12L promotes tissue regeneration in vivo as effectively as FR-HMGB1. Accordingly, dBB12L can be developed for clinical translation.

SUMMARY

[0083] Reduced High Mobility Group Box 1 (HMGB1) protein binds to CXC Ligand 12 (CXCL12) and signals through CXC Receptor 4 (CXCR4) to promote tissue and regeneration and accelerates repair by transitioning stem and progenitor cells to GAien. However, local conversion of FR-HMGB1 to the disulfide form (DS-HMGB1) may result in deleterious inflammation through signaling via Toll-Like Receptors 2 (TLR-2) and 4 (TLR-4), and the Receptor for Advanced Glycation End Products (RAGE). Therefore, before considering administration of HMGB1 to promote tissue regeneration in clinical practice, it is important to engineer the molecule to eliminate these potentially deleterious proinflammatory effects.

[0084] We have identified the residues involved in formation of the HMGB1-CXCL12 heterocomplex using a combination of peptide arrays, biolayer interferometry and nuclear magnetic resonance. Combining these data with the available literature on the sites of interaction with the proinflammatory receptors, we designed a construct comprising two HMG B Boxes in tandem (dBB12L) which has similar stability and conformation to that of a wild- type fully-reduced HMGB1 construct without the C-terminal acidic tail. As shown herein, dBB12L does not signal through TLR-2 or TLR-4, even in the presence of their co-ligands, and has greatly decreased RAGE binding. Furthermore, the dBB12L construct retains regenerative activity equivalent to FR-HMGB1 in vivo.

[0085] A comprehensive review of the patent landscape and the scientific literature has identified U.S. Patent Application Publication US 2015/0203551 Al, which describes the substitution of cysteines with serines to prevent TFR-4 signalling; however, this construct has been shown to lead to excessive cardiac fibrosis following MI (17). Furthermore, this construct has a slower dissociation for RAGE compared to FR-HMGB1, resulting in RAGE remaining bound for longer after equilibrium, as shown herein in Figure 6B); therefore these substitutions were avoided in our constructs. U.S. Patent Application Publication US 2009/0069227 A9 stipulates that HMGB1 constructs that promote stem cell migration and proliferation must include amino acids 1-187 (0-186 in our data, with 0 being the N-terminal Met) and U.S. Patent 9,623,078 refers to peptides limited to amino acids 1-44 (0-43 for our data) for cardiac regeneration. U.S. Patent Application Publication US 2009/0202500 Al discloses methods for tissue repair but only refers to full-length (1-215) wild-type HMGB1 (0-214 for our data). The dBB12L construct presented herein has no RAGE binding or TFR-4/2 signalling, is 177 amino acids long, and includes amino acid substitutions that have not been previously described. Therefore, the constructs presented herein do not fall within the scope of the prior art.

CFINICAF APPLICATIONS

[0086] This invention provides polypeptides and methods to harness endogenous regenerative processes to enhance tissue repair. The polypeptides function similarly to fully reduced wild type HMGB1 which promotes tissue regeneration by forming a heterocomplex with two CXCL12 molecules, which in turn signals via CXCR4, likely two adjacent CXCR4 receptors on the cell surface.

[0087] Our data show that a polypeptide of the invention (dBB12L) acts in a similar way. Therefore, it is contemplated that dBB12L will promote the regeneration of tissues that rely on CXCR4+ cells for repair. Such tissues include tissues where repair is primarily dependent on stem and progenitor cells, such as skeletal muscle and the haemopoietic system, as well tissues where repair is largely dependent on existing mature cells, e.g., cardiomyocytes in the adult mammalian heart.

[0088] Potential clinical indications:

[0089] Heart following myocardial infarction. This indication is ripe for a clinical trial. Globally, ischemic heart disease affects 153 million people (101), with the loss of >105,000,000 Disability Adjusted Life Years in 2017 (102). Every year 205,000 people in the UK (103) and 805,000 in USA suffer from myocardial infarction (MI), 38% of them experiencing ST-elevation MI (STEMI) (101). Following MI, approximately 30-40% of individuals develop heart failure, affecting 38 million worldwide. Despite US healthcare expenditure for heart failure of >$30 billion in 2012, projected to increase to $70Bn by 2030, 5 -year survival is only -60%, which is worse than most cancers (101). The main target population are patients following MI, especially those at risk of developing heart failure (104). A novel therapeutic that limits cardiac damage, promotes regeneration following MI and prevents the development of heart failure would dramatically reduce morbidity and mortality, and massively reduce healthcare burden. Definitive data using an established (105-108) permanent ligation murine MI model that reliably leads to cardiomyocyte necrosis (109) show that a single iv dose of FR-HMGB1 at the time of injury leads to enhanced survival [83% for animals treated with FR-HMGB1 compared to 52% in the group treated with PBS (placebo], and compared to controls, -16% improvement of absolute cardiac ejection fraction and -60% reduction in infarct size compared to PBS controls over 5 weeks (Figure 7F).

[0090] In a skeletal injury model, the optimal dose of FR-HMGB1 was 0.75 mg/kg (Figure 7C) and is effective even if administered iv up to 5 hours post injury (109) (Figure 7D), despite a very short half-life (Figure 7E). Following myocardical infarction, reperfiiion of the ischemic cardiac muscle should be achieved as soon as possible. For example, following STEMI patients should undergo percutaneous intervention. Data show that administration of HMGB1 as soon as possible and maximally up to 5 hours post-injury will preserve the damaged myocardium and promote regeneration.

[0091] While native FR-HMGB1 promotes functional recovery post MI (Figures 7F-7I), local conversion to the disulfide form promotes thrombus formation and propagation via RAGE, TFR-2 and TFR-4 (110). Constructs reported by others such as 3S-HMGB1 that retain RAGE binding (Figure 6B) result in excessive fibrosis and impairment of function following MI (111). FR-HMGB1 also binds RAGE, albeit to a lesser extent than DS-HMGB1 and, therefore, would not be suitable for clinical use. HMGB1 signalling via TLR-2 plays a key role in ischaemia reperfusion injury following myocardial infarction (112) and thrombosis (110), The inventors a have shown the key role of TLR-2 in human atherosclerosis (113). TLR-4 signalling is also crucial in myocardial reperfusion injury (114). The redox conditions in the ischemic and inflamed microcirculation of the damaged heart following myocardial infarction will promote conversion of FR-HMGB1 to the disulfide form (DS-HMGB1), which is a central mediator of thrombosis (110). There is no approved therapy for promoting cardiac regeneration following MI. Reports purporting to show regenerative effect of hematopoietic stem cells have been discredited (115), and killed cells are just as effective by triggering an immune response (116). Even with other cell types, including pluripotential stem cells, significant challenges remain, including arrhythmogenesis, immunosuppression, scalability, batch variability, delivery, long-term viability and efficacy (115, 117). Large scale trials of cell-based therapies showed no significant improvement in function, with arrhythmias reported in patients (118- 120).

[0092] The absence of a significant stem population (121) and limited epicardial progenitor cells (122) in the adult heart, together with an understanding that the majority of new cardiomyocytes following injury are derived from existing cardiomyocytes, has shifted focus to promoting regeneration by manipulating endogenous pathways (121). This includes adenoviral transduction of multiple transcription factors(105, 108), manipulation of developmental pathways such as Hippo (106) or Meisl (123), addition of growth factors such as neuregulin (124), IGF/HGF (125) or FSTL1 (126), or manipulation of miRNAs (127). These approaches have significant shortcomings: adenoviral transduction and growth factors (IFGF1/HGF) require intracardiac injection or topical patch application (FSTL1), manipulation of developmental pathways carries oncogenic risk (128) and viral transduction of miRNA199- a in pigs resulted in fatal arrhythmias (127). An alternative strategy for stimulating cardiac regeneration by promoting clearance of immune cells requires repeated injection of VEGF-C (129). Inhibition of MAP4K4 promotes myocardial survival and limits infarct size, but there was no regenerative effect (130). To date, none of these strategies have progressed to clinical trials.

[0093] This invention provides a unique solution which targets endogenous processes to promote cardiomyocyte survival and regeneration of multiple tissues. It overcomes the many hurdles associated with cell therapies, including anti-fibrotic CAR T cells (131), such as prohibitive expense (132, 133). Since FR-HMGB1 acts via the cell surface receptor CXCR4, it is not expected to have off target effects associated with targeting intracellular processes, e.g. by adenoviral transduction of transcription factors or miRNA. HMGB1 inhibition increased infarct size following ischemia reperfusion injury (134) and whilst local upregulation (135, 136) or intramyocardial injection of FR-HMGB1 has been shown to be effective in both mice (111, 137, 138) and sheep (139), our data indicate iv administration is efficacious and more likely to reach all target cells. The engineered double Box B construct of the invention which avoids deleterious proinflammatory signaling should be safe.

[0094] A group in Milan described an HMGB1 analogue (3S-HMGB1) where 3 cysteines are replaced with serines to negate TLR-4 signaling (140). Whilst they claimed that 3S- HMGB1 is superior to FR-HMGB1 in promoting tissue regeneration (141), the present inventors have not found this to be the case (142). Importantly, 3S-HMGB1 promoted fibrosis in a murine MI model, with deterioration in cardiac function, whereas FR-HMGB1 promoted tissue regeneration and improved left ventricular ejection fraction (111). The inventors have found that 3S-HMGB1 remains bound to RAGE for longer than either DS-HMGB1 or FR- HMGB1 (Figure 5A). Therefore, whilst FR-HMGB1, 3S-HMGB1 and DS-HMGB1 bind equivalent amounts of RAGE at equilibrium, overtime levels of RAGE bound by 3S-HMGB1 are comparable to the proinflammatory disulfide HMGB1 (DS-HMGB1) and higher than to FR-HMGB1. The inventors double Box B construct eliminates undesirable proinflammatory signaling whilst retaining regenerative activity equivalent to FR-HMGB1.

[0095] Additional applications:

[0096] Fractures. Fractures occur following injury. However, one of the commonest skeletal ‘injuries’ is joint replacement or arthroplasty. The inventors propose that dBB12 can be used to promote healing following fracture or arthroplasty, thereby reducing the risk of potential complications such as loosening of components.

[0097] Brain and nervous system. dBB12L may be used to improve patient outcomes following stroke. Other potential indications include Parkinson’s disease and dementia.

[0098] Lung. dBB12L is contemplated to improve outcomes following lung injury, for example, following Covid-19 or in patients with idiopathic pulmonary fibrosis.

[0099] Liver. 30% of people in the USA are estimated to suffer from non-alcoholic liver disease. 60% of these go on to develop non-alcoholic steatohepatitis and 20% of those develop liver cirrhosis. Treatments are being developed to limit and prevent liver damage from tehse conditions. The inventors propose that dBB 12L to be used in combination with these treatments to promote liver regeneration.

[00100] Gut. dBB12L may be used to promote healing of the gut, for example, following surgery or patients with inflammatory bowel disease such as ulceractive colitis in combination with treatments to control inflammation.

[00101] Kidney. dBB12L may be used to promote regeneration of the kideny, thereby potentially avoiding the need for dialysis or kidney transplantation.

[00102] Skin. dBB12L may be used to promote wound healing eg following surgery, bums or patients with ulcers eg diabetic ulders.

[00103] Pancreas. dBB 12L may be used to improve outcomes in patients with type 1 diabetes mellitus by promoting regeneration of islet cells.

[00104] Bone marrow. dBB12L may promote regeneration of the haemopoetic system e.g. following chemotherapy, thereby preventing severe potentially life thereatening neutropenia.

[00105] The inventiors have previously shown that FR-HMGB1 is effective even if adminsitered up to 2 weeks before injury (142). Since dBB12L is equally efficacious to FR- HMGB1 (Figure 7J) it is contemplated that this polypeptide may be used prophylactically, for example, by the military or for sports injuries or before elective surgery or chemotherapy.

MATERIALS AND METHODS

E. coli strains

[00106] Mach-1 T1R cells (Invitrogen, no antibiotic resistance or induction, BL21(DE3)-R3- pRARE2 (in-house BL21 derivative, chloramphenicol resistance 36 μg/mL , T7-polymerase lac induction [87]) and BL21(DE3)-R3-pRARE2-BirA (in vivo biotinylation derivative of the above, additional spectinomycin resistance 50 μg/mL) were sourced from chemically competent stocks made in-house.

Bacterial culture media

[00107] SOC: 20 g/L tryptone, 5 g/L yeast extract, 0.5 g/L NaCl, 0.1862g/L KC1 were autoclaved and supplemented with 4.132 g/L MgCI₂ and 20 mM glucose.

[00108] LB (Luria Bertani): 10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl, pH 7.2, autoclave sterilized. 2% w/v agar powder was added to make LB agar plates. [00109] TB (Terrific Broth): 12 g/L tryptone, 24 g/L yeast extract, 4 g/L glycerol, 12.5 g/L K₂HPO₄, 2.35 g/L KH₂PO₄., autoclave-sterilized

[00110] TB supplement: 1.6% w/v glycerol, 1% glucose, 25 mM (NH4)2S04, 10 mM MgSO₄, 10X trace metals, 0.22 μM sterile filtered.

[00111] Trace metal solution: 50 mM FeCI₃ (13.5 g/L), 20 mM CaCI₂ (2.94 g/L), 10 mM MnCI₂ (1.96 g/L), 10 mM ZnSO₄ (2.88 g/L), 2 mM CoCI₂ (0.48 g/L), 2 mM CuCI₂ (0.34 g/L), and 2 mM NiCI₂ (0.48 g/L), in 0.1 M HC1, 0.22 μM sterile-filtered.

[00112] M9 minimal medium: 16 g/L Na₂HPO₄, 4 g/L K₂HPO₄, 1 g/L NaCl, pH 7.2-7.3 and 2.5 g/L FeSO₄, 0.25 mg/L ZnCI₂, 0.05 mg/L CuSO₄, 0.25 g/L EDTA, 1 mM MgSO₄ were autoclaved and supplemented with 4 g/L glucose, 1 g/L U-99% ¹⁵NH4C1 (Cambridge Isotopes), 0.3 mM CaCI_2,1.5 mg/L D-biotin and 1.5 mg/L Thiamine-HCL from sterile filtered stocks.

Plasmids

[00113] Plasmids were sourced from the SGC libraries [87]. All plasmids contain a 6xHis tag with a TEV-cleavage site; pNIC-Bio3 and pDsbC-HT-CBio also have C-terminal biotinylation epitopes (which can be removed with a stop codon). Plasmid DNA was linearized by restriction enzyme digestion: BfuAl (3h, 60°C) for pNIC-CTHF or Bsal (2h, 31°C). Cut vector DNA was purified with a PureLink PCR kit and treated with T4 DNA polymerase (NEB M0203) in the presence of 0.25 mM dGTP (pNIC-CTHF) or dCTP as per manufacturer protocols.

Cloning

[00114] HMGB1 constructs were sourced from the Mammalian Gene Collection (purified as plasmid from Machl cells grown overnight in LB medium with antibiotics). Constructs were amplified via PCR: a program of 95°C /10' , 25x (95°C/ 30”, 52°C/L, 0.5-1.5 min at 68°C), 68°C/10'was used. Reaction consisted of 5 μL Herculase II buffer, 1 μM of each primer, 6 μg/mL plasmid template, 1 μM dNTP mixture and 1 unit Herculase II polymerase (Agilent 600679; supplied with buffer and 100 μM dNTP stocks) in 25 μL final volume. PCR products were purified before further use (PureLink kit, ThermoFisher K310001).

[00115] Amplified coding sequences (alleles) were cloned into the destination vector via ligation independent cloning (LIC). The insert was treated with T4 DNA polymerase in the presence of a cognate nucleotide to that used for the vector (10 μL reaction volume), and 2 μL was mixed with 1 μL of treated vector and annealed for 30‘. 40 μL ice-cold Mach-1 cells (for storage) or 20 μL BL21(DE3)-R3-pRARE2/ BL21 (DE3)-R3 -pRARE2-BirA cells (for expression) were added and heat-shocked for 45” at 42°C before chilling in ice. Recovery was performed for 2 hours in SOC medium at 37°C prior to plating on selective media with 5% sucrose and antibiotics. After 24 h, positive colonies were picked and screened with MyTaq polymerase according to manufacturer protocols with specific sequencing primer pairs for bands of the correct molecular weight. Positive transformants were grown overnight in 1 mL of 2X LB (double concentration of LB) with antibiotics and stocked with 12% glycerol v/v at -80°C.

[00116] 3 S-HMGB1 mutant sequence was generated in a similar manner. PCR was performed separately to generate a S23-S45 and a SI 06 fragment, which were annealed via PCR; 5 μL of each purified PCR product substituted the primers and template in this reaction. The process is summarized in Figure 14.

[00117] CXCL12 constructs were cloned with an in-frame SUMO protease site N-terminal to the mature protein to allow for periplasmic secretion with an N-terminal fusion protein in the pDsbC-HT-CBio vector (DsbC-SUMO-CXCL12) to avoid addition of N-terminal residues to the protein which could affect its activity [88,89] whilst obtaining folded, oxidized CXCL12 via the DsbC fusion protein system [90]. All mutants were verified by sequencing (SourceBioscience). The sequence for HMGBl-dBB was designed in silico by codon- optimizing a Box B 89-174 sequence according to E. coli BL21-DE3 genome (assembly ASM956vl) placed after the native HMGB1 Box B sequence, and synthetized in vitro by Twist Bioscience (San Francisco, USA) cloned in pNIC-CTHF.

Recombinant protein expression

[00118] 20 mL of overnight culture of HMGB1 -expression strain transformants, grown from a fresh agar plate streak, were inoculated into 1 L of TB (or M9) medium with supplement and allowed to grow up to OD 2.0 at 37°C with 0.45 RCF orbital shaking (OD 0.6 for M9 medium). Precultures used for production of 15N labelled HMGB1 were first spun down at 1000 RCF for 5 ‘and washed in M9 medium. Once the target OD was reached, were cooled to 18°C before addition of 0.5 mM or 0.25 mM IPTG (for HMGB1 and CXCL12 proteins respectively) and grown for 16 hours before harvesting at 4000 RCF. For biotinylated proteins, 10 mM D-biotin in PBS was added before induction and again 1 hour before cell harvesting.

Recombinant HMGB1 purification

[00119] Pellets of induced HMGB1 -expressing cells were resuspended at 14 g/L in 1 MNaCl, 5% glycerol, 50 mM HEPES pH 7.5, 10 mM Imidazole (Buffer A) supplemented with 1 : 1000 protease inhibitors (Calbiochem Set III, Merck 539134), 3 μg/mL Benzonase-MBP, 1 mM MgSO₄0.5 mg/L lysozyme (Sigma L6876) and 0.5% v/v Triton-XlOO before freezing at - 80°C; from this point onwards all steps took place at 4°C. Thawed pellets were spun down at 6780 RCF for 45' and the supernatant was loaded into pre-equilibrated nickel-His GraviTrap (GE Healthcare) 1 mL columns. After drip-through, columns were washed with 10 CV of 1 M NaCl, 50 mM HEPES pH 7.5 and 1.5 CV of 0.4 M NaCl, 20 mM HEPES pH 7.5, 1 mM MgSCE, and 3 μg/mL Benzonase-MBP solution to digest remaining DNA for 30‘. Contaminants were washed with 15 CV of 0.5 M NaCl, 5% glycerol, 50 mM HEPES pH 7.5 (Buffer B) supplemented with 30 mM imidazole before elution directly into a PD- 10 column (GE Healthcare; equilibrated in Buffer B + 20 mM imidazole) with 2.5 mL of Buffer B + 500 mM imidazole. Proteins were eluted from the column with 3.5 mL of Buffer B + 20 mM imidazole before tag removal with 1:20 OD TEV-GST protease over 16 h.

[00120] Proteases and further contaminants were removed by recirculating the protein solutions over the same GraviTrap column used to purify initially (equilibrated in Buffer B+20 mM imidazole). For biotinylated proteins, streptavidin-XT resin was used instead to select biotinylated molecules only: after 30 min of incubation in the resin, sample was allowed to drip through, washed with 30 CV of buffer A and 1 CV of buffer B + 100 mM D-biotin, and eluted by incubation in 3 CV of the same buffer for 2 h. Proteins were further purified by size exclusion chromatography (SEC) (Superdex S75 10/300-0.35 ml./min or 16/600-1.2 mL/min flow rate) in either 10 mM HEPES pH 7.5 + 150 mM NaCl for biophysics work or cell-culture grade PBS for cell and animal work. Recombinant proteins were flash-frozen for storage, adding 1 mM TCEP in the case of reduced HMGB1 proteins.

Recombinant CXCL12 purification

[00121] Outer membranes of cells expressing DsbC-SUMO-CXCL12 were lysed by osmotic shock [91]. Pellets were resuspended at 40 g/L in 1 M sucrose, 0.2 M Tris-HCl pH 8.0, 1 mM EDTA, 1 mg/mL lysozyme, 2X cOmplete protease inhibitor set (COEDTAF-RO, Roche), 50 mM Imidazole and 3 μg/mL benzonase. This was stirred for 45 min at room temperature before adding 4 volumes of ice-cold 18.2 mΩ water and mixed for a further 10 min before adding 1 mM MgS04. This was centrifuged for 1 hour at 16000 RCF, 4°C, and the supernatant loaded at 10 mL/min into Ni-NTA Superflow columns (Qiagen, 30761) on an Akta Xpress FPLC system; 1 column was used for every 6 L of cells. Proteins were eluted via an imidazole gradient (10-25 mM over 10 CV, and 25-500 mM over 8 CV) in Buffer B, and 1: 10 OD of Ulp-1 protease were added before dialysis in 100 volumes of 0.2 M NaCl, 20 mM HEPES pH 8.0 (Buffer Ac) overnight. On the next day, the protein was loaded into CaptoS columns (CaptoS ImpAct, GE 17-3717-47) at 2.3 mL/min and eluted in a gradient of 0.2-1.5 M NaCl in 20 mM HEPES pH 8.0 to separate cut CXCL12 from DsbC and Ulp-1. Proteins were further purified via SEC in the same way as HMGB1 and flash-frozen for storage.

Removal of endotoxins

[00122] Endotoxin was removed in all cases before size-exclusion chromatography via phase separation with Triton Tx-114 [92]. A 2% v/v of TX-114 was added to recombinant protein solutions, homogenized for 20' with orbital shaking at 2000 RCF at 4°C, and separated for 5' at 37°C before pelleting the detergent phase at 8000 RCF, 10', 25 °C. The supernatant was mixed with 5% w/v of SM-2 Biobeads (BioRad, 152-8920), cleaned with 2% TX-114 for 2 hours and regenerated with 30 CV of methanol, 30 CV of endotoxin-free 18.2 ihW water and 30 CV of endotoxin-free PBS. This was incubated for 4 hours at room temperature to adsorb remaining Triton and PEG [93] before injection onto a sanitized SEC system (with 0.5 M NaOH contact over 12 h, followed by 0.2 M acetic acid/20% ethanol contact over 6 hours and equilibration in cell-culture grade PBS) to fully remove leftover polymer contaminants whilst performing size exclusion. The absence of Triton and PEG was verified by lack of their respective charge state species in ESI/QTOF-MS mass spectrometry [94]. LPS content of the recombinant proteins was assessed via the LAL method (GenScript ToxinSensor L000350). Samples were approved for cell and animal use when they contained < 4 EU LPS/mg protein.

Enzyme production

[00123] TEV-GST protease (GST-fusion protein), Benzonase-MBP, and Ulp-1 protease were produced from transformants in storage at the SGC collection [87]; all had 200 μg/mL ampicillin resistance. TEV and Ulp-1 were purified as per the protocols described for HMGB1 with only one IMAC step, whereas Benzonase-MBP was purified from outer membrane lysates obtained as with CXCL12 and isolated with use of amylose resin (NEB, E0821) as per manufacturer protocols. In both cases, the resulting proteins were concentrated to 10 mg/mL in 50 mM HEPES pH 7.5, 0.3 M NaCl, 10% glycerol. GST-TEV protease and Ulp-1 were flash-frozen with liquid nitrogen and supplemented with 0.5 mM TCEP during purification; Benzonase-MBP was supplemented with 50% glycerol and 2 mM MgCI₂ and stored at -20°C.

Peptide arrays

[00124] Membranes with FMOC-coupled 15-mer peptides of human HMGB1 (Uniprot P09429) or CXCL12 (Uniprot P48061, excluding secretion signal) were printed by Dr. Sarah Picaud at the SGC upon request following published protocols [40]. The membranes were rehydrated at 20-25°C with 95% and 70% ethanol, equilibrated with PBST (PBS IX + 0.05% Tween-20, 3x), and blocked with 10% BSA/PBST for 8 h. 1 μM of the partner His-tagged protein construct was added (in PBS) and allowed to bind for 24 hours at 4°C. Excess BSA and protein was removed with 3 washes in PBST; all washes lasted 1 min unless otherwise stated. To detect bound proteins, the membranes were treated with 1:3000 dilution of Qiagen anti- PentaHis HRP conjugate (Qiagen 34460) and excess antibody removed with 3 washes in PBST for 20'. Bound antibody was then quantified via chemiluminescence (Pierce ECL substrate - 32109): the membrane was covered in substrate solution and placed between two clear plastic sheets before incremental imaging at 2' intervals in a LAS-4000 camera (chemiluminescence settings). The intensity of the peptides in each membrane was measured in Image J and normalized to the 10-His control. Alanine mutagenesis scans were performed in the same manner, with the printed peptides consisting of those identified during the initial peptide array. Residues whose mutation to alanine resulted in higher intensity changes than those observed for alanine positions in the sequence were considered as significant contributors to CXCL12 binding.

Biolaver interferometry (BLI)

[00125] Pre-hydrated streptavidin Octet Biosensors (ForteBio 18-5019) were coated with 4 μM solutions of biotinylated HMGB1 proteins in 10 mM HEPES, pH 7.5, 150 mM NaCl (Base buffer-BB) plus, 0.5 mM TCEP (60 sec baseline, 60 sec binding). Nonspecific binding was minimized by incubation for 3' in BB+ 1% BSA + 0.05% Tween-20 (Kinetics Buffer, KB) prior to kinetics assays. Interaction with CXCL12 was measured by performing incremental immersion of the sensors in solutions with increasing CXCL12 concentration (0-150 μM, in 1:2 dilutions) in KB (60 sec baseline, 500 sec association, 420 sec dissociation, 180 sec reduction in BB + 0.5 mM TCEP). An OctetRed 384 instrument was used for these experiments. Kinetics data were extracted with DataAnalysis 9.0 (ForteBio). Response at equilibrium R_Eq was plotted against concentration in a Michaelis-Menten saturation plot to calculate kD/B_max (saturation plot of concentration against rEq). Kinetic rates (association rate; kon and dissociation rate; k_0ff) were derived from the direct measurements of each parameter from all the association and dissociation steps in the interferogram, and fitted to a horizontal line (mean) across all measurements. Data from each replicate run were pooled in the same manner to calculate the overall mean. For measuring RAGE binding kinetics to HMGB1, 15 μg/mL RAGE-Fc in PBS+ 0.1% BSA + 0.02% Tween-20 was immobilized on the surface of anti-IgG biosensors (AHC, 18-5060) over 30 sec and dipped in serial concentrations of each HMGB1 construct (60 sec baseline, 200 sec association/dissociation) and fitted in the same manner to derive kinetic parameters.

Nuclear magnetic resonance (NMR)

[00126] ¹⁵N-labelled recombinant HMGB1 constructs in 10 mM HEPES 150 mM NaCl pH 7.5 (identical buffering and ionic strength to BLI experiments) were supplemented with 5% v/v D2O, and pipetted with a glass Pasteur pipette into a 5 mm Shigeimi tube, sealed with paraffin. Final volumes were > 330 μL. CXCL12 was added in the same buffer and final volumes were adjusted to avoid modification of the referencing. Signal locking, tube shimming, and nuclei tuning were manually performed via Bruker TopSpin software. Water signal was suppressed by acquiring a 'H spectra with power level 1 (PI) = estimated pulse calibration (pulsecal). When a single peak was observed, a PI value of 4 times the initial was used as a baseline and adjusted until a symmetric peak could be observed in the ¹H spectrum. NMR experiments were performed after these calibration steps ('H-NMR. ¹⁵N-HSQC, ¹⁵N- NOESY-HSQC, ¹⁵N-TOCSY-HSQC Peaks in ¹⁵N-HSQC spectra were assigned based on published NMR tables for HMGB1 1-184 (BMRB 15418) and our own NOESY/TOCSY data for each construct analyzed. To measure CXCL12 binding, the chemical shift position and volume of identified peaks was tracked across different molar equivalences of CXCL12 (these are listed in the relevant image) via CCPNMR 3.0 Analyze's Chemical Shift Tracking module. For peak intensity changes, the median intensity change in each set was considered as a baseline. Full chemical shift tables and experiment parameters are detained at the end of this section.

Mass spectrometry

[00127] For protein identification via MS/MS and tryptic digest, bands from SDS-PAGE gels were excised and submitted to the SGC open access MS platform and analysed by Dr. Rod Chalk, Dr. Tiago Moreira and Oktawia Borkowska, as published [94,95]. Data analysis (peptide mapping) to annotate protein identity was performed with the MASCOT search engine, against the Uniprot (reference protein sequences) and SGC (construct sequence) databases. Native ESI/MS experiments were performed by manual injection in volatile buffer (50 or 200 mM ammonium acetate, pH 6.5) into an ESI/QTOF instrument (Agilent Q-TOF 6545) at 360 μL/h. Signal was acquired for at least 10 counts (30 sec) once a steady ionic stream was observed in the total ion chromatogram. For denaturing experiments, samples were diluted to 1 mg/mF in 0.2% formic acid and injected via HPFC (Agilent 1100 HPFC) and eluted in a mobile phase of formic acid/methanol, as described [94]. Each continuous distribution of charge states was considered a distinct conformation; charge states (Z) were assigned according to the formula where mW= (mW/Z -proton mass )* Z. Surface areas were derived from the formulas proposed in the literature [96,97] which resulted in the formula ln(SASA)= ln(M/Z)*0.6897-4.063 for native MS and ln(SASA)= ln(M/Z)*0.9024-5.9013 for denatured samples. At least three independent injections were performed for all MS samples. All solutions in these experiments were made with HPLC water (electrochemical grade) and solvents.

SEC surface area quantitation

[00128] To correlate SEC chromatograms with surface area, standard sets with known structures (BioRad 1511901) were run through a Superdex 75 pg, 10/300 column used in these experiments. SASA for the proteins contained in these and the GE-supplied calibration curve standards were derived from published PDB structures (BSA, 3V03;Ovalbumin, 1JTI;Myoglobin, 2VlI;RNAseA, lA5P;Aprotinin, lNAG;Vitamin B12, 3BUL) and correlated with retention volume by nonlinear least squares fitting (SASA= 331.2*RV²- 1 . 19c4*RV+l 08c5). All experiments compared between HMGB1 samples were performed in the same buffer as native MS (200 mM ammonium acetate, pH 6.5); injections were performed at 1 mg/mL to avoid saturation of signal and all samples were eluted at 0.4 mL/min.

RAGE binding ELISA assay

[00129] 384-well protein-binding ELISA plates (Santa Cruz Biotechnology, sc-206072) were coated with 50 mE of 40 nM solutions in PBS (+ 0.5 mM TCEP for FR-HMGB1 constructs) of HMGB1 constructs for 24 h, at 4°C, including FL/DS HMGB1 full-length controls and blank, with 4 replicates of each. Nonspecific binding was blocked by incubation with 10% BSA in PBS for 2 h, at 20-25°C. A concentration range of RAGE-Fc chimera protein (BioTechne, 1145-RG; 0-640 nM in 1:4 dilutions) was added in 10% BSA/PBS and allowed to bind for 2 h, at 4°C. Bound FC chimera was detected by incubation with Anti-Human IgG HRP (Agilent Dako P021402-2) diluted 1: 10000 in 1% BSA/PBS for 2 h, at 20-25°C. Between each of these 3 steps, the plate was washed with 100 μL PBST, 3 times.

[00130] To detect bound antibody, 25 μL of TMB substrate (ThermoFisherN301) was added to each well; the reaction was allowed to develop in the dark until the FL-DS-HMGB1 control developed a clear concentration-dependent color gradient before stopping the reaction with 25 μL of 0.5 M H2SO4. OD 450 was measured as a readout (FluoStar OMEGA, BMG Labtech) and plotted as a saturation fit against 2x RAGE-Fc concentration (as the chimera is a RAGE dimer)

TLR-4 and TLR-2-mediated NF-KB signaling reporter assay

[00131] HEK-Dual cells (Invivogen) expressing human TLR-2 and CD 14 or murine TLR-4, MD-2 and CD14 were maintained in DMEM (Gibco), supplemented with 10 % FBS (Gibco), 1 % L-Glutamine (Gibco), and 1 % penicillin/streptomycin (Gibco), in standard tissue culture conditions (37oC; 5% C02). To determine if FR-HMGB1, DS-HMGB1 and dBB12L induces activation of TLR-4 and TLR-2 signaling, 104 TLR-4 and TLR-2 HEK-Dual cells were plated in triplicate into wells of a 96 well plate and stimulated with 10 μg/mL 1 HMGB1 and (X concentration) FSL-1 for TLR-2 and 10 ng/mL LPS for TLR-4. 24 hours after stimulation, NF- kb activity was determined my measuring the induced levels of secreted embryonic alkaline phosphatase (SEAP).

Monocyte total NF-KB secretion assay

[00132] Human monocytes (StemCell Technologies) were maintained in DMEM (Gibco), supplemented with 10% FBS (Gibco) in standard tissue culture conditions (37°C; 5% C02). To determine if FR-HMGB1, DS-HMGB1 and dBB12L induces proinflammatory cytokine production, 10⁵ human monocytes were plated in triplicate into wells of a 96-well plate and stimulated with 10 μg/mL HMGB1 and 50 ng/mL LPS or 10 ng/mL LTA. 24 hours after stimulation, TNF levels were determined by Enzyme-linked immunosorbent assays (ELISA) (Abeam).

Transcriptomic analysis

[00133] Mice were treated systemically with an i.v. injection of 30 pg FR-HMGB1 in 50 μL of PBS vehicle, or PBS only control. Injury cell are from BaCI₂ injured mice as described below. Alert cells are from the uninjured contralateral side of BaCI₂ injured mice. Murine muscle stem cells (mMuSCs) were defined and freshly isolated according to previously reported protocols. Muscle cell suspensions were created by mincing thigh muscles and enzymatically digesting with collagenase 800 U/ml (Worthington-Biochem) and dispase 1 U/mL (Gibco). Thereafter, all suspensions were strained through 70 pm and 40 pm filters (Greiner Bio-One) and stained with respective antibodies. mMuSC, CD31-CD45-Sca- 1- VCAM1⁺, were isolated by fluorescence activated cell sorting (FACS) using BD FACSAria III machine. RNA, extracted from freshly FACS isolated mMuSCs, was sent to RNA-seq analysis using Lexogen 3 ’kit library prep and sequenced using HiSeq400 (Illumina). FASTQ files were assessed using FASTQC followed by the generation of TPM values with kallisto v0.42.4. TPM values were summed to obtain gene-level expression values using tximport and differential expression analysis was undertaken with DeSEQ2. GO enrichment of differentially expressed genes was performed using the R package ‘clusterProfiler’[98] with a Benjamini-Hochberg multiple testing adjustment and a false-discovery rate cut-off of 0.1. Visualization was performed using the R packages ‘ggplot2' and ‘igraph' .

In vivo mouse muscle injury model

[00134] C57BL/6 inbred mouse strain, females of 11-12 weeks of age were purchased from Charles River UK and housed in the local Biological Safety Unit (BSU) at the Kennedy Institute. Acclimatization period was 1-2 weeks. All protocols performed on live animals have been approved by the UK Home Office (PPU 30/3330 and PPU P12F5C2AF) as well as the local animal facility named persons and are registered under the appropriate project and personal licenses under ASPA regulations. All consumables were surgically certified; recombinant proteins were endotoxin-free. Surgeries were done in a clean environment separate from culling facilities. All animals were monitored for 6 hours post-operative ly, and daily for the following 3 days; monitoring was then transferred to the NVS/NACWO.

[00135] Surgeries were performed as described previously ([7,12]). Animals were anesthetized by aerosolized 2% isoflurane, given analgesia, transferred to a warming pad and the right lower hindlimb was disinfected with povidone iodine and the tail with 70% ethanol if intravenous injection was performed. 50 μL of 1 .2% BaCI₂ (Sigma) was injected into and along the length of the tibialis anterior (TA) muscle to induce cell death. Mice were euthanized and lower limbs removed at the times indicated, fixed in 4% paraformaldehyde (Santa Cruz Biotechnology) for 24 h. The TA muscles were dissected and further fixed for 24 hours before being embedded in paraffin and sectioned. Sections (5 pm) were stained with hematoxylin and eosin to identify fibers with central nuclei and imaged with an Olympus BX51 using a lOx ocular/ 40x objective lens. The cross-sectional area (CSA) of the fibers from at least 4 images per mice was manually measured using the FIJI distribution of Image J2 software (NIH). Data were grouped per mice. Mice were injected with HMGB1 constructs (46 nM/kg, resuspended in PBS) or PBS vehicle control intramuscularly or intravenously at the time of injury or after injury for the optimal time administration of HMGB1 constructs after injury.

In vivo mouse cardiac injury model

[00136] C57BU/6 female mice were subject to surgery between 10-14 weeks old, with body weight between 25-30 g. All mice had either an intravenous injection of FR-HMGB1 (46 nM/kg, resuspended in PBS) or vehicle control just before surgery. Buprenorphine (buprenorphine hydrochloride; Vetergesic) was delivered as a 0.015 mg ml solution via intraperitoneal injection at 20 min before the procedure to provide analgesia. They were anaesthetized with 2.5% isoflurane and externally ventilated via an endotracheal tube. Cardiac injury was induced by permanent ligation of the left anterior descending coronary artery (LAD) via a thoracotomy. Experimenters were blind to treatment groups for subsequent cardiac cine- MRI and analysis. Mice were housed and maintained in a controlled environment. All surgical and pharmacological procedures were performed in accordance with the Animals (Scientific Procedures) Act 1986, UK.

Cardiac cine-MRI and analysis

[00137] Cardiac cine-MRI was performed post-LAD ligation at 7T using a Varian DDR system. Briefly, mice were anaesthetised with 2% isoflurane in 02, and positioned supine in a custom animal handling system with homeothermic control. Prospectively gated proton cardiac images were acquired with a partial Fourier accelerated spoiled gradient echo CINE sequence (TR 5.9 ms, TE 2.2 ms, 30 kHz bandwidth, 30° FA, approximately 20-30 frames gated to the R wave with a 4 ms postlabel delay; 20% partial acquisition; 4 averages) with a 72 mm volume transmit/4 channel surface receive coil (Rapid Biomedical GmbH) in order to acquire two and four chamber long-axis views and a short axis stack for functional quantification (128x128 matrix; 25.6 mm^2 FOV; 0.2 mm resolution in-plane). Non-acquired partial Fourier data was reconstructed via the method of projection onto convex sets prior to a simple, cartesian, DFT. Blinded image analysis was performed using ImageJ (NIH). Left ventricular mass, volumes and ejection fraction were calculated as previously described¹. The relative infarct size was calculated from the average of the endocardial and epicardial circumferential lengths of the thinned, akinetic region of all slices, measured at diastole, and expressed as a percentage of the total myocardial surface [99] .

Statistical analysis

[00138] All calculations were performed with GraphPad Prism (v. 8.41). For kinetics experiments (BFI/RAGE EFISA), all fits were performed via nonlinear least squares. For the RAGE EFISA, as each concentration of RAGE was independent from the rest of the wells, all data were considered as one kinetics fit; in BFI however each sensor was considered an independent fit for the purposes of calculation. Comparisons between parameters were performed via the AUC method. Mouse muscle injury model data were analyzed as a nested ANOVA, where each sub-column comprises all the muscle CSA values for a given animal, and each group contains all the animals in order to separate biological variation from treatment effect. If the equal variance assumption could not be met in either case data were analyzed via a Kruskal-Wallis test; for nested ANOVA equal numbers of data from each animal would be randomly selected to avoid skewing. Any other data were analyzed via one-way ANOVA if the heteroskedasticity plot and Q/Q plot supported the equal variance assumption [100]; these were also verified by Spearman's Test. For multivariate experiments (e.g. cardiac experiments) a two-factor ANOVA was used under the same assumptions (no dataset violated heteroskedasticity in this case). Post-hoc comparisons were weighted by Holms-Sidak correction (ANOVA family tests) or Dunn's method (Kruskal-Wallis). The test selected in each case is noted under each figure legend. Significance legends: n.s; not significant, *; p < 0.033,

**; p< 0.002, ***; p< 0.0002, ****; p< 0.0001.

NMR chemical shift tables

HMGB1-C028 (94-162, biotinylated) titration with CXCL12A-c021 at 0, 0.42, 0.82 and 1.42 molar equivalents, Figures 3A-3B.

[00139] Due to protein amount limitations, titration was performed by sequential addition of CXCL12 to HMGB1 samples, resulting in sample dilution. As the calculations in the chemical shift tracking module in CCPNMR are independent of peak, this does not alter the results; the median change has been indicated in the volume comparisons. 3D experiments (Day 1)

These spectra are not shown due to their 3D nature; data available if requested. Sample is that of Baseline 1.

HMGB1-C038 (89-174, biotinylated) titration with CXCL12A-c021 at 0, 0.42, 0.82 and 1.42 molar equivalents, Figures 3A-3B

[00140] Due to protein amount limitations, titration was performed by sequential addition of CXCL12 to HMGB1 samples, resulting in sample dilution. As the calculations in the chemical shift tracking module in CCPNMR are independent of peak, this does not alter the results; the median change has been indicated in the volume comparisons.

REFERENCES

1. Weissman IL: Stem cells: Units of development, units of regeneration, and units in evolution. Cell 2000, 100:157-168.

2. Gratwohl A, Pasquini MC, Aljurf M, Atsuta Y, Baldomero H, Foeken L, Gratwohl M, Bouzas LF, Confer D, Frauendorfer K, et al.: One million haemopoietic stem-cell transplants: a retrospective observational study. Lancet Haematol 2015, 2:e91-el00.

3. Sharma R, Khristov V, Rising A, Jha BS, Dejene R, Hotaling N, Li Y, Stoddard J, Stankewicz C, Wan Q, et al.: Clinical -grade stem cell-derived retinal pigment epithelium patch rescues retinal degeneration in rodents and pigs. Sci TranslMed 2019, 11:1-14.

4. Ojeh N, Pastar I, Tomic-Canic M, Stojadinovic O: Stem cells in skin regeneration, wound healing, and their clinical applications. IntJMol Sci 2015, 16:25476-25501.

5. Domb AJ, Nudelman R: In vivo and in vitro elimination of aliphatic polyanhydrides. Biomaterials 1995, 16:319-23.

6. Trainor N, Pietak A, Smith T: Rethinking clinical delivery of adult stem cell therapies. Nat Biotechnol 2014, 32:729-735.

7. Lee G*, Espirito Santo, A. I.*, Zwingenberger Cai L, Vogl T, Feldmann M, Horwood N, J.K-K C, Nanchahal J: HMGB1 accelerates the regeneration of multiple tissues by transitioning stem cells to G(Alert). Proc Natl Acad Sci USA (* authors Contrib Equal to this Proj 2018, doi: 10.1073/pnas.1802893115.

8. Oppenheim JJ, Yang D: Alarmins: Chemotactic activators of immune responses. Curr Opin Immunol 2005, 17:359-365.

9. Yang D, Rosa G De, Tewary P, Oppenheim JJ: Alarmins Link Neutrophils and Dendritic Cells. Trends Inmunol 2009, 30:531-537.

10. Belgrano FS, De Abreu Da Silva IC, Bastos De Oliveira FM, Fantappie MR, Mohana- Borges R: Role of the acidic tail of high mobility group protein B1 (HMGB1) in protein stability and DNA bending. PLoS One 2013, 8: 1-12.

11. Joshi SR, Sarpong YC, Peterson RC, Scovell WM: Nucleosome dynamics: HMGB1 relaxes canonical nucleosome structure to facilitate estrogen receptor binding. Nucleic Acids Res 2012, 40:10161-10171.

12. Rodgers JT, King KY, Brett JO, Cromie MJ, Charville GW, Maguire KK, Brunson C, Mastey N, Liu L, Tsai C-R, et al.: mTORCl controls the adaptive transition of quiescent stem cells from GO to G(Alert). Nature 2014, 509:393-6.

13. Youn JH, Kwak MS, Wu J, Kim ES, Ji Y, Min HJ, Yoo JH, Choi JE, Cho HS, Shin JS: Identification of lipopolysaccharide-binding peptide regions within HMGB1 and their effects on subclinical endotoxemia in a mouse model. Eur J Immunol 2011, 41:2753- 2762.

14. Huttunen HJ, Fages C, Kuja-Panula J, Ridley AJ, Rauvala H: Receptor for advanced glycation end products-binding COOH-terminal motif of amphoterin inhibits invasive migration and metastasis. Cancer Res 2002, 62:4805-4811.

15. Anggayasti WL, Ogino K, Yamamoto E, Helmerhorst E, Yasuoka K, Mancera RL:

The acidic tail of HMGB1 regulates its secondary structure and conformational flexibility: A circular dichroism and molecular dynamics simulation study. Comput Struct Biotechnol J 2020, 18:1160-1172. Watson M, Stott K, Thomas JO: Mapping Intramolecular Interactions between Domains in HMGB1 using a Tail-truncation Approach. JMol Biol 2007, 374: 1286— 1297. Venereau E, Casalgrandi M, Schiraldi M, Antoine DJ, Cattaneo A, De Marchis F, Liu J, Antonelli A, Preti A, Raeli L, et al.: Mutually exclusive redox forms of HMGB1 promote cell recruitment or proinflammatory cytokine release. J Exp Med 2012, 209:1519-28. Cecchinato V, D’Agostino G, Raeli L, Nerviani A, Schiraldi M, Danelon G, Manzo A, Thelen M, Ciurea A, Bianchi ME, et al.: Redox-Mediated Mechanisms Fuel Monocyte Responses to CXCL12/HMGB1 in Active Rheumatoid Arthritis. Front Immunol 2018, 9:1-12. Ferrara M, Chialli G, Ferreira LM, Ruggieri E, Careccia G, Preti A, Piccirillo R, Bianchi ME, Sitia G, Venereau E: Oxidation of HMGB1 Is a Dynamically Regulated Process in Physiological and Pathological Conditions. Front Immunol 2020, 11: 1-13. Bonaldi T, Talamo F, Scaffidi P, Ferrera D, Porto A, Bachi A, Rubartelli A, Agresti A, Bianchi ME: Monocytic cells hyperacetylate chromatin protein HMGB1 to redirect it towards secretion. EMBOJ 2003, 22:5551-5560. Kim YH, Kwak MS, Park JB, Lee SA, Choi JE, Cho HS, Shin JS: N-linked glycosylation plays a crucial role in the secretion of HMGB1. J Cell Sci 2016, 129:29- 38. Stark K, Philippi V, Stockhausen S, Busse J, Antonelli A, Miller M, Schubert I, Hoseinpour P, Chandraratne S, Von Bruhl ML, et al.: Disulfide HMGB1 derived from platelets coordinates venous thrombosis in mice. Blood 2016, 128:2435-2449. Ahrens I, Agrotis A, Topcic D, Bassler N, Chen YC, Bobik A, Bode C, Peter K: HMGB1 binds to activated platelets via platelet-expressed receptor for advanced glycation end products (RAGE) and is highly expressed in platelet rich coronary artery thrombi. Thromb Haemost 2015, 14:994-1003. Yu M, Wang H, Ding A, Golenbock DT, Latz E, Czura CJ, Fenton MJ, Tracey KJ, Yang H: HMGB1 signals through toll-like receptor (TLR) 4 and TLR2. Shock 2006, 26:174-179. van Beijnum JR, Buurman WA, Griffioen AW: Convergence and amplification of toll like receptor (TLR) and receptor for advanced glycation end products (RAGE) signaling pathways via high mobility group B1 (HMGB1). Angiogenesis 2008, 11:91— 99. Lotze MT, Tracey KJ: High-mobility group box 1 protein (HMGB1): nuclear weapon in the immune arsenal. Nat Rev Immunol 2005, 5:331-42. Sakaguchi M, Murata H, Yamamoto K, Ono T, Sakaguchi Y, Motoyama A, Hibino T, Kataoka K, Huh N: TIRAP, an adaptor protein for TLR2/4, transduces a signal from RAGE phosphorylated upon ligand binding. PLoS One 2011, 6:e23132. Venereau E, Schiraldi M, Uguccioni M, Bianchi ME: HMGB1 and leukocyte migration during trauma and sterile inflammation. Mol Immunol 2013, 55:76-82. He M, Bianchi ME, Coleman TR, Tracey KJ, Al-abed Y, Unit CD, Biology C: Exploring the biological functional mechanism of the HMGB1 / TLR4 / MD-2 complex by surface plasmon resonance. [date unknown], Yang H, Hreggvidsdottir HS, Palmblad K, Wang H, Ochani M, Li J, Lu B, Chavan S, Rosas-Ballina M, Al-Abed Y, et al.: A critical cysteine is required for HMGB1 binding to Toll-like receptor 4 and activation of macrophage cytokine release. Proc Natl Acad Sci USA 2010, 107:11942-7. Tirone M, Tran NL, Ceriotti C, Gorzanelli A, Canepari M, Bottinelli R, Raucci A, Maggio S Di, Santiago C, Mellado M, et al.: High mobility group box 1 orchestrates tissue regeneration via CXCR4. J Exp Med 2018, doi: 10.1084/jem.20160217. Di Maggio S, Milano G, De Marchis F, D’Ambrosio A, Bertolotti M, Palacios BS,

Badi I, Sommariva E, Pompilio G, Capogrossi MC, et al.: Non-oxidizable HMGB1 induces cardiac fibroblasts migration via CXCR4 in a CXCL 12-independent manner and worsens tissue remodeling after myocardial infarction. Biochim Biophys Acta - Mol Basis Dis 2017, 1863:2693-2704. Teng W, He X, Shan L, Fan C, Peng S, Wang X, Shan Z, Wang W, Liu X, Liu S, et al.: Glycyrrhizin, a Direct HMGB1 Antagonist, Ameliorates Inflammatory Infiltration in a Model of Autoimmune Thyroiditis via Inhibition of TLR2-HMGB1 Signaling. Thyroid 2017, 27:722-731. Kwak MS, Lim M, Lee YJ, Lee HS, Kim YH, Ho Youn J, Choi JE, Shin J-S: HMGB1 Binds to Lipoteichoic Acid and Enhances TNF-α and IL-6 Production through HMGB1 -Mediated Transfer of Lipoteichoic Acid to CD 14 and TLR2. J Innate Immun 2015, 7:405-416. Aucott H, Sowinska A, Harris HE, Lundback P: Ligation of free HMGB1 to TLR2 in the absence of ligand is negatively regulated by the C-terminal tail domain. Mol Med 2018, 24:19. Das N, Dewan V, Grace PM, Gunn RJ, Tamura R, Tzarum N, Watkins LR, Wilson IA, Yin H: HMGB1 Activates Proinflammatory Signaling via TLR5 Leading to Allodynia. Cell Rep 2016, 17:1128-1140. Arumugam T, Ramachandran V, Gomez SB, Schmidt AM, Logsdon CD: S100P- derived RAGE antagonistic peptide reduces tumor growth and metastasis. Clin Cancer Res 2012, 18:4356-4364. LeBlanc PM, Doggett TA, Choi J, Hancock MA, Durocher Y, Frank F, Nagar B, Ferguson TA, Saleh M: An immunogenic peptide in the A-box of HMGB1 protein reverses apoptosis-induced tolerance through RAGE receptor. JBiol Chem 2014, 289:7777-7786. Rauvala H, Rouhiainen A: Physiological and pathophysiological outcomes of the interactions of HMGB1 with cell surface receptors. Biochim Biophys Acta - Gene Regul Mech 2010, 1799:164-170. Nady N, Min J, Kareta MS, Chedin F, Arrowsmith CH: A SPOT on the chromatin landscape? Histone peptide arrays as a tool for epigenetic research. Trends Biochem Sci 2008, 33:305-313. Mollica L, De Marchis F, Spitaleri A, Dallacosta C, Pennacchini D, Zamai M, Agresti A, Trisciuoglio L, Musco G, Bianchi ME: Glycyrrhizin binds to high-mobility group box 1 protein and inhibits its cytokine activities. Chem Biol 2007, 14:431-41. Schiraldi M, Raucci A, Munoz LM, Livoti E, Celona B, Venereau E, Apuzzo T, De Marchis F, Pedotti M, Bachi A, et al.: {HMGB1} promotes recruitment of inflammatory cells to damaged tissues by forming a complex with {CXCL12} and signaling via { CXCR4.}. J Exp Med 2012, 209:551-563. De Leo F, Quilici G, Tirone M, Mannella V, De Marchis F, Preti A, Gori A, Casalgrandi M, Mezzapelle R, Bianchi M, et al.: Diflunisal targets the HMGB1/CXCL12 heterocomplex and blocks immune cell recruitment. bioRxiv 2019, doi: 10.1101/563890. Fassi EMA, Sgrignani J, D’Agostino G, Cecchinato V, Garofalo M, Grazioso G, Uguccioni M, Cavalli A: Oxidation State Dependent Conformational Changes of HMGB1 Regulate the Formation of the CXCL12/HMGB1 Heterocomplex. Comput Struct Biotechnol J 2019, 17:886-894. Bianchi ME, Falciola L, Ferrari S, Lilley DM: The DNA binding site of HMG1 protein is composed of two similar segments (HMG boxes), both of which have counterparts in other eukaryotic regulatory proteins. Embo J 1992, 11:1055-1063. Li J, Wang H, Mason JM, Levine J, Yu M, Ulloa L, Czura CJ, Tracey KJ, Yang H: Recombinant HMGB1 with cytokine -stimulating activity. J Immunol Methods 2004, 289:211-223. De Leo F, Quilici G, Tirone M, Mannella V, De Marchis F, Preti A, Gori A, Casalgrandi M, Mezzapelle R, Bianchi M, et al.: Diflunisal targets the HMGB1/CXCL12 heterocomplex and blocks immune cell recruitment - SM. bioRxiv

2019, doi: 10.1101/563890. He M, Bianchi ME, Coleman TR, Tracey KJ, Al-abed Y : Exploring the biological functional mechanism of the HMGB1 / TLR4 / MD-2 complex by surface plasmon resonance. Mol Med 2018, 24: 1-9. Deng M, Tang Y, Li W, Wang X, Zhang R, Zhang X, Zhao X, Liu J, Tang C, Liu Z, et al.: The Endotoxin Delivery Protein HMGB1 Mediates Caspase-11 -Dependent Lethality in Sepsis. Immunity 2018, 49:740-753x7. Ottestad W, Rognes IN, Pischke SE, Mollnes TE, Andersson U, Eken T: Biphasic Release of the Alarmin High Mobility Group Box 1 Protein Early After Trauma Predicts Poor Clinical Outcome. Crit Care Med 2019, 47:e614-e622. Yang H, Wang H, Wang Y, Addorisio M, Li J, Postiglione MJ, Chavan SS, Al-Abed Y, Antoine DJ, Andersson U, et al.: The haptoglobin beta subunit sequesters HMGB1 toxicity in sterile and infectious inflammation. J Intern Med 2017, 282:76-93. Yang H, Wang H, Levine YA, Gunasekaran MK, Wang Y, Addorisio M, Zhu S, Li W, Li J, de Kleijn DP V, et al.: Identification of CD 163 as an antiinflammatory receptor for HMGB1 -haptoglobin complexes. JCI insight 2016, 1:1-14. Sadek H, Olson EN: Toward the Goal of Human Heart Regeneration. Cell Stem Cell

2020, 26:7-16. Brooks M: Stem cell research: time for a dose of realism. BMJ 2017, 356. Vagnozzi RJ, Maillet M, Sargent MA, Khalil H, Johansen AKZ, Schwanekamp JA, York AJ, Huang V, Nahrendorf M, Sadayappan S, et al.: An acute immune response underlies the benefit of cardiac stem cell therapy. Nature 2020, 577:405-409. Forbes SJ, Rosenthal N: Preparing the ground for tissue regeneration: From mechanism to therapy. Nat Med 2014, 20:857-869. Lane SW, Williams DA, Watt FM: Modulating the stem cell niche for tissue regeneration. Nat Biotechnol 2014, 32:795-803. Zhang Y, Desai A, Yang SY, Bae KB, Antczak MI, Fink SP, Tiwari S, Willis JE, Williams NS, Dawson DM, et al.: Inhibition of the prostaglandin-degrading enzyme 15 -PGDH potentiates tissue regeneration. Science (80- ) 2015, 348. Antczak MI, Zhang Y, Wang C, Doran J, Naidoo J, Voruganti S, Williams NS, Markowitz SD, Ready JM: Inhibitors of 15-Prostaglandin Dehydrogenase To Potentiate Tissue Repair. JMed Chem 2017, 60:3979-4001. Ren X, Zhao M, Lash B, Martino MM, Julier Z: Growth Factor Engineering Strategies for Regenerative Medicine Applications. Front Bioeng Biotechnol 2020, 7:1-9. Yamakawa S, Hayashida K: Advances in surgical applications of growth factors for wound healing. Burn Trauma 2019, 7:1-13. Mitchell AC, Briquez PS, Hubbell JA, Cochran JR: Engineering growth factors for regenerative medicine applications. Acta Biomater 2016, 30: 1-12. Hori O, Brett J, Slattery T, Cao R, Zhang J, Chen JX, Stem D, Schmidt AM: The Receptor for Advanced Glycation End Products (RAGE) Is a Cellular Binding Site for Amphoterin. J Biol Chem 1995, 270:25752-25761. Stark K, Philippi V, Stockhausen S, Busse J, Antonelli A, Miller M, Schubert I, Hoseinpour P, Chandraratne S, Von Bruhl ML, et al.: Disulfide HMGB1 derived from platelets coordinates venous thrombosis in mice - SM. Blood 2016, 128:2435-2449. Vogel S, Bodenstein R, Chen Q, Feil S, Fed R, Rheinlaender J, Schaffer TE, Bohn E, Frick JS, Borst O, et al.: Platelet-derived HMGB1 is a critical mediator of thrombosis. J Clin Invest 2015, 125:4638-4654. Maugeri N, Campana L, Gavina M, Covino C, De Metrio M, Panciroli C, Maiuri L, Maseri A, D’Angelo A, Bianchi ME, et al.: Activated platelets present high mobility group box 1 to neutrophils, inducing autophagy and promoting the extrusion of neutrophil extracellular traps. JThromb Haemost 2014, 12:2074-2088. Ibrahim ZA, Armour CL, Phipps S, Sukkar MB: RAGE and TLRs: Relatives, friends or neighbours? Mol Immunol 2013, 56:739-744. Festoff BW, Sajja RK, van Dreden P, Cucullo L: HMGB1 and thrombin mediate the blood-brain barrier dysfunction acting as biomarkers of neuroinflammation and progression to neurodegeneration in Alzheimer’s disease. J Neuroinflammation 2016, 13:1-12. Hreggvidsdottir HS, Lundberg AM, Aveberger A-C, Klevenvall L, Andersson U, Harris HE: High mobility group box protein 1 (HMGBl)-partner molecule complexes enhance cytokine production by signaling through the partner molecule receptor. Mol Med 2012, 18:224-30. Livoti E: Experimentally validated computational docking to characterize protein- protein interactions. [date unknown], Fassi EMA, Sgrignani J, D’Agostino G, Cecchinato V, Garfalo M, Grazioso G, Uguccioni M, Cavalli A: Oxidation state dependent conformational changes of HMGB1 regulates the formation of the CXCL12/HMGB1 heterocomplex. bioRxiv Biochem 2019, doi : 10.1101/555946. Drury LJ, Ziarek JJ, Gravel S, Veldkamp CT, Takekoshi T, Hwang ST, HevekerN, Volkman BF, Dwinell MB: Monomeric and dimeric CXCL12 inhibit metastasis through distinct CXCR4 interactions and signaling pathways. Proc Natl Acad Sci U S A 2011, 108:17655-17660. Chang S, Li Y, Yuan F, Qu M, Song Y, Zhang Z, Yang GY, Wang Y : Monomeric CXCL12 outperforms its dimeric and wild type variants in the promotion of human endothelial progenitor cells min function. Biochem Biophys Res Commun 2017, 488:303-310. Ziarek JJ, Kleist AB, London N, Raveh B, Montpas N, Bonneterre J, St-onge G, Dicosmo-ponticello CJ, Koplinski CA, Roy I, et al.: Structural basis for chemokine recognition by a G protein - coupled receptor and implications for receptor activation. Sci Signal 2017, 5756. Nady N, Min J, Kareta MS, Chedin F, Arrowsmith CH: Supplementary Materials: A SPOT on the chromatin landscape? Histone peptide arrays as a tool for epigenetic research. Trends Biochem Sci 2008, 33:305-313. Veldkamp CT, Peterson FC, Pelzek AJ, Volkman BF: The monomer - dimer equilibrium of stromal cell-derived factor- 1 (CXCL12) is altered by pH, phosphate, sulfate, and heparin. Protein Sci 2005, 1: 1071-1081. Akira S: Toll-like Receptors and Innate Immunity. In Edited by Dixon FJBT-A in I. Academic Press; 2001:1-56. Lee S, Piao C, Kim G, Kim JY, Choi E, Lee M: Production and application of HMGB1 derived recombinant RAGE-antagonist peptide for anti-inflammatory therapy in acute lung injury. Eur J Pharm Sci 2018, 114:275-284. Xu D, Young J, Song D, Esko JD: Heparan sulfate is essential for high mobility group protein 1 (HMGB1) signaling by the receptor for advanced gly cation end products (RAGE). J Biol Chem 2011, 286:41736-41744. Schlessinger K, Hall A, Tolwinski N: Wnt signaling pathways meet Rho GTPases. Genes Dev 2009, 23:265-77. Bryja V, Cervenka I, Cajanek L: The connections of Wnt pathway components with cell cycle and centrosome: side effects or a hidden logic? Crit Rev Biochem Mol Biol 2017, 52:614-637. Gurevich DB, Nguyen PD, Siegel AL, Ehrlich O V., Sonntag C, Phan JMN, Berger S, Ratnayake D, Hersey L, Berger J, et al.: Asymmetric division of clonal muscle stem cells coordinates muscle regeneration in vivo. Science (80- ) 2016, 353. Benjamin EJ, Muntner P, Alonso A, Bittencourt MS, Callaway CW, Carson AP, Chamberlain AM, Chang AR, Cheng S, Das SR, et al.: Heart Disease and Stroke Statistics-2019 Update: A Report From the American Heart Association. 2019. Limana F, Esposito G, D’Arcangelo D, Di Carlo A, Romani S, Melillo G, Mangoni A, Bertolami C, Pompilio G, Germani A, et al.: HMGB1 attenuates cardiac remodelling in the failing heart via enhanced cardiac regeneration and miR-206-mediated inhibition of TlMP-3. PLoSOne 2011, 6:1-11. Foglio E, Puddighinu G, Germani A, Russo MA, Limana F: HMGB1 Inhibits Apoptosis Following MI and Induces Autophagy via mTORCl Inhibition . J Cell Physiol 2017, 232:1135-1143. Bauza M del R, Gimenez CS, Locatelli P, De Lorenzi A, Hnatiuk A, Capogrossi MC, Crottogini A, Cuniberti L, Olea FD: High-dose intramyocardial HMGB1 induces long- term cardioprotection in sheep with myocardial infarction. Drug Deliv Transl Res 2019, 9:935-944. Savitsky P, Bray J, Cooper CDO, Marsden BD, Mahajan P, Burgess-Brown NA, Gileadi O: High-throughput production of human proteins for crystallization: The SGC experience. J Struct Biol 2010, 172:3-13. Yang OO, Swanberg SL, Lu Z, Dziejman M, McCoy J, Luster AD, Walker BD, Herrmann SH: Enhanced inhibition of human immunodeficiency virus type 1 by Met- stromal -derived factor lbeta correlates with down-modulation of CXCR4. J Virol 1999, 73:4582-4589. Crump MP, Gong JH, Loetscher P, Rajarathnam K, Amara A, Arenzana-Seisdedos F, Virelizier JL, Baggiolini M, Sykes BD, Clark-Lewis I: Solution structure and basis for functional activity of stromal cell-derived factor- 1; dissociation of CXCR4 activation from binding and inhibition of HIV-1. EMBO J 1997, 16:6996-7007. Turchetto J, Sequeira AF, Ramond L, Peysson F, Bras JLA, Saez NJ, Duhoo Y, Blemont M, Guerreiro CIPD, Quinton L, et al.: High-throughput expression of animal venom toxins in Escherichia coli to generate a large library of oxidized disulphide- reticulated peptides for drug discovery. Microb Cell Fact 2017, 16:6. Thein M, Sauer G, Paramasivam N, Grin I, Linke D: Efficient subfractionation of gram-negative bacteria for proteomics studies. JProteome Res 2010, 9:6135-6147. Aida Y, Pabst MJ: Removal of endotoxin from protein solutions by phase separation using triton X-l 14. J Immunol Methods 1990, 132:191-195. Teodorowicz M, Perdijk O, Verhoek I, Govers C, Savelkoul HFJ, Tang Y, Wichers H, Broersen K: Optimized Triton X-l 14 assisted lipopolysaccharide (LPS) removal method reveals the immunomodulatory effect of food proteins. PLoS One 2017, 12:e0173778. Chalk R: Mass Spectrometric Analysis of Proteins. Humana Press, New York, NY ; 2017:373-395. Chalk R, Berridge G, Shrestha L, Strain-Damerell C, Mahajan P, Yue W, Gileadi O, Burgess-Brown N: High-Throughput Mass Spectrometry Applied to Structural Genomics. Chromatography 2014, 1:159-175. Kaltashow IA, Mohimen A: Electrospray ionization mass spectrometry can provide estimates of protein surface areas in solution. Anal Chem 2005, 77:5370-5379. Testa L, Brocca S, Grandori R: Charge-surface correlation in electrospray ionization of folded and unfolded proteins. Anal Chem 2011, 83:6459-6463. Yu G, Wang LG, Han Y, He QY : ClusterProfiler: An R package for comparing biological themes among gene clusters. Omi A J Integr Biol 2012, doi: 10.1089/omi.2011.0118. Carr CA, Stuckey DJ, Tatton L, Tyler DJ, Hale SJM, Sweeney D, Schneider JE, Martin-Rendon E, Radda GK, Harding SE, et al.: Bone marrow-derived stromal cells home to and remain in the infarcted rat heart but fail to improve function: An in vivo cine-MRI study. Am J Physiol - Hear Circ Physiol 2008, 295. Lee S, Lee DK: Multiple Comparison Test and Its Imitations What is the proper way to apply the multiple comparison test? KJA. Korean J Anesth 2018, doi:10.4097/kja.d.18.00242. E. J. Benjamin, P. Muntner, A. Alonso, M. S. Bittencourt, C. W. Callaway, A. P. Carson, A. M. Chamberlain, A. R. Chang, S. Cheng, S. R. Das, F. N. Delling, L. Djousse, M. S. V. Elkind, J. F. Ferguson, M. Fomage, F. C. Jordan, S. S. Khan, B. M. Kissela, K. F. Knutson, T. W. Kwan, D. T. Fackland, T. T. Fewis, J. H. Fichtman, C. T. Fongenecker, M. S. Foop, P. F. Futsey, S. S. Martin, K. Matsushita, A. E. Moran, M. E. Mussolino, M. O'Flaherty, A. Pandey, A. M. Perak, W. D. Rosamond, G. A. Roth, U. K. A. Sampson, G. M. Satou, E. B. Schroeder, S. H. Shah, N. F. Spartano, A. Stokes, D. F. Tirschwell, C. W. Tsao, M. P. Turakhia, F. B. VanWagner, J. T. Wilkins, S. S. Wong, S. S. Virani, E. American Heart Association Council on, C. Prevention Statistics, S. Stroke Statistics, Heart Disease and Stroke Statistics-2019 Update: A Report From the American Heart Association. Circulation 139, e56-e528 (2019). G. B. D. DAFYs, H. Collaborators, Global, regional, and national disability-adjusted life-years (DAFYs) for 359 diseases and injuries and healthy life expectancy (HAFE) for 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet 392, 1859-1922 (2018). BritishHeartFoundation, bhf.org.uk/what-we-do/our-research/heart-statistics/heart- statistics-publications/cardiovascular-disease-statistics-2019, (2020). T. J. Cahill, R. K. Kharbanda, Heart failure after myocardial infarction in the era of primary percutaneous coronary intervention: Mechanisms, incidence and identification ofpatients at risk. World J Cardiol 9, 407-415 (2017). F. Qian, Y. Huang, C. I. Spencer, A. Foley, V. Vedantham, F. Fiu, S. J. Conway, J. D. Fu, D. Srivastava, In vivo reprogramming of murine cardiac fibroblasts into induced cardiomyocytes. Nature 485, 593-598 (2012). M. Xin, Y. Kim, F. B. Sutherland, M. Murakami, X. Qi, J. McAnally, E. R. Porrello,

A. I. Mahmoud, W. Tan, J. M. Shelton, J. A. Richardson, H. A. Sadek, R. Bassel- Duby, E. N. Olson, Hippo pathway effector Yap promotes cardiac regeneration. Proc Natl Acad Sci USA 110, 13839-13844 (2013). J. P. Feach, T. Heallen, M. Zhang, M. Rahmani, Y. Morikawa, M. C. Hill, A. Segura,

J. T. Willerson, J. F. Martin, Hippo pathway deficiency reverses systolic heart failure after infarction. Nature 550, 260-264 (2017). T. M. A. Mohamed, Y. S. Ang, E. Radzinsky, P. Zhou, Y. Huang, A. Elfenbein, A. Foley, S. Magnitsky, D. Srivastava, Regulation of Cell Cycle to Stimulate Adult Cardiomyocyte Proliferation and Cardiac Regeneration. Cell 173, 104-116 el 12 (2018). S. Hashmi, S. Al-Salam, Acute myocardial infarction and myocardial ischemia- reperfusion injury: a comparison. IntJ Clin Exp Pathol 8, 8786-8796 (2015). K. Stark, V. Philippi, S. Stockhausen, J. Busse, A. Antonelli, M. Miller, I. Schubert, P. Hoseinpour, S. Chandraratne, M. F. von Bruhl, F. Gaertner, M. Forenz, A. Agresti, R. Coletti, D. J. Antoine, R. Heermann, K. Jung, S. Reese, I. Faitinen, M. Schwaiger, A. Walch, M. Sperandio, P. P. Nawroth, C. Reinhardt, S. Jackel, M. E. Bianchi, S. Massberg, Disulfide HMGB1 derived from platelets coordinates venous thrombosis in mice. Blood 128, 2435-2449 (2016). S. Di Maggio, G. Milano, F. De Marchis, A. D'Ambrosio, M. Bertolotti, B. S. Palacios, I. Badi, E. Sommariva, G. Pompilio, M. C. Capogrossi, A. Raucci, Non-oxidizable HMGB1 induces cardiac fibroblasts migration via CXCR4 in a CXCL 12-independent manner and worsens tissue remodeling after myocardial infarction. Biochim Biophys Acta 1863, 2693-2704 (2017). J. Mersmann, F. Iskandar, K. Latsch, K. Habeck, V. Sprunck, R. Zimmermann, R. R. Schumann, K. Zacharowski, A. Koch, Attenuation of myocardial injury by HMGB1 blockade during ischemia/reperfusion is toll-like receptor 2-dependent. Mediators Inflamm 2013, 174168 (2013). C. Monaco, S. M. Gregan, T. J. Navin, B. M. Foxwell, A. H. Davies, M. Feldmann, Toll-like receptor-2 mediates inflammation and matrix degradation in human atherosclerosis. Circulation 120, 2462-2469 (2009). J. Oyama, C. Blais, Jr., X. Liu, M. Pu, L. Kobzik, R. A. Kelly, T. Bourcier, Reduced myocardial ischemia-reperfusion injury in toll-like receptor 4-deficient mice. Circulation 109, 784-789 (2004). K. R. Chien, J. Frisen, R. Fritsche-Danielson, D. A. Melton, C. E. Murry, L L. Weissman, Regenerating the field of cardiovascular cell therapy. Nat Biotechnol 37, 232-237 (2019). R. J. Vagnozzi, M. Maillet, M. A. Sargent, H. Khalil, A. K. Z. Johansen, J. A. Schwanekamp, A. J. York, V. Huang, M. Nahrendorf, S. Sadayappan, J. D. Molkentin, An acute immune response underlies the benefit of cardiac stem cell therapy. Nature 577, 405-409 (2020). C. E. Murry, W. R. MacLellan, Stem cells and the heart-the road ahead. Science 367, 854-855 (2020). S. Bollini, N. Smart, P. R. Riley, Resident cardiac progenitor cells: at the heart of regeneration. JMol Cell Cardiol 50, 296-303 (2011). P. K. Nguyen, E. Neofytou, J. W. Rhee, J. C. Wu, Potential Strategies to Address the Major Clinical Barriers Facing Stem Cell Regenerative Therapy for Cardiovascular Disease: A Review. JAMA Cardiol 1, 953-962 (2016). T. J. Cahill, R. P. Choudhury, P. R. Riley, Heart regeneration and repair after myocardial infarction: translational opportunities for novel therapeutics. Nat Rev Drug Discov 16, 699-717 (2017). H. Sadek, E. N. Olson, Toward the Goal of Human Heart Regeneration. Cell Stem Cell 26, 7-16 (2020). N. Smart, S. Bollini, K. N. Dube, J. M. Vieira, B. Zhou, S. Davidson, D. Yellon, J. Riegler, A. N. Price, M. F. Lythgoe, W. T. Pu, P. R. Riley, De novo cardiomyocytes from within the activated adult heart after injury. Nature 474, 640-644 (2011). A. I. Mahmoud, F. Kocabas, S. A. Muralidhar, W. Kimura, A. S. Koura, S. Thet, E. R. Porrello, H. A. Sadek, Meisl regulates postnatal cardiomyocyte cell cycle arrest. Nature 497, 249-253 (2013). K. Bersell, S. Arab, B. Haring, B. Kuhn, Neuregulinl/ErbB4 signaling induces cardiomyocyte proliferation and repair of heart injury. Cell 138, 257-270 (2009). S. Koudstaal, M. M. Bastings, D. A. Feyen, C. D. Waring, F. J. van Slochteren, P. Y. Dankers, D. Torella, J. P. Sluijter, B. Nadal-Ginard, P. A. Doevendans, G. M. Ellison, S. A. Chamuleau, Sustained delivery of insulin-like growth factor- 1/hepatocyte growth factor stimulates endogenous cardiac repair in the chronic infarcted pig heart. J Cardiovasc Transl Res 7, 232-241 (2014). K. Wei, V. Serpooshan, C. Hurtado, M. Diez-Cunado, M. Zhao, S. Maruyama, W.

Zhu, G. Fajardo, M. Noseda, K. Nakamura, X. Tian, Q. Liu, A. Wang, Y. Matsuura, P. Bushway, W. Cai, A. Savchenko, M. Mahmoudi, M. D. Schneider, M. J. van den Hoff, M. J. Butte, P. C. Yang, K. Walsh, B. Zhou, D. Bernstein, M. Mercola, P. Ruiz- Lozano, Epicardial FSTL1 reconstitution regenerates the adult mammalian heart. Nature 525, 479-485 (2015). K. Gabisonia, G. Prosdocimo, G. D. Aquaro, L. Carlucci, L. Zentilin, I. Secco, H. Ali,

L. Braga, N. Gorgodze, F. Bernini, S. Burchielli, C. Collesi, L. Zandona, G. Sinagra,

M. Piacenti, S. Zacchigna, R. Bussani, F. A. Recchia, M. Giacca, MicroRNA therapy stimulates uncontrolled cardiac repair after myocardial infarction in pigs. Nature 569, 418-422 (2019). A. Vujic, N. Natarajan, R. T. Lee, Molecular mechanisms of heart regeneration. Semin Cell Dev Biol, (2019). J. M. Vieira, S. Norman, C. Villa Del Campo, T. J. Cahill, D. N. Barnette, M. Gunadasa-Rohling, L. A. Johnson, D. R. Greaves, C. A. Carr, D. G. Jackson, P. R. Riley, The cardiac lymphatic system stimulates resolution of inflammation following myocardial infarction. J Clin Invest 128, 3402-3412 (2018). L. R. Fiedler, K. Chapman, M. Xie, E. Maifoshie, M. Jenkins, P. A. Golforoush, M. Bellahcene, M. Noseda, D. Faust, A. Jarvis, G. Newton, M. A. Paiva, M. Harada, D. J. Stuckey, W. Song, J. Habib, P. Narasimhan, R. Aqil, D. Sanmugalingam, R. Yan, L. Pavanello, M. Sano, S. C. Wang, R. D. Sampson, S. Kanayaganam, G. E. Taffet, L. H. Michael, M. L. Entman, T. H. Tan, S. E. Harding, C. M. R. Low, C. Tralau-Stewart, T. Perrior, M. D. Schneider, MAP4K4 Inhibition Promotes Survival of Human Stem Cell-Derived Cardiomyocytes and Reduces Infarct Size In Vivo. Cell Stem Cell 24, 579-591 e512 (2019). H. Aghajanian, T. Kimura, J. G. Rurik, A. S. Hancock, M. S. Leibowitz, L. Li, J. Scholler, J. Monslow, A. Lo, W. Han, T. Wang, K. Bedi, M. P. Morley, R. A. Linares Saldana, N. A. Bolar, K. McDaid, C. A. Assenmacher, C. L. Smith, D. Wirth, C. H. June, K. B. Margulies, R. Jain, E. Pure, S. M. Albelda, J. A. Epstein, Targeting cardiac fibrosis with engineered T cells. Nature 573, 430-433 (2019). N. Trainor, A. Pietak, T. Smith, Rethinking clinical delivery of adult stem cell therapies. Nat Biotechnol 32, 729-735 (2014). M. J. Foglia, K. D. Poss, Building and re-building the heart by cardiomyocyte proliferation. Development 143, 729-740 (2016). S. Oozawa, S. Mori, T. Kanke, H. Takahashi, K. Liu, Y. Tomono, M. Asanuma, I. Miyazaki, M. Nishibori, S. Sano, Effects of HMGB1 on ischemia-reperfusion injury in the rat heart. CircJ72, 1178-1184 (2008). T. Kitahara, Y. Takeishi, M. Harada, T. Niizeki, S. Suzuki, T. Sasaki, M. Ishino, O. Bilim, O. Nakajima, I. Kubota, High-mobility group box 1 restores cardiac function after myocardial infarction in transgenic mice. Cardiovasc Res 80, 40-46 (2008). Y. Nakamura, S. Suzuki, T. Shimizu, M. Miyata, T. Shishido, K. Ikeda, S. Saitoh, I. Kubota, Y. Takeishi, High Mobility Group Box 1 Promotes Angiogenesis from Bone Marrow-derived Endothelial Progenitor Cells after Myocardial Infarction. J Atheroscler Thromb 22, 570-581 (2015). F. Limana, G. Esposito, D. D'Arcangelo, A. Di Carlo, S. Romani, G. Melillo, A. Mangoni, C. Bertolami, G. Pompilio, A. Germani, M. C. Capogrossi, HMGB1 attenuates cardiac remodelling in the failing heart via enhanced cardiac regeneration and miR-206-mediated inhibition of TIMP-3. PLoS One 6, el9845 (2011). E. Foglio, G. Puddighinu, A. Germani, M. A. Russo, F. Limana, HMGB1 Inhibits Apoptosis Following MI and Induces Autophagy via mTORCl Inhibition. J Cell Physiol 232, 1135-1143 (2017). M. D. R. Bauza, C. S. Gimenez, P. Locatelli, A. De Lorenzi, A. Hnatiuk, M. C. Capogrossi, A. Crottogini, L. Cuniberti, F. D. Olea, High-dose intramyocardial HMGB1 induces long-term cardioprotection in sheep with myocardial infarction. Drug Deliv Transl Res 9, 935-944 (2019). M. E. Bianchi, Casalgrandi, M., Venereau, E. J., Brunelli, S., HMGB1 variants and uses thereof. European Patent Office - patentimages.storage.googleapis.com/cf/3c/6c/3b9d7ca70fl94a/EP2877248Bl. pdf

(2012). M. Tirone, N. L. Tran, C. Ceriotti, A. Gorzanelli, M. Canepari, R. Bottinelli, A.

Raucci, S. Di Maggio, C. Santiago, M. Mellado, M. Saclier, S. Francois, G. Careccia, M. He, F. De Marchis, V. Conti, S. Ben Larbi, S. Cuvellier, M. Casalgrandi, A. Preti, B. Chazaud, Y. Al-Abed, G. Messina, G. Sitia, S. Brunelli, M. E. Bianchi, E.

Venereau, High mobility group box 1 orchestrates tissue regeneration via CXCR4. J Exp Med 215, 303-318 (2018). G. Lee, A. I. Espirito Santo, S. Zwingenberger, L. Cai, T. Vogl, M. Feldmann, N. J. Horwood, J. K. Chan, J. Nanchahal, Fully reduced HMGB1 accelerates the regeneration of multiple tissues by transitioning stem cells to GAlert. Proc Natl Acad Sci USA 115, E4463-E4472 (2018).

Claims

CLAIMS What is claimed is:

1. A polypeptide represented by the following formula:

H2N-A-X-B-A-X-B-HOOC wherein A represents consecutive amino acids, the sequence of which (1) includes a sequence identical to the sequence of amino acids 90 - 93 of wild type HMGB1, (2) has at its amino terminal end, between one and six consecutive amino acids, the sequence of which is identical to the sequence of the corresponding one to six amino acids preceding amino acid 90 in wild type HMGB1, and optionally (3) has a methionine at the amino terminus; wherein X represents consecutive amino acids, the sequence of which is identical to the sequence of amino acids 94 - 162 of wild type HMGB1; and wherein B represents consecutive amino acids, the sequence of which (1) includes a sequence identical to the sequence of amino acids 163 - 168 of wild type HMGB1 and (2) has at its carboxy terminal end, between one and six consecutive amino acids, the sequence of which is identical to the sequence of the corresponding one to six amino acids following amino acid 168 in wild type HMGB1; and wherein each - represents a peptide bond between each of A and X, X and B,

B and A, A and X, and X and B.

2. A polypeptide of claim 1, wherein the methionine is present at the amino terminus.

3. A polypeptide of claim 1 or 2, wherein A has at its amino terminal end, one amino acid corresponding to amino acid 89 of wild type HMGB1.

4. A polypeptide of any one of claims 1-3, wherein B has at its carboxy terminal end, six amino acids corresponding to amino acids 169-174 of wild type HMGB1.

5. A composition comprising the polypeptide of any one of claims 1-4 and a carrier.

6. A pharmaceutical composition of claim 5, wherein the polypeptide is present in a therapeutically or prophylactically effective amount and the carrier is a pharmaceutically acceptable carrier.

7. A method of treating a subject suffering from, or at risk for developing, a condition which would be alleviated by promoting regeneration of a tissue or cells that rely upon CXCR4⁺ cells for repair which comprise administering to the subject the polypeptide of any one of claims 1-4 in an amount effective to promote regeneration of the tissue or a therapeutic or prophylactic dose of the pharmaceutical composition of Claim 6.

8. A method of claim 7, wherein the condition is myocardial infarction and the tissue is a cardiac tissue / myocardium.

9. The method of claim 8, wherein the polypeptide is administered within 5 hours of the myocardial infarction.

10. A method of claim 7, wherein the condition is a fracture and the tissue is a bone.

11. A method of claim 7, wherein the condition involves liver damage and the tissue is liver tissue.

12. A method of claim 7, wherein the condition involves damage to the brain or nervous system and includes stroke, Parkinson’s disease and dementia.

13. A method of claim 7, wherein the condition involves damage to the lung.

14. A method of claim 7, wherein the condition involves the gut and includes surgery and inflammatory bowel disease.

15. A method of claim 7, wherein the condition involves damage to the skin and includes surgical procedures, bums and ulcers.

16. A method of claim 7, wherein the condition involves the pancreas including type 1 diabetes and the cells are islet cells.

17. A method of claim 7, wherein the condition is neutropenia for example following chemotherapy and the tissue is bone marrow.

18. A method of claim 7, wherein the condition is kidney failure and the tissue is kidney tissue.