Digital tissue repair model

Abstract

To understand the regeneration process, a digital agent-based model composed of segments and growing circles modelling ECM fibers and differentiating cells from adipose tissue respectively was developed5 to reproduce the tissue repair process and allowed the identification of key factors that control the emergence of tissue architecture. It integrated three critical parameters in ECM organization: frequency of fiber synthesis as well as the probability of cross-linking and unlinking. The inventors have identified the key biomechanical factors and developed a predictive model of tissue repair outcome for a subject and new therapeutic approach to drive tissue regeneration targeting ECM cross-linking, in particular, at 10 an early and transient stage after injury. The present disclosure relates to a method of predicting tissue repair outcome of an injured tissue in a subject in need thereof and therapeutic use of extracellular cross- linking fibers inhibitors for regenerative healing of injured tissues.

Description

[0001] DIGITAL TISSUE REPAIR MODEL

[0002] TECHNICAL FIELD

[0003] The present disclosure relates to a method of predicting tissue repair outcome of an injured tissue in a subject in need thereof and therapeutic use of extracellular cross-linking fibers inhibitors for regenerative healing of injured tissues.

[0004] BACKGROUND

[0005] Regenerating normal tissue structure after injury is a central goal of regenerative medicine in adult mammals. Overall, striking regenerative properties are present in newborn mammals, but this capability rapidly declines and disappears a few days after birth. This loss is a major issue in spontaneous tissue repair since injury in adults usually leads to scarring rather than regeneration. While scarring rapidly blocks bleeding and reconstitutes a protective barrier (Ferguson, M. W. J. et al. Philos. Trans. R. Soc. B BioL Sci. 359, 839-850 (2004)), it has detrimental effects on tissue function. These adverse effects are due to the lack of recovery of tissue architecture and function due to inadequate extracellular matrix (ECM) structuring (Theocharis, A. D., et al. Adv. Drug Deliv. Rev. 97, 4-27 (2016)).

[0006] Classically, a putative therapeutic target is based on the identification of elements that differentiate regenerative from non-regenerative repair. However, these two repair outcomes are usually studied by comparing different species. Adult spiny mice (Acomys Cahirinus), which show regeneration in adults, have been increasingly used to investigate scar- free regeneration compared to Mus Musculus, the most commonly used mammal in research (Brant, J. O., et al. PLOS ONE 10, e0142931 (2015)). Recently, Sinha et al used the different regenerative capabilities of reindeer antler velvet compared to back skin, which forms fibrotic scars, to highlight the importance of the coupling between fibroblast and immune cells in repair processes (Sinha, S. et al. Cell 185, 4717-4736. e25 (2022)). Despite the power of these approaches, they can lead to context-specific insights rather than identifying factors specific to repair processes across tissues and species. Identification of regenerative medicine treatments based on these comparative studies may, therefore, fail to activate dormant regenerative capacities in adult mammal tissues across species.

[0007] To overcome these limitations, an original inducible model of adipose tissue (AT) regeneration in adult mammals was developed and provides a convenient model where regeneration and scarring can be investigated in the same tissue and developmental stage in animals with a shared genetic background (Labit, E. et al. Sci. Rep. 8, 12170 (2018); Rabiller, L. et al. NPJ Regen. Med. 6, 63 (2021); Rabiller, L. et al. NPJ Regen. Med. 6, 41 (2021); Berthezene, C. D. et al. Int. J. MoL Sci. 22, 7336 (2021)). In this model, a large resection of subcutaneous AT spontaneously drives tissue repair toward scar healing that can be switched toward tissue regeneration following a treatment with antagonist of opioid receptors (Labit, E. et al. Sci. Rep. 8, 12170 (2018); Rabiller, L. et al. NPJ Regen. Med. 6, 63 (2021)).

[0008] The complete recovery of tissue architecture during tissue regeneration depends on interactions between cells and ECM fibrillar components, which provide spatial information for cells. ECM organization is highly dynamic and results from a combination of three factors: its composition, its quantity, and temporally dynamic cross-links between fibrillar components. However, the dynamics and nature of these changes in adult tissue after injury requires further systematic investigations.

[0009] To address this complex biological question in a simpler way, and understand the regeneration process as a whole, a digital agent-based model (ABM), composed of two agents (segments and growing circles modelling ECM fibers and differentiating cells from AT respectively) was developed. This model qualitatively reproduced the tissue repair process and allowed the identification of key factors that control the emergence of tissue architecture such as mechanical cues (Peurichard, D. et al. J. Theor. BioL 429, 61 — 81 (2017); Peurichard, D. et al. J. Theor. BioL 469, 127-136 (2019)). It integrated three critical parameters in ECM organization: frequency of fiber synthesis as well as the probability of crosslinking and unlinking. Changing parameters showed that simple mechanical interactions between adipocytes and fibers may support the emergence of a lobular tissue architecture, similar to the one observed in vivo (Barreau, C. et al. Obesity 24, 1081-1089 (2016)).

[0010] However, it is noteworthy that the interdependence of the three parameters to generate ECM mechanical forces leading to tissue organization makes it difficult to biologically determine the respective importance of each parameter and to identify key biomechanical factors controlling the emergence of AT architecture. Therefore, there remains a need to determine the main parameters involved in the tissue repair outcome prediction.

[0011] SUMMARY

[0012] In the present application, the inventors have developed a predictive model of tissue repair. This model mimics both regeneration and scar healing-like architectures. They performed high-throughput multiparametric simulations (i.e. in silico experiments) in which each ECM parameter was independently changed over a wide range of values and used Machine Learning (ML) models to reveal that ECM cross-linking was the most important parameter explaining tissue repair outcome in silico. The digital model predicted that an early and transient decrease of ECM cross-linking after an injury is necessary and sufficient to drive tissue regeneration. Temporal calibration and in vivo experiments validated this digital model-based hypothesis and specified the kinetic. Indeed, by demonstrating that regeneration in adult mammals is unlocked by early and transient inhibition of ECM cross-linking, whereas increased cross-linking drives repair towards scar healing, the present study positions the digital model as a predictive tool for tissue regeneration. The present disclosure relates to a method of predicting tissue repair outcome for a subject having injured tissue, preferably lobular tissue, comprising: determining a set of parameter values relative to the subject, wherein the set of parameters are indicative of : o a level of synthesis of extra-cellular matrix (ECM) components, o an ability of the ECM components to bind, and o a capacity of the ECM components to be degraded, and predicting, based on the set of parameter values, an outcome of the tissue repair among at least the following outcomes: regenerative healing and scarring healing.

[0013] In a preferred embodiment, said parameter values indicative of an ability to bind of the ECM components comprise a gene expression level of at least one gene selected from the group consisting of: LOX-1, LOXL-1, LOXL-2, LOXL-3, LOXL-4, TGM1, TGM2, TGM3, TGM4, TGM5, TGM6, and TGM7 in a subject sample, parameter values indicative of a capacity of the ECM components to be degraded comprise a gene expression level of at least one gene selected from the group consisting of : MMP1; MMP8; MMP13, MMP2; MMP9; MMP3; MMP10; MMP7; MMP26; MMP14; MMP16, trypsin-2, macrophage elastase, neutrophil elastase, cathepsin K, cathepsin B; cathepsin L, cathepsin S, Meprin A subunit alpha, Meprin A subunit beta, RECK, TIMP-1, TIMP-2, TIMP-3, and TIMP-4 in a subject sample, and parameter values indicative of a level of synthesis of ECM components comprises a gene expression level of at least one gene selected from the group consisting of : COL1 Al, COL1 A2, Transforming Growth Factor beta (TGF-P), Endothelin-1 (ET-1), Platelet-Derived Growth Factor B (PDGF-BB), Interleukin- lb (IL- lb), Interferon g (IFN-g) and Basic Fibroblast Growth Factor (bFGF) in a subject sample, preferably wherein subject sample is a subject injured tissue sample.

[0014] In a more preferred embodiment, a lower expression of at least one gene selected from the group consisting of LOX-1, LOXL-1, LOXL-2, LOXL-3, LOXL-4, TGM1, TGM2, TGM3, TGM4, TGM5, TGM6, and TGM7 compared to a control value is indicative of a lower ability to bind of the ECM components, a lower expression of at least one gene selected from the group consisting of: MMP1; MMP8; MMP13, MMP2; MMP9; MMP3; MMP10; MMP7; MMP26; MMP14; MMP16; macrophage elastase, neutrophil elastase, trypsin-2, cathepsin K, cathepsin B; cathepsin L, cathepsin S, Meprin A subunit alpha, Meprin A subunit beta, and RECK compared to a control value and / or a higher expression of at least one gene selected from the group consisting of: TIMP-1, TIMP-2, TIMP-3, and TIMP-4 compared to a control value is indicative of a higher capacity of the ECM components to be degraded, and a lower expression of at least one gene selected from the group consisting of: COL1 Al ; COL1 A2, Transforming Growth Factor beta (TGF-P), Endothelin-1 (ET-1), Platelet-Derived Growth Factor B (PDGF-BB), Interleukin- lb (IL- lb), Interferon g (IFN-g) and Basic Fibroblast Growth Factor (bFGF) compared to a control value is indicative of a lower frequency of ECM components synthesis. In some embodiment, said tissue repair outcome is predicted by application to the determined set of parameters valued of a trained model configured to classify the type of tissue repair outcome among a plurality of classes comprising at least a regenerative healing class and a scarring healing class, preferably wherein the model has been trained by supervised learning on a training dataset formed by data pairs where each pair comprises a set of parameter values, and a corresponding tissue repair class and the training dataset comprises a plurality of data pairs for at least one set of parameter values.

[0015] In a preferred embodiment, the training of the model iurther comprises a step of obtaining the data pairs of the training dataset, by performing, for each set of parameter values, numerical simulation of tissue repair, and determining, from said numerical simulation, the corresponding tissue repair class, wherein the numerical simulation is performed by an agent-based model, in which: a tissue comprising a lesion is simulated, wherein the tissue is represented by two types of agents, comprising segments modelling ECM fibers and growing circles representing differentiating cells, and a set of interactions are defined between the two types of agents, wherein a repair index is computed from the outcome of the numerical simulation, and the tissue repair class is determined from the value of the repair index, preferably wherein the repair index from the following parameters:

[0016] An average alignment of bundles of fibers,

[0017] A number of cells,

[0018] A number of clusters of cells,

[0019] An elongation rate of the clusters of cells.

[0020] In a preferred embodiment, the repair index is computed as follows:

[0021] Where:

[0022] AE quantifies a variation in elongation rate of the clusters of cells before and after tissue repair, Eoquantifies an elongation rate of the clusters of cells before tissue repair, A1VCquantifies a variation in the number of clusters of cells before and after tissue repair, Ncoquantifies a number of clusters of cells before tissue repair,

[0023] A l quantifies a variation in average alignment of bundles of fibers before and after tissue repair, d0quantifies an average alignment of bundles of fibers before tissue repair, A / V£quantifies a variation in the number of cells before and after tissue repair, NEOquantifies a number of cells before tissue repair. In a more preferred embodiment, the repair index is computed only on the area of the tissue corresponding to the lesion. In a particular embodiment, the set of parameters further comprises a value of time from the appearance of the lesion, a parameter describing intensity of the lesion and / or at least a parameter indicative of a distance between the edges of the lesion and / or a parameter indicative of a level of degradation of the tissue within the lesion.

[0024] The present disclosure also relates to a computer-program product comprising code instructions for implementing the method according to any of the preceding steps, when it is executed by a processor.

[0025] In another aspect, the present disclosure relates to a Lysyl oxidase (LOX) inhibitor for use in the regenerative healing of an injured tissue, preferably injured lobular tissue, more preferably adipose tissue, in a subject in need thereof.

[0026] In a preferred embodiment, the present disclosure relates to a Lysyl oxidase (LOX) inhibitor for use in the regenerative healing of an injured tissue, preferably injured lobular tissue, more preferably adipose tissue, in a subject in need thereof, wherein the injured tissue is previously predicted to have a scarring repair outcome by a method as described above. In a preferred embodiment, said inhibitor is administered no more than 6 days after the injury. In some embodiments, the lesion, is a round or square lesion wherein the ratio of cumulative length of the edges of the injury on the surface area of the injury is higher than 2.5 mm or for a lesion with two open edges wherein the distance between the 2 edges is higher than 5 mm.

[0027] The LOX inhibitor according to the present disclosure is preferably selected from the group consisting of: P-aminopropionitrile (BAPN), -aminopropionitrile fumarate (BAF), P-aminopropionitrile maleate (BAM), P-aminopropionitrile monohydrate (BAMH), aminomethylenethiophene (AMT), 5- (naphthalen-2-ylsuIfonyl)thiophen-2-yl)methamine 1, 2-aminomethylene-5-suIponylthiazole (AMTz), 5-(naphthalen-2-ylsulfonyl)thiazol-2-yl)methanamine 2, PXS-5505 (CAS No.: 2409963-83-1), PXS- 5382 and Simtuzumab.

[0028] LEGEND FIGURES

[0029] Figure 1: In silico modelling of AT repair outcomes. A) Cellular (left panel) and fibrillar (right panel) factors used in the agent-based in silico model of AT repair. Rmin corresponds to the minimal radius of cell. B) Simulation results after injury leading to scar repair is defined by a repair index higher than 2 (upper panel), unstructured tissue is defined by a repair index from 0.7 to 2 (middle panel) and regenerative repair is defined by a repair index below 0.7 during tissue repair steps (lower panel).

[0030] Figure 2: In silico model description. A) Table presenting the three fiber parameters in the in silico model of AT repair and their biological counterpart. B) Table presenting the quantifiers extracted from simulations and their description. C) Formula of the 'repair index' as a weighted norm of the quantifiers before and after injury. EO = 0.6, A0= 0.2, NC0= 10 and NEO = 0.2 were chosen for all simulations and correspond to the value of the quantifiers difference between regular AT morphology and a scar.

[0031] Figure 3: In silico modelling: fiber cross-linking plays a major role in tissue repair outcome prediction. A) 3D scatter plot of repair index over a wide range of values of three parameters used in in silico modelling (Pf, vf, vad). B) ML pipeline for the predictive model. C) RF SHAP-values when varying each parameter for a regenerative outcome. The color of each violin represents the value of the associated parameter — dark grey for high values and light grey for low values of the parameter. A positive SHAP value means that the parameter contributes positively to the prediction, whereas a negative SHAP value means that the parameter contributes negatively to the prediction. The parameter ranking indicates the importance in the class (repair outcome) prediction: the parameter at the top is the most important for predicting the class. D) RF SHAP-values according to each parameter for the scarring class. E) RF SHAP dependency plot for parameter values in the regenerative class. F) RF SHAP dependency plot of each parameter value for the scarring class.

[0032] Figure 4: ML analysis. A) Histogram representing the number of simulations for each class of the dataset. Simulations were split into train and test groups (80:20) for each class (regenerative, unstructured, scar). B) Table of the accuracy score of each parameter for the two ML algorithms used in the dataset analysis. C) Confusion matrix of the predicted class attributed by Random Forest algorithm and the real class. D) Confusion matrix of the predicted class attributed by XGBoost algorithm and the real class. E) XGBoost SHAP-values according to each parameter for the regenerative class, f XGBoost SHAP-values according to each parameter for the scarring class. G) Table of the feature importance score of each parameter for the two ML algorithms used in the dataset analysis. H) Total number of cross-links over simulation time upon regenerative repair, unstructured tissue and scar repair conditions (n=50 for each condition). Data are expressed as mean +1- SEM and analysed by Tukey's test. **** p< 0.0001.

[0033] Figure 5: In silico modelling: Identification and validation of an early and transient treatment window where fiber cross-linking modulation can drive tissue repair outcomes. A) Total number of cross-links over time upon regenerative repair, unstructured tissue and scar repair conditions (n=7 for each condition). B) Simulation results after injury for a Pf =0.7 corresponds to scar repair. A transient Pf decrease (Pf=0.1 ) during the first 6 days leads to regenerative repair. C) Simulation results after injury for a Pf =0.2 corresponding to regenerative repair or for a transient Pf increase (Pf=0.8) during the first 6 days leads to scarring repair. D) Histogram of the repair index when Pf is constant throughout the simulation (scar and regenerative repair) or for a transient Pf decrease (Pf=0.1) or increase (Pf=0.8) during the first 6 days. (n=6 for each condition). E) Total number of cross-links over time upon regenerative and scar repair or for a transient increase or decrease in Pf. Data are expressed as mean +1- SEM and analysed by Tukey's test. **** p< 0.0001. Figure 6: In vivo validation: Early and transient fiber cross-linking modulation drives tissue repair outcomes. A) Morphological image of AT after scarring repair (Vehicle), upon transient inhibition of cross-links (BAPN), under regenerative conditions (Nal-M) and upon transient increase of cross-links (Nal-M + Genipin). Images were taken 1 -month post-injury. B) Box-plot of light transmittance after Vehicle, BAPN, Nal-M and Nal-M+Genipin treatments. C) Representative fluorescence (300 pm thick tissue sections) images of regenerative (Nal-M and BAPN) and scarring conditions (Vehicle and Nal-M+ Genipin) 1 -month post-injury. The two lines represent the repaired area. Red, magenta and green colors correspond to adipocytes, nucleus and collagen respectively. Scale bars represent 2,000 pm (panel A) and 100 pm (fluorescence images). D) Histogram of the repaired area (%) based on the ratio of adipocyte area compared to the whole repaired area upon regenerative and scarring repair. E) Boxplot of fractal dimension under regenerative and scarring conditions 3 days postinjury. Data are expressed as mean +1- SEM. Data analysed by Tukey's test. * p<0.05; ** p<0.01; *"* p< 0.0001.

[0034] Figure 7 : Schematic representation of a computing device for implementing a method for predicting tissue repair outcome according to embodiments.

[0035] DETAILED DESCRIPTION

[0036] In the present application, the inventors have developed a digital model to act as a predictive model of tissue repair.

[0037] The present disclosure relates to a method of predicting tissue repair outcome for a subject having an injured tissue comprising: determining a set of parameters values relative to the subject, wherein the set of parameters are indicative of: o a level of synthesis of extracellular matrix (ECM) components, o a capacity of the ECM components to be degraded, o an ability of the ECM components to bind, predicting tissue repair outcome among the following classes: regeneration (regenerative- healing) and scar-healing.

[0038] Tissue repair, also named wound healing is the process of restoring tissue architecture and function after an injury. It involves two processes: regenerative and scar healing.

[0039] Regeneration is the process of replacing damaged or lost tissue with new tissue that has the same structure and function as the original tissue. Regeneration is the ideal outcome of tissue repair because it restores the tissue to its original state without any loss of function. Scarring, on the other hand, is the process of rapidly replacing damaged or lost tissue with fibrous connective tissue. Scarring is characterized by the formation of a fibrous scar tissue that lacks the structure and function of the original tissue.

[0040] The injured tissue according to the present disclosure can be any tissue of a subject. In a preferred embodiment, said tissue is lobular tissue. The lobular tissues are tissues that are divided into small units called lobules. Lobular tissues can be pulmonary lobule tissue, hepatic lobule tissue, breast lobule tissue or adipose tissue.

[0041] The pulmonary lobules, for example, are the smallest functional units of the respiratory tree of the lungs. They are composed of several bronchioles, small blood vessels called capillaries, as well as structures called alveoli. The very thin wall of each of these alveoli is in contact with the lung’s capillary vessels, which allows gas exchange. The pulmonary lobules are surrounded by connective tissue and are separated from each other by partitions of this same tissue. Hepatic lobules tissue is composed of building blocks of liver tissue, consisting of a portal triad, hepatocytes arranged in linear cords between a capillary network, and a central vein. Breast lobule tissue is composed of several small ducts that converge to form a terminal duct lobular unit. Adipose tissues are made up of lobules of juxtaposed adipocytes, separated by richly vascularized connective septa. In a preferred embodiment, said tissue is an adipose tissue.

[0042] The terms "subject" and "patient" are used interchangeably herein and refer to both human and nonhuman animals. As used herein, the term “patient” denotes a mammal, such as a rodent, a feline, a canine, and a primate. Preferably, a patient according to the invention is a human. According to the present disclosure, said subject is a subject having an injured tissue (i.e., a tissue with a lesion).

[0043] The term “subject sample” means any biological sample derived from a patient. Examples of such samples include tissue sample, cell samples, organs, biopsies, preferably an injured tissue sample. The tissue sample used in the context of the present disclosure is typically obtained from an injured tissue biopsy, ft can e.g. be a fresh or a preserved sample such as a frozen sample, or any sample preserved by other means.

[0044] According to a preferred embodiment of the present disclosure, the injury (lesion) is not a small injury (lesion), fn a particular embodiment, if said lesion is a round or square lesion, the ratio of cumulative length of the edges of the injury on the surface area of the injury is preferably higher than 2.5 mm or for a lesion with two open edges, the distance between the 2 edges is preferably higher than 5 mm.

[0045] Frequency of ECM fiber cross-linking

[0046] An ability of extracellular matrix (ECM) components to bind, which can be expressed for instance as a frequency of ECM fiber cross-linking, may be evaluated by determining the gene expression level by any techniques suitable to estimate the content in RNA (northern-blot, RT-QPCR, transcriptomics approaches) m a subject sample of at least one gene selected from the group consisting of: LOX-1, LOXL-1, LOXL-2, LOXL-3, LOXL-4, TGM1, TGM2, TGM3, TGM4, TGM5, TGM6, and TGM7.

[0047] In particular, a lower expression of at least one gene, preferably2, 3, 4, 5, 6, 7, 8, 9, 10, 11, or 12 genes selected from the group consisting of LOX-1, LOXL-1, LOXL-2, LOXL-3, LOXL-4, TGM1, TGM2, TGM3, TGM4, TGM5, TGM6, and TGM7 compared to a control value is indicative of a lower ability of ECM components to bind.

[0048] Human LOX-1 gene, also named Oxidized low-density lipoprotein receptor 1 (ORL1) gene (Gene ID: 4973, updated on 23 November 2023) encodes a low density lipoprotein receptor that belongs to the C- type lectin superfamily (UniProtKB: P78380, updated on November 8, 2023). This protein may be involved in the regulation of Fas-induced apoptosis. This protein may play a role as a scavenger receptor. Alternate splicing results in multiple transcript variants (NCBI Reference Sequence: NM 002543.3, updated on 18 September 2023, NCBI Reference Sequence: NM_001172632.1, updated on 22 September 2023, NCBI Reference Sequence: NM_001172633.1 updated on 22 September 2023).

[0049] Human Lysyl oxidase like 1 (LOXL1) gene (Gene ID: 4016, updated on 23 November 2023), human Lysyl oxidase like 2 (LOXL2) gene (Gene ID: 4017, updated on 23 November 2023), human Lysyl oxidase like 2 (LOXL3) gene (Gene ID: 84695, updated on 25 November 2023) and human Lysyl oxidase like 2 (LOXL4) gene (Gene ID: 84171, updated on 23 November 2023) encode each a member of the lysyl oxidase family of proteins, LOXL1 protein (UniprotKB: Q08397, updated on 08 November 2023), LOXL2 protein (UniprotKB: Q9Y4K0, updated on 08 November 2023), LOXL3 protein (UniprotKB: P58215, updated on 08 November 2023) and LOXL4 protein (UniprotKB: Q96JB6, updated on 08 November 2023) respectively. The prototypic members of the family are essential to the biogenesis of connective tissue, encoding extracellular copper-dependent amine oxidase that catalyze the first step in the formation of crosslinks in collagen and elastin. The corresponding transcripts are LOXL1 transcript (NCBI Reference Sequence: NM 005576.4, last updated on 14 October 2023), LOXL2 transcript (NCBI Reference Sequence: NM 002318.3, last updated on 16 November 2023), LOXL3 transcript (NCBI Reference Sequence: NM 032603.5, last updated on 26 March 2001), LOXL4 transcript (NCBI Reference Sequence: NM 032211.7, last updated on 04 June 2023).

[0050] Human transglutaminase 1 (TGM1) gene (Gene ID: 7051, updated on November 23, 2023), human transglutaminase 2 (TGM2) gene (Gene ID: 7052, updated on November 23, 2023), human transglutaminase 3 (TGM3) gene (Gene ID: 7053, updated on November 23, 2023), human transglutaminase 4 (TGM4) gene (Gene ID: 7047, updated on November 23, 2023), human transglutaminase 5 (TGM5) gene (Gene ID: 9333, updated on November 23, 2023), human transglutaminase 6 (TGM6) gene (Gene ID: 343641, updated on November 23, 2023), human transglutaminase 7 (TGM7) gene (Gene ID: 116179, updated on November 23, 2023) encode human protein-glutamine gamma-glutamyltransferase K (TGM1) protein (UniprotKB: P22735, updated on 8 November 2023), human protein-glutamine gamma-glutamyltransferase 2 (TGM2) protein (UniprotKB: P21980, updated on 8 November 2023), human protein-glutamine gamma-glutamyltransferase E (TGM3) protein (UniprotKB: Q08188, updated on 8 November 2023), human protein-glutamine gamma-glutamyltransferase 4 (TGM4) protein (UniprotKB: P49221, updated on 8 November 2023), human protein-glutamine gamma-glutamyltransferase 5 (TGM5) protein (UniprotKB: 043548, updated on 8 November 2023), human protein-glutamine gamma-glutamyltransferase 6 (TGM6) protein (UniprotKB: 095932, updated on 8 November 2023) and human protein-glutamine gamma- glutamyltransferase Z (TGM7) protein (UniprotKB: Q96PF1, updated on 8 November 2023) respectively that catalyzes the crosslinking of proteins by epsilon-gamma glutamyl lysine isopeptide bonds. The corresponding transcripts are TGM1 transcript (NCBI Reference Sequence: NM 000359.2, last updated on 14 November 2023), TGM2 transcript (NCBI Reference Sequence: NM_001323316.2, NCBI Reference Sequence: NM_004613.3, NCBI Reference Sequence: NM_198951.3, updated on 30 Oct 2023), TGM3 transcript (NCBI Reference Sequence: NM_003245.3, last updated on 10 September 2023), TGM4 transcript (NCBI Reference Sequence: NM_003241.4, last updated on 12 February 2023), TGM5 transcript (NCBI Reference Sequence: NM_003241.4, NCBI Reference Sequence: NM 201631.4, last updated on 15 and 16 November 2023), TGM6 transcript (NCBI Reference Sequence: NM_001254734.2, NCBI Reference Sequence: NM 198994.2, last updated on 16 November 2023), TGM7 transcript (NCBI Reference Sequence: NM 052955.3, last updated on 25 December 2022).

[0051] Frequency of ECM fiber delinking

[0052] The capacity of the ECM components to be degraded, which can be expressed for instance as a frequency of extracellular matrix fiber delinking, may be evaluated by determining the gene expression level in a subject sample of at least one gene selected from the group consisting of: MMP1; MMP8; MMP13, MMP2; MMP9; MMP3; MMP10; MMP7; MMP26; MMP14; MMP16; macrophage elastase, neutrophile elastase, trypsin-2, cathepsin K, cathepsin B; cathepsin L, cathepsin S, Meprin A subunit alpha, Meprin A subunit beta, RECK, TIMP-1, TIMP-2, TIMP-3, and TIMP-4.

[0053] In particular, a lower expression of at least one gene, preferably 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 or 16 genes selected from the group consisting of: MMP1; MMP8; MMP13, MMP2; MMP9; MMP3; MMP10; MMP7; MMP26; MMP14; MMP16; macrophage elastase, neutrophil elastase, trypsin-2, cathepsin K, cathepsin B; cathepsin L, cathepsin S, Meprin A subunit alpha, Meprin A subunit beta, and RECK compared to a control value and / or a higher expression of at least one gene, preferably 2, 3 or 4 genes selected from the group consisting of: TIMP-1, TIMP-2, TIMP-3, and TIMP-4 compared to a control value is indicative of a higher capacity of ECM components to be degraded.

[0054] Human Matrix metallopeptidase 1 (MMP1) gene (Gene ID: 4312, updated on November 23, 2023), human matrix metallopeptidase 2 (MMP2) gene (Gene ID: 4313, updated on November 28, 2023), matrix metallopeptidase 3 (MMP3) gene (Gene ID: 4314, updated on November 27, 2023), human matrix metallopeptidase 7 (MMP7) gene (Gene ID: 4316, updated on November 23, 2023), human matrix metallopeptidase 8 (MMP8) gene (Gene ID: 4317, updated on November 23, 2023), human matrix metallopeptidase 9 (MMP9) gene (Gene ID: 4318, updated on November 28, 2023), human matrix metallopeptidase 10 (MMP10) gene (Gene ID: 4319, updated on November 23, 2023), human matrix metallopeptidase 13 (MMP13) gene (Gene ID: 4322, updated on November 30, 2023), human matrix metallopeptidase 14 (MMP14) gene (Gene ID: 4323, updated on November 27, 2023), human matrix metallopeptidase 16 (MMP16) gene (Gene ID:4325, updated on November 23, 2023), and human matrix metallopeptidase 26 (MMP26) gene (Gene ID: 56547, updated on November 23, 2023), encode Interstitial collagenase, also named MMP1 protein (UniprotKB: P03956, updated on 8 November 2023), 72 kDa type IV collagenase, also named MMP2 protein (UniprotKB: P08253, updated on 8 November 2023), Stromelysin- 1, also named MMP3 protein (UniprotKB: P08254, updated on 8 November 2023), Matrilysin, also named MMP7 protein (UniprotKB: P09237, updated on 8 November 2023), Neutrophil collagenase, also named MMP8 protein (UniprotKB: P22894, updated on 8 November 2023), Matrix metalloproteinase-9, also named MMP9 protein (UniprotKB: P14780, updated on 8 November 2023), Stromelysin-2, also named MMP10 protein (UniprotKB: P09238, updated on 8 November 2023), Collagenase 3, also named MMP13 protein (UniprotKB: P45452, updated on 8 November 2023), Matrix metalloproteinase- 14, also named MMP14 protein (UniprotKB: P50281, updated on 8 November 2023), Matrix metalloproteinase- 16, also named MMP 16 protein (UniprotKB: P51512, updated on 8 November 2023), and Matrix metalloproteinase-26, also named MMP26 protein (UniprotKB: Q9NRE1, updated on 8 November 2023). Proteins in this family are involved in the breakdown of extracellular matrix in normal physiological processes, such as embryonic development, reproduction, and tissue remodeling, as well as in disease processes, such as arthritis and metastasis. The encoded preproprotein is proteolytically processed to generate the mature protease. This secreted protease breaks down the interstitial collagens, including types I, II, and III. Alternative splicing results in multiple transcript variants, at least one of which encodes an isoform that is proteolytically processed. The corresponding transcripts are MMP1 transcript (NCBI Reference Sequence: NM 002421.4, last updated on 20 November 2023), MMP2 transcript (NCBI Reference Sequence: NM_001127891.3, NCBI Reference Sequence: NM_001302508.1, NCBI Reference Sequence: NM_001302509.2, NCBI Reference Sequence: NM 001302510.2, NCBI Reference Sequence: NM 004530.6, updated on 4 December 2023), MMP3 transcript (NCBI Reference Sequence: NM_002422.5, last updated on 27 November 2023), MMP7 transcript (NCBI Reference Sequence: NM_002423.5, last updated on 15 October 2023), MMP8 transcript (NCBI Reference Sequence: NM_001304441.2, NCBI Reference Sequence: NM_001304442.2, NCBI Reference Sequence: NM_002424.3, last updated on 16 September 2023), MMP9 transcript (NCBI Reference Sequence: NM 004994.3, last updated on 04 December 2023), MMP10 transcript (NCBI Reference Sequence: NM 002425.3, last updated on 01 January 2023), MMP13 transcript (NCBI Reference Sequence: NM_002427.4, last updated on 04 December 2023), MMP14 transcript (NCBI Reference Sequence: NM_004995.4, last updated on 27 November 2023), MMP16 transcript (NCBI Reference Sequence: NM_005941.5, last updated on 21 November 2023), MMP26 transcript (NCBI Reference Sequence: NM 021801.5, last updated on 28 December 2022).

[0055] Human Cathepsin K (CTSK) gene (Gene ID: 1513, updated on November 23, 2023), is transcriptionally expressed in CTSK mRNA (NCBI Reference Number: NM 000396.4, updated on November 14, 2023) that encodes a lysosomal cysteine proteinase involved in bone remodeling and resorption (UniprotKB: P43235, updated on 8 November 2023).

[0056] Human Cathepsin B (CTSB) gene (Gene ID: 1508, updated on November 29, 2023) encodes a member of the Cl family of peptidases, Cathepsin B (UniprotKB: P07858, updated on 8 November 2023). Alternative splicing of this gene results in multiple transcript variants (NCBI Reference Sequences: NM_001908.5, NM_147780.4, NMJ47781.4, NMJ47782.4, NM_147783.4, last updated on 4 December 2023). At least one of these variants encodes a preproprotein that is proteolytically processed to generate multiple protein products.

[0057] Human Cathepsin L (CTSL) gene (Gene ID: 1514, updated on November 23, 2023) encodes lysosomal cysteine proteinase that plays a major role in intracellular protein catabolism, Cathepsin L (UniprotKB: P07711, updated on 8 November 2023). Alternative splicing of this gene results in multiple transcript variants (NCBI Reference Sequences: NM 145918.3, NM 001912.5, NM 001257972.2, NM_001257971.2, last updated on 09 October 2023). At least one of these variants encodes a preproprotein that is proteolytically processed to generate multiple protein products.

[0058] Human Cathepsin S (CTSS) gene (Gene ID: 1520, updated on November 23, 2023) encodes a lysosomal cysteine proteinase Cathepsin S (UniprotKB: P25774, updated on 8 November 2023), that participates in the degradation of antigenic proteins to peptides for presentation on MHC class II molecules. The mature protein cleaves the invariant chain of MHC class II molecules in endo lysosomal compartments and enables the formation of antigen-MHC class II complexes and the proper display of extracellular antigenic peptides by MHC-II. The mature protein also functions as an elastase over a broad pH range. When secreted from cells, this protein can remodel components of the extracellular matrix such as elastin, collagen, and fibronectin. Alternative splicing of this gene results in multiple transcript variants (NCBI Reference Sequences: NM_001199739.2, NM_004079.5, last update 27 August 2023). At least one of these variants encodes a preproprotein that is proteolytically processed to generate multiple protein products.

[0059] Human TIMP metallopeptidase inhibitor 1 (TIMP1) gene (Gene ID: 7076, updated on November 23, 2023), human TIMP metallopeptidase inhibitor 2 (TIMP2) gene (Gene ID: 7077, updated on November 23, 2023), human TIMP metallopeptidase inhibitor 3 (TIMP3) ) gene (Gene ID: 7078, updated on November 23, 2023), human TIMP metallopeptidase inhibitor 4 (TIMP4) gene (Gene ID: 7079, updated on November 23, 2023), encode natural inhibitors of the matrix metalloproteinases (MMPs), human metalloproteinase inhibitor 1 protein (UniprotKB: P01033, updated on November 8, 2023), human metalloproteinase inhibitor 2 protein (UniprotKB: P16035, updated on November 8, 2023), human metalloproteinase inhibitor 3 protein (UniprotKB: P35625, updated on November 8, 2023), human metalloproteinase inhibitor 4 protein (UniprotKB: Q99727, updated on November 8, 2023), a group of peptidases involved in degradation of the extracellular matrix. Alternative splicing of this gene results in multiple transcript variants (TIMP1 (NCBI Reference Sequence: NM 003254.3, last updated on 21 November 2023), TIMP2 (NCBI Reference Sequence: NM_003255.5, last updated on 30 Oct 2023, TIMP3 (NCBI Reference Sequence: NM 000362.5, last updated 27 Aug 2023), TIMP4 (NCBI Reference Sequence: NM 003256.4, last updated on 14 August 2023).

[0060] Human Macrophage elastase gene, also known as matrix metallopeptidase 12 gene (Gene ID: 4312, updated on 12 Dec. 2023) encodes a member of the peptidase M10 family of matrix metalloproteinases (MMPs) (UniprotKB: P39900, updated on 08 Nov. 2023, transcript: NCBI Reference Sequence: NM 002426.6, updated on 12 Dec. 2023) that degrades soluble and insoluble elastin.

[0061] Human Neutrophil elastase, also named elastase, neutrophil expressed (ELANE) (Gene ID: 1991, updated on 23 Nov 2023) encodes protease (UniprotKB: P08246, updated on 08 Nov. 2023, transcript: NCBI Reference Sequence: NM_001972.4, updated on 12 Nov. 2023) that hydrolyzes proteins within specialized neutrophil lysosomes, called azurophil granules, as well as proteins of the extracellular matrix. The enzyme may play a role in degenerative and inflammatory diseases through proteolysis of collagen-IV and elastin.

[0062] Human Trypsin-2, also named serine protease 2 (PRSS2) (Gene ID: 5645, updated on 23 November 2023) encodes an enzyme (UniprotKB: P07478, updated on November 08, 2023) that cleave peptide bonds that follow lysine or arginine residues. This enzyme is able to cleave across the type II collagen triple helix in rheumatoid arthritis synovitis tissue, potentially participating in the degradation of type II collagen-rich cartilage matrix. Alternative splicing results in multiple transcript variants (NCBI Reference Sequence: NM_002770.4, last updated 14 Aug 2023). .

[0063] Human Meprin A subunit alpha (MEP1A) gene (Gene ID: 4224, updated on 23 November 2023) encodes subunit alpha of meprin A (UniprotKB: Q16819, updated on November 08, 2023, transcript (NCBI Reference Sequence: NM_005588.3, last updated on 13 Nov 2023) predicted to be involved in proteolysis.

[0064] Human Meprin A subunit beta (MEP1B) gene (Gene ID: 4225, updated on 23 November 2023) encodes subunit beta of meprin A (UniprotKB: Q16820, updated on November 08, 2023, transcript (NCBI Reference Sequence: NM_001308171.2, last updated 8 Jul 2023, NM_005925.3, last updated on 16 Nov 2023) predicted to be involved in proteolysis and involved in tissue remodeling due to its capability to degrade extracellular matrix components. Human reversion inducing cysteine rich protein with kazal motifs (RECK) gene (Gene ID: 8434, updated on 23 November 2023) encodes a cysteine-rich, extracellular protein (UniprotKB: 095980, updated on 8 November 2023, transcript: NCBI Reference Sequence: NM 021111.3, last updated on 08 nov 2023) with protease inhibitor-like domains. This membrane-anchored glycoprotein may serve as a negative regulator for matrix metalloproteinase-9. Several transcript variants encoding different isoforms have been found for this gene.

[0065] Frequency of fiber synthesis

[0066] The level of synthesis of ECM components, which can be expressed for instance as a frequency of ECM fiber synthesis, may be evaluated by determining the gene expression level in a subject sample of at least one gene, preferably 2, 3, 4, 5, 6, 7 or 8 genes selected from the group consisting of: COL1 Al ; COL1A2, Transforming Growth Factor beta (TGF-P), Endothelin-1 (ET-1), Platelet-Derived Growth Factor B (PDGF-BB), Interleukin- lb (IL-lb), Interferon g (IFN-g) and Basic Fibroblast Growth Factor (bFGF).

[0067] In particular, a lower expression of at least one gene selected from the group consisting of: COL1A1 ; COL1A2, Transforming Growth Factor beta (TGF-P), Endothelin-1 (ET-1), Platelet-Derived Growth Factor B (PDGF-BB), Interleukin- lb (IL-lb), Interferon g (IFN-g) and Basic Fibroblast Growth Factor (bFGF) compared to a control value is indicative of a lower frequency of fiber synthesis.

[0068] Human collagen type I alpha 1 chain (COL1A1) gene (Gene ID: 1277, updated on 4 december 2023) and human collagen type I alpha 2 chain (COL1A2) gene (Gene ID: 1278, updated on 23 November 2023) encode the pro-alphal (UniprotKB: B6VJR6, updated on 14 december 2022, transcript: GenBank: FM165586.1, last updated on 26 Jul 2016) and 2 (UniprotKB: P08123, updated on 08 November 2023, transcript: NCBI Reference Sequence: NM 000089.4, last updated on 22 Oct 2023) chains of type I collagen. Type I is a fibril-forming collagen found in most connective tissues and is abundant in bone, cornea, dermis and tendon.

[0069] Human Transforming Growth Factor beta 1 (TGFbeta 1) gene (Gene ID: 7040, updated on 5 December 2023) encodes a secreted ligand (UniprotKB: P01137, updated on 08 November 2023, transcript: NCBI Reference Sequence: NM_000660.7, last updated on 04 Dec 2023) of the TGF-beta (transforming growth factor-beta) superfamily of proteins. Ligands of this family bind various TGF-beta receptors leading to recruitment and activation of SMAD family transcription factors that regulate gene expression. The encoded preproprotein is proteolytically processed to generate a latency-associated peptide (LAP) and a mature peptide, and is found in either a latent form composed of a mature peptide homodimer, a LAP homodimer, and a latent TGF-beta binding protein, or in an active form consisting solely of the mature peptide homodimer. Human Endothelin-1 (ET-1 or EDN1) gene (Gene ID: 1906, updated on 23 November 2023) encodes a preproprotein that is proteolytically processed to generate a secreted peptide (UniprotKB: P05305, updated on 08 November 2023, transcripts: NCBI Reference Sequence: NM 001168319.2 and NM 001955.5, last updated 02 Oct 2023) that belongs to the endothelin / sarafotoxin family. Alternative splicing results in multiple transcript variants.

[0070] Human Platelet-Derived Growth Factor B (PDGF-B) gene (Gene ID: 5155, updated on 27-Nov-2023) encodes platelet-derived growth factor subunit B (UniprotKB: P01127, updated on 08 November 2023, transcripts: NCBI Reference Sequence: NM_002608.4 and NM_033016.3, last updated on 27 Nov 2023), which can homodimerize, or alternatively, heterodimerize with the related platelet-derived growth factor subunit A. These proteins bind and activate PDGF receptor tyrosine kinases, which play a role in a wide range of developmental processes.

[0071] Human Interleukin-1 beta (IL-lb) gene (Gene ID: 3553, updated on December 12, 2023) encodes a member of the interleukin 1 cytokine family. This cytokine (UniprotKB: P01584, updated on 8 November 2023, transcript: NCBI Reference Sequence: NM_000576.3, last updated on 04 dec 2023) is produced by activated macrophages as a proprotein, which is proteolytically processed to its active form by caspase 1 (CASP1 / ICE). This cytokine is an important mediator of the inflammatory response, and is involved in a variety of cellular activities, including cell proliferation, differentiation, and apoptosis. The induction of cyclooxygenase-2 (PTGS2 / COX2) by this cytokine in the central nervous system (CNS) is found to contribute to inflammatory pain hypersensitivity. Similarly, IL- IB has been implicated in human osteoarthritis pathogenesis.

[0072] Human Interferon g (IFN-g) gene (Gene ID: 3458, updated on 4 December 2023) encodes a soluble cytokine that is a member of the type II interferon class. The encoded protein (UniprotKB: P01579, updated on 8 November 2023, transcript: NCBI Reference Sequence: NM_000619.3, last updated on 04 Dec 2023) is secreted by cells of both the innate and adaptive immune systems. The active protein is a homodimer that binds to the interferon gamma receptor which triggers a cellular response to viral and microbial infections.

[0073] Human Basic Fibroblast Growth Factor (bFGF) gene, also known as FGF2 gene (Gene ID: 2247, updated on 23 November 2023), encodes a member of the fibroblast growth factor (FGF) family. FGF family members bind heparin and possess broad mitogenic and angiogenic activities. This protein (Uniprot KB: P09038, updated on 08 November 2023, transcript: NCBI Reference Sequence: NM 002006.6, last updated on 22 Nov 2023) has been implicated in diverse biological processes, such as limb and nervous system development, wound healing, and tumor growth. The mRNA for this gene contains multiple polyadenylation sites, and is alternatively translated from non- AUG (CUG) and AUG initiation codons, resulting in five different isoforms with distinct properties. The term “gene expression level” or “gene signature” refers to a pattern of expression of at least one gene of the set of genes as described above in a subject sample that is specific to the frequency of extracellular matrix fiber cross-linking, frequency of extracellular matrix fiber delinking or frequency of fiber synthesis and is indicative to tissue repair outcome of injured tissue of a subject.

[0074] The term " determining the gene expression level of at least one gene” of the set of genes as described above means that the gene expression level of at least one gene of the set of genes is assessed in a subject sample.

[0075] The gene expression level of said genes may be determined by any suitable methods known by the person skilled in the art in a patient sample.

[0076] Usually, these methods comprise measuring the quantity of mRNA or protein in a patient sample. Methods for determining the quantity of mRNA are well known in the art. For example, the mRNA contained in the sample is first extracted according to standard methods, for example using lytic enzymes or chemical solutions or extracted by nucleic-acid-binding resins following the manufacturer's instructions. The extracted mRNA is then detected by hybridization (e.g., Northern blot analysis) and / or amplification (e.g., RT-PCR) by using primer pairs and probes specific to said genes as described in the examples of the present disclosure. Quantitative or semi-quantitative RT-PCR is preferred. In another particular embodiment, the mRNA expression level is measured by RNA seq method.

[0077] In some embodiments, the expression level of said genes measured for example by quantitative RT-PCR are normalized by subtracting from each gene, the expression levels of housekeeping genes determined in the same experiment and the gene expression level may correspond to the normalized gene expression level of one gene or to the sum of normalized gene expression level of the set of genes as described above. In some embodiment, gene expression level is determined by measuring the quantity of protein. The quantity of the protein may be measured, for example, by semi-quantitative Western blots, enzyme- labelled and mediated immunoassays, such as ELISAs, biotin / avidin type assays, radioimmunoassay, immunoelectrophoresis, mass spectrometry or immunoprecipitation or by protein or antibody arrays.

[0078] As used herein, the term “control value” may refer to the gene as described above in biological sample obtained from a general population or from a selected population of subjects. For example, the general population may comprise apparently healthy subjects, such as individuals who have not lesion in corresponding tissue. The term “healthy subjects” as used herein refers to a population of subjects who do not suffer from any known condition, and in particular, who are not affected with any lesions in corresponding tissue.

[0079] In another particular embodiment, said control value may be a “threshold value” or “cut-off value” determined experimentally, empirically, or theoretically. The threshold value may be established based upon comparative measurements between patients having a tissue with and without lesion. Typically, the optimal sensitivity and specificity (and so the threshold value) can be determined using a Receiver Operating Characteristic (ROC) curve based on experimental data. For example, after determining the gene expression level in a group of reference, one can use algorithmic analysis for the statistic treatment of the measured values in samples to be tested, and thus obtain a classification standard having significance for sample classification. In a particular embodiment, Receiver operating characteristic (ROC) analysis was performed to calculate the gene expression level cut-off value of each gene using tissue samples with lesion and / or without lesion. The gene expression level values offering the highest sensitivity and specificity were selected as cut-off points. This algorithmic method is preferably done with a computer. Existing software or systems in the art may be used for the drawing of the ROC curve, such as: MedCalc 9.2.0.1 medical statistical software, SPSS 9.0, ROCPOWER.SAS, DESIGNROC.FOR, MULTIREADER POWER. SAS, CREATE-ROC.SAS, GB STAT VIO.O (Dynamic Microsystems, Inc. Silver Spring, Md., USA), R software (CRAN project) etc.

[0080] According to the present disclosure, the threshold value can be determined for at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 or 12 genes or for each gene selected from the group consisting of: LOX-1, LOXL-1, LOXL- 2, LOXL-3, LOXL-4, TGM1, TGM2, TGM3, TGM4, TGM5, TGM6, and TGM7.

[0081] According to the present disclosure, the threshold value can be determined for at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22 or 23 genes or for each gene selected from the group consisting of: MMP1; MMP8; MMP13, MMP2; MMP9; MMP3; MMP10; MMP7; MMP26; MMP14; MMP16; macrophage elastase, neutrophil elastase, trypsin-2, cathepsin K, cathepsin B; cathepsin L, cathepsin S, Meprin A subunit alpha, Meprin A subunit beta, RECK, TIMP-1, TIMP-2, TIMP-3, and TIMP-4.

[0082] According to the present disclosure, the threshold value can be determined for at least 1, 2, 3, 4, 5, 6, 7, or 8 genes or for each gene selected from the group consisting of: COL1A1 ; COL1A2, Transforming Growth Factor beta (TGF-P), Endothelin-1 (ET-1), Platelet-Derived Growth Factor B (PDGF-BB), Interleukin- lb (IL-lb), Interferon g (IFN-g) and Basic Fibroblast Growth Factor (bFGF).

[0083] According to the present disclosure, the gene expression level of said biomarkers can be determined by measuring the relative amount of product genes expressed in injured tissue sample, and preferably by performing a single sample GSEA (ssGSEA) in GSVA algorithm, by calculating a gene set enrichment score per sample as the normalized difference in empirical cumulative distribution functions of gene expression ranks inside and outside the gene set.

[0084] Prediction Model

[0085] Once the gene signatures as described above is determined and the corresponding set of parameters values is determined, the repair outcome of the injured tissue may be determined by applying a trained machine-learning model on said set of parameters. With reference to figure 7, the method is in that case implemented by a device 1 comprising a computer 10, which may include one or more processors, such as for instance a Computer Processing Unit or a Graphical Processing Unit, and at least one memory 11, having stored thereon code instructions which are executed by the computer 10 for implementing the method. The memory may also store the trained model configured for predicting tissue repair outcome. According to embodiments, the trained machine-learning model is configured to classify a tissue repair outcome, from said set of parameters values, among a plurality of pre-established classes comprising at least a regenerative healing class and a scarring healing class. In embodiments, the plurality of pre- established classes may comprise at least one additional class corresponding to unstructured tissue repair and which corresponds to an intermediate state between scarring healing and regenerative healing.

[0086] The model may be trained by supervised learning on a training dataset formed by data pairs, where each pair comprises a set of parameter values and a corresponding tissue repair outcome, belonging to one of the pre-established classes. In embodiments, the training dataset may comprise a plurality of data pairs for at least one set of parameter values.

[0087] The trained machine-learning model may be a classification model. According to non-limiting examples, the trained machine-learning model may be of the type Random Forest or Extreme Gradient Boosting.

[0088] In embodiments, the training of the model comprises a step of constituting the training dataset, i.e. obtaining the data pairs of the training dataset. Said data pairs can be obtained by performing numerical simulation of tissue repair and determining, from said numerical simulation, the corresponding tissue repair class.

[0089] With reference to figure la, the numerical simulation is performed by application of an agent-based model representing in a simplified manner, in a 2D bounded domain, the tissue in which a lesion is formed, wherein the tissue is represented a plurality of agent types representing respective components of the tissue to be modelled. Rules and interactions between the agents types are also defined to model the growth of the tissue.

[0090] In embodiments, the agent-based model comprises two types of agents: segments of fixed length, modeling ECM fibers, and growing circles, representing differentiating cells, represented by their centers and radii.

[0091] The growing circles representing differentiating cells are inseminated randomly in the modelling domain, with an initial radius Rmin. The cells grow regularly by a radius increase until reaching a radius Rmax. According to a non- limiting example, when the agent-based model represents adipose tissue, the growing circles represent adipocytes and the growth of these cells can reach up to 100 pm. Further, the model is based on a defined set of interactions between the two types of agents. In embodiments, at least the following interactions are defined: cells are modelled as incompressible and non-overlapping, Fiber-fiber and fiber-cell interactions are modelled via soft repulsion potentials, allowing some interpenetration of the agents,

[0092] Intersecting fibers are able to link or unlink following random (Poisson) processes of frequencies Pf and vad, respectively. Pf corresponds in the agent-based model to an ability of the ECM components to bind, while vad corresponds in the agent-based model to a capacity of ECM components to be degraded.

[0093] The numerical simulation comprises a first step of morphogenesis, started from a random distribution of fibers linked with Pf probability. New cells are randomly inseminated into the domain with a minimal radius and allowed to grow until the maximal radius Rmax is reached. The cell insemination process stops when the maximal number of cells is reached.

[0094] Tissue injury is then modelled by removing at least some components located in a region of the model domain. In embodiments, the lesion may be modelled by removing all components in a region. In other embodiments, the region devoid of components may also comprise an inner region where some components remain.

[0095] The removal of components creates a gradient at the border of the wounds that activates the production of an ‘injury signalling chemical’. This chemical is produced at the front of the tissue (borders of the wound), allowed to diffuse in the tissue, and has a finite lifetime. This signal inhibits the insemination of new cells and locally activated the insemination of new fibers modelled as a Poisson process of frequency vf. Upon insemination, new fibers are automatically linked with a proportion (Pf) of their intersecting neighbours, and unlinked with frequency vad.

[0096] With reference to figure 2a, the set of parameters forming the input of a numerical simulation, comprises at least: a cross linking probability Pf, representative of an ability of the ECM components to bind, an unlinking probability vad, representative of a capacity of the ECM components to be degraded, and a probability of synthesis of new fibers vf, representative of a level of synthesis of ECM components.

[0097] Thus a plurality of numerical simulations are run using the agent-based model based on varying values of the above parameters. In embodiments, a plurality of numerical simulations are run for at least one set of parameters values, to obtain a plurality of outcome for each set of parameters values and hence a plurality of data pairs with the same input values. In embodiments, the set of parameter values takes into account a time from the appearance of the lesion. The time from appearance of the lesion may be an additional parameter of the set. Alternatively, the values of the parameters mentioned above may be related to time and vary with time. For instance, a simulation may be performed with an initial value of Pf for a certain duration from the appearance of the lesion, and then with a different value of Pf.

[0098] In embodiments, the set of parameters values further comprise at least one parameter describing intensity of the lesion, which may include for instance a parameter indicative of a size of the lesion and / or a parameter indicative of a level of degradation of the tissue within the lesion. A parameter indicative of the size of the lesion may be expressed for instance as a ratio of cumulative length of the edges of the injury on the surface area of the injury, or for a lesion with two open edges, a distance between the two edges. A parameter indicative of a level of degradation of the tissue may for instance be a binary parameter representing whether all components have been removed in the region corresponding to the lesion, or not.

[0099] In order to determine the repair class corresponding to a set of parameters values, the method comprises computing a repair index regarding the structure obtained at the end of the numerical simulation and which forms the outcome of said simulation. The repair class can then be determined from the value of the repair index. The repair index quantifies a difference between the structure of the tissue before and after tissue repair. The higher the repair index, the more the repaired tissue differs from the initial tissue. In embodiments, the repair index is computed over the whole simulation domain. Alternative, the repair index may be computed only on the portion of the domain corresponding to the lesion. This enables considering only the lesion zone for assessing tissue repair and hence better quantifying the tissue repair.

[0100] With reference to figure lb, are shown different outcomes of tissue repair and corresponding value of repair index, starting from a same initial tissue but with varying parameters values for the numerical simulation.

[0101] With reference to figure 2b and 2c, the repair index may be computed from the following parameters, which are computed on the modelled domain at the end of the numerical simulation:

[0102] An average alignment of bundles of fibers,

[0103] A number of cells,

[0104] A number of clusters of cells,

[0105] An elongation rate of the clusters of cells.

[0106] According to a non-limiting example, the repair index may be computed as: Where:

[0107] AE quantifies a variation in elongation rate of the clusters of cells before and after tissue repair, Eoquantifies an elongation rate of the clusters of cells before tissue repair, NCquantifies a variation in the number of clusters of cells before and after tissue repair, Ncoquantifies a number of clusters of cells before tissue repair,

[0108] Ad quantifies a variation in average alignment of bundles of fibers before and after tissue repair, Aoquantifies an average alignment of bundles of fibers before tissue repair, NEquantifies a variation in the number of cells before and after tissue repair, NEOquantifies a number of cells before tissue repair.

[0109] Based on the plurality of simulations, the classes regarding tissue repair may be defined from the values of the repair index. According to a non-limiting example, and considering the formula for the repair index given above, a first class corresponding to regenerative repair may be defined, corresponding to a repair index comprised between 0 and 0.7, a second class corresponding to unstructured tissue repair may be defined, corresponding to a repair index comprised between 0.7 and 2, and a third class corresponding to scar repair may be defined corresponding to a repair index comprised between 2 and 4.5.

[0110] As indicated earlier, a plurality of numerical simulations are so performed, a repair index is computed on the outcome of each simulation and a corresponding tissue repair class is associated to each simulation outcome, thereby enabling the obtention of data pairs. Preferably, at least 1000 data pairs are so obtained. The prediction model is then trained by supervised learning on the obtained data pairs, in order to be able to predict a repair class from the input set of parameters values.

[0111] Once trained the modelled may be used to predict tissue repair outcome for a patient, by providing values of the set of parameters comprising at least: a level of synthesis of extracellular matrix (ECM) components, corresponding to a frequency of fiber synthesis as disclosed above, a capacity of the ECM components to be degraded, corresponding to a frequency of ECM fiber delinking as disclosed above, an ability of the ECM components to bind, corresponding to a frequency of ECM fiber cross-linking as disclosed above.

[0112] When the model has been trained on data pairs comprising at least one additional parameter, such as a time elapsed since the appearance of the lesion and / or an indicator of an intensity of the lesion, said parameters may also be determined for the lesion of interest of the patient and the model may be interrogated with these additional parameters.

[0113] Therapeutic use In connection with above methods of predicting tissue repair outcome, the present disclosure relates to the treatment of the injured tissue in a subject in need thereof.

[0114] In a preferred embodiment the present disclosure relates to the treatment of the injured tissue in a subject in need thereof wherein said injured tissue is previously predicted to have a scarring healing outcome by the method for predicting tissue repair outcome as described above.

[0115] Indeed, subject having an injured tissue predicted to be repaired by scarring healing may benefit from certain treatments that allow to change the tissue repair outcome and induce regenerative healing of the tissue.

[0116] Indeed, the inventors have shown that the administration of extra-cellular matrix cross-linking inhibitors such as LOX inhibitors can enhance regenerative healing, in particular when administered at early stage (e.g., 6 days) after the injury.

[0117] Therefore, the present disclosure relates to an extra-cellular matrix cross-linking inhibitors such as a Lysyl oxidase (LOX) inhibitor for use in the regenerative healing of an injured tissue in a subject in need thereof, preferably previously predicted to have an injured tissue with a scarring repair outcome by a method as described above.

[0118] In other terms, the present disclosure relates to a method for regenerative healing of an injured tissue in a subject in need thereof comprising: administering a therapeutically effective amount of an extracellular matrix cross-linking inhibitors such as a LOX inhibitor to said subject.

[0119] In a preferred embodiment, the present disclosure relates to a method for regenerative healing of an injured tissue in a subject in need thereof comprising: predicting tissue repair outcome of the injured tissue by the method as described above, administering a therapeutically effective amount of an extra-cellular matrix cross-linking inhibitors, such as a LOX inhibitor to said subject in need thereof, previously predicted to have an injured tissue with a scaring repair outcome.

[0120] The present disclosure also relates to the use of an extra-cellular matrix cross-linking inhibitors such as a LOX inhibitor in the manufacturing of a medicament for use in the regenerative healing of an injured tissue in a subject in need, preferably previously predicted to have an injured tissue with a scarring repair outcome by a method as described above.

[0121] As used herein, a "therapeutically effective amount" or an "effective amount" means the amount of a composition that, when administered to a subject for treating a state, disorder or condition is sufficient to affect a treatment. The therapeutically effective amount will vary depending on the compound, formulation or composition, the disease and its severity and the age, weight, physical condition and responsiveness of the subject to be treated. As used herein, the term “treatment”, “treat” or “treating” refers to any act intended to ameliorate the health status of patients such as therapy, prevention, prophylaxis and retardation of the disease. In certain embodiments, such term refers to the amelioration or eradication of a disease or symptoms associated with a disease. In other embodiments, this term refers to minimizing the spread or worsening of the disease resulting from the administration of one or more therapeutic agents to a subject with such a disease. According to the present disclosure, the term treatment refers to the reparation of the tissue lesion by regeneration of the tissue instead of scaring.

[0122] According to the present disclosure, extra-cellular matrix cross-linking inhibitors can be as non-limiting examples, inhibitor of lysyl oxidase (LOX).

[0123] By “LOX inhibitor” is meant any agent able to decrease at least one member of lysyl oxidase family comprising in mammals five family homologs: lysyl oxidase (LOX) and lysyl oxidase-like 1 to 4 (LOXL1-LOXL4) and specifically at least one member of lysyl oxidase family expression and / or biological activity, in particular enzymatic activity using for example a fluorescence-based assay as disclosed in Holt et al. 2006, Nature Protocols, 1, 2498-2505, in particular that results in an decrease of extra-cellular matrix, in particular collagen cross-linking. Collagen cross-linking can be evaluated in vitro with 3D organitypic co-culture model as disclosed in Chitty et al. 2023, Nature Cancer 4, 1326— 1344 (2023).

[0124] Some examples of LOX inhibitors are: P-aminopropionitrile (BAPN), -aminopropionitrile fumarate (BAF), P-aminopropionitrile maleate (BAM), P-aminopropionitrile monohydrate (BAMH), aminomethylenethiophene (AMT), 5-(naphthalen-2-ylsulfonyl)thiophen-2-yl)methamine 1, 2- aminomethylene-5-sulponylthiazole (AMTz), 5-(naphthalen-2-ylsulfonyl)thiazol-2-yl)methanamine 2, PXS-5505 (CAS No.: 2409963-83-1), PXS-5382 and Simtuzumab.

[0125] According to the present disclosure, the tissue can be any tissue of a subject, preferably said injured tissue is a lobular tissue such as pulmonary, hepatic, breast, adipose tissue, more preferably adipose tissue.

[0126] In the present application, the inventors showed that a lower frequency of ECM cross-linking at early stage, preferably no more than 6 days after injury induces regenerative healing in the injured tissue.

[0127] Therefore, in a preferred embodiment, said ECM cross-linking inhibitor such as LOX inhibitor is administered no more than 10, 9, 8, 7, preferably 6 days after the injury.

[0128] According to a preferred embodiment of the present disclosure, the injury is not a small injury. In a particular embodiment, if said lesion is a round or square lesion, the ratio of cumulative length of the edges of the injury on the surface area of the injury is preferably higher than 2.5 mm or for a lesion with two open edges, the distance between the 2 edges is preferably higher than 5 mm. The LOX inhibitor described herein may be administered by any means known to those skilled in the art, including, without limitation, intravenously, orally, intra-tumoral, intra-lesional, intradermal, topical, intraperitoneal, intramuscular, parenteral, subcutaneous and topical administration. Thus, the compositions may be formulated as an injectable, topical, ingestible, or suppository formulation. Administration of the immunosuppressive agent to a subject in accordance with the present invention may exhibit beneficial effects in a dose-dependent manner. Thus, within broad limits, administration of larger quantities of the compositions is expected to achieve increased beneficial biological effects than administration of a smaller amount. Moreover, efficacy is also contemplated at dosages below the level at which toxicity is seen.

[0129] It will be appreciated that the specific dosage of LOX inhibitor administered in any given case will be adjusted in accordance with the composition or compositions being administered, the volume of the composition that can be effectively delivered to the site of administration, the disease to be treated or inhibited, the condition of the subject, and other relevant medical factors that may modify the activity of the compositions or the response of the subject, as is well known by those skilled in the art.

[0130] For example, the specific dose of LOX inhibitor for a particular subject depends on age, body weight, general state of health, diet, the timing and mode of administration, the rate of excretion, medicaments used in combination and the severity of the particular disorder to which the therapy is applied. Dosages for a given patient can be determined using conventional considerations, e.g., by customary comparison of the differential activities of the compositions described herein and of a known agent, such as by means of an appropriate conventional pharmacological protocol. The compositions can be given in a single dose schedule, or in a multiple dose schedule.

[0131] Suitable dosage ranges for a LOX inhibitor may be of the order of several hundred micrograms of the agent with a range from about 0.001 to 10 mg / kg / day, preferably in the range from about 0.01 to 1 mg / kg / day.

[0132] The invention will now be exemplified with the following examples, which are not limitative.

[0133] EXAMPLES

[0134] 1. METHODS

[0135] Animals

[0136] All experiments in mice were performed on 7-week-old male mice. C57BL / 6 OlaHSD mice were obtained from Envigo Laboratories and bred in a conventional facility at the Faculty of Pharmacy in Toulouse. Animals were group-housed (3 to 5 per cage) in a controlled environment (12-hour light / dark cycles at 21 °C) with unrestricted access to water and a standard chow diet. The animals were maintained in accordance with the ARRIVE guidelines of the European Community Council. Mice were killed by cervical dislocation. All experiments were carried out in compliance with European Community Guidelines (2010 / 63 / UE) and approved by the MESRI (Ministry of Higher Education, Research andinnovation) and the French ethics committee.

[0137] Subcutaneous AT resection

[0138] Control mice used for the baseline control were not subjected to surgery. For mice that underwent unilateral resection of subcutaneous AT, animals were anaesthetized by inhalation of isoflurane 2.5%. A single abdominal incision was then made to access and excise 35% of the right AT between the lymph node and groin. The left AT did not undergo a surgical procedure and was thus used as an internal control. For these two groups (control and resected), the skin was closed with 3 suture points.

[0139] In vivo treatments

[0140] Mice were treated with 50 NI of naloxone methiodide (NaL-M) (daily subcutaneous injections, 17 mg / kg, N129, Sigma Aldrich) or NaCI (daily subcutaneous injections) from day 0 to 3 after AT resection. For crosslink regulation, mice were treated with 50 NI of lysyl oxidase inhibitor, p-aminopropionitrile (daily subcutaneous injections, 150 mg / kg, A3134, Sigma Aldrich) from day 0 to 6 after AT resection. Genipin treatment was administrated daily as a co-injection with NaL-M during the first 3 days after AT resection and Genipin only (daily subcutaneous injection, 5 mg / kg, G4796, Sigma Aldrich) from day 4 to 6 after AT resection.

[0141] Regeneration evaluation by light transmittance analyses

[0142] One-month post-resection, mouse AT were fixed (PFA 4%, 24h) and placed on a coverglass. The samples were observed through a binocular microscope and pictures were taken using a 48MP camera (Nikon, D5000). All images were opened in Imaged software and the contrast was normalized using the histogram equalization function. Regions of interest (ROI) were drawn to include the injury area and the mean gray value in this ROI were calculated. Higher mean gray values mean that the tissues were not reconstructed since the light can go through the injury area and lower values were associated with reconstructed tissue where the injury area is darker thanks to the presence of new adipocytes.

[0143] Fluorescence and Second Harmonic Generation (SHG) imaging

[0144] Mice AT were fixed (PFA 4%, 24h), embedded in agarose gel (Sigma A0169), and cut into 300 pm tissue sections with a Vibroslice for Campden Instruments. Tissue sections were incubated for 1.5 hours in PBS / 0.2% triton at room temperature and then 2 hours in Bodipy 493 (Invitrogen 03922) in the dark. After several washes, tissue sections were incubated for 2 hours in Draq5 (Thermo fisher, 62251).

[0145] Imaging was performed using a Biphotonic Laser Scanning Microscope (LSM880 Carl Zeiss, Jena, Germany) with an objective lens LCI 'Pan Apochromat110X / 0.45 or ID C-Apochromaf 40x / l.l. Bodipy and Draq5 were excited using 488 and 633 nm lasers, and SHG was excited using the 800 nm biphoton laser.

[0146] Fractal dimension (FD) analyses

[0147] For FD analysis, SHG tiles were individually opened in ImageJ and background subtractions were performed before binarizing all images from the Z-stack. FD were then estimated using the box-counting method in ImageJ. The software considered box-counting in two dimensions, allowing quantification of pixel distribution in this space. The FD was based on a series of grids with different sizes (boxes) over an image and the recorded data (counting) for each successive box size. The results were expressed as the FD of the object that is DF = Log N / Log r; N, where N is the number of boxes that cover the pattern, and r is the magnification, or the inverse of the box size. FD were thereby calculated using the ImageJ software set between 0 and 2, with 0 corresponding to an image without signal (0 pixels) and 2 to an image full of signal.

[0148] Picrosirius Red (PSR) Staining

[0149] AT were collected from uninjured mice 3 and 10 days after injury and fixed (PFA 4%, 24h). AT were dehydrated using the following successive alcohol treatment protocol: 10 minutes in 70% EtOH, 15 minutes in 95% EtOH, 35 minutes in 100% EtOH and 35 minutes in Bioclear solution. Then AT samples were embedded in paraffin for 24 hours. 5 pm sections were cut, deparaffined by successive baths in Bioclear and 100% EtOH, and then rinsed in distilled water. Tissue sections were incubated in Fast Green / Citrate buffer solution (0.04%) for 15 minutes, rinsed in distilled water, incubated in Fast Green / Picrosirius Red (0.1%) during 15 minutes, washed in distilled water and dehydrated in 2 minutes EtOH and 10 minutes Bioclear solution and mounted with Eukitt Mounting Medium. Tissue sections were imaged using Lyon Platform Imaging (CIQLE) with a Zeiss AxioScan 7 microscope.

[0150] Machine learning analyses

[0151] The dataset was composed of 2,218 simulations from 405 combinations of parameters. The dataset was split into train (80%) and test (20%) groups, and two ML methods (RandomForest Classifier and XGBoost) were used to predict the final result of simulations (regenerative, unstructured, and scar repair). To analyse the explainability of the prediction results, a SHAP model was applied to calculate the influences and interactions of each parameter on the output of the predictive model.

[0152] Statistical analyses

[0153] The number of animals used in each study is indicated directly on the figure or in the figure legends. Mice were randomized allocated to the different groups and investigators were blinded to analyses. All results are given as means SEM for the barplot or curve plot and median ± min / max for the boxplot. Statistical differences were measured using an ANOVA test when there were more than two groups and Tukey-Hsd post-hoc test was used to determine statitical differences between each group. All statistical analyses were carried out using RStuido software. p< 0.05 was considered as significant. The following symbols for statistical significance were used throughout the manuscript: *p< 0.05; **p < 0.01; ***p < 0.001, **** p < 0.0001.

[0154] Mathematical modelling

[0155] The 2D mathematical model features cells described as 2D spheres represented by their centers and radii and fibers described as segments of fixed length. The model was implemented on a 2D square domain with periodic boundary conditions. Both agents (cells and fibers) interacted via mechanical repulsive interactions. Cells were modeled as incompressible, i.e., non-overlapping, while fiber-fiber and fiber-cell interactions were modeled via soft repulsion potentials, allowing some interpenetration of the agents. Intersecting fibers were able to link or unlink following random (Poisson) processes of frequencies vlink and vunlink, respectively. In contrast, linked fiber pairs were constrained to maintain the position of their cross-link during motion. Linked fibers were subjected to an alignment force at their junctions. The model was divided into two steps (morphogenesis and reconstruction). The morphogenesis step was started from a random distribution of fibers linked with Pf probability. New cells were randomly inseminated into the domain with a minimal radius and allowed to grow with linear volumic growth until a maximal radius was reached. The cell insemination process was stopped when the maximal number of cells was reached. As shown in Berthezene, C. D. et al. Int. J. MoL Sci. 22, 7336 (2021), this model was able to spontaneously generate tissue architectures at equilibrium consisting of lobular cell structures in organized fiber networks which served as a basis for the reconstruction step.

[0156] The reconstruction steps were studied starting from morphogenesis simulations, described above, as a starting point. Tissue injury was modelled by removing all the components (cells and fibers) located in a region of the simulation square domain. This created a gradient at the border of the wounds that activates the production of an 'injury signalling chemical’. This chemical was produced at the front of the tissue (borders of the wound), allowed to diffuse in the tissue, and had a finite lifetime. This signal inhibited the insemination of new cells and locally activated the insemination of new fibers modelled as a Poisson process of frequency vf. Upon insemination, new fibers were automatically linked with a proportion (Pf) of their intersecting neighbours, and unlinked with frequency vad. The wound was then filled with new fibers and the injury signalling chemicals disappeared, enabling insemination of new cells modelled as a Poisson process. The frequency of new cell insemination depended on: (a) The density of injury-signalling molecules: new cells were more likely to appear where the density of injury-signalling chemical is low, (b) the number of fiber links in the ECM: new cells were more likely to appear in regions with low ECM fiber links, and (c) the density of existing agents: new cells were more likely to appear where existing fibers were already present. As in the morphogenesis step, new cells grew according to a linear volumic growth up to a preset maximal radius.

[0157] 2. RESULTS

[0158] Fiber cross-linking plays a major role in tissue repair outcome prediction. To set-up a predictive model of AT repair (regenerative versus scar repair), the inventors upgraded a previous conceptual ABM of tissue injury. Briefly, in this conceptual model, a tissue-lesion was modelled by removal of segments (ECM fibers) and growing circles (adipose cells) (Fig. la). The subsequent repair process was obtained by seeding new fibers and the growth of new cells. Cells and fibers interact through mechanical repulsive forces, but when two fibers intersect, they can also form cross-links with a defined probability (Pf), and can thus resist the pressure of growing cells. The inventors mimicked ECM remodelling by setting up fiber synthesis (vf), cross-linking (with a probability, PO) and unlinking (vad), thus controlling both the density and cross-linking of fibers in the network (Fig. la right panel and Fig. 2a). Cross-linked fibers were also subjected to an alignment mechanism at their junction to model the ability of long fibers (those made up of several cross-linked fiber units) to resist bending. Thanks to these properties, digital fibers generate a network from which both cell clustering and tissue architecture can emerge. To evaluate tissue structure at the end of simulations, a repair index was calculated that captures the numbers and spatial organization of cells and fibers in the reconstructed tissue compared to the initial non-injured tissue (Fig. 2 b, c). As described in a previous conceptual study, the model gave rise to three distinct classes of repair outcomes. Regenerative repair with a repair index from 0 to 0.7 was characterized by a reconstructed tissue similar to the non-injured tissue in terms of number of adipocytes and lobular organization. In contrast, unstructured repair (repair index from 0.7 to 2) did not reproduce lobular organization. Scar repair (repair index from 2 to 4.5) was characterized by a high number of fibers and a few adipocytes (Fig. lb).

[0159] To go further, a sensitivity analysis of the three parameters vf, Pf and vad that control dynamic organization of the fiber network has been performed using the conceptual model (Fig. 3a). Each ECM parameter was independently changed over a wide range of values, generating a dataset of 405 parameter combinations. Each combination of the parameters vf (1 to 9), Pf (0.1 to 0.9), and vad (0.001 to 0.1) was simulated six times. These 2,218 simulations provided repair index values ranging from 0 to 4.5 (243 of them being shown in 3D space, Fig. 3a). The sensitivity study, represented on the 3D scatter plot, revealed that with a low fiber unlinking frequency (vad), the combination of specific Pf and vf values drove tissue repair towards the three repair outcomes. In contrast, a high frequency of fiber unlinking led mainly to regenerative repair (Fig. 3a).

[0160] Because the three ECM-related parameters interact together to ensure fiber density and organization, the specific contribution of each parameter is hard to define. To tackle this complex issue, the inventors applied an explainable ML strategy that integrated all types of interactions considering one parameter at a time. They randomly split simulations into training (80%) and test (20%) groups according to classes (regenerative, unstructured, and scar repair) (Fig. 3b). Both groups were balanced among the different classes (Fig 4a). The inventors trained and compared class predictions (regenerative, unstructured, scar) using two algorithms, Random Forest (RF) and Extreme Gradient Boosting (XG Boost), with 5-fold cross-validation. Both RF and XG Boost predicted the different repair classes with an accuracy of 77% (Fig. 4b). The classification results were summarized using confusion matrices (Fig. 4c, d). To define the contribution of each variable in repair outcomes predictions, Shapley Additive exPlanations (SHAP) was applied to RF. SHAP graphs represent explanations for regenerative (Fig. 3c) and scar (Fig. 3d) repair class predictions. Parameters (vf, Pf and vad) were ranked according to their importance in the prediction, the parameter at the top being the most important to predict the class. Similar results were obtained with XG Boost algorithm (Figure 4 e, f).

[0161] RF and XG Boost results showed that Pf, vad and vf values predict the outcome of tissue repair, respectively 43 and 37 %, 31 and 39 % and 25 and 24 % (Figure 4g). These analyses suggested that Pf and vad, both impacting the number of fiber cross-links, were the most important analysed factors in predicting tissue repair outcome.

[0162] Low values of Pf and vf and high values of vad had a positive impact on regenerative class prediction (Fig. 3c). More precisely, setting Pf from 0.1 to 0.2, vad from 0.1 to 0.05 and vf from 1 to 3 seems to systematically drive tissue repair toward regenerative repair (Fig. 3d). In contrast, high values of Pf (from 0.7 to 0.9) and vf (from 7 to 9) as well as low values of vad (from 0.001 to 0.01) had a positive impact on the scar repair class prediction (Fig. 3e, 1).

[0163] The inventors next evaluated the temporal changes in fiber cross-linking resulting from interactions between Pf and vad. They recorded the number of cross-links over time. This demonstrated that the three repair outcomes followed the same inverted exponential curve. They found that regenerative repair curve was associated with low number of cross-link values. The present analysis indicated that the outcome of tissue repair could be driven by modulating the number of fiber cross-links immediately following lesion and in a short temporal window since these values plateaued in the earliest time points of simulations (Epochs) (Fig. 4h).

[0164] Early and transient modulation of fiber cross-linking is sufficient to drive digital tissue repair outcomes.

[0165] To determine how the kinetic of cross-link number impacts tissue repair, the inventors had to temporally calibrate the ABM, by using the present in vivo model of AT regenerations. In the ABM, the uninjured tissue is composed of lobular structures filled with adipocytes. In vivo, after an injury, the wound is filled with fibers within 3 days and adipocytes emerge from 10 days. The inventors focused on simulations for which the combination of parameters (vf, Pf and vad) allowed wound closure and adipose cell emergence in a kinetic that matched with in vivo observations. After this calibration, the number of fiber cross-links plateaued at 6 days post-injury (Fig. 5a), suggesting that day 0 to day 6 postinjury may constitute the temporal window during which fiber cross-linking can be modulated to drive tissue repair outcome. The inventors thus tested whether an early and transient change in Pf value is sufficient to reverse the final expected outcome of scar or regenerative simulations. To this end, they used the specific Pf values obtained with SHAP analysis (Fig. 3e, 1). Decreasing Pf from 0.7 to 0.1 during the first six days after injury switched tissue repair trajectories from scar to regenerative repair (Fig. 5b). In contrast, increasing Pf from 0.2 to 0.8 during the first 6 days after injury switched tissue repair trajectories from regenerative to scar repair (Fig. 5c). These results were confirmed by the quantification of the repair index (Fig. 5d). As expected, changes in the outcome of tissue repair were associated with a switch in the profiles of the curves representing number of fiber cross-links (Fig. 5e). Taken together, these in silico experiments performed on a calibrated digital tissue model show that early and transient modulation of fiber crosslinking could be sufficient to drive the outcome of tissue repair.

[0166] Temporally specific modulation of fiber cross-linking is sufficient to drive tissue repair outcomes in vivo.

[0167] The inventors then investigated whether the in silico model reflected the biology of tissue repair. According to the digital experiments, they tested whether an early and transient decrease or increase in fiber cross-linking led to tissue regeneration or scarring respectively using the previously validated in vivo model of AT repair. Indeed, as previously described, treatment with NaCI or Naloxone Methiodide (NalM), an antagonist of opioid receptors, after partial resection of subcutaneous AT leads one -month post-lesion to scar healing or regeneration respectively. During the 6 days following AT resection, NaCI (scarring) and NalM (regenerative) treated mice were treated with beta aminopropionitrile (BAPN) or with Genipin, drugs that inhibit and enhance fiber cross-linking, respectively.

[0168] One month after resection, AT repair was analysed by macroscopic evaluation, light transmittance and quantified fluorescence experiments. BAPN significantly induced AT macroscopic regeneration in scarring mice (Fig. 6a) and a decrease in light transmittance compared to vehicle treated mice, suggesting the presence of new cells in reconstructed area (Fig. 6b). In addition, fluorescence of 300 pm thick tissue sections revealed the presence of mature adipocytes organized as lobules in the repaired area of BAPN-treated mice (Fig. 6c). This was confirmed by a higher percentage of reconstruction in BAPN compared to vehicle treated mice (Fig. 6d). In contrast, Genipin inhibited macroscopic regeneration and induced

[0169] AT scarring in regenerative mice (Fig. 6a), associated with an increase in light transmittance, suggesting the absence of new cells in the reconstructed area (Fig. 6b). In addition, reconstructed area was characterized by the presence of a dense fiber network and the absence of adipocytes (Fig. 6c), as well as a lower reconstruction percentage (Fig. 6d).

[0170] To evaluate ECM organization during treatment windows, the inventors then calculated Fractal Dimension (FD) values on SHG images 3 days post-injury. FD describes the amount of space and selfsimilarity of structure and is highly sensitive to changes in collagen organization. FD values were higher in regenerative than scarring mice. Moreover, ECM cross-linking inhibition through BAPN treatment is sufficient to switch FD values of scarring mice close to regenerative ones (Fig. 6e). These results clearly confirm that regeneration and scarring are characterized by specific and distinct early ECM organization patterns.

[0171] Overall, these in vivo data validated both the digital model upgrade and its derived-hypotheses. In addition, it demonstrated that early and transient fiber cross-linking modulation drives tissue repair outcomes.

[0172] DISCUSSION

[0173] The present study provides biological validation of an innovative hypothesis derived from a calibrated digital model used as a digital twin. The inventors validated in vivo that regeneration in adult mammals can be induced by early and transient inhibition of ECM fiber cross-linking after injury. The ability to perform high throughput in silico experiments before implementing in vivo experiments for biological validation greatly reduced the time and number of animals needed to explore the experimental space.

[0174] In this study, the initial conceptual digital model was significantly upgraded to develop a predictive version. The systematic investigation and the calibration of the conceptual model led to an independent and temporal exploration of the parameters, focusing the questions on the role of mechanical forces during the early steps of repair in shaping tissues. The calibration was based on morphological observations of tissue repair process (wound closure and adipocyte emergence kinetics) in the previously described in vivo model of AT regeneration, since reliable methods to measure cross-linking and / or crosslinking activity biologically are currently lacking.

[0175] The present in silico results show that dynamic ECM cross-linking and unlinking rates determine the final repair outcome. ECM cross-linking is known to condition ECM topology and physical properties and thus facilitate ECM stabilization. Dynamic remodelling by changing the connectivity of the ECM network can confer either fluid or solid-like properties to tissues, as previously described in the development of embryos and adult tissues (Palmquist, K. H. et al. Cell 185, 1960-1973. el 1 (2022)). In all these studies, a weakly-cross-linked network behaves like a fluid and allows cell rearrangement within the ECM. This transition from a solid to fluid-like behavior also occurs at the tissue level during lung branching morphogenesis, where increased local fluidity enables branching to expand. In contrast, extensive collagen cross-linking is observed in tissue fibrosis which results in an increasingly stiff and less compliant matrix. The importance of ECM cross-linking in repair processes is also consistent with previous studies showing that uterus regeneration only occurs within an appropriate range of decellularized matrix cross-links number.

[0176] Although this connectivity appears to be fundamental to the repair outcome, the timing of the interaction is also critical. The present digital tissue repair model demonstrates that early, transient and tight restriction of ECM cross-link numbers within a given range is mandatory and sufficient to allow cells to self-organize into an optimal architecture and guide tissue repair towards regeneration.

[0177] The present results also indicate that tissue reconstruction after injury seems to result from a selforganization of cells from an initial morphological template, as it has recently been described in skin development. Indeed, tissue reconstruction in adult mammals seems to be a self-organizing process where an initial scaffold leads spontaneously to scarring. Altering this initial scaffold sets the stage for the self-organization of cells and recovery of the tissue architecture existing before injury.

[0178] Previously, the inventors demonstrated that regeneration in adult tissues is observed after transient and early treatment with an antagonist of endogenous opioids during the 3 days following injury. Similar results were also obtained on mouse pancreas confirming the conclusions. They also showed that inhibition of resident macrophages and management of the post-injury inflammatory phase could explain the regenerative effect of endogenous opioids inhibition. The present study suggests that similar regenerative effects can be obtained without directly managing inflammation and opioid signalling. In both cases, the studies strongly suggest that tissue regeneration is possible but inhibited in adult mammals. The present study opens new therapeutic approaches targeting ECM cross-linking while preserving pain management.

[0179] Finally, the present discovery relies on three elements: i) a biological model relevant to analysis of repair processes in adult mammals, ii) a calibrated digital model of the biological tissue, iii) dynamic round trips between digital and in vivo models. Taken together, these three elements make the present model close to the definition of a digital twin to reveal the complex biomechanical cues controlling tissue architecture emergence, impairment, and recovery after injury. This highlights the management of crosslinking during the first steps after injury as a relevant target for plastic and reconstructive surgery and regenerative medicine. In addition, the present work reveals the relevance of digital strategies to delineate fine biological processes with a minimal number of in vivo experiments.

Claims

CLAIMS1. A method of predicting tissue repair outcome for a subject having injured tissue, preferably lobular tissue, comprising: determining a set of parameter values relative to the subject, wherein the set of parameters are indicative of : o a level of synthesis of extra-cellular matrix (ECM) components, o an ability of the ECM components to bind, and o a capacity of the ECM components to be degraded, and predicting, based on the set of parameter values, an outcome of the tissue repair among at least the following outcomes: regenerative healing and scarring healing.

2. The method according to claim 1, wherein: parameter values indicative of an ability to bind of the ECM components comprise a gene expression level of at least one gene selected from the group consisting of : LOX-1, LOXL- 1, LOXL-2, LOXL-3, LOXL-4, TGM1, TGM2, TGM3, TGM4, TGM5, TGM6, and TGM7 in a subject sample,Parameter values indicative of a capacity of the ECM components to be degraded comprise a gene expression level of at least one gene selected from the group consisting of : MMP1; MMP8; MMP13, MMP2; MMP9; MMP3; MMP10; MMP7; MMP26; MMP14; MMP16, trypsin-2, macrophage elastase, neutrophil elastase, cathepsin K, cathepsin B; cathepsin L, cathepsin S, Meprin A subunit alpha, Meprin A subunit beta, RECK, TIMP-1, TIMP-2, TIMP-3, and TIMP-4 in a subject sample, andParameter values indicative of a level of synthesis of ECM components comprises a gene expression level of at least one gene selected from the group consisting of : COL1A1, COL1A2, Transforming Growth Factor beta (TGF-[3), Endothelin-1 (ET-1), Platelet- Derived Growth Factor B (PDGF-BB), Interleukin- lb (IL-lb), Interferon g (IFN-g) and Basic Fibroblast Growth Factor (bFGF) in a subject sample, preferably wherein subject sample is a subject injured tissue sample.

3. The method according to claim 2, wherein:A lower expression of at least one gene selected from the group consisting of LOX- 1, LOXL-1, LOXL-2, LOXL-3, LOXL-4, TGM1, TGM2, TGM3, TGM4, TGM5, TGM6, and TGM7 compared to a control value is indicative of a lower ability to bind of the ECM components,a lower expression of at least one gene selected from the group consisting of: MMP1; MMP8; MMP13, MMP2; MMP9; MMP3; MMP10; MMP7; MMP26; MMP14; MMP16; macrophage elastase, neutrophil elastase, trypsin-2, cathepsin K, cathepsin B; cathepsin L, cathepsin S, Meprin A subunit alpha, Meprin A subunit beta, and RECK compared to a control value and / or a higher expression of at least one gene selected from the group consisting of: TIMP-1, TIMP-2, TIMP-3, and TIMP-4 compared to a control value is indicative of a higher capacity of the ECM components to be degraded, and a lower expression of at least one gene selected from the group consisting of: COL1A1 ; COL1A2, Transforming Growth Factor beta (TGF-|3), Endothelin-1 (ET-1), Platelet- Derived Growth Factor B (PDGF-BB), Interleukin- lb (IL-lb), Interferon g (IFN-g) and Basic Fibroblast Growth Factor (bFGF) compared to a control value is indicative of a lower frequency of ECM components synthesis.

4. The method according to any one of claims 1 to 3 wherein said tissue repair outcome is predicted by application to the determined set of parameters valued of a trained model configured to classify the type of tissue repair outcome among a plurality of classes comprising at least a regenerative healing class and a scarring healing class.

5. The method according to any one of claims 1 to 4, wherein the model has been trained by supervised learning on a training dataset formed by data pairs where each pair comprises a set of parameter values, and a corresponding tissue repair class and the training dataset comprises a plurality of data pairs for at least one set of parameter values.

6. The method according to claim 5, wherein the training of the model further comprises a step of obtaining the data pairs of the training dataset, by performing, for each set of parameter values, numerical simulation of tissue repair, and determining, from said numerical simulation, the corresponding tissue repair class, wherein the numerical simulation is performed by an agent-based model, in which: a tissue comprising a lesion is simulated, wherein the tissue is represented by two types of agents, comprising segments modelling ECM fibers and growing circles representing differentiating cells, and a set of interactions are defined between the two types of agents, wherein a repair index is computed from the outcome of the numerical simulation, and the tissue repair class is determined from the value of the repair index.

7. The method according to claim 6, wherein the repair index from the following parameters:An average alignment of bundles of fibers,A number of cells,A number of clusters of cells,An elongation rate of the clusters of cells.

8. The method according to claim 6 or 7, wherein the repair index is computed as follows:Where:AE quantifies a variation in elongation rate of the clusters of cells before and after tissue repair, Eoquantifies an elongation rate of the clusters of cells before tissue repair,AACquantifies a variation in the number of clusters of cells before and after tissue repair, Ncoquantifies a number of clusters of cells before tissue repair,AA quantifies a variation in average alignment of bundles of fibers before and after tissue repair, Aoquantifies an average alignment of bundles of fibers before tissue repair,A.Vg quantifies a variation in the number of cells before and after tissue repair, NEOquantifies a number of cells before tissue repair.

9. The method according to any of claims 6 to 8, wherein the repair index is computed only on the area of the tissue corresponding to the lesion.

10. The method according to any of the preceding claims, wherein the set of parameters further comprises a value of time from the appearance of the lesion.

11. The method according to any of the preceding claims, wherein the set of parameters further comprises a parameter describing intensity of the lesion.

12. The method according to claim 11, wherein the set of parameters comprises at least a parameter indicative of a distance between the edges of the lesion and / or a parameter indicative of a level of degradation of the tissue within the lesion.

13. A computer-program product comprising code instructions for implementing the method according to any of the preceding steps, when it is executed by a processor.

14. A Lysyl oxidase (LOX) inhibitor for use in the regenerative healing of an injured tissue in a subject in need thereof, preferably wherein the lesion tissue is previously predicted to have a scarring repair outcome by a method according to any one of claims 1 to 11.

15. The LOX inhibitor for use according to claim 14 wherein said inhibitor is administered no more than 6 days after the injury.

16. The LOX inhibitor for use according to claim 14 or 15 for regenerative healing of a round or square lesion wherein the ratio of cumulative length of the edges of the injury on the surface area of the injury is higher than 2.5 mm or for a lesion with two open edges wherein the distance between the 2 edges is higher than 5 mm.

17. The LOX inhibitor for use according to any one of claims 14 to 16 wherein said inhibitor is selected from the group consisting of: P-aminopropionitrile (BAPN), -aminopropionitrile fumarate (BAF), P-aminopropionitrile maleate (BAM), P-aminopropionitrile monohydrate (BAMH), aminomethylenethiophene (AMT), 5-(naphthalen-2-ylsulfonyl)thiophen-2- yl)methamine 1, 2-aminomethylene-5-sulponylthiazole (AMTz), 5-(naphthalen-2- ylsulfonyl)thiazol-2-yl)methanamine 2, PXS-5505 (CAS No.: 2409963-83-1), PXS-5382 and Simtuzumab.

18. The LOX inhibitor for use according to any one of claims 14 to 17 wherein said injured tissue is lobular tissue, preferably adipose tissue.

Images