WO2012160031A1 - Method for prognosing solid tumor outcome - Google Patents

Abstract

The present invention concerns an in vitro method of prognosing outcome of a cancer in a patient afflicted with a solid tumor by determining the level of expression of thrombomodulin in a sample of the solid tumor, as well as an in vitro method for screening for a compound for the treatment of a solid tumor comprising tumor-associated stromal cells which express both thrombomodulin and protein JAM-C.

Description

METHOD FOR PROGNOSING SOLID TUMOR OUTCOME

The present invention concerns an in vitro method of prognosing outcome of a cancer in a patient afflicted with a solid tumor by determining the level of expression of thrombomodulin in a sample of the solid tumor, as well as an in vitro method for screening for a compound for the treatment of a solid tumor comprising tumor-associated stromal cells which express both thrombomodulin and protein JAM-C.

The invention further pertains to an in vitro method for selecting a patient afflicted with a cancer suitable to be treated with a compound which induces death of and/or inhibits tumor-associated stromal cells which express both thrombomodulin and protein JAM-C comprising determining the level of expression of thrombomodulin in a sample of the solid tumor.

The invention is also relative to compounds which inhibit and/or induce death of tumor-associated stromal cells expressing thrombomodulin for use as a medicament in the treatment of solid tumor which comprises tumor-associated stromal cells which express both thrombomodulin and protein JAM-C.

Finally, the invention pertains to an in vitro method for obtaining stromal cells expressing both thrombomodulin and protein JAM-C, and to stromal cells isolated therefrom.

BACKGROUND OF THE INVENTION

Despite the advances in tumor biology and in treatment options, the mortality of cancer patients is high and cancer remains a leading cause of death.

Traditionally, tumor development has been linked to multiple genetic alterations in somatic cells and has therefore been considered as a multiple step process along which a cell acquires several mutations before becoming cancerous cell, i.e. a cell which has defects in regulatory pathways that govern normal cell proliferation and homeostasis.

In recent years, the tumor microenvironment has become the focus of intense research, with the understanding that alterations that occur in the stromal environment (also called "tumor stroma") around a tumor might prove useful in prognosis and generate new therapeutic targets.

Several studies have shown that changes in the stromal environment influence cancer initiation, development, progression and can contribute to tumor metastatic potential. Stromal cells are suspected to produce many of the growth signals driving the proliferation of carcinoma cells (Skobe and Fusenig, Proc. Natl. Acad. Sci., 95: 1050- 1055, 1998). Tumor stroma consists of a structural network called extracellular matrix (EMC), and of different cell types such as inflammatory cells, cells that contribute to the vasculature (smooth muscle cells, endothelial cells and pericytes) and fibroblasts.

Fibroblasts are among the most abundant stromal cell types in the microenvironment of solid tumors, being particularly prominent in carcinomas of breast pancreas, colon and prostate. Recent studies tend to show that tumor-associated fibroblasts (tumor stromal cells) significantly enhance tumor growth through releasing of growth factor and chemokines (Kalluri and Zeisberg, Nat. Rev. Cancer, 6: 392-401 , 2006; Orimo et a\., Cell, 121 : 335-348, 2005; Yang et al., Cancer Res., 65 : 8887-8895, 2005 ; see Allen and Jones, Journal of Pathology, 223 : 162-176, 201 1 for review). Further, tumor-associated fibroblasts have the particularity of remaining constantly activated.

Molecular characteristics of activated fibroblasts have been extensively studied. Particularly, it was shown that activated fibroblasts are positive for the smooth muscle cells markers Desmin, a-smooth-muscle-actin (aSMA), γ-smooth-muscle-actin (ySMA) and the cytoskeletal protein that controls stress fiber integrity, Palladin 41 g, the intermediate filament Vimentin, the small mucin-like transmembrane protein podoplanin, the collagen type 1 maturation enzyme prolyl-4 hydroxylase (P4H), stromelysin-3, the adhesion molecule cadherin-1 1 and the dipeptidyl peptidase fibroblast activation protein (FAP) (see Allen and Jones, Journal of Pathology, 223 : 162-176, 201 1 for review). Other study showed that the tumor promoting effect of cancer associated fibroblasts is at least partially mediated by a high expression of SDF-1 that exerts direct effects on carcinoma cells and indirect effects by promoting angiogenesis.

In spite of the progress in the use of the above markers as indicators of the stromal fibroblasts with an activated fibroblast phenotype, there are still open questions concerning the identification of activated fibroblasts, since stromal fibroblasts with an activated fibroblast phenotype display heterogeneity and most of the markers are actually staining subpopulations and not all activated fibroblasts uniformly (Sugimoto et al., Cancer Biol. Ther., 5: 1640-1646, 2006). Since stromal cells play a critical role in determining outcome, as well as in determining response of the disease, it is of particular interest of identifying the subpopulations of tumor-associated stromal cells which are implied in stimulating tumor growth. DESCRIPTION OF THE INVENTION

Studying the function of lymph nodes stromal cells in the mechanisms of regulation of lymphocyte entry and exit from the lymph nodes, the Inventors have characterized a subpopulation of stromal cells (i.e. Fibroblastic Reticular cells, abbreviated by "FRCs") which differs from other subpopulation of FRCs by its ability to secrete homeostatic chemokines.

The inventors have also shown that this FRC subpopulation, hereinafter called FRC^DP (for double positive), has the particularity of expressing both JAM-C and thrombomodulin, and that JAM-C controls homeostatic cytokine secretion in this subpopulation. Since JAM-C and thrombomodulin are conjointly expressed only in this specific subpopulation of stromal cells, they represent specific biomarkers useful to identify these stromal cells that represent the prominent source of chemokines in lymph nodes.

Because stromal cells are suspected to produce many of the growth signals promoting the proliferation of tumor, the inventors then hypothesized that a stromal cell subpopulation with a phenotype similar to FRC^DP would exist in certain solid tumors and would be responsible of tumor growth due to chemokines secretion.

The inventors have shown that FRC^DP when admixed with tumor cells before engraftment in mouse animal model promote tumor growth in vivo. On the contrary, cells expressing Thrombomodulin disappear from regressing tumors of anti-JAM-C treated animals.

Thus, the inventors have demonstrated for the first time that thrombomodulin is a new biomarker of chemokine secreting stromal cells associated to poor prognosis in cancerology.

Because tumor-associated stromal cells expressing Thrombomodulin and JAM-C promote tumor growth via chemokines secretion, the higher the number of these cells is, the poorer the prognosis is. Hence, measuring thrombomodulin expression level in a sample from a solid tumor, possibly combined with measure of JAM-C expression, provides information about the likely outcome of a cancer.

Thrombomodulin is an integral membrane protein expressed on the surface of endothelial cells. Human thrombomodulin precursor is 575 amino acids long (GenBank accession number NP 000352, sequence present in the database at the date of filing), whereas mouse Human thrombomodulin precursor is 577 amino acids long (GenBank accession number NP 033404, sequence present in the database at the date of filing). Thrombomodulin is known as a member of the protein C anticoagulant system which serves as the primary natural anticoagulant system of the capillary bed. The anticoagulant function of Thrombomodulin is believed to be mediated by interaction with thrombin and protein C to form a complex which catalyzes the activation of protein C (Weiler et Isermann, J. Thromb. Haemost, 1 : 1515-1524, 2003). Thrombomodulin has been shown to also exert anti-inflammatory effects (for review see (Weiler and Isermann, J. Thromb. Haemost, 1 : 1515-1524, 2003).

To date, thrombomodulin has been described as a vascular specific protein.

It has been shown that a selective collapse of tumor vasculature and a resultant inhibition or tumor growth can be achieved by the administration of a combination of an agent that induces a procoagulant state, such as thrombomodulin inhibitor, and cytokines (for instance TNF-beta) or an inducer of cytokine production. However, neither an agent that induces a procoagulant state, nor a lymphotoxin alone, results in a total selective loss of blood flow to tumors in tumor-bearing animals (WO 95/34319).

JAM-C, for "Junctional Adhesion Molecule C", is a protein 310 amino acids long in human (GenKank accession number NP 1 16190.3 displays human isoform 1 precursor sequence, sequence present in the database at the date of filing). It is an adhesion molecule expressed by endothelial cells (ECs) that plays a role in tight junction formation, leukocyte adhesion, and transendothelial migration.

JAM-C had been shown to have a proangiogenesis effect stimulating the formation of new blood vessels that irrigate the tumor. It has been shown that an anti-JAM-C antibody inhibits angiogenesis, and consequently reduces tumor growth since the tumor is deprived in nutrients and oxygen usually provided by blood (Lamagna et al., Cancer Research, 65(13): 5703-5710, 2005).

Recently, a study showed that JAM-C also exists in soluble form (SJAM-C), and that soluble form plays a role in angiogenesis too, suggesting that modulation of sJAM-C may provide a novel route for controlling pathological angiogenesis (Rabquer et al., J. Immunol., 185(3): 1777-1785, 2010).

Now, the inventors have found that thrombomodulin can also be expressed by tumor associated stromal cells, and that these tumor associated stromal cells secrete chemokines which drive solid tumor growth. Prognosis of patients suffering from solid tumor

Therefore, the present invention provides an in vitro method of prognosing outcome of a cancer in a patient afflicted with a solid tumor, said method comprising:

a) measuring the expression level of thrombomodulin in a biological sample from a solid tumor from said patient, said biological sample comprising tumor-associated stromal cells; and

b) optionally deducing from the result of step a) the prognosis of said patient.

In one embodiment, the in vitro method of prognosing outcome of a cancer in a patient afflicted with a solid tumor further comprises a step prior to step a) consisting of providing or obtaining a biological sample from a solid tumor from said patient, said biological sample comprising at least tumor-associated stromal cells, and optionally tumor cells. By "patient afflicted with a solid tumor" is meant an individual suffering from any type of cancerous (malignant) tumor that is a localized mass of tissues, on the opposite of non- solid tumor like leukemia. The individual to be prognosed in the frame of the invention may be any mammal, for instance a dog, a cat, a horse, a non human primate, a human, a rat, a mouse. In a preferred embodiment, the individual is a human.

The solid malignant tumor may be for instance carcinomas, adenocarcinomas, sarcomas, malignant melanomas, mesotheliomas, blastomas. The carcinoma or adenocarcinoma may for example correspond to a bladder, a colon, a kidney, an ovary, a prostate, a lung, an uterus, a breast or a prostate carcinoma or adenocarcinoma. The blastoma may for example correspond to a neuroblastoma, a glioblastoma or a retinoblastoma. The cancer is preferably selected from the group consisting of prostate cancer (e.g. prostate adenocarcinoma), lung cancer (e.g. squamous cellular carcinoma), breast cancer (e.g. infiltrated ductal carcinoma), ovary cancer (e.g. serous papillary carcinoma), uterus cancer (squamous cellular carcinoma), brain cancer (e.g. astrocytoma), colon cancer (e.g. colon adenocarcinoma), colorectal cancer, rectal cancer (e.g. rectal adenocarcinoma), cancer of the striated muscle (e.g. rhabdomyosarcoma), thyroid cancer, testicular cancer. In a most preferred embodiment, the cancer is selected from the group consisting of lung cancer, prostate cancer, ovary cancer, uterus cancer, brain cancer, colon cancer, colorectal cancer, rectal cancer and cancer of the striated muscle, bladder cancer, liver cancer, kidney cancer, thyroid cancer.

As used throughout the present specification, the term "thrombomodulin" is meant to encompass any naturally occurring isoform of the thrombomodulin protein naturally encoded by the genome of the patient to be prognosed, allelic variants thereof and splice variants thereof. The thrombomodulin to be measured will depend on the patient to be prognosed (dog, cat, horse, non human primate, primate, rat, mouse...) and will be easily determined by the person skilled in the art. For instance, when the patient is a human, the term thrombomodulin is intended to mean any naturally occurring isoform of the thrombomodulin protein naturally encoded by human genome, including the protein having an amino acid sequence corresponding to amino acids 22 to 575 of SEQ ID NO: 1 (GenBank accession number NP 000352, sequence present in the database at the date of filing), allelic variants thereof and splice variants thereof.

In a preferred embodiment, the patient is human and the thrombomodulin to be measured is a thrombomodulin protein of SEQ ID NO: 1 .

Methods for measuring the expression level of thrombomodulin are well-known by one skilled in the art.

The expression level of thrombomodulin may be measured at the nucleic acid level (e.g. through RT-PCR) or at the level of the protein (e.g. through immunofluorescence - immunocytochemistry or immunohistochemistry-, flow cytometry or by western blot).

The protocols provided in the examples of the present application may for instance be used.

In a preferred embodiment, step a) of measuring the expression level of thrombomodulin is carried out on stromal cells extracted from the biological sample comprising tumor-associated stromal cells.

Methods for extracting (isolating) tumor-associated stromal cells from a solid tumor biological sample are well-known by one skilled in the art and are in particular described in Allinen. M. et al. (Cancer Cell, 6(1 ): 17-32, 2004). For instance, a biological sample from a solid tumor is harvested from the patient to be prognosed, dissected, digested in collagenase solution (about 200 units/ml) at 37 °C, then mechanically dissociated by trituring, filtered through a 70 μηι cell strainer, and centrifuged. The cells pellet is resuspended and then stromal cells expressing the membrane protein Thrombomodulin are purified. For instance, these cells may be purified by magnetic immunosorting with a negative sort for PECAM-1 and Lyve-1 expression and a positive sort for Thrombomodulin (using magnetic beads coated with anti-Thrombomodulin antibodies). They can also be purified using flow cytometry after having been labelled with anti-PECAM-1 , anti-Lyve-1 , anti-Thrombomodulin and anti-PDGFR-oc antibodies. Alternatively, tumor-associated stromal cells expressing thrombomodulin may be purified on the basis of expression of JAM-C and PDGFR-oc and the lack of PECAM-1 and Lyve-1 expression. In another embodiment, double labelling of both JAM-C and thrombomodulin may be used to isolate tumor-associated stromal cells that lack expression of PECAM-1 and Lyve-1 .

In another preferred embodiment, step a) of measuring the expression level of thrombomodulin is carried out on a section of the solid tumor sample comprising tumor- associated stromal cells and tumor cells. To perform the measure, the section is stained for thrombomodulin (for instance with an anti-thrombomodulin antibody conjugated with fluorescent dye) and quantification of the specific staining of stromal cells is determined, for instance by immunofluorescence (see examples of the present application).

In a specific embodiment of the in vitro method of prognosing outcome of a cancer in a patient afflicted with a solid tumor, expression level of thrombomodulin is normalized using a housekeeping gene expressed by stromal cells, for instance Beta actin, Beta-2- Microglobulin, Glyceraldehyde-3-phosphate-deshydrogenase Hydroxymethyl-bilane synthase, Hypoxanthine Phosphoribosyl-transferase I, Ribosomal protein L13a, Succinate deshydrogenase complex, subunit A, TATA box binding protein, Ubiquitin C (Vandesompele. J et al, Genome Biology, 2002). In particular, when determined by RT- PCR, the expression level of thrombomodulin may be normalized with the average of the expression level of the housekeeping genes in order to make mRNA expression values comparable in between different samples and experiments (for instance to minimize biases due to the different size and number of cells comprised in the samples to be analyzed).

Further, it is possible that in some solid tumors, tumor cells themselves express thrombomodulin (Liu PL et al.,. Mol Carcinog., 49(10):874-81 , 2010 ; Tamura A et al., Lung Cancer, 34(3):375-82, 2001 ; Ogawa H et al., Cancer Letter, 28;149(1 -2):95-103, 2000)

Therefore, in embodiments of the in vivo method of prognosing outcome of a cancer in a patient afflicted with a solid tumor of the invention wherein measurement of step a) performed directly on a biological sample comprising tumor-associated stromal cells and tumor cells is positive for thrombomodulin expression, result should be confirmed in isolated tumor-associated stromal cells from the biological sample using the method for extracting (isolating) tumor-associated stromal cells from a solid tumor biological sample described above and antibodies directed against PCAM-1 , Lyve-1 , PDGFR-oc, and JAM-C.

The inventors have found that tumor-associated stromal cells expressing both thrombomodulin and JAM-C secrete chemokines that promote tumor growth, in particular SDF-1 alpha (also called CXCL12), CCL21 and CCL19. Therefore, patient afflicted with a solid tumor that does not comprise, or comprises a low level of, tumor-associated stromal cells expressing both thrombomodulin and JAM-C is likely indicative of a good prognosis. On the contrary, the higher the number of these specific tumor-associated stromal cells is, the poorer the prognosis is.

Hence, the measurement of no or low expression levels of thrombomodulin in step a) is indicative of a good prognosis, whereas the measurement of high expression levels of thrombomodulin in step a) is indicative of a poor prognosis.

As used throughout the present specification, the term "poor prognosis" refers to a patient that is likely to present a short life-expectancy, and/or that is likely to develop metastases, and/or that is likely to relapse, and/or that is likely not to respond, or poorly respond, to treatments.

As indicated above, the level of expression of thrombomodulin measured in step a) provides an evaluation of the prognosis of the outcome of a cancer in a patient afflicted with a solid tumor. Therefore, in a preferred embodiment of the in vitro method of the invention, in order to evaluate the prognosis, the value obtained in step a) is applied to a standard calibration curve showing a relationship between concentration of thrombomodulin in the biological sample and the probable outcome (short life-expectancy, metastases development, relapse...).

The standard calibration curve can be obtained by determining the level of thrombomodulin in a large cohort of patients whose outcome is known.

In one embodiment, the biological sample is obtained prior to the patient receiving an anticancer therapy.

In another embodiment, the biological sample is obtained after a treatment with an anti-cancer agent is initiated. Preferably, the anti-cancer agent is a compound suitable for the treatment of a solid tumor comprising tumor-associated stromal cells which express both thrombomodulin and protein JAM-C according to the invention (see compounds as described below).

Method of identifying agents for the treatment of a solid tumor comprising tumor- associated stromal cells expressing thrombomodulin and protein JAM-C

The inventors have found that treatment with an antibody that reduces the tumor content of stromal cells expressing thrombomodulin reduces tumor-growth. Accordingly, to obtain new anti-cancer treatment, it is highly desirable to identify further compounds able to either induce death of these cells, or at least able to inhibit secretion of chemokines by tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C, knowing that tumor growth is promoted by these secreted chemokines.

Accordingly, another subject of the invention relates to an in vitro method for screening for a compound suitable for the treatment of a solid tumor comprising tumor- associated stromal cells which express both thrombomodulin and protein JAM-C, the method comprising the steps of:

a) bringing into contact a candidate compound with stromal cells expressing both thrombomodulin and protein JAM-C ;

b) determining whether said candidate compound inhibits chemokines expression by, and/or induces death of, tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C;

wherein the candidate compound which inhibits chemokines expression by, and/or induces death of, tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C is identified as a compound suitable for the treatment of a solid tumor comprising tumor-associated stromal cells which express both thrombomodulin and protein JAM-C.

Stromal cells expressing both thrombomodulin and protein JAM-C usable to carry out the in vitro method for screening of the invention are preferably tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C isolated from solid tumor expressing thrombomodulin, or fibroblastic Reticular Cells expressing both thrombomodulin and protein JAM-C (called FRCs^DP) isolated from lymph nodes.

Determining inhibition of chemokines expression may be performed at the nucleic acid level (e.g. through RT-PCR) or at the level of the protein (e.g. by western blot).

In a particularly interesting embodiment of the in vitro method for screening according to the invention, chemokines whose expression is determined are SDF-1 alpha, CCL21 and/or CCL19.

"Death cell" is intended to mean apoptotic cell death as well as oncotic cell death. Death of cells can readily be determined by one skilled in the art. Indeed, many methods for measuring apoptosis and oncosis are known in the art, and are for example reviewed in the following document: Apoptosis: Methods and Protocols, Second Edition (Methods in Molceular Biology). P. Erhard (Editor), Ambrus Toth (Editor), Springer Protocols, Humana Press.

For instance, the staining of cells with vital dyes is a commonly used approach to quantify cell death. Further, since cleavage of genomic DNA into oligonucleosome-length fragments by endogenous nucleases is an hallmark of apoptosis, apoptotic death can be measured by detecting/measuring fragmented DNA (e.g. terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end labelling of DNA fragments ("TUNEL") and in situ en labelling ("ISEL") tehcniques). Apoptosis can also be determined by measuring lactate dehydrogenase enzyme release. The candidate compound may correspond to any type of compound. It may for example correspond to a small molecule or an antibody or a fragment thereof, an aptamer, an aptamer conjugated with a cytotoxic agent. In a preferred embodiment, the test compound is a small molecule and a library of small molecules is screened with the method according to the invention.

Once a compound which inhibits chemokines expression by, and/or induces death of, tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C, has been identified, a pharmaceutical composition comprising the compound and a pharmaceutically acceptable carrier may be manufactured. The pharmaceutical composition may be tested in a non-human animal model afflicted with a solid tumor comprising tumor-associated stromal cells and tumor cells in order to confirm in vivo its efficacy.

Treatment of a solid tumor comprising tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C

Since tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C promote tumor growth, these cells represent a valuable therapeutic target for treating solids tumor comprising such cells. Then, compounds which induce death of these cells, or which inhibit secretion of chemokines by these stromal cells are useful for treating solid tumors comprising tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C.

Therefore, the present invention provides a method of treating a solid tumor comprising tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C, the method comprises the step of administering to an individual in need thereof an effective amount of at least one compound which induces death of tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C, and/or which inhibits secretion of chemokines by these tumor associated stromal cells, or a composition comprising such a compound.

The invention also relates to a compound which inhibits secretion of chemokines by, and/or induces death of, tumor-associated stromal cells expressing thrombomodulin for use as a medicament in the treatment of solid tumor, wherein the solid tumor comprises tumor-associated stromal cells which express both thrombomodulin and protein JAM-C.

By "method of treating a solid tumor comprising tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C " is meant a method aiming at curing, improving the condition and/or extending the lifespan of an individual suffering from a solid tumor comprising tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C.

By "effective amount" is meant an amount sufficient to achieve a concentration of compound which is capable of treating the disease to be treated. Such concentrations can be routinely determined by those of skilled in the art. The amount of the compound actually administered will typically be determined by a physician, in the light of the relevant circumstances, including the condition to be treated, the chosen route of administration, the actual compound administered, the age, weight, and response of the individual patient, the severity of the patient's symptoms, etc.

By "individual in need thereof" is meant an individual suffering from or susceptible of suffering from the disease to be treated. The individual to be treated in the frame of the invention may correspond to any mammal, e.g. a dog, a cat, a horse, a non human primate, a human. In a preferred embodiment, the individual is a human.

The compound which induces death of tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C may correspond to any type of molecule, such as small molecules, antibodies naturally capable of inducing lysis of tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C (e.g. complement- dependent lysis, cell-dependent lysis), antibodies conjugated with a cytotoxic molecule, and aptamers, in particular aptamers conjugated with a cytotoxic molecule.

As used herein, the terms "antibody" and "immunoglobulin" have the same meaning and are used indifferently in the present invention. The term "antibody" refers to immunoglobulin molecules and immunologically active portions of immunoglobulin molecules, i.e., molecules that contain an antigen binding site that immunospecifically binds an antigen. As such, the term antibody encompasses not only whole antibody molecules, but also antibody fragments as well as variants (including derivatives) of antibodies and antibody fragments.

In particular, the antibody according to the invention may correspond to a polyclonal antibody, a monoclonal antibody (e.g. a chimeric, humanized or human antibody), a fragment of a polyclonal or monoclonal antibody or a diabody.

In natural antibodies, two heavy chains are linked to each other by disulfide bonds and each heavy chain is linked to a light chain by a disulfide bond. Each chain contains distinct sequence domains. The light chain includes two domains, a variable domain (V_L) and a constant domain (C_L). The heavy chain includes four domains, a variable domain (V_H) and three constant domains (C_H1 , C_H2 and C_H3, collectively referred to as C_H). The variable regions of both light (V_L) and heavy (V_H) chains determine binding recognition and specificity to the antigen. The specificity of the antibody resides in the structural complementarity between the antibody combining site and the antigenic determinant. Antibody combining sites are made up of residues that are primarily from the hypervariable or complementarity determining regions (CDRs). They refer to amino acid sequences which, together, define the binding affinity and specificity of the natural Fv region of a native immunoglobulin binding site. The light and heavy chains of an immunoglobulin each have three CDRs, designated L-CDR1 , L-CDR2, L-CDR3 and H-CDR1 , H-CDR2, H-CDR3, respectively. Therefore, an antigen-binding site includes six CDRs, comprising the CDR set from each of a heavy and a light chain V region.

Framework regions (FRs) refer to amino acid sequences interposed between

CDRs, i.e. to those portions of immunoglobulin light and heavy chain variable regions that are relatively conserved among different immunoglobulins in a single species, as defined by Kabat et at., 1991 (Kabat et at., 1991 , Sequences of Proteins Of Immunological Interest, National Institute of Health, Bethesda, Md). As used herein, a "human framework region" is a framework region that is substantially identical (about 85%, or more, in particular, 90%, 95% or 100%) to the framework region of naturally occurring human antibody.

The term "monoclonal antibody" or "mAb" as used herein refers to an antibody molecule of a single amino acid composition, that is directed against a specific antigen and which may be produced by a single clone of B cells or hybridoma, or by recombinant methods.

A "humanized antibody" is a chimeric, genetically engineered, antibody in which the CDRs from a mouse antibody ("donor antibody") are grafted onto a human antibody ("acceptor antibody"). Thus, a humanized antibody is an antibody having CDRs from a donor antibody and variable region framework and constant regions from a human antibody. The use of antibody components derived from humanized monoclonal antibodies obviates potential problems associated with the immunogenicity of murine constant regions.

"Antibody fragments" comprise a portion of an intact antibody, preferably the antigen binding or variable region of the intact antibody. Examples of antibody fragments include Fv, Fab, F(ab')₂, Fab', Fd, dAb, dsFv, scFv, sc(Fv)₂, CDRs, diabodies and multi- specific antibodies formed from antibodies fragments.

The term "Fab" denotes an antibody monovalent fragment having a molecular weight of about 50,000 and antigen binding activity, and consisting of the V_L, V_H, C_L and C_H1 domains. The Fv fragment is the N-terminal part of the Fab fragment and consists of the variable portions of one light chain and one heavy chain.

The term "F(ab')₂" refers to an antibody bivalent fragment having a molecular weight of about 100,000 and antigen binding activity, which comprises two Fab fragments linked by a disulfide bridge at the hinge region.

The term "Fab"' refers to an antibody fragment having a molecular weight of about 50,000 and antigen binding activity, which is obtained by cutting a disulfide bond of the hinge region of the F(ab')₂ fragment.

The term "Fd" refers to an antibody fragment consisting of the V_H and C_H1 domains.

The term "dAb" (Ward et al., 1989 Nature 341 :544-546) refers to a single variable domain antibody, i.e. an antibody fragment which consists of a V_H or V_L domain.

A single chain Fv ("scFv") polypeptide is a covalently linked V_H::V_L heterodimer which is usually expressed from a gene fusion including V_H and V_L encoding genes linked by a peptide-encoding linker. "dsFv" is a V_H::V_L heterodimer stabilised by a disulfide bond.

Divalent and multivalent antibody fragments can form either spontaneously by association of monovalent scFvs, or can be generated by coupling monovalent scFvs by a peptide linker, such as divalent sc(Fv)₂.

The term "diabodies" refers to small antibody fragments with two antigen-binding sites, which fragments comprise a V_H domain connected to a V_L domain in the same polypeptide chain (V_H-V_L). By using a linker that is too short to allow pairing between the two domains on the same chain, the domains are forced to pair with the complementarity domains of another chain and create two antigen-binding sites.

Antibodies according to the invention may be produced by any technique known in the art, such as, without limitation, any chemical, biological, genetic or enzymatic technique, either alone or in combination. The antibodies of this invention can be obtained by producing and culturing hybridomas.

"Aptamers" are a class of molecule that represents an alternative to antibodies in term of molecular recognition. Aptamers are oligonucleotide or oligopeptide sequences with the capacity to recognize virtually any class of target molecules with high affinity and specificity. Such ligands may be isolated through Systematic Evolution of Ligands by

Exponential enrichment (SELEX) of a random sequence library, as described in Tuerk C. and Gold L, Science, 1990, 249(4968):505-10. The random sequence library is obtainable by combinatorial chemical synthesis of DNA. In this library, each member is a linear oligomer, eventually chemically modified, of a unique sequence. Possible modifications, uses and advantages of this class of molecules have been reviewed in Jayasena S.D., Clin. Chem., 1999, 45(9):1628-50. Peptide aptamers consists of a conformational^ constrained antibody variable region displayed by a platform protein, from combinatorial libraries by two hybrid methods (Colas et al., Nature, 1996,380, 548- 50).

In a preferred embodiment, the compound which induces death of tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C is selected from the group consisting of: an anti-thrombomodulin antibody, an anti-JAM-C antibody, an aptamer which specifically binds to thrombomodulin and which is conjugated with a cytotoxic molecule, and an aptamer which specifically binds to JAM-C and which is conjugated with a cytotoxic molecule. Preferably, the compound which induces death of tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C is an cytotoxic anti-thrombomodulin antibody or an aptamer which specifically binds to thrombomodulin and which is conjugated with a cytotoxic molecule.

In the frame of the present invention, the term "cytotoxic molecule" is intended to mean any molecule which is armful for a cell, and which can be conjugated with an antibody or with an aptamer.

For instance, a cytotoxic molecule can be selected from the group consisting of:

- a toxin, for instance ricin (in particular ricin A chain), abrin, gelonin, calicheamicin, diphtheria toxin, Pseudomonas exotoxin, Cholera Toxin, saporin, pokeweed antiviral protein, a-sarcin, restrictocin, human pancreatic ribonuclease A (for review Choudhary. S, et al, Drug Discovery Today, April 201 1 ) ;

- a radioisotope, for instance yttrium 90 (⁹⁰Y), Iodine 131 ( ³ l), Rhenium 188 ( ⁸⁸Re), Holmium 166 ( ⁶⁶Ho), Technetium 99m (⁹⁹Tc^m), Samarium 153 ( ⁵³Sm), Lutetium 177 ( ⁷⁷Lu), Gallium 67 (⁶⁷Ga), Thallium 201 (²⁰ TI), Indium 1 1 1 ( η) (For review Labelling Techniques of Biomolecules for Targeted Radiotherapy, IAEA TECDOC Series No. 7359, 2003);

- maytansinoids (in particular DM1 and DM4);

- Anthracycline drugs, Duocarmycin analogs, taxoids (For review see Kovtun. Y. V.et al., Cancer Letters, 2007)

The compounds according to the invention are capable of inhibiting chemokines expression by, and/or induce death of, tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C in vivo and/or in vitro. The agent may inhibit chemokines expression and/or induce death of tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C by at least about 10%, preferably by at least about 30%, preferably by at least about 50%, preferably by at least about 70, 75 or 80%, still preferably by 85, 90, 95, or 100%.

The method of treating according to the invention preferably corresponds to a combination chemotherapy. Indeed, while the compounds of the invention induce death of the tumor-associated stromal cells which secrete tumor-growth promoting chemokines, or at least inhibit expression of these tumor-growth promoting chemokines, they do not target the tumor cells. Therefore, it is suitable that at least one of the following anti-cancer agents targeting the tumor cells be administrated to an individual in combination with the compounds of the invention (simultaneously or sequentially):

- an alkylating agent such as Cyclophosphamide, Chlorambucil and Melphalan;

- an antimetabolite such as Methotrexate, Cytarabine, Fludarabine, 6- Mercaptopurine and 5- Fluorouracil;

- an antimitotic such as Vincristine, Paclitaxel (Taxol), Vinorelbine, Docetal and

Abraxane;

- a topoisomerase inhibitor such as Doxorubicin, Irinotecan, Platinum derivatives, Cisplatin, Carboplatin, Oxaliplatin;

- a hormonal therapy drug such as Tamoxifen;

- an aromatase inhibitor such as Bicalutamide, Anastrozole, Examestane and

Letrozole;

- a signaling inhibitor such as Imatinib (Gleevec), Gefitinib and Erlotinib;

- a monoclonal antibody such as Rituximab, Trastuzumab (Herceptin) and Gemtuzumab ozogamicin;

- a biologic response modifier such as Interferon-alpha;

- a differentiating agent such as Tretinoin and Arsenic trioxide; and/or

- an agent that block blood vessel formation (antiangiogenic agents) such as Bevicizumab, Serafinib and Sunitinib.

In addition, the method of treating solid tumor according to the invention may be associated with a radiation therapy and/or surgery.

Method for selecting patients afflicted with a cancer suitable to be treated with a compound targeting tumor-associated stromal cells expressing thrombomodulin and protein JAM-C

Now that the inventors have shown that tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C are particularly interesting targets for treating a subset of solid tumor (i.e. solid tumor comprising tumor-associated stromal cells expressing thrombomodulin and protein JAM-C), it is important to be able to select patients suitable to be treated with compounds which inhibit chemokines expression by, and/or induces death of, tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C.

Thus, the present invention also concerns an in vitro method for selecting a patient afflicted with a solid tumor suitable to be treated with a compound which inhibit chemokines expression by, and/or induces death of, tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C, comprising:

a) measuring the expression level of thrombomodulin, and optionally of JAM-C, in a biological sample from a solid tumor from said patient, said biological sample comprising at least tumor-associated stromal cells, and optionally tumor cells ;

b) selecting the patient expressing thrombomodulin as suitable to be treated with a compound which inhibit chemokines expression by, and/or induces death of, tumor- associated stromal cells expressing both thrombomodulin and protein JAM-C.

The in vitro method for selecting a patient afflicted with a solid tumor suitable to be treated with a compound which inhibit chemokines expression by, and/or induces death of, tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C according to the invention can be carried out just after the cancer is diagnosed, or during an anti-cancer treatment to follow up evolution of the disease, i.e. to test for emergence, or on the contrary for disappearance, of tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C.

Method for obtaining stromal cells secreting chemokines and cells therefrom

Another object of the invention relates to an in vitro method for obtaining stromal cells secreting chemokines comprising a step a) consisting of extracting stromal cells expressing both thrombomodulin and protein JAM-C from a biological sample in which stromal cells expressing both thrombomodulin and protein JAM-C are present. In one embodiment, the in vitro method for obtaining stromal cells secreting chemokines further comprises a step prior to step a) consisting of providing or obtaining a biological sample in which stromal cells expressing both thrombomodulin and protein JAM-C are present. In a preferred embodiment, the biological sample in which stromal cells expressing both thrombomodulin and protein JAM-C are present is a lymph node tissue and the stromal cells secreting chemokines extracted are Fibroblastic Reticular cells expressing both thrombomodulin and protein JAM-C.

In another preferred embodiment, the biological sample in which stromal cells expressing both thrombomodulin and protein JAM-C are present is a solid tumor tissue which comprises tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C.

Methods for extracting (isolating) tumor-associated stromal cells from a solid tumor biological sample are well-known by one skilled in the art and are in particular described in Allinen. M. et al. (Cancer Cell, 6(1 ): 17-32, 2004). For instance, a biological sample from a solid tumor is harvested from the patient to be prognosed, dissected, digested in collagenase solution (about 200 units/ml) at 37 °C, then mechanically dissociated by trituring, filtered through a 70 μηι cell strainer, and centrifuged. The cells pellet is resuspended and then stromal cells expressing the membrane protein Thrombomodulin are purified. For instance, these cells may be purified by magnetic immunosorting with a negative sort for PECAM-1 and Lyve-1 expression and a positive sort for Thrombomodulin (using magnetic beads coated with anti-Thrombomodulin antibodies). They can also be purified using flow cytometry after having been labelled with anti-PECAM-1 , anti-Lyve-1 , anti-Thrombomodulin and anti-PDGFR-oc antibodies. Alternatively, tumor-associated stromal cells expressing thrombomodulin can be purified on the basis of expression of JAM-C and PDGFR-oc and the lack of PECAM-1 and Lyve-1 expression.

It is to be noted that isolated stromal cells secreting chemokines characterized in that they express both thrombomodulin and protein JAM-C, in particular fibroblastic reticular cells and tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C, are also part of the present invention.

Kits according to the invention

The invention further discloses kits that are useful in the above methods. Such kits comprise means for measuring the expression level of thrombomodulin, and optionally means for measuring the expression level of JAM-C.

They can be used, e.g. for prognosing the outcome of a cancer in a patient, for designing a treatment regimen, for monitoring the progression of the cancer, and/or for monitoring the response of the individual to a drug (i.e. "drug monitoring").

The kit may further comprise means for measuring the expression level of other cancer markers, such as e.g. means for measuring the expression level of VEGF. Optionally, the kit may further comprise means for measuring the expression level of some housekeeping genes.

In a preferred embodiment, the kit according to the invention comprises, in addition to the means for measuring the expression level of thrombomodulin, a control sample comprising a known amount of thrombomodulin.

The kits according to the invention may for example comprise, in addition to the means for measuring the expression level of thrombomodulin and/or JAM-C, one of (i) to (iii) below:

i. a standard calibration curve showing a relationship between concentration of thrombomodulin and/or JAM-C in the biological sample and the probable outcome

(short life-expectancy, metastases development, relapse...);

ii. a negative control sample indicative of the expression level of thrombomodulin and/or JAM-C in a healthy individual;

iii. instructions for the use of said kit in prognosing solid tumor.

Such a kit may for example comprise (i) and (ii), (i) and (iii), (ii) and (iii), or (i), (ii) and

(iii).

Means for measuring the expression level of thrombomodulin and JAM-C are well- known in the art. They include, e.g. reagents allowing the detection of thrombomodulin and JAM-C mRNAs by real-time quantitative-PCR, such as primers specific for thrombomodulin and JAM-C. When the kit comprises means for real-time quantitative- PCR thrombomodulin mRNA and/or JAM-C mRNA detection, the kit may further comprise a second reagent, labeled with a detectable compound, which binds to thrombomodulin mRNA and/or JAM-C mRNA synthesized during the PCR, such as e.g. SYBER GREEN reagents.

Means for measuring the expression level of thrombomodulin or JAM-C may also include antibodies specifically binding to thrombomodulin or JAM-C. Such means can be labeled with detectable compound such as fluorophores or radioactive compounds. For example, the probe or the antibody specifically binding to thrombomodulin or to JAM-C may be labeled with a detectable compound. Alternatively, when the kit comprises an antibody, the kit may further comprise a secondary antibody, labeled with a detectable compound, which binds to an unlabelled antibody specifically binding to thrombomodulin or JAM-C.

The means for measuring the expression level of thrombomodulin and/or JAM-C may also include reagents such as e.g. reaction, hybridization and/or washing buffers. The means may be present, e.g., in vials or microtiter plates, or be attached to a solid support such as a microarray as can be the case for primers and probes. The kit may for example include the anti-thrombomodulin polyclonal goat IgG manufactured by R&D system® (clone number 461714) as a mean for measuring the expression level of thrombomodulin.

Advantageously, the kits of the invention consists of means for measuring the expression level of thrombomodulin and JAM-C, and optionally one of (i) to (iii) as recited above.

The present invention will be further illustrated by the additional description and drawings which follow, which refer to examples illustrating the isolation and characterization of a subset of stromal cells expressing chemokines, and their role in promoting tumor growth. It should be understood however that these examples are given only by way of illustration of the invention and do not constitute in anyway a limitation thereof. BRIEF DESCRIPTION OF THE FIGURES

- Figure 1 illustrates antibody (mAb) LS17-9 characterization

Fig. 1 .A: Histogram plots illustrating the comparative expression by flow cytometry of JAM-C, PECAM-1 and LS17.9 on LyEnd-5 and bEnd-2 cell lines. The LS17.9 monoclonal antibody stains specifically the LyEnd-5 cell line above the first decade.

Fig.l .B: Silver staining of immunoprecipitated material from lysates obtained from 100 x 106 LyEnd-5 cells. Cell lysates were immunoprecipitated using isotype control antibody (9B5) or LS17.9 mAb. Specific bands are observed between 75 and 1 15 kDa. The peptide sequence identified by mass spectrometry matches with mouse thrombomodulin protein of GenBank accession number NP 033404.1 .

Fig. 1 .C: Histogram plots comparing reactivities of commercially available anti- Thrombomodulin (dashes profiles) and LS17.9 (plain grey profiles) antibodies on the indicated mouse endothelial cell lines and 293T cells transfected with empty pcDNA3 vector or vector containing thrombomodulin cDNA.

Fig. 1 .D: Indicated cell lysates were immunoprecipitated with LS17.9 mAb and immunoblotted using commercially anti-mouse thrombomodulin coupled to biotin (ref.BAF3894, R&D system). Blots were probed with streptavidin conjugated to horseradish peroxidase (Jackson Immunoresearch) and immunoreactive bands were visualized by enhanced chemoluminescence (Pierce). - Figure 2 illustrates a Dot plot showing Thrombomodulin expression on endothelial (PECAM-1 ) and lymphatic cells (Lyve-1 ). Cell suspensions obtained from collagenase dissociated LNs were gated on nonhematopoietic cells (CD45Neg) and analysed for PECAM-1 , Lyve-1 and thrombomodulin (LS17.9) expression.

- Figure 3 illustrates characterization of the cell subsets of the fibroblastic compartment CD45^NegPecam-1 ^NegLyve-1 ^Neg.

- Figure 4 illustrates the phenotypic characterization of primary fibroblastic reticular cells. Histogram plots show cell surface expression of PDGFRa, Thrombomodulin, JAM-

C, VCAM-1 , ICAM-1 , Gp38, CD45 and Pecam-1 on primary cultured Fibroblastic Reticular Cells.

- Figure 5 illustrates characterization of FRCs^DP (fibroblastic reticular cells expressing both Thrombomodulin and JAM-C).

Fig. 5.A: Adhesion of lymphocytes to primary stromal cells under the indicated conditions.

Fig. 5.B: Transmigration of naive T cells toward medium containing FCS (grey bar) or medium containing stromal cells (white bar). The negative control consists in Medium without FCS added in the lower chamber (black bar). Results are expressed as percent of input +/- SEM. Representative data from 3 independent experiments are shown.

- Figure 6 illustrates the relative expression of SDF-1 oc, CCL19, CCL21 , JAM-C, PDGFRa and ocSMA by FRCs^DP and FRCs^DN (fibroblastic reticular cells that do not express Thrombomodulin and JAM-C). qPCR analysis of SDF-1 oc, CCL19, CCL21 , PDGFRa, JAM-C and aSMA in freshly isolated sorted FRCs. Results are normalized to GAPDH. Data represent mean +/- SEM. Representative data from 3 independent experiments are shown. ^** p< 0,01 .

- Figure 7 illustrates the results of a study of the mechanisms controlling chemokine secretion in FRCs^DP.

Fig. 7.A: Histogram plot showing JAM-C expression on primary fibroblastic reticular cells upon transfection with control siRNA (grey filled profile) or siRNA against JAM-C (dashed profile).

Fig. 7.B: SDF-1 a secretion quantification by untransfected primary fibroblastic reticular cells (black bars) and cells transfected with control siRNA (grey bars) or two different siRNA against JAM-C (1 and 2, white bars). Fig. 7.C: The graph shows the number of cells in individual wells containing untransfected primary fibroblastic reticular cells (black bars) or cells transfected with control siRNA (grey bars) or siRNA against JAM-C (white bars) at the time of supernatant harvesting for SDF-1 a quantification. Results are expressed as mean value +/- SEM obtained on six independent wells. One representative experiment is shown.

Fig. 7.D: Quantification of intranodal homeostatic chemokine content in Jam-C deficient mice (white bars) as compared to wild type littermate control mice (black bars). n=3. ^* p< 0.05. - Figure 8 illustrates the results of a study of the physiological function of JAM-C expression in LNs. Intranodal homeostatic chemokine content was determined in mice treated with monoclonal antibodies directed against JAM-C. qPCR analysis of SDF-1 oc and CCL21 chemokines expression in LN of mice treated with an anti-JAM-C antibody. Results are normalized to GAPDH. Data represent mean +/- SEM. Representative data from 3 independent experiments are shown. ^** p<0.01 and ^*** p<0.001 .

- Figure 9 illustrates the quantification of tumor mass in syngenic C57BI/6 mice injected subcutaneously with B16F10 tumor cells and the indicated stromal cell subset. Three weeks after engraftment, tumors were resected and weighted.

- Figure 10 illustrates the quantification of tumor mass in syngenic C57BI/6 mice injected subcutaneously with B16F10 tumor cells in the presence or absence of stromal cells. Two weeks after engraftment, tumors were dissected and weighted. Results are expressed as mean tumor weights +/- SEM obtained in two independent experiments. ^** p<0.01

- Figure 1 1 illustrates the quantification of IHC staining with LS17-9 recognizing Thrombomodulin and Pecam recognizing endothelial cells on 26 sections of six different tumors. The Thrombomodulin decrease is twice as much as the reduction in intratumoral vascularization visualized by anti-PECAM staining. ^*** p<0.001

Example 1 : Materials and Methods

Generation of mAbs against surface markers of LyEnd.5

The panel of LS antibodies against the middle-T transformed lymphatic endothelial cell line (hereafter abbreviated to "LyEnd.5") was generated starting from membrane preparations of 180 x 10⁶ cells obtained using polytron and differential centrifugation as previously described by Aurrand-Lions et al. (Immunity, 5: 391 -405, 1996). Male fisher rats were immunized by intraperitoneal injection of 300μΙ of membrane preparation mixed with RIBI adjuvant (SIGMA). Two days after the final boost, splenocytes were fused to Sp2/0 cells, and hybridomas were selected in HAT-containing medium. Resistant clones were screened by flow cytometry according to their specific reactivity with LyEnd.5. The LS17-9 mAb that is specifically directed LyEnd.5, expressed by one of the screened clones, is further described in this study. It is of lgG_2a isotype and was purified using beads coupled to protein G according to manufacturer instructions. Cell lines

The middle-T transformed lymphatic endothelial cell line (LyEnd.5), brain-derived endothelial cell line (bEnd.2), thymic-derived endothelial cell line (tend.V⁺ and tend.V^") (14) and 293T cells (ATCC), were cultured in DMEM supplemented with 10% fetal calf serum (FCS).

LN primary stromal cells were obtained as follows. LNs were harvested and triturated with needles before digestion for 30 min at 37°C in 1 ml RPMI 1640 medium containing collagenase I (1 mg/ml, SIGMA), DNAse I (40 μg/m\, SIGMA) and 2% FCS under agitation. Digestion was stopped with 9 mL of RPMI medium containing 2% FCS. After centrifugation, cells were suspended in PBS containing EDTA (5 mM), filtered through a 70-μηι cell strainer (BD Biosciences), washed twice and plated in DMEM medium containing 10% FCS. After 24 hours, non-adherent cells were removed. When cells reached 80% confluency, cells were splitted ½ and used within 15 passages. For experiments using siRNA, cells were transfected using Lipofectamine RNAiMax (Invitrogen) and siRNA against JAM-C (J-044688-09 and J-044688-10, Dharmacon) or control non-targeting siRNA pool (D-001 1810-10-05, Dharmacon). Medium was changed after 24 hours and cells were counted at the time of supernatant harvesting to control growth.

Mice

C57BL/6 mice were purchased from Janvier, France. Jam-C deficient mice were obtained by heterozygous crossing of SV/129 and C57BI/6 mice. Generation of chimeric Ubiquitin-EGFP mice has been previously described (see Bajenoff et al., Immunity, 25: 989-1001 , 2006).

For experiment studying the effect of antibodies in vivo, mice were treated every day over three days with intraperitoneal (i.p) or intravenous (i.v) injection of 150 μg of antibodies (9B5, 13H33 or PBS). All experiments were performed in agreement with the French Guidelines for animal handling.

Flow Cytometry and Histology

Data were acquired on a BD LSRM SORP using DIVA software and analysed with

FlowJo 7.2.2 software. Antibodies used are listed below in Tablel .

Clone Class Origin

Endothelial markers

Pecam-1 390 rat lgG2a Ebioscience

GC51 rat lgG2a Home made

Lyve-1 X polyclonal rabbit RELIATech

Adhesion molecules

JAM-C 13h33 rat lgG2a Home made

JAM-C 501 X polyclonal rabbit Home made

JAM-C X polyclonal goat IgG R&D system

!CAM-1 YN1/1 .4 rat lgG2b Ebioscience

VCAM-1 429 rat lgG2a Ebioscience

Other Stromal markers

Gp38 8.1.1 hamster IgG Ebioscience aSMA 1A4 mouse lgG2a SIGMA

PDGFRa APA-5 rat lgG2a Ebioscience

THBD LS17.9 rat lgG2a Home made

THBD 461714 polyclonal goat IgG R&D system

Hematopoietic markers

CD45 30-F-11 rat lgG2b Ebioscience

CD3 145-2C11 hamster IgG Ebioscience

CD4 RM4-5 rat lgG2a Ebioscience

CD8 53-6.7 rat lgG2a Ebioscience

CD19 1D3 rat lgG2a Ebioscience

CD69 H1.2F3 hamster IgG Ebioscience

Chemokine marker

CCL21 X polyclonal goat IgG R&D system

Table 1 : Antibodies used to carried out flow cytometry and histology studies

For FRC population sorting, cells suspensions were obtained from peripheral and mesenteric LNs of five mice. A combination of hematopoeitic antibodies, conjugated to Phycoerythrin was used for depletion with anti rat magnetic beads (Dynal, Invitrogen) and the remaining cells were stained with PDGFR ^APC and LS17.9^AF88 and sorted using FacsAria. For immunohistochemistry, 7pm LN cryosections were fixed in methanol (5mn, -20°C), dried and rehydrated in PBS 0.5% BSA before processing for immunostaining. All secondary probes were from Invitrogen and Jackson Immunoresearch. Images were acquired using a Zeiss LSM510 Meta confocal microscope. Adhesion and Transmigration assay

For adhesion assay, 1 10⁵ naive or activated lymphocytes were plated on stromal cell monolayer during 1 hour. After washes, cells were recovered by trypsinization and quantified by cell counting upon staining with antibodies against CD45, CD3, CD4, CD8 and CD19, Lymphocyte activation was achieved by overnight incubation (37°C) of naive lymphocytes with PMA (20ng/ml) and lonomycine ( g/m!). Pertussis toxin (Ptx) treatment consisted in 6 hours incubation of lymphocytes (37°C) with Ptx 100ng/ml.

For transmigration assay, 5 x 10³ stromal cells were plated in the bottom chamber of

HTS transwell 96 well permeable support with polycarbonate membrane of 5 μηι pore size (Corning). Two days later, lymphocytes were added in the top chamber of transwells and absolute cell numbers of transmigrated cells were quantified. Enzyme-linked Immunosorbent Assay

CCL19, CCL21 and SDF1a were quantified by ELISA according to manufacturer's instructions (Duoset kit, R&D Systems). For quantification of LN chemokine contents, individual peripheral LNs were resuspended in 30μΙ, pooled and disrupted using Polytron. The soluble fractions were separated from nuclear and membrane fractions by centrifugation. ELISAs were performed on soluble fractions and results were expressed as ng/ml.

RT-PCR analysis

Total RNA were isolated using RNAeasy microkit (QIAGEN) and reverse transcribed using Superscript II kit (Invitrogen).

Q-PCR was performed using Power Sybr green PCR master mix (Applied Biosystem) and primer pairs listed below in Table 2.

Table 2: Primer pairs used to carry out Q-PCR (corresponding respectively to SEQ

ID Nos: 2 to 15 of the sequence listing).

Immunohistology studies of tumor sections

To quantify Thrombomodulin and PECA positive (vascular volume) fractions, tumor sections were fixed with acetone/methanol (1 :1 ) for 5 minutes at -20°C, dried, and hydrated in PBS/0.2% gelatin/0.05% Tween 20. Sections were then incubated with LS17- 9 mAb against thrombomodulin and a polyclonal rabbit antibody against PECAM-1/CD31 ( Ref: Piali L. et al. J. Cell. Biol., 130: 451-60, 1995) for 1 hour at room temperature. After three washes in PBS they were incubated with secondary antibodies coupled to Texas- Red or FITC and specifically directed against rabbit or rat antibodies (Jackson Immunoresearch). Six tumor samples for each experimental condition (9B5 Control or H33 anti-JAM-C treated mice) were used. For each tumor, twenty six pictures were acquired using LSM510 microscope (Zeiss) and analyzed using Metamorph software. The thrombomodulin and vascular volume fraction were quantified by determining the area of Thrombomodulin or PECAM-1 -positive staining across high power field pictures. Results are expressed as mean area in pixels per High Power Field (HPF) +/- SEM. The percentage of loss of staining is obtained by applying the formula: % Loss= [Mean area obtained on tumors of H33 anti-JAM-C treated mice/Mean area obtained on tumors of 9B5 treated animals] ^χ 100.

Tumor graft

Five C57BI/6 mice were injected subcutaneously with 5x105 B16/F10 tumor cells or

5x105 B16/F10 cells admixed with 1 x105 primary cultured FRDDP in their left and right flank, respectively. Two weeks after injection, tumors were dissected and weighted. Results are expressed as mean tumor weights +/- SEM obtained in two independent experiments (Figure 10). The same experiment was repeated using 2x103 freshly isolated FRCDP and FRCDN (see Material and Methods section of the manuscript) admixed with 5x105 B16/F10 tumor cells and engrafted s.c. in the left and right flank of the six mice, respectively. Mice engrafted with B16xF10 tumor cells alone were used as control. Two weeks after injection, tumors were dissected, weighted and results were expressed as mean tumor weights +/- SEM (Figure 1 1 ).

Statistical analysis

Statistical significance was determined with non parametric Mann-Withney U test or one way ANOVA with Bonferroni post-test using the Prism software. Example 2: Thrombomodulin is a new marker of Fibroblastic Reticular Cells expressing JAM-C

In order to reveal heterogeneity between JAM-C expressing cells of lymph nodes, the inventors generated a panel of monoclonal antibodies (mAbs) directed against surface markers of the sinus derived JAM-C expressing cell line, LyEnd-5. The mAbs were tested by flow cytometry for their ability to recognize the LyEnd-5 cells but not the JAM-C expressing vascular cell line, bEnd-2. Among 408 hybridomas, only 1 1 % of the reactivities were specifically directed against the LyEnd-5 cell line. Among these, the monoclonal antibody (mAb) LS17-9 was selected for further studies and found to be specifically directed against mouse thrombomodulin, as shown in Figure. 1 A-D. The Thbd expression profile on LN sections was then studied by immunohistochemistry (IHC) and compared with JAM-C. It was found that Thbd was not expressed by HEVs expressing JAM-C and partially colocalized with JAM-C on fibroblastic stromal cells of LNs (data not shown). Similarly, a partial colocalization with Pecam-1 and Lyve-1 was observed indicating that Thbd expression on endothelial and lymphatic cells is heterogeneous.

To determine if fibroblastic expression of Thbd was associated to chemokine secreting cells in the T cell area, LS17-9 staining was combined with CCL21 . It was found a close association between CCL21 and Thbd expression indicating that Thbd is most likely expressed by FRCs. To address this issue, LN sections of chimeric irradiated ubiquitin promoter-GFP transgenic animals reconstituted with wild-type bone marrow cells were stained with the LS17.9 mAb. In these mice, the GFP-expressing cells in the T cell paracortex are FRCs, as described by Bajenoff et al. (Immunity, 25: 989-1001 , 2006). Some GFP-expressing cells in the T cell area were stained with LS17.9 mAb demonstrating that Thbd is a new marker of FRC heterogeneity.

To refine the cellular distribution of Thbd and JAM-C on LN stromal cells, a collagenase-based protocol and flow cytometry have been used. As expected from IHC results, Thbd expression was restricted to the non-hematopoietic CD45^Neg compartment and was heterogeneously expressed by endothelial and lymphatic cells expressing PECAM-1 and Lyve-1 respectively, as shown in Figure 2 When gated on the fibroblastic compartment CD45^NegPecam-1 ^NegLyve-1 ^Neg and combined with the FRC marker PDGFRoc, it was found that Thbd divided FRCs into three subsets: one minor population expressing intermediate levels of PDGFRoc and Thbd, and two major subsets: double negative (PDGFRoc^NegThbd^Neg)and double positive (PDGFRoc^PosThbd^Pos) subsets referred as FRCs^DN and FRCs^DP in the remaining of the manuscript (Fig. 3 top right panel). Expression of JAM-C was restricted to FRCs^DP (Fig. 3 lower panels).

Example 3: FRCs^DP support naive T cell adhesion and secrete homeostatic chemokines

To investigate the function of FRCs, primary stromal cells from LNs were isolated by long-term culturing primary adherent cells. All isolated cell lines had a typical fibroblastic morphology and four out of five cell lines expressed PDGFRoc, Thbd and JAM-C which was the expected phenotype for FRCs^DP (see Figure 4).

Consistent with previous reports (see Katakai et at., The Journal of experimental medicine, 200: 783-795, 2004; and Farr et al., The Journal of experimental medicine, 176: 1477-1482, 1992), they all expressed high levels of VCAM-1 and gp38 and variable levels of ICAM-1 (see Figure 4).

FRCs^DP were then tested for their ability to support naive T cell adhesion and it was found that 22% of naive T lymphocytes adhered to the stromal cells whereas the adhesion level of activated T cells was only 4% (see Figure 5.A). Adhesion was reduced by 30% when naive T cells were pre-treated with pertussis toxin suggesting the involvement of cytokine secretion in the preferential adhesion of na^'ive T cells to FRC^DP. This was consistent with the twofold increase in naive T cell transmigration toward stromal cell conditioned medium as compared to control medium (see Figure 5.B) suggesting that FRCs^DP secrete homeostatic chemokines acting on naive T cells. FRC^DP and FRC^DN cells were then freshly isolated by cell sorting and tested by quantitative reverse PCR for SDF- 1 a, CCL19 and CCL21 mRNA levels. The transcripts encoding the homeostatic chemokines SDF-1 oc, CCL19 and CCL21 were highly enriched in the FRC^DP fraction as compared to the FRC^DN cells (see Figure 5.C).

In agreement with the sorting strategy, transcripts encoding JAM-C and PDGFRoc were enriched in FRC^DP cells in contrast to the ocSMA transcript (see Figure 6), which was significantly enriched in the minor subset of FRCs expressing intermediate levels of Thbd and PDGFRoc (not shown).

This shows that Thbd and PDGFRoc are two valuable markers to isolate FRC^DP secreting homeostatic chemokines from LNs.

Example 4: JAM-C controls FRC^DP chemokine secretion in vitro and in vivo

In order to explore the mechanisms controlling chemokine secretion in FRCs^DP, homeostatic chemokine secretion was assessed in primary stromal cells. Constitutive secretion of SDF1 a was detected and significantly reduced when expression of JAM-C was silenced by means of two different siRNA (purchased from Dharmacon, siRNAs J- 044688-09 and J-044688-10) which did not affect stromal cell growth or survival (see Figure 7.A-C).

To extend our findings to the physiological function of JAM-C expression in LNs, intranodal homeostatic chemokine content was determined in mice treated with monoclonal antibodies directed against JAM-C. The intranodal chemokine content of treated mice was significantly reduced, ranging from 25-30% reduction for SDF-1 oc to 50% reduction for CCL21 (Fig. 8).

These results indicate that mRNA levels encoding homeostatic chemokines in FRC^DP are specifically targeted by anti-JAM-C treatment. This was further confirmed by in situ staining for CCL21 on lymph node sections obtained from control and anti-JAM-C treated mice (data not shown) and correlated with the significant reduction in intranodal SDF-l oc, CCL19 and CCL21 contents observed in Jam-C deficient mice (Fig. 7.D et 8).

Example 5: Tumor-associated stromal cells expressing both thrombomodulin and JAM-C promote tumor growth

Because tumor-associated stromal cells are suspected to produce many of the growth signals promoting the proliferation of tumor, the inventors then hypothesized that a stromal cell subpopulation with a phenotype similar to FRC^DP would exist in certain solid tumors and would be responsible of tumor growth due to chemokines secretion.

Hence, the inventors looked to see if FRC^DP had the property to promote tumor growth. For this purpose, FRC^DP and FRC^DN were sorted and admixed with different tumor cell lines before engraftment.

Tumor mass was measured after two to three weeks depending on the tumor cell line used.

Using the B16F10 tumor cell line, it was found a two to three fold increase in syngenic tumor growth when FRC^DP were used as compared to FRC^DN and to the absence of FRC cells, as illustrated in Figures 9 and 10.

These results indicate that stromal cells expressing Thrombomodulin have the unique property to secrete chemokines and to promote tumor growth in vivo.

The inventors have previously shown that anti-JAM-C treatment inhibits tumor growth most likely by inhibiting tumoral angiogenesis (Lamagna et al, Cancer Research, 2005). However, the step at which angiogenesis was inhibited was not precisely defined. Therefore, the inventors tested if thrombomodulin expressing cells were the target of anti- JAM-C treatment.

To this end, mice engrafted with FRC^DP / tumor cells mixture were treated with anti-

JAM-C or with an antibody that does not target FRC^DP cells (isotype control 9B5). After X days of treatment, tumor samples were harvested and sections of isotype control or anti- JAM-C treated mice were stained for thrombomodulin and PECAM, a vascular specific marker (Fig. 1 1 ). The quantification of staining showed a dramatic decrease in thombomodulin expression (-55%), whereas PECAM was only reduced by 25%. These results indicate that anti-JAM-C treatment is more specifically directed against thrombomodulin expressing cells than endothelial cells in tumor microenvironment.

Altogether, these data clearly show that the anti-tumor effects of anti-JAM-C treatment previously disclosed not only targets endothelial cells of the vessels irrigating solid tumor as suggested by Lamagna et al. (Cancer Research, 65(13): 5703-5710, 2005), but also targets a subset of tumor-associated stromal cells which promote tumor growth by secreting chemokines. Similar results can be achieved by treating tumor engrafted mice with an anti-thrombomodulin antibody, in particular LS17-9 mAb, instead of an anti- JAM-C antibody.

Claims

1 . An in vitro method of prognosing outcome of a cancer in a patient afflicted with a solid tumor, said method comprising:

a) measuring the expression level of thrombomodulin in a biological sample from a solid tumor from said patient comprising at least tumor-associated stromal cells ; and

b) optionally deducing from the result of step a) the prognosis of said patient.

2. The method of claim 1 , wherein step a) of measuring the expression level of thrombomodulin is carried out on tumor-associated stromal cells from the biological sample.

3. The method of claim 1 or 2, wherein expression of thrombomodulin is indicative of a poor prognosis, the higher the expression level of thrombomodulin is, the poorer the prognosis is.

4. An in vitro method for selecting a patient afflicted with a solid tumor suitable to be treated with a compound which inhibits chemokines expression by, and/or induces death of, tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C, comprising:

a) measuring the expression level of thrombomodulin in a biological sample from a solid tumor from said patient, said biological sample comprising at least tumor-associated stromal cells ;

b) selecting the patient expressing thrombomodulin as suitable to be treated with a compound which inhibit chemokines expression by, and/or induces death of, tumor- associated stromal cells expressing both thrombomodulin and protein JAM-C.

5. The method of claim 4, wherein the compound which inhibits chemokines expression by, and/or induces death of, tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C is selected from the group consisting of: an anti- thrombomodulin antibody, an anti-JAM-C antibody, an aptamer which specifically binds to thrombomodulin and which is conjugated with a cytotoxic molecule, and an aptamer which specifically binds to JAM-C and which is conjugated with a cytotoxic molecule.

6. An in vitro method for screening for a compound suitable for the treatment of a solid tumor comprising tumor-associated stromal cells which express both thrombomodulin and protein JAM-C, the method comprising the steps of:

a) bringing into contact a candidate compound with stromal cells expressing both thrombomodulin and protein JAM-C ;

b) determining whether said candidate compound inhibits chemokines expression by, and/or induces death of, tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C;

wherein the candidate compound which inhibits chemokines expression by, and/or induces death of, tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C is identified as a compound suitable for the treatment of a solid tumor comprising tumor-associated stromal cells which express both thrombomodulin and protein JAM-C.

7. A compound which inhibits secretion of chemokines by, and/or induces death of, tumor- associated stromal cells expressing thrombomodulin for use as a medicament in the treatment of solid tumor, wherein the solid tumor comprises tumor-associated stromal cells which express both thrombomodulin and protein JAM-C

8. A compound which inhibits and/or induces death of stromal cells expressing thrombomodulin according to claim 7, wherein said compound is selected from the group consisting of an anti-thrombomodulin antibody, an anti-JAM-C antibody, an aptamer which specifically binds to thrombomodulin and which is conjugated with a cytotoxic molecule, and an aptamer which specifically binds to JAM-C and which is conjugated with a cytotoxic molecule.

9. An in vitro method for obtaining chemokines secreting stromal cells comprising a step consisting of extracting stromal cells expressing both thrombomodulin and protein JAM-C from a biological sample in which stromal cells expressing both thrombomodulin and protein JAM-C are present.

10. An in vitro method for obtaining stromal cells secreting chemokines according to claim 9, wherein the biological sample in which stromal cells expressing both thrombomodulin and protein JAM-C are present is a lymph node tissue, and wherein the stromal cells secreting chemokines extracted are Fibroblastic Reticular cells expressing both thrombomodulin and protein JAM-C

1 1 . An in vitro method for obtaining stromal cells secreting chemokines according to claim 9, wherein the biological sample in which stromal cells expressing both thrombomodulin and protein JAM-C are present is a solid tumor tissue which comprises tumor-associated stromal cells expressing both thrombomodulin and protein JAM-C.

12. An isolated secreting chemokines stromal cell expressing both thrombomodulin and protein JAM-C.

13. An isolated secreting chemokines stromal cell expressing both thrombomodulin and protein JAM-C according to claim 12, wherein said cell is obtainable by an in vitro method as defined in any of claims 9 to 1 1 .

14. A kit comprising:

a) means for detecting the amount and/or expression level of thrombomodulin; and b) optionally, instructions for the use of said kit in prognosing solid tumor.