A Role for Stroma-Derived Annexin A1 as Mediator in the Control of Genetic Susceptibility to T-Cell Lymphoblastic Malignancies through Prostaglandin E₂ Secretion

Cancer susceptibility is essentially attributable to multiple low-penetrance genes. Using interspecific consomic and congenic mice between the tumor-resistant SEG/Pas and the tumor-sensitive C57BL/6J strains, a region on chromosome 19 involved in the genetic resistance to γ-irradiation–induced T-cell lymphomas (Tlyr1) has been identified. Through the development of nonoverlapping subcongenic strains, it has been further shown that Anxa1 may be a candidate resistance gene on the basis of its differential expression in thymus stroma cells after γ-radiation exposure. In addition, thymus stroma cells of thymic lymphomas exhibited a significant reduction in the expression levels of Anxa1. Interestingly, the activity of Anxa1 relies on prostaglandin E₂ (PGE₂) induction that brings about apoptosis in thymocytes. In fact, in vitro transfection experiments revealed that PGE₂ production was enhanced when HEK 293 cells were transfected with full-length cDNAs of Anxa1, with PGE₂ production in the cells transfected with the allele of the resistant strain (Anxa1^Tyr) being higher than that in cells transfected with the allele of the susceptible strain (Anxa1^Phe). Furthermore, the presence of this compound in the medium induced apoptosis of immature CD4⁺CD8⁺CD3^low cells in a dose-dependent manner. These results improve our knowledge of the molecular mechanisms triggering T-cell lymphoblastic lymphoma development while highlighting the relevance of the stroma in controlling genetic susceptibility and the use of PGE₂ as a new therapeutic approach in T-cell hematologic malignancies. [Cancer Res 2009;69(6):2577–87]

There is convincing evidence from twin studies that the risk of cancer in humans has a strong genetic component (1). Whereas rare highly penetrant germ-line mutations in tumor suppressor genes strongly predispose to familial forms of cancer, it is also known that a large proportion of cancer risk is attributable to multiple common alleles of minor susceptibility genes, sometimes referred to as low-penetrance genes (2). For such reason, the identification of genes capable of modifying individual tumor susceptibility is a main issue for understanding the genetic basis of cancer predisposition.

The detection of genetic variants with weak phenotype effects in humans is difficult because cancer susceptibility does not in general segregate as single Mendelian traits. The genetic heterogeneity and variable etiology of carcinogenesis in humans adds a further difficulty to this challenge. On the other hand, mouse models, and in particular those making use of congenic strains, have been shown to have a bearing on the identification of cancer susceptibility genes (3, 4). Although more than 100 cancer susceptibility loci have been mapped in mice (5), relatively few of these have been translated into specific genes (6–10).

Genetic loci controlling susceptibility to ionizing radiation–induced mouse thymic lymphomas have previously been mapped at chromosomes 4, 5, and 16 (refs. 11–13). A study of the γ-radiation–induced thymic lymphoma (RITL) predisposition with the help of interspecific consomic strains led us to identify a tumor resistance locus on mouse chromosome 19 (Tlyr1; ref. 14). By analyzing congenic mice derived from interspecific consomic strains for chromosome 19, we mapped Tlyr1 to a region of ∼14 Mb flanked by D19Mit85 and D19Mit13 microsatellite markers. Tlyr1 lies adjacent to another distal RITL susceptibility region (15) and excludes the candidacy of some genes like Pten (16) and Fas (17, 18) for which a role in RITL predisposition has been already proposed.

In this article, we focus on Tlyr1, emphasizing the identification of putative candidate genes through the generation of nonoverlapping subcongenic strains. We examined the connections between the different subcongenic strains and thymic lymphoma resistance and explored the candidacy of specific genes through the analysis of their expression patterns in an early response to γ-irradiation and during RITL development. We found that RITL resistance is clearly influenced by the levels of expression in thymus stroma cells of the gene encoding the annexin A1 protein (Anxa1) that is located at the proximal region of Tlyr1.

Mice. C57BL/6J mice were purchased from The Jackson Laboratory. Mice congenics for SEG/Pas (Mus spretus) in the critical Tlyr1 region (named as B6.Tlyr1^SEG) were generated as we described elsewhere (19). Animal experiments were carried out according to the European Commission Guidelines (Directive 86/609/CEE) on the use of laboratory animals.

Genotyping of subcongenic mice. Mice were genotyped for 14 microsatellite markers arranged along the Tlyr1 region. These markers were amplified using primers and PCR conditions obtained from the Mouse Genome Database.⁷

Thymic lymphoma induction. Mice were exposed to whole-body fractionated (4 × 1.75 Gy) γ-irradiation at weekly doses, starting at 4 to 5 wk of age. Treated mice were observed at weekly intervals beginning 12 wk after completion of γ-irradiation treatment and up to 25 wk (the latency period for these tumors) and phenotyped for the presence or absence of thymic lymphomas as we previously described (14).

Thymic cell fractionation. Thymus samples were mechanically dispersed and strained through a nylon mesh (BD Biosciences) to isolate the thymocytes. Stroma-enriched cell fractions were then obtained by collagenase digestion as previously described (20). Stromal cells (CD45⁻) were afterward isolated by immunomagnetic separation using anti-CD45–conjugated paramagnetic microbeads, following the manufacturer's instructions (Miltenyi Biotech).

Standard reverse transcription-PCR. Transcriptional expression of the genes mapped at the Tlyr1a region was carried out by conventional reverse transcription-PCR (RT-PCR). The SuperScript First-Strand Synthesis System (Invitrogen) was used to perform RT-PCR reactions. The gene encoding the glucose-6-phosphate-1-dehydrogenase (G6pd) was used as an internal control. Supplementary Table S1 shows the primer sequences used. The primers were designed by TibMolBiol. RT-PCR products were visualized by direct ethidium bromide staining in 1.5% agarose gels.

DNA sequencing. RNA was extracted from thymuses using TriPure Reagent (Roche). Based on the Anxa1 cDNA sequence,⁸

a couple of primers were designed: 5′-CTCTAAAAATGGCAATGGTATCAG-3′ (forward) and 5′-TTGCAGAATAGTTGGGATGT-3′ (reverse). Reverse transcription was done using the Superscript First-Strand Synthesis System (Invitrogen), followed by PCR using the Expand High Fidelity PCR System (Roche) and the aforementioned primers. cDNA sequencing reactions were done on an ABI Prism 310 Automated Sequencer (Applied Biosystems).

Quantitative real-time RT-PCR. The quantification of the transcriptional levels of thymus-expressing genes was done by real-time RT-PCR with a LightCycler instrument (Roche). RT-PCR reactions were carried out in total RNA using the one-step LightCycler SYBR Green I kit (Roche). The primers used were those described in Supplementary Table S1. Relative expression values were calculated as the mRNA amount of each gene relative to that of G6pd (used as reference) and normalized to the relative expression of a nontreated thymus sample, using the LightCycler Relative Quantification software (Roche).

Loss-of-heterozygosity analysis of Anxa1. PCR/single-strand conformational polymorphism analysis for loss of heterozygosity (LOH) detection was done using the primers 5′-AAGGAGAAAGGGGACAGACG-3′ (forward) and 5′-AATAAAGGAACAGCATCGCC-3′ (reverse). After PCR, samples were mixed (1:1) with loading buffer containing 0.08 N NaOH and 95% formamide. Samples were heated at 95°C for 10 min and subjected to 0.5× mutation detection enhancement polyacrylamide gels (BioWhittaker) in 0.6× Tris-borate EDTA buffer. After electrophoresis, DNA fragments were silver stained using a standard protocol.

Western blotting. Proteins were extracted from cell lysates using TriPure Reagent (Roche), separated on 8% to 12% SDS-PAGE, and electrotransferred onto nitrocellulose membranes (Bio-Rad). They were then incubated with anti-ANXA1 antibody (R&D Systems) at 1:5,000 dilution. Detection of β-actin with a monoclonal anti–β-actin antibody (Sigma) at 1:10,000 dilution was used as control. Next, an incubation with secondary antibodies coupled to horseradish peroxidase was carried out with a donkey anti-goat antibody for ANXA1 (Santa Cruz Biotechnology) and a sheep anti-mouse antibody for β-actin (GE Healthcare) at 1:1,000 dilution. Bands were detected using the ECL Western Blotting Detection kit (GE Healthcare). Protein levels were densitometrically quantified using the Scion Image program (Scion Corp.) and calculated as the amount of ANXA1 protein relative to that of β-actin. These were then normalized to the amount of ANXA1 in a nontreated thymus sample.

Cloning of Anxa1 cDNA sequences and transient transfections. Purified DNA fragments containing the Anxa1 full-length cDNA were doubly digested with HindIII and BamHI and cloned into pcDNA3 plasmid (Invitrogen). The primers used for cloning were (generated restriction sites underlined) for Anxa1HindIII, 5′-AAGCTTCTCTAAAAATGGCAATGGTATCAG-3′ (forward), and for Anxa1BamHI, 5′-GGATCCTTGCAGAATAGTTGGGATGT-3′ (reverse). Using the Expand High Fidelity PCR System (Roche) and the aforementioned primers, a fragment of 1,081 bp, which includes the entire coding sequence, was amplified. Transfections were done in human embryonic kidney (HEK) 293 cells using Lipofectamine (Invitrogen). HEK 293 cells were cultured in DMEM, supplemented with 10% fetal bovine serum (FBS), 2 mmol/L l-glutamine (all from Invitrogen), 0.1 mg/mL ampicillin (Roche), and 64 μg/mL gentamicin (Sigma). The mouse thymic epithelial cell line 427 (kindly provided by Dr. Barbara Knowles, The Jackson Laboratory, Bar Harbor, ME) was used as control in Western blots to identify the mouse form of ANXA1 (21).

Determination of prostaglandin E₂ production by transfected cells. PGE₂ was determined in cell culture supernatants by an ELISA assay using the ACE competitive EIA kit (Cayman Chemical). In this assay, the target (PGE₂) competes with a PGE₂-acetylcholinesterase (ACE) conjugate (PGE₂ tracer) for a limited amount of PGE₂ monoclonal antibody attached to the well. Both HEK 293 cells and supernatants were recovered 24 h after transfection. PGE₂ concentrations were determined spectrophotometrically by measuring the amount of PGE₂ tracer bound to the well, which is inversely proportional to the amount of free PGE₂.

T-cell cultures and PGE₂ treatments. Freshly isolated mouse thymocytes and mouse Thy278 T cells (kindly provided by Dr. Ingo Schmitz, University of Düsseldorf, Düsseldorf, Germany) were grown in RPMI 1640 supplemented with 10% FBS, 100 units/mL penicillin, 100 μg/mL streptomycin, 2 mmol/L l-glutamine (all from Invitrogen), 1 mmol/L sodium pyruvate, 1% nonessential amino acids (both from BioWhittaker), and 0.05 mmol/L 2-mercaptoethanol (Merck). Human Karpas-45, Peer, and Jurkat T cells were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ) and cultured as we described elsewhere (17). PGE₂ (Cayman Chemical) was added to the culture medium at the time of plating. Serial dilutions of PGE₂, ranging from 0.01 to 100 μmol/L, were used. Cells were harvested at 8, 16, and 24 h after treatment.

Immunofluorescence staining and flow cytometry analysis. Freshly isolated thymocytes from nontreated and 1.75 Gy–treated thymuses were examined by immunofluorescence staining and flow cytometry analysis. We used FITC-conjugated rat anti-mouse CD4 and phycoerythrin-conjugated rat anti-mouse CD8a (both from BD Pharmingen) for a two-color flow analysis of CD4- and CD8-positive cells on a Coulter Epics XL-MCL flow cytometer (Beckman Coulter). In addition, biotin-conjugated rat anti-mouse CD4, phycoerythrin-conjugated rat anti-mouse CD8a, and FITC-conjugated hamster anti-mouse CD3e (all from BD Pharmingen) were used for a three-color flow analysis on a FACSCalibur flow cytometer (BD Biosciences). All antibodies were used at 1:100 dilution.

Quantification of apoptosis by terminal deoxyribonucleotidyl transferase–mediated dUTP nick end labeling assay. Cells were fixed in 3.7% formaldehyde in PBS for 20 min at 4°C and then permeabilized in 0.1% Triton X-100 sodium citrate for 5 min at 4°C. Staining was done in suspension cells by terminal deoxyribonucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) assay using a commercially available kit (Roche). The percentage of TUNEL-positive cells was determined using either a Coulter Epics XL-MCL flow cytometer or a FACSCalibur flow cytometer.

Cell cycle analysis by propidium iodide flow cytometry. Cells were harvested and fixed in ice-cold 70% ethanol. On staining, cells were suspended in PBS containing 50 μg/mL RNase A (Sigma) for 15 min at room temperature. Last, propidium iodide (Sigma) staining solution at a final concentration of 50 μg/mL was added. The DNA content was estimated using a Coulter Epics XL-MCL flow cytometer or a FACSCalibur flow cytometer.

Statistical analysis. The differences in RITL incidence between groups were examined for statistical significance by χ² test. P < 0.0125 [Bonferroni correction (α/n), with α= 0.05 and n (number of comparisons) = 4] was considered significant. The Kolgomorov-Smirnov test was used to test expression data sets for normality, and the Levene test was used for homogeneity of variances. For multiple comparisons, statistical significance was determined using a one-way ANOVA analysis with a Bonferroni comparison post-test. All statistical tests were carried out using the SPSS software (version 14.0).

High-resolution mapping of Tlyr1. A high-resolution genetic map of Tlyr1 was constructed to aid in the identification of candidate RITL resistance genes for this region. Initially, we worked on the selection of nested recombinant haplotypes (NRH) for Tlyr1 by backcrossing male mice heterozygous congenics for SEG/Pas (M. spretus) in this region [named as (B6.Tlyr1^SEG × B6) F1] with females of the susceptible C57BL/6J strain, followed by a final intercross to reach homozygosity. We generated three different subcongenic strains (hereafter named as NRH1, NRH2, and NRH3) covering the entire Tlyr1 region, each of them carrying a nonoverlapping SEG/Pas chromosome segment (Tlyr1a, Tlyr1b, and Tlyr1c; Fig. 1A). To analyze RITL incidence, mice of congenic B6.Tlyr1^SEG and subcongenic NRH strains, as well as C57BL/6J mice as reference, were subjected to fractionated sublethal whole body γ-irradiation (4 × 1.75 Gy). To exclude the well-documented sex difference in RITL development (13), only treated females were analyzed. Although all of the three NRH regions contributed to a reduction of RITL incidence, the association was only conclusive for NRH1 (Fig. 1A). These data allowed us to restrict our search to Tlyr1a, a SEG/Pas chromosome segment of ∼3.2 Mb delimited by the D19Mit41 and D19Mit96 markers.

The gene encoding the annexin A1 protein is a putative candidate gene for Tlyr1. Once Tlyr1a was selected, an in silico analysis² indicated nine annotated genes mapping to this region. Later on, we characterized the thymus expression specificity of all these genes by RT-PCR. Only six of these genes showed expression in the thymus (Fig. 1B). To find out whether there existed a pattern of differential gene expression in an early response to 1.75 Gy of γ-irradiation, the transcriptional expression of these genes was determined by quantitative real-time RT-PCR in thymuses from NRH1 and C57BL/6J mice. Only the Anxa1 gene was shown to exhibit a significant differential expression (Fig. 1C). Apart from quantitative differences, the comparative analysis of the coding sequence of Anxa1 between the NRH1 and the C57BL/6J strains evidenced a nonconservative T689A single nucleotide polymorphism (GenBank accession no. EU684130) that resulted in a Tyr230Phe variation in the ANXA1 protein, which might reflect a different biological activity of both alleles (Anxa1^Tyr in NRH1 and Anxa1^Phe in C57BL/6J). The accumulation of qualitative and quantitative differences clearly indicates that Anxa1 could be a good candidate gene for the Tlyr1 region.

There is a higher expression of Anxa1 in the thymus stroma of NRH1 mice in early response to a single sublethal γ-ray dose. Because the thymus is a complex organ incorporating thymocytes and stromal cells, we analyzed the allele expression profiles of this gene in separate thymus stroma cells and thymocytes. Notably, in nontreated mice, the Anxa1 gene was expressed in thymus stroma cells, whereas it was weakly perceptible in thymocytes (Table 1). In addition, treatments with a single sublethal dose (1.75 Gy) induced higher RNA levels of Anxa1 in thymus stroma cells from NRH1 mice as compared with C57BL/6J mice (Table 1). These results were confirmed at the protein level (Table 1; Supplementary Fig. S1).

The expression of Anxa1 is also altered in the thymic stroma of T-cell lymphoblastic lymphoma–bearing mice. Exposure of mice to fractionated 1.75-Gy doses of γ-irradiation elicits thymic lymphomas, in particular T-cell lymphoblastic lymphomas, which are characterized by uncontrolled expansion of immature thymocytes (22). To get further insight into the role of Anxa1 in RITL development, we examined the expression of this gene in the two cell fractions from thymuses of treated mice that developed thymic lymphoma (RITL-bearing mice; in all cases T-cell lymphoblastic lymphomas) as well as from thymuses of treated mice that did not develop thymic lymphoma after the latency period. Irrespective of their genotype, thymus stroma cells of RITL-bearing mice exhibited a significant reduction in the mRNA levels of Anxa1 compared with the values observed in thymus stroma cells from nontreated control mice (Table 1). By contrast, no differences were observed in thymus stroma cells from irradiated mice that did not develop thymic lymphomas after the latency period. The amount of ANXA1 protein was also found to be significantly reduced in thymus stroma cells from all the RITL-bearing mice analyzed (Table 1; Supplementary Fig. S1).

It should be stressed that sequencing analysis of Anxa1 done on its cDNA did not reveal any mutation in the tumor samples (data not shown). However, LOH studies with this gene revealed frequent allele losses in tumors of NRH1 heterozygous mice (10 of 22, 45.4%), indicating that the reduction of Anxa1 expression found in thymic lymphomas could be attributed, at least in part, to the deletion of one allele (Supplementary Fig. S2).

γ-Irradiation results in differential G₁ arrest and apoptosis in thymocytes of NRH1 mice. Given that ionizing irradiation results in cell cycle arrest and apoptosis induction, we wondered whether different levels of Anxa1 expression, detected in thymus stroma cells of thymuses from 1.75 Gy–treated NRH1 and C57BL/6J mice, could be involved in causing differences in the cell cycle distribution and/or the apoptotic response of their respective thymocyte fractions. To this end, we studied the progression through the cell cycle and the apoptosis rate in freshly isolated thymocytes from these mice collected 24 hours after γ-ray treatment. Cell cycle was analyzed by flow cytometry following the staining of cells with propidium iodide (Fig. 2A). We detected a significant increase of propidium iodide–positive NRH1 thymocytes at the G₁ phase by comparison with those of the C57BL/6J, evidencing a differential effect of γ-irradiation in G₁ arrest between the two strains. Interestingly, this G₁ arrest was also accompanied by a significant increase in the sub-G₁ population of NRH1 thymocytes compared with that of the C57BL/6J strain.

Next, we studied the induction of apoptosis by TUNEL assay and flow cytometry (Fig. 2B). The treatment with 1.75 Gy significantly induced more thymocyte apoptosis in NRH1 than in C57BL/6J mice. The percentage of TUNEL-positive cells detected in γ-radiation–treated thymocytes was closely similar to sub-G₁ cell percentages in the same samples (Fig. 2A and B), indicating that the sub-G₁ fraction may basically consist of apoptotic thymocytes, as pointed out elsewhere (23). Interestingly, the Anxa1 expression levels in thymus stroma cells correlated positively either with the rate of thymocyte arrest at the G₁ phase of the cell cycle (R² = 0.948, P = 0.018) or with thymocyte apoptosis (R² = 0.963, P = 0.013).

It is known that immature thymocytes, particularly CD4⁺CD8⁺ cells, readily undergo apoptosis in response to γ-irradiation (24). If Anxa1 expression and thymocyte apoptosis are two events positively correlated, then increased expression levels of Anxa1 should specifically result in a reduction of the number of CD4⁺CD8⁺ cells. We found a significant higher depletion of 1.75 Gy γ-radiation–treated CD4⁺CD8⁺ thymocytes in NRH1 mice as compared with C57BL/6J (Fig. 2C and D), which might reflect the different levels of Anxa1 expression existing in the thymus stroma cells of both strains (R² = 0.956, P = 0.02).

The Anxa1^Tyr allele induces a higher production of PGE₂ in transfected cells. Because thymic epithelial cells transfected with Anxa1 cDNA increase the production of PGE₂ (25), and this molecule is able to induce apoptosis of immature CD4⁺CD8⁺ thymocytes (26), we performed in vitro experiments using human HEK 293 cells transfected with an expression vector (pcDNA3) containing the full-length cDNA of mouse Anxa1 in either the Anxa1^Tyr or Anxa1^Phe allelic variant. We used these cells for three main reasons: First, they produce very low levels of endogenous ANXA1 protein. Second, the human form of ANXA1 can be easily distinguished from the mouse one. Finally, HEK 293 cells are extremely easy to culture and transfect. The functionality of each allele was determined by quantifying the amount of extracellular PGE₂ produced by the transfected HEK 293 cells in a competitive ELISA assay. The production of ANXA1 protein by the transfected cells was examined by Western blot analysis as a quality control measure of transfection. As shown in Fig. 3A, the amount of ANXA1 protein increased ∼2.5-fold in the ANXA1-overexpressing cells with no significant differences between the two types of Anxa1 transfectants. However, PGE₂ production in Anxa1^Tyr-transfected cells was significantly higher (∼1.3-fold increase) than in Anxa1^Phe-transfected cells (Fig. 3B). Because the amounts of ANXA1 protein were kept constant, the differences in the production of PGE₂ may be attributed to a distinct biological activity of the two Anxa1 alleles.

PGE₂ induces apoptosis in immature CD4⁺CD8⁺CD3^low thymocytes. To find out whether a causal link exists between the amount of PGE₂ and the ability of primary thymocytes to undergo apoptosis, we first analyzed the apoptotic response of primary C57BL/6J thymocytes in vitro by TUNEL assays after being subjected to serial dilutions of PGE₂ ranging from 0.01 to 100 μmol/L. To select the most appropriate conditions, we monitored this process at 8, 16, and 24 hours after treatment. The results showed that PGE₂-treated C57BL/6J thymocytes significantly underwent more apoptosis than those in nontreated cultures, these differences being highest at 16 hours (Supplementary Fig. S3). The effect of PGE₂ on thymocyte apoptosis became significantly evident at a concentration as low as 0.01 μmol/L, was similar at 0.1 and 10 μmol/L, and was much higher at 100 μmol/L (Fig. 4A). Our data also revealed that the administration of a given dose of PGE₂ in the medium produced similar induction of apoptosis in primary NRH1 thymocytes (data not shown). Next, we investigated which thymocyte subpopulations were affected by exposure to PGE₂. To this end, cultures of primary C57BL/6J thymocytes were treated during 16 hours with PGE₂ and then analyzed by flow cytometry to determine the expression of CD4, CD8, and CD3 cell surface antigens. PGE₂ administration triggered a selective loss of immature CD4⁺CD8⁺CD3^low thymocytes (Fig. 4B). Because it had been suggested that PGE₂ might also interfere in the process of lymphoid cells by mediating growth arrest (27), we also explored this possibility by determining the number of thymocytes at different phases of the cell cycle. Compared with the control, PGE₂-treated primary thymocytes did not show changes in cell cycle distribution, but they did exhibit statistically significant increases in the number of cells at the sub-G₁ region in a dose-dependent manner (Fig. 4C).

To show whether the induction of apoptosis by PGE₂ was also functional in transformed thymocytes, we analyzed apoptosis events in the murine Thy278 thymic lymphoma cell line, consisting of immature CD4⁺CD8⁺ thymocytes (28). Significant apoptosis was only detected 24 hours after treatment with 100 μmol/L PGE₂ (Fig. 5A). Changes in cell cycle distribution of Thy278 cells were only evident after treatment with 100 μmol/L PGE₂, with a significant arrest of cells at the G₁ phase (Fig. 5B).

To investigate the ability of PGE₂ to induce apoptosis and cell cycle arrest in human T-lymphoma cells, we analyzed three T-acute lymphoblastic leukemia cell lines, which represent different stages of T-cell differentiation: Karpas-45 (CD4⁺CD8⁺CD3⁻), Peer (CD4⁺CD8⁺CD3⁺), and Jurkat (CD4⁺CD8⁻CD3⁺). Significant variations in the apoptotic rate were only observed in Karpas-45 cells 24 hours after treatment with 100 μmol/L PGE₂ (Fig. 5A). In this case, the induction of apoptosis was accompanied by a significant arrest of cells at G₂-M (Fig. 5B).

It is widely known that minor susceptibility genes are an essential component of heritability on individual resistance/susceptibility to cancer (2, 6). In this article, the use of NRH-subcongenic mice between a thymic lymphoma–susceptible strain (C57BL/6J) and a thymic lymphoma–resistant one (SEG/Pas) allowed us to identify a critical region on the proximal part of mouse chromosome 19 that confers significant resistance to γ-radiation–induced thymic lymphomas. Due to differential expression profiles in an early response to γ-irradiation and reduced expression in thymic lymphomas, we proposed Anxa1, whose locus maps to the critical region, as a possible thymic lymphoma resistance candidate gene (Fig. 1). Its candidacy was also supported by the fact that the allelic variants from susceptible and resistant mice encode protein variants involving functional domains of the ANXA1 protein.

ANXA1 has been described as a calcium and phospholipid-binding protein that participates in a variety of inflammatory pathways, in the control of cell proliferation, and in the regulation of death signaling (29), but the mechanisms through which this is accomplished in the thymus remain to be elucidated. It is known that thymocyte apoptosis is mediated by thymic epithelial cells, which constitute the major component of the so-called thymus stroma (30). We hereby report that Anxa1 is expressed exclusively by thymus stroma cells (Table 1) and that thymocytes from γ-radiation–treated NHR1 subcongenics and C57BL/6J mice undergo G₁ arrest and apoptosis in an Anxa1 dose–dependent manner (Fig. 2A and B). This is in agreement with previous findings evidencing that thymic microenvironmental factors can control thymocyte apoptosis from irradiated mice after preventive treatments for thymic lymphoma (31). Consistent with published studies (32, 33), we found that CD4⁺CD8⁺ cells are the most susceptible thymocytes to radiation-induced apoptosis. Interestingly, the degree of CD4⁺CD8⁺ cell reduction differs between the NRH1 and C57BL/6J strains (Fig. 2C). Because the more resistant strain (NHR1) is the one exhibiting the highest levels of Anxa1 expression and undergoing the highest rate of cell cycle arrest and apoptosis, these results suggested a possible causal relationship between levels of Anxa1 expression, cell cycle arrest, and apoptotic induction. This led us to wonder about the mechanisms of cell cycle arrest and apoptosis induced by Anxa1.

Previous studies reported that ANXA1 may increase the production of PGE₂ (25). It is also known that PGE₂ is able to induce thymocyte apoptosis and may interfere with the proliferation of thymocytes by mediating growth arrest (26, 27). These premises led us to initially evaluate the ability of NRH1-Anxa1^Tyr and C57BL/6J-Anxa1^Phe alleles as potential inducers of PGE₂. In vitro transfection experiments revealed that PGE₂ production was, in fact, enhanced when epithelial HEK 293 cells were transfected with Anxa1 cDNA, and that the cDNA from the Anxa1^Tyr allele was able to induce a far higher amount of PGE₂ than that of Anxa1^Phe (Fig. 3). Thus, it seems reasonable to think that the sequence variations existing between the two alleles are responsible, at least in part, for the differences in PGE₂ production.

Having confirmed that Anxa1 was able to induce PGE₂ production, we next investigated whether primary thymocytes underwent cell cycle arrest and apoptosis as an outgrowth of differences in the production of PGE₂ by thymus stroma cells. Quantification of apoptosis in primary thymocytes revealed that these cells died on PGE₂ exposure at physiologic (0.01 μmol/L) and nonphysiologic (0.1–100 μmol/L) concentrations in a dose-dependent manner (Fig. 4A). In addition, the effect of these PGE₂ concentrations on apoptosis involved a selective and progressive reduction of CD3 low-expressing CD4⁺CD8⁺ thymocytes (Fig. 4B), suggesting a CD3-dependent rank of action of PGE₂-induced apoptosis. Our data confirm previous experiments reporting apoptosis induction in CD4⁺CD8⁺CD3^low thymocytes after an in vivo administration of PGE₂ in C57BL/6J mice (26). Because mouse CD4⁺CD8⁺ thymic lymphoma Thy278 cells express CD3 at intermediate levels (28) and human Karpas-45 are human CD4⁺CD8⁺CD3⁻ T-lymphoma cells (34), it is not surprising that physiologic concentrations of PGE₂ had no effect on these cells.

Our results are also in line with the existence of distinct apoptotic pathways operating alternatively during the development of double positive thymocytes. For example, mouse double-positive PD1.6 thymic lymphoma cells (CD4⁺CD8⁺CD3⁺) undergo thymic epithelial cell–derived glucocorticoid–mediated apoptosis (35), whereas Thy278 cells die by apoptosis on T-cell receptor/CD3 stimulation (27). We have recently shown that the Cd95/Cd95L system mediates induced apoptosis of immature thymocytes and thymic lymphoma cells after exposure to γ-irradiation (18).

At the same time, it is worth noting that PGE₂ treatment is able to induce G₁ arrest in the Thy278 cell line and G₂ arrest in Karpas-45 cells (Fig. 5B). These results indicate that the cellular response of transformed T cells to PGE₂ is quite complex and suggest the existence of distinct molecular mechanisms of cell cycle arrest in different cell lines. It has been reported that differences in the gene expression profile may be related to the activation of a specific cell cycle arrest (at G₁ or G₂) in human malignant B-lymphocyte lines after exposure to ionizing irradiation (36). The existence of a distinct gene status in the transformed Thy278 and Karpas-45 cells might help to explain the differences found in the cell cycle distribution. Further studies will be necessary to confirm this hypothesis.

As regard to cancer, the Anxa1 gene has been found to be frequently down-regulated or up-regulated in many human solid tumors (37–43), as well as in B-cell lymphomas (44). Remarkably, alterations of this gene were always restricted to the tumor cells per se, but this does not seem to be the case in RITL samples, where the expression of Anxa1 was altered only in the tumor-associated stroma cells (Fig. 2). Several studies point to the engagement of the tumor-associated stroma in the development of certain solid tumors (45–47). A recent work pointed out the importance of bone marrow stromal cells to prevent the apoptosis of lymphoma cells in non-Hodgkin's lymphoma (48). In mice, we have shown that γ-irradiation is able to induce T-cell lymphoblastic lymphomas in the setting of a gene-altered thymic microenvironment (49). The results reported here highlight the engagement of the stroma in the development of T-cell hematologic malignancies. The contribution of tumor stroma sensitivity as a determinant of radiation-induced tumor growth delay has been reported elsewhere (50). From our results, it seems reasonable to postulate that the expression of Anxa1 below a certain threshold level (as detected in RITL samples) could be favoring T-cell lymphomagenesis, and that its overexpression (as detected in response to a single sublethal dose of γ-rays) might have a protective role over surrounding thymocytes, enhancing their apoptotic signals.

In summary, we show here that ANXA1, a protein previously characterized as a mediator of the anti-inflammatory activity of glucocorticoids in the host defense system, might play a critical role in the induction of apoptosis of immature CD4⁺CD8⁺ thymocytes through the production of PGE₂ by thymus stroma cells. These results expand our understanding of the mechanisms underlying the genetic susceptibility to thymic lymphomagenesis and open new perspectives for PGE₂ as a potential drug in the therapy of T-cell hematologic malignancies characterized by an uncontrolled proliferation of immature CD4⁺CD8⁺ thymocytes.

No potential conflicts of interest were disclosed.

Grant support: European Commission contract no. FI6R-CT-2003-508842 (J. Santos) and Spanish Ministry of Education and Science contract no. SAF-2006-09437 (J. Fernández-Piqueras).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Arturo Morales for critical reading of the manuscript and Immaculada Ors for her technical assistance.