Telomerase functions beyond telomere maintenance in primary cutaneous T-cell lymphoma

Human telomeres consist of tandem repeats of a DNA sequence (5′-TTAGGG-3′) that terminates in a 3′ single-stranded overhang and is organized in a higher-order structure built and stabilized by shelterin protein complex.¹  Such a cap is thought to sequester the end of the chromosomes away from the double-strand break repair apparatus and prevent aberrant chromosomal fusions and recombinations. Telomeres also perform other functions, which include transcriptional silencing of genes located close to them, as well as ensuring correct chromosome segregation during mitosis.²  Because of the end-replication problem, loss of telomeric sequences occurs during each cell division,³  resulting in the gradual and predictable erosion of telomeres (reviewed in Cech⁴ ). This process reduces the chromosome’s ability to form functional capped telomere structures.⁵  When telomere length (TL) reaches a certain limit, telomeres become dysfunctional and elicit a p53-dependent DNA damage checkpoint response^6,7  which, depending on the cell type, triggers either replicative senescence or apoptotic cell death. In stem cells, germ-cell lineages, and in nearly 90% of tumors, this proliferative limit is overcome by activating telomerase, a reverse transcriptase enzyme (TERT) that uses an RNA molecule (TERC) as a template to add DNA repeats onto the 3′ chromosomal ends.^8-10  Telomere dysfunction and telomerase deregulation have been found to be associated with genomic instability and disease progression.¹¹  Therefore, TL and telomerase expression may constitute prognostic tools and potential targets for anticancer drugs.¹²  This concept, first established in solid tumors,^13,14  can now be applied in most hematologic malignancies.^15,16 

Relatively few studies on telomere status and telomerase activity (TA) in T-cell malignancies have been published.^12,17  The aim of this study was to investigate such parameters in cutaneous T-cell lymphomas (CTCLs), since they were not previously assessed in neoplastic cells.^18-20  We focused on common subtypes of CTCL: 2 aggressive forms, transformed mycosis fungoides (T-MF) and Sézary syndrome (SS) with a median survival of ∼2 years, and primary cutaneous anaplastic large cell lymphoma (C-ALCL), with a favorable prognosis (10-year disease-related survival exceeding 90%).²¹  In these lymphoproliferations, the neoplastic T cells involve the skin as a primary site, but with disease progression, abnormal cells may also infiltrate blood, lymph nodes, lungs, heart, liver, and spleen, except for SS, which arises suddenly as an aggressive leukemic variant.²¹  Whereas recent cytogenetic studies on T-MF, SS, and C-ALCL reported the presence of genetic aberrations,^22-29  the key molecular targets underlying CTCL pathogenesis remain elusive. This lack of insight into the biology of CTCL has hindered the development of true targeted therapies.^30,31 

In this context, we explored TL and TA in CTCL by using cell lines (Hut78, Se-Ax, My-La, FE-PD) and patient cells (SS, T-MF, and C-ALCL) to better understand the role played by telomeres and telomerase in pathogenesis and to evaluate the relevance of these biomarkers as target candidates for diagnostic and therapeutic purposes.

Fifty-seven patients with CTCL, according to the criteria of the World Health Organization and the European Organisation for Research and Treatment of Cancer,^21,32  were selected in the CTCL cohort of Aquitaine. This study complied with the Declaration of Helsinki and was approved by the local ethics committee. Each patient gave written consent. Twenty-four SS (supplemental Table 1, available on the Blood Web site.), 19 T-MF, and 14 C-ALCL samples were analyzed and compared with peripheral blood samples from 49 age-matched healthy donors (25 males, 24 females) recruited from the Etablissement Français du Sang in Bordeaux, France.

Experiments were performed on either peripheral blood mononuclear cells or 2 SS cell lines (Hut78 [LGC Standards] and Se-Ax [kindly provided by Dr K. Kaltoft, Aarhus, Denmark]), 1 T-MF cell line (My-La [Dr K. Kaltoft], a CD30⁺ALK^– Hodgkin-like ALCL), FE-PD (Prof G. Delsol, Toulouse, France), and a T-cell lymphoblastic leukemia cell line (1301; Sigma-Aldrich). Since 1301 exhibited unusually long telomeres and was telomerase positive, it was used as a positive control.

DNA and RNA were extracted from cell lines, white blood cells, and fresh-frozen tumor biopsies of T-MF and C-ALCL patients. Genomic DNA was prepared from cells by using a salt extraction procedure³³  and from tumor biopsies by using proteinase K digestion followed by a standard phenol/chloroform extraction procedure. Total RNA was isolated by using TRIzol reagent (Invitrogen).

Relative TL was measured from genomic DNA by using a standard quantitative polymerase chain reaction (qPCR) assay.³⁴  For interphase quantitative fluorescence in situ hybridization (IQFISH), cell nuclei were prepared and spread onto glass slides using standard cytogenetic procedures.³⁵  The hybridization was performed according to manufacturer’s instructions (DakoCytomation) with minor modifications.

Slide scanning, nuclei identification, fluorescence intensity, and nuclei surface measurements were performed by using a motorized Axio Imager Z2 microscope (Zeiss) equipped with a motorized 8-slide scanning stage (Märzhäuser Wetzlar), a high-resolution monochrome camera (CoolCube 1m), and Metafer software (MetaSystems). MetaCyte is a single-cell analysis software module integrated into the Metafer platform that enables setting of capture and exposure parameters and determination of target cells, nuclei selection criteria (eg, minimum and maximum nucleus area, maximum relative concavity depth, maximum aspect ratio), and image analysis features (eg, contour area, total signal intensity, ratio value). We collected data from 1000 nuclei per slide.

To correlate the chromosome content and the neoplastic T-cell status of SS patient cells, we performed sequential labeling by using immunofluorescence (IF) followed by interphase FISH. Images were acquired after each labeling method. To delineate neoplastic SS T cells, CD158e/K (KIR3DL1/DL2 mouse anti-human; magnetic-activated cell sorting; Miltenyi Biotec) and CD3 (rabbit anti-human; Dako) antibodies revealed by anti-mouse antibody conjugated with AlexaFluor 488 and anti-rabbit antibody conjugated with AlexaFluor 594 (Molecular Probes), respectively, were used. Twenty images were collected by using the image analysis workstation (Isis; Metasystems), and microscope coordinates of each image were recorded to relocate the same area after FISH assay. As in SS patients (cases 2 and 17), a partial chromosome 8 gain was observed using multicolor FISH and array comparative genomic hybridization (data not shown). We performed dual-color FISH analysis with LSI c-MYC SpectrumOrange (8q24.12-q24.13) and CEP8 SpectrumGreen (8p11.1-q11.1) (Abbott Molecular, Inc.) with a view toward evaluating chromosome count in normal and neoplastic T cells, knowing that normal T cells are CD3⁺ and CD158e/k^– and that neoplastic SS T cells are both CD3⁺ and CD158e/k⁺. As soon as IF images were acquired, slides were used in FISH assays. After FISH assay, nuclei were relocated according to immunostaining coordinates, and c-MYC/chromosome 8 status was evaluated for the same nuclei.

To correlate chromosome count in SS cells and TL, we performed sequential labeling by using an IQFISH assay followed by a FISH assay using LSI c-MYC SpectrumOrange and CEP8 SpectrumGreen. Twenty images of the same area containing approximately 100 nuclei were collected (Isis; Metasystems).

A 3-color FISH strategy using RP11-117B23, RP11-619F12, and RP11-6461O14 was applied. Fluorescence signals were analyzed in 100 nuclei, and metaphase spreads for each sample were studied (Isis; Metasystems).

hTERT messenger RNA (mRNA) levels were determined by quantitative reverse-transcription PCR (qRT-PCR) normalized to the amount of TATA box-binding protein (TBP) transcripts, displayed as hTERT/TBP ratios.

TA was assessed by using the TRAP assay (TRAPeze RT telomerase detection kit; Chemicon) according to manufacturer’s instructions, with some modifications.

To inhibit telomerase subunit expression, 3 short hairpin RNA (shRNA) constructs, shTERC, shhTERT1, and shhTERT2, targeting hTERC or hTERT transcripts, respectively, and an shScramble (nontargeting) used as control, were cloned into pLK0.1-Puro (Addgene) vector at AgeI/EcoRI sites. To overexpress hTERT, two lentiviral constructs were cloned by inserting hTERT complementary DNA or DsRed2 (used as control) into two attB sites by gateway cloning in a lentivector backbone (hPGK/AttR sites/mPGK/Puro). For shRNA sequences, lentiviral production, and titration, refer to supplemental Data.

The day before the infection, cells were seeded at a density of 1.5 × 10⁵ cells per well into 12-well plates. Cells were then infected with lentiviruses for telomerase inhibition or overexpression, as described above. At 5 and 10 days after infection, cell proliferation was measured by a direct cell-counting assay.

Western blot analysis was performed by following standard protocols. Membranes were hybridized with antibodies against hTERT (Millipore), Ki67 (Dako), and tubulin (Sigma-Aldrich). Proteins were visualized by enhanced chemiluminescence (GE Healthcare) and autoradiography, according to the manufacturer's instructions.

Senescence markers p53, p21waf1, and γ-H2AX were analyzed by qRT-PCR and IF. p53 and p21waf1 were analyzed by qRT-PCR and γ-H2AX by IF, in 1301 and My-La cells in which telomerase expression was modulated (inhibition and overexpression).

For the soft agar assay, 50 000 cells (1301, My-La, Hut78, FE-PD) overexpressing hTERT or with telomerase inhibition (shTERC, shTERT1, and shTERT2) and their respective controls were put in 6-well plates in RPMI containing 10% fetal bovine serum and 0.45% agarose (overlay) on the top of an agarose underlay (RPMI + 10% fetal bovine serum and 0.75% agarose). Cells were fed twice a week with 500 µL of fresh RPMI medium, and the colonies (>8 cells) were counted after 2 weeks. Three different fields were scored from each well. The experiments were performed in triplicate, and the results were expressed as mean ± standard deviation. Significant difference (vs control) was determined by Student t test.

A total of 1 × 10⁶ My-La cells overexpressing hTERT or infected with control vector were injected subcutaneously in NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ mice. Tumor volume was measured twice a week over 20 days in 8 animals. Mayer’s hematoxylin labeling and CD3 immunohistochemistry were performed on mice tumor sections.

Data were analyzed according to the number of dependent variables and the nature of dependent and independent variables. The Shapiro-Wilk test was used to test the normality of our data. The Wilcoxon Mann-Whitney or the Friedman test was used for nonparametric statistical analysis. These tests were suitable for a sample number below 30. To remove outlier values, Dixon’s test was applied. The significance level was taken at P ≤ .05. The statistical tests were performed by using MedCalc software. More details on the methods can be found in the supplemental Data.

This study explored TL and telomerase status in CTCL subtypes. By using a qPCR assay, we found that in SS (P = .03) and T-MF (P = .04), two aggressive subtypes of CTCL, TL was shorter compared with that in healthy donors, whereas in C-ALCL, TL was similar to that of healthy donors used as controls (P = .75) (Figure 1). To strengthen this observation, SS patient cells and CTCL cell lines were investigated by using another TL assay. Hence, IQFISH validated the data previously obtained by qPCR. Indeed, SS patient tumor cells and CTCL cell lines had significantly shorter TL than cells from age-matched healthy donors (P = 2.61 × 10⁻⁷) (Figure 1). Patients with SS were elderly (mean age, 73 years), so we checked that their age was not the cause of telomere shrinkage (P = .58) (Figure 1).

Metafer Metacyte analysis software allowed us to visualize cell nuclei and also to measure their fluorescence (Figure 2A-B) and to count nuclei with weak fluorescence (short telomeres) (Figure 2C-D) as well as neighborhood nuclei with strong fluorescence intensity (long telomeres) (Figure 2A-D). We observed that age-matched healthy donor nuclei consistently showed homogeneous fluorescence intensity signals within nuclei (Figure 2A) and a unimodal distribution of TL (Figure 2C), whereas nuclei from SS patients exhibited strong and weak fluorescence intensity signals (Figure 2B) associated with a bimodal distribution of TL (Figure 2D). We wondered whether a relationship between chromosomal content and TL could be established. To answer this question, genomic abnormalities were investigated in SS patient cells and cell lines, taking into consideration chromosome aberrations previously reported in SS patients by our group²⁷  and others.^23-26,28,29  For cases 2 and 17, AU (data not shown), c-MYC and centromere 8 fluorescent probes were chosen to investigate SS nuclei chromosomal status in combination with antibodies to delineate SS cells (Figure 2E-H) or with peptide nucleic acid telomeric probes to evaluate TL (Figure 2I-J). Image analysis revealed that abnormal chromosome content was restricted to both neoplastic SS T cells (Figure 2E-H) and to nuclei with short telomeres, whereas cells with normal chromosome content showed long telomeres (Figure 2I-J).

Investigation of the hTERT gene region at 5p15.33³⁶  by FISH in 24 SS patients revealed that cells from 71% of patients (17/24) exhibited a diploid and nonrearranged hTERT status (Figure 3A). Numerical and structural alterations of the hTERT locus were rarely encountered in our series. Among the 7 cases (7/24) with an abnormal hTERT region (Figure 3B), only 2 cases presented a gain restricted to the hTERT locus. The other numerical abnormalities were due to the presence of whole or partial chromosome 5 gain (Figure 3C).

We studied hTERT mRNA levels in SS patient cells (n = 16), cell lines (n = 5), and healthy donors (n = 35) by using qRT-PCR (Figure 3D). We observed an increase in mRNA level in SS cells compared with cells from healthy donors, but this was not statistically significant (P = .29). All cell lines tested expressed hTERT mRNA except Hut78, an SS cell line. Interestingly, Se-Ax (the other SS cell line studied) also exhibited a lower hTERT mRNA level than T-MF and ALCL cell lines (Figure 3D).

In contrast with healthy lymphocytes, TA was detected in CTCL patient cells (6/7 SS [P = .038], 6/8 T-MF [P = .001], and 3/4 C-ALCL [P = .016]) and in all cell lines (Hut78, My-La, FE-PD, and 1301) investigated (Figure 4). However, SS patient cells and the tested SS cell line Hut78 exhibited the lowest TA level compared with T-MF and C-ALCL patient cells and cell lines.

Three lentiviral shRNA vectors were constructed to inhibit telomerase expression: shTERC, shhTERT1, and shhTERT2. They were introduced into Hut78, My-La, FE-PD, and 1301 cell line nuclei. To enhance hTERT expression, a viral infection assay was used. Both control curves (Scramble and DsRed) were closed to each other. Whatever the cell line, telomerase inhibition led to a significant cell proliferation inhibition and to a massive cell death as early as 5 days after lentiviral transduction (P < .05) (Figure 5A, dotted lines). Moreover, in all cell lines tested, downregulation of hTERT (catalytic subunit of telomerase) had substantially more pronounced effects than downregulation of hTERC (the RNA template) at both 5 and 10 days after lentiviral transduction. Conversely, lentivirus-mediated hTERT overexpression led to a significant increase in cell proliferation (P < .05) (Figure 5A, solid lines).

hTERT silencing and ectopic expression were determined by using western blot analysis (Figure 5B-C). The impact of hTERT inhibition (Figure 5B) and overexpression (Figure 5C) on cell proliferation was confirmed by the detection of downregulation (Figure 5B) or upregulation (Figure 5C), respectively, of the Ki67 proliferation marker. We wondered whether TL was involved in the cell proliferation variations we observed upon modulation of telomerase expression. To test this hypothesis, we investigated the relative TL in the same cell lines by qPCR (supplemental Table 2). With only a short time for cell culture, no significant variation in TL was observed in cell lines.

Because 1301 and My-La cells expressed a higher hTERT endogenous level than Hut78 and FE-PD and strongly responded to hTERT inhibition, they were tested for senescence markers upon telomerase inhibition. Hence, telomerase inhibition led to upregulation of p53 and p21 mRNA levels (Figure 5D) and γ-H2AX staining (supplemental Figure 8).

The above-mentioned data showed that hTERT played a significant role in CTCL cell proliferation. To further investigate the impact of this protein on the transformed properties in vitro, we used a soft agar clonogenicity assay with Hut78, FE-PD, My-La, and 1301 cells that overexpressed hTERT compared with control vector. While Hut78 and FE-PD cells did not form colonies in the tested conditions, control 1301 cells formed colonies, and the introduction of hTERT enhanced the colony formation potentiality (threefold increase). Control My-La cells were unable to form colonies, but they became able to form colonies (nearly 40 colonies per field) upon hTERT overexpression (Figure 6A). In contrast, telomerase inhibition impeded colony formation only in cells endogenously able to form colonies such as 1301.

On the basis of our observations concerning the role of hTERT on in vitro proliferation, survival, and clonogenicity, we investigated its effects on the in vivo tumorigenic properties of My-La, a CTCL cell line, in immunodeficient mice. My-La cells that did or did not overexpress hTERT (control vector vs hTERT vector) were xenografted into NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ mice (Figure 6B). Tumor formation was clearly detected as early as 10 days following the inoculation of My-La cells. We observed a major impact on tumor development induced by ectopic hTERT expression. Indeed, in all mice grafted with hTERT overexpressing My-La cells, both tumor volume (Figure 6B) and weight (Figure 6C) were strongly increased compared with mice grafted with control My-La cells, which barely developed tumors in the tested conditions.

Tumor measurement and weight demonstrated an impressive tumor growth (Figure 6D-2 compared with Figure 6D-1) of 19.25 mm³ per day when hTERT was overexpressed compared with control (1.19 mm³ per day). Histologic sections from My-La cell–derived tumors showed a large number of mitotic cells when hTERT was overexpressed (Figure 6D-3) whereas in control mice, tumors were undetectable. The positive CD3 labeling showed that tumors were composed of T cells (Figure 6D-4).

In aggressive subtypes of CTCL, T-MF, and SS, TL was shorter than in healthy donor lymphocytes and in a CTCL subtype associated with good prognosis (C-ALCL). The CTCL cell lines Hut78, Se-Ax, FE-PD, and to a lesser extent My-La, showed a TL that matched that of their corresponding lymphoma subtype of origin. In agreement with our observations, telomere erosion was associated with disease progression and aggressiveness as well as chromosome abnormalities in human tumors.^11,37-39  Indeed, data from our group^22,27  and from others^{23-26,28,29,40-43}  showed the presence of more complex chromosome aberrations in aggressive CTCL than in C-ALCL.^28,44  When telomeres become critically short, it is believed that their protective function is compromised and that cells follow the flow diagram process (Figure 7) with replicative senescence, crisis, escape from crisis, and maintenance of TL.⁴⁵  Our data support this theoretical process and also suggest that aggressive CTCLs are passed-over crisis cells. Furthermore, we demonstrated here that short telomeres are restricted to neoplastic T cells, especially in SS cells. This critical point in elderly patients was not previously addressed by the rare investigations of telomere status in patients with CTCL.^18-20 

Because the ability to maintain TL is accomplished by activating telomerase in most tumor cells,^8-10  we investigated the status of the catalytic subunit of the human telomerase hTERT in SS patient cells. The vast majority of SS cells displayed a diploid hTERT locus and presented a low level of hTERT expression. This low level of hTERT mRNA was associated with TA and short telomeres in SS cells. This suggests that low levels of hTERT transcripts are sufficient to account for basal TA and to maintain functional short telomeres. This could represent an Achilles’ heel of such tumor cells since inhibition of hTERT alone led to a drastic drop in TA in SS cell lines. Moreover, TA in SS patient cells seemed to be correlated with the percentage of SS cells in S phase of the cell cycle (data not shown). A putative link between cell proliferation and TA was previously reported in primary human mammary epithelial cells.^46,47  By using an extensive panel of hTERT mutants, Mukherjee et al⁴⁷  were able to decipher the diversity of biologic functions attributed to hTERT. They demonstrated that the ability of hTERT to enhance cell proliferation required hTERT catalytic activity but was uncoupled from telomere elongation. While these elegant observations were made in cultured human mammary epithelial cells, our study directly addressed the association between hTERT expression and cell proliferation in SS cells.

To further investigate hTERT function in CTCL, we inhibited hTERT activity in cell lines (Hut78, My-La, FE-PD, and 1301) by using hTERT shRNA constructs. The inhibition of hTERT led to a drastic reduction of proliferation, which was associated with activation of senescence markers, and drove rapidly to cell death, emphasizing the role of telomerase in cell proliferation and survival. This was also observed by inhibition of TERC expression, the RNA component of telomerase. Nonetheless, the cell proliferation defect was more pronounced when inhibiting the hTERT catalytic component than when inhibiting the TERC RNA component, suggesting that hTERT may exert TERC-independent effects on cell proliferation. In contrast, enhanced hTERT expression led to an increase of cell proliferation. Changes in cell proliferation occurred without measureable variations in TL, suggesting that telomerase could regulate cell proliferation independently of its roles in telomeres, as reported recently.⁴⁸  The cell cycle distribution of CTCL cell lines, with and without telomerase, was not significantly altered, but cells overexpressing hTERT progressed more quickly through the cell cycle (data not shown). In addition, since hTERT inhibition led to cell death, we focused on the impact of hTERT overexpression on tumorigenicity both in vitro by using a soft agar clonogenicity assay and in vivo by using xenografts in immunodeficient mice. Upregulation of hTERT in 1301 and My-La cells significantly enhanced colony formation capacity as well as tumor development. This tumor formation was clearly detectable as early as 10 days after inoculating immunodeficient mice with My-La overexpressing hTERT. At the end of the experiment, large tumors were observed with cells overexpressing hTERT (mean weight, 342 mg), whereas no tumor was detectable in mice grafted with control cells.

In summary, this study showed that aggressive CTCL subtypes exhibited short telomeres. This consistent short length was restricted to neoplastic T cells. We observed that CTCL cells were telomerase positive. In these cells, we assumed that telomerase expression could be involved in maintaining functional telomeres. Beyond this role, telomerase might exert functions on proliferation and tumorigenicity in vitro and in vivo, as shown by proliferation and soft agar assays and xenografts in immunodeficient mice. Further investigations are necessary to improve our knowledge of the molecular pathways implicated in these additional telomerase functions. Nevertheless, the data presented here adds an appealing perspective for the development of treatment targeting telomerase functions in such lymphomas, especially in aggressive CTCL cells.

The authors are very grateful to B. Rousseau, Animalerie A2 Université Bordeaux, for helping with tumor measurements and to A. Giese, histology platform Cancéropole Grand Sud Ouest-Site de Recherche Intégrée sur le Cancer (SIRIC) Bordeaux Recherche Intégrée en Oncologie (BRIO) for technical assistance (Bordeaux, France). The authors thank Dr Canamero, Centro Nacional de Investigaciones Oncológicas (Madrid, Spain) and Dr Segal-Bendirdjian, Institut National de la Santé et de la Recherche Médicale UMR-S 1007 (Paris, France) for their advice regarding hTERT antibodies, Dr Vial for providing the CD158e/K antibody (Centre Hospitalier Universitaire [CHU] Bordeaux, France), Dr V. Gire for her kind support (Unité mixte de recherche 5237, Montpellier, France), Dr R. M’kacher for helpful discussion (Commissariat à l'énergie atomique, Direction des Sciences du Vivant/Institut de radiobiologie cellulaire et moléculaire, Fontenay aux Roses, France), and Dr. C Bedichaud, Etablissement Français Du Sang Bordeaux (France). The authors also thank C. Martin for secretarial assistance, the volunteer blood donors, and the tumor bank (CHU Bordeaux).

This work was supported by grants from Société Française de Dermatologie, Ligue Contre le Cancer, Conseil Regional d’Aquitaine (No. 20081302003), Caractérisation des Anomalies Cytogénétiques des Lymphomes T cutanés par obtention de xénogreffes, CHU Bordeaux, and SIRIC BRIO. Lentiviral vectors production was conducted in the P3 facility, Bergonié Institute (Bordeaux, France). L. A. was supported by Programme Hospitalier de Recherche Clinique Essai multicentrique de phase II évaluant le bénéfice thérapeutique du lenalidomide (Revlimid) dans les lymphomes cutanés primitifs B diffus à grandes cellules “type-jambe” en rechute ou réfractaires à un traitement initial associant Rituximab et polychimiothérapie, Centre Hospitalier Universitaire Bordeaux 2011/28-REV-LEG.

Contribution: E.C. conceived and designed the study, performed IQFISH experiments, DNA extractions, and cell culture, analyzed and interpreted the data, and wrote the manuscript; L.A. performed lentiviral, western blot, soft agar, and xenografts experiments; M.P.-C. performed immunostaining, FISH experiments, slide scanning, and quantitative image analysis as well as statistical analysis; J.F. performed qPCR analysis, TRAP assay, and RNA and protein extractions with Y.I.; D.C. designed qPCR primers and analyzed qPCR data; E.L. performed and analyzed array comparative genomic hybridization experiments; A.B. and J.F. performed qPCR analysis; W.S. and E.C. performed IQFISH; F.R. and M.P.-C. performed hTERT FISH experiments; A.P.-L., B.V., and M.B.-B. provided CTCL patients; A.P.-L. and M.B.-B. provided patient samples and clinical data; F.B. performed cell cycle analysis; P.D. participated in the design of the study and data interpretation; J.-P.M. supervised all the research; and L.A., M.P.-C., D.C., P.D., M.B.-B., and J.-P.M. assisted in writing the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Edith Chevret, Histologie et pathologie moléculaire des tumeurs, EA 2406, Université Bordeaux, 146 Rue Leo Saignat, 33076 Bordeaux, France; e-mail: edith.chevret@u-bordeaux2.fr.