If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
i3S - Instituto de Investigação e Inovação em Saúde, Universidade do Porto, Porto, PortugalInstitute of Molecular Pathology and Immunology of the University of Porto, Porto, PortugalGraduate Program in Areas of Basic and Applied Biology, Instituto de Ciências Biomédicas Abel Salazar, Universidade do Porto, Porto, Portugal
i3S - Instituto de Investigação e Inovação em Saúde, Universidade do Porto, Porto, PortugalInstitute of Molecular Pathology and Immunology of the University of Porto, Porto, PortugalCentre for Biomedical Research, University of Algarve, Faro, Portugal
1 Present address: Center for Research in Health Technologies and Information Systems, Universidade do Algarve, 8005-139 Faro, Portugal. 2 Present address: Escola Superior de Saúde, Universidade do Algarve; Algarve Biomedical Center Research Institute, 8005-139 Faro, Portugal.
Cdkn2a deletions occur in T-ALL driven by the ETV6::JAK2 fusion protein when the pre–TCR is absent.
Germline Cdkn2a inactivation favors malignant transformation of DN3 thymocytes.
Mouse T-ALL originating from Cdkn2a+/− DN3 thymocytes often presents Cdkn2a loss of heterozygosity.
Further secondary events occur in T-ALL originating from Cdkn2a+/− DN3 thymocytes.
CDKN2A deletion is the most frequent genetic alteration in T-cell acute lymphoblastic leukemia (T-ALL), occurring across all molecular and immunophenotypic subtypes. CDKN2A encodes two functionally unrelated tumor suppressor proteins, ARF and INK4a, which are critical regulators of cell cycle and proliferation. Arf has been reported to suppress T-ALL development in post–β-selection thymocytes, but whether CDKN2A acts as a tumor suppressor gene in immature, pre–β-selection thymocytes remains to be elucidated. Resorting to a Rag2-deficient model of T-ALL, driven by the ETV6::JAK2 fusion, we report that Cdkn2a haploinsufficiency at early stages of T-cell development facilitates leukemia development. ETV6::JAK2;Rag2−/−;Cdkn2a+/− T-ALL arose from pre–β-selection thymocytes because the thymocyte differentiation arrest caused by Rag2 deficiency was maintained in preleukemic mice. Furthermore, somatic loss of the Cdkn2a wild-type allele was frequently observed in ETV6::JAK2;Rag2−/−;Cdkn2a+/− T-ALL cells, suggesting a selective advantage for total inactivation of Cdkn2a. Both Cdkn2a-sufficient and Cdkn2a-defective T-ALL presented additional genetic alterations, such as Notch1 mutations and gains of chromosomes 13 and 15. These data indicate that Cdkn2a acts as a gatekeeper for leukemogenesis from the most immature stages of thymocyte development.
Pre–T-cell receptor (pre–TCR) expression in CD4−/CD8− double-negative (DN) thymocytes, following successful Tcrb locus rearrangements, is critical for the β-selection checkpoint. The pre–TCR is composed of a TCRβ chain paired with the invariant pTα protein and CD3 family proteins. Although pTα expression has been documented in a high proportion of human T-cell acute lymphoblastic leukemia (T-ALL) [
]. Further indicating that pre–TCR expression can be a major driver of thymocyte malignant transformation, compensatory genetic alterations were observed in pre–TCR-deficient ETV6::JAK2-driven leukemias, including chromosome 15 gains and chromosome 4 microdeletions encompassing the Cdkn2a locus [
The CDKN2A locus encodes two distinct tumor suppressor proteins, ARF and INK4a, which are crucial to regulating cell survival and proliferation. CDKN2A deletions are common in solid and hematological malignancies, being frequent in T-ALL and specific subtypes of B-cell ALL [
] reported that Arf prevents the malignant transformation of thymocytes after β-selection when thymocytes undergo massive proliferation after TCRβ rearrangements. However, because CDKN2A loss has been detected across all maturation subtypes of human T-ALL [
] were obtained from the National Cancer Institute Mouse Repository (strain number 01XB1). Mice were bred and maintained at the i3S barrier animal facility (high efficiency particulate air filtration of incoming air, differential pressure, and disinfection or sterilization of room equipment and supplies) under 12:12-hour light:dark cycles and with food (2014S diet; Envigo) and water ad libitum. All experimental procedures were approved by the i3S ethics committee and Portuguese authorities (Direção-Geral de Agricultura e Veterinária) and followed recommendations from the European Commission (Directive 2010/63/UE) and the local Portuguese authorities (Decreto-Lei n°113/2013). Both female and male mice were used for all experiments. Mice were monitored for signs of disease (i.e., dyspnea, lethargy, enlarged lymph nodes, enlarged abdomen, and paraplegia) and euthanized by CO2 inhalation upon reaching predefined experimental end points. Mice of different genotypes from the same litter were kept together in the same cages, and monitoring for signs of disease was performed blindly. Adult mice that were euthanized without leukemia were censored in Kaplan–Meier survival curves.
Low-coverage Whole Genome Sequencing
For low-coverage whole genome sequencing, the Ion Xpress Plus Fragment Library kit (ThermoFisher Scientific) was used for library preparation, and sequencing was performed using the Ion S5XL system (ThermoFisher Scientific). For copy number analysis, sequencing data were aligned with the mouse reference genome (Genome Reference Consortium Mouse Build 38 or mm10 reference genome) and further analyzed using QDNAseq (RRID:SCR_003174) package, R software. A bin size of 30 kb was used.
Statistical analysis was performed with GraphPad Prism 6.0 software (RRID:SCR_002798). Unpaired Student t test was used for comparisons between the two groups. Log-rank test was used to compare the survival of different groups. Sample numbers are indicated in the figure legends. p < 0.05 was considered statistically significant.
RESULTS AND DISCUSSION
To determine whether Cdkn2a inactivation could act as a tumor suppressor in pre–β-selection thymocytes, we investigated a mouse model of T-ALL arising from thymocytes arrested at the DN stage. ETV6::JAK2;Rag2−/− mice develop late-onset thymic lymphomas (median of 9 months) with malignant cells expressing CD4 and CD8 [
]. Because Rag2 deficiency imposes a strict arrest at the CD4−/CD8−/CD25+/CD44− DN stage 3 of thymocyte development, we assessed if the arrest could be overcome by ETV6::JAK2 transgenic expression. Thymocyte cellularity of preleukemic ETV6::JAK2 and Rag2−/− mice remained very low, similar to that of Rag2−/− mice. Eight of nine mice retained the DN3 block (Figure 1A,B), indicating that the ETV6::JAK2 transgene did not systematically rescue the DN3 block caused by Rag2 deficiency. One of nine ETV6::JAK2;Rag2−/− thymuses presented DN4 cells (CD4−/CD8−/CD25−/CD44−), but these expressed aberrantly high levels of the immature thymocyte marker CD24, indicating they were already transformed thymocytes. These data show that ETV6::JAK2 fusion protein does not bypass the β-selection block caused by Rag2 deficiency and transforms malignant cells at DN3 or earlier stages of thymocyte development.
Because chromosome 4 deletions encompassing the Cdkn2a locus were previously described in pre–TCR-deficient ETV6::JAK2 T-ALL [
], we performed Cdkn2a locus quantitative PCR in ETV6::JAK2;Rag2−/− (pre–TCR-deficient) T-ALL. Cdkn2a copy number loss, encompassing all exons, was found in one of six cases (Supplementary Figure E1). Compiling these results with reported array comparative genomic hybridization data [
], we verified that Cdkn2a loss occurred only in pre–TCR-deficient and not in pre–TCR-proficient T-ALL (3/15 vs. 0/11 cases analyzed; Figure 1C). Together, these data hint that Cdkn2a inactivation can cooperate with the ETV6::JAK2 fusion protein in the malignant transformation of pre–β-selection thymocytes.
Next, to determine the impact of germline Cdkn2a genetic inactivation on pre–TCR-deficient T-ALL, we bred ETV6::JAK2;Rag2−/− mice with the Cdkn2a constitutive knockout allele to generate cohorts carrying normal or haploinsufficient Cdkn2a copy number. Although not statistically significant, ETV6::JAK2;Rag2−/−;Cdkn2a+/− mice developed lymphoma/leukemia tendentially faster and with higher frequency than ETV6::JAK2;Rag2−/− mice (median survival of 53 weeks vs. 72 weeks and survival proportions of 19.7% vs. 49.2%, respectively; Figure 1D). ETV6::JAK2;Rag2−/−;Cdkn2a+/− mice presented thymic lymphomas, reduced dissemination to the spleen and lymph nodes, and immature immunophenotype, that is, aberrant expression of CD4 and CD8, as well as high levels of CD24 and CD25, similar to that of ETV6::JAK2;Rag2−/− mice (Figure 1E,F). Of note, we could not investigate the impact of complete Cdkn2a deficiency on ETV6::JAK2;Rag2−/− T-ALL development, because Rag2−/−;Cdkn2a−/− mice developed rapid B-cell precursor leukemia, a phenotype similar to that of Rag2−/−;Arf−/− mice [
We surmised that the reduced latency of T-ALL in ETV6::JAK2;Rag2−/−;Cdkn2a+/− mice could be caused by a DN3 block bypass before leukemia onset. However, we found that Cdkn2a haploinsufficiency did not rescue the DN3 developmental block. Indeed, pre-leukemic ETV6::JAK2;Rag2−/−;Cdkn2a+/− mice presented similar thymic atrophy, thymocyte hypocellularity, and DN3 thymocyte developmental arrest as Rag2−/−;Cdkn2a+/− or Rag2−/− mice (Supplementary Figure E2A,B).
To determine whether leukemogenesis was associated with somatic inactivation of Cdkn2a, we assessed the Cdkn2a copy number status in ETV6::JAK2;Rag2−/−;Cdkn2a+/− leukemic cells. Indeed, six of 10 cases showed loss of the wild-type (WT) allele (Figure 2A), indicating a selective advantage for complete Cdkn2a loss. Because ETV6::JAK2;Rag2−/− leukemias presented increased frequency of DNA copy number alterations [
], we assessed the genome-wide copy number status of T-ALL from Cdkn2a-sufficient and -haploinsufficient mice by low-coverage whole genome sequencing. We observed recurrent chromosomal numerical alterations in both ETV6::JAK2;Rag2−/− (occurring in 83% of cases, most frequently chromosome 15 gains) and ETV6::JAK2;Rag2−/−;Cdkn2a+/− T-ALL (found in 50% of cases, most frequently chromosome 4, 13, and 15 gains; Figure 2B). In two cases (nos. 53 and 128), Cdkn2a loss of heterozygosity (LOH) co-occurred with the gain of chromosome 4, which harbors the Cdkn2a locus. From the low-coverage sequencing, there was no indication of Cdkn2a deletion in these cases with chromosome 4 gain. These findings suggest that the chromosome carrying the WT Cdkn2a allele was lost, resulting in LOH, and the chromosome carrying the mutant allele was triplicated. This conjecture is reminiscent of uniparental isodisomies of chromosome 9 segments harboring CDKN2A detected in human T-ALL [
]. These data suggest that complete Cdkn2a inactivation favors ETV6::JAK2-induced leukemogenesis and that further genetic alterations are still required for the malignant transformation of pre–TCR-deficient thymocytes.
Because Notch1 mutations occur frequently in human and mouse T-ALL, including ETV6::JAK2 mouse T-ALL [
], we assessed whether the frequency of Notch1 mutations was influenced by Cdkn2a allele loss. Notch1 PEST domain (exon 34) mutations, which account for most Notch1 mutations reported in murine T-ALL [
], were detected with similar frequency (∼60%) among ETV6::JAK2, ETV6::JAK2;Rag2−/−, and ETV6::JAK2;Rag2−/−;Cdkn2a+/− leukemias, with no correlation with age of onset (Figure 2C). Notch1 PEST domain mutations in ETV6::JAK2;Rag2−/−;Cdkn2a+/− T-ALL occurred independently of Cdkn2a copy number status because four cases carried both Cdkn2a LOH and Notch1 mutation, two cases carried only Cdkn2a LOH, three cases carried only Notch1 mutations, and one case neither alteration (Figure 2D and Supplementary Table E1). This indicates that Cdkn2a complete inactivation did not relieve the selective pressure for Notch1 mutation acquisition and these two types of genetic alterations can complement each other in leukemogenesis.
In summary, using the ETV6::JAK2-driven T-ALL mouse model, we show that Cdkn2a suppresses leukemogenesis in immature pre–β-selection thymocytes. This notion is supported by two main pieces of evidence: (1) germline loss of 1 Cdkn2a copy aggravated leukemogenesis in ETV6::JAK2;Rag2−/− mice without bypassing the DN3 thymocyte developmental block; and (2) Cdkn2a somatic allele loss was frequently observed in ETV6::JAK2;Rag2−/−;Cdkn2a+/− T-ALL cells, which suggests a selective advantage for total inactivation of Cdkn2a. The fact that 40% of cases retained the WT allele indicates that either Cdkn2a haploinsufficiency conferred a selective advantage for leukemic cells or Cdkn2a is inactivated through other mechanisms, such as epigenetic silencing, previously reported in Notch1-driven mouse T-ALL [
]. Although Cdkn2a genetic inactivation facilitated thymocyte leukemogenesis, Cdkn2a-defective leukemic cells still presented other genetic alterations, such as Notch1 mutations and chromosomal gains. This indicates that Cdkn2a deletion does not fully compensate for the absence of a pre-TCR signaling complex, and secondary alterations, such as Notch1-activating mutations, are required for leukemogenesis.
Although Cdkn2a haploinsufficiency favored leukemogenesis, leukemia onset in ETV6::JAK2;Rag2−/−;Cdkn2a+/− mice occurred much later than in pre–TCR-proficient ETV6::JAK2 transgenic mice (median survival of 14 weeks; p < 0.0001, log-rank test). The chance for the acquisition of genetic alterations in Rag2-deficient pre–β-selection thymocytes is purportedly very low, due to the absence of Rag-mediated mutagenic recombination events [
] and a low rate of cell division. Although at this stage we cannot determine which events in which order are required for leukemogenesis, we posit DN3 thymocytes expressing the ETV6::JAK2 fusion protein undergo malignant transformation upon acquisition of an odd oncogenic event that either inactivates Cdkn2a completely or cooperates with reduced INK4a/ARF tumor suppressive activity. Loss of Arf has been shown to promote leukemia self-renewal [
], so this might be a crucial event in pre–β-selection thymocytes. Notch1 mutations or other events (e.g., chromosome 13 or 15 trisomy) may bolster the self-renewal or malignant phenotype of leukemic cells with haploinsufficient or complete inactivation of Cdkn2a. Notch1 mutations in our model are likely secondary events, paralleling human T-ALL, where NOTCH1 mutations are often found in minor subclones [
Our results support the notion that Cdkn2a acts as a brake for leukemogenesis at the early stages of thymocyte development before the formation of the pre-TCR complex. Other reports indicated that Cdkn2a is an important tumor suppressor in post–β-selection thymocytes [
] allow us to conclude that Cdkn2a acts as a gatekeeper for leukemogenesis from the most immature until the post–β-selection stages of thymocyte development.
Conflict of Interest Disclosures
N.R.S. has received research funding from Gilead Sciences, Portugal, and Fundação AstraZeneca, Portugal, after applying in competitive calls for projects. The other authors do not have any conflicts of interest to declare in relation to this work.
We thank Ron DePinho for providing Cdkn2a mutant mice through the NCI Mouse Repository. We thank Helena Ferreira for help with R software, Sara Alves and members of i3S Intercellular Communication and Cancer group, Carla Oliveira, Marta Ferreira, and Patricia Oliveira for fruitful discussions about this work. We thank Jacques Ghysdael for critically reading this manuscript. We acknowledge the support of Ana Mafalda Rocha and the Genomics i3S Scientific Platform for technical help with low-coverage whole genome sequencing, the Animal Facility, Translational Cytometry and Histology and Electron Microscopy (member of the Portuguese Platform of Bioimaging; PPBI-POCI-01-0145-FEDER-022122) i3S Scientific Platforms. This work was supported by a fellowship (Bolsa LLA 21) from Associação Portuguesa Contra a Leucemia and Sociedade Portuguesa de Hematologia in partnership with Amgen Biofarmacêutica, Lda. This work was supported by European Regional Development Fund (ERDF), through COMPETE 2020 – Operational Program for Competitiveness and Internationalization, Portugal2020, and Fundação para a Ciência e a Tecnologia (FCT; POCI-01-0145-FEDER-007274; PTDC/MED-ONC/32592/2017), and by ERDF, through the Norte Portugal Regional Program (NORTE2020), Portugal2020 (NORTE-01-0145-FEDER-000029). T.A.C. was a recipient of FCT (PD/BD/114129/2015) and Liga Portuguesa Contra o Cancro – Núcleo Regional do Norte fellowships. Authors Ghezzo (SFRH/BD/80503/2011) and Fernandes (SFRH/BD/75137/2010) were recipients of FCT fellowships.
T.A.C. designed, performed, and analyzed experiments and created the figures and wrote the manuscript; I.P.-L., F.A.-D., M.N.G., M.T.F., and T.C. performed experiments; and N.R.S. designed the study, performed experiments, and wrote the manuscript.
The data that support the findings of this study are available from the corresponding author upon request. Raw unaligned sequencing reads (fastq-format) that support the findings of this study have been deposited in the Sequence Read Archive under the accession number PRJNA843967.