RNA-Seq reveals the existence of a CDKN1C-E2F1-TP53 axis that is altered in human T-cell lymphoblastic lymphomas

Precursor T-cell lymphoblastic neoplasms are aggressive haematological malignancies that mainly develop in children (in particular adolescent males) but also in adults. They derive from maturing thymocytes leading to excessive lymphoblastoid cells in the bone marrow and other lymphoid organs. Clinically, T-cell acute lymphoblastic leukaemia (T-ALL) and T-cell lymphoblastic lymphoma (T-LBL) are two subgroups differing by the extent of bone marrow infiltration. T-ALL manifests with extensive bone marrow and blood affectation, whereas a mass lesion in the thymus/anterior mediastinum with less than 25% of lymphoblasts in the bone marrow characterizes T-LBL [1].

As in other cancers, the loss of cell cycle control plays a prominent role in the pathogenesis of these malignancies that is primarily attributed to loss of CDKN2A (which encodes the tumour suppressor protein p16INK4A) and, to a lesser extent, loss of RB1 or CDKN1B (which encodes p27/KIP1 protein) and aberrantly high levels of CCND2 (encoding cyclin D2) [2]. Downregulation of CDKN1C (which encodes p57/KIP2 protein) by promoter hypermethylation has been detected with very low frequency in paediatric T-ALL and more often in adult patients. However, the biological and clinical impact of hypermethylation and/or loss of CDKN1C expression remain uncertain [3]. In addition to T-ALL, downregulation of CDKN1C has been observed more frequently in a wide variety of human tumours associated with a strengthening of cell proliferation [4, 5].

In addition, numerous studies have reported that E2F1 overexpression has clinical relevance in many types of cancers [6]. However, to the best of our knowledge, E2F1 alterations have not been so far implicated in the development of precursor T-cell neoplasms.

Moreover, the gene encoding TP53 protein, a main downstream effector of E2F1, is frequently targeted in human tumours by gene mutations [7, 8]. Apart from the canonical full-length transcript, it should be noted that alternative splicing of TP53 and the use of alternate promoter might result in multiple transcript variants and isoforms [9] and, interestingly, abnormal expression of TP53 isoforms has been reported in many cancers as head and neck, acute myeloid leukaemia (AML) and breast tumours [10] but not in T-cell lymphoblastic neoplasms.

The potential nexus between these three genes has been demonstrated in mice. Some authors [11] have shown in mouse models that inactivation of the Cdkn1c gene (also termed as p57 ^KIP2 ) results in thymocyte development arrest at DN3 (Double-Negative 3) to DN4 cells transition, due to hyper-activation of the E2f-Tp53 pathway. Furthermore, the loss of Cdkn1c accelerates the development of thymic lymphomas in the absence of the Tp53 gene.

To assess whether the axis CDKN1C/E2F1/TP53 plays a role in human T-cell lymphoblastic lymphomas, we investigated the mutational status and the expression levels of these three genes using Next-Generation Sequencing (NSG) approaches. Interestingly, RNA-Sequencing analysis revealed reduced levels of CDKN1C mRNA in almost all analysed T-LBL samples, which may be accompanied by increased expression of E2F1 and overexpression of the TP53 transcript variant encoding the ∆133TP53 isoform. Deregulation of these genes is executed by epigenetics mechanisms and deregulation of specific miRNAs.

It is well established that CDKN1C and E2F1 are two critical controllers of the cell cycle. The overexpression of CDKN1C may cause cell cycle arrest in human tumour cell lines [30, 31], and this inhibitory effect may be reversed by siRNAs against the CDKN1C gene [32]. In contrast, knockdown of E2F1 by RNA interference impairs proliferation of rat glioma cells [33]. Importantly, previous experimental work in mice reported that conditional T cell-specific deletion of Cdkn1c gene induced a differentiation block in mouse immature thymocytes that is caused by hyperactivation of E2f1 and Tp53 and may be predisposed to thymic lymphoma development. Moreover, Cdkn1c ablation led to the development of aggressive thymic lymphomas with a reduced latency in a Tp53-null background. Thus, these results suggested a critical role for the Cdkn1c-E2f1-Tp53 axis in mouse thymic lymphoma development [11, 34].

Our results show that all analysed human T-LBL samples exhibited a strong downregulation of CDKN1C. In addition, most of them also exhibited upregulation of E2F1 (6/8 in the exploratory cohort and 6/10 in the extended cohort), which may be accompanied by impairment of TP53 function in some cases (4/6 in the exploratory cohort and 6/10 in the extended cohort) (Fig. 1; Additional file 3: Table S3 and Additional file 4: Table S4). Thus, our data are consistent with the existence and deregulation of a CDKN1C-E2F1-TP53 axis in human T-LBL. However, it should be noted that our study is largely based on the expression of these genes at the transcriptional level. The relationship between mRNA and protein expression levels is dependent on the combined outcomes of mRNA stability, translation, and protein degradation. Notwithstanding, it has been reported that at least 30 to even 85% of the variation in protein levels can be attributed to variation in mRNA expression [35]. Other authors [36] reported that differentially expressed mRNAs correlate significantly better with their protein product than non-differentially expressed mRNAs, therefore providing some optimism for the usefulness on inferences from mRNA expression in general.

Concerning the mechanisms by which these genes are deregulated, it is well known that CDKN1C is subject to a complex regulation involving the cooperation of a CpG island at its promoter region and distal regulatory elements, such as the imprinting control region Kv-Differentially Methylated Region 1 (KvDMR1) in the promoter of the noncoding KCNQ1OT1 [37, 38]. Although the biological and clinical impact of CDKN1C hypermethylation is rather uncertain, aberrant DNA methylation of CDKN1C in its promoter region has been reported in lymphoid malignancies of B and T-cell phenotype [39, 40]. However, CDKN1C has been reported downregulated in other type of cancer cells mainly by histone modifications operating in critical regions of its promoter [41, 42]. We initially focused on promoter hypermethylation to explain downregulation of this gene in our sample series of T-LBL, but despite a substantial reduction in the levels of mRNA in almost all samples in the exploratory cohort (7/8), only two samples (840 and 521) (2/8) exhibited significant hypermethylation density (Fig. 4), and six out of eight (including tumor 840 with promoter hypermethylation) exhibited upregulation of one or two miRNAs selected for CDKN1C regulation (miR-211–3p and miR-222-3p). Thus, downregulation of CDKN1C in two samples (33 and 346) should be explained by a different transcriptional mechanism.

Besides this epigenetic mechanism, regulation by miRNAs might be an additional way contributing to determine CDKN1C transcript levels in T-LBLs. Results reported here are in line with those reported in the literature describing miR-25, miR-221 and miR-222 as direct regulators of CDKN1C expression in a wide variety of solid tumours, showing a new mechanism responsible for CDKN1C downregulation in carcinogenesis [43‐45]. In this context, our findings suggest that aberrant expression of miR-221 and miR-222 may have an oncogenic function in T-LBL development by targeting CDKN1C. However two samples (33 and 346) showed a pronounced downregulation of CDKN1C in the absence of significant changes in miRNA expression (Figs. 5 and 6) or promoter CpG methylation, thus indicating that the mechanism regulating the expression of this gene is far more complex.

Overexpression of E2F1 may promote proliferation or cell cycle progression by increasing the transcription of genes that contribute to G1-S transition [46]. Notwithstanding at the same time it may also induce apoptosis by multiple pathways, some of which induce stabilization and activation of the TP53 protein [47]. Our microRNA analysis also revealed a consistent deregulation of seven miRNAs in T-LBLs, miR-203a and miR-205-5p being the most representative downregulated microRNAs (Figs. 5 and 6). Interestingly, downregulated miRNAs showed higher fold changes than upregulated microRNAs. miR-205-5p is known to be down-regulated in melanoma and its expression inversely correlated with that of E2F1 [48].

Concerning impairing of TP53 function, we found overexpression of the human Δ133p53αisoform in 4 samples from the exploratory cohort, from which three also exhibited downregulation of the isoform encoding full length TAp53α protein isoforms (Figs. 1 and 2). It has been demonstrated that ∆133p53α does not exclusively function in a dominant-negative manner toward TAp53α, the full-length TP53 isoform [49], but it also inhibits TP53-dependent apoptosis [50]. Finally, two tumours (192 and 521) showed increased amounts of the TP53β transcript, which encodes a C-terminal truncated protein that downplay TP53 capacity to induce apoptosis [9, 51]. These changes in the expression levels of full length and shorter isoforms may be sustained, at least in part, by deregulation of 17 miRNAs, with particular reference to miR-200a-3p and miR-375 that exhibited very high levels of downregulation in all samples in the exploratory cohort (Figs. 5 and 6).

But impairment of the TP53 function could be also attributed to the overrepresentation of the arg72 allele in our sample series (Fig. 2). It is known that the TP53 gene is not only frequently mutated in human tumours [7], but it also contains several functional polymorphisms, being by far the most common a proline (Pro) to arginine (Arg) change at codon 72 in the TP53 protein [10]. Several studies have reported preferential retention of arg72 allele in squamous cell carcinomas of the vulva [52], head and neck [53], and esophagus [54]. Considering tumour tissue DNA, Schneider-Stock et al. [55] found a significantly higher frequency of the arg72 allele in colorectal tumours and reported that the presence of this allele correlates with the malignant potential of the tumour. Similar results were also reported in urinary tract cancers [56] and lung cancer [57]. The arg72 allele was also related with increased risk for bladder cancer [58].

Our results indicate the existence of a CDKN1C-E2F1-TP53 axis that is disrupted in a significant fraction of human T-LBLs. If, as expected, the consequences of the deregulation of the CDKN1C-E2F1-TP53 axis were the same as those experimentally demonstrated in mouse models, deregulation of this axis in human T-LBL might serve as a biomarker to predict the aggressiveness of T-LBL development as depicted in Fig. 7. Furthermore, these findings would provide the basis to the development of potential therapeutic strategies based on the use of microRNAs (mimics or antagomirs) to target CDKN1C or E2F1 deregulation that allow to rescue normal thymocyte differentiation and normal levels of thymocytes proliferation in patients with T-LBL. Blocking E2F1 expression by RNA interference might represent a promising therapeutic approach in this type of tumours. Future studies with new samples series of T-LBL should be done to be sure that the differences detected at the mRNA level translate into the protein level, and to confirm that the deregulation of this axis in human samples is really predictive of clinical outcome.