Background
Burkitt lymphoma (BL) is a highly aggressive B-cell non-Hodgkin lymphoma characterized by peculiar clinical, morphological, immunophenotypical, cytogenetic, and gene expression profile features [
1]. The current World Health Organization (WHO) classification of tumors of hematopoietic and lymphoid tissue assesses that no single parameter can be used as the gold standard to achieve the diagnosis but that a combination of clinical, histological, immunophenotypical and genetic criteria is necessary [
1]. The presence of the
MYC-associated translocation [t(8;14)
MYC/Immunoglobulin heavy chain gene (
IGH)] or variants is necessary to confirm all but the most classic cases. However, in the cases of otherwise typical BL, in which an evident
MYC translocation cannot be detected by the standard procedures, the diagnosis of BL can still be made [
1]. Five to ten percent of BL cases show no translocation, both by classical cytogenetics and molecular methods like fluorescence
in situ hybridization (FISH) analysis [
2,
3]. This may be due to technical failure of FISH, as these cases may present with a very small excision of
MYC and insertion of the gene into one of the
IG loci, which is missed by the available probes [
4]. Another option is that the breakpoint is localized far outside the region covered by the currently available FISH probes [
4]. Even though none of the techniques currently used to diagnose genetic changes can unambiguously rule out all of
MYC translocations [
4], some observations suggest that mechanisms other than translocation are responsible for elevated MYC protein expression in BL even in the absence of genomic rearrangements [
5,
6]. Amplification, rearrangement or hypomethylation of the
MYC oncogene are genetic alterations frequently occurring in many cancers, as carcinoma of the cervix, colon, breast, lung and stomach [
7‐
11], and causing
MYC to be activated and over-expressed. Previous studies, by integrating structural and functional genomics to catalogue the broad of somatic mutations in BL [
12‐
14] have found that the most mutated gene in BL is
MYC itself (70 % of cases approximately). Moreover, there is increasing evidence that MYC protein over-expression may occur in tumors without apparent gene alterations [
15] and it has been suggested that a dysregulated expression of microRNAs (miRNAs) may represent one of the mechanisms leading to
MYC overexpression in BL cases lacking a classical
MYC translocation, through either a direct or indirect mechanism [
5,
6]. In recent years, lymphoma studies have uncovered various mechanisms by which miRNAs influence their target genes [
16] and it has become clear that alterations in the expression of miRNAs contribute to the pathogenesis of most, if not all, human malignancies [
17].
All the mechanisms leading to MYC over-expression, affect the expression of its downstream target genes that are involved in various cellular processes such as cell proliferation, cell growth, apoptosis, differentiation, and stem-cell self-renewal, presumably through DNA over-replication [
18]. In addition,
MYC amplifies the existing gene expression program and can also control global chromatin structure by regulating histone acetylation [
19].
Increasing information identifies other essential pathways that are activated in the pathogenesis of BL and highlights the fact that MYC translocation alone is insufficient to drive lymphomagenesis. Therefore BL cases lacking the typical MYC translocation, but expressing MYC at the protein level, may represent a good model for a more detailed description of MYC regulation. In this paper we investigated the microRNA profile of MYC translocation-positive and MYC translocation-negative BL cases in order to uncover possible differences at the molecular level. We found that MYC translocation-positive and -negative BL cases are slightly different in terms of microRNA and gene expression, and we validated our findings at the mRNA and protein levels. Interestingly, in MYC translocation-negative BLs we found over-expression of DNA methyltransferase (DNMT) family members, consistent to hypo-expression of hsa-miR-29 family. This finding suggests an alternative way for the activation of lymphomagenesis in these cases, based on global changes in methylation landscape, aberrant DNA hypermethylation, lack of epigenetic control on transcription of targeted genes, and increase of genomic instability. In addition, we observed the over-expression of another MYC family gene member, MYCN that may therefore represent an additional mechanism for malignant transformation.
Our findings may be helpful to explain the pathogenetic mechanisms of tumors in which overexpression of MYC is independent of a chromosomal translocation or a gene amplification.
Methods
Ethics
This study was approved by the ethics committee of the University of Siena, Italy and of Lacor Hospital, Uganda. Study participants or their legal guardians provided written informed consent.
Case selection
109 Burkitt lymphoma cases, enrolled in the International Network for Cancer Treatment and Research (INCTR) study on African BL, were used for this study. All cases were recorded in childhood and diagnosed as BL by an expert panel on histological slides stained with Haematoxylin and Eosin (H&E) and Giemsa, and by immunophenotyping, according to the WHO classification [
1,
20]. Ten cases did not show the typical t(8;14), t(8;2) and t(8;22)
MYC-translocations at FISH analysis (
MYC translocation negative in the following) by using both dual-fusion probes and split-signal probes for
IGH and Immunoglobulin light chain gene (
IGL)
loci as well as an LSI
IGH/
MYC CEP 8 Tri-color dual-fusion probe (Vysis, Abbott Molecular IL, USA). FISH analysis using
BCL2 and
BCL6 probes was also negative. All cases were otherwise completely typical in term of clinical presentation (age: median 7, range 4–10; female/male ratio: 4/6; nodal/extra-nodal ratio: 2/8), morphology and immunophenotype (CD10+, BCL6+, BCL2-, CD38+, CD44-, Ki-67 100 %) to make a diagnosis of BL.
The analysis of the EBV status was performed by
in situ hybridization for EBV-encoded RNA (EBER) as previously reported [
6]. In particular, 8/10
MYC translocation negative cases were EBV-positive, whereas the positivity to the virus was detected in 90 % of
MYC translocation positive cases.
Unfortunately, RNA extracted from formalin-fixed and paraffin-embedded (FFPE) material precluded next generation sequencing (NGS) studies in most cases, which was therefore performed only in one case, whose fresh tissue was available.
For gene expression analysis, RecoverAll™ Total Nucleic Acid Isolation Kit (Life Technologies, Carlsbad, California, USA) was used to extract total RNA from FFPE tissues. Up to five 10 μm sections were processed per reaction. FFPE samples were deparaffinised using a series of xylene and ethanol washes. Next, they were subjected to a rigorous protease digestion with an incubation time tailored for recovery of total RNA. RNA was purified using a rapid glass-fiber filter methodology that includes an on-filter DNAse treatment and were eluted into the low salt buffer provided. On the other hand, for miRNA analysis RNA was extracted from FFPE sections of primary tumors and reactive lymph nodes using the miRNeasy FFPE Kit (Qiagen, Milan, Italy), according to the manufacturer’s instructions.
The amount and quality of RNA were evaluated by measuring the OD at 260 nm and the 260/230 and 260/280 ratios using a Nanodrop spectrophotometer (Celbio, Milan, Italy). The quality of RNA was also checked using a Bioanalyzer 2100 (Agilent, CA, USA).
Next generation sequencing
High-throughput RNA sequencing produced about 66 million of 75 bp paired ends reads (theoretical coverage calculated on Ref Seq transcriptome 84X). Chromosomal translocations were detected using a bioinformatic pipeline that combines results from three different fusion-detection tools (deFuse, Chimerascan and Tophat Fusion) [
21‐
23] and filtered on non-tumor controls using previously sequenced control reactive lymph nodes.
MYC gene expression was estimated in one
MYC translocation
-negative sample and in other 21 endemic Burkitt lymphomas using the transcripts parts per million (TPM) calculation method [
24].
Single Nucleotide Variants (SNVs) and short insertions and deletions (Indels) were called using the Genome Analysis Toolkit (GATK) [
25] after mapping quality score recalibration and local realignment around indels. All of the mutations detected were filtered using tresholds based on quality, coverage and strand of the mapped reads and according to variants already present in public databases (Hapmap, dbSNP and 1000genome project) [
26]. The Annovar tool [
27] was used for functional annotation of variants, including exonic functions and aminoacid changes. All the mutations found in the
MYC gene, including variations in intergenic, intronic and UTR regions, were manually checked and explorated using the Integrative Genomic Viewer 2.03 (IGV) visualization tool [
28].
MicroRNA array profiling
MiRNA profiling was performed by an external facility (Exiqon, Copenhagen, Denmark). The samples were labelled using the miRCURY™ Hy3/Hy5 Power labelling kit and hybridized on the miRCURY™ LNA Array (5th Generation arrays, hsa, mmu and rno, Exiqon).
Raw data was then received and analyzed in our laboratories. Briefly, signals quantified by microarrays were processed with a normalization pipeline using MIDAS v2.22 software [
29]: bad channels (intensity values less than 1) were filtered prior to normalization, and all the spots with a signal/noise value less than 2 were marked as “bad” and excluded from analysis (background correction). Signals were normalized using the global Lowess (Locally weighted scatterplot smoothing) regression algorithm [
30] with a smooth parameter of 0,33, which has been found to produce the best within-slide normalization to minimize the intensity-dependent differences between the dyes. Statistical Analysis was performed using MeV v4.7.4 on a dataset including only human miRNA annotated on miRBase [
31]. Unsupervised hierarchical clustering on dataset was used on Pearson correlation of log2(Hy3/Hy5) intensities and all of the samples and miRNAs were clusterized using average linkage method. Principal Component Analysis (PCA) was also used to discriminate the different biological samples on the basis of the distances of a reduced set of new variables (Principal Components). Differentially expressed miRNAs between the two groups (
MYC translocation-positive
versus MYC translocation-negative) were identified with a two-tailed
T-test with Welch approximation for different variance among groups and with different stringency criteria for false discovery rate (adjusted Bonferroni correction and no correction). Results of the test were filtered considering as differentially expressed only miRNAs with adjusted
p-value less than 0,05 and fold change in absolute value greater than 1 [fold change = mean (group A) - mean(group B)].
Quantitative Real-Time Polymerase Chain Reaction (RT-qPCR)
Quantitative RT-PCR was performed to validate results of both miRNA and gene expression profiling, and to assess relative expression of
MYC in ten
MYC translocation positive and ten
MYC translocation negative cases. For validation of differentially expressed miRNAs identified by profiling, RNA samples were reverse transcribed using the Universal cDNA synthesis kit (Exiqon, Copenhagen, Denmark), according to the manufacturer’s instructions. RT-qPCR amplification was performed using microRNA LNA™ PCR primer sets (Exiqon, Copenhagen, Denmark) specific for hsa-miR-29a-b, hsa-miR-513a-5p, and hsa-miR-628-3p, and using hsa-Let-7c as a reference gene. Validation of genes potentially targeted by the differentially expressed miRNAs (DNA (cytosine-5)-methyltransferase 1 (
DNMT1), 3 alpha (
DNMT3A), 3 beta (
DNMT3B) was also carried out by RT-qPCR using FluoCycle SYBR green (Euroclone, Celbio, Italy) in 10
MYC-translocation positive and 10
MYC-translocation negative cases according to manufacturer’s instructions. Non-neoplastic lymph nodes were meant as a negative control;
HPRT was used as housekeeping gene. Primer sequences were designed using Primer-BLAST [
32] and are reported in Table
1. Differences in gene expression were calculated using the ΔΔCt method [
33].
Table 1
Primers used for RTqPCR. Primer sequences for DNMT1 amplified a region of 88 bp. Primers for DNMT3a amplified a region of 68 bp; Primers for DNMT3b amplified a region of 68 bp; Primers for MYC amplified a region of 129 bp; Primers for HPRT amplified a region of 191 bp
DNMT1-FORWARD | 5’-CGACTACATCAAAGGCAGCAACCTG-3’ |
DNMT1-REVERSE | 5’-TGGAGTGGACTTGTGGGTGTTCTC-3’ |
DNMT3A-FORWARD | 5’-TAT TGATGAGCGCACAAGAGAGC-3’ |
DNMT3A-REVERSE | 5’-GGGTGTTCCAGGGTAACATTGAG-3’ |
DNMT3b-FORWARD | 5’-GGCAAGTTCTCCGAGGTCTCTG-3’ |
DNMT3b-REVERSE | 5’-TGGTACATGGCTTTTCGATAGGA-3’ |
MYC-FORWARD | 5’-AGCGACTCTGAGGAGGAAC-3’ |
MYC-REVERSE | 5’-TGTGAGGAGGTTTGCTGTG-3’ |
HPRT-FORWARD | 5’-AGCCAGACTTTGTTGGATTTG-3’ |
HPRT-REVERSE | 5’-TTTACTGGCGATGTCAATAAG-3’ |
Immunohistochemistry
Immunohistochemistry analysis for MYC (Abcam; dilution 1:200), DNMT1 (BD Biosciences: dilution 1:50), DNMT3A (Abcam; dilution 1:100), DNMT3B (Imgenex; dilution 1:200) and NMYC (ThermoScientific; dilution:1:100) was performed on Bond III automated immunostainer (Leica Microsystem, Bannockburn, IL, USA), with controls in parallel. No epitope retrieval was exploited. Ultravision Detection System using anti-Polyvalent HRP (LabVision, Fremont, CA, USA) and diaminobenzidine (DAB, Dako, Milan-Italy) as a chromogen was employed. The expression level of the proteins was evaluated in the ten
MYC translocation-positive and ten
MYC translocation-negative cases used for the RT-qPCR analysis, to validate results. Immunoreactivity was assessed by two investigators and cases with discrepancy were re-viewed to obtain a concordance ratio of more than 90 %. It is noteworthy that the definition of MYC positivity by immunohistochemistry is not universally standardized. However, the literature reports that having at least 40 % of malignant lymphocytes with nuclear MYC expression is considered positive [
34]; therefore we used this cut-off to discriminate positive and negative cases. For DNMT1 and DNMT3A, the cut-off level was based on modified Choi et al. system considering only the proportion of neoplastic cells showing a nuclear positivity [
35]. The expression of DNMT1, DNMT3A and DNMT3B was considered absent/low if only 0–10 % of tumor cells were stained; intermediate whether the positivity was present in 11–50 % of neoplastic cells, and high when the immmunoreactive cells were >50 %. For N-MYC, only nuclear staining was considered positive with no cut-off level.
Discussion
BL is an aggressive B-cell lymphoma with a characteristic clinical presentation, morphology and immunophenotype [
1]. Over the past years, the typical translocation, involving the
MYC oncogene and its variants, has been considered the molecular hallmark of this tumor. However, transcriptional and genomic profiling aimed to distinguish BL
versus DLBCL revealed the existence of BLs without evident
MYC translocation clustering with molecular BL. A recent paper reported that BLs lacking
MYC translocation share a peculiar pattern of chromosome 11q aberration [
38]. The significantly lower expression of
MYC in such cases supported the view that
MYC is not genomically activated, and the clinical, morphologic, and molecular characterizations of these cases suggest that they represent a distinct subset of
MYC-negative high-grade B-cell lymphomas with features resembling but not identical to BL. Yet, these findings do not explain the mechanisms through which some classic BL cases lack the typical genetic translocation involving
MYC but do express
MYC at the mRNA and the protein level [
5,
6]. Dysregulation of
MYC expression may be due to additional mechanisms, other than common genomic abnormalities, such as a miRNA imbalance [
39,
40]. So far, no data is available concerning the miRNA profile of
MYC translocation-negative cases, besides the evidences previously reported by our group [
5,
6]. In this study, we further explored the miRNA profile of BLs carrying or not the classical translocations involving the
MYC gene.
Interestingly, when we compared the miRNA profiling of MYC translocation-positive versus MYC translocation-negative BL cases, we identified four miRNAs differentially expressed, of which hsa-miR-513a-5p and hsa-miR-628-3p were up-regulated and two miR-29 family members (hsa-miR-29a and hsa-miR-29b) were down-regulated in BL cases lacking the MYC translocation.
Of note, microarray-based miRNA analysis turned out to be quite specific and robust in this study. In fact, all of the genes tested were successfully validated by RT-qPCR.
Hsa-miR-628-3p and hsa-miR-513a-5p are less referred in the literature, whereas, more is known about the miR-29 family [
41]. Interestingly, miR-29 family members have been related to malignant transformation, and it has been demonstrated that their down-regulation contributes to MYC-induced lymphomagenesis
in vivo and
in vitro models [
42,
43]. Thus, hsa-miR-29 family members down-regulation may represent an appealing possible mechanisms able to determine MYC up-regulation and sustain its expression at mRNA and protein level also in the absence of a translocation. Interestingly, a link between the miR-29 family by
MYC has been recently reported [
44], as repression of miR-29 by
MYC through a corepressor complex with HDAC3 and EZH2 is observed in aggressive B-cell lymphomas [
43]. This miRNA family may represent a novel target for tailored therapies as
in vitro and mouse studies suggest increasing miR-29 expression by combined inhibition of HDAC3 and EZH2. Such an approach could help treat
MYC-overexpressing cancers [
44]. In addition, it has been recently demonstrated that hsa-miR-29b directly binds to
DNMT3A and
DNMT3B, and regulates indirectly
DNMT1 by targeting Sp1, a transactivator of the gene [
36,
45]. In this scenario, over-expression of
DNMT family members, due to hypo-expression of hsa-miR-29 family members, may elicit a role in inducing carcinogenesis [
46]. The finding that DNMTs were up-regulated in
MYC translocation-negative BLs suggests an alternative way for the activation of lymphomagenesis in these cases, based on global changes in methylation landscape and loss of epigenetic control. Hsa-miR29a may favor this process by a synergistic hypermethylating effect [
47]. In this regard, future studies exploring the global methylation patterns of BL with or without
MYC translocation are definitely warranted.
We were also intrigued by the observation that another member of the
MYC family,
MYCN, was potentially dysregulated in BL cases lacking
MYC translocation. Literature reports that MYC and N-MYC possess similar ability to induce cell proliferation and transformation although MYC may be more effective in some contexts
. Over-expression of specific
MYC family genes is frequently associated with particular types of human tumors [
4];
MYCN deregulation is almost exclusively associated with solid tumors and only rarely observed in lymphomas. Nonetheless, both N-MYC and MYC are expressed in pro-B cells, and it has been demonstrated that N-MYC can support normal B-cell development in the absence of MYC [
48‐
50]. Over-expression of either MYC or N-MYC under the control of the B cell-specific Eμ enhancer results in development of pro-B cell lymphomas [
51]. Finally, complex
MYCN/
IGH translocations frequently arise in mice deficient for p53, showing that, in this genetic background, the endogenous N-MYC can compete with MYC as a pro-B cell oncogenic translocation/amplification target [
52]. Based on our findings (i.e. over-expression of N-MYC at the mRNA and protein levels in
MYC translocation-negative cases) one should hypothesize that in BL cases lacking
MYC translocation N-MYC may represent an alternative cooperating mechanisms in contributing to malignant transformation. Interestingly, two of the differentially expressed miRNAs (miR-513a-5p and miR-628-3p) have been recently reported dysregulated in human neuroblastomas, in which aberrant expression of
MYCN is quite common [
53,
54]. Of note, miR-628-3p expression seems even to correlate with tumors prognosis in such cases [
55]. Altogether this observation suggests that
MYCN aberrant expression itself may impact gene and microRNA expression pattern in BL cases lacking the typical
MYC translocation. A large body of evidence has documented the existence of an active cross-talk between
MYC itself and miRNAs machinery, suggesting the existence of a feedback loop between
MYC and specific miRNAs [
56]. This, in turn, might be the cause of a differential gene expression and of functional alterations of neoplastic cells [
40]. The difference in has-miR29 family members expression we detected between MYC translocation-positive and MYC-translocation negative BL samples might be related to the lower MYC protein level among cases lacking the MYC-translocation.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
DFG, AMR, LL and PPP conceived and designed the experiments; OA, FF, ML, GS. performed the experiments; DFG, AMR, FF analyzed the data; BC, NM, EM, RBJ, contributed reagents/materials/analysis tools; DFG, LL, AMR and PPP draft the paper, LL and SAP were responsible for funding. All authors read and approved the final manuscript.