Abstract
Background: MicroRNAs are regulators of gene expression implicated in vital cellular processes including differentiation, cell growth and apoptosis. Distinct microRNA signatures have been identified for many malignancies including follicular lymphoma (FL). However, no microRNA expression profile characteristic of FL subtypes, e.g. FL with B-cell lymphoma-6 (BCL6) locus rearrangement (FLBCL2+/BCL6+, FLBCL2−/BCL6+) or FL with diffuse growth pattern have been reported. Materials and Methods: MicroRNA signatures from 44 cases of FL were generated employing a quantitative real-time polymerase chain reaction approach. 15 cases of diffuse FL and 15 cases of FLBCL2+/BCL6+/FLBCL2−/BCL6+ were compared against 14 cases of typical FLBCL2+/BCL6−. Results: Numerous microRNAs were found to be differentially expressed between FLBCL2+/BCL6+ and FLBCL2−/BCL6+, as well as diffuse FL, when compared to typical cases of FL. Up-regulation of several oncogenic microRNAs as well as down-regulation of tumor-suppressor microRNAs was identified. Cluster analysis, however, revealed no microRNA signatures distinct from the reference group for either subtype. Conclusion: These results indicate an involvement of microRNAs in the pathogenesis of FL and its subtypes. Marked de-regulation of oncogenic RNAs and tumor suppressors appears to correspond with a more aggressive phenotype frequently observed in FLBCL2+/BCL6+, FLBCL2−/BCL6+ and diffuse FL.
Chromosomal translocations are common events in B-cell non-Hodgkin's lymphomas; in particular translocations involving an oncogene and one of the immunoglobulin genes, associating the former with the promotor of an enhancing locus, are common (1, 2). For certain types of lymphomas, the occurrence of a specific chromosomal aberration is considered to be essential for tumor development e.g. t(14;18)(q32;q21) involving the B-cell lymphoma-2 gene (BCL2) and immunoglobin (Ig) loci in follicular lymphoma (FL).
FL constitutes the second most common type of B-cell non-Hodgkin's lymphoma (3). Its clinical course is usually indolent, with median overall survival ranging between eight and ten years. FL is a disorder most common among the elderly, with patients usually presenting in the seventh decade of life for initial diagnosis (4, 5). Recent advances with regard to more targeted-therapy approaches have, however, further improved outcome in patients with FL (6). In addition to t(14;18)(q32;q21) as the molecular hallmark of FL, chromosomal rearrangements affecting the BCL6 locus constitute one of the most common cytogenetic findings, present in 5-15% of cases (7, 8). The impact of these aberrations on clinical course and outcome remains a subject of dispute, as Jardin et al. described a morphologically- and immunophenotypically-distinct BCL6 rearranged subtype of FL without the common t(14;18) (FLBCL2−/BCL6+) feature which appeared not to differ from typical cases of FL with regard to overall survival (9). Akasaka et al., however, postulated that patients harboring a BCL6 rearrangement at the time of diagnosis, including those with concurrent BCL2 translocation (FLBCL2+/BCL6+) had significantly elevated rates of secondary progression to diffuse large B-cell lymphoma which has been repeatedly associated with poor survival rates and shorter time-to-progression (10, 11). Another feature diversifying the spectrum of FL are the different types of growth patterns observed, ranging from strictly follicular to diffuse proliferation of neoplastic cells (without remaining follicular dendritic cell networks visualized by immunohistochemical staining for the “low-affinity“ receptor for IgE (CD23)) (12). While Harris et al. describe a significantly shortened overall survival for patients with diffuse FL, Klapper et al. detected a similar trend which failed to reach statistical significance (3, 13).
In the recent past, microRNAs have been identified as key elements in the regulation of gene expression (14). This species of small non-coding RNAs appears to interfere with numerous biological processes, including differentiation, cell growth, apoptosis and pathogenesis of malignancies, by negatively-regulating one or several target genes through translational repression or target degradation (15, 16). Quantitative expression of microRNAs was revealed to differ between normal and neoplastic tissues and it is currently postulated that distinct microRNA signatures exist for all types of tissues and tumors. To date, insights into the specific microRNA expression profile of FL remain fairly limited. Yet recently, our group and Wang et al. independently proposed an initial draft of a microRNA signature for FL (17, 18).
The aim of the present study was to determine particularities of microRNA signatures in a comparative approach using quantitative real-time polymerase chain reaction (qRT-PCR) investigating cases of FL with diffuse growth pattern and cases of FL harboring chromosomal translocations affecting the BCL6 locus. The current study group constitutes a subgroup of a cohort of previously published cases of FL that was recently examined for associations between cytogenetic, morphological and immunophenotypic features at our institution (19). Both subtypes were contrasted with a reference group composed of typical FL with BCL2 translocation lacking BCL6 translocations (FLBCL2+/BCL6−) from the same cohort.
Materials and Methods
Tumor samples. Ten cases harbored both the characteristic t(14;18)(q32;q21) as well as an additional translocation affecting the BCL6 locus and five cases had only the latter aberration. Fifteen samples were characterized as FL with a significant diffuse growth pattern (exhibiting diffuse proliferation in a fraction of 25% or higher) and 14 cases exhibited all typical morphological, immunophenotypic and cytogenetic characteristics of FL (t(14;18)(q32;q21) but no aberration affecting BCL6, of grade I or II, and with follicular growth pattern. The latter group served as a reference group for further analysis. All samples were collected as part of standard clinical care, and all studies were approved by the Ethics Commission (ID: 08-134) at the University of Luebeck.
Immunohistochemistry. Immunophenotypic analysis was performed as described elsewhere (20). In addition to hematoxylin and eosin (HE), Giemsa, Gomori and periodic acid-Schiff (PAS) stains, immunohistochemical stains for CD3, CD4, CD8, CD10, CD20, CD23, CD68, IgD, BCL2, BCL6, kappa and lambda light chains, and Ki67 (MIB1) were evaluated. Staining results were evaluated qualitatively for CD10, CD20, CD23 (on tumor cells), BCL2, BCL6, kappa and lambda, whereas CD3, CD4, CD8, CD68, Ki67 and IgD staining was detected quantitatively as described elsewhere using antibodies obtained from Lab Vision, Thermo Fisher Scientific Inc. (Fremont, CA, USA) and Dako (Glostrup, Denmark) (21, 22).
Fluorescence in situ hybridization (FISH) for BCL2 and BCL6. Chromosomal breakpoints were analyzed by means of FISH using dual-color break apart probes for 18q21 (BCL2) and 3q27 (BCL6) (Abott Vysis, Des Plaines, IL, USA) as described elsewhere (19).
Nucleic acid isolation and cDNA synthesis. Total RNA was isolated from four 20-μm sections of formalin-fixed, paraffin-embedded tissues using the Ambion Recover All kit (Ambion, Austin, TX, USA) according to the manufacturer's protocol. In order to assess microRNA expression, a total of 10 ng of RNA was reverse-transcribed using the TaqMan® universal PCR master mix, No AmpErase® UNG-kit and the TaqMan® microRNA reverse transcription kit from Applied Biosystems (Foster City, CA, USA). All reactions were diluted 1:60 and stored at −20°C.
qRT-PCR. Primers and probes for quantification of microRNA expression were obtained from Applied Biosystems. Expression signatures of 376 microRNAs were generated employing a qRT-PCR approach on a Roche LightCycler 480 System (Roche, Mannheim, Germany), according to the manufacturer's instructions.
Statistical analysis. MicroRNA expression data were normalized using quantile normalization. In order to determine statistically significant differences in the expression of microRNAs between two groups, the nonparametric Mann–Whitney U-test was applied. Due to the highly explorative nature of the current study, no adjustment of the significance level according to the Bonferroni method was performed and results were considered significant with p-values of 0.05 or less (two-sided). All tests were performed using SPSS ver. 19 (IBM, Armonk, NY, USA). Cluster analysis was performed according to the average linkage method using Cluster 3.0 by Michael Eisen (http://rana.lbl.gov/EisenSoftware.htm).
Results
Clinicopathological features of FLBCL2+/BCL6+, FLBCL2−/BCL6+ and DFL compared to the reference group. Comprehensive morphological, immunophenotypic and cytogenetic features were previously published (19). Briefly, FLBCL2+/BCL6+ & FLBCL2−/BCL6+ differed significantly from typical FLBCL2+/BCL6− concerning four parameters. The extension of residual follicular dendritic cell networks (visualized by CD23 staining; p=0.002), as well as immunohistochemical reactivity for CD10 (p=0.017) were significantly reduced. Proliferative activity (as indicated by both Ki67 staining in ‘hot spots’, as well as overall), however, was significantly increased in cases harboring 3q27 rearrangements (p=0.024 and p=0.016). Diffuse FL exhibited a significantly decreased immunoreactivity for CD10 (p=0.002) and BCL2 (p=0.017), as well as increased CD23 reactivity (p=0.035), and, by definition, decreased residual follicular dendritic cell networks (p=0.001). Moreover, diffuse FL revealed a significantly lower rate of bone marrow involvement (p=0.025).
MicroRNA signatures in follicular lymphomas with diffuse growth pattern. MicroRNA expression analysis revealed a subset of 25 (25/376) microRNAs significantly altered between diffuse FL and the reference group of typical FLBCL2+/BCL6. Fourteen microRNAs exhibited an at least two-fold de-regulation (ΔCT >1). Out of these, five were shown to be significantly overexpressed in diffuse FL, whereas we identified nine microRNA to be down-regulated (Table I).
MicroRNA signatures in FLBCL6+. Comparative microRNA signature analysis revealed a subset of 30 (30/376) microRNAs to be differentially expressed to a statistically significant extent between FL harboring chromosomal rearrangements affecting BCL6 (FLBCL2+/BCL6+ are FLBCL2−/BCL6+) and the reference group of typical FLBCL2+/BCL6−. A total of 21 microRNAs exhibited an at least two-fold de-regulation (ΔCT>1). Out of these, 12 microRNAs were shown to be significantly up-regulated in FLBCL6+ cases and nine were down-regulated in the same cohort (Table II).
Next, we compared BCL6-rearranged cases of FL not harboring BCL2 translocations with the reference group in order to assess BCL6 translocation-specific impact on microRNA signatures. We found increased expression of nine microRNAs and reduced expression levels of six microRNAs (Table III).
Hierarchical clustering analysis of microRNA expression data. Unsupervised cluster analysis based on the average linkage method was performed, incorporating qRT-PCR data of microRNAs differentially expressed between the different subgroups and the reference group. Despite there being significant disparities for several microRNAs between FLBCL2+/BCL6+, FLBCL2−/BCL6+ and diffuse FL compared to the reference group (see Tables I, II and III), the cluster analyses did not separate either morphologically (diffuse FL) or cytogenetically (FLBCL2+/BCL6+, FLBCL2−/BCL6+) defined groups.
Discussion
In the light of the widespread estimation that a distinct microRNA expression signature characterizes most, if not all, neoplastic disorders, insight into the correlation between cancer-related microRNA expression levels and the cytogenetic and ultimately morphological phenotype of FL remains profoundly-constrained. From a merely genomic perspective, FL is predominantly characterized by the translocation t(14;18), resulting in de-regulated BCL2 expression that is considered an initiating event in FL lymphomagenesis. Translocations of the BCL6 oncogene or up-regulation of inhibitors of apoptosis other than BCL2 have been described as additional or alternate triggering events (23, 24). The generation of microRNA profiles established in the current study was aimed at the identification of characteristics unique to FLBCL2+/BCL6+, FLBCL2−/BCL6+ and diffuse FL.
MicroRNA expression signatures vary greatly throughout B-cell maturation from naïve B-lymphocytes and germinal center-associated stages to plasma cells and post-germinal memory cells (25, 26). Results obtained in our current study appear equivocal with regard to these previously published findings, at least in part. Although the majority of differentially expressed microRNAs support there being a late-germinal center microRNA phenotype, we found a considerable number of microRNAs deviating from this expectation in FLBCL2+/BCL6+, FLBCL2−/BCL6+ and diffuse FL (Tables I and II).
Diffuse FL were shown to differ significantly from cases of typical FL with regard to the expression of individual microRNAs. Several microRNAs previously characterized as so-called oncomiRs were shown to be expressed at elevated levels e.g. hsa-miR-21, hsa-miR-424 and hsa-miR-452. Moreover, numerous microRNAs previously found to possess tumor-suppressive activity were detected at significantly repressed levels: e.g. hsa-miR-133a, hsa-miR-139 and hsa-miR-148a. These findings are well in line with published results hinting at a more aggressive phenotype of diffuse FL when compared to typical cases of FL (13). Despite these substantial contributions to our understanding over the pathogenesis of diffuse FL derived from comparative microRNA expression analysis, we detected multiple de-regulated microRNAs with divergent reported functions with regard to cell-cycle regulation (27).
The aberrantly elevated expression of has-miR-9 appears particularly curious in the context of its opposing effects reported in different types of neoplasias. Elevated levels have so far only been described for germinal center-derived lymphomas, whereas down-regulation was found in several types of cancer, including melanoma and renal cell carcinoma (17, 18, 28, 29). Functional investigations associating has-miR-9 expression with suppression of nuclear factor-κB signaling seem to be in keeping with these findings (30). In the light of these observations and recent findings by Jima et al. (21), we emphasize on the possibility of elevated hsa-miR-9 expression merely reflecting the late germinal center origin of diffuse FL and not being the overexpression of a tumor-suppressor microRNA.
Other microRNAs with equivocal functions and implication in malignancies found to be differentially expressed in our study include hsa-miR-135a, an inhibitor of both the adenomatous polyposis coli (APC) and the Janus-kinase 2 (JAK2) genes, and hsa-miR-184, a potent tumor-suppressor in cholangiocellular carcinoma but at the same time a supposed oncomiR in head and neck squamous cell carcinoma (31-34).
In order to assess the impact of BCL6 gene rearrangements on microRNA signatures, we performed a comparative analysis of three groups consisting of typical FLBCL2+/BCL6− (n=14), FLBCL2+/BCL6+ (n=10) and FLBCL2−/BCL6+ (n=5). The latter two were subsequently regrouped for initial investigations. Out of the initially identified 30 microRNAs, 21 exhibited significant aberration by at least two-fold. A subset of 12 microRNAs was revealed to be significantly overexpressed in BCL6-rearranged FL: hsa-miR188, hsa-miR202, hsa-miR381, hsa-miR433, hsa-miR501, hsa-miR572, hsa-miR584, hsa-miR596, hsa-miR612, hsa-miR302c, hsa-miR373# and hsa-miR526b. In this group, we identified multiple tumor suppressors, as well as putative oncogenic microRNAs.
The most striking deregulation was observed for hsa-miR-612, an insufficiently characterized microRNA reported to be down-regulated in HL-60 cell reacting to treatment with all trans-retinoic acid and leading to reinstatement of physiological differentiation (35). On the other hand, more recent findings indicate its strong tumor-suppressive function in hepatocellular carcinoma (36). Hsa-miR-302c as another strongly expressed microRNA in FL harboring BCL6 translocations, was recently found to repress differentiation and assist in the acceleration of cell-cycle progression of cancer cells (37).
Nine microRNAs were significantly down-regulated in FL cases harboring BCL6 translocations: hsa-miR33, hsa-miR138, hsa-miR153, hsa-miR184, hsa-miR224, hsa-miR375, hsa-miR135a, hsa-miR200a and hsa-miR517c. Within this group, we again identified both putative tumor suppressors and oncomiRs.
Hsa-miR200a was among the most significantly down-regulated microRNAs in our analysis and was previously found to be a potent tumor-suppressor, mediating tumor growth restriction via the Wingless-type MMTV integration site family, member 1/β-catenin pathway (38, 39).
Much like diffuse FL, FLBCL2+/BCL6+ and FLBCL2−/BCL6+ were also characterized by a significant down-regulation of hsa-miR-135a and hsa-miR-184, with elusive pathogenetic implications.
Overall, the comparative analysis of microRNA signatures remains largely inconclusive with regard to their effect on biologically aggressive behavior in FLBCL2+/BCL6+ and FLBCL2−/BCL6+, only slightly favoring an induction of oncogenic microRNAs and a down-regulation of tumor suppressors. Thus, our results appear to be in keeping with published findings associating BCL6 gene rearrangements with a slightly inferior prognosis (with variable degrees of statistical significance in the literature) mostly due to higher rates of transformation to diffuse large B-cell lymphoma in FL.
In a second step, we compared BCL6-rearranged cases of FL without BCL2 translocations with the reference group in order to assess the BCL6 translocation-specific impact on microRNA signatures (see Table III) and found significantly induced expression of nine microRNAs, most of which are currently considered to possess oncogenic potential: hsa-miR21, hsa-miR197, hsa-miR205, hsa-miR432, hsa-miR612, hsa-miR193b, hsa-miR302b, hsa-miR519a and hsa-miR525# (40-42). Additionally, we found significantly reduced expression of six microRNAs predominantly implicated in tumor-suppressive processes” hsa-miR28, hsa-miR153, hsa-miR184, hsa-miR597, hsa-miR302a and hsa-miR30a-3p (43-45).
As our results did not enable reliable separation of cytogenetically- or morphologically-defined subgroups of FL by means of unsupervised cluster analysis, it appears likely that the use of microRNA signatures alone does not possess sufficient power for the differential diagnosis of these sub-entities but may very well complement established methodologies for more refined approaches to molecular classification of FL.
Despite the substantial advances in the comprehension of the impact of microRNA expression patterns on the morphological and clinical phenotype of different FL subtypes, we believe that our current findings necessitate further studies on a more extensive cohort of FL and its various subtypes in order to broaden our understanding over the microRNA-mediated molecular differences, potentially leading to more refined and targeted-therapy approaches.
Acknowledgements
We thank Annette Aufseß, Biggi Steinfeldt and Tanja Oeltermann for their skilled and dedicated technical assistance.
- Received January 30, 2014.
- Revision received March 6, 2014.
- Accepted March 7, 2014.
- Copyright© 2014 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved