MicroRNA expression in Epstein-Barr virus-associated post-transplant smooth muscle tumours is related to leiomyomatous phenotype

Epstein-Barr virus (EBV)-associated post-transplant smooth muscle tumours (PTSMT) are rare complications. In our previous molecular analysis, we have evaluated the expression of regulatory microRNA which are known to be EBV-related (miR-146a and miR-155) but found no deregulation in PTSMT. In this current analysis, we aimed to characterize the expression profiles of several hundred microRNA. Tissue samples from PTSMT and uterine leiomyomas were analysed by quantitative real-time PCR for the expression of 365 mature microRNA. PTSMT and leiomyomas share a highly similar microRNA profile, e.g. strong expression of miR-143/miR-145 cluster and low expression of miR-200c. Among EBV-related microRNA (miR-10b, miR-21, miR-29b, miR-34a, miR-127, miR-146a, miR-155, miR-200b, miR-203 and miR-429) only miR-10b and miR-203 were significantly deregulated. The expression pattern of microRNA in PTSMT is not associated with EBV infection but reflects the leiomyomatous differentiation of the tumour cells.


Introduction
Epstein-Barr virus (EBV)-associated diseases are often associated with acquired or congenital immunosuppression or immunodeficiency, e.g. bone marrow and solid organ-transplanted patients are at a higher risk. Up to 10% of transplant recipients develop post-transplant lymphoproliferative disorders (PTLD) while EBV-associated post-transplant smooth muscle tumours (PTSMT) are rare complications (<1% of transplant patients) [1,2]. Neoplastic spindle cells in PTSMT express leiomyogenous marker proteins such as smooth muscle actin and desmin, and the majority of tumour cells is positive for EBER. We could previously show that PTSMT differ from conventional leiomyosarcomas by their lack of marked atypia, unusual sites of involvement (>50% in the recipient or donor liver) and defined EBV association [2].
The molecular pathobiology of this rare neoplastic entity is not fully understood and only few experimental analyses have addressed this issue. Ong et al. [3] have analysed cell cycle factors, cytokines and gene promoter methylation in PTSMT and found an activated mTOR/ Akt cell cycle pathway by demonstrating phosphorylated mTOR in tumour cells. In our previous analysis, we have evaluated the expression of EBV-associated human genes in PTSMT, including transcription, cell cycle and apoptosis factors and cytokines/cytokine receptors [2]. We found that the transcription factor v-myc myelocytomatosis viral oncogene homolog (avian) (MYC) is significantly upregulated in PTSMT. In addition to mRNA, we have analysed micro-RNAs which are known to be expressed in an EBV-related fashion (miR-146a and miR-155) but in PTSMT we found low expression levels and no delimitable deregulation. MicroRNAs are non-coding RNA molecules of 20-25 nucleotides in length [4,5]. These small RNA molecules can bind semi-complementarily to the 3′-untranslated region (3′-UTR) of target mRNAs and repress translation or target mRNA for degradation. The microRNA genes can be present as single gene or gene clusters (different microRNA species are encoded on the same chromosome segment). Furthermore, microRNA families represent different microRNA genes with different precursor forms but very similar mature microRNA with no or minor differences in their nucleotide sequence.

Tissue specimens
All available PTSMT samples from our tissue archive were evaluated; these PTSMT cases have been characterized earlier [2]. Five EBV + PTSMT samples from four patients, including two tumours from one patient (#4) were analysed (Additional file 1: Table S1). Controls: seven EBVbenign uterine leiomyomas. Formalin-fixed and paraffin-embedded (FFPE) samples were retrieved from the archives of the Institute of Pathology (Hannover Medical School/MHH, Hannover, Germany). The retrospective analysis of the samples has been approved by the local ethics committee of the Hannover Medical School (MHH).

Laser microdissection of the PTSMT compartment and gene expression analysis
Tissue from FFPE blocks with >90% tumour cells were cut and processed for further PCR analysis. In blocks with <90% aberrant cells, the PTSMT compartments were laser microdissected using a SmartCutPlus-System (MMI, Glattbrugg, Switzerland), as previously described [2].
A set of 365 mature microRNA and corresponding endogenous controls were analysed by quantitative real-time PCR (Pool A, Applied Biosystems, Carlsbad, CA, USA). In brief, cells were digested in proteinase K and RNA was extracted with phenol/chloroform [2,18,19]. Synthesis of cDNA from microRNA, subsequent pre-amplification of cDNA and real-time quantitative PCR with a 7900HT Fast Real-Time PCR system were performed according to the manufacturers' instructions (Applied Biosystems).

Data analyses
The sample-and detector-specific evaluation of amplification curves was accomplished with the software RQ Manager 1.2 (Applied Biosystems). C T values established in this manner were converted into ΔC T values and into 2 -ΔCT values (normalized to mean of endogenous control genes). Statistical analysis was performed with Prism 5.0 (GraphPad Software, San Diego, California, USA) by applying the Mann-Whitney test for two-group comparison. P values < 0.05 were considered as statistically significant.

Similar microRNA expression profile in PTSMT and leiomyomas
Cluster analysis of the expression profile of 365 micro-RNA revealed that PTSMT and leiomyomas share a highly similar profile (Additional file 2: Table S2); cluster analysis could not discriminate between the two tumours.

Leiomyomatous phenotype-associated microRNA expression in PTSMT
MicroRNA expression analyses in uterine leiomyomas and leiomyosarcomas were introduced only a few years ago (Additional file 4: Table S4) [6][7][8][9][10][11][12][13][14][15][16][17]. The majority of studies have evaluated patient-derived leiomyomas as a model disease of neoplastic smooth muscle proliferation. Among different studies and different analytical methods, microRNA expression patterns of patient-derived tumour samples showed a set of microRNA which are recurrently deregulated in comparison to normal uterus wall cells. Similar to PTSMT, decreased expression of miR-150, miR-200c and miR-221 and increased expression of miR-21 and let-7 family members have become particularly evident in leiomyomas [6][7][8][9][10][11][12][13][14][15][16][17]. It has been demonstrated that mesenchymal cells which differentiate into smooth muscle cells in vitro change their microRNA expression patterns, e.g. down-regulation of 13q31.3-clustered miR-17/miR-18a/miR-20a and up-regulation of miR-181a/miR-181c paralogs [16]. Furthermore, in smooth muscle cells, it has been shown that the 5q32-encoded miR-143/miR-145 cluster is co-expressed [38,39]. Both microRNA target a network of factors to promote vascular smooth muscle cell differentiation and repress proliferation [38,39]. We could previously demonstrate high expression of miR-143 and even higher expression of miR-145 in pulmonary vessel wall cells [40]. A similar expression pattern of high miR-143 and higher miR-145 could also be found in PTSMT but also in leiomyomas. Because PTSMT can be found next to vessels (e.g. manifestation in cerebral sinus), it is thought that the aberrant founder cells might be derived from a vessel wall [2,41]. However, due to very similar miR-143/miR-145 expression patterns in PTSMT, uterus wall-derived leiomyomas and pulmonary vessels, the high expression of these two microRNA does not prove a vessel wall origin of PTSMT but reflects the smooth muscle differentiation.
In PTSMT and leiomyomas, many microRNA are expressed at low or very low levels, which makes it likely that protein translation of potential target mRNA types is not inhibited. The problem for target prediction, but simultaneously the hallmark of microRNA/mRNA biology, is the characteristic semi-complementary binding of the seven nucleotides at the 3′-end of the mature microRNA (so-called seed sequence) to corresponding mRNA-nucleotides of the 5′-UTR [4,5]. This semicomplementary binding is sufficient to induce a biological effect, the inhibition of mRNA/protein translation. As a result, one microRNA can bind to several 5′-UTR-mRNA and vice versa one 5′-UTR-mRNA can be targeted by several microRNA. In our previous analysis, we have evaluated the expression of several mRNA transcripts in PTSMT and leiomyomas, including MYC, vascular endothelial growth factor A (VEGFA), nuclear factor of kappa light polypeptide gene enhancer in B-cells 1 (NFKB1), tumour protein p53 (TP53), transforming growth factor, beta receptor II (TGFBR2) and transforming growth factor, beta 1 (TGFB1) [2].
Among several EBV-associated human factors, we found only MYC to be significantly increased in PTSMT [2]. The mRNA of this transcription factor can be regulated by miR-150, miR-143 and miR-145 [42,43] but no PTSMT-specific inverse correlation was found.
VEGFA can be negatively regulated by miR-200c and other microRNA. In leiomyomatous cell lines, miR-200c interaction with VEGFA has been shown [7] and accordingly, in leiomyomas as well as in PTSMT, very low levels of miR-200c correlate with increased levels of VEGFA [2].
In many different tumour types, miR-21 is aberrantly expressed, because this microRNA can target several signal networks, either directly by binding to different types of mRNA from similar signal cascades or indirectly via deregulation of factors down/up-stream to the factors directly suppressed by miR-21 [44,45]. In particular, miR-21 is a negative regulator of TP53 signalling and simultaneously a promoter of NFKB1 signalling [44,45]. Increased expression of miR-21 has been previously demonstrated in leiomyomas [6,10,15,16] which we could confirm in our analysis. In smooth muscle cells, miR-21 is involved in regulation of apoptosis and TGFBR2/TGFB1 signalling [6,10,13]. TGFBR2-3′UTR has an miR-21 binding site and can therefore directly be regulated by miR-21 in smooth muscle cells [10]. In vitro studies also suggested an indirect regulatory interaction between miR-21 and TGFB1; of note, TGFB1 is not a direct target of miR-21 [10]. Furthermore, inhibition of miR-21 expression in smooth muscle cells indirectly increases caspase 3 and caspase 7 activity in vitro; both caspases have no miR-21 binding site [6,13]. The miR-21 expression was lower in PTSMT than leiomyomas but the difference was not significant. In addition, in our previous analysis we found no differences in the expression of miR-21-related NFKB1, TP53, TGFBR2 or TGFB1 between PTSMT and leiomyomas [2]. Therefore, our in situ-derived results do not reveal a PTSMT-specific deregulated miR-21 signal cascade, but an expression pattern related to smooth muscle phenotype.
Members of the let-7 family are increased in leiomyomas [15] and smooth muscle cell lines, e.g. let-7b [13,16]. We found that in PTSMT and leiomyomas, the strongest expressed let-7 paralog was let-7b. In leiomyomas and leiomyosarcomas, let-7c shows an inverse correlation with its target mRNA high mobility group AT-hook 2 (HMGA2) [11,17]. Both genes, let-7c and HMGA2, are expressed in association with size of leiomyomas [11], while in PTSMT, irrespective of the size, let-7c was almost absent.
In summary, in addition to leiomyomas and leiomyosarcomas, PTSMT is the third smooth muscle tumour type in which the microRNA expression profile could be evaluated. The expression pattern of microRNA in PTSMT is not associated with EBV infection (presumably due to lack of strong LMP1 expression) but reflects the leiomyomatous differentiation of the tumour cells.