Epigenetic silencing of a long non-coding RNA KIAA0495 in multiple myeloma

While the putative KIAA0495 promoter region was found embedded in a CpG island (Additional file 2: Figure S1), MSP primers were designed to study methylation of this CpG island in a panel of healthy controls [peripheral (N = 10) and marrow (N = 3) buffy coat, CD138-sorted healthy plasma cell (N = 1)], and myeloma cell lines (N = 10) (Table 1). Direct sequencing analysis of M-MSP products from bisulfite-treated positive control showed expected conversion of unmethylated cytosine into uracil (turned to thymidine after PCR), with methylated cytosine remained unchanged, indicating complete bisulfite conversion and MSP specificity (Fig. 1a). None of 14 healthy controls showed KIAA0495 methylation (Fig. 1b). On the other hand, in myeloma cell lines, KMS-12-PE, LP-1, NCI-H929, OPM-2, and OCI-MY5 were partially methylated, whereas MOLP-8, RPMI-8226, U-266, WL-2, and JJN-3 were completely unmethylated for KIAA0495 (Fig. 1c). Moreover, the MSP methylation statuses of controls and cell lines were confirmed by quantitative bisulfite pyrosequencing (Fig. 1d). These data suggested that methylation of KIAA0495 was tumour-specific, consistent with other tumour suppressor protein-coding genes and non-coding miRNAs in myeloma [17‐20].

To determine whether KIAA0495 methylation was associated with silencing of KIAA0495 expression, the expression level of KIAA0495 was measured by semi-quantitative RT-PCR and quantitative real-time PCR (qPCR), and correlated with the KIAA0495 methylation status as detected by MSP in myeloma cell lines. Of the 10 myeloma cell lines, KIAA0495 methylation was associated with lower expression as detected by both semi-quantitative RT-PCR (P = 0.0079; Fig. 2a) and quantitative real-time PCR (P = 0. 000155; Fig. 2b).

Moreover, myeloma cell lines with KIAA0495 methylation were treated with a hypomethylating agent, 5-AzadC, followed by pyrosequencing and qPCR analyses. Upon 5-AzadC treatment, LP-1 and OCI-MY5 cells, which were partially methylated for KIAA0495, showed progressive demethylation of the KIAA0495 promoter, as evidenced by the decreased methylation percentage on a stretch of seven consecutive CpG dinucleotides, together with concomitant re-expression of the KIAA0495 transcript, as illustrated by the increased expression relative to the untreated control (LP-1: P = 0.0013; OCI-MY5: P = 0.0013; Fig. 2c). Nevertheless, when 5-AzadC-treated cells were continually cultured with fresh medium in the absence of 5-AzadC for a further 6 days, methylation of the KIAA0495 promoter was restored, with simultaneous suppression of the KIAA0495 expression (LP-1: P = 0.0331; OCI-MY5: P = 0.0447; Fig. 2c). Taken together, the methylation of KIAA0495 was inversely correlated with KIAA0495 expression level in myeloma cell lines, similar to the methylation-mediated silencing of KIAA0495 demonstrated in glioma cell lines and primary oligodendroglial tumour cells [16], suggesting that methylation of the promoter-associated CpG island emerged to be one of the mechanisms resulting in the regulation of lncRNAs in cancer cells.

To examine if methylation of KIAA0495 was also detected in primary samples, methylation of KIAA0495 was studied in 61 primary samples at diagnosis and 16 primary samples at relapse by MSP. However, none of these samples showed methylation of KIAA0495 (Fig. 3), indicating methylation of KIAA0495 was rarely detected in primary myeloma samples. Hence, similar to the studies of miR-9-1, miR-9-3, and miR-124-1 [21, 22], these data suggested that methylation of certain tumour suppressor non-coding miRNAs or lncRNAs was acquired in vitro during continuous culture of myeloma cells, hence not pathogenic. Therefore, methylation was not the mechanism leading to the progressive downregulation of KIAA0495 from normal plasma cell to MGUS to symptomatic myeloma [15], suggestive of other mechanisms, such as histone modification and miRNA-mediated repression, may come into play. For instance, in gastric cancer cells, histone deacetylation was found to be associated in the downregulation of a tumour suppressor lncRNA, FENDRR [23]. Moreover, in breast cancer cells, miR-21 was shown to target and hence downregulate the expression of a tumour suppressor lncRNA, GAS5 [24]. Furthermore, while KIAA0495 resides on 1p36, which has been shown frequently deleted in newly diagnosed patients with myeloma [25], the loss of expression of KIAA0495 may be associated with chromosome deletion. Last but not least, inactivation of KIAA0495 may also be accounted by haploinsufficiency, which has been demonstrated in the TP53 in myeloma [26].