Non-Hodgkin Lymphoma (NHL) is a class of cancer that originates in the lymphatic system, caused by an over-proliferation of malignant B-cells. NHL is one of the most common cancers in the United States, and lymphoma is the fifth most common cancer in Australasia (Abba et al., 2016), therefore posing a significant health burden (Calin et al., 2002; Cancer Council, 2020). The five-year survival rate is an estimated 71% (Allemani et al., 2018); however, both outcomes and pathogenesis vary greatly between NHL subtypes. Of the more than 40 subtypes of NHL, the two most common subtypes consist of the indolent follicular lymphoma (FL), and the more aggressive diffuse large B-cell lymphoma (DLBCL). Aggressive subtypes account for approximately 60% of cases, and a significant number of NHL patients suffer relapse on various treatments, with refractory NHL having a much poorer prognosis despite access to extensive chemotherapy and immunotherapy regimes (Michot et al., 2018; Sarkozy and Sehn, 2018; Ayers et al., 2020). Overall, this makes NHL an intractable disease to manage and treat, highlighting the need for reliable and subtype-specific diagnostic and prognostic biomarkers.

Understanding the cell of origin also plays a significant role in understanding NHL classification, progression, and prognosis (Alizadeh et al., 2000). B-cell lymphomas account for approximately 80% of NHL cases (Shankland et al., 2012), with DLBCL making up 25% of NHL diagnoses (Teras et al., 20162016; Siegel et al., 20192019). In 2000, gene expression profiling further defined DLBCL into two molecular subtypes; germinal centre B-cell like (GCB) and activated B-cell like (ABC) (Alizadeh et al., 2000). Also known as non-GCB, ABC demonstrated a poorer response to standard immunochemotherapy when compared to DLBCL cases with no specific subtype. This has been in part attributed to two oncogenic mechanisms seen in this subtype, encompassing the prevention of apoptosis and blockage of terminal differentiation (Blenk et al., 2007). These new subtypes of DLBCL, termed high-grade B-cell lymphoma, double-hit lymphoma, or triple hit lymphoma are associated with MYC and BCL2 and/or BCL6 rearrangements (Blenk et al., 2007) which are believed to contribute to their more aggressive oncogenesis.

Dysregulation of miRNAs is a hallmark of both cancer initiation and metastasis, through the regulation of gene expression via post-transcriptional repression and mRNA degradation (Di Lisio et al., 2012; Martin-Guerrero et al., 2015; Bradshaw et al., 2016; Getaneh et al., 2019). It is well established that miRNAs can act as both oncogenes and tumour suppressors in several cancers (Esquela-Kerscher and Slack, 2006; Di Lisio et al., 2012; Caramuta et al., 2013; Martin-Guerrero et al., 2015; Bradshaw et al., 2016; Getaneh et al., 2019), including NHL, via targeted repression of regulatory factors involved in processes such as cellular proliferation and migration (Xiao et al., 2008; Lawrie et al., 2009; Culpin, 2010). Previous studies have identified numerous miRNAs as having oncogenic potential in NHL (Amodio et al., 2015; Gado et al., 2019), as well as several being identified to be differentially expressed in tissues and fluids of individuals with these diseases. It has been previously established that miR-92a-3p has oncogenic potential, implicated in several cancers, including NHL, proposing that dysregulation of this miRNA may play a role in disease development (Culpin, 2010; Amodio et al., 2015; Gado et al., 2019). In both leukemia and DLBCL, miR-92 was seen to be downregulated, raising the question of whether this same downregulation is seen in indolent NHL subtypes such as FL (Amodio et al., 2015; Gado et al., 2019). The oncogenic miR-17–92 cluster, which includes miR-92a-3p, is a known driver of Burkitt lymphoma (BL), wherein miR-17 is associated with a poor prognosis and decreased overall survival rate (Robaina et al., 2016). miR-17–92 regulates multiple cellular pathways within NHL subtypes that favour malignant transformation, cellular proliferation, and cell survival (Dal Bo et al., 2015). In GCB DLBCL, the mir-17–92 microRNA cluster has been shown to be over-expressed when compared to ABC DLBCL (Lenz et al., 2008). Comparatively, the miR-92a-5p mature miRNA is not well described in NHL. With these relationships in mind, oncogenic or tumour suppressive miRNAs may be feasible targets for therapeutic agents, with the possibility of slowing or even halting malignancy (Wang and Wu, 2009), (Arif et al., 2020; Otoukesh et al., 2020). As recently reviewed in an earlier publication by our group, miRNAs can be epigenetically regulated, and even self-mediate these epigenetic modulations in various cancers (Arif et al., 2020). Aberrant DNA methylation in cytosines and adenines of mature miRNAs can lead to downregulation and suppression of the specific miRNA and, in NHL, a greater likelihood of lymphomagenesis (Calin and Croce, 2006; Esquela-Kerscher and Slack, 2006; Abba et al., 2017; Arif et al., 2020). These relationships support the possibility of miRNAs as being both functionally relevant in NHL and possible therapeutic targets.

The TET2 gene primarily functions as a tumour suppressor gene, immune regulator, and driver of DNA repair, as well as playing a key role in DNA demethylation (Fang et al., 2012). The TET2 protein catalyses the conversion of 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5-hmC), playing an integral role in transcriptional regulation. It is plausible that a hypermethylated genome may be due to a combination of factors that include high expression of DNA methyltransferases (DNMTs) or low expression of TET proteins (Fang et al., 2012; Asmar et al., 2013). TET2 is also involved in the recruitment of O-GlcNAc transferase OGT to CpG-rich sites to promote histone H2B GlcNAcylation (Asmar et al., 2013; Wang et al., 2018). TET2 is highly expressed in stem and progenitor cells, and in T-helper cells, whereas loss of TET2 gene function in bone marrow cells leads to increased immature B-cells resulting in lymphomagenesis (Fang et al., 2012; Asmar et al., 2013; Chiba, 2017). In NHL, TET2 expression is often reduced and regulated by methylation (Chiba, 2017). Various pathways are hypothesised to be regulated by miR-92a-TET2 including; JAK-STAT signalling, human pluripotent stem cells (hPSCs) differentiation, balanced B-cell terminal differentiation, and oxidative demethylation (miRTarBase, 2021; mimirna, 2021). To date, the prognostic significance of TET2 dysregulation in NHL is still not well characterised and further understanding of epigenetic drivers of TET2 regulation in lymphoma requires additional investigation.

In the last decade, both the role of miRNAs in malignancy and their viability as biomarkers for malignancy has been a rapidly expanding area of research. The mechanisms behind miRNA expression are not fully understood; however, aberrant methylation of promoter regions of miRNA genes has been previously implicated in aberrant miRNA expression in several cancers (Weber et al., 2007; Suzuki et al., 2010; Suzuki et al., 2012; Harada et al., 2015), including lymphoma (Stumpel et al., 2011; Yim et al., 2012). The expression profiles of miRNAs have been previously presented as novel biomarker panels for NHL (Khare et al., 2017; Solé et al., 2017), but the role of DNA methylation in miRNA expression regulation in NHL is not well explored. In this study, the methylation levels of CpGs in CGIs spanning the promoter regions of the miR-17–92 cluster host gene and the TET2 gene were quantified by bisulfite pyrosequencing and correlated with the expression levels of TET2 mRNA and mature miR-92a-3p and miR-92a-5p miRNAs. Gene expression was assayed in NHL B-cell lines, NHL tumour samples, and healthy PBMCs. Methylation of promoter CGIs was measured in the DNA of several NHL B-cell lines, in NHL tumour samples, in healthy PBMCs, and in whole blood gDNA of a retrospective NHL case-control cohort.

Increased expression of both the −3p and −5p mature miRNAs was observed in Mino cells when compared to controls, and additionally, upregulation of −5p was observed in Raji and Toledo cells lines. Previous studies have identified miR-92a overexpression in BL, MCL, and DLBCL cell lines, and our study replicates these findings (He et al., 2005; Ji et al., 2011). Higher expression of miR-92a has also been previously identified in MCL tumours (Husby et al., 2013; Roisman et al., 2016), reflecting the increased expression identified in Mino cells in this study. Increased expression of miR-92a-3p was also observed in our DLBCL tumour samples when compared to controls. Increased expression of miR-92a has previously been found to cause lymphatic malignancy in mice (Xiao et al., 2008), with downregulation of miR-92a implicated in the inhibition of tumour growth in an MCL mouse model (Rao et al., 2012). Overexpression of miR-92a and has been previously implicated in the reduction of overall survival and event-free survival in NHL patients (Yan et al., 2019), reinforcing the relevance of our findings.

Differential methylation was observed in the upstream promoter CGI of MIR17HG between cell lines, and between cell lines and healthy controls; however, this difference may not be biologically significant as the difference between mean methylation is within the detection limit margin of error (Gruber et al., 2002; Tost and Gut, 2007). Therefore, it seems unlikely that methylation of CpGs in these specific regions in the MIR17HG promoter region contribute to the differential miR-92a expression between NHL cell lines. A similar trend was identified in the NHL tumour samples; however, again the difference between mean methylation is within the detection limit margin of error. Aberrant methylation of the MIR17HG promoter has been previously correlated with dysfunctional expression of miRNAs in the miR-17–92 cluster in lung biopsies of individuals diagnosed with pulmonary fibrosis (Dakhlallah et al., 2013), and in both human samples and mouse models of bronchopulmonary dysplasia (Rogers et al., 2015; Robbins et al., 2016). The epigenetic mechanisms for the regulation of the miR-17–92 cluster expression in cancer, and in NHL specifically, are not well explored and further investigation is therefore required to identify the role of miR-92a in malignancy and specifically in lymphomagenesis. In NHL patient whole blood gDNA, statistically significant increased levels of mean methylation were observed in region 1 of the MIR17HG CGI when compared to healthy controls. Although the mean difference between the two cohorts was not substantial, several individuals in the NHL patient cohort presented high levels of genomic DNA methylation across these targeted regions. Aberrant methylation of levels were observed in individuals diagnosed with specific subtypes of NHL when compared to controls, BL being the most prominent. An aggressive subtype of NHL, the BL cell line Raji was identified to exhibit increased methylation in these same regions in the cell line populations, and previous studies have reported a similar increase in methylation levels to those identified in this study in other miRNA promoter regions, in both BL cell lines and BL tumours (Mazzoccoli et al., 2018; Mazzoccoli et al., 2019). Increased methylation was also observed between CLL individuals and controls in region 1 of the MIR17HG, and a previous study identified that aberrant methylation of miRNA promoter regions in circulating B-cells of CLL patients was associated with abnormal miRNA expression when compared to healthy B-cells. Increased expression of miR-92a has been implicated in CLL malignancy, specifically in lymph node proliferation centres (Szurián et al., 2017). As the case-control cohort in this study is comprised of whole blood gDNA rather than specifically PBMC or B-cell DNA these previous results cannot be replicated, but the similar rates of increased methylation of DNA can be considered, and further investigation into miRNA promoter methylation in CLL is required. It should be considered that, although differential methylation was observed between subtypes in this study, DNA hypermethylation has long been regarded as a hallmark of cancer, and the aberrant methylation observed in these regions may be a result of genome-wide increases in DNA methylation rather than a consequence of dysregulation of specific genes or promoter regions, endorsing further investigation into the role of DNA methylation in these specific NHL subtypes.

The findings of this study support previous conclusions regarding the significance of miR-92a in B-cell malignancy and reinforce the importance of investigating its specific role in NHL. Several previous studies have identified miR-92a as being implicated in dysregulation and suppression of tumour suppressors, including PTEN (Xiao et al., 2008) and PHLPP2 (Rao et al., 2012) in MCL models and VHL in CLL B-cells (Ghosh et al., 2009). This same may be true of TET2, wherein our studies identified a trend of dysregulation in some cells, along with additional novel findings which warrant further investigation. Differential expression of TET2 was seen across NHL cell lines and malignant tissues compared to controls, with specifically increased expression of TET2 in Mino, Raji, and Toledo cell lines alongside DLBCL and FL tumour tissue compared to controls. However, when taking into account the increased miR-92a-5p expression in the cell lines, NHL tumour tissue, and increased miR-92a-3p expression in the Mino cell line compared to controls, it is suggestive that miR-92a-5p and −3p may not be strong negative regulators of TET2 within these subtypes. Further assessment of the functional role of miR-92a on TET2 expression is required, feasibly with the utilisation of miRNA mimics and inhibitors, to determine the effect of miR-92a-5p and −3p expression and inhibition in NHL cell models.

SU-DHL-4 and Raji cell lines showed significant hypermethylation in the studied CpGs within the TET2 upstream promotor region when compared to controls and to Mino and Raji cell lines. This novel finding correlated with expected downregulation in TET2 expression in the SU-DHL-4 cells, along with a similar trend observed in the Raji cells compared to Mino and Toledo cell lines in previous studies (Chiba, 2016; Chiba, 2017). Although TET2 mRNA generally reported increased expression in cell lines and tumours, with the exception of SU-DHL-4, when compared to controls in our findings, TET2 methylation in our target region negatively correlated with expression in the NHL cell lines, indicating that the CpGs assayed in this region may be involved in the regulation of TET2 transcription and expression. TET2 overexpression in NHL is not well explored when compared to well-documented trends of downregulation in TET2 within various haematological malignancies (Chiba, 2017); however, previous studies have established this same trend of TET2 over-expression in CLL as observed in our study (Hernández-Sánchez et al., 2014). A possible explanation for this finding is TET2 overexpression in cells of some aggressive and indolent NHL subtypes is an innate immune response, involving tumour suppressor and DNA repair pathways, to the specific subtype of malignancy. Examination of TET2 protein and of downstream pathways of TET2 related expression regulation may assist in determining the translational consequence of TET2 overexpression in these NHL samples and cell lines.

Previous studies have identified TET2 variants in DLBCL as being associated with malignancy via the hypermethylation of pro-tumorigenic genes (Liu et al., 2017). Genotyping of the TET2 gene in the specific NHL cell models and tumours samples may therefore provide greater understanding into the potential presence of disease-causing TET2 mutations, which may contribute to disease phenotypes; as highlighted by differential expression DLBCL cell lines SU-DHL-4 and Toledo, both of which are DLBCL cell models. DNA methylation studies of CpGs in the TET2 upstream promotor region across both NHL tumour-derived and control PBMC samples showed consistently low levels of methylation, and a similar trend was observed in the methylation of genomic DNA in the NHL case-control cohort. Differential methylation was observed in the TET2 regulatory region between the case and control cohort; however, this difference may again not be biologically significant (Gruber et al., 2002; Tost and Gut, 2007). It is therefore unlikely that localised, as well as genome-wide, aberration of DNA methylation in CpGs in this region are directly involved in TET2 expression regulation in these NHL cohorts.

Clinical data from specific DLBCL tumour samples presented interesting characteristics and relationships when both TET2 and miR-92a expression were considered (Table 1). One tumour sample displayed Reed-Sternberg-like cells during the histological examination, while our studies showed significantly decreased TET2 expression compared to other DLBCL and FL tumour samples, as well as compared controls considered (Supplementary Figure S1E). Hypermethylation within our studied CpGs was not noted, therefore a likely cause of decreased TET2 expression in this outlier is the presence of a functional mutation in TET2. Reed-Sternberg-like cells have been previously associated with TET2 mutations (Venanzi et al., 2021) along with diminished 5 hmC levels (Siref et al., 2020). Genotyping the tumour samples would assist in determining whether a polymorphism or variant in TET2 is the cause of downregulated TET2 levels, thereby further supporting previous findings regarding TET2 as a potential marker and driver of malignancy in NHL.

The role of TET2 in NHL has been previously documented; however, the mechanisms driving TET2 expression regulation, such as DNA methylation, are not well explored. We identified in this study that aberrant methylation of several CpGs in the regulatory region of TET2 correlated with TET2 mRNA expression in NHL cell models. These findings are novel and may indicate that methylation of these regions may play a functional role in TET2 expression. This same relationship was not observed in NHL tumour samples or in NHL patient genomic DNA. Although a target of miR-92a, and aberrant miR-92a-3p and miR-92a-5p expression was observed in NHL cell models and NHL tumours, an association between miR-92a and TET2 was not observed in this study. Further assessment of the functional role of miR-92a-TET2 in human cell models is required; as is, ultimately, assessment in clinical cohorts. Although TET2 may not be comprehensively regulated by miR-92a, these novel findings regarding dysregulated methylation of TET2 regulatory regions in NHL is an intriguing relationship warranting further investigation and reinforces the prominence of TET2 as an element in NHL pathogenesis and malignancy.