Loss of the GPI-anchor in B-lymphoblastic leukemia by epigenetic downregulation of PIGH expression

Abstract

Adult B-lymphoblastic leukemia (B-ALL) is a hematological malignancy characterized by genetic heterogeneity. Despite successful remission induction with classical chemotherapeutics and novel targeted agents, enduring remission is often hampered by disease relapse due to outgrowth of a pre-existing subclone resistant against the treatment. In this study, we show that small glycophosphatidylinositol (GPI)-anchor deficient CD52-negative B-cell populations are frequently present already at diagnosis in B-ALL patients, but not in patients suffering from other B-cell malignancies. We demonstrate that the GPI-anchor negative phenotype results from loss of mRNA expression of the PIGH gene, which is involved in the first step of GPI-anchor synthesis. Loss of PIGH mRNA expression within these B-ALL cells follows epigenetic silencing rather than gene mutation or deletion. The coinciding loss of CD52 membrane expression may contribute to the development of resistance to alemtuzumab (ALM) treatment in B-ALL patients resulting in the outgrowth of CD52-negative escape variants. Additional treatment with 5-aza-2'-deoxycytidine may restore expression of CD52 and revert ALM resistance.

Figure 1

Loss of GPI/CD52‐expression is due to absence of PIGH mRNA expression in B‐ALL samples at diagnosis. A, Four representative flow cytometric analyses of GPI‐anchor (FLAER) and CD52 membrane expression on B cells (CD19+ CD3−, left panels) or T cells (CD3+ CD19−, right panels) in MNC samples taken at diagnosis from patients with B‐ALL. The percentages of GPI/CD52‐negative cells are indicated. B, Representative example of mRNA expression analysis for the 28 genes comprising the GPI‐anchor synthesis pathway on equimolar amounts of cDNA from GPI/CD52‐negative (GPI^neg) and GPI/CD52‐positive (GPI^pos) B cells purified from MNC sample ALL‐04. C, mRNA expression analyses of the PIGH protein‐coding region (717 bp) performed on equimolar amounts of cDNA from purified GPI^neg and GPI^pos B cells from PB (n = 4, left panels) or BM (n = 4, right panels) samples of patients with B‐ALL, using GAPDH (351 bp) as a loading control (M defines the marker lane)

Figure 2

Retroviral transduction with PIGH, and genomic analysis of the PIGH gene loci in the GPI^neg and GPI^pos B‐ALL subcultures. A and B, Flow cytometric analysis of GPI‐anchor (FLAER) and CD52 membrane expression in (A) GPI^neg and (B) GPI^pos subcultures of cell lines Leiden‐ALL‐BV (BV‐GPI^neg and BV‐GPI^pos) and Leiden‐ALL‐HP (HP‐GPI^neg and HP‐GPI^pos) retrovirally transduced with an empty control construct (mock) or with constructs encoding PIGA or PIGH, coupled to tNGFR as marker gene. Transduction efficiency ranged between 13.6% and 52.9%. Dot plots are gated on tNGFR positive cells. C, Schematic representation of analyses performed on DNA and mRNA isolated from GPI^neg and GPI^pos subcultures of cell lines Leiden‐ALL‐BV (BV‐GPI^neg and BV‐GPI^pos) and Leiden‐ALL‐HP (HP‐GPI^neg and HP‐GPI^pos). The PIGH genomic locus (chromosome 14:67601700‐67588700, GRCh38.p7) is shown, numbered black boxes represent coding exons connected by straight black lines representing intronic regions. SNPs rs12893796 and rs11547225 (dbSNP build 144) are depicted by triangles. PCR amplicons are numbered and indicated as stripped lines connecting the relevant primers (arrows). mRNA transcription variants are depicted by boxes connected with curved lines and identified by their Ensemble transcript number. The protein coding transcript is in black, predicted protein coding mRNA transcription variants are in gray. Closed boxes illustrate the protein coding region and open boxes the 5′ and 3′ untranslated regions. D and E, Gel electrophoresis results for the indicated PCR amplifications on (D) mRNA or (E) DNA isolated from the GPI^neg and GPI^pos subcultures (GAPDH as loading control). F, Sanger sequencing results from nested PCR amplifications VII (Leiden‐ALL‐BV) and VI (Leiden‐ALL‐HP) with assembly GRCh38.p7 as reference sequence. SNPs rs12893796 and rs11547225 are presented as gray boxes containing the allelic variants

Figure 3

Epigenetic down regulation of PIGH gene transcription in GPI^neg B‐ALL subcultures. A, ChIP‐qPCR analysis of the relative presence of histone marks H3K4me3 and H3K27me3 at the PIGH gene location in the GPI^neg subcultures compared with the GPI^pos subcultures of Leiden‐ALL‐BV and Leiden‐ALL‐HP. PIGH DNA was quantified in H3K4me3 and H3K27me3 ChIP samples by qPCR and compared with a nonimmunoprecipitated DNA reference sample (refDNA) and normalized for a positive control gene (GAPDH for H3K4me3 and MYOD1 for H3K27me3). Bars represent the relative presence of PIGH DNA in the GPI^neg compared with the GPI^pos subcultures (whiskers represent SD; *P < .05, **P < .01, ***P < .001, NS = not significant; two‐sided unpaired T‐test). Primer set 1 (targeting the region 246‐371 bp downstream of the TSS) and primer set 2 (329‐496 bp downstream of the TSS) used for qPCR of H3K4me3 targeted the PIGH promoter region. Primer set 1 used for qPCR of H3K4me3 doubled as primer set 1 for qPCR of H3K27me3. Primer set 2 (4486‐4570 bp downstream of the TSS) and primer set 3 (6261‐6373 bp downstream of the TSS) used for qPCR of H3K27me3 targeted the PIGH gene body. B, MS‐MCA curves for the PIGH promoter region (−212 bp to +101 bp relative to the TSS, CpGs −19 to +15) on bisulfite converted DNA from the GPI^pos and GPI^neg B‐ALL subcultures (solid black line). Bisulfite converted DNA from purified healthy donor B cells served as a control. MS‐MCA curves for unmethylated reference DNA (striped gray line, peak T _m 83.4°C) and methylated reference DNA (dotted gray line, peak T _m 89.0°C) are plotted in each graph. C, Sanger sequencing analysis of the methylation state of 24 individual CpGs in the PIGH promoter region (−161 bp to +51 bp relative to the TSS, CpGs −16 to +8) on bisulfite converted DNA from the GPI^pos and GPI^neg B‐ALL subcultures and from purified healthy donor B cells. The levels of methylation of individual CpGs were determined by the ratio between the height of the cytosine signal and the thymine signal. D, Percentages of GPI‐anchor positive cells in the GPI^neg B‐ALL subcultures following treatment with or without 0.5 μM 5‐aza for 15 days as analyzed by flow cytometry counterstaining with FLAER