Loss of Pigh Expression Frequently Results in a GPI-Negative Subclone Lacking CD52 Membrane Expression, Conferring Alemtuzumab Resistance to B Cell Acute Lymphoblastic Leukemia

To improve treatment outcome of patients with B cell acute lymphoblastic leukemia (B-ALL), several immunotherapeutic approaches have been developed in recent years. E.g., direct targeting of CD19 or CD20 by (bispecific) antibodies or chimeric antigen receptors result in effective control of the disease. In contrast, introduction of alemtuzumab which targets the glycophosphatidylinositol (GPI) anchored CD52 protein has been less successful. Despite its profound lymphodepleting potential, the clinical efficacy was unsatisfactory. Similar to development of escape variants following anti-CD19 and anti-CD20 treatment, this might have been due to the outgrowth of CD52 negative B-ALL escape variants. Indeed, in previous studies (Nijmeijer et al, 2010) we have found outgrowth of CD52 negative escape variants after alemtuzumab treatment in a mouse model engrafted with human B-ALL. Further analysis showed that these variants expressed normal CD52 mRNA levels, but lacked CD52 membrane expression which was found to be due to the loss of GPI anchor expression, as confirmed by loss of staining with the GPI-specific aearolysin FLAER. The aim of the current study was to further unravel the mechanism underlying loss of CD52 membrane expression in adult B-ALL. To study whether the relatively frequent development of these CD52 negative/GPI anchor deficient escape variants during alemtuzumab treatment was the result of outgrowth of pre-existing GPI negative cells, we analyzed 10 Peripheral Blood (PB) and 8 Bone Marrow (BM) samples from B-ALL patients at the moment of diagnosis. GPI negative cells were present in 7 out of 10 (70%) PB and 5 out of 8 (63%) BM samples, and comprised between 0.01% and 4.98% of the B-ALL population as analyzed by flow cytometry using FLAER and CD52 counterstaining (detection limit 100 cells per million). Interestingly, these obvious GPI negative populations were not found in other B cell malignancies such as chronic lymphocytic leukemia (n=5), mantle cell lymphoma (n=5), hairy cell leukemia (n=6), or in healthy donors (n=5). To investigate the mechanism of GPI loss, gene expression analysis was performed for the 26 genes that comprise the GPI anchor biosynthesis pathway. GPI positive and GPI negative B-ALL populations (n=7) were purified by fluorescent activated cell sorting (FACS). Recurrent loss of PIGH mRNA expression, but of none of the other genes involved in GPI anchor biosynthesis pathway, was found in GPI negative cells but not in GPI+ cells in all cases. To validate the relevance of this finding, GPI negative and GPI+ B-ALL cell cultures were generated from diagnosis material (n=2) and subsequently transduced with a retroviral construct encoding PIGH, PIGA (another member of the GPI-N-acetylglucosaminyl-transferase complex, which catalyzes the first step in the GPI synthesis) or an empty vector. Restored GPI anchor expression and coinciding CD52 membrane expression was observed in the GPI negative B-ALL cultures upon transduction with the PIGH, but not PIGA or empty vector. To explore the mechanism underlying the recurrent loss of PIGH mRNA expression, we performed extensive DNA screening of the GPI negative B-ALL cultures to discover possible mutations or indels in the promoter region, gene body, or at splice sites and compared this to GPI+ B-ALL cultures from the same individual. These analyses revealed that both alleles of the PIGH gene were present, unmutated and intact. To investigate whether epigenetic regulation could be the cause of PIGH deficiency, we measured promoter CpG methylation by bisulfite sequencing, comparing GPI negative and GPI+ B-ALL cultures (n=2). This analysis revealed that the region surrounding the transcription start site (-200bp up to +100bp) was heavier methylated in the GPI negative B-ALL cultures compared to the GPI+ counterparts. Also, a 14 day treatment of GPI negative B-ALL cultures with the demethylating agent 5-Azacitidine resulted in re-expression of the GPI anchor.

In conclusion, the majority of B-ALL patients presented a CD52/GPI negative, alemtuzumab resistant, B-ALL population in samples taken at diagnosis. These cells lost PIGH expression, a key component in GPI anchor synthesis. This is not due to genomic instability, but rather to epigenetic down regulation. Combining epigenetic modulatory drugs with alemtuzumab might be a promising therapeutic strategy to prevent outgrowth of CD52/GPI negative escape variants in B-ALL.