Immune Checkpoint Molecules Regulate Paroxysmal Nocturnal Hemoglobinuria Clonal Evolution

In paroxysmal nocturnal hemoglobinuria (PNH), the acquisition of somatic PIGA mutations provides a selection advantage to hematopoietic stem cells from aplastic anemia (AA) immune attack. In addition to this immune evasion mechanism, HLA mutations, deletions, and somatic UPD involving 6p21 locus have been also described in AA, all of which implicated in such adaptive clonal escape. It is well-known that blood cells derived from a PIGA-mutant clone lack membrane-bound complement regulators (e.g., CD55, CD59) and exhibit an increased vulnerability to complement attack. Nevertheless, PIGA-mutant cells may still be susceptible targets for AA T cell-mediated immune attack. In conjunction with mutational analysis, HLA protein expression may be investigated by flow cytometry to identify a haploinsufficient (clonal) vs. possible adaptive down-modulation. Indeed, the absence of these otherwise constitutionally expressed proteins represents an excellent marker of clonality. Additionally, in an attempt to further hinder the creation of a successful immunological synapsis, other mechanisms, including modulation of immune checkpoints expression, may be also involved in such AA/PNH intertwine. Whether these immune escape mechanisms are convergent or instead represent alternative pathways is a subject of intense research.

Here, we comprehensively explore the dynamics governing PNH immune ontogenesis by combining HLA/PIGA genomics and flow cytometric measurements of HLA (A, B, C, DQ, DR, DP) and the immune checkpoint B7/CD28H family (PD1, PDL1, PDL2, VISTA, B7-H6, HHLA2, CD28H), stimulatory (GITR, CD30L) and inhibitory immune checkpoints (CD47) in CD45⁺CD15⁺CD59⁺ and CD59^- fractions. A deep targeted sequencing panel covering HLA classical loci along with an in-house developed pipeline was also applied to assess fine HLA aberrations (mutational and copy number events), whereas multi-amplicon deep sequencing was deployed to identify PIGA mutational status.

A total of 27 PNH patients classified as fully blown hemolytic PNH (n=15) or PNH/AA overlap (n=12) were included in this study with additional 4 healthy controls (HC). Overall, 41% of our patients harboured concomitant myeloid clones (most common hits were in DNMT3A, TET2, ASXL1, BCOR), while 58% of cases carried somatic HLA alterations, involving allelic losses or mutations. The latter were nonsense (33%), missense (11%), or 5' or 3' UTR (56%) hits with a median VAF of 11% (3-95%). Somatic lesions (including losses) were found in HLA-A (10%), B (30%), DQA1 (10%), or DQB1 (50%) loci. Since HLA aberrations may constitute the extreme pole of immune evasion and a marker of enhanced immune selection, we performed sorting/sequencing experiments on CD59⁺ vs CD59^- cells. A similar distribution and clonal burden of HLA-mutant subclones was found between the 2 fractions, indicating that PNH cells do not provide absolute immune privilege.¹

Next, we compared the expression of HLA and other immunomodulatory molecules across HC, CD59⁺ cells from PNH patients and their corresponding CD59^- cells (GPI+ and GPI-, respectively). Using MFI as a readout, GPI- cells showed similar expression levels of HLA-DP, -DR, PDL2, HHLA2 and GITR as compared to HC cells, while these molecules were significantly 1.5 to 5-fold upregulated on corresponding GPI+ cells consistent with their trigger/target function. This conclusion was also supported by the higher levels of CD28H found on GPI+ cells vs. GPI- (median MFI, 13.3 vs. 6.0; p<0.0001). Of note is that GPI- cells exhibited lower levels of HLA-DQ (0 vs. 7.0, p=0.01), B7H6 (0 vs. 39, p<0.0001) and CD30L (0 vs. 41, p <0.0001) vs. HC cells and upregulation of VISTA (460 vs. 227, p=0.006), suggesting that PNH clones may exhibit additional (to PNH phenotype) immune deterrence pathways. HLA alterations resulted in downregulation of HLA-DQ (0 vs. 29, p=.03), -DR (20 vs. 136, p=.03), B7H6 (0 vs. 45, p=.02), and CD30L (7.5 vs. 52, p=.02), as well as upregulation of HLA-B (2671 vs. 1909, p=.03) and VISTA (561 vs. 291, p=.02) in GPI- vs. GPI+, while no difference was noted in HLA^WT cases.

In conclusion, differential expression of immunomodulatory molecules on PNH vs. residual phenotypically normal cells in PNH patients and HC highlights various stimulatory and immune escape mechanisms operating in this disease. Understanding these processes may enable the identification of therapeutically relevant immune pathways.