Changes in N-Glycosylation Induced By Somatic Hypermutation Modulate the Antigen Reactivity of the Immunoglobulin Receptors in CLL Stereotyped Subset #201

The IGHV4-34 gene is intrinsically autoreactive due to carrying a germline(GL)-encoded (super)antigenic motif binding various self (and exogenous) antigens, while it is one of the few IGHV genes that contain a GL-encoded N-glycosylation (N-glyc) site. IGHV4-34 is overrepresented in chronic lymphocytic leukemia (CLL), particularly in cases expressing B cell receptor immunoglobulin (BcR IG) with a significant load of somatic hypermutation (SHM; 'mutated' CLL, M-CLL). Moreover, a large fraction of IGHV4-34 M-CLL cases are clustered in different stereotyped subsets, of which the best studied is subset #4, the largest within M-CLL, defined by the expression of IgG-switched IGHV4-34/IGKV2-30 BcR IG with a distinctive SHM imprint. Considerably smaller than subset #4 is subset #201, defined by the expression of IGHV4-34/IGLV1-44 BcR IG of the IgMD isotype. Subset #201 is noteworthy owing to recurrent replacement SHMs that frequently lead to the creation of novel N- glyc motifs within the VH domain. This may be functionally relevant, considering that N-linked glycosylation is a widespread post-translational modification that is largely SHM-induced during antigen-specific immune responses and can modulate antibody (Ab) affinity towards antigen. That said, nothing is yet known about the antigen reactivity of subset #201 BcR IG and whether/how it could be affected through SHM-induced changes of N-linked glycosylation. In order to obtain insight into this issue, 4 subset #201 clonotypic IGs were expressed as recombinant monoclonal Abs (mAbs) of the mu isotype in HEK293 human cells, in either the authentic SHM state ('wildtype', WT-mAbs) or after reverting specific SHMs that altered N-glyc sites (R-mAbs) by site-directed mutagenesis. Since not all N-glyc motifs are eventually glycosylated, we used the NetNglyc 1.0 Server (http://www.cbs.dtu.dk/services/NetNGlyc/) for the prediction of N-glycan occupancy. Binding to MEC1 B CLL, Jurkat T and HEK293 cells was assessed by flow cytometry. Reactivity against nuclear Hep-2 cell extract, nDNA, actin, myosin, thyroglobulin (TG), β-amyloid, carbonic anhydrase, F(ab')2 and the non-self hapten trinitrophenyl was tested by ELISA. Non-subset #201 M-CLL mAbs (n=14, including 3 subset #4 mAbs), were used as controls. None of the subset #201 WT-mAbs displayed reactivity in any of the ELISAs. However, unlike most CLL mAbs, all subset #201 WT-mAbs bound to live MEC1 cells, while also exhibiting reactivity to HEK293 cells that was significantly higher when compared to non-subset #201 M-CLL (p=0.0095) or subset #4 (p=0.05); additionally, 1/4 subset #201 mAb displayed weak binding to Jurkat T cells. Three of 4 subset #201 mAbs bore a novel N-glyc site introduced by SHM in codons VL CDR1 36-38 of the clonotypic lambda light chains. Reversion to the GL in one such mAb resulted in enhanced binding to all 3 cell lines [fold change (FC) of binding of the R- vs WT-mAb to MEC1, Jurkat and HEK293: 1.3, 7.9 and 3.3, respectively) and in strong anti-TG activity. The GL-encoded N-glyc site in VH CDR2 57-59, that has been reported to be mostly unoccupied, was targeted by SHM in 2/4 subset #201 mAbs: reversion to GL decreased binding to both MEC1 and HEK293 cells (FC: -8 and -1.4 respectively). Finally, in 2/4 cases, SHM at codons VH FR3 67-68 inserted an N-glyc site that, however, is not predicted to acquire N-glycans. Reversion to GL enhanced the binding of one of these mAbs to MEC1 and HEK293 cells (FC: 2.1 and 5.6, respectively). The same mAb bore an additional predicted N-glyc site introduced by SHM at VH FR3 90-92; reversion of this change to GL augmented binding to both MEC1 and HEK293 cells (FC: 4.1 and 9.7, respectively). Double reversion of both aforementioned SHMs conferred further increased binding than any of the single reversions, implying a synergistic effect. Acquisition of novel N-glyc sites is not an intrinsic characteristic of either M-CLL in general or IGHV4-34 M-CLL in particular and its high incidence in subset #201 implies a selective process likely due to distinct (auto)antigenic pressure. Indeed, subset #201 mAbs exhibit an antigen reactivity profile that differs from that of typical polyreactive mAbs, including natural autoantibodies and other CLL mAbs, binding selectively to viable lymphoblastoid cell line cells and human HEK293 epithelial cells. These results further emphasize the importance of SHM in shaping the distinct (auto)antigenic recognition profile of CLL mAbs.