Antigen selection reflected in the subclonal architecture of the B-cell receptor immunoglobulin gene repertoire in splenic marginal zone lymphoma

Abstract

Almost one-third of all splenic marginal zone lymphoma (SMZL) cases express B-cell receptor immunoglobulin (BcR IG) encoded by the IGHV1-2*04 gene, implicating antigen selection in disease ontogeny. Evidence supporting this notion mostly derives from low-throughput sequencing approaches, which have limitations in capturing the full complexity of the BcR IG gene repertoire. This hinders the comprehensive assessment of the subclonal architecture of SMZL as shaped by antigen selection. To address this, we conducted a high-throughput immunogenetic investigation of SMZL aimed at the comprehensive characterization of the somatic hypermutation (SHM) and intraclonal diversification within the IG genes. We identified significant differences in the SHM and ID profiles between cases expressing the IGHV1-2*04 gene and those expressing other IGHV genes. Specifically, IGHV1-2*04 cases displayed (i) targeted SHM resulting in recurrent replacement SHMs, and (ii) significantly more pronounced intraclonal diversification, reflecting ongoing antigen selection. Overall, our findings suggest that SMZL cases expressing the IGHV1-2*04 gene have a distinct immunogenetic signature shaped by microenvironmental pressure on the clonotypic BcR IG, corroborating the idea that this group may represent a distinct molecular variant of SMZL.

Figure 1

Differential SHM targeting in IGHV1‐2*04 cases versus cases utilizing other IGHV genes. The topology of replacement SHMs reveals the existence of different patterns in IGHV1‐2*04 cases compared to cases expressing the IGHV3‐23 or IGHV4‐34 genes. Rows in each table represent the codons of the clonotypic BcR IG heavy chain sequence from the start of VH FR1 until the end of VH FR3. The sequence of the first 26 codons of VH FR1 was not available in 19 IGHV1‐2*04 rearrangements that were PCR‐amplified utilizing VH FR1 rather than leader primers; this part of the graph is depicted with a light gray color. Columns in each table represent individual SMZL cases in each group.

Figure 2

Differential distribution of recurrent replacement SHMs in IGHV1‐2*04 cases versus cases utilizing other frequent IGHV genes. The IGHV1‐2*04 group shows a higher frequency of recurrent replacement SHMs occurring both in the dominant clonotype as well as in the end clonotype in the context of intraclonal diversification. Some of these SHMs, namely M39I, A87V, G36D, and R95T, were found to occur simultaneously in a large number of IGHV1‐2*04 cases. Co‐occurrence of recurrent replacement SHMs was significantly more limited in the other sample groups, being evident only in two samples of the IGHV3‐23 group. In each individual heatmap, rows represent different recurrent replacement SHMs, while columns represent individual SMZL cases of each group. A lighter shade of color indicates that the SHM was present in the dominant clonotype, while a stronger shade of color indicates that the SHM was acquired in the end clonotype (i.e., the one with the highest number of SHMs acquired in the context of intraclonal diversification of the IG genes).

Figure 3

In silico BcR IG modeling and clustering in IGHV1‐2*04 cases support a differential impact for recurrent replacement SHMs. The 3D BcR IG models of the SMZL cases carrying ID‐related recurrent SHMs were grouped in six different clusters. “L1” and “L2” indicate the two different light chains that were paired with each heavy chain to eliminate the possible impact of the light chain. Each separate pair of 3D BcR IG models with a given heavy chain and a different light chain was assigned to the same cluster, indicating that the overall structure of the 3D BcR IG models is largely unaffected by the light chain partner of the IGHV1‐2*04 heavy chain. At the cluster level, C1 comprised the “SHM‐free” IGHV1‐2*04 models (i.e., those devoid of any SHM) as well as those carrying the G36D and M39I mutations individually. Cluster C2 models were characterized by the presence of the R95T mutation, either as a single mutation or in combination with one of the G36D or the M39I mutations. Clusters C3 and C4 consisted of the models carrying the G36D‐M39I and G36D‐M39I‐R95T SHM combinations, respectively. Clusters 5 and 6 were formed by all 3D BcR IG models carrying the A87V, again either individually or in combination with the other analyzed mutations.