MATCHED PATTERN DISCOVERY ACROSS PAIRED IMMUNOGLOBULIN HEAVY AND LIGHT CHAINS IN CLL REVEALS UNIQUE SUBSET-DEFINING AMINO ACID ASSOCIATIONS

ABSSUB-5462

Background: Immunogenetic analyses have helped to understand the pathogenesis of CLL and also to define subgroups with strikingly different behavior and outcome on top of unprecedented B-cell receptor (BcR) restriction (stereotypy), cementing the idea that antigenic elements select the leukemic clones. Even though recognition of certain antigenic epitopes may be critically dependent on conformations achieved by the interaction of both the IG heavy chain (HC) and light chain (LC) proteins, the focus to date has been on the IG HC gene, with studies on CLL IG LCs lagging behind. That said, previous studies from our group have demonstrated that CLL LCs can be as important as HCs, and that even very slight sequence alterations in LCs can be selected for and likely confer a functional advantage to the clone.

Aims: We here aimed to gain an even deeper view of CLL IG primary structures by searching for pattern associations between paired HC and LC gene rearrangements of 1331 CLL cases, the largest series to date, focusing on the main antigen-binding sites i.e. the HC and LC CDR3s.

Methods: We applied exhaustive amino acid pattern discovery that initially detected patterns within the VH and VK/VL CDR3s before associating these patterns with each other, either within the same chain (intra-association) or across chains (heavy-light inter-association).

Results: Overall, 14,544 patterns were discovered, of which 10,050 concerned the VH CDR3s; the remaining involved the VK (n=2528) or VL (n=1966) CDR3s, respectively. The discovered patterns varied in length from 1 to 22 amino acids. A total of 248,626 patterns were discovered with at least 2 occurrences across VH-VK chains, and 161,940 such patterns across VH-VL chains, being detected in 84.2% and 67.6% of the respective datasets. Starting from VH CDR3 patterns characteristic of stereotyped subsets, we explored whether certain VH/VK or VH/VL associations could be unique to particular subsets. We identified subset-biased and, most interestingly, subset-specific associations, i.e., matched patterns that were restricted to specific subsets. Of note, several of the subset-specific associations concerned dipeptides or even single amino acids within the VH and VK or VL CDR3, respectively. Examples include: (i) Q at VH CDR3 position 4 and S at VK CDR3 position 5 was detected only in subset #1 (clan I IGHV genes/IGKV1(D)-39), being present in 54/55 cases (98.1%); (ii) RY at VH CDR3 positions 12-13 and M at VK CDR3 position 1 was detected only in subset #4 (IGHV4-34/IGKV2-30), being present in all 26 (100%) subset #4 cases; (iii) DV at VH CDR3 positions 8-9 and D/P at VL CDR3 position 8 was detected only in subset #2 (IGHV3-21/IGLV3-21), being present in 48/51 (94%) subset #2 cases.

Summary/Conclusion: In conclusion, we herein demonstrate that CLL stereotyped subsets can now be more accurately defined based on IG gene usage, CDR3 length and pivotal short amino acid patterns or, remarkably, even single residues with a precise offset in both the HC and LC chain CDR3s. This finding draws further molecular analogies between stereotyped BcR IGs and the IGs expressed by mouse B-1 cells, supporting the notion that at least certain CLL clones may be derived from a B cell population intermediary to a true innate immune system and the conventional adaptive B cell immune system and hence functionally similar to what has previously been suggested for mouse B-1 cells.

Keywords: None

http://www.ehaweb.org/