B cells originate in the bone marrow where precursors pass through a series of highly regulated processes to generate a functional B-cell receptor (BCR). IGHV-D-J and IGLV-J rearrangements are achieved via a combinatorial process with associated amino acid additions/deletions at recombination sites leading to an immense variety of structurally diverse BCRs. Due to the large number of possibilities, potentially harmful BCRs recognizing self-antigens can be generated. In the early stages of B cell development a number of distinct selection mechanisms eliminate higher affinity auto-reactive B cells.

Chronic lymphocytic leukemia (CLL) is a CD5+ B cell malignancy. Besides having a unique phenotype not always assignable to a defined B-cell subset, many CLL B cells express auto-reactive BCRs; those that do not can often be shown to have derived from self-reactive clones. In addition, certain CLL patients express virtually identical IGHV-D-J rearrangements that can be organized into categories called "stereotyped BCRs". To date, stereotyped BCRs have not been conclusively shown to exist in the normal B-cell repertoire. This might be due to strict autoreactive censoring that eliminates stereotyped BCRs during B-cell ontogeny and/or because the number of IGHV-D-J rearrangements sequenced from individuals has not been sufficient to detect infrequent Igs. Therefore, using ultradeep, next generation IGHV-D-J sequencing we searched for CLL-specific stereotyped BCR rearrangements in the early phases of normal B cell maturation, before all mechanisms of selection are complete.

We sorted several B-cell populations from bone marrow samples obtained from healthy patients undergoing hip replacement without an autoimmunity or chronic infection history. The MiSeq platform was used to determine the IGHV-D-J repertoire of pro-, pre-, immature, transitional, naive, and memory B cells, and of plasmablasts. Sequences were analyzed using the pRESTO bioinformatic tool and R. Using VDJ rearrangement and CDR3 length criteria we focused on sequences potentially similar to those in CLL. Sequences were submitted to an online tool (ARReST/Assign subset) to be attributed to a particular CLL stereotyped subset based on a degree of certitude. We identified 156 sequences belonging to defined CLL stereotyped subsets with varying levels of similarity (extreme to low), most residing in early/antigen-independent compartments such as pre-B cells and immature, transitional, and naive B cells.

When stereotyped rearrangements of all degrees of certainty were considered, the greatest number of CLL-like stereotypes were found among the most immature B-cell populations, with frequency decreasing with advanced maturation (e.g., 0.04% in pro-B cells vs. 0.028 in naive and 0.01% in memory B cells). This is consistent with the notion that auto-reactive stereotyped rearrangements are eliminated from the normal repertoire. However, when using only those IGHV-D-J rearrangements with extreme or very high certitude, CLL-like stereotyped sequences were found at the highest frequency in the naive B-cell subset (0.02% vs 0.0058% in immature and 0.007 in pre-B cells). In addition, we identified at least one sequence for a majority of the most frequent 18 stereotyped subsets, although the proportional representation of each subset in our data is different from that seen in CLL. Specifically, the highest scoring recurrent subsets in our analysis were subsets 3 (61% of total CLL sequences), 5 (9.5%) and 64B (7%), which are not the most common in the known CLL cohort. Rather, we found only a few sequences belonging to the most common subsets such as subsets 1 and 2 (4.5% and 2% of the total CLL sequences), and none belonging to subset 4, although this subset is always isotype class-switched and IGHV-mutated and hence would not be expected to be found in the developing or naive repertoire.

Finding a significant presence of CLL-like stereotyped sequences when using ultra-deep DNA sequencing techniques in early phases of B-cell development and in the naive B-cell fraction suggests that CLL-like B cells are present among normal B cell populations albeit at very low frequencies. Moreover the presence of only certain of the most prevalent stereotypes such as subset 3 and 5, respectively the 5th and 6th biggest subsets, suggests these clones might not exhibit high auto-reactive affinity allowing them to more readily escape mechanisms of selection.


Stamatopoulos:Abbvie: Honoraria, Other: Travel expenses; Gilead: Consultancy, Honoraria, Research Funding; Novartis: Honoraria, Research Funding; Janssen: Honoraria, Other: Travel expenses, Research Funding.

Author notes


Asterisk with author names denotes non-ASH members.