High-throughput sequencing reveals novel features of immunoglobulin gene rearrangements in Burkitt lymphoma

TO THE EDITOR:

We thank Ralf Küppers for his interest in our article¹  and for his thoughtful commentary.²  We address each of the points he raises below.

First, Küppers claims that, in contrast to what has been demonstrated in mice,³  not all normal human B-cell progenitors undergo concurrent biallelic D_HJ_H rearrangement during B-cell development, and he cites an analysis performed with Southern blot techniques of 26 B-cell lines derived from 3 individuals to support this claim.⁴  We analyzed publicly available IGH high-throughput sequencing (HTS) data derived from analysis of B-cell genomic DNA from 60 healthy individuals (available at https://clients.adaptivebiotech.com/immuneaccess), and we conclude that the data strongly suggest that a very high fraction of normal human B cells do indeed undergo biallelic D_HJ_H rearrangement. These IGH HTS data were acquired using the same sequencing platform used for our studies.⁵  A total of 2 604 887 unique IGH nucleotide sequences, comprising productive V_HD_HJ_H, nonproductive V_HD_HJ_H, and incomplete D_HJ_H rearrangements, were observed in the HTS data from the 60 individuals. Of these, 1 326 781, or 50.9%, were productive V_HD_HJ_H rearrangements, and the rest were either nonproductive V_HD_HJ_H or incomplete D_HJ_H rearrangements. The fact that the total number of unique productive V_HD_HJ_H, nonproductive V_HD_HJ_H, and incomplete D_HJ_H rearrangements detected in these 60 individuals was almost exactly twice the number of unique productive V_HD_HJ_H rearrangements strongly supports biallelic D_HJ_H rearrangement in most developing human B cells.

Second, Küppers points out that it is physiological for human B cells, and not an abnormal feature of Burkitt lymphoma (BL) cells, that nonproductive V_HD_HJ_H rearrangements undergo somatic hypermutation (SHM) as efficiently as productive V_HD_HJ_H rearrangements. He reports that this observation is attributable to the fact that nonproductive rearrangements are transcribed and that transcription is a prerequisite for being targeted by somatic hypermutation. In our study, however, we observed evidence of comparable hypermutation of incomplete D_HJ_H rearrangements, which are not transcribed. Hypermutation on incomplete D_HJ_H rearrangements has not been reported previously, to our knowledge. We interpret this novel finding as evidence of an aberrant mutational process that is broadly active across the immunoglobulin loci in BL tumors.

Third, while one early report suggested that MYC translocation into the IGH locus may occur preferentially within the J_H gene segment in endemic BL and preferentially within the S_μ switch region in sporadic BL,⁶  subsequent studies have demonstrated that the translocation breakpoints in both endemic and sporadic BL regularly occur in the switch and J_H regions.^7-10  Therefore, the possible breakpoint locations on chromosome 14 span a genomic interval of ∼280 kb, from the J_H region through the switch region. For the V_HD_HJ_H or D_HJ_H rearrangement carried on the IGH allele into which MYC is translocated to be undetectable by our HTS platform, the translocation breakpoint would have to occur within the <200-bp interval that is amplified and sequenced by our sequencing strategy. If the breakpoint occurs outside this extremely narrow <200-bp window, our sequencing strategy would still detect the IGH rearrangement on this allele, even if it is adjacent to MYC. While disruption of the polymerase chain reaction (PCR) target region by the MYC; IGH translocation is theoretically possible, it is a highly unlikely event, and it is therefore improbable that it could explain the lack of detection of a second rearranged IGH allele in 85% of the BL tumors we studied that carried clonal immunoglobulin rearrangements.

Fourth, Küppers points out that mutations within the primer-binding regions may interfere with PCR amplification of IGH rearrangements and thereby impede their detection by HTS. The amplification primer binding-sites in our sequencing strategy are deliberately located in regions of FR3 and the J_H gene segment that are conserved and infrequently targeted by SHM, in an effort to diminish this phenomenon. We note that the existence of unique single-nucleotide polymorphisms in the V_H region or even novel IGHV gene segments present only within the Ugandan and Ghanaian populations that we studied could also have impaired PCR amplification, and thus the detection, of IGH rearrangements in the BL tumors. To address this possibility, we used RNA sequencing (RNA-seq) to analyze in an unbiased manner the expressed IGH repertoire in a subset of the BL tumors on which we reported. In 7 out of 12 cases, we found that the same dominant sequence detected by our targeted sequencing approach using genomic DNA was detectable by RNA-seq.^{1(suppl Fig7)} Moreover, our RNA-seq data, generation of which does not involve PCR amplification of the IGH locus, did not identity any high-frequency V_HD_HJ_H rearrangements that were not detected by our genomic DNA-based sequencing strategy. This suggests that neither SHM nor novel genetic variation present within the populations that we studied interfered with PCR primer binding and that the primers used for amplification of IGH gene segments in our sequencing strategy efficiently captured the full range of IGH rearrangements carried in the BL tumors.

For the reasons outlined above, we believe that the detection of only one rearranged IGH allele in 35 out of 41 BL tumors (85%) carrying clonal IGH rearrangements across 2 independent cohorts is likely biologically real and not a technical artifact. Moreover, we believe that our observation of hypermutation of incomplete, nontranscribed D_HJ_H rearrangements is indicative of an aberrant mutational process that affects the immunoglobulin loci in BL tumor cells.