To the editor:
Rearrangements involving the type I cytokine receptor subunit CRLF2 occur in 5% to 7% of all adult and pediatric B-cell precursor acute lymphoblastic leukemias (B-ALL) and in 60% of B-ALL in children with Down syndrome.1-4 CRLF2 rearrangement places full-length CRLF2 under alternate transcriptional control and can result from either an intrachromosomal CRLF2-P2RY8 deletion or a CRLF2-IGH translocation. All of the reported CRLF2-P2RY8 deletions involve V(D)J-type breaks at highly localized heptamer sequences upstream of CRLF2 (Figure 1A) and within intron 1 of P2RY8.2-4 In contrast, the breakpoints upstream of CRLF2 in CRLF2-IGH translocations distribute over a 24 717-bp region in a patchy manner (Figure 1A). This patchy clustering differs from the random distribution of breakpoints over multikilobase domains that we described for translocations including TEL-AML1 and BCR-ABL that occur within multipotent progenitors before V(D)J recombination.5 Among CRLF2 breakpoints, there is a moderately prominent 311-bp cluster from positions 1 307 403 to 1 307 713, which contains 6 of 19 described breakpoints (Figure 1A).4 Comparing this CRLF2 cluster to the remaining 24 406 bp, there is a 36-fold “enrichment” of breakpoints in the region.
Distribution of breakpoints on CRLF2. (A) Each 
denotes an individual CRLF2 breakpoint sequenced from a B-ALL with a CRLF2-IGH translocation. The ■ denotes the position of the CRLF2 breakpoints sequenced from B-ALL with CRLF2-P2RY8 intrachromosomal deletions. The arrow labeled “CRLF2” indicates the position and transcriptional direction of the CRLF2 gene. Numbering is per the March 2006 (hg18) build from the UCSC Genome Browser (http://genome.ucsc.edu/). (B) Overall frequency of CRLF2 breakpoints at various distance intervals from CpG. (C) Overall frequency of CRLF2 breakpoints at various distance intervals from CAC. Proportions of CRLF2 breakpoints at distances of 0 bp, 1-2 bp, 3-4 bp, 5-8 bp, and > 8 bp from CpG (B) or CAC (C) are graphed for the CRLF2 region. The distribution for actual leukemia breakpoints is shown in black, and that for a random distribution between the farthest breakpoints is shown in gray. If the black and gray bars parallel one another, then the patient breakpoints appear random in their distribution relative to the specified motif. However, when they follow opposite trends (ie, the gray bars rise with increasing distance from the specified motif while the black bars fall), then the breakage process appears to concentrate around the motif.
Distribution of breakpoints on CRLF2. (A) Each denotes an individual CRLF2 breakpoint sequenced from a B-ALL with a CRLF2-IGH translocation. The ■ denotes the position of the CRLF2 breakpoints sequenced from B-ALL with CRLF2-P2RY8 intrachromosomal deletions. The arrow labeled “CRLF2” indicates the position and transcriptional direction of the CRLF2 gene. Numbering is per the March 2006 (hg18) build from the UCSC Genome Browser (http://genome.ucsc.edu/). (B) Overall frequency of CRLF2 breakpoints at various distance intervals from CpG. (C) Overall frequency of CRLF2 breakpoints at various distance intervals from CAC. Proportions of CRLF2 breakpoints at distances of 0 bp, 1-2 bp, 3-4 bp, 5-8 bp, and > 8 bp from CpG (B) or CAC (C) are graphed for the CRLF2 region. The distribution for actual leukemia breakpoints is shown in black, and that for a random distribution between the farthest breakpoints is shown in gray. If the black and gray bars parallel one another, then the patient breakpoints appear random in their distribution relative to the specified motif. However, when they follow opposite trends (ie, the gray bars rise with increasing distance from the specified motif while the black bars fall), then the breakage process appears to concentrate around the motif.
We previously reported that some translocations occurring during B-cell ontogeny preferentially localize to the dinucleotide CpG, with 30% or more occurring directly at a CpG and 70% or more within 8 bases of a CpG.5,6 Of 19 total CRLF2 breakpoints, 7 are directly at CpGs (P < .001) and breakpoints are much closer to CpGs than expected by random chance (P < .001). Thirteen breakpoints are within 8 bp of a CpG, far more than if breakpoints were randomly distributed throughout the region (Figure 1B). In contrast, breakpoints are not significantly clustered around ▼CAC or ▼CACA (P > .1 in all tests, Figure 1C; ▼ indicates the target site for breakage by the V(D)J recombinase), the essential sequence motif that defines V(D)J recognition signal sequences (RSSs). Methods have been described previously.5
Similar to BCL2-IGH translocations described previously, at least 18 of the 19 IGH breaks in CRLF2-IGH translocations are compatible with standard V(D)J recombination, in that the IGH junction is within 30 bp 5′ of an RSS. In addition, 18 of 19 CRLF2-IGH junctions contain nontemplated nucleotide additions consistent with the activity of terminal deoxynucleotide transferase (TdT). Thus, the CRLF2 region from CRLF2-IGH translocations shares all the key features of the 4 regions previously described as having CpG-type breaks: (1) preferential localization to the dinucleotide sequence CpG, with breakpoints on either side (ie, 5′ or 3′) of CpG, (2) significantly weaker or no preferential localization to any other dinucleotide motif or ▼CAC, (3) propensity to cluster into zones of 20 to 600 bp, and (4) evidence of occurrence at the pro-B/pre-B stage, based on the involvement of IGH breaks from V(D)J recombination and the presence of TdT additions.
With this letter, the list of pathologic chromosomal rearrangements involving a CpG-type mechanism now consists of the BCL2-IGH translocation of follicular lymphomas, the BCL1-IGH translocation of mantle cell lymphomas, the MALT1-IGH translocation of MALT lymphomas,6 the E2A-PBX1 translocation of B-ALL, and the CRLF2-IGH translocation of B-ALL. Despite the differences in stage of arrested differentiation between the 5 diseases, all 5 translocations appear to occur within pro-B/pre-B cells.
Authorship
Contribution: A.G.T. designed and performed research and wrote the manuscript; A.Y. designed and performed research; and D.M.W. and M.R.L. revised the manuscript.
Conflict-of-interest disclosure: The authors declare no competing financial interests.
Correspondence: David M. Weinstock, Department of Medical Oncology, Dana-Farber Cancer Institute, Harvard Medical School, 44 Binney St, Dana 510B, Boston, MA 02115; e-mail: dweinstock@partners.org.
