Exome Sequencing Of Purified Hodgkin Reed-Sternberg Cells Reveals Recurrent Somatic Mutations In Genes Responsible For Antigen Presentation, Chromosome Integrity, Transcriptional Regulation and Protein Ubiquitination

Most genomic studies of classical Hodgkin lymphoma (CHL) have been confined to cell lines due to the difficulty of isolating sparsely distributed Hodgkin and Reed-Sternberg (HRS) cells from reactive background tissue. One approach to evaluate primary cases has been to microdissect HRS cells from fresh-frozen tissue biopsies, which has been used for gene expression profiling and array comparative genomic hybridization to assess copy number alterations (Hartmann et al., Haematologica 93, 2006; Brune et al, J Exp Med 205, 2008; Steidl et al., Blood 120, 2012; Steidl et al., Blood 116, 2010; Tiacci et al., Blood 120, 2012). However, laser capture microdissection is technically challenging, does not provide a pure tumor cell population, and yields very small amounts of nucleic acids that may not be adequate for deep sequencing. Typically, generating whole-exome sequence data from fewer than 10⁴ cells, including single cells, requires whole genome amplification (WGA), and is of a quality suitable to detect large scale copy-number alterations, but not nucleotide level information to identify point mutations.

We used a flow cytometric cell isolation method, which has enabled rapid isolation of thousands of viable HRS cells from primary CHL tumors (Fromm, et al., Am J Clin Pathol 126, 2006). Here we developed a new ultra-low-input DNA exome sequencing protocol which we combined with flow cytometry using CD64, CD95, CD30, CD5, CD20, CD15, CD40, and CD45, to produce what is to our knowledge the first full exome deep sequencing study of primary cases of Hodgkin lymphoma. We obtained a sequence depth of over 10X with >90% coverage of the target exome in HRS, as well as tumor-infiltrating T cells (also sorted and used as somatic control), in ten primary cases of CHL and performed mutation, copy number variation, and loss-of-heterozygosity (LOH) analysis.

We have identified 61 recurrent mutations, and 12 genes had somatic mutations in over 30% of cases. Non-synonymous or splice site mutations were seen in genes involved in antigen presentation (B2M), chromosome integrity (BCL7A), NF-kB activation (A20/TNFAIP3) and protein ubiquitination (HECW2 and UBE2A). We also obtained high-resolution copy number variation data indicating specific regions of gains and losses and intragenic chromosomal breakpoints. The most common genetic alterations from the combined analyses were in A20, present either as mutations or LOH, and alterations in b2-microglobulin (B2M), which were found in 80% of the cases sequenced. Alterations of B2M were inactivating and bi-allelic, leading to a lack of expression of MHC class I protein complex on the cell surface. The alterations include start codon mutations, exon-one splice donor site mutations, out of frame first-exon deletions and gene loss through chromosome-level deletion (LOH). Where possible, mutations were confirmed at the RNA level and resulted in lack of B2M protein expression documented by immunohistochemistry in an expanded cohort, where a total of 20 of 27 cases (74%) lacked B2M expression. Ectopic expression of B2M in a CHL cell line induced MHC class 1 expression, indicating that this genetic alteration is singly responsible for this defect in antigen presentation. In addition, B2M inactivation was highly prevalent in nodular sclerosis CHL, but was not found in any of the five cases of mixed cellularity CHL examined, indicating that these two types of CHL may belong to two different genetic categories.

We report the first exome deep sequencing of purified HRS cells from CHL tumor specimens, and reveal consistent alterations in important biological processes. The methodology developed allows exome sequencing from very low DNA input (10 ng), which has broader applications such as using FNA specimens from multiple tumor types. The genomic landscape of CHL revealed commonalities and differences with other lymphoma subtypes, like the consistent presence of A20 alterations that can lead to NF-kB activation. Inactivating mutations in B2M explain the complete lack of MHC class I expression, which likely affect the microenvironment. Since B2M expression is normal in all the cases evaluated that were histologically classified as mixed cellularity CHL, it is possible that this genetic alteration and resulting lack of B2M protein expression is a more accurate biomarker of CHL subtype than histological distinction.

No relevant conflicts of interest to declare.