Retrovirus insertion site analysis of LGL leukemia patient genomes

Abstract

Background: Large granular lymphocyte (LGL) leukemia is an uncommon cancer characterized by sustained clonal proliferation of LGL cells. Antibodies reactive to retroviruses have been documented in the serum of patients with LGL leukemia. Culture or molecular approaches have to date not been successful in identifying a retrovirus.

Methods: Because a retrovirus must integrate into the genome of an infected cell, we focused our efforts on detecting a novel retrovirus integration site in the clonally expanded LGL cells. We present a new computational tool that uses long-insert mate pair sequence data to search the genome of LGL leukemia cells for retrovirus integration sites. We also utilize recently published methods to interrogate the status of polymorphic human endogenous retrovirus type K (HERV-K) provirus in patient genomes.

Results: Our data show that there are no new retrovirus insertions in LGL genomes of LGL leukemia patients. However, our insertion call tool did detect four HERV-K provirus integration sites that are polymorphic in the human population but absent from the human reference genome, hg19. To determine if the prevalence of these or other polymorphic proviral HERV-Ks differed between LGL leukemia patients and the general population, we used a recently developed tool that reports sites in the human genome occupied by a known proviral HERV-K. We report that there are significant differences in the number of polymorphic HERV-Ks in the genomes of LGL leukemia patients of European origin compared to individuals with European ancestry in the 1000 genomes (KGP) data.

Conclusions: Our study confirms that the clonal expansion of LGL cells in LGL leukemia is not driven by the integration of a new infectious or endogenous retrovirus, although we do not rule out that these cells are responding to retroviral antigens produced in other cell types. However, our computational analyses revealed that the genomes of LGL leukemia patients carry a higher burden of polymorphic HERV-K proviruses compare to individuals from KGP of European ancestry. Our research emphasizes the merits of comprehensive genomic assessment of HERV-K in cancer samples and suggests that further analyses to determine contributions of HERV-K to LGL leukemia are warranted.

Keywords: Genomic insertion; HERV-K; Large granular lymphocyte leukemia; Retrovirus; Visualization tool.

Fig. 1

Utilizing long insert mate pair reads to localize retrovirus integrations. Reference human genome is shown as a blue line with the location of a novel inserted retrovirus in a patient sequence, in orange, indicated by a dotted vertical orange line. Long insert mate pair reads are linked by gray dotted lines, with the read derived from the new retrovirus, which will not map, shown in orange, it’s mate that maps to the human reference genome shown in blue. Depending on the length of the retrovirus, which typically is 6–10 kbp, some mate pairs may span the entire inserted virus and hence both mate pairs will originate from the host (light blue), resulting in mate pairs that map at a distance shorter than the expected insert distribution of 5–12 kbp. A retrovirus insertion site is suggested by a combination of several features of mate pair mapping including short insert intervals and discordant or broken mate pairs. The insert length and depth of mapped reads are key signals in our retrovirus insertion pipeline (see Additional file 1: Supplementary Methods; Figure S1). The unmapped reads (orange in the figure) from discordant mate pairs at each called insertion site are assembled and used to determine the sequence of a candidate retrovirus

Fig. 2

Linear discriminant analysis based on HERV-K status of T-LGL-EUR patients and KGP super populations. Linear discriminant analysis (LDA) was conducted on data generated by a comprehensive analysis of polymorphic HERV-Ks in an individual genome [18]. a Data is based on three HERV-K states of ‘absence,’ ‘solo LTR’, ‘provirus’ or b. The n/T ratio of each known HERV-K provirus for T-LGL leukemia patients of European ancestry and the 28 individuals from KGP super populations with high coverage data. The ratio indicates the proportion of k-mers derived from a person’s WGS dataset (n) that are 100% match to a set of unique k-mers (T) characterizing each HERV-K provirus. The improved resolution of T-LGL-EUR patients from other individuals using n/T likely reflects that alleles of HERV-K contribute to population differentiation. The symbols and colors for each KGP populations and T-LGL-EUR leukemia patients are indicated in the key on the right

Fig. 3

Histogram of the number of polymorphic HERV-K proviruses identified in LGL leukemia patients compared to individuals of European origin from KGP. Data are shown for 51 LGL patients (blue) and for the subset of 40 patients with T-LGL leukemia of European ancestry (T-LGL-EUR, orange). Data for the 505 EUR individuals (gray) from the KGP data is from Li et al. [18]

Fig. 4

The prevalence of combinations of polymorphic HERV-K provirus in KGP populations and T-LGL-EUR leukemia patients. The combinations of polymorphic HERV-K provirus evaluated are indicated at the top right of each panel. a The prevalence of three polymorphic HERV-K proviruses that include chr12: 58721242–58,730,698 in KGP individuals and T-LGL-EUR patients. b The prevalence of three polymorphic HERV-K, excluding chr12: 58721242–58,730,698, in KGP individuals and T-LGL-EUR leukemia patients. Coordinates are referenced to hg19. Bubble size is proportional to the number of individuals in the population and color gradient represents prevalence from 0 to 100%. Absolute values are given in the text on the right for each population. KGP population abbreviations are given in Table 1 and additional information can be found at (http://www.internationalgenome.org/category/population/)