Extensive somatic L1 retrotransposition in colorectal tumors

Abstract

L1 retrotransposons comprise 17% of the human genome and are its only autonomous mobile elements. Although L1-induced insertional mutagenesis causes Mendelian disease, their mutagenic load in cancer has been elusive. Using L1-targeted resequencing of 16 colorectal tumor and matched normal DNAs, we found that certain cancers were excessively mutagenized by human-specific L1s, while no verifiable insertions were present in normal tissues. We confirmed de novo L1 insertions in malignancy by both validating and sequencing 69/107 tumor-specific insertions and retrieving both 5' and 3' junctions for 35. In contrast to germline polymorphic L1s, all insertions were severely 5' truncated. Validated insertion numbers varied from up to 17 in some tumors to none in three others, and correlated with the age of the patients. Numerous genes with a role in tumorigenesis were targeted, including ODZ3, ROBO2, PTPRM, PCM1, and CDH11. Thus, somatic retrotransposition may play an etiologic role in colorectal cancer.

Figure 1.

Genomic distribution of L1 insertions. Outer rings show the density of detected insertion sites for reference (gray) and nonreference (black) L1s. The approximate locations of the 72 PCR-validated somatic insertions are indicated by dots inside the circle. Note that 69 of the 72 insertions were successfully sequenced.

Figure 2.

PCR validation scheme of L1-seq results. (A) The three-step PCR validation scheme and location of primers used. Triangles symbolize TSD. (B) PCR validation of the 3′ junction (ins. 7). This insertion is in tumor 1 of the eight DNA samples that had been pooled for Illumina sequence analysis (left), while the right panel shows it is present exclusively in the tumor, but not in the normal colon. The higher molecular weight band visible above the ins. 7 empty site PCR product in the tumor is a highly truncated L1. (T) Tumor; (N) normal colon; (FS) filled site PCR product; (ES) empty site PCR product.

Figure 3.

Distribution of somatic L1 insertions in tumors. Insertions in black were detected by L1-seq, while the insertion in tumor 10 in white was detected by RC-seq only.

Figure 4.

Analysis of factors influencing L1 activity. (A) L1 CpG promoter methylation status performed by quantitative bisulfite PCR analysis. (N) Normal tissue; (T) tumor tissue; (*) MSI. Replicates of four were done for each data point. (Error bars) Standard deviations. (B) MSI analysis. 6% TBE gel depicting the status of five microsatellite repeats (BAT25, BAT26, D2S123, D17S250, D17346 in descending order) in normal and tumor tissue from two different patients. Tumor tissue “6” contained additional bands and gel shifts compared with the normal tissue, indicating MSI. Samples from “8” demonstrated no differences suggestive of MSI. (C) Correlation of L1 activity with age of the patient at time of surgery. See text for details.