ALT Cancer Cells Share Telomere End Structure With Normal Cells Despite Rogue Lengthening

Approximately 15% of human cancers bypass replicative senescence not through telomerase activation but via the alternative lengthening of telomeres (ALT) pathway — a recombination-based mechanism that uses existing telomeric DNA as a template. While ALT has been increasingly studied for its therapeutic vulnerabilities, fundamental questions about the physical structure and sequence composition of ALT telomere ends have remained unanswered, in large part because telomere repeats are notoriously difficult to sequence with precision. This NIH study tackles that gap head-on using END-seq, an unbiased sequencing platform originally designed to map DNA double-strand breaks genome-wide.

The core insight enabling this approach is that chromosome ends structurally resemble one-ended double-strand breaks, which END-seq captures through ligation of biotinylated adapters followed by next-generation sequencing. When applied to telomerase-positive human cell lines (HeLa and RPE1-hTERT), virtually all (99.9%) telomeric END-seq reads mapped to the C-rich strand, confirming that the method faithfully captures the 5′ chromosome terminus. Motif analysis then revealed that 78% of HeLa and 65% of RPE1 chromosome ends terminate with ATC-5′, a finding consistent with prior STELA-based measurements on defined chromosome ends. Crucially, this ATC-5′ bias was validated across multiple human cell lines, mouse tissues, and a canine cell line, indicating evolutionary conservation of this terminal sequence.

To confirm that END-seq genuinely reads the physical 5′ terminus rather than reporting a sequencing artifact, the team treated chromatin in agarose plugs with T7 exonuclease prior to END-seq. T7 progressively resects 5′ ends of double-stranded DNA, and as predicted, this treatment significantly randomized the detected terminal sequences, reducing the ATC-5′ bias toward an equal distribution across the six possible telomeric hexamer phases. Conversely, depletion of POT1 — the shelterin component known to regulate the 3′ overhang and indirectly control 5′ end precision — also partially randomized the terminal sequence. These paired experiments provide strong mechanistic and technical validation that END-seq is reading true chromosomal termini.

When END-seq was applied to a panel of ALT-positive cell lines (U2OS, SAOS2, VA13, and G292), the same ATC-5′ terminal bias was observed as in telomerase-positive cells (ranging from ~55–78%), and POT1 depletion similarly randomized the terminus in ALT cells. This is a striking finding: despite the chaotic, recombination-driven nature of ALT telomere elongation, the final processed chromosome end retains canonical structure and POT1-dependent regulation, suggesting that end-processing is decoupled from the mechanism of telomere elongation.

The study's second major contribution involves mapping single-stranded DNA (ssDNA) within ALT telomeres using S1 nuclease coupled with END-seq (S1-END-seq). S1 nuclease degrades single-stranded nucleic acids, including ssDNA bubbles within otherwise duplex DNA. In non-ALT cells, S1 treatment at telomeres produced minimal signal. In contrast, all four ALT-positive cell lines showed markedly elevated S1-END-seq signal at telomeres, indicating substantially more ssDNA within ALT telomeres. Strand-specific analysis showed this ssDNA was enriched on the G-rich strand, consistent with displacement loops (D-loops) or other recombination intermediates. This ssDNA signature robustly distinguished ALT-positive from ALT-negative cell lines and may represent both a biomarker for ALT status and a targetable vulnerability, since extensive telomeric ssDNA could sensitize cells to agents that exploit replication stress or RPA exhaustion.