HERV-K HML-2 transcripts in human cells: RNA sequencing findings and limitations
Mobile DNA · 2026
In plain language
This study examined RNA from human endogenous retroviruses in a single laboratory-grown cell line, not in cats or FeLV. It described the number of RNA reads assigned to different groups and some features of those molecules. The technique has limitations that make it difficult to recover complete molecules and identify their origins. Its relevance to FeLV is methodological and indirect only: the supplied text provides no direct evidence about disease mechanisms or clinical outcomes.
Scientific detail
The study in Mobile DNA (2026; doi: 10.1186/s13100-026-00400-4) used direct single-molecule long-read RNA sequencing to characterize HERV-K HML-2 (HK2) transcription and splicing patterns in vitro in a single human cell line, NCCIT. METHODS describes four experiments, with one library each for Experiments 1 and 3 and two libraries each for Experiments 2 and 4; RESULTS instead describes four libraries and four biological replicates from independent cell cultures. Total RNA was extracted from 1 × 10⁶ to 3 × 10⁶ cells. The pooled sequencing dataset contained 3,661,998 reads. The discrepancy between sections prevents a consistent accounting of libraries, and the supplied text omits a substantial central passage.
All four claims in the knowledge object have evidence level D. In the pooled NCCIT sample, HERV-H was the most abundantly transcribed HERV group, with 24,031 assigned reads; HK2 accounted for 9,193. These values are read assignments from pooled human cell-line data, not counts of infected cells, animals or independent samples. They do not establish a corresponding transcriptional pattern in FeLV. Estimated median poly(A) tail lengths of HK2 proviral transcripts overlapped those of the selected housekeeping genes. This observation does not directly demonstrate equivalent RNA stability, translation efficiency or functional importance, and it provides no feline-specific evidence.
In most detected NCCIT transcripts containing parts of HK2 solo LTRs, the authors identified the solo LTR as providing a polyadenylation signal and site. This is their interpretation of transcript structures in a human cell line; it neither establishes the same mechanism in FeLV nor demonstrates downstream protein expression or biological effects.
The authors identify RNA degradation during handling as a cause of 5’ truncation, limiting the recovery of complete transcripts and identification of their origins. Full-length HK2 transcripts are particularly susceptible to 5’ under-detection; this limitation does not show that every detected isoform is artifactual. Other limitations include lower throughput and yield, higher nucleotide-level error rates than short-read sequencing, and selection bias against non-polyadenylated RNAs. Closely related and polymorphic integrations complicate locus assignment. Poly(A) tail estimates do not directly measure transcript stability or translation. No feline or FeLV experiments are reported in the supplied text: applying this approach to feline retroviral transcripts would be an inference requiring feline-specific validation.
What we know
In a single human cell line studied in vitro, pooled data showed more reads assigned to HERV-H than to HK2. Estimated median poly(A) tail lengths of HK2 proviral transcripts overlapped those of the selected housekeeping genes. In most detected transcripts containing parts of HK2 solo LTRs, the authors identified those LTRs as providing a polyadenylation signal and site. The claims have evidence level D and are subject to the stated limitations of transcript recovery and assignment.
What we don't know
These results do not establish corresponding patterns or mechanisms in FeLV. Estimated poly(A) tail lengths do not demonstrate equivalent RNA stability, translation efficiency or functional importance, and the solo LTR interpretation does not demonstrate downstream protein expression or biological effects. The supplied text provides no feline or FeLV experiments and no direct evidence about FeLV disease mechanisms or clinical outcomes. Applying the method to cats requires feline-specific validation. The omitted central passage and inconsistent library counts limit assessment of the study.
Evidence
EvidenceLevel D — Indirect evidence: other species or retrovirusConfidence 100/100
HERV-H was the most abundantly transcribed HERV group in the pooled NCCIT sample, with 24,031 reads; HK2 accounted for 9,193 reads.
Excerpt verified in the source — verbatim, in the source's own language“From the total retrotranscriptome, 64% were of HERV-H origin ( n = 24,031) (see Additional File 3), which constituted the most actively transcribed endogenous retrovirus group in our sample. Around 25% were HK2 transcripts ( n = 9,193), with the rest of ~ 11% ( n = 4,341) belonging to other annotated HERV families’ insertions within our database.”
These are read-assignment results from pooled human cell-line data, not counts of infected cells, animals or independent samples. They do not establish a corresponding transcriptional pattern in FeLV.
EvidenceLevel D — Indirect evidence: other species or retrovirusConfidence 100/100
The authors identify 5’ truncation caused by RNA degradation during handling as a limitation of recovering complete transcripts.
Excerpt verified in the source — verbatim, in the source's own language“However, due to inevitable RNA degradation during handling, the 5’ end of many reads (especially the longer ones) is truncated and therefore sequencing stops before reaching the actual 5’ end of the produced transcript. Thus, high-integrity input RNA is required for full-length reads.”
Incomplete 5’ coverage limits reconstruction of full-length transcripts and identification of their origins. This is a stated methodological caveat, not evidence that every detected isoform is artifactual.
EvidenceLevel D — Indirect evidence: other species or retrovirusConfidence 99/100
Estimated median poly(A) tail lengths of HK2 proviral transcripts overlapped those of the selected housekeeping genes.
Excerpt verified in the source — verbatim, in the source's own language“The range of the estimated adenines per HK2 derived proviral transcript, as depicted in Fig. 4 , suggests that the median poly(A) tail length estimates overlap with that of the housekeeping genes, and seem independent of sample size and transcript direction.”
The finding concerns estimated tail lengths. It does not directly demonstrate equivalent RNA stability, translation efficiency or functional importance, and it provides no feline-specific evidence.
EvidenceLevel D — Indirect evidence: other species or retrovirusConfidence 100/100
In most detected NCCIT transcripts containing parts of HK2 solo LTRs, the authors identified the solo LTR as providing a polyadenylation signal and site.
Excerpt verified in the source — verbatim, in the source's own language“In our study, in most transcripts of NCCITs that included parts of solo LTRs, the solo LTR was utilized for the provision of a polyadenylation signal and site (see Additional File 5).”
This is the authors’ interpretation of transcript structures in a human cell line. It does not establish the same mechanism in FeLV or demonstrate downstream protein expression or biological effects.
Provenance
Source: HERV-K HML-2 transcriptional profiling and splicing pattern unveiled by direct single-molecule long-read RNA sequencing.
AVA