A Master Gene Loose in a Poxvirus

On 24 September 2026, Pu Gao, Ellen Pritham, Cédric Feschotte and Cheng Sun reported an oddity in Science: the human noncoding RNA gene BC200 is sitting inside the genome of molluscum contagiosum virus, the poxvirus behind small, waxy skin papules in children. It is there twice, from two independent insertions. Both carry the signatures of LINE-1–mediated retrotransposition, and according to the Cornell release, both were acquired within roughly the last 100,000 years. Feschotte frames this as a broken rule: elements domesticated into cellular genes are supposed to stop moving. “Somehow evolution hasn’t been able to untangle these two things.” The locus becomes much easier to read under a name the rodent literature coined thirty-two years ago. BC200 is a master gene, in the technical sense Kim, Martignetti, Shen, Brosius and Deininger defined in PNAS in 1994.

Start with the gene’s anatomy. BC200 is about 200 nucleotides long, in three parts: a 5′ domain that is a monomeric Alu (derived ultimately from 7SL RNA), an A-rich linker, and about 40 nucleotides of unique 3′ sequence. RNA polymerase III transcribes it, so it carries its own internal promoter and needs no flanking regulatory context. It is a transcription unit that still works wherever it is copied. Kuryshev and colleagues sequenced the orthologous locus in lemur, galago and tarsier in 2001 and found no BC200 in any of them, so the founding Alu monomer landed after the anthropoids split off, 35–55 million years ago. The same paper counted more than 200 BC200 pseudogenes in the human genome and used the drift in their consensus sequence as a clock for changes in the parent gene. That is the master-gene idea in a sentence. Kim et al. had shown that in each rodent species, the dispersed ID elements track that species’ own BC1 gene, so the copies read as a stratigraphy of one source locus. Shen, Brosius and Deininger showed in 1997 why that locus in particular: BC1 transcripts prime their own reverse transcription, and the priming depends on the structure of the unique 3′ region. At this locus, being a good gene and being a good template are the same physical fact.

Read that way, the new paper completes a prediction. Its abstract reports that BC200 “functioned as a ‘master’ source of L1-mediated germline retrotransposition events throughout anthropoid evolution, spawning hundreds of lineage-specific insertions.” It also reports insertion polymorphisms still segregating in living humans and insertions found in single individuals, so the family is still growing. Kuryshev’s 200-plus pseudogenes were the fossil record, and Gao et al. found the process still running. The virus is almost incidental. Molluscum infects only keratinocytes, BC200 is expressed at low levels outside the brain, and L1 is active in skin. Put a self-priming Pol III transcript in a tissue where L1 machinery and a viral replication complex share a nucleus, and two captures in 100,000 years is ordinary arithmetic. The bigger story is convergence. Rodent BC1 and primate BC200 fill the same niche, as small dendritic RNAs implicated in local translational control, yet they are not homologs. BC1’s 5′ domain derives from a tRNA^Ala, while BC200’s is an Alu monomer. Two lineages independently promoted a SINE to a neuronal gene, and in both lineages the promoted SINE went on to seed a pseudogene family. Built twice from unrelated parts, “functional gene that is also a master source locus” looks like an attractor in mammalian genomes.

That suggests two concrete bets. The first ports Kim et al.’s 1994 experiment to primates. Sequence BC200-derived insertions lineage by lineage, and each lineage’s pseudogene consensus should track that lineage’s own BC200 allele, with the human-specific and polymorphic insertions carrying the modern human sequence. The second concerns the 120-nucleotide processed Alu monomer that Booy et al. reported in 2024 as unusually stable. It lacks the 3′ region that does the priming, so it should be a poor retroposition substrate, and the junctions of the polymorphic insertions should map to full-length BC200 rather than to BC120. If the new insertions instead turn out to be 5′-truncated at about 120 nucleotides, the self-priming account is wrong and the L1 machinery is doing the work unaided.