Functional Annotation Report: Mouse Rtl3 (Retrotransposon Gag-like protein 3)

Gene: Rtl3 (synonyms Mar3, MArt3, Zcchc5, SIRH9)
UniProt: Q6P1Y1 (RTL3_MOUSE) · 553 aa · ~61 kDa
Organism: Mus musculus (Mouse)
Chromosomal location: X chromosome
Ensembl: ENSMUSG00000047686

Summary

Mouse Rtl3 encodes a "domesticated" (co-opted) retrotransposon Gag-like protein derived from an ancient Ty3/gypsy (sushi-ichi / Metaviridae-type) LTR retrotransposon. It belongs to the Mar/Mart / Sushi-ichi-Related Retrotransposon Homolog (SIRH) / RTL family of neogenes that arose during early mammalian evolution and have since been maintained by purifying selection despite having lost the capacity for autonomous retrotransposition. The gene symbol is unambiguous for this target: the UniProt description, the Ty3/gypsy Gag/capsid + CCHC-zinc-finger domain architecture, and the mouse organism all align precisely with the literature discussed below. This is, however, a poorly characterized gene — the primary literature directly addressing Rtl3 function is thin, and much of what can be stated rests on structural/evolutionary inference and a small number of functional studies, some performed on the human ortholog (Q8N8U3).

The primary molecular function of the Rtl3 protein is not enzymatic or transport-related but is best described as zinc-dependent nucleic-acid binding. The canonical 553-aa product is a Gag-only module — it contains a retroviral capsid (CA)-like fold and a single, strongly conserved CCHC-type ("knuckle") nucleocapsid zinc finger (residues ~523–537), but it lacks the aspartyl-protease and the rest of the Pol region that autonomous retroelements use for self-processing and replication. Consequently Rtl3 cannot self-package or retrotranspose; its retained capsid + zinc-finger core instead provides a scaffold for single-stranded RNA/DNA binding (GO:0003676 nucleic acid binding; GO:0008270 zinc ion binding). Notably, the RTL3 locus retains an ancient programmed −1 ribosomal frameshift site between the capsid and protease domains that is differentially used across the family, so protease-extended products may exist in some contexts even though the reference isoform is capsid-only.

Functionally, Rtl3 acts as a low-abundance, tissue-restricted transcriptional / gene-regulatory factor. The two best-supported biological roles are (1) modulation of muscle satellite-cell self-renewal during postnatal myogenesis (UniProt-curated nuclear localization from Alonso-Martin et al. 2016), and (2) promotion of the SOX-9 / COL2A1 chondrogenic program in cartilage, demonstrated for the human ortholog by Ball et al. 2021, where deletion of the C-terminal CCHC zinc finger abolished regulatory activity. Unlike its closest relatives PEG10 and RTL1, Rtl3 is not a placental gene and is biallelically expressed (non-imprinted). Its site of action is most plausibly the nucleus for gene regulation, although antibody-based atlas data describe a cytoplasmic, granular pattern, leaving the precise subcellular localization genuinely unresolved.

Gene / Protein Identity Verification

Before presenting findings, the mandatory identity check:

Caveat on literature depth: Rtl3-specific primary literature is limited. Several of the strongest functional claims derive from (a) the human ortholog RTL3 (Q8N8U3), (b) family-wide surveys, or (c) structural/evolutionary inference. This is flagged throughout.

Key Findings

Finding 1 — Rtl3 is a domesticated, X-linked, purifying-selected Ty3/gypsy retrotransposon neogene

Rtl3/Mart3/Zcchc5 is a member of the Mar/Mart (SIRH) family of neofunctionalized Ty3/gypsy retrotransposon-derived genes. UniProt Q6P1Y1 assigns the 553-aa, ~61 kDa protein to this family and places the gene on the X chromosome. Two foundational surveys by Brandt et al. in 2005 (PMID: 15716091; PMID: 16093683) identified 11 Mar/Mart genes in human and mouse that evolved under purifying selection, retain intact open reading frames, but have lost retrotransposition capacity. The family arose after the bird–mammal split (~310 Mya) but before the eutherian–marsupial divergence (~170 Mya). Youngson et al. 2005 (PMID: 16155747) showed Rtl3 has strongest homology to the Fugu sushi-ichi element and, importantly, is biallelically expressed (non-imprinted) — distinguishing it from imprinted paralogs such as Peg10 and Rtl1.

"Mart genes have lost their capacity of retrotransposition through non-functionalizing rearrangements having principally affected long terminal repeats and pol open reading frame. Most Mart genes are located on the X chromosome in different mammals." — Brandt et al. 2005 (PMID: 15716091)

The maintenance of an intact ORF under purifying selection, despite loss of transposition machinery, is the hallmark of gene domestication: the host has co-opted the element to perform an endogenous cellular function.

Finding 2 — The protein combines a Gag capsid-like domain with a C-terminal CCHC zinc finger that confers nucleic-acid binding and is required for its regulatory activity

UniProt Q6P1Y1 annotates:
- a retroviral Gag capsid-like region (Pfam PEG10-RTL1_C-capsid / RTL1-8_LDOC; InterPro IPR063300, IPR032567, IPR032549),
- a CCHC-type zinc finger at residues 523–537 (Pfam zf-CCHC),
- an N-terminal coiled coil (res 2–43), and
- large intrinsically disordered regions.

GO molecular functions are nucleic acid binding (GO:0003676) and zinc ion binding (GO:0008270). Brandt 2005 (PMID: 16093683) noted conservation of the Gag CCHC zinc-finger motif in three Mar proteins, implying nucleic-acid binding. Critically, function hinges on this zinc finger: Ball et al. 2021 (PMID: 33043724) showed that deletion of the C-terminal CCHC zinc-finger domain (ΔRTL3) abolished RTL3's gene-regulatory activity on the SOX-9/COL2A1 axis in chondrocytes. This is the single most direct piece of evidence linking the conserved structural module to a measurable cellular output.

Finding 3 — Rtl3/Zcchc5 is a nuclear transcriptional regulator in postnatal myogenesis and chondrocyte COL2A1 expression

UniProt curates the FUNCTION as "May function as a transcriptional regulator; plays a role in postnatal myogenesis, may be involved in regulation of satellite cell self-renewal," with SUBCELLULAR LOCATION: Nucleus, drawn from Alonso-Martin et al. 2016 (PMID: 27446912). In that study, Zcchc5 is expressed in embryonic/fetal myogenic progenitor cells (12.5–17.5 dpc) but down-regulated in adult/quiescent satellite cells, and ZCCHC5 overexpression decreased satellite-cell self-renewal capacity — a specific, directional phenotype.

Independently, Ball et al. 2021 (PMID: 33043724) showed that the human ortholog RTL3 is strongly expressed in cartilage/chondrocytes/synoviocytes, that siRNA knockdown and CRISPR deletion reduced SOX-9 and COL2A1 expression and matrix synthesis, and that RTL3 co-regulates COL2A1 together with SOX-9. Mass spectrometry has detected the mouse protein in brain (PMID: 24129315; protein-level evidence). By contrast it is undetectable in mouse placenta (Henke et al. 2015, PMID: 25888968), unlike the placental paralogs Rtl1/Peg10.

Finding 4 — Structural/bioinformatic analysis: a Gag-only core (coiled-coil + capsid + single CCHC) with an acidic, proline-rich disordered linker resembling a transcription-factor activation domain

Comparative analysis of the 553-aa mouse sequence (Q6P1Y1) versus the 475-aa human ortholog (Q8N8U3) revealed:

1. A single retroviral nucleocapsid "knuckle" CCHC motif that is near-identical across species — mouse C523-LYCGHPGHFARDC vs human CLYCGYPGHFARDC (consensus CX₂CX₄HX₄C).
2. No aspartyl-protease catalytic motif (D-[ST]-G-[AG]) in the reference protein, confirming retention of only the Gag module (capsid + nucleocapsid zinc finger) and loss of the Pol/protease found in some other SIRH genes.
3. A central region (res 38–274) that is 46% Pro+Glu+Ser (19% Pro, 19% Glu) — an acidic, proline-rich low-complexity linker resembling transcription-factor activation domains, consistent with a gene-regulatory role.
4. AlphaFold (AF-Q6P1Y1) regional confidence (pLDDT):

The folded capsid + zinc-finger core is markedly more conserved between mouse and human than the divergent central linker — a classic signature of a functionally constrained structural module flanked by a rapidly evolving regulatory region.

Finding 5 — Low-abundance, narrowly expressed gene with an unsettled subcellular localization

Bgee multi-assay expression calls for mouse Rtl3 (ENSMUSG00000047686; Affymetrix/EST/ISH/RNA-seq/scRNA-seq) return predominantly "not_expressed" / low expression scores (~34–41/100) across brain, testis, gonad, synovial joint, liver, spleen and digestive tract — indicating a lowly and context-restricted gene rather than one with a single dominant tissue. This is consistent with the literature: undetectable in placenta (PMID: 25888968), restricted to fetal myogenic progenitors and down-regulated in adult satellite cells (PMID: 27446912), and cartilage/chondrocyte-enriched for the human ortholog (PMID: 33043724).

Localization is genuinely unresolved. UniProt curates Nucleus (from the myogenesis functional study, PMID: 27446912), consistent with a transcriptional role. However, the Human Protein Atlas antibody data describe ubiquitous cytoplasmic expression in a granular pattern for human RTL3 — a pattern more reminiscent of Gag proteins that assemble RNP granules. These two observations are not easily reconciled and are discussed under Limitations.

Finding 6 — Rtl3's closest paralog by its conserved module is PEG10, placing it among the capsid-forming, RNA-binding domesticated Gag proteins

A Smith-Waterman local alignment (BLOSUM62) of mouse Rtl3 against the mouse RTL/Mart/LDOC paralogs, scored over the conserved Gag capsid + zinc-finger region, ranked similarity as:

Thus the retained folded module of Rtl3 is most similar to PEG10, the best-characterized SIRH/RTL gene. Like LDOC1 and the short RTLs — but unlike PEG10 and RTL1 — Rtl3 is Gag-only (no Pol/protease), so it cannot self-package autonomously. Related Ty3/gypsy-derived Gag proteins are experimentally capsid-forming and RNA-binding: Arc "forms virus-like protein capsids that can transfer RNA between neurons" and is "derived from an ancient retrotransposon of the Ty3/gypsy family and contains homology to the retroviral Gag polyproteins" (Hantak et al. 2021, PMID: 33485691); PEG10 and Arc "structurally demonstrate significant similarity with Gag retroviral proteins" and can be engineered into capsid particles for mRNA delivery (Al Othman et al. 2025, PMID: 38798208). By analogy, Rtl3's capsid + CCHC core is a plausible RNA-binding / RNP-scaffolding module, even if autonomous capsid assembly is not proven for Rtl3 itself.

Finding 7 — Authoritative 2024 evolutionary analysis: RTL3 is a domesticated Metaviridae capsid gene retaining an ancient, differentially used capsid–protease ribosomal frameshift

Henriques, Young, … Malik 2024 (Mol Biol Evol 41:msae061; PMID: 38507667), analyzing a complete human genome plus 25 vertebrate genomes, place the SIRH/RTL family among capsid (CA) genes domesticated from Metaviridae (Ty3/gypsy-type retrovirus-like retrotransposons). They report:

"In the SIRH/RTL family, we find diverse amino-terminal domains … and differential utilization of an ancient programmed ribosomal frameshift in RTL3 between the domesticated CA and protease domains."

They further find these domesticated capsid genes evolve under purifying selection, that two families are unique to placental mammals, and that some members show rapid evolution. This refines the within-project structural observation (Finding 4) that the single annotated 553-aa mouse isoform has no in-frame aspartyl-protease motif: the reference protein is capsid(Gag)-only, but the RTL3 locus retains an ancestral −1 programmed ribosomal frameshift capable of extending translation from the capsid frame into a domesticated protease domain, with usage differing across the family / lineages. This is a distinctive, named feature of RTL3 specifically.

Mechanistic Model / Interpretation

Integrating the seven findings yields the following coherent model of Rtl3 as a domesticated Gag-derived nucleic-acid-binding transcriptional modulator:

      Ancestral Ty3/gypsy (sushi-ichi / Metaviridae) LTR retrotransposon
                   │
    domestication (~170–310 Mya, mammalian lineage)
    loss of LTRs + Pol/protease → purifying selection
                   │
                   ▼
   ┌───────────────────────── Rtl3 / Mart3 / Zcchc5 (553 aa) ─────────────────────────┐
   │                                                                                   │
   │  [Coiled coil]   [Acidic/Pro-rich disordered linker]   [Gag CAPSID fold]  [CCHC] │
   │   res 2–43           res 38–274 (46% P+E+S)             res ~370–510    523–537  │
   │   pLDDT 89.9          pLDDT 40.4 (disordered)           pLDDT 70.9      pLDDT 88  │
   │   oligomer?        activation-domain-like               RNA/ssDNA        Zn²⁺,    │
   │                                                          scaffold      ssNA bind  │
   │                                                                                   │
   │        ⟵─ latent −1 programmed ribosomal frameshift → (domesticated protease) ─⟶ │
   │              (reference isoform = capsid-only; frameshift differentially used)    │
   └───────────────────────────────────────────────────────────────────────────────┘
                   │
 nucleic-acid binding (GO:0003676) + Zn²⁺ (GO:0008270)
                   │
    ┌──────────────────────┴──────────────────────┐
    ▼                                              ▼
   NUCLEAR gene regulation                      (cytoplasmic granular pattern,
   (UniProt-curated)                             HPA antibody — unresolved)
    │
   ┌────────┴────────────────────────────────┐
   ▼                                          ▼
 Postnatal myogenesis:                 Chondrogenesis (human RTL3):
 ↓ satellite-cell self-renewal         co-regulates SOX-9 → COL2A1
 (Alonso-Martin 2016)                  CCHC deletion abolishes activity
                       (Ball 2021)

Primary molecular function: Rtl3 is not an enzyme or transporter. Its defining biochemical activity is zinc-dependent single-stranded nucleic-acid (RNA/ssDNA) binding through the combination of a retroviral capsid fold and a single, strongly conserved CCHC nucleocapsid zinc finger. The capsid fold provides an oligomerization/scaffolding surface (by homology to PEG10/Arc), while the CCHC knuckle provides sequence-tolerant ssNA contact. There is no evidence for a catalyzed reaction in the canonical isoform; the only latent enzymatic potential is the frameshift-accessible domesticated protease domain, whose usage and substrate (if any) remain undemonstrated.

Cellular role: Rtl3 behaves as a transcriptional / gene-regulatory factor. The acidic, proline-rich disordered central region resembles a transcription-factor activation domain, and the CCHC zinc finger is strictly required for regulatory output. The protein modulates two differentiation programs: it restrains satellite-cell self-renewal in postnatal muscle (so its down-regulation in adult quiescent satellite cells is part of maintaining the stem pool), and it promotes the SOX-9/COL2A1 chondrogenic program in cartilage. Whether it binds DNA, mRNA, or acts via an RNP complex is not established; its most likely mode is nuclear RNA/chromatin-associated regulation.

Site of action: Most plausibly nuclear (UniProt-curated, from the functional myogenesis study), consistent with a transcriptional role. The conflicting cytoplasmic-granular antibody pattern may reflect a genuine dual localization (as seen for some Gag-derived RNP proteins), antibody cross-reactivity, or context-dependent shuttling — it is not resolved.

Pathways: The concrete pathway linkage is the SOX-9 → COL2A1 cartilage extracellular-matrix program (Ball 2021), where Rtl3/RTL3 acts as a co-regulator alongside the master chondrogenic transcription factor SOX-9. In muscle, it intersects the satellite-cell self-renewal vs. differentiation decision. Broad pleiotropic/placental roles that dominate its paralogs (PEG10, RTL1) are explicitly absent for Rtl3.

Evidence Base

Consistency / conflict assessment: The structural (F2, F4), evolutionary (F1, F7), and functional (F3) lines of evidence are mutually reinforcing: a conserved capsid+CCHC module under purifying selection, with the zinc finger experimentally required for gene regulation. The one genuine conflict is subcellular localization (nuclear per functional study vs. cytoplasmic-granular per antibody atlas; F5). A secondary limitation is that the most mechanistic functional study (Ball 2021) used the human ortholog, not mouse.

Limitations and Knowledge Gaps

1. Sparse direct literature. No dedicated mouse Rtl3 loss-of-function (knockout) phenotype is available. The myogenesis role rests largely on one study with overexpression/expression data (PMID: 27446912); the chondrogenesis role is from the human ortholog (PMID: 33043724).
2. Unresolved subcellular localization. Nuclear (curated) vs. cytoplasmic-granular (antibody atlas) — not reconciled. Antibody specificity for this low-abundance protein is a concern.
3. Molecular target(s) unknown. Whether Rtl3 binds DNA, specific mRNAs, or acts through protein–protein interactions is undetermined. No direct RNA/DNA target, binding motif, or affinity has been measured. "Transcriptional regulator" is inferred, not biochemically defined.
4. Frameshift product uncharacterized. The RTL3-specific programmed −1 frameshift (PMID: 38507667) implies a potential capsid–protease fusion isoform whose existence, abundance, catalytic activity, and substrate in mouse are unknown.
5. Capsid assembly not demonstrated for Rtl3. Capsid/RNA-binding behavior is inferred from PEG10/Arc; Rtl3 itself has not been shown to form particles or bind a defined nucleic acid in vitro.
6. Low abundance / context restriction makes endogenous detection and functional assays technically difficult, which likely explains the thin literature.

Proposed Follow-up Experiments / Actions

1. Mouse Rtl3 loss-of-function. Generate a conditional/global Rtl3 knockout and phenotype muscle regeneration (satellite-cell self-renewal) and cartilage/chondrocyte compartments, to test the two proposed roles in the native organism.
2. Define molecular targets. Perform CLIP-seq (RNA targets) and CUT&RUN/ChIP-seq (chromatin occupancy) on tagged Rtl3 in a relevant cell type (chondrocytes, myogenic progenitors) to determine whether it binds RNA, DNA, or both, and identify a binding motif.
3. Resolve localization. Endogenous tagging (CRISPR knock-in of a fluorescent/epitope tag) plus fractionation to settle the nuclear vs. cytoplasmic-granular question and test for context-dependent shuttling.
4. Zinc-finger structure–function. Recombinant capsid+CCHC core: test in vitro ssRNA/ssDNA binding, Zn²⁺ dependence, and whether point mutations in the CCHC knuckle (vs. ΔRTL3 deletion) abolish binding — linking the Ball 2021 functional requirement to a biochemical activity.
5. Test the frameshift isoform. Use ribosome profiling / targeted proteomics to detect any CA–protease frameshift product in mouse tissues; if present, assay the domesticated protease for activity and substrate.
6. Capsid-assembly assay. Express the Gag/capsid core and test for virus-like particle formation and RNA packaging (as done for PEG10/Arc), to determine whether Rtl3 retains self-assembly despite lacking Pol.
7. SOX-9 pathway epistasis. In mouse chondrocytes, test whether mouse Rtl3 phenocopies human RTL3 in the SOX-9 → COL2A1 axis, establishing cross-species conservation of the mechanistic role.

Bottom line

Mouse Rtl3 is a domesticated, non-autonomous retrotransposon Gag-like protein whose primary molecular function is zinc-dependent single-stranded nucleic-acid binding (via a conserved capsid fold + CCHC zinc finger), repurposed by the host as a low-abundance, tissue-restricted transcriptional/gene-regulatory factor acting most plausibly in the nucleus — with documented roles in restraining muscle satellite-cell self-renewal and promoting the SOX-9/COL2A1 chondrogenic program — and, unlike its closest relatives PEG10/RTL1, it is not imprinted and not a placental gene.

Artifacts

• OpenScientist final report
• OpenScientist final report

Citations

1. PMID:15716091
2. PMID:16093683
3. PMID:16155747
4. PMID:33043724
5. PMID:27446912
6. PMID:24129315
7. PMID:25888968
8. PMID:33485691
9. PMID:38798208
10. PMID:38507667