DNASE2 Phagotroph Distribution Hypothesis — Deep Research Report

Target gene: DNASE2 (human), UniProt O00115
Family: InterPro IPR004947 = PANTHER PTHR10858 = Pfam PF03265
Hypothesis slug: dnase2-phagotroph-distribution
Focus type: free_text (evolutionary / phylogenetic distribution)
Source review: genes/human/DNASE2/DNASE2-ai-review.yaml

Summary

The seed hypothesis proposes that DNase II family enzymes — which digest the DNA
of engulfed cells in animal phagolysosomes — are present in phagotrophic protists
(Dictyostelium discoideum, Acanthamoeba, choanoflagellates), and that this
distribution is "consistent with digestion of phagocytosed DNA being an ancestral
eukaryotic function." The requested test was the phylogenetic distribution of
DNase II homologs across eukaryotic reference proteomes.

I evaluated this by tabulating InterPro/UniProt protein counts for family
IPR004947 (equivalently PANTHER PTHR10858 / Pfam PF03265) across major
eukaryotic lineages, with protein-level identity audits of surprising hits and a
proteome-coverage control for the key negative example. The verdict is partially
supported. Two of the three named phagotroph examples are genuine:
Dictyostelium discoideum encodes a real ~14-member family expansion and
Acanthamoeba castellanii encodes 3 bona fide members. The family is indeed
ancient and broadly eukaryotic (5,031 eukaryotic proteins across 1,423 reference
proteomes), reaching Amoebozoa, land plants, ciliates and stramenopiles. However,
the third named example — choanoflagellates — is refuted: both well-sequenced
species (Monosiga brevicollis, Salpingoeca rosetta) lack the family entirely,
and this is not a proteome-coverage artifact.

More importantly, the evolutionary interpretation is only weakly supported.
Presence of the family does not track phagotrophy: non-phagotrophic land plants
retain a robust complement (72 proteins) while several phagotrophic lineages
(kinetoplastids, Giardia/Fornicata, and the choanoflagellates themselves) have
lost it. The pattern is best explained as an ancient eukaryotic family subject
to many independent lineage-specific losses, not as a clean marker of an
ancestral phagocytic-DNA-digestion function. For curation, the evolutionary
free-text should be narrowed and the choanoflagellate claim corrected; this
observation is orientation/context only and should not by itself drive any GO
term change on the human DNASE2 record.

Key Findings

Finding 1 — The DNase II family is patchily distributed across eukaryotes: present in Amoebozoan phagotrophs but absent from choanoflagellates

Using the EBI InterPro and UniProt REST APIs, I tabulated protein counts for
family IPR004947 (= PTHR10858 = PF03265) across major eukaryotic taxa. The
family totals 5,031 proteins in Eukaryota, distributed across 1,423
reference proteomes — confirming a deep, broad eukaryotic presence.

The choanoflagellate absence was explicitly tested against a proteome-coverage
control: Monosiga brevicollis (9,331 UniProt proteins) and Salpingoeca
rosetta (12,235 UniProt proteins) both have well-populated proteomes, so a count
of zero DNase II proteins reflects a genuine evolutionary loss, not missing
sequence data.

This finding simultaneously supports the hypothesis for the two named amoebae
and refutes it for choanoflagellates. It also undercuts the deeper inference:
phagotrophy does not predict presence, because several phagotrophic lineages lack
the family while non-phagotrophic land plants retain it.

The function the hypothesis attributes to the family is itself well established.
The DNase II review states that "Its lysosomal localization and ubiquitous tissue
distribution suggested that this enzyme played a role in the degradation of
exogenous DNA encountered by phagocytosis"
(PMID: 14644493). In an invertebrate
gut, a DNase II homolog digests engulfed/dietary bacterial DNA — DNase II
homologs "both fail to replace nuc-1 in degrading bacterial DNA in intestine"
(PMID: 19809494). These confirm the
engulfed-DNA digestion role beyond mammals but do not establish that this
function is ancestral or phagotrophy-linked across the eukaryotic tree.

Finding 2 — Protein-level identity checks confirm genuine DNase II in amoebae and plants, and show the sole fungal "hit" is spurious — strengthening the loss pattern

Because InterPro signature matches can occasionally be incidental, each surprising
hit was inspected at the protein-annotation level:

• Dictyostelium discoideum hits are annotated "Deoxyribonuclease II" and
  "Deoxyribonuclease II family protein" (e.g., C7G058, Q54FV5) — confirming a
  genuine ~14-member family expansion rather than low-quality matches.
• Viridiplantae hits include real, functionally named members:
  "Cell-death-related nuclease 7", "Plancitoxin-1", and
  "Deoxyribonuclease-2-alpha" — bona fide DNase II enzymes in a
  non-phagotrophic lineage.
• The single Fungi hit (A0A2H9TI56) is annotated "Repressor of RNA polymerase
  III transcription" — an incidental/spurious signature match, not a true
  DNase II. Fungi are therefore effectively devoid of the family.

This audit removes the one apparent counter-example to fungal loss, confirms that
plant retention is real, and verifies that the Dictyostelium expansion is
authentic. Together these sharpen the conclusion that the presence/absence pattern
is driven by repeated independent losses of an ancient family rather than by
phagotrophic lifestyle.

Finding 3 — The catalytic domain is not strictly eukaryote-specific

An outgroup query of IPR004947 returned Bacteria 66, Archaea 0, Viruses 6
against Eukaryota 5,031. The family is overwhelmingly eukaryotic but not strictly
eukaryote-restricted; sparse bacterial members suggest a deeper domain origin or
horizontal transfer. This is a minor qualifier to the seed's phrase "ancestral
eukaryotic function": the catalytic domain may predate or extend beyond
eukaryotes even though the lysosomal/phagolysosomal DNA-digestion role is a
eukaryotic elaboration. It does not change the core verdict.

Mechanistic Model / Interpretation

The molecular function under test is narrow and well-defined: DNase II–family
enzymes are acid (lysosomal/phagolysosomal) endonucleases that cleave DNA at
low pH without divalent-cation dependence, degrading DNA delivered into the
lysosome — from apoptotic/engulfed cells, phagocytosed microbes, or dietary
sources. The hypothesis takes this well-supported function and makes an
evolutionary extrapolation: that the enzyme's distribution should mark
phagotrophy as an ancestral eukaryotic activity.

The data support a more parsimonious model:

 Ancient DNase II family (domain predates/spans eukaryotes;
             sparse bacterial members)
                     │
┌────────────────────────────┼────────────────────────────┐
│                            │                            │
   RETAINED                     RETAINED                       LOST (independently,
   in many phagotrophs          in NON-phagotrophs             many lineages)
   (Dictyostelium 14,           (land plants 72,               Choanoflagellata 0
    Acanthamoeba 3,              stramenopiles 62)             Fungi ~0 (spurious)
    ciliates 55)                                               Kinetoplastida 0
                                               Apicomplexa 0
                                               Giardia/Fornicata 0
                                               Rhodophyta 0

   ⇒ Distribution = "ancient family + many lineage-specific losses"
   ⇒ NOT "clean marker of phagotrophic DNA digestion"

Key logical points:

1. Presence in phagotrophs is consistent with — but not diagnostic of — an
   ancestral phagocytic role. Dictyostelium and Acanthamoeba fit, but the
   inference is confounded by the exceptions below.
2. Retention in non-phagotrophic plants shows the family is maintained for
   functions unrelated to phagocytosis (e.g., programmed-cell-death nucleases).
3. Loss in multiple phagotrophs (choanoflagellates, kinetoplastids, Giardia)
   shows phagotrophy can proceed without this specific family — other acid
   nucleases can substitute.
4. Sparse bacterial members indicate the catalytic domain's deep origin,
   offering an alternative to a strictly eukaryotic-phagotrophic framing.

The correct framing for curation is therefore evolutionary context, not
mechanistic evidence for the human gene's GO annotations. Human DNASE2 function
(lysosomal DNA catabolism, clearance of DNA from engulfed apoptotic cells and
erythroid nuclei) is established by direct mammalian and invertebrate experiments;
the cross-eukaryote distribution neither adds to nor subtracts from those
annotations.

Evidence Base

GO Curation Implications

Likely curation action (lead — requires curator verification): Narrow and
correct the evolutionary free-text; make no GO term change driven by this
hypothesis.

• No MF change. Deoxyribonuclease II activity (GO:0004531; acid,
  cation-independent endonuclease) is established by direct biochemistry and is
  unaffected by the distribution survey. Retain.
• No BP change. DNA catabolism in the context of engulfed/apoptotic DNA
  (e.g., GO:0006308 "DNA catabolic process"; apoptotic-cell-clearance–associated
  DNA degradation) is supported by direct metazoan genetics (PMIDs 19809494,
  26182365, 41805323), not by the distribution survey. Do not add a
  phagocytosis-pathway BP term to the human gene on evolutionary grounds alone.
  Retain existing.
• No CC change. Lysosome localization (GO:0005764) is independently
  established. Retain.
• Edit the evolutionary free-text. As written it (a) names choanoflagellates
  as a positive example when they lack the family, and (b) implies phagotrophy
  predicts presence when the distribution does not support that. Treat as
  non-core orientation that should not alter the gene's GO annotation set.
• Over-annotation caution. Lineage-specific expansions (Dictyostelium ×14)
  mean family-level presence must not be transferred as 1:1 orthology of human
  DNASE2 or of a specific sub-function.

This is a free-text / evolutionary-context item, not a function-assignment
item for the human gene.

Mechanistic Scope

• Direct gene-product activity (in scope, well-supported): an acid-optimum,
  divalent-cation-independent endonuclease that cleaves dsDNA within the
  lysosome/phagolysosome. This is the immediate, cell-autonomous molecular
  function, documented biochemically in mammals and genetically in C. elegans.
• Downstream phenotypes (out of scope for this evolutionary hypothesis):
  failure of DNA clearance → undegraded cytosolic DNA → cGAS–STING/interferon
  activation → autoinflammation (human AIPCS,
  PMID: 42901880); innate-immune gene
  upregulation in C. elegans
  (PMID: 26251453). These are
  loss-of-function consequences, not evidence for the ancestral-function claim.
• Evolutionary distribution is an orthology/inference layer, not a direct
  functional assay; it constrains ancestry but cannot assign the ancestral
  function.

Conflicts and Alternatives

1. Choanoflagellate counter-evidence (direct conflict). The seed names
   choanoflagellates as a positive example, but Monosiga brevicollis and
   Salpingoeca rosetta have zero members in well-covered proteomes — refuting a
   named example.
2. Phagotrophy does not predict presence (interpretive conflict). Phagotrophic
   kinetoplastids and Giardia lack the family; non-phagotrophic land plants
   retain it. The "ancestral phagotroph" framing is confounded.
3. Paralog expansion / annotation inflation. Counts reflect family-signature
   hits, including paralogs; raw counts are not a phylogeny and can overstate
   independent acquisitions. The identity audit mitigated but did not eliminate
   this.
4. Spurious signature matches (database carry-over). The lone fungal hit was a
   mis-annotated "Repressor of RNA Pol III," showing raw InterPro counts can
   contain false positives; single-hit "presence" needs identity checks.
5. Deep (pre-eukaryotic) origin. Sparse bacterial members indicate the domain
   predates eukaryotic phagotrophy, an alternative to the "ancestral eukaryotic
   phagocytic function" story.

Limitations and Knowledge Gaps

Proposed Follow-up Experiments / Actions

Discriminating analyses

• Profile-HMM (PF03265) search of raw choanoflagellate, fungal, and additional
  protist genomes to firm up presence/absence beyond InterPro annotation recall.
• Reconciled gene tree of PTHR10858 with Dollo/ML parsimony to count
  independent losses and test the LECA-origin hypothesis.
• Phylogenetic comparative test (Pagel / phylogenetic logistic regression) of
  family presence vs phagotrophic trait — the decisive test of the seed's core
  inference.
• BUSCO completeness controls for all "absent" lineages.
• Functional assay: knock out/down DNase II in Dictyostelium and test for
  persistence of undigested DNA in phagosomes — a direct test of the
  ancestral-function claim.

Curation leads (verify before applying)

• Action: Soften/correct the free-text evolutionary statement; remove or
  reverse the claim that choanoflagellates carry DNase II. Suggested replacement:
  "DNase II (IPR004947 / PTHR10858 / PF03265) is an ancient, broadly distributed
  eukaryotic family subject to many independent losses. It is present in
  phagotrophic amoebae such as Dictyostelium discoideum (a ~14-member expansion)
  and Acanthamoeba castellanii, and in ciliates, stramenopiles, and land plants,
  but is absent from choanoflagellates, kinetoplastids, Giardia, Apicomplexa, and
  (apart from a spurious hit) fungi; sparse bacterial members also exist. Family
  presence does not track phagotrophy, so the distribution reflects an ancient
  family with repeated lineage-specific losses rather than a diagnostic marker of
  ancestral phagocytic DNA digestion."
• Candidate references: PMID: 14644493
  (phagocytic exogenous-DNA role, review);
  PMID: 19809494 &
  PMID: 26182365 (engulfed/apoptotic
  DNA digestion, C. elegans);
  PMID: 41805323 (phagosomal DNA
  clearance, macrophages). Add as BP/MF support; keep evolutionary distribution as
  a computed note citing InterPro IPR004947 / PANTHER PTHR10858.
• Candidate GO: retain MF deoxyribonuclease II activity (GO:0004531), CC
  lysosome (GO:0005764), BP DNA catabolism in apoptotic/engulfment context; do
  not elevate the evolutionary narrative to a GO annotation.
• Suggested question for curator: Should the review assert a protist-conserved
  ancestral phagotrophic function, given that the distribution is patchy and the
  flagship choanoflagellate example is negative?

Hypothesis Scorecard

Report generated from a 3-iteration autonomous investigation. Distribution counts
were obtained from the EBI InterPro and UniProt REST APIs for family IPR004947 /
PTHR10858 / PF03265 (UniProtKB 2026-10); all presence/absence calls for named
lineages were cross-checked at the protein-annotation level, and the
choanoflagellate absence was controlled against proteome coverage. Results are
reported conservatively: database signature counts are not a formal phylogeny, no
gene tree or ancestral-state reconstruction was run, and the phagotrophy-linkage
inference remains statistically untested. Provenance artifacts:
/tmp/outputs/dnase2_taxonomic_distribution.csv,
/tmp/outputs/dnase2_distribution.png.

Artifacts

• OpenScientist final report
• OpenScientist final report