NOTE-1241: Sørensen, Stiller, Kröninger & Nowack — Protein import into bacterial endosymbionts and evolving organelles

Contribution

It gathers, system by system, the direct evidence that host-encoded proteins reach the inside of bacterial endosymbionts. Before it, the textbook line was that endosymbionts are “genetically autonomous”: they take metabolites from the host but not proteins. The review shows two cases of import at scale and a scatter of single proteins in others. It also separates two things the record had run together: gene transfer from the endosymbiont to the nucleus, and genetic integration by import. The second happens without the first.

Key insight

An organelle, on the authors’ definition, is an endosymbiont “in which essential endosymbiont genes have been functionally replaced by nuclear genes with products that are imported into the endosymbiont”. The replacing genes need not come from the endosymbiont. In Paulinella the hundreds of imported proteins “derive from EGTs and HGTs, but mostly from the host itself”. In the nitroplast, “Evidence for EGT has not been found”, yet more than 350 nucleus-encoded proteins are imported. Hence: “It is important to note that EGT is not a precondition for the genetic integration of an endosymbiont.” What makes the organelle is a host protein that reaches inside and a symbiont gene that is then lost.

Assumptions

• Proteomics of isolated endosymbionts detects import. A host protein enriched in an endosymbiont fraction is taken as imported. For the diatom, the host membrane is lost on isolation, so enrichment shows targeting “at least across this outermost host membrane”, not translocation into the bacterium.
• Antibody localization shows import. Single-protein cases (RlpA4 in Buchnera, the GOT enzymes in Nardonella, ColA in Sodalis, MurF in Moranella) rest on immunofluorescence or immunogold microscopy. The weevil case is flagged: “Further experiments are needed to confirm the import of these proteins”.
• Absence of evidence is weak. Proteome coverage is partial: 6.5% of predicted nucleus-encoded proteins in the diatom, and 1940 of 34 616 predicted aphid proteins (5.6%). Systems with few imported proteins may have more.
• A metabolically integrated endosymbiont “should, in theory, be viable outside the host when supplemented with the right metabolites”. This is the test that separates the two kinds of integration. It is stated, not applied to any system.

Key results

• Mitochondria and plastids. Their genomes encode “no more than 66 and 240 proteins”. Their import channels are of mixed origin: Tom40 has no identified bacterial precursor, Toc75 is related to bacterial BamA, and the Tim17/22/23 family’s ancestor “remains obscure”. Bacteria do not have import systems to start from, since export systems “cannot be simply operated in reverse”. So organellogenesis needed “a major restructuring of the bacterial envelope”.

• Envelope loss in reduced symbionts (Figure 2A, KEGG presence/absence). Some endosymbionts lost phospholipid, PG or LPS synthesis, and Sec, Tat and BAM. In genomes under 0.2 Mbp “often all of the above-mentioned envelope-related functions are missing”, so they “completely depend on host-derived lipids and presumably protein complexes”. Only the caption of Figure 2 was read.

• The chromatophore of Paulinella.

  • Acquired ”~ 90–140 million years ago” from a cyanobacterium of the Synechococcus/Prochlorococcus clade.
  • Genomes of ~1 Mbp with “878–860 protein‐coding genes” lost “around 2/3 of their original coding capacity”.
  • About 60 EGT-derived and 170 HGT-derived nuclear genes add to the complementarity.
  • Radiolabelling with ribosome-specific inhibitors showed the EGT-derived photosystem I subunits PsaE and PsaK made on host ribosomes and assembled into chromatophore photosystem I. PsaE was also seen in the Golgi.
  • Proteomics found “hundreds” of imported proteins in two classes: short ones (under 90 aa) without obvious signals, and long ones (over 250 aa) with a ~200 aa bipartite “chromatophore transit peptide” whose first ~50 aa are cleaved and the rest retained.
  • About 20% of long imported proteins have genetic-information roles (DNA polymerase I, helicases, a ligase, transcription factors), and a family of ~20 octotrico peptide repeat proteins suggests control of chromatophore gene expression. “In sum, the chromatophore is a new genetically integrated photosynthetic organelle.”
  • The route runs through the secretory pathway, “very different” from TIC/TOC and TIM/TOM. The outer chromatophore membrane is “interpreted” from micrographs as host-derived.

• The nitroplast (UCYN-A in Braarudosphaera bigelowii).

  • About 91 million years of coevolution. Genome ~1.4 Mbp, without photosynthesis or carbon fixation, with all nif genes.
  • One UCYN-A per host cell, dividing with it and vertically transmitted. But “when nitrogen compounds are externally provided in culture, B. bigelowii can lose UCYN‐A”.
  • No EGT found. More than 350 nucleus-encoded proteins imported, many filling pathway gaps. They include a methionine-tRNA formyltransferase that “likely replaces” the lost fmt gene.
  • 62% of enriched host proteins carry a C-terminal extension, the “UCYN-A transit peptide”, presumably cleaved on import. Its C-terminal position suggests proteins are folded before recognition.
  • “the nitroplast shows a similar integration level as the chromatophore”.

• Spheroid bodies in Epithemia. 12–35 million years old, genomes 2.5–3.0 Mbp. Vertically transmitted, two to 16 per cell. Only six host proteins enriched, none of bacterial origin. They “apparently have not reached the same level of genetic integration”.

• Kinetoplastibacterium in Angomonas. Genome 0.8 Mbp, one symbiont per host, division synchronized. Seven targeted host proteins found with fluorescent fusions. Two localize at the endosymbiont division site: a predicted PG hydrolase and a dynamin-like protein, which suggests “nuclear control over endosymbiont division”. One EGT gene is known, and its product goes to glycosomes, not the symbiont.

• Insects.

  • Buchnera (0.64 Mbp): no host protein “reliably enriched” in the symbiont by proteomics. One protein, RlpA4, was localized to Buchnera’s cytoplasm with antibodies. It is encoded by a gene horizontally transferred from a bacterium other than Buchnera. “no functional EGT occurred in the pea aphid”. Bacteriocyte-specific cysteine-rich peptides (BCRs) kill E. coli in vitro. Whether they enter Buchnera has “yet to be demonstrated”.
  • Nardonella in the red palm weevil (0.20–0.23 Mbp, 196–231 genes): two host transaminases complete its one remaining job, tyrosine synthesis, and were seen in its cytoplasm. Import is not yet confirmed.
  • Sodalis in the cereal weevil (4.51 Mbp, replacing Nardonella 20 million years ago): the host antimicrobial peptide ColA enters the symbiont’s cytoplasm and regulates its growth. “relatively recently captured endosymbionts can already be targeted by host effector proteins”.
  • The mealybug: the imported protein is MurF, found in the cytoplasm of Moranella (0.54 Mbp), the inner symbiont, not Tremblaya (0.14 Mbp). Its gene came by HGT. It must cross five membranes. One protein doing so makes it “in our opinion likely” that a general mechanism exists.

• Rhizobia. Medicago targets about 700 NCR peptides to its bacteroids, which control division, endoreduplication and translation. Yet “we would not consider rhizobia as genetically integrated organelles”. They are not vertically transmitted, keep their soil life, and no endosymbiont function has been replaced by a nuclear gene. Effector manipulation is “an effective alternative to the genetic integration that is always linked to a permanent host/endosymbiont association with vertical transmission”.

• A scenario. “A scenario for progressing organellogenesis that appears plausible in the light of the available, though limited, data”:

  1. The host first targets effector proteins into the endosymbiont.
  2. A nuclear protein redundant with a symbiont function gains import. Mutations in the symbiont copy can then fix.
  3. If metabolite import cannot replace the process, the endosymbiont becomes dependent.
  4. A general import mechanism with a targeting signal “likely spreads quickly”, bringing hundreds of proteins.

  Whether EGT contributes depends on how many genes are replaced, the host’s access to HGT, and the rate of nuclear integration of symbiont DNA.

Claims

id	claim	strength	support
C1	Hundreds of nucleus-encoded proteins are imported into the Paulinella chromatophore and the nitroplast, and replace lost endosymbiont genes	strong	proteomics of isolated organelles, targeting signals conserved across many proteins, cited primary studies
C2	Gene transfer from the endosymbiont is not a precondition for genetic integration	strong for the nitroplast, moderate in general	no EGT found in B. bigelowii; Paulinella’s imported proteins mostly host-derived; two cases
C3	Single host-encoded proteins enter some insect endosymbionts and replace lost functions	moderate for Buchnera and Moranella, weak for Nardonella	antibody localization of one or two proteins each; import in the weevil “needs confirmation”
C4	Protein import into endosymbionts is more common than generally realized	weak to moderate	the authors’ belief from scattered cases and partial proteome coverage; stated as “we believe”
C5	Endosymbionts under about 0.2 Mbp depend on host lipids and probably host proteins to build their envelope	moderate for lipids, weak for proteins	gene absence (Figure 2A); compensation mechanisms “remain unclear”
C6	Effector targeting precedes and leads into genetic integration	weak	a scenario called plausible “in the light of the available, though limited, data”
C7	Genetic integration requires a permanent, vertically transmitted association; effector control alone does not produce it	moderate as an argument	rhizobia as the contrast case; one counter-example class, no test
C8	Different systems evolved different import mechanisms (secretory route and N-terminal crTP in Paulinella, C-terminal uTP in the nitroplast, signal peptides in insects and legumes)	moderate	signal comparisons (Figure 4); the mechanisms beyond the signals are unknown

Concepts

• Genetic integration: essential endosymbiont genes “functionally replaced by nuclear genes with products that are imported into the endosymbiont”. The authors’ mark of an organelle.
• Metabolic integration: the endosymbiont receives only metabolites, though host effectors may still act on it. In theory such a symbiont is viable outside the host with the right metabolites.
• EGT: endosymbiotic gene transfer, from the endosymbiont to the nucleus. Distinguished from HGT, transfer from other bacteria.
• ETP: endosymbiont-targeted protein.
• crTP and uTP: the chromatophore transit peptide (N-terminal, bipartite, partly retained) and the UCYN-A transit peptide (C-terminal, cleaved).
• Effector: a host protein that manipulates the endosymbiont without replacing a lost function (NCRs, ColA, perhaps BCRs).

Connections

It is the review that McCutcheon et al. (LIT-1571, read in NOTE-1213) point to for import. The two agree that the smallest symbionts lack envelope genes. This review sharpens the import evidence that McCutcheon et al. cite. The aphid protein is RlpA4 in Buchnera, and the mealybug protein is MurF in Moranella. Both symbionts are above 0.5 Mbp, well above McCutcheon’s 200 kb tipping point. The Paulinella primary studies (LIT-1557, LIT-1621) and the nitroplast paper (LIT-1562) are cited here and are unread in this record. The mealybug HGT genes are LIT-1569’s subject.

Bearing on the record

• QUESTION-277, the gene-transfer-and-import route. The question calls the route “Genes move to the host nucleus and their products are imported back”. This reading corrects that. Import is the route. Transfer from the endosymbiont is optional and was minor in Paulinella and absent in the nitroplast. The replacing genes are mostly the host’s own, or come from third-party bacteria. So the route moves C (control by import) and G (functions now on the host genome), and the G move does not need the symbiont’s genes to travel. The C move includes control of the symbiont’s DNA replication, expression and division. The inference that this is one route with two variants is mine.
• QUESTION-277, the order of dimensions. The authors’ scenario puts C first, as effectors, and D after, when a redundant imported protein lets the symbiont copy decay. Their rhizobia argument says genetic integration is “always linked” to vertical transmission (R). So on their account R is a condition for the C-then-D sequence to end in an organelle, and C without R (rhizobia) stops at manipulation. This is an order to test, not one shown. The mapping to the question’s letters is mine.
• QUESTION-277, directional dependence. B. bigelowii can lose UCYN-A when given fixed nitrogen. The organelle-level integration of the symbiont therefore coexists with a host whose dependence is conditional on its environment. D is asymmetric even here. The inference is mine.
• CASE-174 (Paulinella). The case’s figures are from its unread primary sources. This review gives a later age (90–140 million years, against the case’s “about 60”), about 60 EGT and 170 HGT genes (against “more than thirty”), and names two imported photosystem I subunits, PsaE and PsaK (the case says three). It says EGT played “only a circumstantial role”. The case title’s “genes moved to the host nucleus and their proteins imported back” overstates transfer.
• CASE-167 (the nitroplast). Confirmed: coordinated division, vertical transmission, more than 350 imported proteins. Added: no EGT, a C-terminal targeting signal on 62% of imported proteins, and a host that can lose the symbiont in culture.
• CASE-171 and CASE-173. The mealybug protein is MurF, in Moranella, not Tremblaya. The aphid protein is RlpA4, from a horizontally transferred gene, and proteomics found no host protein reliably enriched in Buchnera. Both cases’ notes from NOTE-1213 should say so.
• THEORY-1049. Consistent with its envelope claim. But the single imported proteins it cites as evidence of host supply are in symbionts above both its thresholds. The best case near 200 kb is Nardonella, whose import is unconfirmed. So import into a tipping-point endosymbiont is still not shown.
• Filed as THEORY-1100 and THEORY-1107. No instruction for machine-learning practice.

Limitations

• A review by one group, one of whose authors holds a fellowship named for chromatophores (Acknowledgements). The two systems with import at scale are its strongest evidence, and the others are cited from primary studies not read here.
• Single-protein evidence is antibody-based. Cross-reactivity is a general risk. The review reads ColA antibodies binding Nardonella as a sign of ColA’s wide reach; that it could also be cross-reaction is my reading, not the paper’s. No single-protein case has a mechanism.
• “More common than generally realized” is a belief, stated as such, from partial proteomes. No system has been shown to lack import.
• The scenario is a narrative. No system has been followed through its steps, and the order of effector targeting and dependence is not observed anywhere.
• Figures were not seen. Figure 2A’s gene matrix and Figure 3’s membrane counts were read only from captions and text.
• The Paulinella gene count is printed “878–860”, and the genome fraction lost is “around 2/3”, where the record’s case says 26% kept. The two use different reference genomes, presumably. The review does not say which.

Open questions

• Is any protein imported into a symbiont under 200 kb? Tremblaya and Nardonella are the candidates.
• How does a protein cross the bacterial inner membrane in any of these systems? Only signals are known.
• Do effectors in fact come first? A system caught with effectors but no replaced function, then followed, would show it. Angomonas, with genetic tools and seven ETPs, is the authors’ suggested model.
• Can a metabolically integrated symbiont be grown outside its host with supplements? That is the authors’ own criterion, untested.
• Does genetic integration ever arise without vertical transmission? The rhizobia contrast says no, but it is one contrast.