Contribution
It draws a line inside the class of reduced endosymbionts. The earlier picture (LIT-1559, unread here) was of degenerate genomes that “retain only the most essential functions”, with information processing the last to go. This paper shows that past about 200 kb, even translation, “the most tightly conserved process”, starts to erode. Past about 400 kb, the cell can no longer build or control its own boundary. It turns that observation into four concrete open problems about how the host must be supplying what the symbiont no longer encodes.
Key insight
Genome reduction comes in two thresholds, and past the second the symbiont cannot be self-sufficient in the most basic sense: “Tipping-point endosymbiont genomes encode no genes for making their own cell membranes or for producing the lipid molecules from which membranes are assembled or the enzymatic machinery that enables transport”. “But their membranes exist … and apparently work, so where do they come from?” The authors’ answer is the host, “just as in mitochondria, chloroplasts, and plastids”. If so, a bacterium that was never given a protein-import system is being built and supplied by its host in the ways that define an organelle. The paper calls host-directed protein import “a major milestone in the evolution of a bona fide organelle”.
Assumptions
- Genome content is a proxy for capacity. A gene absent from the genome is a function the symbiont lacks on its own. Gene presence was called from annotations, the literature and HMMER searches. Some genomes are drafts, so “the exact gene patterns should be considered tentative” (Figure 2 caption).
- The genome set is representative. Figure 2’s organisms were “manually” selected from GenBank: every genome under 1 Mb, but only the largest and smallest of each genus.
- The 200 kb cutoff is “somewhat arbitrary” (Figure 3 caption, on Zinderia just missing it).
Key results
-
Diverse origins. Bacteria with genomes under 1 Mb are “almost exclusively” symbionts. They come from 10 bacterial groups, and those under 200 kb from 5 (Alpha-, Beta- and Gammaproteobacteria, Bacteroidia, Verrucomicrobiota), whose common ancestor is near the origin of cellular life. The authors read the shared features as convergent.
-
First threshold, about 400 kb (Figure 2). Genes for fatty acids, phospholipids and cell wall, for transport across the envelope (BAM complex, sec translocon), and for envelope proteins (ATP synthase) are “generally lost in concert”. Below about 400 kb, endosymbionts “almost universally lose the autonomous ability to build and to control the transport of molecules across their cellular envelope”. Severe erosion can start at 0.6–0.7 Mb in candidate phyla radiation bacteria.
-
Second threshold, about 200 kb (Figure 2). Above it, genomes keep enough tRNAs to decode all codons, enough aaRSs to charge them, and about 50 ribosomal proteins. Below it, many of each are lost. One cicada endosymbiont lacks tRNA genes for eight amino acids it still encodes. Up to 20 ribosomal proteins can be missing.
-
Composition (Figure 3). A PCA of amino-acid frequencies across 98,966 proteomes:
- PC1, 85% of the variance, tracks genomic GC content, recovered from amino-acid frequencies alone;
- tipping-point endosymbionts lie “well off the main cloud”;
- extremophiles (except halophiles) do not.
The authors conclude that “mutational and population genetic forces can be more powerful in driving outlier proteome compositional biases than physical forces”. Zinderia’s proteome is almost 50% lysine, isoleucine and asparagine, against bacterial averages of 5%, 6% and 4%. The analysis is “preliminary”.
-
The four mysteries, each with the host as the candidate.
- Envelope. Many tipping-point endosymbionts show three bilayers: two inherited, one from the host’s engulfment. Building them may use membrane contact sites, as organelles do. Pathogen-style effector secretion cannot be the route, since the symbionts can no longer secrete effectors.
- Transport. Sap-feeding insects have evolved symbiosis-specific amino-acid transporters by gene duplication, and enrich the vacuolar membrane with them. How molecules cross the symbiont’s own two membranes “is unknown”.
- Translation.
- No bacterial tRNA gene transferred to a host nucleus has been reported as imported by an endosymbiont or organelle.
- No new transfer of aaRS genes has been found.
- The candidates are imported host tRNAs, imported aaRSs, or imported charged tRNAs.
- Ribosomes may be shrinking to a functional core, or acquiring host-encoded proteins as mitochondrial ribosomes did.
- Protein function. Endosymbiont RNAs and proteins are predicted to be less stable. Endosymbionts are more heat-sensitive than their hosts. High GroEL expression may buffer this.
-
Import already observed. Host-encoded proteins have been found by immunofluorescence in the cytoplasm of aphid and mealybug endosymbionts (references 55 and 61, not read here). These studies looked at “just one protein each”, and the mechanism is unknown. Large-scale import is established in pre-tipping-point endosymbionts of protists (references 59–60), which is the Paulinella case.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Below about 400 kb, endosymbionts almost always lose the genes to build their envelope and to transport across it, and these losses go together | moderate | the Figure 2 matrix over a hand-picked set; patterns “tentative” |
| C2 | Below about 200 kb, translation-related genes (tRNAs, aaRSs, ribosomal proteins) erode, where above it they are retained | moderate | Figure 2; the threshold’s precise value is “somewhat arbitrary” |
| C3 | Tipping-point endosymbionts have among the most compositionally biased proteomes in bacteria, beyond extremophiles | moderate | a PCA of about 99,000 proteomes; “preliminary” |
| C4 | Mutational and population-genetic forces can shape proteome composition more strongly than physical environment | weak to moderate | inferred from C3, given the stable intracellular environment |
| C5 | The membranes, transport and missing translation components of tipping-point endosymbionts are supplied by the host | weak | the organelle analogy, three-membrane micrographs, and single-protein import observations; the mechanisms are unknown |
| C6 | Host-encoded proteins are imported into insect endosymbionts | moderate for existence, weak for scale | two cited single-protein studies, and an expectation that “the total number … is likely to be much higher” |
Concepts
- Tipping-point endosymbiont: one with a genome under about 200 kb, where translation genes begin to be lost.
- Host-beneficial endosymbiosis: “an organism stably maintains an unrelated organism inside some or all of its cells”, to the host’s benefit.
- Three-membrane envelope: the two inherited membranes of a gram-negative ancestor, plus the host-derived vacuolar membrane.
- Membrane contact site: a protein junction between closely spaced membranes through which lipids pass, used by mitochondria and plastids.
Connections
It updates McCutcheon & Moran’s 2012 review (LIT-1559, unread). The mealybug symbiont Tremblaya is its Figure 4, which links it to LIT-1569 and LIT-1572, both unread. Its protein-import mystery is the subject of Sørensen et al. (LIT-1560, unread), and its Shigenobu-era picture of Buchnera (LIT-1596, unread) is a symbiont well above the tipping point.
Bearing on the record
- QUESTION-277, regulatory control (C) and gene distribution (G). The question placed host control with gene transfer and import, the Paulinella route (CASE-174). This reading shows control arising another way in insect symbioses. Below the thresholds, a symbiont that cannot build its membrane or charge its tRNAs is supplied by the host whether or not genes moved. The host’s control is then forced by the symbiont’s losses, not won by transfer. G and C are therefore not independent: losses in the symbiont’s genome create the host’s role. The inference is mine.
- CASE-173 (aphid–Buchnera). The case says the abstracts it rests on report no gene transfer to the host. This reading adds that a host-encoded protein has been found in Buchnera’s cytoplasm (its reference 55). It does not say the protein is a transferred Buchnera gene: the paper speaks of horizontally transferred bacterial genes on host genomes in general. A dated note goes on the case.
- CASE-171 (mealybugs). Tremblaya is a tipping-point endosymbiont with a three-membrane envelope (Figure 4), and a host-encoded protein has been seen in a mealybug endosymbiont. A dated note goes on the case.
- CASE-167 (nitroplast) and CASE-174 (Paulinella). These are the “pre-tipping-point” endosymbionts in protists where import is established “on a large scale”. The paper’s milestone language, import as the mark of “a bona fide organelle”, is the criterion those cases rest on.
- Filed as THEORY-1049. No instruction for machine-learning practice.
Limitations
- Thresholds from a curated sample. The 400 kb and 200 kb boundaries are read off a hand-picked matrix with draft genomes. No statistical test locates them.
- Composition analysis preliminary, by the authors’ word. PC2 is uninterpreted.
- Host supply is analogy. Every proposed mechanism (contact sites, imported tRNAs, imported aaRSs, host ribosomal proteins) is a hypothesis. The only observations are three-membrane micrographs and two single-protein import studies, neither read here.
- Gene absence is not function absence. The paper’s own point about promiscuous enzymes and host supply cuts both ways: a gene absent from the symbiont may be present elsewhere in the system.
Open questions
- Is the 200 kb boundary sharp, and what sets it? The paper locates it without a mechanism.
- How many host proteins are imported into tipping-point endosymbionts, and by what machinery? Proteomics of isolated symbionts, as done for Paulinella, would answer it.
- Do insect hosts import tRNAs into their symbionts?
- Does host supply precede the symbiont’s losses, allowing them, or follow them as rescue? This is the order question QUESTION-277 asks of host compensation.