Contribution
Before this letter, Buchnera’s role was known from nutritional and physiological studies, and its small genome from a physical map. This letter gives the whole sequence. It turns “the symbiont supplies amino acids” into a gene-by-gene account of what the symbiont can and cannot make, and of what it can no longer do for itself. In the authors’ words, it is “the first case where genomic evolution of a mutualistic organism is revealed at the genomic level”.
Key insight
Buchnera is reduced like a parasite but dependent in a different way. Parasites with small genomes lose the pathways for nutrients they take from the host. Buchnera keeps the pathways for nutrients it gives the host, and loses those the host can supply. So the gene set is complementary to the host’s: “Buchnera provides the host with what the host cannot synthesize, and conversely, the host provides the symbiont with what Buchnera cannot synthesize.” Beyond nutrients, it also seems to lack the means to make its own membrane lipids and to regulate its own genes. The authors conclude that it “is similar to organelles”.
Assumptions
- Gene content is capacity. A function is taken as absent when no gene for it was found by GeneHacker (an HMM gene finder) and BLAST against a non-redundant database. A function is taken as present when its genes are. Expression and enzyme activity are not measured.
- E. coli and H. influenzae as reference. The DnaA box upstream of gidA is assumed to be the replication origin “as E. coli and Haemophilus influenzae are closely related to Buchnera”.
- Which amino acids are essential to aphids is taken from earlier work (refs 3, 15), not tested here.
- The ancestor’s gene set is inferred cladistically. A gene is placed in the last common ancestor of Buchnera, E. coli and H. influenzae when an orthologue pair spans sister groups or a descendant and an outgroup. The authors note that the true ancestral set “may be larger”.
- One strain. All of it is Buchnera APS from one long-established parthenogenetic clone of Acyrthosiphon pisum, 2,000 aphids dissected.
Key results
- The genome.
- One circular chromosome of 640,681 bp, “the smallest of the completely sequenced genomes” at the time except Mycoplasma genitalium (580,070 bp). It is 1.10 times M. genitalium’s.
- Two plasmids: pLeu (7,786 bp, 7 ORFs, with leuABCD) and pTrp (at least two tandem repeats of trpEG; the assembly converged at 7,258 bp, two units).
- G+C content 26.3%. No insertion sequences, phage-related sequences or significant repeats. One copy of each rRNA, 32 tRNA genes.
- 583 ORFs, average 988 bp, covering 88% of the genome. My recomputation gives 90% of the chromosome alone, and 88% of the chromosome and both plasmids, so the 88% probably counts the plasmids. The inference is mine.
- Predicted proteins are basic: average pI 9.6 against 7.2 in E. coli and 7.3 in H. influenzae. Lysine usage is twice E. coli’s (data not shown).
- A subset of E. coli. 500 ORFs were assigned a function, 79 match hypothetical proteins of other bacteria, and 4 are unique to Buchnera. The closest matches are generally E. coli’s, and E. coli’s operon gene order is well conserved. Hence “the Buchnera genome is a subset of the E. coli genome”.
- Amino acids (Fig. 2).
- 54 genes for amino-acid biosynthesis in the text; Table 1 lists 55, and its subcategories sum to 55. The difference is not explained.
- Genes for the amino acids essential to the aphid are present. Those for the non-essential ones “are almost completely missing”. Cysteine is an exception: the sulphate-reduction pathway and cysteine synthesis are complete, and “insects cannot reduce sulphate to sulphide”.
- Gaps inside retained pathways: ilvE, the usual last step for valine, leucine and isoleucine, is absent, “so the reaction may take place by some other aminotransferase”; dapC “has never been identified in any organisms”; HisC may substitute for TyrB in phenylalanine synthesis.
- Glutamate and aspartate, which Buchnera cannot make, are precursors of essential amino acids. So the pathways “are not only complementary, but also mutually dependent”. This fits earlier evidence that aphids recycle amino groups as glutamine, which Buchnera uses.
- Pantothenate is made from pyruvate, but the pathway on to coenzyme A is missing. Animals can make CoA from pantothenate but not pantothenate itself.
- Nucleotide pathways are “almost complete”, unlike in obligate parasites. Whether for host or self “is not known”.
- Transport. Few transporter genes (18 transport/binding in Table 1), and only a few ABC transporters. PTSs for glucose and mannitol; no other substrate-specific transporter found. The sec secretion genes are kept. Flagellar basal-body genes are present but fliC (filament) and chemotaxis genes are not, and no flagellum or motility has been seen. The authors suggest the basal body “serves as a transporter structure”, by analogy with Salmonella and Yersinia. This is a suggestion, not a finding.
- Energy. Glycolysis, the pentose phosphate pathway and aerobic respiration (nuo and cyo operons, F0F1 ATP synthase) are complete. The TCA cycle is not, apart from the 2-oxoglutarate dehydrogenase genes. No ubiquinone synthesis, no fermentation, no anaerobic respiration.
- DNA repair. recA is missing, though recBCD is present: “the first organism found to have recBCD without recA”. uvrABC is missing; mfd is kept. lexA and umuCD are missing, and so are methylation and restriction genes. The authors read the genome as “vulnerable to DNA damage”, or as using the residual genes differently.
- Cell surface.
- LPS genes are missing except lpcA and kdtB. Only a few lipoprotein and outer-membrane-protein genes.
- Phospholipid synthesis genes are “completely missing … except that for cardiolipin synthetase (cls)”. The authors offer two possibilities: Buchnera “either imports phospholipid from the host or synthesizes it, employing relevant enzymes transferred from the host cell, like mitochondria do”. Neither is tested.
- Table 1 still lists 6 fatty-acid biosynthesis genes and 15 murein and peptidoglycan genes.
- Regulation.
- No two-component systems. No transcriptional regulators except dnaA. No leader sequences or attenuation. No cyaA or crp, despite PTSs. Two sigma factors only, rpoD and rpoH. CsrA might regulate carbohydrate metabolism after transcription.
- “Buchnera is unique in lacking regulatory genes, but having their regulatees.” M. genitalium and Rickettsia prowazekii lost both. The authors suggest the loss follows from “the homeostatic environment in which Buchnera has been housed for so long”.
- Reductive origin. Over orthologue groups from 23 complete genomes, the most frequent subtree places Buchnera as diverging from the lineage to E. coli and H. influenzae after R. prowazekii, “although a certain number of trees support” Buchnera as E. coli’s closest relative. The Buchnera gene set, “excluding a few genes”, is a small subset of the inferred ancestral set, and no Buchnera-specific duplicated gene was found. Hence the small genome is “the result of reductive evolution”, not a small ancestor.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | The Buchnera APS chromosome is 640,681 bp with 583 ORFs, a subset of the E. coli gene set | strong | whole-genome shotgun at about 7-fold coverage, gaps closed by primer walking, error probability under 0.01% by CONSED; restriction fragments match the physical map |
| C2 | Buchnera keeps the biosynthetic genes for amino acids essential to aphids and lacks almost all of those for non-essential ones, cysteine excepted | strong as gene content | annotation (Fig. 2, Table 1); the essential/non-essential split is taken from the literature |
| C3 | Host and symbiont amino-acid pathways are mutually dependent, since glutamate and aspartate from the host are precursors of essential amino acids | moderate | gene content plus earlier physiological evidence on glutamine recycling; no flux measured here |
| C4 | Buchnera lacks phospholipid and LPS synthesis, nearly all transcriptional regulators, and recA, uvrABC and the SOS genes | strong as gene content | annotation; absence by failure to find a homologue |
| C5 | The small genome arose by loss from a larger common ancestor with E. coli and H. influenzae | moderate | cladistic inference of the ancestral gene set over 23 genomes; tree topology not unanimous |
| C6 | Buchnera “cannot survive outside the eukaryotic cell” and depends on the host for its membrane lipids | weak as shown here | inferred from missing genes; no culture attempt or lipid analysis reported |
| C7 | The mutualism “is so obligate that neither can reproduce independently” | not tested here | asserted in the opening sentence with a citation (ref. 2) |
| C8 | Loss of regulators follows from Buchnera’s long life in a homeostatic environment | weak | a suggestion; no test |
Concepts
- Bacteriocyte: one of the 60–80 “huge cells” in almost all aphid species in which Buchnera lives.
- Complementarity and syntrophy: the authors’ terms for host and symbiont each making what the other cannot, with exchange of products.
- Mutually dependent pathways: pathways in which one partner’s product is the other’s precursor, as with glutamate and aspartate feeding Buchnera’s essential-amino-acid synthesis.
- Reductive evolution: here, a genome that is a subset of a larger inferred ancestor’s, with no lineage-specific duplications.
Connections
The letter rests on earlier work it cites without detail: Buchnera as a new genus (ref. 1), its polyploidy, “more than 100 genomic copies per cell” (ref. 4), the physical map giving a genome a seventh of E. coli’s (ref. 5), and a phylogenetic dating of the symbiosis to 200–250 Myr ago with co-speciation of hosts and symbionts (ref. 7). It mentions AT bias as a feature of endocellular bacteria (ref. 11) but does not explain it. It does not mention drift, Muller’s ratchet, population size or transmission bottlenecks. Its only causal suggestions for loss are the “homeostatic environment” and a “prolonged intracellular life, sheltered from attack”. The drift account in the record comes from LIT-1573, Moran 1996, which is not cited here. Pál et al. (LIT-1585, not reached) is registered as simulating gene loss on a metabolic network to predict this genome’s gene content.
Bearing on the record
- CASE-173. The case’s first paragraph is drawn from this paper’s
abstract and mostly stands. Four corrections:
- “So obligate that neither can reproduce independently” is the opening sentence, with a citation. This paper does not show it.
- The paper says nothing about drift or the ratchet. The case’s “dependence through degradation … largely by drift” rests on LIT-1573 alone, which is Unfetched.
- “Lacks those for the ones it can” needs “almost all”: cysteine is made.
- The paper does not look at the host genome, so it could not report gene transfer from symbiont to host, and does not. It does raise the reverse: host enzymes, or host phospholipid, supplied to the symbiont, “like mitochondria do”. That is a hypothesis.
- THEORY-1049. Buchnera APS, at 641 kb, already lacks phospholipid synthesis (except cls) and almost all LPS synthesis, and the authors already proposed in 2000 that the host supplies membrane lipids. It keeps fatty-acid, murein and sec genes. So in this genome envelope erosion has begun well above the 400 kb threshold, but the full set is not gone. That fits a threshold for complete loss rather than for the start of loss. The inference is mine.
- QUESTION-277. On the five dimensions:
- G is the paper’s subject: essential amino-acid synthesis is split between the genomes, and interlocked by shared precursors.
- D is shown for the symbiont only, and only from gene content. The host’s dependence is asserted from the literature. D’s directionality is therefore not settled by this paper.
- C: the symbiont has lost almost all its own regulators but kept the genes they regulated. Nothing here shows the host regulating them. Loss of self-regulation is not host control. The inference is mine.
- R: maternal transmission is stated, citing earlier work.
- S: not addressed.
- The route it documents is reduction from a larger ancestor. The paper names no mechanism for it beyond relaxed demands of a homeostatic niche.
- THEORY-1051. The paper gives no data on Buchnera’s births and deaths inside the host, the variable that model makes decisive. It reports polyploidy (over 100 genome copies per cell), from earlier work.
- Filed as THEORY-1103. No instruction for machine-learning practice.
Limitations
- Annotation only. Every functional claim is from gene presence or absence. Absent genes could be divergent homologues the searches missed; present genes could be inactive.
- One strain, one host clone. Nothing here says how general the pattern is across Buchnera of other aphids.
- The host side is not studied. The host genome, host transporters and host contributions are outside the paper. “Viable only in … the bacteriocyte” is an inference from gene content, not an attempt at culture.
- Figures in extraction. Figures 1–3 were read from their captions and the text here. Which pathway steps are pink in Figures 2 and 3 could not be checked.
- Small inconsistencies. 54 amino-acid genes in the text against 55 in Table 1. “Covering 88% of the whole genome” matches the chromosome and plasmids, not the chromosome alone. “Chaemotaxis” in Table 1.
Open questions
- Does the host supply phospholipids to Buchnera, and by what route? Lipid analysis of isolated Buchnera membranes would show it.
- What, if anything, regulates Buchnera’s retained biosynthetic genes once its own regulators are gone? Measuring their expression in response to the host’s diet would bear on control (C).
- Is the flagellar basal body a transporter, as suggested?
- How did Buchnera lose so much? The paper’s homeostatic-environment suggestion and the drift account of LIT-1573 make different predictions about which genes go first. The paper does not test either.