Evolutionarily recent dual obligatory symbiosis among adelgids indicates a transition between fungus- and insect-associated lifestyles

Profftia and Vallotia are related to free-living bacteria and fungus-associated endosymbionts

Previous 16S rRNA-based phylogenetic analyses suggested an affiliation of Profftia with free-living gammaproteobacteria and a close phylogenetic relationship between Vallotia and betaproteobacterial endosymbionts of Rhizopus fungi [14]. Biased nucleotide composition and accelerated sequence evolution of endosymbiont genomes [2, 3] often result in inconsistent phylogenies and may cause grouping of unrelated taxa [55, 56]. Thus, to further investigate the phylogenetic relationships of the A. laricis/tardus symbionts, we used conserved marker genes for maximum likelihood and Bayesian phylogenetic analyses.

Phylogenetic analysis of 45 single-copy proteins demonstrated that Profftia opens up a novel insect symbiont lineage most similar to Hafnia species and an isolate from the human gastrointestinal tract within the Hafniaceae, which has been recently designated as a distinct family within the Enterobacteriales [57] (Fig. S2). Hafnia strains are frequently identified in the gastrointestinal tract of humans and animals and were also found in insects [58, 59]. The phylogenomic placement of Profftia in our analysis is in agreement with previous 16S rRNA-based analyses [14].

Vallotia formed a monophyletic group with Mycetohabitans endofungorum and M. rhizoxinica, endosymbionts of Rhizopus fungi within the Burkholderiaceae [60, 61] with strong support in phylogenetic analyses based on a concatenated set of 108 proteins (Figs. 1 and S3; previous taxonomic assignments of the fungus-associated symbionts were as Burkholderia/Paraburkholderia endofungorum and rhizoxinica, respectively). Interestingly, Vallotia and M. endofungorum appeared as well-supported sister taxa within this clade. This implies a closer phylogenetic relationship between Vallotia and M. endofungorum and a common origin of adelgid endosymbionts from within a clade of fungus-associated bacterial symbionts. Lengths of branches leading to the fungus-associated endosymbionts were similar to those of free-living bacteria in the data set; however, Vallotia had a remarkably longer branch marking a rapid rate of sequence evolution characteristic of obligate intracellular bacteria [2, 3]. M. endofungorum and M. rhizoxinica have been identified in the cytosol of the zygomycete Rhizopus microsporus, best known as the causative agent of rice seedling blight [61, 62]. The necrotrophic fungus secretes potent toxins, rhizoxin and rhizonin, which are produced by the endosymbionts. The bacterial partners are obligatory for their host as they tightly control its sporulation, while they benefit from host nutrients and spread with the fungal spores [63, 64]. Additionally, related bacterial strains have also been found in association with Rhizopus fungi worldwide in a diverse set of environments, including other plant species, soil, food, and even human tissues [65, 66].

Taken together, phylogenomic analyses support that Profftia and Vallotia open up novel insect symbionts lineages most closely related to free-living bacteria within the Hafniaceae and a clade of fungus-associated endosymbionts within the Burkholderiaceae, respectively. Given the well-supported phylogenetic positioning of “Candidatus Vallotia tarda” nested within a clade formed by Mycetohabitans species, we propose the transfer of “Candidatus Vallotia tarda” to the Mycetohabitans genus, as “Candidatus Mycetohabitans vallotii” (a detailed proposal for the re-classification is given in the Supplementary Material).

Vallotia and Profftia are evolutionary young symbionts of adelgids

The complete sequence of the Profftia chromosome had a length of 1,225,795 bp and a G + C content of 31.9% (Table 1). It encoded for 645 proteins, one copy of each rRNA, 35 transfer RNAs (tRNAs), and 10 non-coding RNAs (ncRNAs). It had tRNAs and amino acid charging potential for all 20 standard amino acids. However, protein-coding sequences (CDSs) made up only 52.4% of the genome, and 21 pseudogenes indicated an ongoing gene inactivation.

The Vallotia chromosome had a length of 1,123,864 bp. It had a G + C content and a coding density of 42.9 and 64.9%, respectively. However, a 72,431-bp-long contig showed a characteristically lower G + C content (36.1%) and contained only 46.2% putative CDSs. This contig had identical repeats at its ends, and genome annotation revealed neighboring genes for a plasmid replication initiation protein, and ParA/ParB partitioning proteins, which function in plasmid and chromosome segregation between daughter cells before cell division [67]. We thus assume that this contig corresponds to a circular plasmid of Vallotia. Vallotia has three rRNA operons, similarly to its close relative, M. rhizoxinica [68]. In total, the Vallotia genome encoded 780 proteins (29 on the putative plasmid), 41 tRNAs, and 52 predicted pseudogenes (5 on the putative plasmid).

The host-restricted lifestyle has a profound influence on bacterial genomes. Relaxed purifying selection on many redundant functions and increased genetic drift can lead to the accumulation of slightly deleterious mutations and the proliferation of mobile genetic elements [69,70,71,72]. Disruption of DNA repair genes can increase mutation rates, which promote gene inactivation [73]. Non-functional genomic regions get subsequently lost, and ancient obligate endosymbionts typically have tiny (≪0.8 Mb), gene-dense genomes with AT-biased nucleotide composition [2, 74, 75]. Facultative symbionts also possess accelerated rates of sequence evolution but have larger genomes (>2 Mb) with variable coding densities following the age of their host-restricted lifestyle [76]. The number of pseudogenes in Vallotia and Profftia is higher than in ancient intracellular symbionts, which suggests an intermediate state of genomic reduction [2]. The only moderately reduced size and AT bias together with the low protein-coding density of the Vallotia and Profftia genomes was most similar to those of evolutionary young co-obligate partners of insects [76], for instance, “Ca. Pseudomonas adelgestsugas” in A. tsugae [23], Serratia symbiotica in Cinara cedri [77, 78], and the Sodalis-like symbiont of Philaenus spumarius, the meadow spittlebug [79].

The evolutionary link between Vallotia and fungus-associated endosymbionts

High level of genomic synteny between Vallotia and M. rhizoxinica

Intracellular symbionts usually show a low level of genomic similarity to related bacteria. Rare examples of newly emerged bacteriocyte-associated symbionts of herbivorous insects pinpoint their source from plant-associated bacteria [4], gut bacteria [5], and other free-living bacteria [6].

Genome alignments showed a low level of collinearity between the genomes of Profftia and its closest relatives. Among the relatives of Vallotia, a closed genome is available for M. rhizoxinica [68]. We therefore mostly focused on this fungus-associated symbiont as a reference for comparison with Vallotia.

The Vallotia chromosome showed a surprisingly high level of synteny with the chromosome of M. rhizoxinica (Fig. 2A). However, its size was only ~40% of the fungus-associated symbiont chromosome. The putative plasmid of Vallotia was perfectly syntenic with the larger of the two plasmids of M. rhizoxinica (pBRH01), although the Vallotia plasmid was >90% smaller in size (72,431 bp versus 822,304 bp) [68]. Thus, the Vallotia plasmid showed a much higher level of reduction than the chromosome, which together with its lower G + C content and gene density suggests differential evolutionary constraints on these replicons.

The conservation of genome structure contrasts with the elevated number of transposases and inactive derivatives making up ~6% of the fungus-associated symbiont genome [68]. Transition to a host-restricted lifestyle is usually followed by a sharp proliferation of mobile genetic elements coupled with many genomic rearrangements [80,81,82]. However, mobile genetic elements get subsequently purged out of the genomes of strictly vertically transmitted symbionts via a mutational bias toward deletion and because of lack of opportunity for horizontal acquisition of novel genetic elements [71, 74]. Independent origins of the fungus and adelgid symbioses from free-living precursors would have likely resulted in extensive genome rearrangements due to the accumulation of mobile genetic elements, as seen, for instance, between different S. symbiotica strains in aphids [81]. In contrast to the fungus-associated symbiont, mobile elements are notably absent from the Vallotia genome, suggesting that they might have been lost early after the establishment of the adelgid symbiosis conserving high collinearity between the fungus- and adelgid-associated symbiont genomes. M. rhizoxinica is transmitted also horizontally among fungi and might have more exposure to foreign DNA, therefore at least part of the mobile elements could possibly be inserted into its genome after the host switch of the Vallotia precursor [61, 62].

The observed high level of genome synteny between Vallotia and M. rhizoxinica genomes is consistent with the phylogenetic position of Vallotia interleaved within the clade of Rhizopus endosymbionts. This points toward a direct evolutionary link between these symbioses and a symbiont transition between the fungus and insect hosts.

Shrinkage of the insect symbiont genome

Deletion of large genomic fragments—spanning many functionally unrelated genes—represents an important driving force of genome erosion especially at early stages of symbioses when selection on many functions is weak [3, 83]. Besides, gene loss also occurs individually and is ongoing, albeit at a much lower rate, even in ancient symbionts [75, 84, 85]. Both small and large deletions could be seen when comparing the Vallotia and M. rhizoxinica genomes. Several small deletions as small as one gene were observed sparsely in the entire length of the Vallotia genome within otherwise collinear regions. The largest genomic region missing from Vallotia encompassed 165 kbp on the M. rhizoxinica chromosome (Fig. 2B). The corresponding intergenic spacer was only 3843-bp long on the Vallotia genome between a phage shock protein and the Mfd transcription-repair-coupling factor, present both in Vallotia and M. rhizoxinica. Interestingly, this large genomic fragment included the large rhizoxin biosynthesis gene cluster (rhiIGBCDHEF), which is responsible for the production of rhizoxin, a potent antimitotic macrolide serving as a virulence factor for R. microsporus, the host of M. rhizoxinica [86]. A homologous gene cluster was also found in Pseudomonas fluorescens, and it has been suggested that it has been horizontally acquired by M. rhizoxinica [68, 86]. The rhi cluster is also present in M. endofungorum, therefore it was most likely already present in the genome of the common ancestor of the fungus- and adelgid-associated symbionts and got subsequently lost in Vallotia. Rhizoxin blocks microtubule formation in various types of eukaryotic cells [86, 87], thus the loss of this gene cluster in ancestral Vallotia could have contributed to the establishment of the adelgid symbiosis. However, this large deleted genomic region also contained several transposases and many other genes, such as argE and ilvA, coding for the final enzymes for ornithine and 2-oxobutanoate productions, which were located adjacent to each other at the beginning of this fragment. The largest deletion between the plasmids encompassed nearly 137 kbp of the megaplasmid of M. rhizoxinica and involved several non-ribosomal peptide synthetases (NRPS), insecticidal toxin complex (Tc) proteins, and a high number of transposases among others. M. rhizoxinica harbors 15 NRPS gene clusters [68] in total, all of which are absent in Vallotia. NRPSs are large multienzyme machineries that assemble various peptides, which might function as antibiotics, signal molecules, or virulence factors [88]. Insecticidal toxin complexes are bacterial protein toxins, which exhibit powerful insecticidal activity [89]. Two of such proteins are also present in the large deleted chromosomal region in close proximity to the rhizoxin biosynthesis gene cluster (Fig. 2B); however, their role in M. rhizoxinica remains elusive.

The Vallotia genome encodes a subset of functions of the fungus-associated endosymbionts

The number of protein-coding genes of Vallotia is less than one-third of those of M. rhizoxinica and M. endofungorum, although metabolic functions are already reduced in the fungus-associated endosymbionts compared to free-living Burkholderia species [68] (Figs. S4 and S5). When compared to the two genomes of the fungus-associated endosymbionts, only 53 proteins were specific to Vallotia (Fig. S6). All of these were short (on average 68 amino acid long) hypothetical proteins and most of them showed no significant similarity to other proteins in public databases. Whether these Vallotia-specific hypothetical proteins might be over-annotated/non-functional open reading frames or orphan genes with a yet unknown function [90, 91] needs further investigation. Four genes were present in Vallotia and M. rhizoxinica but were missing in M. endofungorum. These encoded for BioA and BioD in biotin biosynthesis, NagZ in cell wall recycling, and an MFS transporter. Fifteen genes, including, for instance, the MreB rod-shape-determining protein, glycosyltransferase and hit family proteins, genes in lipopolysaccharide, lipoate synthesis, and the oxidative pentose phosphate pathway, were shared between Vallotia and M. endofungorum only. The rest of the Vallotia genes, coding for 91% of all of its proteins, were shared among the fungus- and insect-associated endosymbionts.

Comparing the genes present in both endosymbionts to those shared only by the fungus-associated endosymbionts (Fig. S7), we can infer selective functions maintained or lost during transition to insect endosymbiosis. Translation-related functions have been retained in the greatest measure in the group shared by all endosymbionts. Functions, where higher proportion of genes were specific to the fungus endosymbioses, were related to transcription, inorganic ion transport and metabolism, secondary metabolite biosynthesis, signal transduction, intracellular trafficking, secretion, vesicular transport, and defense mechanisms. Most of the proteins specific to either of the fungus-associated symbionts were homologous to transposases and integrases, transcriptional regulators, or had an unknown function.

Fungus-associated endosymbionts encode a high number of transcriptional regulators (~5% of all genes in M. rhizoxinica) [68], but Vallotia has retained only a handful of such genes, which is a feature similar to other insect symbionts and might facilitate the overproduction of essential amino acids [75, 92].

M. rhizoxinica is resistant against various β-lactams and has an arsenal of efflux pumps that might provide defense against antibacterial fungal molecules, the latter might also excrete virulence factors to the fungus cytosol (type I secretion) [68]. Besides, M. rhizoxinica encodes several genes for pilus formation; adhesion proteins; and type II, type III, and type IV secretion systems, which likely play a central role in host infection and manipulation in the bacteria–fungus symbiosis [68, 93, 94]. However, all of the corresponding genes are missing in Vallotia. Thus, neither of these mechanisms likely play a role in the adelgid symbiosis. Indeed, we could not even detect remnants of these genes in the Vallotia genome, except for a type II secretion system protein as a pseudogene. Loss of these functions is consistent with a strictly vertical transmission of Vallotia between host generations. Transovarial transmission likely does not require active infection mechanisms, and the endosymbionts rather move between the insect cells in a passive manner via an endocytic/exocytic vesicular route [12, 95]. In contrast, M. rhizoxinca is also able to spread horizontally among fungi and re-infect cured Rhizopus strains under laboratory conditions [61, 62].

Differential reduction of metabolic pathways in Vallotia and Profftia

Although compared to their closest free-living relatives both Vallotia and Profftia have lost many genes in all functional categories, both retained the highest number of genes in translation-related functions (Fig. S4). Besides, functions related to cell division, nucleotide and coenzyme transport and metabolism, DNA replication and repair, posttranslational modification, and cell envelope biogenesis are reduced to a lesser extent in both endosymbionts. As a consequence, most of the genes of Vallotia and Profftia are devoted to translation and cell envelope biogenesis, which make up higher proportions of their genomes than in related bacteria (Fig. S5). Retention of a minimal set of genes involved in central cellular functions such as translation, transcription, and replication is a typical feature of reduced genomes, even extremely tiny ones of long-term symbionts [75]. However, ancient intracellular symbionts usually miss a substantial number of genes for the production of the cell envelope and might rely on host-derived membrane compounds [96,97,98].

Based on pathway reconstructions, both Vallotia (Fig. S8) and Profftia (Fig. S9) have a complete gene set for peptidoglycan, fatty acid, and phospholipid biosynthesis and retained most of the genes for the production of lipid A, LPS core, and the Lpt LPS transport machinery. Besides, we found a partial set of genes for O antigen biosynthesis in the Vallotia genome. Regarding the membrane protein transport and assembly, both adelgid endosymbionts have the necessary genes for Sec and signal recognition particle translocation and the BAM outer membrane protein assembly complex. Profftia also has a complete Lol lipoprotein trafficking machinery (lolABCDE), which can deliver newly matured lipoproteins from the inner membrane to the outer membrane [99]. In addition, Profftia has a near-complete gene set for the Tol-Pal system; however, tolA has been pseudogenized suggesting an ongoing reduction of this complex. Further, both adelgid endosymbionts have retained mrdAB and mreBCD having a role in the maintenance of cell wall integrity and morphology [100, 101]. The observed well-preserved cellular functions for cell envelope biogenesis and integrity are consistent with the rod-shaped cell morphology of Profftia and Vallotia [14], contrasting the spherical/pleomorphic cell shape of ancient endosymbionts, such as Annandia in A. tsugae and Pineus species [10, 11, 15].

Regarding the central metabolism, Vallotia lacks 6-phosphofructokinase but has a complete gene set for gluconeogenesis and the tricarboxylic acid (TCA) cycle. TCA cycle genes are typically lost in long-term symbionts but are present in facultative and evolutionarily recent obligate endosymbionts [79, 82, 102]. Interestingly, Vallotia does not have a recognized sugar transporter. Similarly to M. rhizoxinica [68], a glycerol kinase gene next to a putative glycerol uptake facilitator protein is present on its plasmid. However, the latter gene has a frameshift mutation and a premature stop codon in the first 40% of the sequence and whether it can still produce a functional protein remains unknown.

Profftia can convert acetyl-CoA to acetate for energy but lacks TCA cycle genes, a feature characteristic to more reduced genomes, such as, for instance, Annandia in A. tsugae [23]. Profftia has import systems for a variety of organic compounds, such as murein tripeptides, phospholipids, thiamine, spermidine and putrescine, 3-phenylpropionate, and a complete phosphotransferase system for the uptake of sugars.

NADH dehydrogenase, ATP synthase, and cytochrome oxidases (bo/bd-1) are encoded on both adelgid symbiont genomes. However, Vallotia is not able to produce ubiquinone and six pseudogenes in its genome indicate a recent inactivation of this pathway (Fig. S10).

Profftia retained more functions in inorganic ion transport and metabolism, while Vallotia had a characteristically higher number of genes related to amino acid biosynthesis (see its function below) and nucleotide transport and metabolism (Fig. S4). For instance, Profftia can take up sulfate and use it for assimilatory sulfate reduction and cysteine production, and it has also retained many genes for heme biosynthesis (Fig. S9). However, it cannot produce inosine-5-phosphate and uridine 5’-monophosphate precursors for the de novo synthesis of purine and pyrimidine nucleotides and thus would need to import these compounds.

Notably, although core genes in DNA replication and repair [70] are well preserved, multiple pseudogenes may indicate an ongoing erosion of DNA repair functions in the genomes. These include genes for the UvrABC nucleotide excision repair complex in both adelgid symbionts, helicases (recG, recQ), mismatch repair genes (mutL, mutS; the MutHLS complex is also missing in Profftia), and alkA encoding a DNA glycosylase in Vallotia.

Taken together, their moderately reduced, gene-sparse genomes but still versatile metabolic capabilities support that Vallotia and Profftia are evolutionarily recently acquired endosymbionts. This is following their occurrence in lineages of adelgids, which likely diversified relatively recently, ~60 and ~47 million years ago, respectively, from the remaining clades of Adelgidae [8].

Vallotia and Profftia are both obligatory nutritional symbionts

Complementary functions in essential amino acid provision

Vallotia and Profftia complement each other’s role in the essential amino acid synthesis, thus have a co-obligatory status in the A. laricis/A. tardus symbiosis (Fig. 3). Although Vallotia likely generates most essential amino acids, solely Profftia can produce chorismate, a key precursor for the synthesis of phenylalanine and tryptophan. Profftia is likely responsible for the complete biosynthesis of phenylalanine as it has a full set of genes for this pathway. It can also convert chorismate to anthranilate; however, further genes for tryptophan biosynthesis are only present in the Vallotia genome. Thus, Vallotia likely takes up anthranilate for tryptophan biosynthesis. Anthranilate synthase (trpEG), is subject to negative feedback regulation by tryptophan [103], thus partition of this rate-limiting step between the co-symbionts can enhance overproduction of the amino acid and might stabilize dual symbiotic partnerships at an early stage of coexistence. The production of tryptophan is partitioned between Vallotia and Profftia similarly as seen in other insect symbioses [77, 78, 104], and it is also shared but is more redundant between the Annandia and Pseudomonas symbionts of A. tsugae [23]. The Vallotia–Profftia system generally shows a lower level of functional overlap between the symbionts and is more unbalanced than the Annandia–Pseudomonas association. In the latter, redundant genes are present also in the synthesis of phenylalanine, threonine, lysine, and arginine, and Annandia can produce seven and the Pseudomonas partner five essential amino acids with the contribution of host genes [23].

The Vallotia genome encodes for all the enzymes for the synthesis of five essential amino acids (histidine, leucine, valine, lysine, threonine). ArgG and tyrB among the essential amino acid synthesis-related genes are only present on the plasmid of Vallotia, which might be a reason that the plasmid is still part of its genome. However, neither of the endosymbionts can produce ornithine, 2-oxobutanoate, and homocysteine de novo, which are key for the biosynthesis of arginine, isoleucine, and methionine, respectively. The corresponding functions are also missing from the Annandia–Pseudomonas system [23]. These compounds are thus likely supplied by the insect host, as seen for instance in aphids, mealybugs, and psyllids, where the respective genes are present in the insect genomes and are typically overexpressed within the bacteriome [97, 105, 106]. The metC and argA genes are still present as pseudogenes in Vallotia suggesting a recent loss of these functions in methionine and arginine biosynthesis, respectively.

In most plant sap-feeding insects harboring a dual symbiotic system, typically the more ancient symbiont provides most of the essential amino acids [77, 107]. Given its prominent role in nutrient provision and its presence in both larch- and Douglas fir-associated adelgids, Vallotia might be the older symbiont. Loss of functions in chorismate and anthranilate biosynthesis might have led to the fixation of Profftia in the system.

Vallotia and Profftia have more redundant functions in non-essential amino acid production (Fig. 3). Only Profftia can produce cysteine and tyrosine, while none of the symbionts can build up glutamine, thus this latter amino acid is likely supplied by the insect bacteriocytes.

The presence of relevant transporters can complement missing functions in amino acid synthesis (Fig. 3). For instance, Profftia has a high-affinity glutamine ABC transporter and three symporters (BrnQ, Mtr, TdcC), which can import five among the essential amino acids that can be produced by Vallotia. Vallotia might excrete isoleucine, valine, and leucine via AzICD, a putative branched-chain amino acid efflux pump [108], and these amino acids could be taken up by Profftia via BrnQ and would be readily available also for the insect host.

B vitamin provision by Vallotia

Regarding the B vitamin synthesis, Vallotia is likely able to produce thiamine (B₁), riboflavin (B₂), pantothenate (B₅), pyridoxine (B₆), biotin (B₇), and folic acid (B₉) (Fig. S11). Although Vallotia misses some genes of the canonical pathways, alternative enzymes and host-derived compounds might bypass these reactions, as detailed in the Supplementary Material. Profftia has only a few genes related to B vitamin biosynthesis. Three pseudogenes (ribAEC) in the riboflavin synthesis pathway indicate that these functions might have been lost recently in this symbiont (Fig. S11).

Source: Ecology - nature.com