by Christoph Weigel
The trouble with going vegan As we know from our herbivorous mammalian relatives, the ruminants, you need to host certain microbes in your guts to subsist on plant polysaccharides other than starch. Think salad or grass. Such 'non-starch' polysaccharides – cellulose, pectin, and a few others – are commonly known as 'fibers'. Fibers are, for the most part, indigestible by vertebrates, but are readily metabolized by many bacteria and fungi. Facing similar digestibility issues with 'non-starch' polysaccharides – except for cellulose and their own chitin 'hardware' – herbivorous insects likewise harbor a bunch of competent microbes in their midguts to handle that matter. Take, for example, the 'walking stick', Ramulus artemis, which, like other members of its family, prefers to snack on roses or other members of the huge Rosaceae family (Fig. 1). Studying their midguts should be revealing...
Being vegan The stick and leaf insects, the Phasmatodea, are not only herbivores, but, in fact, are exclusively leaf-feeding, 'obligate folivores' in scientese. This prompted Matan Shelomi and coworkers to study in more detail the plant cell wall degrading enzymes (PCWDEs) of six selected species, including the just mentioned Ramulus artemis. They dissected individual animals and prepared tissue samples of the anatomically distinguishable anterior and posterior midguts. Extracts of the samples were (classically) analyzed for enzyme activities (Fig. 2), but mostly spent on assembling their transcriptomes, i.e., sequencing all transcripts present in the sample at detectable amounts. This, the authors argue, is a sensible approach to assess the metabolic properties of species for which the genome sequences are not available. Only genes that are actively expressed are detected this way, but whether their mRNA is actually translated into protein remains an open question. Also, discrimination between host and symbiont mRNAs is hard – a clear limitation of this approach when studying PCWDE activities in insect guts. By 'database mining' (basically protein homology and motif searches in data bases, but more refined ) Shelomi et al. identified in their transcriptomes more than twenty PCWDE gene families, representing various isoforms of cellulases, glucanases and pectinases. Most of these PCWDE genes were differentially expressed, i.e., at ~10x higher rates in the anterior midgut as compared to the posterior. They take this as a first molecular hint for one specific function of the anterior midgut: breakdown of large polysaccharide molecules. That these genes were also found in genomic DNA from brain tissue of the sample animals, a tissue that should be essentially symbiont-free, the authors suggest that they are true phasmatodean genes, not symbiont genes. While most of the PCWDE genes showed homology to those of other insects, the pectinases were strikingly homologous to bacterial genes.
Pectinases The enzymatic breakdown of pectin involves a number of enzymes collectively called pectinases (Fig. 3), one of which, the polygalacturonase (PG), comes in different flavors. Although these GH28 family polygalacturonases are structurally and phylogenetically related, they each have either endo-polygalacturonase or exo-polygalacturonase activity, the latter often capable to breaking down their substrate to the monomers. They also differ in kinetic parameters (temperature optimum, pH dependency) and with respect to rhamnogalacturonase activity.
In a subsequent study, Shelomi et al. characterized the GH28 family PG transcripts of the Phasmatodea in more detail, and obtained strong evidence that they were not derived from bacterial symbionts. A majority of the PG transcripts had 3' poly-A tails – a hallmark of eukaryotic mRNAs that is more sparsely found in bacteria – and full-length transcripts encoded signal peptides typical of eukaryotes. Each PG-encoding contig also had multiple matching genomic reads from (symbiont-free) brain tissue that uniquely aligned to them ('contig' and 'reads' is sequenese and refers to partially assembled sequences and singleton sequences, respectively). When cloned in the baculovirus system and expressed in Sf9 insect cells, roughly 90% of 50 tested PG transcripts gave enzymatically active proteins, which indicates that these represented mRNAs of active genes. The expressed proteins allowed the authors to determine individually their exact enzymatic specificity (with assays similar to those shown in Fig. 2 ), and indeed, endo- and exo-polygalacturonase activities were found, both in good correlation to the similarities of the respective sequences.
Pectinase Phylogeny Homology searches for GH28 PG genes in the close phylogenetic 'neighborhood' of the studied species, that is among the Euphasmatodea branch of the Phasmatodea, were unsuccessful. No evidence for such genes was detected in the single sequenced species of the genus Timema, belonging to the Timematodea sister branch of the Euphasmatodea, or among other members of the larger Polyneoptera family (Fig. 4). These findings led the authors to conclude that, unlike the cellulase genes, the GH28 PG genes do not have a deep evolutionary history within the insects, which made a search for their source even more intriguing.
One of the perks of the sequencing craze in recent years is the abundance of protein sequences that now allows one to derive rather robust phylogenies due to now having sufficiently large sample sizes for statistical evaluation. In their first study, Shelomi et al. had compared their GH28 family PGs with 46 other GH28 PGs from bacteria, fungi, nematodes, plants, and two insect families, Hemiptera (cicadas, leafhoppers, aphids & Co.) and Coleoptera (beetles). The Phasmatodea PGs formed a monophyletic group with the PGs of Gammaproteobacteria (and one Betaproteobacterium). Taking a closer look in their new study for a possible bacterial donor for the PGs in the Phasmatodea, the authors find that all Phasmatodea PGs branch off as a subgroup within a monophyletic group of PGs from Enterobacteriaceae, including Erwinia piriflorinigrans and Pantoea stewartii. PGs found in a number of leaf beetles (Chrysomelidae) cluster together with those from Bacteroidetes, while PGs from root-knot nematodes (Nematoda, genus Meloidogyne) cluster with a set of GH28 PGs from Betaproteobacteria and a distinctly different set of GH28 PGs from Enterobacteriaceae. The two sets of enterobacterial PGs share ~30% identity, while the identity within each group rising to >60%. Thus, acquisition of PGs by the nematodes and the leaf beetles via HGT is conceivable, but has not yet been conclusively proven. It seems clear, however, that an ancestor of the extant stick insects (Phasmatodea) picked a GH28 PG gene from an ancestral relative of the Erwinia/Pantoea-type enterobacteria, while the nematodes and leaf beetles drew from different sources.
Perspective Not really an unprecedented breakthrough finding, the presence of a pectinase gene of bona fide bacterial origin in the genomes of several Polyneoptera species – acquired by their common ancestor 60 – 100 Ma ago and retained throughout its subsequent speciation, including gene duplications and specificity changes through mutations – adds to the notion that "long-distance" or even cross-kingdom horizontal gene transfer (HGT) is probably less exceptional than surprised biologists first assumed. The acquisition of a carotenoid desaturase and cyclase gene pair of fungal origin by pea aphids – giving them a pinky-ish phenotype – appeared almost excentric at first (featured here in STC). But many more examples have been found since (one mentioned here in STC). It's a safe bet that even more examples will be found among species – from whatever branch of the tree of life – that happen to share a particular niche. As symbioses among various smaller and larger things turn out to be the rule rather than the exception, it would be surprising to not find some frequency of gene-sharing among these "cohabitants". However, the physicochemical obstacles to moving intact DNA across membranes (conjugative plasmids, phages, and viruses know these obstacles well enough!) have kept promiscuous gene swapping at bay. So, biologists, ease up, You will always find 'species' and not find inextricable 'gene bins'!
Yet 'detecting' horizontally transferred genes in a genome comes at a price. It is impossible to obtain 'experimental evidence' for HGT by meticulous observations and/or carefully controlled biochemical experiments. Instead, biologists have to do lots of math, statistics that is. HGT can only be 'detected' by (meta-)genomic and transcriptomic analyses, and in silico translation of conspicuous sequences into protein. (Spoiler: DNA sequences are of no use here as in a genome any "incoming" foreign genes are streamlined within a few hundred generations, i.e., revamped to match host AT-content, codon usage, mRNA stability requirements, cognate regulatory sequences (promoters), 'decoration' by introns, translation signals, and address labels for exported proteins). Aligning protein sequences to detect similarities is pretty straightforward*), but calculating evolutionary distances – or building robust phylogenetic trees, for that matter – is complicated and phylogeneticists have yet to agree on a statistical 'gold standard'. Clearly, procedures similar to those agreed upon by particle physicist (including the 5-sigma criterion) need to be adopted by biologists. 'Evidence' will thus be supplanted by 'probability'.
Tangential Among hard-core entomologists – the folks studying insects – the 'walking sticks' are known as Phasmatodea, or Phasmids for short. Apparently, there had been an incidence of "Horizontal Term Transfer" three decades ago that went largely undetected. We molecular biologists also knew phasmids: cloning vectors that carried, in addition to the usual (dsDNA) plasmid replication origin, a secondary origin derived from phages like f1 that, given a helper phage, enabled production of single-stranded DNA perfect for Sanger sequencing. Yet, as soon as we learned (while still in the pre-PCR days) to sequence the inserts in double-stranded plasmid vectors using custom-made primers – in both directions at once, without fumbling with phasmids or tedious subcloning in M13 vectors, yeah! – phasmids rapidly fell into oblivion. Thus endeth the confusion of terms.
*) To give an example for what "straightforward" looks like with real alignment data, go to Figure 4 of the Shelomi et al. (2014) paper by clicking the link. You see part of the aligned PG sequences for 28 species/species groups (res. 220 – 315). If you zoom in on a virtual "column" comprising residues 245 – 256 you can "see" also with untrained eyes two sequence blocks that are more similar to each other – they are in fact almost identical – than to any of the other sequences: Phasmatodea (red bar) and Enterobacteriaceae (orange bar).
Shelomi M, Jasper WC, Atallah J, Kimsey LS, Johnson BR (2014). Differential expression of endogenous plant cell wall degrading enzyme genes in the stick insect (Phasmatodea) midgut. BMC Genomics, 15, 917 PMID 25331961 (PMC Free Article)
Shelomi M, Danchin EG, Heckel D, Wipfler B, Bradler S, Zhou X, Pauchet Y (2016). Horizontal Gene Transfer of Pectinases from Bacteria Preceded the Diversification of Stick and Leaf Insects. Sci Rep, 23 (6) 26388 PMID 26814170 (PMC Free Article)