(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Dissecting the invasion of Galleria mellonella by Yersinia enterocolitica reveals metabolic adaptations and a role of a phage lysis cassette in insect killing [1]

['Philipp-Albert Sänger', 'Friedrich-Loeffler-Institut', 'Institut Für Molekulare Pathogenese', 'Jena', 'Stefanie Wagner', 'Elisabeth M. Liebler-Tenorio', 'Thilo M. Fuchs']

Date: 2022-12 

The human pathogen Yersinia enterocolitica strain W22703 is characterized by its toxicity towards invertebrates that requires the insecticidal toxin complex (Tc) proteins encoded by the pathogenicity island Tc-PAI Ye . Molecular and pathophysiological details of insect larvae infection and killing by this pathogen, however, have not been dissected. Here, we applied oral infection of Galleria mellonella (Greater wax moth) larvae to study the colonisation, proliferation, tissue invasion, and killing activity of W22703. We demonstrated that this strain is strongly toxic towards the larvae, in which they proliferate by more than three orders of magnitude within six days post infection. Deletion mutants of the genes tcaA and tccC were atoxic for the insect. W22703 ΔtccC, in contrast to W22703 ΔtcaA, initially proliferated before being eliminated from the host, thus confirming TcaA as membrane-binding Tc subunit and TccC as cell toxin. Time course experiments revealed a Tc-dependent infection process starting with midgut colonisation that is followed by invasion of the hemolymph where the pathogen elicits morphological changes of hemocytes and strongly proliferates. The in vivo transcriptome of strain W22703 shows that the pathogen undergoes a drastic reprogramming of central cell functions and gains access to numerous carbohydrate and amino acid resources within the insect. Strikingly, a mutant lacking a phage-related holin/endolysin (HE) cassette, which is located within Tc-PAI Ye , resembled the phenotypes of W22703 ΔtcaA, suggesting that this dual lysis cassette may be an example of a phage-related function that has been adapted for the release of a bacterial toxin.

Interactions of bacteria with invertebrates took place over a long period of time, and it is assumed that these animals are not only a reservoir for human pathogens, but have also shaped their evolution. We here report that Y. enterocolitica colonizes the midgut of G. mellonella via the insecticidal toxin complex Tc and subsequently migrates through the epithelial cell layer within the first 18 hours of infection. Once reaching the hemolymph, the pathogen grows to high cell densities to finally kill the insect. The massive proliferation is fostered by a set of differentially regulated genes that constitutes an adaptation of Y. enterocolitica to the nutrient-rich environment encountered within the insect larvae. A successful infection not only depends on the insecticidal Tc, but also requires the activity of a phage-related lysis cassette that is involved in Tc release, probably without affecting bacterial cell integrity. We conclude that the investigation of oral invertebrate infections contributes to a better understanding of microbial pathogenicity.

Invertebrates are often used as alternative to mammalian models of infection to study bacterial or fungal pathogenicity and to evaluate therapeutic interventions [ 25 ]. Nematodes have successfully been used to identify virulence-related genes in a broad set of bacterial pathogens [ 16 , 26 ]. On the other hand, insect models such as Galleria mellonella, the Greater wax moth, are considered to provide further insights into pathogen-host-interactions due to their more elaborated innate immune system [ 27 – 29 ]. They also allow subcutaneous injection as well as oral application of bacteria and fungi, in vivo imaging of bacterial cells, monitoring of intracellular gene expression, detection of immune responses, and the investigation of antimicrobial drugs [ 13 , 30 – 34 ]. Subcutaneous infection of G. mellonella larvae demonstrated the insecticidal activity of several Yersinia spp. and identified the enterotoxin YacT of Y. frederiksenii [ 13 , 34 , 35 ]. Moreover, many Yersinia genes, including those contributing to virulence, are up-regulated at lower temperature [ 36 – 38 ], corroborating the hypothesis that invertebrates are a natural host of pathogens and may therefore have fostered their evolution [ 39 ].

Remarkably, two highly conserved phage-related genes, which are not clustered with other phage determinants, are present in all insecticidal pathogenicity islands identified so far in Yersinia strains [ 13 ]. These genes termed holY and elyY are located between tcaC and tccC ( S1A Fig ), and their products were recently shown to act as a holin and an endolysin (HE), respectively [ 21 ]. ElyY revealed an endopeptidase with high substrate specificity that cleaves yersinial murein. Overexpression of HE lyses Y. enterocolitica at 37°C, but not at lower temperature [ 21 ]. Upon deletion of the gene encoding the Lon A protease, which is involved in the thermodependent regulation of Yersinia virulence properties [ 22 , 23 ], we observed lysis of the Y. enterocolitica mutant at 15°C, indicating that this enzyme controls the temperature-dependent activity of the HE cassette [ 24 ]. The biological role of this dual lysis cassette and its potential contribution to the insecticidal activity of Y. enterocolitica has not been elucidated.

Insecticidal Tc proteins are also present in the three human pathogenic Yersinia species (spp.) and in Y. mollaretii [ 13 ]. The pathogen Y. enterocolitica is characterized by a unique lifecycle, as some of its representatives are able to switch between two distinct pathogenicity phases that manifest in invertebrates or mammals [ 14 ]. Strain W22703 (biotype 2, serotype O:9) carries the highly conserved chromosomal pathogenicity island Tc-PAI Ye that encodes two regulators and TcA (tcaA, tcaB1, tcaB2), TcB (tcaC) and TcC (tccC1) toxin complex subunits. TcaA of Y. enterocolitica is essential for toxic activity towards larvae of the tobacco hornworm Manduca sexta and the nematode C. elegans upon oral uptake of cell lysates or living cells [ 15 , 16 ]. Tc mutants of strain T83 were shown to be attenuated in their ability to colonize the gut of orally infected mice [ 17 ], a finding that is in line with the broad cytocidal host spectrum of bacterial toxins [ 18 ]. Y. enterocolitica W22703 produces Tc proteins at environmental temperatures, but not at 37°C [ 15 , 16 ]. The thermodependent activation of insecticidal activity is mainly the result of an antagonism between the regulators TcaR2 and YmoA. The thermolabile TcaR2 is essential and sufficient to activate tc gene transcription at low temperatures [ 19 ], whereas the Yersinia modulator of virulence, YmoA, a Hha-like protein that interacts with the DNA-binding protein H-NS, represses tc gene transcription at 37°C [ 20 ].

Several bacteria are known to successfully colonize and infect invertebrates and to eventually profit from their bioconversion [ 1 ]. Key factors for insect infection are the insecticidal toxin complex (Tc) proteins, which were first purified from Photorhabdus luminescens [ 2 ]. Their oral insecticidal activity is comparable to that of the Bacillus thuringiensis- (Bt-) toxin [ 2 ]. Homologues of the Tc proteins have been described in insect-associated bacteria such as Serratia entomophila and Xenorhabdus nematophilus. 3-D structural analysis of the tripartite Tc suggests a 5:1:1 stoichiometry of the A, B and C subunits, with the A subunit forming a pentamer that associates with a tightly bound 1:1 sub-complex of B and C [ 3 – 5 ]. The TcA subunits are assumed to bind to the membranes of insect midgut cells and harbour a neuraminidase-like region that possibly confers host-specificity [ 5 ]. The B and C proteins of P. luminescens form a large hollow structure encapsulating the toxic and the highly variable carboxyl-terminus of TcC that has recently been demonstrated to ADP-ribosylate actin and Rho-GTPases [ 6 – 8 ]. The attachment of the Tc to the host cell membrane via glycans [ 9 , 10 ] is either followed by receptor-mediated endocytosis or release of the ADP-ribosyltransferase into the target cell [ 3 , 5 , 11 ]. In a pH-dependent manner, the TcA translocation channel is injected into the membrane of the host cell, and conformational changes subsequently allow the toxic component to be released into the translocation channel of TcA and from there into the cytosol [ 5 , 12 ].

Other main functional categories affected by a transcriptional switch upon larva infection were the metabolism and transport of amino acid and carbohydrates, resistance mechanisms, signaling, and motility ( Fig 8 ). Within larvae, Y. enterocolitica down-regulated the biosynthesis pathways of methionine, isoleucine, leucine, histidine, tryptophan, and glutamate ( Table 1 ). In contrast, determinants responsible for the transport of methionine, glycine, glutamate, glutamine, and histidine appeared at higher abundance in vivo in contrast to the control. This finding confirms the assumption that these amino acids are readily available in the surrounding hemolymph. The synthesis of thiamine was repressed, a finding that agrees with the reduced biosynthesis of isoleucine and leucine that require this cofactor. Another large number of up-regulated genes belongs to the categories of carbohydrate metabolism. Y. enterocolitica activated transporters including phosphotransferase systems and/or enzymes involved in the uptake and/or degradation of glycerol, sorbose, mannitol, ribose, xylose, inositol, trehalose, N-acetylgalactosamine, N-acetylglucosamine, sucrose, and glucitol/sorbitol. In addition, the pathogen induced a huge set of genes encoding factors of the nucleotide and lipid metabolisms as well as of the TCA-cycle, pointing to an increased metabolic activity and proliferation within insect larvae. Biofilm formation is not required for the virulence against insect as several biofilm and fimbriae producing genes were repressed and a biofilm repressor gene was transcriptionally activated. Motility appeared to play a pivotal role in insect infection by Y. enterocolita and by Y. entomophaga [ 29 , 42 ], because a large set of genes involved in flagella synthesis was up-regulated. Induction and repression of genes responsible for cell membrane biosynthesis pointed to major rearrangements in particular of the outer membrane within the larvae. Intriguely, several genes encoding factors involved in signaling were transcriptionally activated in the invertebrate, including the lsr operon responsible for autoinducer 2 import, regulation, and degradation, and traI encoding an acyl-homoserine-lactone synthase. No phage genes were induced, but 13 phage genes were repressed in larva-infecting Y. enterocolitica ( S1 Table ).

Owing to their important role in infection, colonization, and killing of G. mellonella demonstrated above, we first analysed the transcriptional activity of genes located on the insecticidal island Tc-PAI Ye . Strikingly, the genes encoding the activator TcaR2 and the two Tc subunits TcaA and TcaB were strongly induced (log 2 FC = 3.9, 5.5, and 3.2, respectively) 12 h p.i., but to lesser extent 24 h p.i. ( S1 Table , S1B Fig ). The endolysin gene elyY was significantly up-regulated after 24 h, but not after 12 h, implying a possible role of this enzyme in the release of the Tc. Beside the tc genes, the pathogen induced 15 virulence genes including those involved in iron acquisition, whereas a set of eight genes of this category was repressed, suggesting a role during infection of other host organisms including mammals.

To delineate the transcriptional profile of Y. enterocolitica during infection of G. mellonella, we enriched and isolated Y. enterocolitica by immunomagnetic separation [ 41 ] from the larvae 12 h and 24 h after infection. These two time-points were chosen due to the results of the time course experiments ( Fig 5A and 5B ). Growth of W22703 cells in minimal medium with glucose as the carbon and energy source served as the reference condition. Following RNA sequencing and read analysis, we identified ~3,600 protein-encoding genes from a genome carrying ~4,000 genes [ 14 ], pointing to a 90% coverage of the transcriptional responses investigated here. Setting the threshold to a log 2 FC ≥ |1.5|, we determined 524 non-redundant transcripts to be significantly more abundant and 301 to be less abundant in contrast to the control ( S1 Table ).

To investigate a possible activity of the insecticidal gene products on cells of the hemolymph, we monitored their morphology 24 p.i. For this purpose, hemolymph preparations of G. mellonella larvae were fixed with methanol and then stained by Giemsa solution. The hemocytes derived from W22703-treated larvae began to form aggregates in comparison with LB medium as control ( Fig 7A ). Cell agglutinations and a fading of the chromatin colour of the cell nucleus occurred more frequently. In contrast, hemolymph preparations of larvae one day after oral infection with W22703 mutants lacking tcaA, HE, tccC, or tcaR2, which encodes the activator of tc genes [ 19 ], showed hemocyte cell morphologies similar to those of the untreated controls ( Fig 7B ).

To further investigate the role of the HE cassette in Tc release, we infected G. mellonella larvae with the reporter strain W22703 tcaA::rfp. Following tissue section and immunostaining with anti-Yersinia antibody and anti-RFP antibody, we detected TcaA in Yersinia-rich hemolymph areas of G. mellonella 24 h p.i. with W22703 tcaA::rfp and W22703 ΔHE tcaA::rfp/pACYC-HE ( Fig 6B ). In the absence of the lysis cassette, however, TcaA::Rfp was not detected despite the presence of W22703 ΔHE tcaA::rfp cells. To test whether or not the promoter of the lysis cassette is active in vivo, we infected G. mellonella larvae with strain W22703 P HE ::rfp that harbours a chromosomal transcriptional fusion of rfp with the HE promoter. Although this strain densely proliferated within the hemolymph, we failed to stain RFP possibly due to no or weak P HE activity, an inference that is in line with the results of the transcriptome analysis ( S1B Fig ). Taken together, these data suggest that the HE cassette is responsible for the transport of the insecticidal Tc.

To determine whether or not the factors encoded on the insecticidal Tc-PAI Ye play a role during the infection process delineated above, G. mellonella larvae were orally infected with Y. enterocolitica W22703 (6.3 × 10 5 CFU), W22703 ΔtcaA (8.3 × 10 5 CFU), W22703 ΔtccC (6.4 × 10 5 CFU), W22703 ΔHE (7 × 10 5 CFU), and W22703 ΔtcaR2 (7.8 × 10 5 CFU). In contrast to the parental strain W22703 that proliferated to high cell numbers in the hemolymph, no mutant cells were detected by FITC-staining of tissue sections made 24 h p.i. ( Fig 6A ). These data confirm that the insecticidal Tc as well as the tc gene activator TcaR2 and the HE lysis cassette are required for full virulence of Y. enterocolitica W22703 towards G. mellonella. In particular, we hypothesize that the Tc-PAI Ye is responsible for midgut colonization and entering of the hemolymph.

No fluorescence signal was obtained when we fed LB medium to larvae as a control ( Fig 5C ). Next, E. coli cultures were applied analogously to the experiments with Y. enterocolitica. Using an anti-E. coli antibody, E. coli cells, however, were not detected in the gut or in the hemolymph of G. mellonella 24 h after infection. To test the specificity of the antibody, E. coli DH5α cells were injected into muscle cells of a chicken leg and shown to be FITC-labeled 24 h p.i. The control experiments indicate that in contrast to Y. enterocolitica, E. coli is not able to survive in G. mellonella.

The tissue sections monitored by fluorescence microscopy show antibody-stained Y. enterocolitica cells in the (A) gut or (B) hemolymph of G. mellonella 4 h, 6 h, 12 h, 18 h, and 24 h after infection. (C) The controls depict the gut area of G. mellonella that were fed with LB (left) or infected with E. coli (middle) 24 h ago. The tissue sections were stained with a Yersinia-specific or an E. coli-specific antibody. Functionality of the anti-E. coli antibody was demonstrated by the application of E. coli into muscle tissue of chicken (right). Cyan-coloured areas in the gut area of G. mellonella are unspecific bonds of the anti-E. coli antibody. Representative preparations are shown; the scale is indicated. 1 = intestinal epithelium, 2 = intestinal lumen, 3 = fat tissue, 4 = hemolymph, 5 = appendix, 6 = Malpighian vessels, 7 = muscle cells, 8 = E. coli, 9 = antibody cross reactions.

The strongest proliferation of W22703 was observed between one and three days p.i. ( Fig 4 ). We hypothesized that Y. enterocolitica starts its infection in the midgut and subsequently invades the tissues of G. mellonella larvae to proliferate in the hemolymph. To dissect the infection process preceding this multiplication in more detail, we performed a time course experiment using larvae infected with 6.3 × 10 5 Y. enterocolitica W22703 cells. Longitudinal sections through the middle of the larvae were prepared 4 h, 6 h, 12 h, 18 h, and 24 h p.i. and stained with FITC-conjugated Yersinia-antibodies. The histological analysis revealed that Y. enterocolitica is present in the midgut with increasing cell numbers until 12 h p.i. At this time-point, W22703 is mainly detected close to glandular epithelial cells of the midgut, implying a tropism of Y. enterocolitica for endodermal tissue ( Fig 5A ) . Eighteen hours p.i., the gut appeared to be Yersinia-free, whereas a high number of cells is now detected in the hemolymph where W22703 proliferated with the next six hours to a high cell density. The larval tissues were completely overgrown with Y. enterocolitica 48 h p.i. when most animals were dead, indicating that the bacteria have started bioconversion of the cadaver ( Fig 5B ) . To verify the finding that Y. enterocolitica W22703 cells are mainly found in the circulating fluid, we isolated 10 μl hemolymph from larvae infected with 1.6 × 10 5 bacteria. 24 h p.i., the animals showed clear signs of melanisation. The brownish hemolymph contained 2.3 × 10 7 to 4.7 × 10 7 CFU as detected on selective agar.

Proliferation within the insect host would indicate a successful infection by Y. enterocolitica. To determine the bacterial load over the time, we infected G. mellonella larvae with the same strains used in Fig 2 and applied similar infection doses. One, three and six days p.i., the homogenate of six animals per time-point was plated on selective LB agar plates, and the CFU were enumerated. Strikingly, when we injected 9.0 ± 0.2 × 10 5 CFU of mutant W22703 ΔtcaA, the total numbers of surviving bacteria rapidly decreased to 1.0 × 10 3 after one day and to eleven CFU after three days, and the strain was completely absent from the larvae six days p.i., probably due to passage through the gut followed by excretion ( Fig 4 ). In contrast, the mutants W22703 ΔHE and W22703 ΔtccC that were applied with 4.0 × 10 5 CFU and 4.0 × 10 5 CFU, respectively, proliferated within the first day p.i. to 2.2 × 10 6 CFU and 2.8 × 10 6 CFU, but were not detected from day three on. This discrepancy suggests that TcaA is involved in adherence to epithelial cells and thus in midgut colonization, without requiring TccC. When larvae were infected with 4.0 × 10 5 CFU of the ΔtcaA and ΔHE mutants, and with 1.4 × 10 6 CFU of strain W22703 ΔtccC, all of which carrying recombinant plasmids that complemented the chromosomally deleted genes, the bacterial burden at days one to six p.i. increased approximately to that of the parental strain W22703 applied with 9.0 × 10 5 CFU, indicating a successful complementation of the gene deletions.

In addition, we monitored the behaviour and the morphology of the larvae each day until six days post infection (p.i.), and again at day nine p.i. ( Fig 3 ). Immediately after oral application, the control group infected with LB medium did not differ from the larvae infected with the three deletion mutants W22703 ΔHE, W22703 ΔtcaA, and W22703 ΔtccC with respect to motility and colour. In contrast, the application of W22703 and strains W22703 ΔHE/pACYC-HE, W22703 ΔtcaA/pACYC-tcaA, and W22703 ΔtccC/pBAD33-tccC resulted in a higher activity of the larvae. After 24 h, a strong melanization in the groups infected with W22703 and with the mutants harbouring complementing plasmids was observed. Similar to the untreated control groups, cocoons surrounded those larvae that had been infected with the deletion mutants, indicating that healthy individuals only are able to produce this protective housing. At days two to three, cocoon formation continued, whereas increasing numbers of larvae infected with W22703, W22703 ΔHE/pACYC-HE and W22703 ΔtcaA/pACYC-tcaA died. At days five to nine, morphological signs of infection did not further enhance. Nine days after infection, the first pupations events were observed.

Larvae were orally infected with W22703, its mutants lacking tcaA, HE, and tccC, and with mutants carrying the plasmids pACYC-HE, pACYC-tcaA, and pBAD-tccC. Application of LB medium served as control. Life span assays were performed for nine days, and the viability of the larvae was monitored each day to determine the survival rate of the larvae. The raw data were plotted by the Kaplan-Meier method. The Kaplan-Meier-plot is based on triplicates with 36 larvae in total per strain. The curves were compared to each other using the log-rank test, which generates a p value testing the null hypothesis that the survival curves are identical. Data were fit to exponential distribution. p values of 0.05 or less were considered significantly different from the null hypothesis (p value W22703 ΔHE/pACYC-HE = 0.0194; p value W22703 ΔtcaA/pACYC-tcaA = 0.0369; p value W22703 ΔtccC/pACYC-tccC = 0.0251). All graphs start at 100%.

Survival assays were performed with larvae of G. mellonella to further investigate the function of Tc-PAI Ye determinants in the interaction of Y. enterocolitica with insects. Recently, we demonstrated that subcutaneous infection of G. mellonella larvae with W22703 (LD 50 ~ 10 4 cells) results in a killing rate similar to that of W22703 ΔtcaA, suggesting that the Tc plays a main role in the initial phases of infection rather than during systemic infection [ 13 ]. Here, we orally infected larvae with 5.7 × 10 5 CFU, 7.8 × 10 5 CFU, 5.6 × 10 5 CFU, and 6.2 × 10 5 CFU, respectively, of W22703 and its mutants W22703 ΔtcaA, W22703 ΔHE, and W22703 ΔtccC, and monitored the larvae for nine days. Larvae infected with Y. enterocolitica strain W22703 exhibited a significantly reduced survival rate with a time to death of 50% (TD 50 ) = 3.67 ± 1.12 days. In the infection experiments with the three mutants lacking tcaA, HE, and tccC, all larvae survived, corresponding to a challenge of the larvae with LB medium ( Fig 2 ). To genetically validate that tcaA, HE, and tccC are essential for the toxicity of W22703 towards G. mellonella, we orally infected larvae with W22703 ΔtcaA/pACYC-tcaA (9.0 × 10 5 CFU), 4.0 × 10 5 CFU (W22703 ΔHE/pACYC-HE), and W22703 ΔtccC/pBAD-tccC (4.0 × 10 5 CFU). Due to the slight leakiness of the pBAD-promoter experienced recently [ 21 ], arabinose was not added to further induce tccC transcription. TD 50 of 2.91 ± 1.46 days, 1.83 ± 0.51 days, and 3.90 ± 0.41 days, respectively, were determined. Thus, the mutants harbouring recombinant plasmids that complement the deletion did not significantly differ in their insecticidal activity from that of the parental strain W22703 after one week, demonstrating that the in trans complementation of ΔtcaA, ΔHE, and ΔtccC fully restored the insecticidal phenotype of W2703. Taken together, these life span assays indicate that the two toxin subunits TcaA and TccC as well as the HE lysis cassette are strictly required for the oral toxicity of Y. enterocolitica strain W22703 towards the insect larvae.

Force-feeding of bacterial cultures was carefully performed by injection with a Hamilton syringe. In the case of accidental tissue perforation, resulting in direct injection of bacteria into the hemocoel and thus early death of larvae, the larvae were excluded from the experiment. Owing to the larva weight of 150–200 mg, a maximum of 5 μl culture or medium were applied. Preliminary experiments to establish the optimal infection dose were conducted within a range of 10 2 to 10 8 colony forming units (CFU) of Y. enterocolitica and showed dose-dependent phenotypes. When 10 7 −10 8 CFU were injected, all larvae died between four and 14 h p.i., and the hemolymph of these cadavers contained Y. enterocolitica cells. No lethality and no melanisation, however, was observed after applying a lower dose of 10 2 to 10 4 cells at least until nine days p.i., and no Y. enterocolitica cells were detected in the hemolymph of the larvae. Finally, 10 5 −10 6 CFU revealed as optimal dose to perform infections of G. mellonella larvae with Y. enterocolitica, a value that corresponds well with those used in subcutaneous applications [ 31 , 40 ].

(A) Underside of a G. mellonella larva. The foregut, the midgut, and the hindgut are indicated by arrows. (B) Dissected digestive tract after instillation with methylene blue. (C) Schematic drawing of the digestive tract. The stomadeal valve (SV) separates foregut lined by cuticular epithelium from midgut lined by glandular epithelium. The proctodeal valve (PV) is located between midgut and hindgut. The distinct epithelium cranial to the PV is labeled in green. The ingesta (I) in the midgut is covered by the peritrophic membrane (PM) and separated from the mucosa by the ectoperitrophic space. The crop and both valves are surrounded by a thick layer of musculature (red). (D) Longitudinal and sagittal histological section along the middle through G. mellonella. (E) Magnification of mouth, esophagus, and crop lined by the cuticular epithelium and surrounded by muscle cells. (F) Magnification of the SV between crop and midgut. (G) Magnification of the midgut lined by glandular epithelium. The ingesta is surrounded by the peritrophic matrix (PM) and separated from mucosa by the ectoperitrophic space. (H) Magnification of the PV between midgut and hindgut lined by cuticular epithelium. Vacuolated columnar epithelial cells line the midgut cranial to the PV. (D)-(H) are paraffin sections stained by hematoxylin and eosin. Sections are indicated by numbers: 1 = mouth, 2 = esophagus, 3 = crop, 4 = glandular intestine, 5 = transition zone, 6 = cuticular intestine, 7 = rectum, 8 = anus. Photos of representative preparations are shown; the scales are indicated.

Discussion

To the best of our knowledge, the oral infection of G. mellonella by a bacterial pathogen performed here is of yet unprecedented resolution with respect to the molecular details, although force-feeding of G. mellonella was already established [43,44]. In contrast to subcutaneous injection in the use of insect larvae as model for bacterial virulence properties towards mammals, oral application mimics natural routes of infection that in particular take place during the bioconversion of animal cadavers by bacteria, fungi, and larvae [45]. In the present study, oral application of Y. enterocolitica demonstrated that the larvae are a powerful tool for dissecting the molecular steps required for a successful and lethal infection of G. mellonella by this enteropathogen. The distinct phases of infection identified here include survival in the gut, adhesion to and penetration of the midgut epithelial cell layer, massive proliferation within the hemolymph, hemocyte deformation, and insect killing. Despite an oral infection dose between 4.0 × 105 and 3.0 × 106 CFU, only few Y. enterocolitica cells were found in the gut according to cell counting and immunostaining, indicating that the majority of the bacterial cells is unable to maintain itself, and that W22703 does not substantially proliferate in the gut. A few Y. enterocolitica W22703 cells were seen in close proximity to glandular epithelial cells of the midgut. They probably cross the epithelial barrier via M-cells and migrate into the underlying tissues. The proliferation data shown in Fig 4 allow the conclusion that TcaA enables the adherence to epithelial cells, whereas the enzymatic active subunit TccC plays a minor role here. The epithelial cell contact is compatible with the finding that the Tc of Y. pseudotuberculosis causes initial membrane ruffling of human colonic epithelial (Caco-2) cells [46]. The passage from the gut to the hemolymph occurs approximately 12 h to 18 h after ingestion, a delay reflecting the time of the invasion process. Once they reach the hemocoel, the open circulatory system of the larvae, the pathogen exhibits a massive proliferation, pointing to excellent growth conditions and nutrient availability in this compartment [29]. Strain W22703 seems to withstand the phagocytic or growth suppressing activities of the hemocytes, probably as a result of Tc activity as suggested by changes of the hemocyte morphology in the presence of TcaA, TccC, and HE.

The in vivo transcriptome of Y. enterocolitica delineates its physiological and biochemical adaptations to the insect. The pattern of up- and down-regulated genes not only point to the relevance of distinct virulence factors, signalling, and motility for a successful infection, but also to the availability of numerous carbohydrates and amino acids in the insect body that fuels the metabolism and thus the proliferation of Y. enterocolitica. The carbon and energy sources are derived from either the diet or the host, such as glycerol, ribose, inositol, N-acetylgalactosamine, or trehalose, which is present in the hemolymph as well as in honey, a component of G. mellonella feeding. The upregulation of genes involved in the uptake and degradation of sorbose and its reduced forms glucitol/sorbitol, of mannitol, of sucrose, and of xylose point to a specific metabolic adaptation of Y. enterocolitica to substrates fed by insects. Sorbose and mannitol are found in plant saps, and xylose is a monomer of hemicelluloses. Sucrose is one of the most concentrated nutrient available for sap-feeding insects [47]. In addition, the genes responsible for N-acetylglucosamine utilization suggests that chitin, a constituent part of the peritrophic matrices that line the inner surface of the gut in many insects [48], is not only degraded as a first step of colonization and invasion of the midgut epithelium, but metabolically utilized by the pathogen. When entering the G. mellonella larvae, strain W22703 in particular down-regulates the genes responsible for the synthesis of methionine, branched-chain amino acids, histidine, tryptophane, cysteine, and glutamate, indicating a sufficient availability of these amino acids within the insect. This is well in line with an increased capacity to import and degrade methionine, proline, glycine, urea, cysteine, threonine, glutamate/aspartate, serine, and histidine. Histidine is one of the most abundant free amino acids in the Hyalophora gloveri fat body [49]. It is worth to note that urease, inositol, and histidine degradation belong to metabolic properties that are common to Y. enterocolitica and P. luminescens [38]. The low temperature-dependent transcription of these and many other factors [37] fits to the proliferation of W22703 in insect larvae. Reprogramming of Y. enterocolitica in vivo activities also includes the increase of lipid import and degradation, and of energy production and conversion. The latter category reflects the massive proliferation in the hemolymph. In addition, the in vivo transcriptional pattern revealed the up-regulation of virulence factors mainly involved in hemolysis and iron scavenging, which are probably specifically directed against insects (S1 Table). A similar response was monitored in the interaction of Y. entomophaga with G. mellonella [35]. Autoinducer-2 import was also found to be up-regulated during insect infection. Given that more than 300 AI-2 regulated genes involved in regulation, metabolic activity, stress response and pathogenicity are known in P. luminescens [50], this points to an important role in signalling during insect infection.

We provide for the first time in vivo evidence for an involvement of the holin/endolysin cassette in Tc release and thus in nematocidal and insecticidal of Y. enterocolitica strain W22703. The dual lysis cassette is highly conserved in the genomes of Yersinia spp. where it is localised between genes encoding Tc subunits, and is also present in the Tc locus of P. luminescens [13]. These findings pointed to a functional role of the phage genes in the insecticidal activity of these bacteria, for example by cell lysis [51,52]. In Y. pestis, it was postulated that the release of the Tc is mediated by a type III secretion system (T3SS) [53], but this hypothesis was recently refuted [54]. This agrees with the lack of virulence plasmid pYV, which encodes a T3SS, in strain W22703 [14].

Thus far, there are only few examples of bacterial toxins that are released into the environment by phage-related factors. In Serratia marcescens, a holin and endopeptidase cassette were identified to be required for the secretion of a chitinase [55]. A further example of a correlation between phage lytic genes and toxicity is the putative coupling of the λ phage lytic cycle and the release of the phage-encoded toxin Stx from Shiga toxin producing E. coli [56,57]. An N-acetyl-ß-D-muramidase similar to phage endolysins was shown to be essential for Salmonella typhoid toxin secretion [58]. The holin-like protein TcdE was shown to be required for Clostridioides difficile toxins TcdA and TcdB secretion via pore formation, and toxin release is independent of bacterial cell lysis [59,60]. The frequent neighbourhood of phage-related lysis factors to bacterial toxins and other secreted factors supports the hypothesis that protein release upon the activities of a dual lysis cassette evolved multiple times and defines a more widespread mechanism that was proposed to be termed the type 10 secretion system [61].

[END]
---
[1] Url: https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1010991

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/