Antimicrobial Peptides Keep Insect Endosymbionts Under Control

See allHide authors and affiliations

Science  21 Oct 2011:
Vol. 334, Issue 6054, pp. 362-365
DOI: 10.1126/science.1209728


Vertically transmitted endosymbionts persist for millions of years in invertebrates and play an important role in animal evolution. However, the functional basis underlying the maintenance of these long-term resident bacteria is unknown. We report that the weevil coleoptericin-A (ColA) antimicrobial peptide selectively targets endosymbionts within the bacteriocytes and regulates their growth through the inhibition of cell division. Silencing the colA gene with RNA interference resulted in a decrease in size of the giant filamentous endosymbionts, which escaped from the bacteriocytes and spread into insect tissues. Although this family of peptides is commonly linked with microbe clearance, this work shows that endosymbiosis benefits from ColA, suggesting that long-term host-symbiont coevolution might have shaped immune effectors for symbiont maintenance.

Cooperative associations between animals and symbiotic bacteria are widespread in nature and common in insects that exploit unusually restricted nutritional resources (1). In many insects, intracellular bacteria (endosymbionts) are transmitted vertically and provide nutrient supplementation to their hosts, thereby improving their adaptive traits and their invasive power (24).

However, maintaining the beneficial nature of this long-term relationship requires both the host and the symbiont to constrain adaptive interactions. Genomic and evolutionary data have shown that major deletions and mutations of genes occur in endosymbionts, some of which are involved in bacterial virulence and host tolerance (57). Data on how host immune systems have evolved to tolerate cooperative bacteria remain scarce and are mainly limited to extracellular associations with environmental and/or horizontal symbiont transmission (8, 9).

To protect permanent endosymbionts from the host’s systemic immune response, and prevent competition with opportunistic invaders, symbionts are sequestered in bacteria-bearing host cells, called the bacteriocytes, which, in some species, group together to form a bacteriome (10). To investigate the immune specificities of bacteriocytes, we have studied associations with Sitophilus genus. This genus of beetle includes three cereal pest species (Sitophilus oryzae, S. zeamais, and S. granarius), all of which share intimate intracellular symbiosis with a Gram-negative γ-Proteobacterium, called Sitophilus primary endosymbiont (SPE) (11, 12). In contrast to Buchnera in aphids, the Sitophilus association with SPE is relatively recent [dating to 20 million years ago (Ma)] (13, 14). The SPE genome has not suffered appreciable genome shrinkage (15) and retains secretion systems and bacterial wall elements that are implicated in bacterial recognition by the host immune receptors (6). Transcriptomic data revealed that SPE induces a strong systemic response when injected into weevil hemolymph (16), whereas permanent infection of bacteriocytes with SPE leads to the up-regulation of only one antimicrobial peptide–encoding gene coleoptericin-A (colA, fig. S1) (16).

In immunohistochemistry experiments with antibody against ColA (14), we found that, in aposymbiotic insects, colA is expressed constitutively in epithelial cells surrounding the intestine and in the fat body, with a high concentration under the cuticle (Fig. 1B and fig. S2). In symbiotic insects, colA is further expressed in all the tissues housing endosymbionts (Fig. 1B). Similarly, follicular cells also expressed ColA as a thin layer around the oocytes, and ColA signals were stronger in bacteriocytes surrounding bacteriomes (Fig. 1B). Thus, a relatively high expression of colA in tissues facing the external environment, and at the boundary of tissues housing endosymbionts, supports the idea that ColA may either prevent pathogen intrusion or retain endosymbionts within the bacteriomes and oocytes. Moreover, ColA appeared to colocalize with endosymbionts in bacteriocytes (Fig. 1B). Using immunogold electron microscopy, we found ColA located inside SPE cytoplasm (Fig. 1C), and some ColA spots were also attached to the bacterial membrane surface. Overall, microscopic observations indicate that ColA expression targets endosymbionts in both somatic and germ cell lines.

Fig. 1

ColA peptide distribution in weevil tissues. (A) Schemes of larva and adult weevils showing bacteriome localizations (in red). (B) (Upper panel) Tissues from aposymbiotic S. zeamais: Left and middle images are cuticle sections stained with a preimmune serum [negative controls can be found in SOM (14)] and with antibody against ColA (anti-ColA), respectively; the right image is a gut section stained with anti-ColA. ColA signals are detected in the fat body, with relatively high intensity under the cuticle and within gut epithelial cells. (Middle and lower panels) Tissues from symbiotic insects stained with anti-ColA. Middle panel: ColA signals can be seen in oocytes and follicular cells (left), in apical bacteriomes of ovaries (middle), and in adult mesenteric caeca (right); lower panel: ColA signals in larval bacteriome (left), within bacteriocytes (middle), and in bacteriome squashes (right) (14). Arrows indicate high ColA signals at the periphery of tissues and show ColA colocalizing with SPE in bacteriocytes and bacteriome squashes. (C). Immunogold staining of SPE with anti-ColA. Bacteriocyte sections are shown with ColA spots inside symbiont cytoplasm and attached to bacterial membranes.

We tested the antimicrobial activity of ColA against microbes. The weevil paralog ColB was used for comparison because the colB gene, unlike colA, is down-regulated in bacteriocytes (16) and ColB shows important sequence identity with ColA (fig. S3). ColA and ColB showed similar bactericidal activity against the Gram-positive Micrococcus luteus and the Gram-negative Escherichia coli (Fig. 2), whereas the yeast Saccharomyces cerevisiae tolerated both peptides (table S1). Both peptides exhibited a wide range of bacteriostatic activity against Gram-negative bacteria. SPE persists in the bacteriocyte because colA expression may lead either to bacterial growth inhibition or dose-dependent bacterial clearance. At bacteriostatic concentrations, both coleoptericins halt E. coli growth, but only ColA inhibited cell division or caused bacterial gigantism (Fig. 2B). The shape of M. luteus was not affected by ColA or by ColB (fig. S4). Thus, ColA and ColB have distinct functions in weevil immunity and symbiosis in regard to their general effect on Gram-negative bacteria.

Fig. 2

ColA and ColB activities against bacteria. (A) ColA (solid line) and ColB (dashed line) activities against E. coli (triangles) and M. luteus (squares). ColA and ColB have a similar inhibitory effect (analysis of variance, P =0.76). They show bactericidal activities against M. luteus at low concentrations. For E. coli, low concentrations of ColA and ColB have bacteriostatic activity, and higher concentrations kill this bacterium. (B) Effect of low concentrations of ColA and ColB on E. coli morphology. Bacteria were incubated in LB broth (left, control), in LB with 8 μM ColA (middle), or in LB with 8 μM ColB (right). Cell gigantism is observed with ColA peptide only. (C) Gram staining of endosymbionts from S. oryzae (rod-shaped, left), S. zeamais (spiral, middle), and R. ferrugineus (filamentous, right). (D) Chromosome visualization of E. coli treated with 8 μM ColA (left), SPE (middle), and Nardonella (right). SPE and Nardonella were isolated from larval bacteriomes of S. oryzae and R. ferrugineus, respectively. Chromosome number was highest in Nardonella.

Since the discovery of coleoptericins (17, 18), their function in the immune system and their role in symbiosis have not yet been explored. One notable observation is that all the coleopteran endosymbionts observed exhibit a similar elongated morphology (Fig. 2C), resembling symbionts in other associations. For example, in Rhizobium-legume symbiosis, Rhizobium elongation has been interpreted to be the result of repeated chromosome DNA replication without cell cytokinesis (19). We measured the relation between bacterial size and genome amplification in E. coli, SPE, and Nardonella, the ancestral endosymbiont of weevils (125 Ma) (13). All bacteria were polyploid, and bacterial size was highly correlated with chromosome number. The highest scores were seen in Nardonella, with 120 chromosomes observed in a giant cell of 200 μm (Fig. 2D). Bacterial division in plants is inhibited by nodule-specific cysteine-rich peptides (20) that induce irreversible elongation of bacteria and render them incapable of multiplying in vitro. By using phylogenetically unrelated molecules, plants and animals target bacterial cytokinesis while preserving DNA replication, hence “domesticating” the bacteria as symbionts.

To elucidate the mechanism by which ColA reaches the bacterial cytoplasm and elicits cell elongation, we used far-Western blotting to identify bacterial molecules targeted by ColA and ColB peptides. ColA specifically interacted with OmpA, OmpC, rp-L2, EF-Ts, and GroEL (fig. S5 and table S2). No interaction was detected with Hsp60, the eukaryotic cytosolic homolog of GroEL. ColB interacted with OmpC and proteins involved in translation, but not with OmpA or GroEL.

As with colicins and phages (21), it is likely that Omps are receptors that allow ColA to enter the cell. Tight attachments of ColA to the SPE membrane (Fig. 1C) support this assumption, and SPE genome sequence analysis showed that SPE encodes a functional ompC gene (table S3). Although the genome sequence is not available, we found that ColA targets Nardonella of the palm weevil Rhynchophorus ferrugineus (fig. S6), suggesting that ColA may have a broad impact on weevil symbioses. Whether Nardonella has retained omps or whether ColA enters the bacterium by other mechanisms remains to be determined.

We propose that after entering the cytoplasm, ColA elicits cell elongation through interaction with GroEL, because ColA, but not ColB, interacts with GroEL protein, and because groEL mutations in E. coli also trigger cell gigantism (22). The absence of any interaction between ColB and GroEL also indicates that ColA may have evolved a specific interaction with GroEL. In this context, it is notable that GroEL is the most abundant protein in insect endosymbionts (23); however, selective up-regulation of groEL has often been interpreted as an adaptive mechanism for protein folding in endosymbionts with a high A+T bias in the genome (24).

We used RNA interference (RNAi) to inhibit colA transcription in the larval weevils. Injection of dsRNA–colA resulted in a significant reduction in the number of bacteriocyte colA transcripts and abundance of ColA peptide for more than 2 weeks (fig. S7). In contrast to the plant-Rhizobium interaction, colA inhibition resulted in the SPE population declining by half (Fig. 3A). However, whether this was due to resumption of cytokinesis or multiplication of small bacteria is unclear. We quantified SPE DNA by quantitative polymerase chain reaction (qPCR) and showed that colA inhibition did not affect bacterial chromosome replication (fig. S8). This, and the decreased number of large-sized cells after RNAi treatment (Fig. 3A), supports the former hypothesis (i.e., resumption of cytokinesis and loss of the elongated form) but does not exclude the latter (multiplication of the small-sized cells) if we consider bacterial turnover and de novo synthesis processes.

Fig. 3

Effects of colA inhibition with RNAi on SPE size and location. (A) Typical forward scatter–area/side scatter–area plots showing size (x axis) and granularity (y axis) of SPE isolated from larvae injected with dsRNA-gfp (left) and with dsRNA-colA (right). Three SPE populations were defined arbitrarily. Small-sized cells (P1) and intermediate-sized cells (P2) significantly increased with dsRNA-colA treatment, whereas the population of large-sized cells (P3) decreased (χ2-test, P < 0.0001; see percentage values). The mean size of P2 and P3 significantly decreased (Mann-Whitney test, P < 0.0001), whereas the mean size of P1 was equal in dsRNA-gfp– and dsRNA-colA–injected larvae (Mann-Whitney test, P = 0.37). (B) FISH visualization of SPE 9 days (upper panels) and 14 days (lower panels) after larvae were injected with dsRNA-gfp (left) and dsRNA-colA (right). colA inhibition resulted in SPE escaping the bacteriome (see arrows). See table S4 for experimental details.

To understand ColA function in symbiosis, we monitored SPE in insects by fluorescence in situ hybridization (FISH) after dsRNA-colA injections. Unexpectedly, several SPE cells exited the bacteriome on the ninth day after treatment (Fig. 3B). This phenomenon increased at day 14, when bacteria were found spread throughout the larval tissues. Nevertheless, symbiont “escape” from the bacteriome did not affect insect mortality under laboratory conditions (table S4), although these data indicate that ColA acts to prevent bacterial tissue invasion.

The weevil ColA peptide demonstrates several properties important to immunity and symbiosis. It appears to act as a first line of defense in insects against microbial intrusion, and the range of bacteriostatic and bactericidal activities of ColA suggests its precise regulation of endosymbiont number and location. The interaction of GroEL with ColA (but not with ColB) supports the idea that long-term coevolution may have selected ColA for this symbiotic function.

Supporting Online Material

Materials and Methods

Figs. S1 to S9

Tables S1 to S6

References (25, 26)

References and Notes

  1. Supporting material is available on Science Online.
  2. Acknowledgments: This work was supported by INRA, INSA de Lyon, the French ANR-06-BLAN-0316 (EndoSymArt) and ANR-2010-BLAN-170101 (ImmunSymbArt), a grant from Région Rhône-Alpes (cluster infectiologie), and the COST action FA0701 (Arthropod Symbioses). The data reported in this paper are posted in the supporting online material (14). colA (EY122872) and colB (EY122826) sequences are published in GenBank. SPE omp sequences (GenBank JN575265, JN575266, JN575267) were provided by C. Dale and R. B. Weiss (University of Utah) from an ongoing SPE genome sequencing and annotation project supported by NSF grant EF-0523818 and the Ministerio de Educación y Ciencia project BFU2006-06003/BMC to A. Moya and A. Latorre (University of Valencia). We thank R. Gil and K. Oakeson for omp sequence analysis; V. E. Shevchik, G. Condemine, and M. Lemaire for supplying E. coli and S. cerevisiae strains; W. J. Miller and B. Loppin for critical reading of the manuscript; and V. James for correction of English. Electron microscopy was carried out in Centre Technologique des Microstructures (UCBL). This paper is an homage to the work of Paul Nardon.
View Abstract

Navigate This Article