Versatile roles of the chaperonin GroEL in microorganism–insect interactions

Authors


Abstract

The chaperonin 60 (Cpn60) is present in all three kingdoms of life and is one of the most conserved proteins in living organisms. The Escherichia coli Cpn60 (GroEL) is the best studied representative of the huge Cpn60 family. It is an essential protein because in conjunction with the chaperonin 10 (Cpn10 or GroES) it forms a protein-folding machine required for correct folding of many proteins and for recycling of misfolded proteins. As many other chaperones, GroEL and GroES are also known as heat-shock proteins (HSPs), since heat stress leads to a strong induction of their expression, a measure to counteract the increase in misfolded proteins as a result of a high nonphysiological temperature. A large amount of literature is available which is dedicated to the elucidation of how protein folding is assisted by this molecular chaperone. However, apart from this primary task, additional so-called ‘moonlighting’ functions of GroEL proteins unrelated to their folding activity have emerged in the past years. In fact, it becomes apparent that GroEL proteins have diverse functions in particular in mutualistic and pathogenic microorganism–host interactions. In this brief review, we describe some of these recent findings focusing on the importance of GroEL for microorganism–insect interactions.

Mutualistic interactions of bacteria and insects: a brief introduction

Insects are the most diverse taxon of living organisms and play key roles in all terrestrial and to a lesser extent also in aquatic ecosystems. Their evolutionary success is often aided by the incorporation of endosymbiotic bacteria that provide nutrients to their hosts such as vitamins or essential amino acids (Douglas, 1998; Zientz et al., 2004; Baumann, 2005). These bacteria have allowed insects to enter ecological niches otherwise unavailable to them as nutritional requirements would not be met without the endosymbionts. Aside from nutritional upgrading of unbalanced food resources, endosymbionts may mediate other ecologically important host traits such as protective functions against pathogens or predators, host plant use or the susceptibility towards pesticides (for reviews see Feldhaar, 2011; Ferrari & Vavre, 2011; Oliver et al., 2014).

Endosymbiotic bacteria are often categorized as obligate (primary) and facultative (secondary) endosymbionts. Especially, primary endosymbionts contribute to the provision of essential nutrients, while the functions of secondary endosymbionts may be more diverse. It was estimated that up to 20% of all insect species may carry primary endosymbionts (Buchner, 1965; Baumann, 2005). Primary endosymbionts are exclusively vertically transmitted via the germ line and most often show a long co-evolutionary history with their hosts, ranging from a few million to more than 200 million years. Many insects may also harbour facultative secondary endosymbionts which often are present in parts of the host population only and generally have a relatively recent symbiotic origin. While primary endosymbionts are typically found intracellularly within specialized cells, the bacteriocytes, and in the reproductive tissue, the secondary endosymbionts may be found extra- or intracellularly in different types of tissue and also in the haemocoel and can be transmitted horizontally and/or vertically (Dale & Moran, 2006). Bacteria constituting the gut microbial communities are by definition also referred to as endosymbionts, although to date it is not clear which proportion of them constitutes a stable element within the insect gut in contrast to transient ‘tourists’ (Engel & Moran, 2013).

A hallmark of primary and secondary endosymbionts is reductive genome evolution as compared to their free-living relatives (McCutcheon & Moran, 2011). Especially primary endosymbionts have small or very small genomes ranging from 800 Kb to astonishing 112 Kb (Bennett & Moran, 2013). Thus, some endosymbionts have genome sizes overlapping with those of cell organelles (McCutcheon, 2010; Reyes-Prieto et al., 2014). Genomic studies have shown that these endosymbionts have lost genes with functions redundant to the host's metabolic capabilities or those that may be deleterious for the host such as virulence factors. In addition, there is a strong tendency to lose factors involved in regulation, recombination and DNA repair.

Especially the limited ability of DNA repair as well as the frequent bottlenecks in endosymbiont population size due to their strictly vertical transmission is considered to explain the observed accumulation of (slightly) deleterious mutations in these bacteria. Thus, primary endosymbionts are under permanent threat to lose traits important for their own or their host's fitness (Moran, 1996). This is supported by the observation that in some cases, primary endosymbionts have been or are being replaced by secondary endosymbionts which may have taken over the biological role of a primary endosymbiont, since due to ongoing erosion of its genetic material, the latter could not satisfy the host's demands anymore (see for example Heddi et al., 1998; Pérez-Brocal et al., 2006; Conord et al., 2008; Lamelas et al., 2011; Koga et al., 2013; Toju et al., 2013). Thus, over evolutionary timescales, these very successful and extremely stable mutualistic interactions may become a dead end for the endosymbiotic bacteria.

While the important biological roles of endosymbionts to their hosts have repeatedly been shown experimentally or inferred from genome studies (Douglas, 1998; Zientz et al., 2004; Baumann, 2005; Feldhaar et al., 2007; Pais et al., 2008; Akman Gündüz & Douglas, 2009; Vogel & Moran, 2011), we are only beginning to understand how endosymbiotic bacteria are stably integrated into insect hosts and how interkingdom crosstalk between the insect hosts and the bacteria they harbour is mediated mechanistically. For example, recent results obtained with tsetse flies, weevils and carpenter ants show that from the host side, maintenance of stable chronic infections with endosymbionts may require specific functions of the insect immune system (Anselme et al., 2008; Wang et al., 2009; Login et al., 2011; Weiss et al., 2012; Ratzka et al., 2013). On the bacterial side, a protein that appears to be a key player in maintaining long-lasting mutualistic bacteria–host interactions is the chaperonin GroEL. GroEL has received increasing attention in the past years as GroEL proteins revealed surprising functions partially unrelated to chaperone activity. Here, we review recent findings on the importance of this chaperonin in bacteria–insect interactions (Fig. 1).

Figure 1.

Proposed roles of the GroEL chaperonin in microorganism–insect interactions, which in part are independent of chaperone function. The ATP-binding bacterial GroEL proteins assemble into a multimeric structure composed of two superposed heptameric rings which, together with the co-chaperonin GroES, build up the functional protein-folding machine (Yebenes et al., 2011).

GroEL as an essential symbiosis factor in bacteria–insect interactions

More than 30 years ago, it was discovered that the primary aphid endosymbiont Buchnera aphidicola massively expresses a single 60-kDa protein which at the time was termed symbionin (Kakeda & Ishikawa, 1991). This protein was then identified to be the chaperonin GroEL. In fact, GroEL may constitute up to 10% of cellular protein of B. aphidicola (Baumann et al., 1996; Poliakov et al., 2011). To date, GroEL overexpression appears to be a general and important feature of primary endosymbiosis in insects. For instance, in the primary endosymbiont Sodalis pierantonius str. SOPE of the weevil Sitophilus oryzae (Oakeson et al., 2014), GroEL is the most highly expressed protein (Charles et al., 1997b). It is also the most abundant protein in the midgut preparations including the bacteriome of the tsetse fly Glossina morsitans harbouring the endosymbiont Wigglesworthia (Aksoy, 1995; Haines et al., 2002), and about 6% of cellular protein of Blochmannia, the primary endosymbiont of ants of the genus Camponotus, was attributed to GroEL (Fan & Wernegreen, 2013). Accordingly, the groEL and groES genes are the most strongly expressed genes found in transcriptomes of Buchnera and Blochmannia (Wilcox et al., 2003; Stoll et al., 2009a). Also in endosymbionts with extremely reduced genomes of below 250 Kb, such as Sulcia muelleri, groEL was found to be expressed at an extremely high level (McCutcheon et al., 2009). Likewise, all other primary endosymbionts with the most reduced genomes of below 250 Kb including Zinderia, Carsonella, Tremblaya, Hodgkinia and Nasuia still encode a groEL homologue in spite of the lack of many genes considered to be absolutely essential for bacterial life (McCutcheon & Moran, 2011; Bennett & Moran, 2013).

Overall, the strong and apparently constitutive overexpression of groEL in primary endosymbionts with reduced genomes suggests that the chaperonin is permanently required in high concentrations to assist in the folding of conformationally damaged proteins and thus mitigates the negative effects of deleterious mutations occurring due to genome erosion (Fares et al., 2002b, 2004). In genomes of primary endosymbionts, groEL is among the most conserved genes. While most primary endosymbionts such as the gammaproteobacteria Blochmannia or Buchnera show a strong bias towards adenine and thymine in most genes and typically have an overall G–C content of between 13% and 30% (McCutcheon & Moran, 2011), groEL and other highly expressed genes are among those that are least biased and still have a relatively high G–C content (Palacios & Wernegreen, 2002; Stoll et al., 2009a). In addition, most sites of the groEL gene of primary endosymbionts show signs of purifying selection, while a few sites – especially those involved in polypeptide and GroES binding – are under positive selection (Fares et al., 2002a, 2005). The proposed function of GroEL in protein stabilization is in line with the recent characterization of the Buchnera anthranilate synthase TrpE after heterologous expression in Escherichia coli. Anthranilate synthase activity in the heterologous host was found to be poor but could be strongly enhanced by overexpression of the E. coli chaperonin, suggesting that the endosymbiont-derived enzyme requires assistance by GroEL for structural stabilization (Huang et al., 2008). Thus, the chaperonin GroEL appears to be a major and indispensable guarantor for the long-lasting and extremely stable relationships between primary endosymbionts and insects.

Under heat stress, many proteins tend to lose their proper conformation leading to loss of function or even aggregation of denatured proteins. As a protective response, many organisms are able to counteract this dangerous situation by overproduction of molecular chaperones (Guisbert et al., 2008). Accordingly, many molecular chaperones are also referred to as heat-shock proteins (HSPs) including the GroEL and GroES proteins. However, it should be mentioned that also other stress conditions including salt or ethanol stress are known to influence groEL expression in some bacteria (Laport et al., 2006; Camarena et al., 2010). A specific sigma factor, RpoH, is recognizing a characteristic consensus promoter motif and is triggering the heat-shock response in many bacteria (Guisbert et al., 2008). In spite of the observed dramatic genome size reduction, many primary endosymbionts still encode this alternative sigma factor raising the question whether a heat-shock response is still occurring in these bacteria. The primary endosymbiont S. pierantonius str. SOPE is able to respond to elevated temperature since induction of the groEL gene was observed under heat-shock conditions (Charles et al., 1997b). In contrast, despite the presence of RpoH, in Blochmannia and Buchnera, no typical heat-shock response is mounted any longer and expression of the groEL gene is not or relatively mildly influenced by high temperature (Wilcox et al., 2003; Stoll et al., 2009b). In fact, in several cases, it was shown that primary endosymbionts can be eliminated by rearing the respective animals at elevated temperature for some time (Ohtaka & Ishikawa, 1991; Sacchi et al., 1993; Heddi et al., 1999; Montllor et al., 2002; Fan & Wernegreen, 2013), and consequences of climate change for endosymbiont-bearing insect populations are discussed (Wernegreen, 2012). Thus, the alternative sigma factor may have been retained in the small genomes mainly because it is required for efficient expression of the groESL operon under ‘normal’ environmental conditions. Sodalis pierantonius str. SOPE appears to be an exception from this rule, but recent investigations indicate that its association with the insect host is quite young as compared to the other endosymbionts and it appears to be a paradigm for early genome evolution in primary endosymbionts (Lefevre et al., 2004; Conord et al., 2008). Accordingly, S. pierantonius str. SOPE still has typical features of free-living bacteria with a relatively large genome of about 4.5 Mb which, however, shows first characteristic signs for ongoing reductive genome evolution such as a very high number of mobile genetic elements involved in genome rearrangements and a multitude of pseudogenes (Charles et al., 1997a; Gil et al., 2008; Oakeson et al., 2014).

In the endosymbionts with the most reduced genomes with less than 250 Kb, even the rpoH gene was lost. These bacteria only retain the single vegetative sigma factor RpoD, which in these microorganisms must have taken over the task to express the groELS operon. This is in line with the observation that in Buchnera and Blochmannia, both possessing relatively large genomes between 650 and 800 Kb, there appears to be a tendency to replace RpoH-dependent promoters by RpoD-dependent promoters (Wilcox et al., 2003; Stoll et al., 2009b). Thus, while the groESL operon was retained in all primary endosymbionts, the transcription machinery required for its expression appears to be changing with increasing genome reduction. Thus, it can be assumed that a heat-shock response is not of relevance anymore for many primary insect endosymbionts. The heat-shock response may not be under strong selection possibly because most host insects are to a certain extent able to actively escape dangerously hot environments by moving to more suitable sites. In addition, a recent study suggests that also secondary endosymbionts may contribute to heat tolerance of the insect host possibly by the provision of protective metabolites (Burke et al., 2010).

GroEL as a target of the insect immune system

Insects entirely rely on an innate immune system to defeat pathogen attack. This system consists of a cellular arm mainly involving haemocytes and a humoral arm mainly characterized by the production of antimicrobial peptides (AMPs) and other effectors. Pathogenic microorganisms are recognized by so-called pattern recognition receptors (PRRs) that perceive microorganism-associated molecular patterns (MAMPs) such as bacterial peptidoglycan fragments and activate signal transduction pathways inducing immune functions (Vallet-Gely et al., 2008). In the case of endosymbiont-bearing insects, the immune system must be able to tolerate the permanent chronic infection by these bacteria, although it may also need to inhibit the spreading of the symbionts into other ‘forbidden’ tissues. MAMPs recognized by PRRs are present in both the endosymbionts but also in pathogenic bacteria. In fact, primary endosymbionts are still recognized by the immune system as shown for instance for Buchnera, Blochmannia and Sodalis endosymbionts in aphids, ants and weevils, respectively (Anselme et al., 2008; Douglas et al., 2011; Ratzka et al., 2011). Thus, a fundamental question is how the immune system of the insect is able to defeat bacterial pathogens, while it allows the mutualistic bacteria to thrive successfully within the host despite the fact that they are recognized as nonself by the immune system (for recent reviews see Ratzka et al., 2012; Login & Heddi, 2013).

Recently, important observations regarding endosymbiont control were made in the weevil Sitophilus zeamais which harbours the primary endosymbiont S. pierantonius str. SZPE (Login et al., 2011). Previous transcriptome analysis revealed a strong and comprehensive immune response when isolated SZPE was injected into the weevil's hemolymph. Thus, despite its long co-evolutionary history with its host, outside of the bacteriocyte tissue, the endosymbiont is recognized as foreign triggering a significant induction of several antimicrobial peptides (AMPs). In contrast, within the bacteriocytes where Sodalis resides, only a single AMP-encoding gene was found to be strongly expressed. This gene codes for the AMP coleoptericin-A (ColA) with antimicrobial activity against Gram-negative and Gram-positive bacteria (Anselme et al., 2008; Login et al., 2011). Interestingly, immunohistochemical analysis with endosymbiont-cured insects demonstrated constitutive expression of ColA in epithelial cells surrounding the intestine and in the fat body. Furthermore, within endosymbiont-bearing weevils, ColA is expressed in all tissues harbouring the endosymbionts including the thin layer of follicular cells surrounding the oocytes. This indicates a role of the AMP in prevention of pathogen infestation but also in confining the endosymbionts within the bacteriocyte tissue and to oocytes (Login et al., 2011).

To investigate a possible role of ColA in endosymbiont control, far-Western blotting experiments have been performed which demonstrated the interaction of ColA with the outer membrane proteins OmpA and OmpC of S. pierantonius str. SZPE. Surprisingly, it was also found to interact with intracellular GroEL protein (Login et al., 2011). The interaction of ColA with GroEL may lead to interference with cell division of the endosymbiont as it has already been shown that temperature-sensitive groEL mutations in E. coli trigger cell elongation (Ogino et al., 2004). In fact, Sodalis cells inside bacteriocytes are very long (up to 200 μm) and highly polyploid. The degree of polyploidy is positively correlated with bacterial size also suggesting a disturbance in cell division (Login et al., 2011; Login & Heddi, 2013). In E. coli, GroEL was primarily found in central regions of the cell overlapping with the FtsZ-ring. Although FtsZ-ring formation is not dependent on GroEL, the cell division machinery is unstable in the absence of the chaperonin. Accordingly, it was suggested that GroEL is either needed for recruitment of cell division proteins or it may act as a stabilizer of the FtsZ-ring, enabling efficient polymerization of the molecular cell division machinery (Ogino et al., 2004). The involvement of GroEL in cell division is further supported by the fact that the folding of an important component of the cell division machinery, FtsE, is strictly dependent on GroEL (Fujiwara & Taguchi, 2007). Thus, it could be speculated that ColA–GroEL interaction may lead to misfolding of FtsE and an incomplete septal FtsZ-ring. Interestingly, ColA does not interact with the human mitochondrial GroEL homologue and chaperonins of the Cpn70 family including bacterial DnaK. This specificity of interaction may be important as the host cells are not affected despite the massive presence of ColA in the tissue and within the bacteriocytes (Login & Heddi, 2013). So far, it is not known whether similar control mechanisms are also used in the case of other primary endosymbionts of insects. However, in those endosymbionts with the most reduced genomes, even the ftsZ gene has been lost, and due to a strongly reduced or lost ability to build a cell wall, there is a tendency to have a coccoid or pleomorphic shape. Thus, future work must show whether GroEL is also an immune target in other endosymbionts or whether the data reported here are a specific adaptation in the weevil–Sodalis symbiosis.

The fact that an AMP targets a chaperonin is not unique. For example, the proline-rich peptides Drosocin (Drosophila melanogaster), Pyrrhocoricin (Pyrrhocoris apterus) and Apidaecin IA (Apis mellifera) specifically bind to DnaK and unspecifically to GroEL (Otvos et al., 2000). Binding of Pyrrhocoricin to DnaK results in killing of bacteria (Kragol et al., 2002). Moreover, it has been shown that incubation of E. coli with Apidaecin IB inhibits bacterial growth and a concomitant decrease in GroEL–GroES production was noted. Growth inhibition could be alleviated by GroEL–GroES overexpression pointing to an interference of the AMP with the chaperone system (Zhou & Chen, 2011). Like ColA, these short proline-rich AMPs do not bind to the human DnaK equivalent Hsp70. Thus, bacterial chaperonins may be interesting targets for novel antimicrobial compounds (Czihal et al., 2012; Johnson et al., 2014).

GroEL as an important virulence factor of entomopathogenic bacteria

In addition to the well-documented function as a molecular chaperone, GroEL proteins are increasingly recognized to exhibit surprising additional so-called ‘moonlighting’ functions (Jeffery, 1999; Henderson et al., 2013). Perhaps the most astonishing feature of members of the GroEL family is that they can have toxic activities against eukaryotes.

Antlions are predators that are able to quickly paralyse their insect prey before they suck out their body fluids. For this purpose, the saliva of the larvae of the antlion Myrmeleon bore contains insecticidal factors that are at least in part produced by endosymbiotic bacteria such as Enterobacter aerogenes and Bacillus cereus (Yoshida et al., 2001; Nishiwaki et al., 2004). Surprisingly, in cultures of an E. aerogenes strain derived from M. bore, a protein was found which rapidly paralysed and killed the German cockroach Blattella germanica when injected in the insect haemocoel. This protein was identified as GroEL (Yoshida et al., 2001). Purified Enterobacter GroEL expressed in E. coli also exhibited this toxic effect on cockroaches, while wild-type E. coli GroEL did not harm the insects, despite the high sequence conservation between the two proteins. Several amino acid residues of the Enterobacter GroEL were found to be involved in toxicity (Val 100, Asn 101, Asp 338, Ala 471), and the substitution of these amino acids in the E. coli GroEL protein also conferred toxicity to the E. coli protein. So far, nothing is known about the mechanism underlying toxicity; however, neither GroES nor ATP binding is required for GroEL toxicity (Yoshida et al., 2001).

The insect pathogen Xenorhabdus nematophila is an endosymbiont of entomopathogenic nematodes of the genus Steinernema residing in the gut of insects (Herbert & Goodrich-Blair, 2007). The nematode is either found in insect hosts or in the soil during its infective juvenile stage waiting to infect a new host. After entering an insect, it is able to reach the haemocoel where it releases the bacteria. The bacteria multiply and release factors that contribute to killing of the host insect thus allowing the worm to complete its developmental cycle. In broth culture, it was found that X. nematophila releases outer membrane vesicles (OMVs) containing insecticidal factors including a GroEL protein (XnGroEL) as a major constituent. Interestingly, the co-chaperonin XnGroES was not found in the OMVs, suggesting a specific export pathway for XnGroEL. How XnGroEL is secreted is not known, but GroEL secretion was previously observed in other bacteria including Helicobacter pylori and Bacillus subtilis (Yang et al., 2011; Gonzales-Lopez et al., 2013). The purified XnGroEL was insecticidal to larvae of the cotton bollworm Helicoverpa armigera after oral administration (Joshi et al., 2008). The XnGroEL but not its E. coli homologue was found to bind to brush border membrane vesicles derived from the larval gut, and binding could be inhibited either by sugars such as N-acetyl-glucosamine and chito-oligosaccharides, or by chitinase treatment of the vesicles. This suggests that the chitinous peritrophic membrane lining the insect gut may be a primary target for binding of XnGroEL. The peritrophic membrane is an important part of the passive host defence system and protects the underlying gut epithelium from toxic compounds and bacterial infestation. Mutation analysis revealed that two amino acids of XnGroEL (Thr 347 and Ser 356) were critical for binding and toxicity of XnGroEL. XnGroEL toxicity is uncoupled from its folding activity, as it does not require the co-chaperonin XnGroES. Meanwhile, insecticidal activities of GroEL homologues of other Xenorhabdus species (X. budapestensis and X. ehlersii) towards the greater wax moth Galleria mellonella were described (Shi et al., 2012; Yang et al., 2012; Yang et al., 2012). Thus, in these entomopathogenic bacteria, GroEL has evolved as an important virulence factor. In the future, the toxicity of XnGroEL may potentially be exploited in agricultural pest management, as its expression in transgenic tobacco plants conferred significant resistance against attack by the cotton bollworm H. armigera (Kumari et al., 2013).

It is interesting to note that the GroEL protein of Sodalis glossinidius, a secondary endosymbiont of the tsetse fly, contains all four amino acid residues crucial for toxicity of the above-mentioned Enterobacter protein, and the GroEL proteins of some primary endosymbionts such as Wigglesworthia and S. pierantonius contain three of the four residues (Haines et al., 2002). So far, no experimental data are available, but this raises the question, whether also endosymbiotic bacteria may possess some toxic potential towards their host at least at certain stages of their interaction. In fact, S. glossinidius encodes and expresses a type III secretion system probably involved in cell invasion by injection of effector proteins into the host cell (Dale et al., 2002), and many insect mutualists are closely related to pathogenic bacteria and have a common evolutionary history indicating a pathogenic relationship at the onset of symbiosis at least for some mutualistic interactions (Goebel & Gross, 2001; Sachs et al., 2013).

GroEL as a transmission factor for plant viruses in insect vectors

Many herbivorous insects are not only themselves pest species, but they can also be vectors for plant viruses. When feeding on plants, they can ingest viruses circulating in the phloem and can subsequently infect plants with viruses found in their saliva. So-called persistent circulative viruses, that is, viruses that after ingestion move about the insect body, are able to traverse the gut wall and to enter the hemolymph. From there, they reach the salivary glands waiting for transmission to a new host plant via saliva secretion during feeding (Hogenhout et al., 2008; Blanc et al., 2011). For several such persistent viruses, GroEL of endosymbiotic bacteria harboured by the host insects appears to play an important role during their infection cycle.

In 1994, van den Heuvel and co-workers first described that a bacterial GroEL protein from the aphid Myzus persicae, possibly derived from the primary endosymbiont B. aphidicola and apparently present also in body parts outside of the bacteriome tissue including the hemolymph of the insect, was associated with purified virus particles of the potato leafroll virus (PLRV), a single-stranded RNA luteovirus (Van den Heuvel et al., 1994). Reduction of the number of endosymbiotic bacteria by feeding the insect host antibiotics drastically reduced virus transmission, indicating a role of GroEL for persistence of the virus in the hemolymph of the aphid possibly by protection of the virus particle against proteolytic attack. Specific in vitro interaction of Buchnera GroEL of the aphid Rhopalosiphum padi but not of E. coli GroEL with another luteovirus, the barley yellow dwarf virus (BYDV), was reported subsequently (Filichkin et al., 1997). However, in a recent investigation, the binding site of BYDV to Buchnera GroEL from two aphid species was characterized. For this purpose, a GroEL peptide library covering the entire GroEL protein was tested for binding of virus particles. The identified binding region of the virus on GroEL was found to be localized in a region that is known to be important for GroEL oligomer formation. Thus, in the presence of virus particles, GroEL assembly may be severely impaired (Bouvaine et al., 2011). Moreover, the presence of Buchnera GroEL in various body parts of aphids was investigated by use of a specific monoclonal antibody. Unexpectedly and in contrast to previous results, GroEL could neither be detected in the hemolymph nor in the gut or fat body (Bouvaine et al., 2011). Taken together, these data are not consistent with the original findings about a role of Buchnera GroEL in stabilizing virus particles during their circulation through the aphid body, and first evidence about the presence of Buchnera GroEL in the hemolymph may have been due to preparation artefacts (Bouvaine et al., 2011). Thus, the role of Buchnera GroEL in plant virus transmission by aphids is still under discussion. Interestingly, recent data indicate that additional Buchnera factors may directly or indirectly be involved in vector competence of aphids (Cilia et al., 2011).

Recently, a GroEL homologue from endosymbiotic bacteria of a whitefly (Bemisia tabaci) population in Israel was found to affect transmission of tomato yellow leaf curl virus (TYLCV), a member of the Geminiviridae family, and this GroEL protein was found to be present in the hemolymph of the whiteflies. In fact, whiteflies treated with antibiotics or fed with anti-GroEL antibodies showed a strongly reduced transmission rate of the virus (Morin et al., 1999, 2000). Whiteflies may harbour a multitude of secondary endosymbionts including Arsenophonus, Cardinium, Hamiltonella, Wolbachia and Rickettsia in addition to their obligate primary endosymbiont Portiera (Gottlieb et al., 2008). The composition of endosymbiont populations was found to correlate with different whitefly biotypes and with their performance in virus transmission. Based on in vitro and in vivo evidence, only GroEL derived from Hamiltonella but not of the other endosymbionts was shown to be involved in TYLCV transmission (Morin et al., 1999; Gottlieb et al., 2010; Su et al., 2013). Whitefly isolates from a location in India which in addition to the primary endosymbiont Portiera harbour only a single secondary endosymbiont, a member of the genus Arsenophonus, and transmit cotton leaf curl virus (CLCuV). Again, only the GroEL protein of the secondary endosymbiont Arsenophonus but not of the primary endosymbiont was involved in virus transmission (Rana et al., 2012). Thus, there is increasing evidence that GroEL proteins of insect endosymbionts in fact can contribute to transmission of certain circulative plant viruses.

Interestingly, the specific interaction between some endosymbiotic GroEL proteins and certain viruses could be exploited successfully to reduce the susceptibility of plants for the respective viruses. Two recent reports show that transgenic plants expressing a GroEL protein derived from endosymbionts of the whitefly B. tabaci exhibit a strongly increased tolerance towards viruses such as TYLCV and cucumber mosaic virus (CMV), but not towards viruses unable to interact with GroEL in vitro. Expression of bacterial GroEL in these plants led to a strong reduction in virus load possibly by trapping the viruses (Akad et al., 2007; Edelbaum et al., 2009).

Conclusions

The diverse roles of GroEL proteins in bacteria–insect interactions reviewed here range from classical chaperone function, absolutely essential in particular for obligate intracellular endosymbionts, to surprising additional ‘moonlighting’ functions unrelated to chaperone activity such as insecticidal activity. So far, very little is known about structure–function relationships of GroEL proteins exhibiting activities apart from chaperone function. Similarly, secretion of GroEL proteins by diverse Gram-negative and Gram-positive bacteria is increasingly reported, but so far, no clues about the secretion mechanisms involved are available. Thus, the molecular characterization of these unexpected talents of GroEL proteins is an important task for future research and will certainly lead to fascinating new insights into evolution of protein function as well as the role of GroEL in mediation interactions of bacteria with other organisms. Finally, some recent investigations indicate that molecular chaperones may be interesting candidates for the identification of novel antimicrobial compounds useful in medicine and for the design of novel antimicrobial strategies in agriculture.

Acknowledgements

Work in the author's laboratory is supported by a grant of the Deutsche Forschungsgemeinschaft (DFG Gr1243/8-1). The authors would like to thank Dagmar Beier for critically reading the manuscript.

Ancillary