Insect blood (hemolymph) contains prophenoloxidase, a proenzyme that is activated to protective phenoloxidase when the insect is damaged or challenged with microorganisms. The Gram-negative bacterium Photorhabdus luminescens kills the lepidopteron insect Manduca sexta by using a variety of toxins. We screened P. luminescens and Photorhabdus asymbiotica cosmid libraries in an Escherichia coli host against previously activated M. sexta hemolymph phenoloxidase and identified three overlapping cosmid clones from P. luminescens and five from P. asymbiotica that suppressed the activity of the enzyme both in vitro and in vivo. Genome alignments of cosmid end sequences from both species confirmed that they contained orthologous loci. We examined one of the cosmids from P. luminescens in detail: it induced the formation of significantly fewer melanotic nodules, proliferated faster within the insect host and was significantly more virulent towards fifth-stage larvae than E. coli control bacteria. Insertional mutagenesis of this cosmid yielded 11 transposon mutants that were no longer inhibitory. All of these were insertions into a single 5.5-kb locus, which contained three ORFs and was homologous to the maltodextrin phosphorylase locus of E. coli. The implications of this novel inhibitory factor of insect phenoloxidase for Photorhabdus virulence are discussed.
Insects mount multiple immune reactions to the presence of microorganisms, including both cellular and humoral responses (Lemaitre & Hoffmann, 2007). Typical cellular responses include phagocytosis and the formation of multicellular hemocyte nodules and capsules (Lavine & Strand, 2002), while humoral responses include the secretion of proteins with both recognition and antimicrobial functions (Uvell & Engström, 2007). An important component of both arms of the insect immune response is the immune-initiated synthesis of melanin by phenoloxidase (Cerenius et al., 2008). Activated phenoloxidase is produced by a serine protease cascade that is initiated upon recognition of invading microorganisms, leading to proteolytic cleavage of its precursor prophenoloxidase (Kanost et al., 2004), which is present in the hemolymph plasma. In addition to causing synthesis of the pigment melanin, prophenoloxidase activation leads to the release of microbicidal reactive intermediates (Zhao et al., 2007) and the production of melanotic nodules around invading microorganisms (Gillespie et al., 1997).
Pathogens have evolved diverse strategies to evade host immune defenses (Hornef et al., 2002; Bhavsar et al., 2007). Photorhabdus luminescens are Gram-negative insect pathogenic bacteria found in association with their specific entomopathogenic nematode vector that lives in the soil. The bacteria are released directly from the gut of invading infective juvenile-stage nematodes into the host insect's body cavity where they proliferate and colonize specific tissues within the insect such as the midgut and fat body (ffrench-Constant et al., 2003). The insect's immune system recognizes the presence of the bacteria and mounts antibacterial responses that slow the progress of the infection, but ultimately fail to prevent the death of the host (Eleftherianos et al., 2006).
Although several toxins encoded in Photorhabdus genomes have been identified and characterized (ffrench-Constant et al., 2007; Waterfield et al., 2008), the mechanisms used by the bacteria to suppress host immune defenses and thereby persist in their host are not fully elucidated. Interestingly, a characteristic of insects killed by Photorhabdus is that unlike insects killed by other means their hemolymph does not blacken (ffrench-Constant et al., 2003). We have previously shown that Photorhabdus infection is associated with the presence in hemolymph of a small molecule phenoloxidase inhibitor (Eleftherianos et al., 2007).
The aim of the present work was to identify the Photorhabdus phenoloxidase inhibitor. We performed in vitro screens of Photorhabdus asymbiotica and P. luminescens cosmid libraries and identified overlapping cosmid clones from both species that contained orthologous loci that suppressed previously activated phenoloxidase from Manduca sexta hemolymph plasma. Further characterization of one representative Escherichia coli cosmid clone from the P. luminescens library revealed that it also suppressed nodule formation, persisted longer within insects and showed increased pathogenicity towards M. sexta larvae. We mapped the genes responsible for these effects on this cosmid using insertional mutagenesis. The potential role of the gene products as a novel phenoloxidase inhibitor molecule during insect infection by P. luminescens is discussed.
Materials and methods
Tobacco hornworm, M. sexta (Lepidoptera: Sphingidae), larvae were kept individually on a wheat germ-based artificial diet at 25 °C and a 17 h light : 7 h dark photoperiod (Reynolds et al., 1985). Newly moulted (day 0) fifth-stage larvae were used for all experiments. Escherichia coli strain EC100 and cosmid clones of P. asymbiotica ATCC 43949 and P. luminescens ssp. laumondii TT01 were grown for 24 h at 37 °C in 5 mL of 2.5% Luria–Bertani liquid broth (LB; 1% tryptone, 0.5% yeast extract and 1% sodium chloride, pH 7.0) with constant shaking, and 1.5% agar (Difco Laboratories, Surrey, UK) was added as required. Relevant antibiotics were supplemented as required by the specific plasmid selection.
Phenoloxidase activity of M. sexta hemolymph was quantified using a microplate enzyme assay (Hall et al., 1995). First, untreated day 0 fifth-stage larvae were chilled on ice for 15 min, surface sterilized with 70% ethanol, cut at the midpoint of the dorsal horn and bled into a prechilled sterile polypropylene tube. Hemolymph was then diluted in a 3 : 1 (v/v) ratio with a 50 mM phosphate-buffered saline (PBS; 0.15 M sodium chloride, 10 mM sodium phosphate buffer, pH 7.4) solution and kept on ice. Hemolymph plasma samples were obtained by centrifuging the hemocytes at 4000 g for 10 min 4 °C. All assays were carried out with a Molecular Devices Thermomax microplate reader with flat-bottom 96-well plates (Nunc, Texas). A reaction mixture containing 10 μL of diluted M. sexta hemolymph plasma, 115 μL of PBS buffer (50 mM) with 10% (w/v) cetylpyridinium chloride (Sigma, Gillingham, UK) in sterile-distilled water was left for 15 min at room temperature with constant shaking to allow the activation of the enzyme. Putative inhibitors (bacterial supernatants or cells) were prepared from bacterial cultures grown for 24 h by centrifugation of cells, which were washed in PBS; supernatants were sterile filtered through 0.2-μm filters (Millipore, UK). Then, 20 μL of bacterial supernatants or washed cells from the cosmid library were added to the reaction mixture, which was left for 30 min at room temperature with constant shaking to allow inhibition of the activated enzyme. Heated supernatants and heat-killed bacteria were prepared by heating at 90 °C for 10 min. Finally, 25 μL of 20 mM 4-methyl catechol (Sigma) was added to initiate the reaction and the final volume of the mixture was made up to 200 μL with sterile-distilled water. The change in absorbance was read at 490 nm for 1 h at room temperature, with a reading taken every 1 min. Each phenoloxidase assay was repeated at least three times.
Cosmid library and screening
The cosmid libraries were prepared from P. asymbiotica ATCC 43939 and P. luminescens strain TT01 genomic DNA by MWG Biotech (Ebersberg, Germany). DNA was physically sheared, size selected for fragments of c. 30 kb and then cloned into pWEB in E. coli strain EC100. For library screening, individual clones were grown for 24 h in 5 mL of LB with shaking. Bacterial supernatants or broths (20 μL containing c. 1 × 103 recombinant E. coli) for each Photorhabdus cosmid clone were then tested for in vitro inhibition of activated phenoloxidase. Control treatments involved supernatants or cells of E. coli EC100 bearing the empty vector pWEB. Three replicates per cosmid were tested and positive clones were rescreened three times. Candidate cosmid clones that inhibited phenoloxidase activity were further characterized by end sequencing.
Insects were injected with bacterial supernatants or cells washed in PBS. Suspensions of 20 μL containing c. 1 × 103E. coli/pWEB or E. coli/Pl_2BF12 cells were injected directly into the hemocoel (body cavity) of M. sexta larvae using a 100-μL disposable syringe with a 30-Gauge needle. Numbers of injected bacteria were estimated by OD600 nm and by plating a known volume of injected suspension on 2.5% LB agar plates. The injected larvae were kept at 28 °C to determine survival or bled 24 h later to collect hemolymph samples. Mortality, defined as an inability to react to poking with a needle, was scored at intervals up to 96 h following the bacterial challenge. Ten insects were used for each treatment and each insect survival bioassay was replicated three times.
Manduca sexta larvae were injected with E. coli/pWEB or E. coli/Pl_2BF12 cells, as above. Nodule formation was assessed 24 h after bacterial infection. Insects were immobilized on ice for 15 min before dissection under 1% (w/v) NaCl solution saturated with phenylthiourea (which prevented general postdissection melanization). Melanized, dark nodules within the hemocoel were observed using a stereomicroscope and an approximate manual count of the nodule number was made with the help of a tally counter. In general, it was easy to recognize individual nodules. Ten insects were used for each treatment and each experiment was repeated three times.
Counting bacteria in vivo
Manduca sexta larvae were injected with bacteria as before and 24 h later were chilled on ice and bled as described previously. A standard volume of hemolymph (500 μL) was taken from each caterpillar. Aliquots of hemolymph (50 μL) were immediately added to 450 μL of prechilled Grace's insect medium (Sigma). To determine the number of surviving bacteria, serial dilutions of the hemolymph were plated onto Petri dishes containing 2.5% LB and 1.5% agar and the number of E. coli/pWEB and E. coli/Pl_2BF12 CFU was recorded 24 h later. The average of two counts per insect of five insects per treatment was calculated. Bacterial survival assays were replicated three times.
Cosmid mutagenesis and sequencing
Insertional mutagenesis used the EZ∷TN<TET1> transposon (Epicentre, Madison). Transposon mutants of individual cosmids were rescreened to test for loss or retention of the phenoloxidase inhibition phenotype. Transposon mutants were also used as entry points in nucleotide sequencing of the positive Pl_2BF12 cosmid clone. Cosmid DNA was prepared on a RoboPrep plasmid preparation robot (MWG Biotech) and sequenced on an ABI3700 nucleotide sequencer (Applied Biosystems). Sequences were assembled using the lasergene software package (DNASTAR, Ebersberg, Germany).
Photorhabdus luminescens cosmid Pl__2BF12 suppresses activated insect phenoloxidase in vitro
Screening cosmid libraries of P. luminescens genomic DNA in E. coli for in vitro inhibition of activated M. sexta hemolymph phenoloxidase identified cosmid Pl_2BF12 and two other overlapping cosmids: Pl_13BB9 and Pl_2BB12, which also had similar effects on the activity of the enzyme (Fig. 4, top). The P. asymbiotica library screen identified five overlapping cosmids: Pa_2CG6, Pa_2CG9, Pa_1CF2, Pa_3BC3 and Pa_1AD2. Alignment of these cosmid end sequences revealed that the cosmids from the two species spanned orthologous regions of the two genomes. For clarity we focused our further research efforts on one representative cosmid clone: Pl_2BF12. Heated or nonheated filtered supernatants of this clone significantly reduced the in vitro levels of previously activated phenoloxidase from naïve fifth-stage M. sexta larvae compared with the same hemolymph incubated with supernatants from control E. coli bacteria carrying the empty vector pWEB (t-test, P<0.0001) (Fig. 1a). Similarly, addition of live or dead washed E. coli/Pl_2BF12 cells, but not E. coli/pWEB cells, to M. sexta cell-free hemolymph led to a significant reduction in phenoloxidase activity (t-test, P<0.0001) (Fig. 1b). There were no significant differences between heated and nonheated supernatants or live and heat-killed cells of the cosmid in terms of their ability to inhibit phenoloxidase (t-test, P>0.05) (Fig. 1a and b).
Photorhabdus luminescens cosmid Pl__2BF12 reduces in vivo phenoloxidase levels and nodulation in M. sexta
We then tested whether the Pl_2BF12 clone was able to suppress phenoloxidase levels in vivo. We found a significant decrease in phenoloxidase activity when insects were injected with supernatants or cells of the P. luminescens cosmid (t-test, P<0.001) (Fig. 2a). This reduction in hemolymph plasma phenoloxidase was reflected by a significant reduction in the number of melanotic nodules formed in tissues of E. coli/Pl_2BF12-infected larvae compared with tissues of E. coli/pWEB-infected individuals (t-test, P<0.05) (Fig. 2b).
Cosmid Pl__2BF12 allows E. coli to persist within and kill M. sexta
Here we looked for changes in survival between insects injected with E. coli containing the P. luminescens cosmid Pl_2BF12 and insects injected with control E. coli containing the empty vector pWEB. We found that there were significant differences in the number of survivors between Pl_2BF12-infected larvae and those challenged with control bacteria from 36 h (χ2 test, P<0.05) and up to 96 h after infection (χ2 test, P<0.001) (Fig. 3a). At the same time points, significantly larger numbers of live E. coli/Pl_2BF12 cells were recovered from infected caterpillars compared with E. coli/pWEB bacteria (one-way anova, P<0.0001), which were eliminated from the insects at 24 h after infection (Fig. 3b).
Sequence and mutagenesis of the P. luminescens cosmid Pl__2BF12 identified a region associated with phenoloxidase inhibition
To further elucidate which genes were responsible for the phenoloxidase inhibition, we focused on the representative P. luminescens cosmid clone Pl_2BF12. The full sequence of the Pl_2BF12 cosmid was derived using cosmid end sequences and the P. luminescens strain TT01 genome sequence. A region of minimum genetic overlap with the other two overlapping P. luminescens cosmid clones was determined in the same way and was shown to contain around 14 intact ORFs. A library of Pl_2BF12 insertion mutants was generated by in vitro mutagenesis, and defined by sequencing out from the transposons. Insertion mutagenesis of Pl_2BF12 led to loss of its ability to inhibit activated phenoloxidase in vitro when any one of three adjacent ORFs was disrupted (Fig. 4, bottom); disruption of other genes on the cosmid had no effect on the phenoloxidase inhibitory phenotype. This identified a c. 5.5-kb locus, homologous to the mal (maltodextrin phosphorylase) locus of E. coli and other Gram-negative bacteria. The MalP gene encodes maltodextrin phosphorylase, the malQ gene encodes amylomaltase, while gene malT encodes the regulator of the mal locus (Fig. 4, middle).
Melanization (darkening) is a normal defensive response of insect hemolymph to infection, being brought about by the activation of PPO to active phenoloxidase, resulting in the production of lethal reactive intermediates at the surface of invading microorganisms (Zhao et al., 2007). Further, phenoloxidase activity is associated with cellular reactions by hemocytes, which wrap around the bacteria to form melanized nodules, thereby isolating the pathogens (Gillespie et al., 1997). The hemolymph of insects infected by the insect pathogenic bacterium P. luminescens has previously been shown to have a reduced ability to melanize (darken) upon bleeding (ffrench-Constant et al., 2003). Thus, inhibition of phenoloxidase is likely to be an adaptation of this bacterium to its pathogenic lifestyle.
Here we identify a cosmid clone Pl_2BF12 from P. luminescens that reduces the levels of previously activated M. sexta hemolymph phenoloxidase in vitro compared with control E. coli bacteria. We also show that cosmid Pl_2BF12 is able to suppress phenoloxidase in vivo and to prevent the formation of melanotic nodules in insect tissues at early time points after infection. We further demonstrate that phenoloxidase inhibition allows a standard laboratory strain of E. coli carrying the P. luminescens cosmid to persist within the insect and eventually to kill it.
Our current results are validated by the fact that screening a cosmid library from the related species P. asymbiotica ssp. asymbiotica in the same way identified five cosmid clones that also inhibited activated M. sexta phenoloxidase in vitro. Interestingly, all five cosmids spanned the equivalent mal gene homologues in P. asymbiotica (our unpublished data). Furthermore, we have recently found that P. luminescens produces a small-molecule antibiotic (3,5-dihydroxy-4-isopropylstilbene) that also acts as a phenoloxidase inhibitor in M. sexta (Eleftherianos et al., 2007). These findings suggest that Photorhabdus bacteria not only possess a large number of genes that contribute to pathogenicity (ffrench-Constant et al., 2007), but can also use multiple means to overcome insect immune defenses by suppressing host phenoloxidase activity.
The importance of the phenoloxidase mechanism in insect resistance to bacterial infections has been documented previously on numerous occasions (Cerenius et al., 2008). Other nematode-associated bacterial pathogens of insects have also been shown to inhibit phenoloxidase (Yokoo et al., 1992). However, it needs to be stressed that all the experiments performed here were carried out in the absence of the nematode symbiont. Although in natural vector-mediated infections the nematode may contribute additionally to insect immune suppression by targeting the phenoloxidase system, this possibility remains to be investigated. It is also worth noting that suppression of the phenoloxidase pathway to evade insect immune responses is a strategy adopted by insect-specific pathogens other than bacteria. For example, it has recently been reported that Microplitis demolitor bracovirus carried by the wasp M. demolitor produces a protein, Egf1.0, which inhibits the phenoloxidase cascade (Beck & Strand, 2007).
However, there are still basic issues that require clarification concerning the mode and site of action of the P. luminescens phenoloxidase inhibitor. We find that the inhibitory factor in Pl_2BF12 supernatants is heat stable, which implies that it is a nonproteinaceous small molecule, and that it is probably present on the outer membrane surface of bacteria because washed cells retained inhibitory activity.
The P. luminescens mal locus that is responsible for the phenoloxidase inhibition in the cosmid encodes two glycoproteins maltodextrin phosphorylase (MalP) and amylomaltase (MalQ) and the transporter/regulator, MalT. MalP catalyzes the phosphorolysis of an α-1,4-glycosidic bond in maltodextrins, removing the nonreducing glucosyl residues of linear oligosaccharides as glucose-1-phosphate (Glc1P). MalQ is a 4-α-glucanotransferase that acts to release glucose from maltodextrins. MalT is essential for the expression of all maltose-inducible functions and is a purely positive regulator that is activated by an inducer and stimulates transcription by activating RNA polymerase (Boos & Shuman, 1998).
It is possible therefore that the phenoloxidase inhibitor elaborated by Pl_2BF12 may in fact be maltodexin polymers or their derivatives. However, it is interesting to note that we tested commercially available Glc1P in the phenoloxidase inhibition assay, but observed only a slight inhibition at very high concentrations (data not shown). An alternative hypothesis is that the Photorhabdus mal operon induces the expression of native E. coli genes, which produce a cryptic phenoloxidase inhibitor. The MalT protein requires maltotriose and ATP as ligands for binding to a dodecanucleotide MalT box that appears in multiple copies upstream of all mal promoters (Boos & Shuman, 1998).
Interestingly, a recent report concerns the isolation of a low-molecular-weight glycoprotein from M. sexta with strong inhibitory activity against M. sexta phenoloxidase and mushroom tyrosinase (Lu & Jiang, 2007). The authors indicate that removal of the carbohydrate moiety from the endogenous inhibitor led to reduced phenoloxidase activity, which implies that glycosylation enhances the interaction between phenoloxidase and the inhibitor compound. It is therefore feasible that the Photorhabdus mal gene products are leading to the glycosylation of an E. coli protein or some component of the phenoloxidase assay system (or a hemolymph component) that reduces phenoloxidase activity. Lastly, it has been reported recently that the metalloprotease PrtS isolated from concentrated 72-h supernatants of P. luminescens TT01 induces a melanization response in M. sexta on injection (Held et al., 2007). It is not clear how this relates to the suppression of melanization that we and others have noted during P. luminescens infection, although it is possible that nonlocalized inappropriate activation of phenoloxidase is itself an advantage to the bacterium. It is also possible that secretion of a phenoloxidase inhibitor may override the activation response to PrtS or that activation and inhibition are temporally separated. It is nevertheless evident from the work reported here that the pathogen is able to manipulate host phenoloxidase in order to successfully promote its own growth and survival within the host.
In conclusion, our present findings further highlight the significance of phenoloxidase inhibitors produced during infection by specialist insect pathogens, such as Photorhabdus, and may be generally relevant in understanding pathogenesis in insects. It is now our aim to isolate and identify biochemically the inhibitory factor from Photorhabdus culture supernatants, perform structural and activity comparisons with other known phenoloxidase inhibitors and study in detail its role in the overall insect immune response and potential interaction with other humoral and/or cellular defense mechanisms.
We thank Sandra Barnes for rearing the insects. This project was supported by a grant to S.E.R. and R.H.ff.-C. from the Exploiting Genomics Initiative of the UK Biotechnology and Biological Sciences Research Council.