Induced expression of the Serratia entomophila Sep proteins shows activity towards the larvae of the New Zealand grass grub Costelytra zealandica


  • Editor: Reggie Lo

Correspondence: Mark R.H. Hurst, Biocontrol and Biosecurity, AgResearch, Canterbury Agricultural and Science Centre, PO Box 60, Lincoln, New Zealand. Tel.: +64 03 3253919; fax: +64 03 3253946; e-mail:


Serratia entomophila and Serratia proteamaculans cause amber disease of the grass grub Costelytra zealandica (Coleoptera: Scarabaeidae). Three genes required for virulence, sepABC, are located on a large plasmid, pADAP. The translated products of the sep genes are members of the toxin complex (Tc) family of insecticidal toxins that reside in the genomes of some Enterobacteriaceae. Each of the sep genes was placed either singly or as various combinations under the control of an inducible arabinose promoter, allowing their inductive expression. Western Immunoblot confirmed that each of the Sep proteins migrated at their predicted size on sodium dodecyl sulphate-polyacrylamide gel electrophoresis gel. Bioassays of sonicated filtrates derived from the various arabinose-induced para-SEP constructs showed that only when sepA, sepB and sepC were coexpressed were amber disease symptoms observed in grass grub larvae. Fourteen days after ingestion of the Sep protein filtrate, ∼64% of the larvae reverted from a diseased to a healthy phenotype. Redosing the revertents with a fresh Sep protein filtrate reinitiated the amber pathotype, indicating that the Sep proteins are needed to be continuously present to exert an effect.


Serratia entomophila and Serratia proteamaculans (Enterobacteriaceae) are the causal agents of amber disease of the New Zealand grass grub, Costelytra zealandica (Coleoptera: Scarabaeidae). The phenology is distinct and host specific, with infected larvae ceasing to feed within 2–5 days of ingesting pathogenic cells. The gut, which is normally dark in colour, clears around this time (Jackson et al., 1993) and the level of the major gut digestive enzymes (trypsin and chymotrypsin) decreases sharply (Jackson, 1995). The clearance of the gut results in the host turning a characteristic amber colour. An infected larvae may remain in this state for a prolonged period (1–3 months) before bacteria eventually invade the haemocoel, resulting in rapid death of the insect.

Two regions of the 153-kb plasmid termed pADAP (amber disease-associated plasmid) have been identified as necessary for causing amber disease symptoms. The afpanti-feeding prophage gene cluster, which encodes an R-type pyocin structure theorized to mediate the transport of toxins to a target site, causes a cessation of feeding by the grass grub larvae (Hurst et al., 2004, 2007). The sep virulence-associated region comprises of three genes designated sepA, sepB and sepC for Serratia entomophila pathogenicity, that are responsible for the amber disease symptoms of gut clearance and amber coloration of the larvae (Hurst et al., 2000). The predicted translated products of the sepABC genes show significant sequence similarity to components of the insecticidal toxins produced by the nematode-associated bacteria Photorhabdus luminescens (Bowen et al., 1998) and Xenorhabdus nematophilus (Morgan et al., 2001). These toxins have been termed Tc toxins by Bowen et al. (1998) for toxin complex, as the three proteins combine to form a complex with insecticidal activity and have been recently reviewed by ffrench-Constant & Waterfield (2005). The completion of genome-sequencing projects such as Yersinia pestis C092 (Parkhill et al., 2001), Pseudomonas syringae pv. tomato DC3000 (Buell et al., 2003), among others, has also revealed the presence of putative insecticidal tc genes.

The sep virulence-associated region has been cloned and the clone (pBM32) is able to induce amber disease symptoms towards grass grub larvae, whether in an Escherichia coli or a pADAP-cured strain of S. entomophila (Hurst et al., 2000). To date, efforts to isolate the Sep proteins from either S. entomophila or the sep virulence-associated clone have been unsuccessful. Here, the construction of vectors allowing the inductive expression of the sep genes singly, or as various combinations is reported.

Materials and methods

Strains and plasmids

The strains and plasmids used in the study are listed in Table 1. Bacteria were grown in Luria–Bertani (LB) broth or on LB agar (Sambrook et al., 1989), at 37°C for E. coli and 30°C for S. entomophila and S. proteamaculans. For E. coli, carbinicillin and chloramphenicol were used at 100 and 30 μg mL−1, respectively, and for S. proteamaculans, carbinicillin and chloramphenicol were used at 400 and 90 μg mL−1, respectively. Where mentioned, arabinose was used at a final concentration of 0.2% for E. coli and 0.4% for S. proteamaculans.

Table 1.   Bacterial strains and plasmids used in the study
Strain or plasmidRelevant genotype or descriptionReferences
Escherichia coli
 DH10BFmcrAΔmrr-hsdRMS-mcrBCΦ80d lacZΔM15ΔlacX74endA1recA1deoara, leu 7697 araD139 galU galK nupG, rpsLorow & Jessee (1990)
 BL21(DE3)FompT hsdSB (rBmB) galdcm (DE3)Studier & Moffatt (1986)
Serratia entomophila
 A1MO2pADAP pathogenic, araGrimont et al. (1988)
Serratia proteamaculans
 4031Plasmid free, nonpathogenic, ara+
Field isolate from the AgResearch Insect Pathogen culture collection (Lincoln, New Zealand)
This study
 pADAPAmber disease associated plasmid (153 kb)Glare et al. (1993)
 pAY2-4Arabinose expression vector, AmRShaw et al. (2003)
 paraAArabinose inducible sepA gene, AmRThis study
 paraBArabinose inducible sepB gene, AmRThis study
 paraCArabinose inducible sepC gene, AmRThis study
 paraABArabinose inducible sepA and sepB genes, AmRThis study
 paraBCArabinose inducible sepB and sepC genes, AmR CmRThis study
 paraABCArabinose inducible sepA, sepB and sepC genes, AmR CmRThis study
 pACΔ10BamHI fragment derived from pBM32-10 cloned into pACYC184, CmRThis study
 pACΔ10BstXIspACΔ10 digested with BstXI and self ligated (omitting the sepA, and sepB genes), CmRThis study
 pACΔ11BamHI fragment derived from pBM32-11 cloned into pACYC184, CmRthis study
 pBM3223-kb BamHI fragment encoding the sep virulence associated region, AmRHurst et al. (2000)
 pBM32-10pBM32 containing mini-Tn10 insertion, AmR KnRHurst et al. (2000)
 pBM32-11pBM32 containing mini-Tn10 insertion, AmR KnRHurst et al. (2000)

DNA manipulation and sequencing

Standard DNA techniques were carried out as described by Sambrook et al. (1989). Plasmid DNA was transferred into S. proteamaculans by electroporation using a Biorad Gene Pulser (25 μF, 2.5 kV and 200 Ω) (Dower et al., 1988). Plasmid templates for DNA sequencing were prepared using the High Pure Plasmid Isolation Kit (Roche Diagnostics GmbH). The DNA was sequenced using a capillary ABI3730 Genetic Analyzer, from Applied Biosystems Inc. (

PCR was undertaken using Thermoprime Plus DNA polymerase (ABgene; Advanced Biotechnologies Ltd), 1.5 mM MgCl2, 0.2 mM each dNTP, 2 μM each primer and 1 μL of DNA in a final volume of 25 μL, adjusted with sterile distilled water. Template DNA was denatured with a preliminary step of 98°C for 2 min, then five PCR cycles of denaturing at 95°C for 15 s, annealing at 55°C for 15 s and elongating at 72°C for 90 s, followed by 30 cycles at 95°C for 15 s 50°C for 15 s, and 72°C for 90 s. The primer sets used were sepAF (5′-AAACATATGTATAATATTGATGATATTCTGG-3′)- sepAR (5′-AAAGAATTCGGCCAGCATGGAGGCAGGGTCC-3′), sepBF (5′-CCCATATGCAAAATCATCAAGACATGGCCATTACTGCC-3′) – sepBR (5′-GACTACCCTCCGTACGTGAACGG-3′), sepCF (5′ AAACATATGAGCACATCCTTGTTCAGTAGCAC-3′) – sepCR (5′ CCAACTGCCCGATAGCAGCCC-3′), araF (5′-TCCATAAGATTAGCGGATCCTAC-3′) and araR 5′-CATGGGGTCAGGTGGGAC-3′). PCR products were purified using the High Pure PCR Product Purification Kit (Roche Diagnostics GmbH) following the manufacturer's instructions.

Construction of the para-SEP constructs

A 430–729-bp region, starting at the initiation codon of each of the sepABC ORFs (refer primer sets; Table 1), was amplified using PCR. The amplicons were cloned into the unique NdeI site (encompassing the ATG initiation codon under the control of the arabinose promoter), and the appropriate 3′-located EcoRI or HindIII site of pAY2-4. The resultant clones were validated by DNA sequencing, for the correct gene fusion relative to the vector, using the araF and araR primers. To complete a functional ORF, the appropriate sep-associated DNA was cloned using restriction enzyme sites, from pACΔ10 and pACΔ11 (Table 1; Fig. 1).

Figure 1.

 (a) Sep virulence-associated clones pACΔ10 and pACΔ11, showing the sep genes and restriction enzyme sites used for cloning each of the sep genes into pAY2-4 (refer text). Abbreviations for restriction enzymes: B, BamHI; Bg, BglII; Bs, BstXI; E, EcoRI; H, HindIII; N, NdeI; S, Sun1; X, XbaI. *, indicates only these restriction sites are shown. (b) The para-SEP constructs; designations are denoted on the right (refer text; Table 1). ▸, denotes a pAY2-4 arabinose-based promoter. Scale bar is indicated.

The pAY2-4 derivative containing the 729-bp sepA amplicon was digested with SunI and ligated to the 6790-bp SunI fragment of pBM32. The resultant construct was digested with BglII and the central BglII region of pACΔ10 was inserted, after which a HindIII self-ligation formed paraA (Fig. 1). The plasmid paraBC was made by digesting the pAY2-4 derivative containing the sepB 430-bp amplicon with HindIII and ligation to HindIII-digested pACΔ11. Digestion of the paraBC construct with XbaI and subsequent self-ligation formed the construct paraB. To make paraC, the pAY2-4 derivative containing the 762-bp sepC amplicon was digested with BstXI and ligated to the BstXI site of pACΔ10BstXIs (Table 1). The construction of paraABC encoding sepA, sepB and sepC was accomplished by ligating the pACΔ11 HindIII region to paraA to form paraABC. By self-ligating the XbaI digested paraABC, paraAB was formed. At each cloning step, the ligation junction points were validated by DNA sequence analysis. All para-SEP constructs were transferred to the E. coli strain BL21(DE3) for protein expression analysis.

Arabinose induction

From a 3 mL overnight culture, 1.0 mL of bacteria was inoculated into 50 mL of LB broth and grown to an OD600 nm of 0.6. The cells were harvested by centrifugation at 8000 g for 10 min, resuspended in 1.0 mL of 0.5 × LB broth and transferred to a fresh 50 mL broth of 0.5 × LB, supplemented with the appropriate antibiotics and arabinose and left at 20°C (0.02 g) for 5 h. Cells were harvested by centrifugation at 8000 g for 3 min and resuspended in 1.2 mL phosphate-buffered saline (10 mM sodium phosphate buffer, pH 7.4; 2.7 mM KCl; 137 mM NaCl). Two 0.6 mL samples were transferred to a 1.7 mL microcentrifuge tube and subjected to three 20 s rounds of sonication using a Sanyo Soniprep 150 Sonicater (18 Ω). The sonicated samples were centrifuged at 16 000 g for 3 min and the supernatant was filter-sterilized through a 0.2 μm Sartorius Minisart® filter to a sterile microcentrifuge tube.

To perform arabinose plate-based expression, 1.5 mL of post arabinose-induced cultures were pelleted (3 min, 8000 g); resuspended in 100 μL of 0.5 × LB supplemented with arabinose and spread onto a 0.5 × LB agar plate containing the appropriate antibiotics and arabinose. The plates were incubated right side up for 24–48 h at 30°C.

Toxin mixing

A sonicated sterile filtrate of E. coli BL21(DE3) paraA was prepared and mixed at room temperature with sonicated E. coli BL21(DE3) paraBC lysate for 30 min. 10 μL−1 was then aliquoted to a 3 mm3 carrot cube that had been predried for 1 h at room temperature and was assessed for activity against grass grub larvae by a standard bioassay.

Sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western immunoblot

Standard SDS-PAGE was performed as described by Laemmli (1970). Proteins were visualized by silver staining according to Blum et al. (1987). Western Immunoblot was under taken using a modified method of Blake et al. (1984) using New Zealand white rabbit polyclonal antibodies generated from manufactured peptides (Mimotopes, 11 Duerdin St, Clayton, Vic. 3168, Australia) based on the amino and carboxyl terminus amino acid residues (SepA, H-MYNIDD-C-HIRYTIIS-OH; SepB, H-MAITAPT-C-GLNDAS-OH; SepC, H-MSTSLFSS-C-LDKRRVM-OH) linked at the central cysteine residue to Keyhole Limpet Hemocyanin as the carrier.

Protein concentration was determined using the Bio-Rad Protein Assay Kit (Bio-Rad Laboratories) based on the method of Bradford (1976).

Standard bioassay

The efficacy of the sonicated sterile Sep filtrates was assessed by the oral injection of 5 μL of filtrate sample through the larval mouth parts, or application of 5 μL of sample (3–5 μg of total protein) to the surface of a 3 mm3 carrot cube, from which the larvae would feed. Twelve second- or third-instar larvae collected from the field were used for each treatment. Inoculated larvae were maintained at 15°C in ice-cube trays. Larvae were fed treated carrot at day 1, and at days 3 and 7 were transferred to fresh trays containing untreated carrot. The occurrence of gut clearance and amber coloration were monitored at days 1, 2, 3, 4, 7 and 14. Controls tested were the wild-type bacterium A1MO2, induced E. coli pAY2-4 and uninduced para-SEP construct; each bioassay was performed a minimum of three times.


Each of the sep genes was expressed either singly or in various combinations. Western Immunoblot of the E. coli arabinose-induced para-SEP constructs showed that they produced proteins of the predicted size: SepA (262 kDa); SepB (156 kDa); and SepC (107 kDa) (Fig. 2a), and that the Sep proteins could be coexpressed by the fusion of the arabinose promoter to the initiation codon of sepA (Fig. 2a). The SepB component is expressed as two forms dependent on which sep gene sepB was expressed. The E. coli arabinose-induced constructs paraABC and paraAB (Fig. 3a) produce a larger SepB protein, while the arabinose constructs expressing sepB and sepBC produce a smaller SepB variant (Fig. 3a). Western Immunoblot of the same para-SEP constructs expressed in S. proteamaculans showed a similar scenario (Fig. 3b).

Figure 2.

 Western Immunoblot of sonicated cell pellets of arabinose-induced Serratia proteamaculans (Sp), Escherichia coli BL21(DE3) (Ec), containing either paraABC (sepA, sepB, sepC) or pAY2-4, (a) M, Bio-Rad Broad Range marker. SepA, SepB and SepC antibodies used to probe are listed at the base of the figure; refer to ‘Materials and methods’ for antibody probe; c. 80-100 ìg of total protein loaded per lane; (b) Silver-stained SDS-PAGE of E. coli BL21(DE3) expressing SepC (paraC). S, supernatant; P, cell pellet; M, Bio-Rad Broad range marker, molecular mass (kDa) are indicated. Arrows denotes the positions of the Sep proteins.

Figure 3.

 Expression of para-SEP constructs in Serratia proteamaculans (Sp), Escherichia coli BL21(DE3) (Ec) and the mobility shift of the SepB protein. (a) Silver-stained SDS-PAGE. Induced para-SEP constructs are denoted as AB (paraAB), B (paraB), BC (paraBC), ABC (paraABC); M, Bio-Rad Broad range marker, molecular mass (kDa) and locations of the Sep proteins are indicted, c. 80–100 μg of total protein loaded per lane; (b) Western Immunoblot of arabinose-induced para-SEP constructs in S. proteamaculans, probed with the SepB antibody; refer to (a) for designations; c. 120–150 μg of total protein loaded per lane. The location of SepB variants is shown. The asterix denotes mobility shift of the SepB protein. Refer to Fig. 2a for validation of negative controls by Western Immunoblot.

Escherichia coli-based para-SEP constructs plated onto arabinose agar had aberrant colony morphologies. If sepA was expressed as an independent entity, growth was reduced with small colonies forming after a 48 h incubation. Conversely, if sepA was expressed in conjunction with sepB (paraAB), or sepB and sepC (paraBC), the growth was less retarded (data not shown).

Bioassay data showed that of the E. coli arabinose-induced para-SEP constructs, only sonicated filtrates derived from the paraABC were able to cause amber disease symptoms (amber coloration and gut clearance) with larvae, taking 24–36 h before symptoms were observed (Table 2). However, over the duration of a 14-day bioassay, 64% of the larvae reverted from amber disease symptoms to a healthy phenotype (Table 2). The expression of the paraABC construct in an E. coli background may have influenced this process by an unknown mechanism. For this reason, the para-SEP constructs were transferred to S. proteamaculans 4031, which, unlike S. entomophila, is able to metabolize arabinose (Table 1). Bioassay data showed that the arabinose-induced sonicated S. proteamaculans (paraABC) filtrate exhibited a similar level of activity relative to E. coli (paraABC), inducing an effect within 24 h postingestion. However, as with E. coli a reversion phenotype was noted, with 63.5% of the diseased becoming healthy after 14 days postingestion (Table 2). The results showed that a repeated dose of the sterile toxin filtrate at the time of reversion could reinitiate the amber pathotype.

Table 2.   Percent Costelytra zealandica larvae amber, after ingestion of arabinose induced filtrates of either Escherichia coli BL21(DE3) para-SEP variants, or Serratia proteamaculans (paraABC) and percent reversion
Treatment*Day 1Day 7Day 14Reversion
  • *

    Refer to ‘Materials and methods’ arabinose induction.

  • Nonarabinose induced culture.

  • Toxin mixing experiment.

E. coli BL21(DE3)
S. proteamaculans
S. entomophila
 Phosphate buffer0.00.00.0 

The ability of the singly expressed SepA protein to be mixed with the coexpressed SepB and SepC (paraBC) components was assessed. Data showed that 37.6% of the larvae turned amber in colour and cleared their gut when fed the combined lysate (Table 2).

To ascertain the specificity of amber disease was not related to the induction of the Sep proteins, the arabinose-induced sonicated S. proteamaculans (paraABC) filtrate was tested against the larvae of the Coeleopteran species, Smith's chafer (Odontria striata), the Red-headed Cockchafer (Adoryphorus couloni) and the Tasmanian grass grub, (Aphodius tasmaniae). No toxic activity was observed against any of these species (data not shown).


The S. entomophila Sep pathogenicity determinants (sepA, sepB and sepC), which are components of a Tc-like insecticidal toxin, have been expressed singly and in various combinations. Bioassay data against grass grub larvae showed that only when sepABC were coexpressed in either an E. coli or S. proteamaculans were amber disease symptoms observed (Table 2). SepC was produced in high amounts if expressed singly (Fig. 2b). This differs from the E. coli-based expression of the X. nematophilus (Sergeant et al., 2003) and P. luminescens (Waterfield et al., 2005) where the respective SepC components (XptB1/TccC1) are weakly expressed, while the respective SepA(XptA1, XptA2/TcdA) and SepB(XptC1/TcdB1) counterparts are expressed at comparatively high levels. Previous studies have shown that when TcdA and XptA1 are expressed as independent entities, they are able to induce toxicity towards there respective targets Manduca sexta (Guo et al., 1999) and Pieris brassicae (Sergeant et al., 2003). The E. coli strain expressing sepA (paraA) was unable to induce amber coloration or gut clearance against first, second or third instar larvae (Table 2). This may reflect the chronic nature of amber disease, where the effects of singly expressed Sep proteins will be too small to be defined by a standard bioassay unless their potentiators (SepB and SepC) are present.

Sergeant et al. (2003) demonstrated that a singly expressed XptA1 could be mixed with the XptC1 and XptB1 components, provided the latter were coexpressed in the same cell. Data showed that on mixing of SepA (paraA) with coexpressed SepB and SepC (paraBC), 37.6% of the larvae turned amber in colour when fed the mixed lysate (Table 2), indicating that the proteins could combine to cause an effect.

Waterfield et al. (2005) observed with the E. coli-based expression of TcdB1 that the protein migrated at a higher molecular weight relative to when it was coexpressed with TccC1. In contrast, the E. coli and S. proteamaculans-based expression of sepB para-SEP variants showed that the larger SepB protein is present when sepB is expressed with either sepA (paraAB) or sepA and sepC (paraABC) and is smaller when expressed as an independent entity or with sepC (paraBC). This may be related to a protein–protein interaction leading to the protection of SepB from E. coli-encoded proteases. Efforts to sequence the amino-terminus of the alternately sized SepB amino terminals in both E. coli and S. proteamaculans are in progress.

Arabinose plate-based expression of the various para-SEP variants in E. coli showed that the induction of sepA as an independent entity (paraA) retarded the growth of the host bacterium. This effect was sequestered by the coexpression of sepA with sepB, or with sepB and sepC (paraBC). This correlates with the observation that during the original mini-Tn10 mutagenesis of the pBM32 sep virulence-associated clone, only two mutations were identified in either the sepB or the sepC genes, compared with nine mutations in the sepA gene (Hurst et al., 2000), indicating that the expression of sepA alone is toxic to the host cell. SDS-PAGE analysis of SepC-sonicated filtrates identified a high amount of SepC in the cell pellet, but little in the supernatant (Fig. 2b). This may reflect SepC similarity to the E. coli Rhs (recombination hot spot) proteins that are believed to be cell wall associated (Wang et al., 1998; Hurst et al., 2000).

Sonicated filtrates derived from arabinose-induced E. coli and S. proteamaculans strains containing paraABC showed that between 7 and 14 days post ingestion of the toxin filtrate, a percentage of the larvae reverted from being symptomatic of amber disease to a healthy phenotype (Table 2). Earlier reports had also noted a reversion phenotype with larvae fed the E. coli strain containing the Sep virulence-associated clone (pBM32) changing from a diseased to a healthy phenotype, a phenomenon thought to be the result of reduced survival dynamics of E. coli in the grass grub gut (Hurst et al., 2000). Alternately, the Afp may be integral to the persistence of the disease process by stopping the grass grub larvae from feeding (Hurst et al., 2004). Spies & Spence (1985) identified that sublethal doses of Bacillus thuringiensis HD-1 crystal endotoxin delivered to M. sexta caused swelling of the midgut goblet and elongation of columnar cells, with some of the columnar cells rupturing. However, gut regeneration occurred and normal insect growth resumed. In the Serratia–grass grub system, no histologial effects have been observed (Jackson et al., 1993). The amber response of larvae administered the arabinose-induced S. proteamaculans (paraABC) toxin filtrate was more rapid than that of the infecting wild-type A1MO2 bacteria (Table 2), indicating that a heightened dose was delivered. In addition, a continued supply of the SepABC toxin filtrate was able to reinitiate the amber pathotype and no lethality was detected over the duration of a 14-day bioassay.

The mechanism behind reversion is yet to be defined, but indicates that the SepABC proteins need to be continuously produced to exert an effect against the grass grub larvae. The Sep proteins may act by titrating out a yet to be defined receptor, necessary for signalling the initiation of gut metabolic function. Alternately, the SepABC proteins may interfere with the transport of the gut metabolic enzymes. Over several days, the SepABC components become degraded, allowing the resumption of normal larval gut metabolic function. In the grass grub–S. entomophila system, this may reflect the chronic nature of the disease.

Amber disease is host specific, only affecting C. zealandica larvae. The testing of the arabinose-induced S. proteamaculans (paraABC) protein filtrate towards other Coeleopterans including the closely related O. striata showed no activity, indicating that the Sep proteins are themselves host specific and that this specificity is not related to the in vivo induction of the sep genes.

Efforts to understand the underlying mechanisms behind the chronic nature and specificity of amber disease are in progress. Having the sep genes under the control of an inducible promoter will facilitate how the SepABC proteins interact to form a complex and help define SepABC–host interactions.


The authors are grateful to Chuck Shoemaker for the provision of the arabinose expression vector pAY2-4 and Richard Townsend for the collection of grass grub larvae. This research was funded by grant C10X0313 of the New Economy Research Fund (NERF), provided by the New Zealand Foundation for Research, Science and Technology.