APPLIED AND ENVIRONMENTAL MICROBIOLOGY,0099-2240/01/$04.00�0 DOI: 10.1128/AEM.67.10.4834–4841.2001

Oct. 2001, p. 4834–4841 Vol. 67, No. 10

Copyright © 2001, American Society for Microbiology. All Rights Reserved.

Isolation and Characterization of Intracellular Protein InclusionsProduced by the Entomopathogenic Bacterium

Photorhabdus luminescensDAVID J. BOWEN† AND JERALD C. ENSIGN*

Department of Bacteriology, The University of Wisconsin, Madison, Wisconsin 53706

Received 16 March 2001/Accepted 3 August 2001

Cells of the entomopathogenic bacterium Photorhabdus luminescens contain two types of morphologicallydistinct crystalline inclusion proteins. The larger rectangular inclusion (type 1) and a smaller bipyramid-shaped inclusion (type 2) were purified from cell lysates by differential centrifugation and isopycnic densitygradient centrifugation. Both structures are composed of protein and are readily soluble at pH 11 and 4 in 1%sodium dodecyl sulfate (SDS) and in 8 M urea. Electrophoretic analysis reveals that each inclusion iscomposed of a single protein subunit with a molecular mass of 11,000 Da. The proteins differ in amino acidcomposition, protease digestion pattern, and immunological cross-reactivity. The protein inclusions are firstvisible in the cells at the time of late exponential growth. Western blot analyses showed that the proteinsappeared in cells during mid- to late exponential growth. When at maximum size in stationary-phase cells, theproteins constitute 40% of the total cellular protein. The protein inclusions are not used during long-termstarvation of the cells and were not toxic when injected into or fed to Galleria mellonella larvae.

Photorhabdus luminescens is a bioluminescent gram-nega-tive, rod-shaped bacterium that was first isolated from a light-emitting insect that had been infected by entomogenous nem-atodes of the family Heterorhabditidae (22, 29). Biochemicaltests and analysis of the 16S rRNA revealed that P. luminescensis related to members of the Enterobacteriaceae in the gammasubdivision of purple bacteria (13, 31, 32).

The bacteria reside in the intestinal tract of the infectivejuvenile (IJ) stage of the nematode, which is the vector fortransmission of the bacteria between insect prey. The IJ pen-etrates the insect, releasing the bacteria into the hemolymph.The bacteria multiply rapidly, killing the insect within 24 to72 h, at which time the dead insect is visibly bioluminescent(23, 25, 29). A 50% lethal dose (LD50) of fewer than 5 cells perinsect has been reported for Galleria mellonella (wax moth)larvae (15). The bacterium produces potent insecticidal toxinsduring growth in the insect as well as in laboratory culture (9,21). The nematode completes several rounds of reproductionwhile feeding on the bacteria in the insect carcass. Within 10 to20 days several thousand IJ progeny, each carrying an inocu-lum of P. luminescens cells, migrate out of the cadaver insearch of new insect prey.

Cells of P. luminescens growing in insect larvae and in cul-ture medium produce phase-bright inclusion proteins withinthe cytoplasm (7, 23). Bacteria of the related genus Xenorhab-dus, associated with entomogenous nematodes of the familySteinernematidiae, also produce two cytoplasmic inclusion pro-teins (11). The genes encoding two inclusion proteins, cipA andcipB, of P. luminescens strain NC1 have been cloned and char-acterized (5). The genes are present at separate loci and show

little nucleotide sequence similarity to each other. Blastsearches using the nucleotide or amino acid sequences of thetwo genes reveal little evidence of homology to any knowngenes, including those encoding the insecticidal crystal pro-teins of Bacillus thuringiensis.

Cultures of P. luminescens exhibit a highly variable pheno-type involving the spontaneous loss of many traits. The vari-ants, termed secondary-phase cells, differ from the originalprimary phase in colony morphology, dye absorption, and bio-chemical utilization and show complete loss of or decrease inantibiotic production, pigmentation, bioluminescence, pro-tease activity, lipase activity, hemolysin production, and theability to support nematode growth (1, 2, 6, 12, 14, 27). Theintracellular inclusion proteins are absent in the secondaryphase cells (5).

The function of the inclusion proteins of Photorhabdus andXenorhabdus is unknown. Because both genera of bacteria areentomopathogens and are associated in a symbiosis with ento-mopathogenic nematodes, logical hypotheses are that the in-clusion proteins are involved in the nematode association or inpathogenesis. The cost of producing the unusually largeamounts of these proteins strongly suggests that the proteinsmust serve an important function for the bacteria.

This report describes the isolation and characterization ofthe two protein inclusions from P. luminescens NC1 and Hmand presents the results of attempts to define their function.

MATERIALS AND METHODS

Bacteria and culture conditions. The inclusion proteins were purified from P.luminescens strains Hm (G. M. Thomas, University of California) and NC1(Wayne Brooks, University of North Carolina). Stock cultures were maintainedon 2% proteose peptone no. 3 (PP3) (Difco Laboratories, Detroit, Mich.) so-lidified with 1.5% Bacto agar (Difco). Cultures were incubated at 30°C for 72 h,stored at room temperature, and transferred at monthly intervals. Two stablesecondary-phase variants were isolated from the primary-phase NC1. They arereferred to as white secondary (nonpigmented) and yellow secondary (yellowpigmentation).

* Corresponding author. Mailing address: Department of Bacteriol-ogy, The University of Wisconsin, Madison, WI 53706. Phone: (608)262-7877. Fax: (608) 262-9865. E-mail: jcensign@facstaff.wisc.edu.

† Present address: Department of Entomology, The University ofWisconsin, Madison, WI 53706.

4834

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

Microscopy. Phase-contrast micrographs were taken with a Zeiss photomicro-scope using Kodak technical pan film (Eastman Kodak Co., Rochester, N.Y.).For the time-lapse study, cells from a 48-h culture were incubated at 30°C on athin layer of PP3 agar on a sterile microscope slide and covered with an oxygen-permeable Teflon membrane. For transmission electron microscopy, the cellswere fixed in 2% glutaraldehyde in 100 mM phosphate buffer at pH 7.4, embed-ded in Duracupan (Sigma), thin sectioned, and stained with lead citrate. Thesectioned cells were viewed under a Jeol-100CX electron microscope. For scan-ning electron microscopy, purified inclusions were suspended in sterile water(sH2O), placed on double-stick tape on a steel post, dried, and coated with goldin vacuo. The samples were examined with a Hitachi S-570 scanning electronmicroscope.

Optimization of inclusion production. The conditions for optimum inclusionproduction in liquid culture (all culture media from Difco) were determined bygrowing cells in 5% yeast extract, 2% neopeptone, 2% casitone, 2% proteosepeptone no. 3, 2.5% nutrient broth, 10% peptone, or 2% Trypticase at 30°C.Cells were examined after 72 h by phase-contrast microscopy.

Isolation of inclusions. A 2-ml suspension of P. luminescens cells in 2% PP3broth was spread on 2% PP3 agar in Pyrex glass baking dishes (18 by 30 cm).After 7 days of incubation at 28°C, 100 ml of sH2O was added, and the cells werescraped from the agar surface with a bent glass rod. The cell suspension wascentrifuged at 5,000 � g for 10 min. The resulting pellet was resuspended insterile phosphate-buffered saline (sPBS) consisting of (per liter) NaCl (8.0 g),KCl (0.20 g), Na2HPO4 (1.15 g), and KH2PO4 (0.2 g) (18) and centrifuged at3,000 � g for 10 min. The cell pellets were resuspended in 10 ml of sPBS andpassed twice through a French press at 10,000 lb/in2. The cell lysate was dilutedto 50 ml in sPBS and centrifuged at 2,000 � g for 20 min. This step was repeatedthree times.

The chalky white pellets that were produced were resuspended in 9 ml ofsPBS, and 3-ml samples were applied to the top of discontinuous Percoll (SigmaChemical Co., St. Louis, Mo.) density gradients (28). The Percoll solutions wereprepared from a stock solution containing 1 part 2.5 M sucrose and 9 parts(vol/vol) Percoll. Step gradients consisted of a 5-ml layer of Percoll stock solutionplaced at the bottom of 30-ml Corex centrifuge tubes, followed by application ofsequential layers of 95, 90, 80, and 70% dilutions of the stock solution. Followingcentrifugation (4 h at 5,000 � g) in a Sorvall type HS-4 swinging bucket rotor at4°C, the two visually distinct inclusion-containing layers were removed separatelyand washed several times in sH2O by centrifugation at 2,000 � g. Each fractionwas then centrifuged through the gradients once more as described. The purifiedinclusions were stored as frozen pellets at �20°C, lyophilized, and stored underdesiccation at room temperature or in sH2O at 4°C.

Solubility of inclusions. A suspension of purified inclusions was made in sH2Oat an optical density at 600 nm (OD600) of 2.0. The pH of samples were adjustedby slowly adding 1.0-�l amounts of 0.1 M HCl or 0.1 M NaOH, and the OD600

of the suspensions was monitored. To determine solubility in sodium dodecylsulfate (SDS), 100 �l of 10% SDS (Calbiochem, San Diego, Calif.) was added to900 �l of inclusion suspension. To determine solubility in urea or EDTA, inclu-sions were resuspended in 1 ml of 8 M urea or 100 mM EDTA at pH 8.0.

Compositional analysis and total inclusion protein content of cells. The pro-tein content of inclusions was determined by the Lowry assay (16), and thecarbohydrate content was estimated by the anthrone reaction (16). Total aminoacid composition was determined at the Biotechnology Instrumentation facilityof the University of California–Riverside, using the Beckman 120C amino acidanalysis system. The percentage of inclusion protein in the cells was determinedusing 7-day-old cells scraped from agar plates. The washed cells were disruptedusing a French pressure cell. The protein content of the lysate was determined.The inclusions were then collected from the lysate by centrifugation and washedtwice with sH2O and subsequent centrifugation, and the protein content of thepelleted inclusions was determined. The percentage of inclusion proteins in thecell was calculated as [(milligrams of inclusion protein)/(milligrams of lysateprotein)] � 100.

Mass spectrometry. Mass determinations were performed at the University ofWisconsin Biotechnology Center on a Bruker Biflex III matrix-assisted laserdesorption-ionization time-of-flight (MALDI-TOF) mass spectrometer.

Protease digestion of intact inclusions. The protein inclusions were digestedwith trypsin (Sigma Chemical Co.), V-8 protease from Staphylococcus aureus(Miles Scientific, Naperville, Ill.), and two different pronase preparations (Cal-biochem). Each digestion reaction contained 1.8 mg of the pure type 1 or 2protein inclusions suspended in 0.9 ml of 100 mM Tris-HCl at pH 7.5, to whichwas added 0.2 mg of protease dissolved in 0.1 ml of the same buffer. The sampleswere digested at 37°C for 4 h on a rocking platform. Five-microliter samples werethen mixed with SDS sample loading buffer, boiled for 5 min, and analyzed bySDS-polyacrylamide gel electrophoresis (PAGE) as described below.

Production of antisera. The density gradient-purified type 1 and 2 inclusionproteins from NC1 were used to immunize New Zealand White rabbits. Prior toemulsification in adjuvant, 500 �g of each inclusion protein was solubilized in 1ml of 10 mM NaOH. Freund’s complete adjuvant was used for the primaryimmunizations, and Freund’s incomplete adjuvant was used for three additionalinjections made at monthly intervals. Serum obtained 10 days after the finalinjections was heated to 56°C for 15 min to inactivate complement and stored at�20°C (18).

Time course of inclusion production. Cells were grown for 16 h at 30°C in 25ml of 2% PP3 broth in a 125-ml flask shaken at 250 rpm. Five milliliters of thisculture was used to inoculate 250 ml of 2% PP3 broth in a 1-liter flask, which wasalso shaken at 250 rpm at 30°C. At various times, samples were removed andwashed in sH2O, and the total protein content of the cells was determined. Thesamples were adjusted to 200 �g of protein per ml, and 10-�l samples (2 �g oftotal protein) were analyzed by SDS-PAGE. Secondary-phase cells grown for96 h at 30°C in 25 ml of 2% PP3 broth in a 125-ml flask with shaking at 250 rpmwere also analyzed by SDS-PAGE.

Gel electrophoresis and Western blot analyses. Cells and purified inclusionproteins were subjected to SDS-PAGE analysis using a protocol designed forhigh resolution of proteins in the 5- to 30-kDa range (33). Proteins were stainedwith 0.1% Coomassie brilliant blue R-250. For Western blot analysis, proteinsseparated by SDS-PAGE were electroblotted onto nitrocellulose membranes in25 mM Tris–192 mM glycine and 20% (vol/vol) methanol. The gels were elec-troblotted for 1 h at 20 V constant voltage in a Genie Blotter (Ideas Scientific,Minneapolis, Minn.). The AuroProbe BLplus and IntenSE BL silver enhance-ment kit (Amersham Life Sciences, Arlington Heights, Ill.) were used accordingto the manufacturers’ instructions to detect antigen on the blots. The primaryantibody was used at a 1:1,000 dilution.

Stability of inclusion proteins during growth and starvation. Cells were grownfor 48 h at 30°C in 50 ml of 2% PP3 broth in 500-ml flasks shaken at 250 rpm.Samples were removed at various times, and microscopic counts were deter-mined using a Petroff-Hausser counting chamber. Viable-cell counts were de-termined by dilution of samples into fresh 2% PP3 broth, plating on 2% PP3agar, and counting colonies after 5 days of incubation at 30°C. For starvationexperiments, the 48-h PP3 cultures were divided into two 25-ml portions. Onesample was transferred to a 250-ml flask and incubated as above. The othersample was centrifuged at 2,000 � g for 5 min at room temperature. The cellswere resuspended in 25 ml of sPBS and incubated as above. Samples wereremoved from the flasks, and microscopic counts and viable-cell counts weredetermined at various times.

Insect toxicity analyses. G. mellonella larvae were obtained from H. C. Coppel(Department of Entomology, University of Wisconsin–Madison) and grown byhis method (26). Samples containing 25 �g of purified type 1 or 2 proteins in 10�l of sH2O were either fed to or injected into last instar G. mellonella larva (9).The inclusion proteins were also solubilized with 10 mM HCl or 10 mM NaOH,filter sterilized with 0.2-�m-pore-sized membrane filters, and then fed or in-jected. Samples (cells plus broth) taken directly from 48-h PP3 cultures were alsofed to and injected into larvae. Freshly prepared (stored at 4°C) and frozeninclusion preparations were used for bioassays.

RESULTS

Production and isolation of inclusion proteins. A variety ofliquid culture media were tested for their effect on inclusionproduction by both P. luminescens strains NC1 and Hm. Theinclusions were visibly evident using phase-contrast microscopyin most cells of both strains after 48 h of growth in 2.5%nutrient broth and 2% neopeptone. Approximately half thecells contained inclusions when grown in 2% Trypticase soybroth. The cells grew well but produced no visible inclusionswhen grown in 2% casitone, 5% yeast extract, or 10% peptone.The best growth medium, in which more than 90% of the cellscontained phase-bright inclusions, was 2% PP3. Photomicro-graphs of cells of strains NC1 and Hm grown on 2% PP3 agarreveal the presence of phase-bright inclusions in the cells (Fig.1A and 1B).

Transmission electron micrographs of thin-sectioned cells ofNC1 and Hm show that both these strains contain two mor-phologically distinct inclusions (Fig. 2A and 2D) The protein

VOL. 67, 2001 PHOTORHABDUS PROTEIN INCLUSIONS 4835

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

inclusions released from cells by French pressure breakagebecame separated into two bands in the Percoll gradients.Scanning electron micrographs of Percoll-separated inclusionsshow the less dense type 1 inclusion to be rectangular (Fig. 2Band 2E). The morphology of the more dense type 2 inclusionsof NC1 and Hm differs. The NC1 type 2 is bipyramidal (Fig.2C), and the Hm type 2 has pointed ends and is elongated inthe middle (Fig. 2F).

Solubility, compositional analyses, and mass spectrometry.The purified type 1 and type 2 inclusions from both strains areinsoluble in water and at neutral pH in PBS or Tris buffer. Theeffect of pH on the solubility of the inclusion structures wastested by slowly increasing and decreasing the pH of an aque-ous suspension. The inclusion structures remained insolublebetween pH 5 and 10. The OD600 of the suspension decreasedby more than 90% at pH 11 or 4; at both pHs, the inclusionsbecome soluble. As the pH was slowly adjusted from 4 and 11toward neutrality, the solutions became cloudy at pH 5 and pH7, respectively, coincident with the formation of an amorphousprecipitate. Both types of inclusions were soluble (greater then90% OD600 decrease) in 8 M urea and 1% SDS. The inclusionswere not soluble in 100 mM EDTA.

Both type 1 and 2 inclusion structures are composed entirelyof protein, with no detectable carbohydrate, even when a10-mg (dry weight) sample of inclusions was analyzed. TheMALDI-TOF mass spectrometer analyses confirmed the ab-sence of glycosylation on the proteins.

The results of amino acid composition analyses of type 1 andtype 2 protein inclusions of strains NC1 and Hm are shown inTable 1. The amino acid compositions of the type 1 and type 2inclusion proteins of strains NC1 and Hm obtained by compo-sitional analysis are similar and correlate closely with the com-position predicted for the cipA and cipB gene products of theHm strain (5). The molecular weights predicted from the aminoacid composition are NC1 type 1, 10,578; NC1 type 2, 10,648;Hm type 1, 10,574; and Hm type 2, 10,700. The amino acidcompositions of the two protein inclusions are quite different,

however. The type 1 protein inclusion contains 0 to 1% cys-teine, 1 to 2% methonine, 20 to 24% leucine, and 4% lysine.The type 2 protein contains 4 to 5% cysteine, 13% methonine,9 to 11% leucine, and 9 to 10% lysine. The type 1 proteincontains approximately 47% hydrophobic amino acids, whilethe type 2 protein contains approximately 42% hydrophobicamino acid residues, with particularly high levels of valine,methonine, isoleucine, and leucine.

The amino acid compositional analysis obtained by acid hy-drolysis of the proteins followed by high-pressure liquid chro-matography (HPLC) showed four lysine residues in the type 1inclusion of both NC1 and Hm. This value differs from the 14lysine residues predicted by the Hm gene sequence (5). Mostlikely this difference is due to an unexplained error in thecompositional analyses.

The percentage of total cell protein attributable to the in-clusion proteins is approximately 40% in the 7-day-old cul-tures.

The density gradient-purified inclusions from both strainswere analyzed by mass spectrometry (MALDI-TOF). The re-sults show that the inclusions are composed almost entirely ofa single-molecular-mass species. The molecular mass of thetype 1 inclusion of the Hm strain is 11,323 Da, which is nearlyidentical to the predicted mass of 11,315 Da for the cipB geneproduct of strain Hm. The molecular mass of NC1 type 1 is11,381 Da and is very close to the molecular mass of the Hmtype 1 inclusion. The molecular masses of the Hm and NC1type 2 inclusions are 11,711 and 11,697 Da, respectively, whichis nearly identical to the predicted 11,700 Da of the cipA geneproduct.

The mass spectrometry data for the type 2 inclusions of bothstrains show a minor shoulder peak indicating components 42mass units larger for the NC1 and 47 mass units larger for theHm inclusions. This mass shift suggests that a small percentageof the protein may contain a posttranslational modification;most likely it is acetylated.

SDS-PAGE analysis and proteolytic degradation. The re-sults of SDS-PAGE analysis of the protein inclusions show thatboth type 1 and type 2 protein inclusions from strains NC1 andHm are apparently composed of single proteins that each havea molecular mass of approximately 10 kDa (Fig. 3). This valuecorrelates well with the mass estimated from amino acid anal-yses and mass estimates for the inclusion proteins.

The degree of proteolytic digestion of type 1 and type 2inclusions of NC1 by four different proteases was analyzed bySDS-PAGE. The type 1 protein inclusion was extensively de-graded by two different pronase preparations, was cleaved intotwo discrete fragments by trypsin, and was not hydrolyzed bythe V8 protease. The type 2 protein inclusion was only slightlydegraded by the pronase preparations, and most of the proteinremained as a single intact band that was not degraded bytrypsin or V8 protease (not shown). Inclusions present in celllysates of both strains were also analyzed for degradation byindigenous cellular proteases. The cell lysates contained highlevels of proteolytic activity, but degradation was not detectedin suspensions of inclusions incubated in the lysates for as longas 1 week (not shown).

Immunological analysis and temporal regulation of inclu-sion protein accumulation. Polyclonal antisera raised againsttype 1 and type 2 inclusion proteins of NC1 were used to

FIG. 1. Phase-contrast photomicrographs of P. luminescens cellsgrown for 7 days on PP3 agar. (A) P. luminescens strain NC1. (B) P.luminescens strain Hm. Magnification, �2,560.

4836 BOWEN AND ENSIGN APPL. ENVIRON. MICROBIOL.

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

determine the time of inclusion protein production in growingcells. Western blot analyses (Fig. 4A and B) revealed that bothtype 1 and 2 proteins are first detected at 16 h (lanes 7, panelsA and B). Both proteins reached high levels in 24-h cells (lanes8). The inclusion protein detected at 0 h (lanes 1, A and B)resulted from the stationary-phase cells used as the inoculum.During the first 12 h of growth, the inclusion proteins werediluted relative to the total protein content of the cells. Micro-scopic examination of the cells confirmed that the proteininclusions were first visible at 16 h of growth. By 24 h, greaterthan 70% of the cells contained small inclusions (Fig. 5).

Two secondary-phase variants were isolated from the pri-mary NC1 strain. The white secondary was deficient in allcharacteristics typical of primary-phase cells. The yellow sec-

ondary-phase variant was deficient in all characteristics exceptpigmentation and antibiotic production. Both secondary vari-ants contained no visible inclusions (not shown). Western blotanalyses did not detect inclusion proteins in the secondary-phase variants (Fig. 4C and 4D, lanes 1 and 2).

The type 1 and type 2 NC1 antisera were used to analyze theimmunological cross-reactivity of the inclusion proteins fromstrains NC1 and Hm (Fig. 4C and 4D). Type 1 antiserumcross-reacts weakly with NC1 type 2 protein (Fig. 4C, lane 6)but strongly recognizes a 10-kDa band, and reacts more weaklywith two proteins of approximately 20 and 30 kDa in the NC1type 1 material (Fig. 4C, lane 5). Type 2 antiserum cross-reactsweakly with NC1 type 1 protein (Fig. 4D, lane 5) but stronglyrecognizes a 10-kDa band and more weakly a protein of ap-

FIG. 2. Transmission electron micrograph of thin sections of cells of P. luminescens NC1 (A) and Hm (D) showing that cells contain two distinctinclusion protein structures. Scanning electron micrographs of protein inclusions separated by density gradient centrifugation. (B and C) Type 1and 2 inclusions of NC1, respectively; (E and F), type 1 and 2 inclusions of Hm, respectively.

VOL. 67, 2001 PHOTORHABDUS PROTEIN INCLUSIONS 4837

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

proximately 20 kDa in the type 2 lane (Fig. 4D, lane 6). Thecross-reactivity and banding patterns produced by the type 1and 2 antisera with NC1 inclusion proteins were nearly iden-tical to the results obtained for the Hm inclusion proteins (Fig.4C and 4D, lanes 3 and 4). The higher-molecular-weight bandsdetected by the antisera may be multimers of the individualsubunits of each inclusion type, since their migration distancesare consistent with those of a dimer and a trimer of the indi-vidual proteins. If this conclusion is correct, these multimersoccur even in the presence of SDS.

Are the protein inclusions nutrient reserves? The possibilitythat the protein inclusions might serve as a reserve nutrientsource for the bacteria was tested. The cells contained visiblecytoplasmic protein inclusions at 16 h (Fig. 5). Growth in 2%PP3 broth reached maximum levels between 24 and 36 h (Fig.6). The direct microscopic counts reached a maximum level of5 � 109 cells/ml at 36 h. The viable-cell counts reached amaximum of 8 � 108 cells/ml at 24 h and decreased steadilyuntil only about 1% of the cells (107/ml) were viable at 192 h.

Protein inclusions were still visible in most of the cells at 192 h.The cells starved in sPBS generally remained viable up to192 h. The ratio of microscopic counts to viable plate countsremained nearly constant throughout the starvation period

FIG. 3. SDS-PAGE analysis (18% gel) of density gradient-purifiedprotein inclusions from P. luminescens strains Hm and NC1. Lane Sta,molecular size markers (2 �g of protein per band); lane 1, Hm type 1protein inclusions; lane 2, Hm type 2 protein inclusions; lane 3, NC1type 2 protein inclusions; lane 4, NC1 type 1 protein inclusions. Eachprotein inclusion sample contained 3 �g of total protein.

FIG. 4. (A and B) Western blot analyses of time of appearanceduring growth of P. luminescens NC1 type 1 and 2 protein inclusions.(A) Cell lysates probed with type 1 inclusion antiserum. (B) Celllysates probed with type 2 inclusion antiserum. Lanes 1, 0-h cells(inoculum); lanes 2, 6-h cells; lanes 3, 8-h cells; lanes 4, 10-h cells; lanes5, 12-h cells; lanes 6, 14-h cells; lanes 7, 16-h cells; lanes 8, 24-h cells.(C and D) Western blot analyses of gradient-purified inclusions andcell lysates from secondary-phase cells. (C) Probed with type 1 inclu-sion antiserm. (D) Probed with type 2 inclusion antiserum. Lanes 1,96-h NC1 yellow secondary cells; lanes 2, 96-h NC1 white secondarycells; lanes 3, type 1 protein inclusions from strain Hm; lanes 4, type 2protein inclusions from strain Hm; lanes 5, type 1 protein inclusionsfrom NC1; lanes 6, type 2 protein inclusions from NC1. Cell lysatelanes contain 2 mg of total protein. Purified protein inclusion lanescontain 0.1 mg of total protein.

TABLE 1. Amino acid composition of P. luminescens Hm and NC1 protein inclusions compared to the amino acid composition predictedfrom the Hm cipA and cipB gene sequences

Amino acida

No. of residues per subunit

Hm type 1NC1 type 1b

Hm type 2NC1 type 2b

Analysisb cipBc Analysisb cipAd

Lysine* 4 14 4 10 11 9Histidine* 4 3 3 1 1 2Phenylalanine* 1 1 1 0 0 1Leucine* 20 18 24 11 11 9Isoleucine* 10 11 9 7 9 6Threonine* 1 1 0 2 3 1Methionine* 2 2 1 13 14 13Valine* 14 16 16 9 10 9Arginine* 1 1 2 2 2 3Tyrosine* 3 2 2 2 2 2Aspartate/asparagine 15 13 10 14 14 12Serine 6 6 6 4 4 3Glutamate/glutamine 6 5 9 5 5 5Proline 2 1 2 3 3 3Glycine 3 3 4 6 7 7Alanine 1 1 2 2 3 3Cysteine 1 0 0 4 4 5Trytophan NDe 0 ND ND 1 NDTotal 94 100 95 95 104 93

a *, essential for Neoplectana glaseri (20).b based and amino acid composition analysis.c Amino acid composition based on cipB gene sequence (5).d Amino acid composition based on cipA gene sequence (5).e ND, not determined.

4838 BOWEN AND ENSIGN APPL. ENVIRON. MICROBIOL.

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

(Fig. 6), and during this time the protein inclusions in the cellswere not noticeably reduced in size (Fig. 5).

Protein inclusions in dividing cells. A possible explanationfor the loss in viability of late-stationary-phase cells is that thelarge protein inclusions in the cytoplasm might interfere withcell division. This possibility is unlikely to be the case, becausetime-lapse phase-contrast micrographs clearly show that a cell

with a large inclusion is capable of cell division (Fig. 7). Thecell elongates and divides on either side of the inclusion. Theinclusion protein remains visible inside the mother cell throughseveral rounds of division. This result also shows that inclusionproteins are not detectably degraded and consumed by divid-ing cells.

Toxicity of protein inclusions. The intact and solubilized P.luminescens protein inclusions did not kill G. mellonella larvae.This was true for both frozen and freshly isolated inclusions.

FIG. 5. Phase-contrast photomicrographs of cells of P. luminescens strain NC1 grown in PP3 broth and during starvation. The first inclusionsare visible at 16 h (numbers on the photomicrographs indicate time [in hours] after inoculation of the culture). After 48 h of growth in PP3, thecells were centrifuged, resuspended in sPBS, and returned to the shaker. The inclusions remain present at all times after the shift to starvationconditions (72 to 192 h).

FIG. 6. Growth and starvation of P. luminescens NC1. Cells wereincubated at 30°C in a shake flask in 2% PP3. At 48 h, a portion of theculture was removed, centrifuged, washed, resuspended in sPBS(starved), and returned to the shaker. Microscopic counts and viable-cell counts were determined at various times after inoculation. Sym-bols: ■, PP3 microscopic count: �, PP3 viable count; Œ, starved mi-croscopic count; ‚, starved viable count.

FIG. 7. Time-lapse photomicrographs of P. luminescens cells grow-ing on slide culture of PP3 agar. Numbers at the upper left of eachframe are the incubation time (in hours).

VOL. 67, 2001 PHOTORHABDUS PROTEIN INCLUSIONS 4839

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

Injection of larvae with several thousand viable P. luminescenscells from a 48-h culture killed the larvae in 24 h.

DISCUSSION

Cells of P. luminescens strains NC1 and Hm each containtwo distinct intracellular protein inclusions that can constituteup to 40% of the total cell protein. The proteins are nearlyidentical in molecular size and solubility properties but differsignificantly in amino acid content, susceptibility to proteasedigestion, and immunological cross-reactivity. The unusuallyhigh content of hydrophobic amino acids in the two proteinclasses (47% for type 1 and 42% for type 2) probably accountsfor their insolubility at neutral pH and solubility at alkaline andacidic pH.

The mass spectrometry and SDS-PAGE analyses show thateach inclusion type is composed of a single protein subunit.The mass spectrometry data also showed that the NC1 type 1inclusion is approximately 66 mass units larger than the Hmtype 1. This is probably due to minor differences in the aminoacid composition of the proteins. All of these analyses com-bined with the amino acid composition analyses show that thetype 1 inclusion is the cipB gene product and the type 2 inclu-sion is the cipA gene product (5).

The biological function of the inclusion proteins is notknown. One possibility suggested by the interesting analogy tothe parasporal insecticidal crystal proteins of Bacillus thurin-giensis (3, 18, 19) is that the proteins are involved in insecttoxicity of P. luminescens. Feeding and injection of G. mel-lonella larvae with both the native and solubilized inclusionproteins did not support this hypothesis. Insect larvae arehighly susceptible to the intact bacterial cells; the injectedlethal dose is 10 to 100 cells. Similarly, secondary-phase cellsthat contain no intracellular protein inclusions are equally vir-ulent when injected into larvae (5).

Another plausible function for the proteins is involvement inthe nematode symbiosis. The Heterorhabditis nematodes growand multiply while feeding on the primary-inclusion-containingcells, but do not grow and multiply with the secondary-phasecells that lack inclusion proteins. The entomopathogenic nem-atode Neoplectana (Steinernema) glaseri requires 10 amino ac-ids for growth (20), and these 10 amino acids account for morethan 60% of the amino acids in the protein inclusions. The type2 inclusion protein is especially rich in methionine, which con-stitutes 13% of the total amino acids. This level of methionineis unusual; the average methionine content of a collection of207 proteins is 1.7% (22). Thus, intracellular protein inclusionsmight serve as a rich supply of essential amino acids for thenematode, although there is no known evidence for the nem-atodes’ obtaining these amino acids from the inclusions. If theprotein inclusions are degraded by enzymes in the nematodeintestine and are essential to nematode development, the nem-atodes would be expected to grow on killed cells. In prelimi-nary studies, we found that the nematodes do not grow andreproduce on heat-, freeze-thaw-, or UV light-killed primary-stage cells that contain protein inclusions (unpublished obser-vations). The requirement for living P. luminescens cells fornematode development indicates that the nature of the asso-ciation between the two organisms is a complex interaction inwhich the inclusion proteins may be just one factor. Two mu-

tants of P. luminescens, each missing just one of the inclusionproteins, did not support nematode growth (5). However,these mutants also acquired some secondary-phase character-istics, which could also explain the inability to support nema-tode growth. Further evidence that this symbiosis is a complexinteraction is the report that a transposon-mediated mutationin a phosphopantetheinyl transferase gene of P. luminescensNC1 results in cells that no longer supported growth and re-production of the nematodes (10). This mutant produced boththe protein inclusions.

The observation that culture broth of P. luminescens NC1contains bacteriocins and phage particles was the basis of spec-ulation that they may be related to the cytoplasmic inclusions(4). The P. luminescens strain NC1 used in this study alsoproduced both of these particles (S. Bintrim, unpublished ob-servations). Western blot analyses using both type 1 and type 2antisera did not detect any immunologically related material inthe culture broth which contained these phage-like structures(unpublished observations).

The inclusions do not appear to be energy or amino acidreserves. The inclusion proteins were not degraded is starvingcells (Fig. 6), and cells incubated on agar media or in brothmedia for several months retained the inclusions.

Another bacterium, Xenorhabdus nematophilus, is symbioti-cally associated with the entomopathogenic nematode Stein-ernema carpocapsae (30). This bacterium, which is related toPhotorhabdus in some characteristics but clearly belongs to adifferent genus (8), also produces two intracellular crystal pro-teins (11). The sizes of these proteins, as estimated by SDS-PAGE analyses, were 22 and 26 kDa, which is twice the size ofthe P. luminescens proteins. The X. nematophilus protein in-clusions are similar in some solubility characteristics to the P.luminescens protein inclusions; for example, they are insolubleat neutral pH but soluble at acidic and alkaline pH, but differin being soluble in 5 mM EDTA, while both of the P. lumine-scens protein inclusions were insoluble at concentrations of upto 100 mM EDTA. Couche et al. suggested that the inclusionproteins of X. nematophilus might be associated with nematodegrowth and reproduction, but no supporting data were pre-sented (11).

It is interesting that two different genera of bacteria involvedin a symbiotic relationship with two different families of ento-mopathogenic nematodes both produce two intracellular pro-tein inclusions. Because the proteins differ in size and otherimportant aspects, it is likely that the two organisms developedthis property independently, although perhaps for a commonpurpose.

ACKNOWLEDGMENTS

This work was supported by funds from the S. C. Johnson Wax Co.and by a USDA Hatch Grant from the College of Agricultural and LifeSciences, University of Wisconsin–Madison.

REFERENCES

1. Akhurst, R. J. 1980. Morphological and functional dimorphism in Xeno-rhabdus spp. bacteria symbiotically associated with the insect pathogenicnematodes Neoaplectana and Heterorhabdus. J. Gen. Microbiol. 121:303–309.

2. Akhurst, R. J. 1982. Antibiotic activity of Xenorhabdus spp., bacteria symbi-otically associated with insect pathogenic nematodes of the families Hetero-rhabditidae and Steinernematidae. J. Gen. Microbiol. 128:3061–3065.

3. Ang, B. J., and K. W. Nickerson. 1978. Purification of the protein crystal

4840 BOWEN AND ENSIGN APPL. ENVIRON. MICROBIOL.

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

from Bacillus thuringiensis by zonal gradient centrifugation. Appl. Environ.Microbiol. 36:625–625.

4. Baghdiguian, S., M. Boyer-Giglio, J. Thaler, G. Bonnot, and N. Boemare.1993. Bacteriocinogenesis in cells of Xenorhabdus nematophilus and Photo-rhabdus luminescens: Enterobacteriaceae associated with entomopathogenicnematodes. Biol. Cell 79:177–185.

5. Bintrim, S. B., and J. C. Ensign. 1998. Insertional inactivation of genesencoding the crystalline inclusion proteins of Photorhabdus results in mu-tants with pleiotropic phenotypes. J. Bacteriol. 180:1261–1269.

6. Bleakly, B., and K. H. Nealson. 1988. Characterization of primary andsecondary forms of Xenorhabdus luminescens strain Hm. FEMS Microbiol.Ecol. 53:241–250.

7. Boemare, N. E., C. Louis, and G. Kuh. 1982. Etude ultrastructurale descrisaux chez Xenorhabdus spp. bacteries infodees aux nematodes entomo-phages Steinernematidae et Heterorhabditidae. C. R. Soc. Biol. 177:107–115.

8. Boemare, N. E., R. J. Akhurst, and R. G. Mourant. 1993. DNA relatednessbetween Xenorhabdus spp. (Enterobacteriaceae), symbiotic bacteria of ento-mopathogenic nematodes, and a proposal to transfer Xenorhabdus lumine-scens to a new genus, Photorhabdus gen. nov. Int. J. Syst. Bacteriol. 43:249–255.

9. Bowen, D. J., and J. C. Ensign. 1998. Purification and characterization of ahigh-molecular-weight insecticidal protein complex produced by the ento-mopathogenic bacterium Photorhabdus luminescens. Appl. Environ. Micro-biol. 64:3029–3035.

10. Ciche, T. A., S. B. Bintrim, A. R. Horswill, and J. C. Ensign. 2001. Aphosphopantetheinyl transferase homolog is essential for Photorhabdus lu-minescens to support growth and reproduction of the entomopathogenicnematode Heterorhabditis bacteriophora. J. Bacteriol. 183:3117–3126.

11. Couche, G. A., and R. P. Gregson. 1987. Protein inclusions produced byentomopathogenic bacterium Xenorhabdus nematophilus. J. Bacteriol. 169:5279–5288.

12. Ehlers, R., S. Stossel, and U. Wyss. 1990. The influence of phase variants ofXenorhabdus spp. and Escherichia coli (Enterobacteriaceae) on the propa-gation of entomopathogenic nematodes of the genera Steinernema and Het-erorhabditis. Rev. Nematol. 13:417–424.

13. Ehlers, A., U. Wyss, and E. Stackebrandt. 1988. 16s rRNA cataloging and thephylogenetic position of the genus Xenorhabdus. Syst. Appl. Microbiol. 10:121–125.

14. Gerritsen, L. J. M., G. DeRaay, and P. H. Smits. 1992. Characterization offorms variants of Xenorhabdus luminescens. Appl. Environ. Microbiol. 58:1975–1979.

15. Giffin, C. T., W. R. Simons, and P. H. Smits. 1989. Activity and infectivity ofHeterorhabditis spp. J. Invertebr. Pathol. 53:107–112.

16. Hansen, R. S., and J. A. Phillips. 1981. Chemical composition, p. 328–364. InP. Gerhardt (ed.), Manual of methods for general bacteriology. AmericanSociety for Microbiology, Washington, D.C.

17. Hofte, H., and H. R. Whitely. 1989. Insecticidal crystal protein of Bacillus

thuringiensis. Microbiol. Rev. 53:242–255.18. Hudson, L., and F. C. Hay. 1980. Practical immunology, 2nd ed., p. 336.

Blackwell Scientific Publications, Oxford, England.19. Ibara, J. E., and B. A. Federici. 1956. Isolation of a relatively nontoxic

65-kilodalton protein inclusion from the paraporal body of Bacillus thurin-giensis subsp. israelensis. J. Bacteriol. 165:527–533.

20. Jackson, G. J. 1973. Neoplectana glaseri: essential amino acids. Exp. Parasi-tol. 34:111–114.

21. Jaroz, J., M. Balcerzak, and H. Skrzypek. 1991. Involvement of larvicidaltoxin in pathogenesis of insect parasitism with the Rhabditoid nematodesSteinernema feltiae and Heterorhabditis bacteriophora. Entomophaga 36:361–368.

22. Khan, A., W. M. Brooks, and H. Hirschmann. 1976. Chromonema heliothidisn. gen., n. sp. (Steinernematidae, Nematoda), a parasite of Heliothis zea(Noctuidae, Lepidoptera) and other insects. J. Nematol. 8:159–168.

23. Khan, A., and W. Brooks. 1977. A chromogenic bioluminescent bacteriumassociated with the entomophilic nematode Chromonema heliothidis. J. In-vertebr. Pathol. 29:253–261.

24. Klapper, M. H. 1977. Frequency of occurrence of each amino acid residue inthe primary structures of 207 unrelated proteins of known sequence. Bio-chem. Biophys. Res. Commun. 78:1018–1024.

25. Milstead, J. E. 1979. Heterorhabditis bacteriophora as a vector for introducingits associated bacterium into the hemocoel of Galleria mellonella larvae.J. Invertebr. Pathol. 33:324–327.

26. Mohamed, M. A., and H. C. Coppel. 1983. Mass rearing of the greater waxmoth, Galleria mellonella (Lepidoptera: Pyralidae), for small-scale labora-tory studies. Great Lakes Entomol. 16:139–141.

27. Paul, V. J., S. Frautschy, W. Fenical, and Nealson. 1981. Antibiotics inmicrobial ecology: isolation and structure assignment of several new anti-bacterial compounds for insect-symbiotic bacteria Xenorhabdus spp.J. Chem. Ecol. 7:589–597.

28. Pharmacia Fine Chemicals. 1980. Percoll methodology and applications.Pharmacia Fine Chemicals, Uppsala, Sweden.

29. Poinar, G. O., Jr. 1975. Description and biology of a new insect parasiticrhabditoid, Heterorhabditis bacteriophora n. gen., n. sp. (Rhabditida: Hetero-rhabditidae n. fam.). Nematologica 21:463–470.

30. Poinar, G. O., and G. M. Thomas. 1966. Significance of Achromobacternematophilus in the development of the nematode DD-136. Parasitology56:385–390.

31. Poinar, G. O., G. M. Thomas, and R. Hess. 1977. Characteristics of thespecific bacterium associated with Heterorhabditis bacteriophora (Hetero-rhabditidae: Rhabditida). Nematologica 23:97–102.

32. Thomas, G. M., and G. O. Poinar. 1979. Xenorhabdus gen. nov., a genus ofentomopathogenic nematophilic bacteria of the family Enterobacteriaceae.Int. J. Syst. Bacteriol. 29:352–360.

33. Thomas, J. O., and R. D. Kornberg. 1978. Chemical cross linking of histones.Methods Cell biol. 18:429–440.

VOL. 67, 2001 PHOTORHABDUS PROTEIN INCLUSIONS 4841

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from