Photorhabdus phase variants express a novel fimbrial locus, mad, essential for symbiosis

Molecular Microbiology (2010) 77(4), 1021–1038 䊏

doi:10.1111/j.1365-2958.2010.07270.x First published online 6 July 2010

Photorhabdus phase variants express a novel fimbrial locus, mad, essential for symbiosis mmi_7270 1021..1038

Vishal S. Somvanshi, Bettina Kaufmann-Daszczuk, Kwi-suk Kim, Shane Mallon and Todd A. Ciche* Department of Microbiology and Molecular Genetics and the Center for Microbial Pathogenesis, Michigan State University, East Lansing, MI 48824, USA.

Summary Fimbriae are adhesive organelles known to enable pathogens to colonize animal tissue, but little is known of their function in mutualistic symbioses. Photorhabdus colonization of Heterorhabditis bacteriophora nematodes is essential for the pair’s insect pathogenic lifestyle. Maternal nematodes acquire Photorhabdus symbionts as a persistent intestinal biofilm prior to transmission to infective juvenile (IJ) stage offspring developing inside the maternal body. Screening 8000 Photorhabdus mutants for defects in IJ colonization revealed that a single fimbrial locus, named mad for maternal adhesion defective, is essential. The mad genes encode a novel usher/ chaperone assembled fimbria regulated by an ON/OFF invertible promoter switch. Adherent Photorhabdus cells in maternal nematode intestines had the switch ON opposite to the OFF orientation of most other cells. A DmadA mutant failed to adhere to maternal intestines and be transmitted to the IJs. Mad fimbriae were detected on TT01 phase ON cells but not on DmadA phase ON cells. Also required for transmission is madJ, predicted to encode a transcriptional activator related to GrlA. Expression of madA–K or madIJK restored the ability of madJ mutant to adhere. The Mad fimbriae were not required for insect pathogenesis, indicating the specialized function of Mad fimbriae for symbiosis.

Introduction Photorhabdus is an enteric bacterium that interacts with nematode and insect hosts with drastically different outcomes (Waterfield et al., 2009). In insects, Photorhabdus is a voracious pathogen killing insects in usually less than Accepted 15 June, 2010. *For correspondence. E-mail ciche@msu. edu; Tel. (+1) 517 884 5359; Fax (+1) 517 353 8957.

© 2010 Blackwell Publishing Ltd

48 h by a combination of septicemia and toxicemia, whereas in Heterorhabditis bacteriophora nematodes, Photorhabdus is a harmless mutualistic symbiont (Ciche et al., 2008). In nature, the pathogenic and mutualistic lifestyles of Photorhabdus are entwined where the ability to infect insects is essential for reproduction of H. bacteriophora nematodes and colonization of nematodes is essential for the symbiont to infect and kill insects. A developmentally arrested stage of the nematode, infective juvenile (IJ), harbour the bacteria in their intestine and regurgitate them in the hemolymph of an insect host, where the bacterium causes insect mortality, grows and produces metabolites essential for nematode reproduction and inhibition of saprophytic microorganisms. Subsequently, in response to a yet unknown signal in the insect hemolymph, IJs exit diapause and grow inside the insect hemolymph to adult hermaphrodites and reproduce (Ciche and Ensign, 2003). The symbiosis is achieved through a selective and sophisticated cell- and stage-specific infection of the nematodes by the symbionts (Ciche et al., 2008). Photorhabdus are maternally transmitted to IJ offspring that develop exclusively inside the maternal body cavity resulting in matricide (the process referred to as ‘transmission’ hereafter). The few symbiont cells that have established a persistent colonization of the posterior intestinal epithelium can be distinguished from the majority of cells transiently present by performing pulse-chase experiments (Fig. 1A). Symbiont transmission is initiated in developing maternal nematodes by the adhesion of Photorhabdus cells to the two most posterior intestinal cells, INT9L and INT9R, at 8–42 h after exiting the IJ stage (Fig. 1B). Subsequently, a persistent biofilm develops in the posterior intestine, and later the symbionts breach the intestinal epithelium by invading the rectal gland cells. After lysis of the rectal gland cells, the symbionts become available to the developing IJs inside the mother’s body and colonize the IJs. Despite these recent insights into the transmission process, the genes and molecular mechanisms involved are largely unknown. To determine the genes involved in the transmission of Photorhabdus from maternal to IJ nematodes, random transposon mutants of GFP-labelled bacteria were screened directly for transmission defects. This study revealed that the maternal adhesion defective (mad)

1022 V. S. Somvanshi et al. 䊏

Fig. 1. The process of Photorhabdus symbiont transmission by Heterorhabditis nematodes. A. The nematode body plan is basically a tube within a tube and is like a flow cell or bioreactor where persistent cells can be detected by pulse-chase experiments. Because the Heterorhabditis intestine can be very permissive for bacteria, many of the ingested cells are visible throughout the intestine (Pulse). To detect persistent cells, nematodes are removed and incubated with non- or differentially labelled symbiont cells (Chase) or starved for 4 h during which time the GFP-labelled transient cells are eliminated. The chased or starved nematodes can be homogenized to isolate persistent colonizing cells. B. Symbiont transmission normally begins with the developmentally arrested IJ that harbours symbiont bacteria (green) in the intestinal lumen. Sensing yet unknown cues in insect hemolymph or lawns of symbiont bacteria the IJs exit diapause and regurgitate and eliminate all intestinal symbiont cells. Then the nematodes ingest symbiont cells where a few adhere and persist (green, transients symbiont cells are not shown) to the posterior intestine and grow as a biofilm. Later, these symbiont cells invade the rectal gland cells and grow inside vacuoles. IJs develop inside the maternal nematode body cavity by intra-uterine hatching and endotokia matricida. The rectal gland cells lyse while the intestine remains intact releasing the symbiont cells to the IJs developing in the maternal body cavity. The symbiont cells now adhere to the pharyngeal–intestinal valve cells and likely invade these cells before exiting and colonizing the IJ intestinal lumen.

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

Phase variation of Mad fimbriae required for symbiosis 1023

fimbrial locus is essential for symbiont transmission, which is one out of 11 predicted fimbrial loci (10 functional, 1 non-functional) present in the Photorhabdus luminescens subspecies laumondii strain TTO1 (hereafter TTO1) genome (Duchaud et al., 2003). The production of fimbriae or pili are known to be important for the colonization of animal tissues or cells (Sauer et al., 2000). However, in contrast to pathogens, examples where fimbriae are known to be important for symbiosis are few (Rendon et al., 2007; Chandra et al., 2008; Lasaro et al., 2009). The study of fimbriae can be complicated because multiple fimbrial loci can be present in the genome, one or more of which may be regulated by phase variation or mutator alleles (Moxon et al., 1994; van der Woude and Baumler, 2004). Phase variation is the ON and OFF expression of genes that is usually regulated by a random and reversible alteration of DNA that results in a phenotypically heterogenous cell population (Henderson et al., 1999). An invertible ON and OFF promoter switch is one mechanism to generate phase variation (Zieg et al., 1977). This study reports the phase variable expression and essential role for the mad fimbrial locus in initiating symbiosis with host nematodes.

Results Isolation of transmission defective mutants GFP-labelled Photorhabdus were used for the mutant screening to facilitate symbiont detection inside the IJ intestine (Fig. 2). Axenic IJs added to the lawns of the mutant symbionts grow and reproduce for two to three generations before IJs develop en masse. IJs exhibit dispersive behaviour by usually moving up the side of the dish onto the lid. Some IJs become trapped in condensate present on the lids where they are easily examined for the presence of symbiont bacteria using a stereofluorescent microscope. In 28 independent mutagenesis experiments, a total of c. 7250 transposon mutants of GFP-labelled Photorhabdus temperata NC1 (NC1) and in three experiments, 750 P. luminescens ssp. laumondii TT01 (TT01) mutants were screened for defects in transmission (TRN) to IJ nematodes. Most of the TRN mutants obtained were in strain NC1 because it was initially more amenable to transposon mutagenesis than TT01. Strain NC1 is transmitted to the same extent as TT01 when in a monoxenic association with H. bacteriophora TT01 nematodes (Table 1; Ciche et al., 2008). Because of the availability of the complete genome sequence, most additional genetic manipulations were performed in TT01 (Duchaud et al., 2003). Twenty-nine mutants of NC1 and one mutant of TT01 were isolated with defects in transmission by IJ nematodes, which had HimarGm insertions in a predicted

Fig. 2. Scheme used for the identification of Photorhabdus transmission deficient mutants. 1. GFP-labelled Photorhabdus were mutated with HimarGm transposon and 2. HimarGm mutants selected for on LBPGm. The mutants were arrayed in grids (3), liquid cultures grown (4) and spread on NA + CO GM (5) and grown overnight (6). IJ nematodes were added (7) and grown 10–14 days (8). Dispersing IJs trapped in condensate present on lids were screened for the presence of intestinal GFP-labelled cells using fluorescence microscopy (9). Mutants that failed to be transmitted (TRN) were isolated (10) and the TRN phenotype verified.

type 6 secretion system (plu2287), type 3 secretion system inner membrane protein SctV (plu3761), stringent starvation protein A (plu4013), sodium/proton antiporter 1 nhaA (plu0587), high affinity branched chain amino acid transport system permease protein livM (plu4096), putative transcriptional regulator abgR (plu3727), few genes encoding hypothetical proteins of unknown function, e.g. plu4602, plu4509, plu1183, and one intergenic region between plu2109 and plu2110. Remarkably, 12 of the 30 mutants, including the TT01 TRN26–25 mutant, contained HimarGm transposon insertions in a single predicted fimbrial or pilin locus (TT01 genome locus tags plu0260–0270) out of 11 predicted fimbrial loci present in the TT01 genome (Fig. 3, Table 1) (Duchaud et al., 2003). The NC1 insertions were located

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

1024 V. S. Somvanshi et al. 䊏

Table 1. Transmission efficiencies of P. temperata NC1 and P. luminescens TT01 TRN mutants and approximate transposon insertion sites.

Strain

Mutant gene

Locus taga

Gene size (bp)

Insertion site (nt)

Transmission efficiency (SD)b

NC1GFP TT01GFP TRN359 TRN583 TRN5–60 TRN10–245 TRN16–158 TRN17–60 TRN18–110 TRN19–124 TRN22–68 TRN23–30 TRN24–55 TRN26–25 (TT01)

n.a. n.a. madA madH madB madF madE madG madH madE madI madF madE madJ

n.a. n.a. plu0261 plu0268 plu0262 plu0266 plu0265 plu0267 plu0268 plu0265 plu0269 plu0266 plu0265 plu0270

n.a. n.a. 606 2631 741 720 657 795 2631 657 1326 720 657 501

n.a. n.a. 225 1710 573 717 n.d. n.d. n.d. 60 102 n.d. 636 147–174

99.5 (0.5) 98.5 (0.5) 0 0 0 0 0 0 0 0 44.5 (7.4) 0 2.3 0

a. Of the P. luminescens ssp. laumondii strain TT01 genome sequence. b. Percent of IJs colonized by Photorhabdus N > 300; for zero, no Photorhabdus were detected for > 5000 IJs examined. SD in parenthesis denotes standard deviation. n.a., not applicable; n.d., not precisely determined.

in the orthologous locus. Ten of these 12 mutants exhibited a total defect in transmission with zero bacterial cells detected in at least 5000 IJs analysed and zero transmission efficiency (defined as percent IJs colonized by one or more symbiont cells), compared with transmission efficiencies of 99.5% (995 colonized/1000 total IJs) and 98.5% (543 colonized/551 total IJs) for NC1GFP and TT01GFP respectively. Mutants TRN22–68 and TRN24–55 had 44.5% and 2.3% transmission efficiencies, respectively, both of which are significantly less than wild type. In summary, the genetic screen clearly identified a single predicted fimbrial locus as being essential for symbiont transmission from maternal nematode to the IJs. Characteristics of the mad fimbrial locus Sequence analysis of the TT01 mad locus revealed that it is predicted to encode a novel usher/chaperone assembled fimbrial organelle and three hypothetical proteins of unknown function all present downstream of an invertible ON/OFF invertible promoter switch (Fig. 3, Table 2). The mad locus is predicted to encode three chaperones, MadB, F and G, which is uncommon for usher/chaperone assembled fimbriae and common to the g-2 clade of fimbrial usher protein secretion pathways of which 987P fimbriae produced by enterotoxigenic Escherichia coli is the most characterized (Edwards et al., 1996; Nuccio and Baumler, 2007). The mad locus is distinct from the fas locus encoding 987P fimbriae and other characterized fimbrial loci in containing seven fimbrial and chaperone genes upstream of the usher madH (Fig. 4). Furthermore, a mad homologue of fasG, which encodes for the 987P adhesin or any other significant homologue to a known adhesin gene was not detected (Khan and

Schifferli, 1994). Two uncharacterized fimbrial loci related to the mad locus were found in the genomes of uropathogenic Proteus mirabilis HI4320 (PMI 2539-2532) and plasmid pSE11-3 (ECSE_PS_0066-0061) of E. coli SE11 isolated from a healthy person (Oshima et al., 2008; Pearson et al., 2008). Three genes, madIJK are located 169 bp downstream of madH and are predicted to encode hypothetical proteins of unknown function, although MadJ contains domain of unknown function (DUF)1401 that is also present in the transcriptional activators CaiF and GrlA (Buchet et al., 1999; Huang and Syu, 2008) (Fig. 3, Table 2). Because two transmission defective mutants contained insertions in either madI or madJ, the function of one or more of these hypothetical proteins is required for transmission. Immediately upstream of madA, predicted to encode the major subunit, is an ON/OFF invertible promoter switch (called madswitch hereafter) consisting of two 36 bp inverted repeats separated by 257 bp and downstream of a FimB-like tyrosine site-specific recombinase encoding gene madR (Fig. 3, Table 2). This is predicted to function as an ON/OFF invertible promoter switch, where a promoter present between the two inverted repeats can flip orientation towards and away from the mad locus catalysed by the MadR and additional site-specific recombinases (e.g. that encoded by plu1991) (Zieg et al., 1977; van der Woude and Baumler, 2004). This predicts that the mad locus is a mutable or a contingency locus that is regulated by a chance recombination event needed to orientate the promoter for expression of the mad genes in a subset of Photorhabdus cells (Moxon et al., 1994). The genomic context of the mad locus is intriguing because it is located between recG and spoT (Figs 3 and

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

Phase variation of Mad fimbriae required for symbiosis 1025

A

mad locus recG

A B C D E F G

R

madH

(w) (+) (+)(+) (+) (+) (+) (+) 274,000

B

276,000

278,000

PON

36 bp Inverted Repeat

280,000

282,000

(+)

284,000

C

36 bp Inverted Repeat

(+)(+)

286,000 288,000

290,000

MadE Adhesin MadD MadC

A

R

madI J K spoT

257 bp

MadH Usher

+ MadR

MadA Major subunit

O.M. 36 bp Inverted Repeat

POFF

36 bp Inverted Repeat

A

R

periplasm =

MadB +

MadF

MadG

Chaperones I.M.

Fig. 3. P. luminescens TTO1 mad fimbrial locus. A. ORFs disrupted by transposon insertion are shown by lollipops. The axis represents genomic position. The (+) symbols below the intergenic regions represent cotranscription of genes; (w) shows weak signal. B. The madswitch invertible ON/OFF promoter upstream of madA. C. Hypothetical model for assembly of Mad fimbrial proteins on the basis of known sequence information. MadB, MadF and MadG are chaperons, MadH usher, MadA, MadC and MadD are fimbrial subunits, and MadE might be the adhesin.

4). This is unusual because most enteric bacteria, including Xenorhabdus nematophila (a genus closely related to Photorhabdus that also shares an insect pathogenic and nematode symbiotic lifestyle), have spoT, trmH (spoU),

recG as a conserved fragment of DNA. No homologue of trmH was detected in the TT01 genome. Mad fimbrial operons are present in two other species, P. temperata NC1 and P. asymbiotica ATCC43949 both of

Table 2. mad fimbriae locus and predicted functions of the ORFs. Name (locus tag)

Size (aa)

Description

Pfam motifs detected (Pfam no., E-value)

MadR (plu0260) MadA (plu0261) MadB (plu0262) MadC (plu0263) MadD (plu0264) MadE (plu0265) MadF (plu0266) MadG (plu0267) MadH (plu0268) MadI (plu0269) MadJ (plu0270) MadK (plu0271)

188 201 246 190 240 218 239 264 876 441 166 169

FimB tyrosine Fimbrial subunit Chaperone Fimbrial subunit Fimbrial subunit Unknown Chaperone Chaperone Usher Unknown Unknown Unknown

Phage_integrase (PF00589, 1e-43) Fimbrial (PF00419, 0.0016) Pili_assembly_N & C (PF00345 & PF02753, 7.5e-30 & 5.4e-06) Fimbrial (PF00419, 0.00093) Fimbrial (PF00419, 0.00057) None detected Pili_assembly_N & C (PF00345 & PF02753, 3.1e-27 & 4.4e-22) Pili_assembly_N & C (PF00345 & PF02753, 3.4e-23 & 3.1e-5) Usher (PF00577,0) None detected DUF1401 (PF07180, 0.002) None detected

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

1026 V. S. Somvanshi et al. 䊏

Fig. 4. Novel characteristics of the mad fimbrial locus. Fimbrial loci were aligned with respect to their usher encoding genes (e.g. fimD top, madH bottom). The open reading frames are depicted as arrows whose fill corresponds to their respective protein families. Gene name is located below the gene and % amino acid identity above (shown % identity > 25%). Type 1, 987P and ECSE_PS_0066-0061 are from E. coli and MAD from Photorhabdus strains. The TT01 mad locus (bottom) is inserted between spoT and recG in the place of trmH, usually present as a three-gene cluster in most Gammaproteobacteria, as it is in X. nematophila (top). The mad locus shows low amino acid identity to assembly proteins (e.g. 31–44% for ushers) and even less for predicted subunits (all but one < 25% identity). The order and content of the mad fimbrial genes is distinct from other known fimbrial loci.

which are associated with Heterorhabditis nematodes (Ciche and Ensign, 2003; Gerrard et al., 2006) (Fig. 4). The sequence of the NC1 mad locus was determined by sequencing DNA flanking the HimarGm insertions of NC1 TRN mutants and by NC1 genome sequence kindly communicated by Dr Lou Tisa, University of New Hampshire. NC1 bacteria, which can colonize H. bacteriophora TT01 nematodes and are associated with H. bacteriophora NC1 nematodes, show high degree of amino acid sequence identity (> 80% identity) in all Mad fimbrial proteins except for MadA (38%). In contrast, 43949 bacteria, which colonize Heterorhabditis gerrardii and are not known to colonize H. bacteriophora, show a high degree of similarity of usher/chaperone assembly genes (> 75% identity), are more divergent in MadC, D, E and I (< 55% identity). MadC, D and possibly E are predicted to be secreted

minor fimbrial proteins, and divergence of these may indicate divergence and host specificity of the TT01/NC1 and 43949 strains to different nematode species. Recently, the mad locus was identified as being absent from P. temperata ssp. temperata XINach, a bacterium associated with Heterorhabditis megidis and not H. bacteriophora nematodes (Gaudriault et al., 2006). However, because 43949 contains the mad locus and was then not known to be nematode associated, it was ruled out as host specificity determinant. These data suggest that the mad locus might function to determine host specificity of the symbiosis. Because mutations affecting transmission were isolated in most genes present in the mad locus and an ON/OFF invertible promoter switch is located upstream of madA, cotranscription of the mad genes was determined (Fig. 5).

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

Phase variation of Mad fimbriae required for symbiosis 1027

Fig. 5. Cotranscription of mad genes indicating that madA–K are expressed as an operon (B–K). The primers were designed to amplify intergenic regions of mad genes using the reverse transcribed RNA (cDNA) as template. madR/madA indicates intergenic region of madR and madA, etc. Controls included DNase I-treated but not reverse transcribed RNA and genomic DNA. PCR products were analysed on a 1% agarose gel. All the PCR products were approximately 500 to 700 bp (DNA ladder not shown). NTC is non-template control.

Except for the intergenic region between madR and madA predicted to contain the invertible promoter switch, madA–madK appear to be cotranscribed. Determination of the defective transmission step of the TRN mutants Transmission of symbiont bacteria in nematodes occurs maternally by a sophisticated multistep process being initiated by adhesion of symbiont cells to the posterior intestine of maternal nematodes (Fig. 1B; Ciche et al., 2008). This is partially explained because the symbionts must breach the maternal intestinal lumen to gain access to IJs that develop inside the maternal body cavity. Pulsechase adherence assays (Fig. 1A) revealed that the TRN mutants containing mutations in the mad locus that were totally defective in transmission were also totally defective in adhesion to and persistence on the posterior intestinal cells (e.g. madJ, Fig. 6B and madA, Fig. 7B). The genes

Fig. 6. Complementation of the adhesive phenotype of TRN26–25 in maternal nematode intestines (A–D) are pulse-chase experiments where only persistent GFP-labelled cells are green and the worms are oriented such that anterior is left, ventral is down, left is out of the plane of the page. A. GFP-labelled TT01 forms a biofilm in the posterior intestine of maternal nematodes. B. No biofilm is visible in TRN26–25 (madJ) mutant. C and D. Expression of the pmadA–K and pmadIJK restores the ability of TRN26–25 to adhere to the posterior intestine (C and D respectively).

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

1028 V. S. Somvanshi et al. 䊏

Fig. 7. Non-colonizing and non-transmitting phenotype of DmadA mutants. A–D. image overlays of Nomarski and epifluorescence (exposure of 800 msec) micrographs of the nematode intestine. Nematodes are oriented anterior-left, ventral-down, left-out of the page of the plane (A and B) and rotated 90° right for (C and D). (A) TTO1 forms persistent biofilm (green) on the posterior maternal nematode intestine at 32–34 h. (B) DmadA mutant shows no colonization. (C) IJ intestine colonized by TTO1. (D) No bacteria observed in IJs grown on DmadA mutant. E. Number of bacteria colonizing 1000 IJs reared on TTO1 as compared DmadA mutants. The pharynx (p), intestine (i) and rectum (r) are indicated.

of this putative fimbrial locus were named mad for maternal adhesion defective. In summary, the putative fimbriae encoded by the mad locus are required for initiation of symbiosis by colonization of the posterior maternal intestinal cells. Complementation of TRN26–25 (madJ) Complementation experiments were conducted using NC1 mutants TRN359 and TRN583 and TT01 mutant TRN26–25 (with transposon insertions in madA, madH and madJ, respectively) containing either the complete (pmadA–K) or partial (pmadIJK) TT01 genes. The ability of TRN26–25 to adhere and persist in the maternal intestine was restored by the complete (pmadA–K) or partial (pmadIJK) gene cluster (Fig. 6). As expected, TT01 bacteria (positive control) were able to fully colonize H. bacteriophora, while TRN26–25 was not able to initiate symbiosis in the maternal nematode intestine. However, IJ nematodes grown on lawns of TRN26–25 pmadA–K and TRN26–25 pmadIJK did not harbour symbiont cells.

This was likely due to plasmid loss by symbiont cells inside nematodes, as Photorhabdus does not retain the plasmids efficiently (more than 60% cells lost the broad host range plasmid pBBR1MCSGFPccdABflmAB overnight under selection, data not shown). Complementation experiments of TRN359 and TRN583 with pmadA–K were unsuccessful, likely because we later determined that this construct had the madswitch in the phase OFF orientation (see below). A secondary promoter may allow enough madIJK to be expressed from pmadIJK or pmadA–K to complement TRN26–25. In summary, the ability of a TT01 madJ mutant (TRN26–25) to adhere to the maternal intestine was restored by complementation with madA–K and madIJK expressed in trans. Determination of the role of TT01 madA in transmission It is possible that mad fimbriae are not required for transmission because the madA–K genes are cotranscribed and only madJ was successfully complemented. This was tested by constructing an in-frame and markerless

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

Phase variation of Mad fimbriae required for symbiosis 1029

Role of the mad locus for pathogenicity to Manduca sexta insects Insect virulence assays for NC1 and TRN359 (madA::HimarGm) were conducted by injecting three dosages (10, 100 and 1000 cfu) of the bacterial strains into the hemocoel of tobacco hornworm, M. sexta larvae, and observing the insect survival at 24, 48 and 72 h. No significant (P = 0.05) difference in insect mortality was observed by the NC1 and TRN359 at all three dosages (Fig. 8A). At the highest dose of 1000 cfu, both the

A Percent survival

100

24 h

48 h 72 h

80 60 40 20 0

10

100 1000 CFU NC1

10

100 1000 CFU TRN359

B 100 Percent survival

deletion of TT01 madA, predicted to encode the major subunit. First madA was replaced with a mutant allele DmadA::GmR by allelic exchange as described in Experimental procedures. Next, this madA::GmR was made non-polar and markerless (designated DmadA) by expression of Flp site-specific recombinase. First, we tested the DmadA::GmR strain for the ability to be transmitted to IJs. No DmadA::GmR cells were detected by fluorescence microscopy or homogenizing and plating IJs nematodes (data not shown). This demonstrated that a madA mutation first identified in NC1 also causes a complete defect in transmission in strain TT01. Because this could be explained because of polar effects of DmadA::GmR onto downstream genes like madJ, most additional analyses focused on the DmadA strain. RT-PCR experiments were performed to determine that the DmadA::GmR mutation is polar in inhibiting transcription of madB/madC and madG/madH, while the DmadA mutation is non-polar and restored in transcription of madB/madC and madG/madH (Fig. S1). The DmadA mutant was identical to TT01 DmadA::GmR and NC1 TRN359 (madA::HimarGm) in its complete inability to adhere and initiate symbiosis in the maternal nematode intestines and to be transmitted to the IJ nematodes (Fig. 7A and B). All (57/57) maternal nematodes reared on DmadA for 32 h totally lacked persistent GFPlabelled bacteria, whereas all (51/51) nematodes cultured on TT01 contained a persistent biofilm on the posterior intestine. Next the ability of DmadA to be transmitted to IJs was determined. No intestinal symbionts were detected in the 1052 IJ nematodes associated with DmadA examined by fluorescence microscopy for GFP-labelled bacteria, whereas 151/156 IJs nematodes associated with TTO1GFP contained fluorescent bacteria in their intestines (Fig. 7C and D). Moreover, plate counts of 1000 homogenized and surface sterilized IJs grown on DmadA resulted in zero colonies. In contrast, c. 990 000 colonyforming units (cfu) were detected from similarly treated IJs propagated on lawns of TT01 (Fig. 7E). In summary, Mad fimbriae are absolutely required for symbiont transmission by IJs at an initial step in the transmission process, adhesion to the maternal nematode intestine.

NC1 TRN359

80 60 40 20 0 0

12

24

36 Hours

48

60

72

Fig. 8. Virulence of P. temperata NC1 (wt) and TRN359 (DmadA::HimarGm) to Manduca sexta. A. Bar graphs showing percent insect mortality at 10, 100 and 1000 cfu of bacterial strains at 24, 48 and 72 h. B. Kaplan-Meier survival curves showed non-significant (P = 0.6333) difference in the insect survival on wild-type and the mutant at the 1000 cfu dose.

mutants and the wild type showed 25% insect survival at 72 h (Fig. 8A). Characterization of insect survival at all three time points and all three dosages by Kaplan-Meier survival analysis revealed non-significant (P > 0.05) difference for wild type and mutant strains; an example survival curve at 1000 cfu is shown in Fig. 8B. The absence of virulence defect to insects suggests that that mad fimbriae have a specialized function in Photorhabdus–Heterorhabditis symbiosis, but not in the virulence of the Photorhabdus to the insects. A previous study identified the mannose resistant fimbrial locus, mrf, as being expressed by P. temperata K122 during insect infection (Meslet-Cladiere et al., 2004). TT01 contains an orthologous mrf locus that may also be expressed during insect infection (Duchaud et al., 2003). Phase variable expression of mad locus Symbionts adhere to the posterior of the maternal intestine at a rate of c. one cell per hour even though the

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

1030 V. S. Somvanshi et al. 䊏

maternal intestine is permissive for viable bacteria (Ciche et al., 2008), which shows that adhesion is a rare event. In addition, the madswitch located immediately upstream of madA suggests that this locus is expressed by phase variants that have the promoter oriented with madA. To determine if the madswitch is flipped ON in symbiont cells adherent to the maternal intestine, the orientation was analysed in TTO1 from 48 h old plate cultures and TT01 cells isolated from the persistent biofilm present on the maternal intestine. Persistent biofilm cells were isolated by homogenizing maternal nematodes associated with GFP-labelled TT01 for 36 h followed by starvation for 4 h during which time the labelled transients were expelled (Fig. 1A). Indeed, the madswitch was flipped in cells adherent to the maternal intestine as compared with TT01 cells present on 48 h LBP agar (Fig. 9A and B). TT01 cells adherent to the maternal nematodes yielded SfcI fragments of 753 and 300 bp while TT01 from LBP agar yielded SfcI fragments of 925 and 128 bp, which is the same as would be predicted from the TT01 genome sequence. Thus, the madswitch is flipped in TT01 cells that established a persistent biofilm in the maternal intestine and is oriented opposite to most TT01 cells on the LBP lawn and those cells used for determining the TT01 genome sequence. Because a minority of the cells adhered to the maternal nematode intestine and had the invertible promoter in an opposite orientation to TT01, it is likely that the TT01 orientation is OFF and the phase variant orientation is ON for mad expression. This was tested by cloning the madswitch from TT01 cells from LBP lawns and TT01 cells from the persistent biofilm in maternal nematodes upstream of a promoterless GFP reporter (Miller et al., 2000). GFP was significantly (P = 0.05) highly expressed [relative fluorescent units (RFU) = 1690] from the madswitch in the predicted ON orientation (753 /300 bp SfcI restriction bands) in E. coli cells (Fig. 9E) as compared with madswitch in the predicted OFF (925/128 bp SfcI bands) orientation (RFU = 626, Fig. 9F), the negative control cells (RFU = 669, Fig. 9C) and the positive control cells (1270 RFU, Fig. 9D). Expression of madswitch ON by TT01 and E. coli was comparable (Fig. 9E and G) suggesting that transcription factors present in both are sufficient for expression of the madswitch. These results suggest that TT01 cells with the capacity to adhere to the maternal intestine have undergone phase variation of the mad locus and that the madswitch is flipped to the ON orientation allowing expression of the mad fimbriae adhesive organelle. We then sought to determine if madswitch flipped ON in TT01 cells in culture as the plates aged. Remarkably, in cells isolated from 60-day-old TT01 plate cultures, the madswitch was flipped ON like the cells that established the persistent biofilm on the maternal nematode intestine. Thus, TT01

cells with the mad phase ON can be isolated from cells adhering to the maternal intestine and from aged LBP plates. Detection of Mad fimbriae on madswitch phase ON cells Because we could isolate TT01 and DmadA cells with the madswitch phase in both ON and OFF orientations, we examined these cells for fimbriae by transmission electron microscopy with expectation that TT01 and not DmadA cells will produce Mad fimbriae when the madswitch is ON while both TT01 and DmadA will express different fimbriae when madswitch is OFF. As mentioned above, most cells have the madswitch OFF during routine culturing (e.g. cells grown for 48 h at 28°C on LBP agar). TT01 cells with the madswitch ON were isolated from colonies that arose from the persistent biofilm present in the maternal nematode intestine. Because of the inability of DmadA to colonize maternal nematodes, DmadA cells with the madswitch ON were isolated from colonies that arose from 60-day-old LBP plates (incubated at room temperature). As predicted both TT01 and DmadA phase OFF cells (from 48 h old plates) expressed other fimbriae, which are possibly mannose resistant fimbriae, Mrf (Meslet-Cladiere et al., 2004) (Fig. 10A and B). In contrast, the presence of long fimbrial fibres radiating from the cells were detected on TT01 madswitch ON cells, but were absent from DmadA madswitch OFF cells (Fig. 10C and D). Analysis of surface proteins produced by TT01 and DmadA madswitch ON and OFF cells revealed a protein corresponding to the predicted size of the MadA that was produced only by TT01 madswitch ON cells (Fig. 10E). Gel extraction of this band and protein identification by liquid chromatography and tandem mass spectrometry revealed the identity of the most abundant protein to be MadA. This suggests that the mad fimbrial locus does encode for a fimbrial adhesive organelle that is regulated by phase variation. Phenotypic characteristics of mad mutants Because phenotypic variants occur in Photorhabdus that are unable to colonize IJ nematodes, phase variable characteristics were determined for TTO1 and TRN26–25 and DmadA mutant strains (Table 3). All the strains absorbed dyes, produced crystal proteins, showed antibiotic activity against Micrococcus luteus, produced siderophores and exhibited lipase activity indicating that the strains are all of the primary phase phenotype. The growth rates of the mutants on LBP were also the same as the wild type. However, strain TRN26–25 (madJ) exhibited significantly more (P = 0.05) swimming and swarming motility as

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

Phase variation of Mad fimbriae required for symbiosis 1031

Fig. 9. Flipping of the madswitch and determination of which phase corresponds to ON expression of mad locus. A. Diagrammatic representation of flipping madswitch and SfcI restriction sites. Arrowheads indicate inverted repeats left and right (IRL, IRR); sizes of restriction products are shown in both orientations of the switch. B. Picture of the 1.4% agarose gel showing SfcI digested madswitch fragments in both orientations. Lane 1 shows orientation of the madswitch in 48 h old TT01 cells (same as the TT01 genomic sequence), lanes 2 and 3 show the orientation of the madswitch flipped in TT01 cells isolated from the persistent biofilm present in maternal nematodes, and from cells isolated from 60-day-old LBP plates respectively. Sizes of DNA bands are indicated. C–G. Grayscale image overlays of Nomarski and GFP epifluorescence (exposure of 50 msec) of E. coli cells containing (C) promoterless GFP (negative control), (D) GFP driven by promoter nptII (Kan) (positive control), (E) GFP fused to madswitch in predicted ON position and (F) GFP fused to madswitch in predicted OFF position. (G) TT01 cells containing GFP fused to madswitch ON position. Whiter cells indicate brighter fluorescence; cells with madswitch ON were overexposed and falsely appear to be of a larger cells size.

compared with the TTO1 and DmadA. Interestingly, TRN26–25 and DmadA produced significantly more biofilm on polystyrene plates compared with TTO1. In summary, mutations in mad fimbrial locus did not affect most of the phenotypic variable characteristics of Photorhabdus.

Discussion The goal of this study was to gain insights into how symbiotic bacteria selectively associate with and are transmitted by animal hosts and how these processes compare with pathogenesis. Both mutualism and pathogenesis can

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

1032 V. S. Somvanshi et al. 䊏

Fig. 10. TEM electron micrographs of negatively stained bacterial cells. Selected fimbriae are indicated by arrows. A and B. Fimbriae are visible in madswitch phase OFF wild-type cells (A) and madswitch phase OFF DmadA (B) suggesting that both express fimbriae different from that encoded by the mad locus (i.e. mrf). C. Fimbriae are visible on TTO1 madswitch phase ON cells isolated from persistent biofilms formed on maternal nematode intestines. D. No fimbriae are visible on madswitch phase ON DmadA cells isolated from 60-day-old LBP cultures. E. Analysis of surface proteins produced by madswitch phase OFF TT01 (lane 1) and DmadA (lane 2), and madswitch phase ON TT01 (lane 3) and DmadA (lane 4) revealed a band of the predicted size of MadA unique to madswitch ON TT01 (arrow), which was identified as MadA by LC and tandem mass spectrophotometry. M.W., molecular weight standards in kDa.

be studied and compared in Photorhabdus. Furthermore, individual GFP-labelled symbiont cells are easily detected inside living nematodes, and large numbers of Photorhabdus mutants can be rapidly and inexpensively screened for transmission defects. Recently, a description of the sophisticated process by which symbiont cells are maternally transmitted to IJs developing inside the maternal body cavity revealed the crucial steps that could be defective during the transmission process (Ciche et al., 2008). The main result of this study was the identification of the

mad fimbrial locus that is essential for symbiont transmission from the maternal nematodes to the IJs and that it is regulated by phase variation. Specifically, mad genes are required for an early initiation step of transmission where symbionts adhere to INT9R and L cells in the maternal nematode intestine. The mad locus and other loci involved in transmission identified here were not identified by a recent study by the Clarke laboratory that identified genes involved in LPS biosynthesis, several of which also had virulence defects (Easom et al., 2010). Fimbriae or pili surface organelles are well-known colonization factors produced by many pathogenic bacteria (Ofec et al., 2003). Fimbriae often function as adhesive organelles allowing bacteria to recognize receptors (e.g. glycans) present on host cells and to initiate infection (Kline et al., 2009). Remarkably, few instances are known where fimbriae are used by symbiont bacteria to adhere to host tissue [e.g. for commensal E. coli (Rendon et al., 2007; Lasaro et al., 2009)]. The study of fimbriae in host– bacterial interactions is often complicated by the fact that some bacteria have several predicted fimbrial loci, and these can be regulated by stochastic events such as phase variation (Holden and Gally, 2004; van der Woude and Baumler, 2004). Indeed, the Photorhabdus genome consists of 11 gene clusters predicted to encode proteins related to known pili and fimbriae biosynthesis (Duchaud et al., 2003), three that are likely expressed by phase variation. A simple explanation for the accumulation of madswitch ON cells on the persistent biofilm on the maternal nematode intestine is that the switch needs to be ON so that the cells can express the Mad fimbrial adhesive organelle required to bind to nematode host receptors. More perplexing is the fact that madswitch ON cells were found to accumulate during aging on LBP plates. This may be due to regulation of switching frequency or selection for one phase under certain environmental conditions. Both mechanisms were recently shown to influence the accumulation of fimbriated P. mirabilis and E. coli cells under oxygen limiting condition (Lane et al., 2009). It remains to be determined if these or other mechanisms cause the accumulation of madswitch ON phase during aging of cells on LBP plates. Thus, screening directly for the transmission defective phenotype likely facilitated the discovery of the Mad fimbriae required for symbiosis. The mad locus genes belongs to the g2-clade of fimbrial operons that are assembled by the classical usherchaperone pathway, and are most similar to the csw, fot and fas fimbrial operons of E. coli (Nuccio and Baumler, 2007). The presence of three chaperones is a unique feature of 3/11 TTO1 fimbrial operons, including the mad operon in addition to the two other TT01 fimbrial operons present in the genome as a tandem duplication (plu0779– 0785 and plu0786–0792) (Duchaud et al., 2003).

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

Phase variation of Mad fimbriae required for symbiosis 1033

Table 3. Phenotypic characteristics of TT01, DmadA and TRN26–25 strains. P. luminescens TTO1 strains Characteristics

TT01 (wt)

DmadA

TRN26–25 (madJ::HimarGm)

Dye absorption Congo Red Methylene Blue Bromothymol Blue MacConkey

+ + + +

+ + + +

+ + + +

Crystal Protein production Swimming motilitya Swarming motilitya Antibiotic production b Lipase Biofilmc Specific growth rate (m) Siderophore productiond

+ + (6.5 ⫾ 0.03) + (1.1 ⫾ 0.1) + (2.9 ⫾ 0.1) + + (0.147 ⫾ 0.05) 0.198 + (2.5 ⫾ 0.1)

+ + (5.8 ⫾ 0.1) + (1.2 ⫾ 0.03) + (2.7 ⫾ 0.3) + + (0.399 ⫾ 0.03)* 0.192 + (1.4 ⫾ 0.07)*

+ + (7.9 ⫾ 0.3)* + (6.0 ⫾ 0.2)* + (2.9 ⫾ 0.07) + + (0.393 ⫾ 0.01)* 0.198 + (2.3 ⫾ 0.2)

a. Diameter of the swimming/swarming motility zone in cm, mean of 2 replicates ⫾ standard deviation of mean. b. Diameter of the M. luteus inhibition zone (cm), mean of 2 replicates ⫾ standard deviation of mean. c. OD590. d. Diameter of the iron utilization zone (cm), mean of 2 replicates ⫾ standard deviation of mean. * Significantly (P = 0.05) different from wild-type TTO1.

However, the predicted fimbrial subunits are highly dissimilar to known fimbrial proteins suggesting that they have divergent function. This is not unexpected as Photorhabdus adheres to only two intestinal cells at only a narrow window of nematode development and not to the Caenorhabditis elegans intestine (Ciche et al., 2008). Possible explanations for this high degree of specificity for symbiont adhesion to the maternal INT9 cells is that a unique receptor is expressed by these cells when adhesion is permitted or barriers to adhesion are present except when adhesion is permitted. It is of interest that non-fimbrial genes cotranscribed with the mad fimbrial genes are also required for adhesion, especially madJ, which may function as a transcriptional activator. MadJ is predicted to contain DUF1401 that is also present in the transcriptional activator of Ler, GrlA (Deng et al., 2004). A homologue of GrlA, SGH (t2936/STY3173), the function of which is currently unknown was detected immediately downstream and in the same orientation of the std fimbrial operon required for intestinal persistence of Salmonella enterica serovar Typhi in mice (Townsend et al., 2001; Weening et al., 2005). In P. mirabilis, coexpression of mrpJ and homologues with at least 10 fimbrial loci is employed to repress motility during fimbrial expression (Li et al., 2001; Pearson and Mobley, 2008). Our finding that a madJ mutant exhibited an increase in motility suggests that it also represses motility when the mad locus is expressed, which might facilitate Mad fimbriae mediated adhesion of symbionts to the intestinal cells. In order to adapt to changing environments, a variety of bacteria undergo phase variation, also known as high

frequency ON/OFF switching of phenotype expression (Henderson et al., 1999). For example, expression of fimA in E. coli is regulated by an invertible region containing the promoter (Abraham et al., 1985). We determined that the mad locus is regulated by phase variation and that cells with the phase ON can be isolated from cells adhering to the maternal nematode intestines or from aged agar media. Although, fimbriae were detected on wild type but not DmadA phase ON cells, additional experiments are required to determine the phase variable regulation of the mad locus. In summary, genetic analysis of Photorhabdus transmission by nematodes revealed a single fimbrial locus, mad, is essential for the maternal transmission of symbionts. The mad locus is divergent from other known fimbrial loci suggesting that it is specialized for symbiont adhesion to host maternal nematode cells. The absence of a virulence defect to insects further supports the hypothesis that mad fimbriae have a specialized function for symbiosis. Future studies of the mad locus and encoded proteins should illuminate how these fimbriae function in symbiosis and why an essential process in the life cycle of the Photorhabdus–Heterorhabditis insect pathogenic symbiosis is regulated by a chance phase variation event.

Experimental procedures Bacterial strains and media List of strains and plasmids used in the study is provided in Table 4. Chemicals were obtained from Sigma-Aldrich (St. Louis, MO) unless otherwise indicated. E. coli was grown at

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

1034 V. S. Somvanshi et al. 䊏

Table 4. Strains and plasmids used. Strains or plasmid

Genotype/relevant characteristics

Source/reference

Nematode M31e

Heterorhabditis bacteriophora strain TT01, inbred line M31e

Ciche and Sternberg (2007)

P. luminescens ssp. laumondii strain TTO1 symbiont of H. bacteriophora TT01 containing Tn7-GFP P. temperata NC1 symbiont of H. bacteriophora NC1 containing Tn7-GFP Transmission defective NC1 mutant (fails to colonize IJ nematodes) NC1GFP madA::HimarGm (maternal adhesion defective) TT01GFP madJ::HimarGm mutant T01GFP madJ::HimarGm mutant + pmadA–K TT01GFP madJ::HimarGm mutant + pmadIJK TTO1 madA::Frt-GmR –Frt TT01 DmadA::Frt

Nematode host

Bacteria Photorhabdus TT01 TT01GFP NC1 NC1GFP TRN16 TRN359 TRN26–25 TRN26–25 + pmadA–K TRN26–25 + pmadIJK DmadA::GmR DmadA E. coli BW29427 EC100 pir-116 BW29427 + pURE10 DH5a Plasmids pURE10 pCR-XL-TOPO pmadA–K pmadIJK pST98As pCP20 pPS856 pCRII pPROBE-GT pPROBE-GTKan pPROBE-GT PmadswitchON pPROBE-GT PmadswitchOFF

dap auxotroph, tra, pir

Ciche et al. (2008) Nematode host Ciche et al. (2008) Ciche et al. (2008) This This This This This This

study study study study study study

R6Kg ori cloning strain Suicide plasmid containing Mini-HimarGm Chemically competent plasmid maintenance strain

K. A. Datsenko and B. L. Wanner, Purdue University Epicentre Biotechnologies D. Lies/D. Newman, Caltech Lab stock

Mini-HimarGm transposon, C9 transposase, R6k ori, GmR, ApR TOPO cloning vector pCR-XL-TOPO containing madA–K from TTO1 pCR-XL-TOPO containing madIJK from TTO1 TetR repressed ISceI, suicide plasmid Flp recombinase under control of bacteriophage l rightward promoter ts, l cI857 ts ori, CmR, ApR GmR flanked by Flp recombinase target (FRT) sites PCR cloning vector Promoter-probe vector with GFP reporter, GmR Constitutively GFP expressing vector, GmR pPROBE-GT vector with madswitch in ON orientation pPROBE-GT vector with madswitch in OFF orientation

D. Lies and D. Newman, Caltech Invitrogen This study This study Posfai et al. (1999) E. coli Genetic Stock Center, Yale University, CT, USA Hoang et al. (1998) Invitrogen Miller et al. (2000) Miller et al. (2000) This study This study

ori, origin of replication initiation; R, resistance; ts, temperature sensitive.

37°C in lysogeny broth (LB) (Bertani, 1951) modified to contain 5 g l-1 NaCl. Agar (1.5%), ampicillin (50 mg ml-1), chloramphenicol (125 mg ml-1), gentamicin (5 mg ml-1) and diaminopimelic acid (DAP) (300 mg ml-1) were added when required. Photorhabdus spp. were grown at 28°C in LB supplemented with 1 g l-1 sodium pyruvate (LBP) or in PP3salt-2% Proteose Peptone no. 3 (Becton, Dickinson and Co., Franklin Lakes, NJ) containing 0.5% NaCl (PP3S). Agar (1.5%), chloramphenicol (15 mg ml-1), gentamicin (0.75 or 5 mg ml-1), streptomycin (40 mg ml-1), ampicillin (100 mg ml-1) and kanamycin (3.75 mg ml-1) were added when required.

Nucleic acid purification and molecular biology Standard molecular techniques were performed as described previously (Sambrook et al., 1989). Bacterial genomic DNA was purified from a 3 ml of culture of Photorhabdus grown in Grace’s insect cell culture medium (Invitrogen; Carlsbad, CA) using DNAeasy tissue kit (Qiagen; Alameda, CA). Plasmid DNA was purified using Qiagen Plasmid Mini or Maxi preps.

Restriction endonucleases and T4 ligase were used as per the manufacturer’s instructions unless otherwise indicated (Invitrogen or New England Biolabs; Ipswich, MA).

Nematode propagation Nematodes of the inbred H. bacteriophora strain M31e were grown on lawns of the GFP-labelled transmission mutant (TRN) 16 (Ciche et al., 2008), totally defective in transmission, on NA-CO [8 g of Nutrient broth, 15 g of agar, 12 ml of corn oil (Mazola) per l] added to one side of split 100 mm diameter Petri dishes for 10–14 days. Then, Ringer’s solution (100 mM NaCl, 1.8 mM KCl, 2 mM CaCl2, 1 mM MgCl2 and 5 mM HEPES pH 6.9) was added to the empty half of the plate to trap the emerging IJs, which were harvested by centrifugation (1200 r.p.m. for 1 min). The IJs were surface sterilized in 1% commercial bleach (Chlorox Ultra, 6.15% NaOCl) for 5 min, washed three times with Ringer’s and then stored in 10 ml of Ringer’s solution in tissue culture bottles. Sterility of the nematode stocks was verified by plating homo-

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

Phase variation of Mad fimbriae required for symbiosis 1035

genates on LBP. Antibiotics were then added to the nematode stocks at the following concentrations: 100 mg ml-1 streptomycin, 100 mg ml-1 ampicillin, 30 mg ml-1 kanamycin and 10 mg ml-1 gentamicin.

Transposon mutagenesis Photorhabdus luminescens strain TT0GFP or P. temperata strain NC1GFP and E. coli strain BW29427 donor cells containing pURE10 were grown overnight. One in 100 of the volume from each culture was transferred to fresh media and grown to an OD600 of 0.6. Cells were harvested by centrifugation, washed three times and resuspended in 500 ml of LB DAP. The recipient and donor suspensions were combined, centrifuged again, resuspended in 50 ml of LB DAP and spotted onto LB DAP plates. After incubation for 6–8 h, the cells were washed off the plates with LB, centrifuged, washed two times and resuspended in 1.5 ml of LB. A total of 100 ml of cells was spread on LB Gm and incubated for 2 days at 28°C. Isolated colonies were transferred to PP3S Gm plates and incubated at 28°C for 48 h.

Screen for transmission defective mutants

transposon was determined by sequencing using MAROUT and GmrFOR primers (Table S1) at the Michigan State University Research Technology Support Facility. DNA sequence was analysed with coliBLAST (http://xbase.bham.ac.uk/ colibase/) to determine the location of the transposon insertions (Altschul et al., 1990).

Complementation Two regions of the mad cluster were cloned using XL PCR kit (Invitrogen): madA through madK and madI through madK using primers Fim F and SpoT R, and Ush F and SpoT R respectively (Table S1). The PCR products were cloned into the TOPO XL PCR vector (Invitrogen) and plasmids pmadA–K and pmadIJK electroporated into TRN359, TRN583 and TRN26–25. The ability of the transformants to persist in the maternal nematode intestine was determined as described below.

Pulse-chase assay to detect persistent symbiont cells in maternal nematode intestines

Mutant strains were cultured statically in 250 ml of PP3S Gm at 28°C for 48 h and then 50 ml was spread on a 60 mm diameter plate (Tritech Research; Los Angeles, CA) containing NA-CO Gm media. After incubation at 28°C for 48 h, an average of 15 axenic IJs in were added to each mutant lawn and incubated at 28°C for 10–12 days or until IJs migrated and became trapped in the condensate present on lids. IJs were visually inspected for colonization (GFP present or absent) using a stereofluorescent MZ16F microscope (Leica Microsystems; Wetzlar, Germany) equipped with an X-cite 120 fluorescence illuminator (EXFO; Quebec, Canada). Potential mutants were isolated and transmission defective (TRN) phenotype verified at least twice in triplicate. The severity of the TRN phenotype was determined by examining c. 5000 IJs for the presence of any GFP-labelled bacteria using stereofluorescent microscopy and the transmission efficiency (colonized IJs/total IJs) determined.

Heterorhabditis bacteriophora associated with TT01 or NC1 have intestines full of fluorescent cells, most of which are transiently present (Fig. 1). Chasing the nematodes with unlabelled, differentially labelled or starving for 4 h causes labelled transient cells to be eliminated leaving only labelled persistent cells. Usually, c. 150 washed axenic M31e nematodes were pulsed for 36 h with GFP-labelled TT01 on NA-chol (like NA + CO, except CO replaced with 10 mg ml-1 cholesterol and 1 g l-1 of sodium pyruvate was added) and then chased for 4 h with TT01. Single cells of persistent green colonizing bacteria were detected in whole living nematodes by using a fluorescent compound microscope (DM5000, Leica) or by growing as green fluorescent colonies from nematode homogenates when examined by stereofluorescent microscopy. This assay was modified for the complementation by growing nematodes first on TRN16 for 16 h prior to pulsing for 24 h on the GFP-labelled test strain and then chasing for 4 h with TT01.

Transposon retrieval and sequencing

Allelic exchange

DNA containing the HimarGm transposons was retrieved by marker rescue (Ciche and Goffredi, 2007). The HimarGm transposon contains an R6Kg origin of replication allowing direct cloning of the transposon along with flanking DNA in strains expressing the p protein. Mutant genomic DNA was digested with SphI, ligated in 250–500 ml to favour intramolecular ligation, precipitated and resuspended in 10 ml of sterile ddH2O. A total of 2.5 ml of the ligation product was electroporated into 15–25 ml of TransforMax EC100D pir-116 electro-competent E. coli cells (EPICENTRE Biotechnologies; Madison, WI) using a Gene Pulser Xcell Electroporation System as per the manufacturer’s instructions (Bio-Rad; Hercules, CA). Transformants were selected for on LB Gm. The presence of the HiMarGm transposon was determined by digesting purified plasmid with SacI that releases an 835 bp fragment from the transposon. DNA flanking the HimarGm

To determine if the mad fimbriae are essential for TT01 transmission in nematodes, an unmarked and in-frame deletion, DmadA, was constructed by first introducing a marked deletion, DmadA::GmR, by allelic exchange followed by excision of the GmR gene by expressing the Flp recombinase. The DmadA::GmR mutant allele was constructed in manner that c. 700 bp upstream and downstream of madA was fused to GmR cassette flanked by Flp recombinase target (FRT) sites so that excision of the GmR by expression of the Flp recombinase would leave only an in-frame 86 bp FRT scar replacing 151–444 bp corresponding to amino acids 51–147 of madA (Choi and Schweizer, 2005). The DNA up- and downstream of madA was amplified using High Fidelity Platinum Taq DNA Polymerase (Invitrogen) by using primers listed in Table S1. The flanking products were then fused to FRTGmR-FRT cassette amplified from pPS856 by fusion PCR.

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

1036 V. S. Somvanshi et al. 䊏

The desired PCR fragment was gel extracted using Zymoclean Gel DNA Recovery kit (Zymo Research Corp; Orange, CA) and cloned into the suicide vector pST98As to create pST98AsDmadA::GmR. Bacteria were grown at 28°C to allow plasmid replication unless otherwise indicated. pST98AsDmadA::GmR was electroporated into GFP-labelled TTO1 and transformants were cultured in LBPGm broth for 16 to 28 h at 37°C before integrants were selected for on LBPGm. Excision of pST98As and madA was selected for by inducing the ISceI homing endonuclease with chlortetracycline (20 mg ml-1). Excision of the GmR was achieved by transforming the DmadA::GmR mutant with pCP20 and inducing expression of the Flp recombinase (Cherepanov and Wackernagel, 1995). The presence of the DmadA allele was determined by PCR.

RT-PCR RNA was isolated from TT01 cells grown in Grace’s Insect medium liquid cultures using Trizol reagent as per manufacture’s instruction (Invitrogen). One microgram of DNAasetreated RNA was reverse transcribed with Thermoscript Reverse Transcriptase (Invitrogen). The resulting cDNA was amplified using primers (Table S1). Controls included DNase I-treated but not reverse transcribed RNA and genomic DNA. PCR products were analysed by agarose gel electrophoresis. This procedure was also used to determine if the DmadA::GmR or DmadA inhibited transcription of downstream mad genes (Fig. S1).

Determination of pathogenicity to insects Pathogenicity to insects was determined by injecting third instar M. sexta larvae (North Carolina State University, Insectary) with an average of 10, 100 or 1000 wild-type or mutant cells and then assessing survival at 24, 36 and 72 h. Overnight cultures were adjusted to OD600 0.1 and appropriate dilutions made in LBP. Ten microlitres was injected behind the first forelimb of the larvae using a Hamilton Syringe with a 28-gauge needle (Hamilton; Reno, NV). M. sexta larvae were placed in wells of 24-well titre dishes containing c. 1 g of Gypsy Moth Wheat Germ Diet (MP Biomedicals; Solon, OH) and incubated at 25°C with a 14 h light and 10 h dark cycle. The mean of insect mortality data was analysed by Student’s t-test (P = 0.05) and using Kaplan-Meier statistical method and Log-rank test using GraphPad Prism, V. 5 (GraphPad Software, La Jolla, CA). P-values of < 0.05 were considered statistically significant.

Determination of madswitch flipping and which orientation is ON and OFF Colony PCR of TT01 cells isolated from LBP lawns or from the persistent biofilm present in maternal nematodes (Fig. 1A) was performed using FimIR F and FimIR R primers (Table S1) that amplify 1053 bp containing the madswitch. Restriction digestion with SfcI is predicted to produce 925 and 128 bp fragments if the madswitch is oriented as the TT01 genome sequence or 753 and 300 bp sized fragments

if it the madswitch flipped. To determine which orientation of the madswitch corresponds to the ON and OFF expression states of the mad locus, both orientations of the madswitch were first cloned upstream of the promoterless GFP reporter of pPROBE-GT (Miller et al., 2000) and transformed into the E. coli DH5a. Fluorescence was detected using a fluorescent compound microscope as above or by resuspending overnight cultures in phosphate-buffered saline (PBS) to OD600 = 0.1 and measuring the RFU using microplate reader (SpectraMax M5, Molecular Devices, Sunnyvale, CA). Sample means were statistically compared using Student’s t-test (P = 0.05).

Transmission electron microscopy The transmission electron microscopy procedure was adopted from Beveridge et al. (2007). The bacteria were grown overnight on LBP at 28°C and scrapped from the plate and resuspended in 1/10 strength PBS. Five microlitres of the resuspended culture was placed on the grid, and dried off by touching the grid with a filter paper (Whatman no. 1). Negative staining was performed by placing 5 ml of the 2% ammonium molybdate on the grid for 15 s, after which the stain was dried and fimbriae were visualized on a JEOL100 CXII (Japan Electron Optics Laboratories; Tokyo, Japan) at the Michigan State University Center for Advanced Microscopy.

Purification and identification of MadA Surface proteins were prepared from madswitch phase ON and OFF TT01 and DmadA cells. Cells from a 48 h lawn were suspended in 1 ml of PBS, washed once in PBS and resuspended in PBS before treating with a tissue homogenizer (Tissue Tearor, BioSpec Products, Bartlesville, OK). The cells were removed by centrifugation, and the supernatant proteins precipitated with 10% trichloroacetic acid, washed twice in cold acetone, dried, resuspended in PBS and quantified by Bradford method. For SDS-PAGE analysis, 40 mg of surface protein was separated on a 4–20% precast gel (Bio-Rad, Hercules, CA). A protein band of the predicted size of MadA was extracted and identified by liquid chromatography and tandem mass spectrometry performed at the Proteomics Facility at Michigan State University.

Phenotypic characterization of mad mutants Because Photorhabdus are known to generate secondary phenotypic variants that do not colonize IJs, the mutant strains were analysed with respect to the phenotypic variable characteristics (Akhurst, 1980; Boemare and Akhurst, 1988). Dye absorption assays were performed on MacConkey agar and NBTA [nutrient agar with 0.004% (w/v) triphenyltetrazolium chloride and 0.025% (w/v) bromothymol blue], LBP agar supplemented with 0.01% Congo red and Eosin Methylene blue agar (Acumedia Manufacturer’s, Baltimore, MD). Crystalline inclusion protein production was determined by phase contrast microscopy of 48 h old culture. Swimming and swarming motility was determined on LBP with 0.35% and 0.8% agar supplemented with 0.025% bromothymol blue and 0.004% triphenyltetrazolium chloride (Givaudan et al., 1995).

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

Phase variation of Mad fimbriae required for symbiosis 1037

Antibiotic production was determined from spot cultures grown for 48 h at 28°C, killed by chloroform exposure, and then overlaid with a M. luteus in 0.7% LB agar, and zone of growth inhibition around Photorhabdus was measured after 24 h at 37°C. Lipase production was determined on LBP supplemented with 0.01% Calcium Chloride and 1% (v/v) Tween 20, 40, 60 and 80. The production of siderophore activity was performed as described previously, except that the chrome azural S (CAS) solution was added to LBP agar (Schwyn and Neilands, 1987). Growth rates were determined from overnight cultures normalized to OD600 = 0.01 in 200 ml in a 96-well polystyrene plate (Corning Incorporated; Corning, NY) in triplicates incubated at 28°C in a microplate reader (SpectraMax M5, Molecular Devices; Sunnyvale, CA). The specific growth rate/h (m) was calculated by using the formula: m = 0.6931/doubling time (hours). Biofilm production was determined as described previously (O’Toole et al., 1999) from 200 ml of normalized cultures (OD600 = 0.01) incubated statically at 28°C for 48 h in 96-well plates. Biofilms were stained with crystal violet, 200 ml of 33% acetic acid was added to release the stain, and absorbance at 590 nm quantified. Each phenotypic test was repeated at least twice, and the data from phenotypic tests were analysed for significant differences using Student’s t-test at P = 0.05.

Acknowledgements We gratefully acknowledge Drs M. Bagdasarian, S. Lindow, D. K. Newman, D. M. Schifferili for plasmids, C. Palmer for help with screening, My-Dang Trang for help with illustrative art, Alicia Pastor for help with transmission electron microscopy and Dr L. Tisa for communicating unpublished P. temperata NC1 mad sequence. Funding was from Center for Microbial Pathogenesis and start-up funds from Michigan State University. We thank Dr L. Kroos, Dr M. Mulks and R. Parker for critical reviews and comments on this manuscript.

References Abraham, J.M., Freitag, C.S., Clements, J.R., and Eisenstein, B.I. (1985) An invertible element of DNA controls phase variation of type 1 fimbriae of Escherichia coli. Proc Natl Acad Sci USA 82: 5724–5727. Akhurst, R.J. (1980) Morphological and functional dimorphism in Xenorhabdus spp. bacteria symbiotically associated with the insect pathogenic nematodes Neoaplectana and Heterorhabditis. J Gen Microbiol 121: 303–310. Altschul, S.F., Gish, W., Miller, W., Myers, E.W., and Lipman, D.J. (1990) Basic local alignment search tool. J Mol Biol 215: 403–410. Bertani, G. (1951) Studies on lysogenesis. I. The mode of phage liberation by lysogenic Escherichia coli. J Bacteriol 62: 293–300. Beveridge, T.J., Moyles, D., and Harris, B. (2007) Electron microscopy. In Methods in General and Molecular Microbiology. Reddy, C.R., Beveridge, T.J., Breznak, J.A., Marzluf, G.A., Schmidt, T., and Snyder, L.R. (eds). Washington D.C.: ASM Press, pp. 54–81. Boemare, N.E., and Akhurst, R.J. (1988) Biochemical and physiological characterization of colony form variants in

Xenorhabdus spp. (Enterobacteriaceae). J Gen Microbiol 134: 751–761. Buchet, A., Nasser, W., Eichler, K., and Mandrand-Berthelot, M.A. (1999) Positive co-regulation of the Escherichia coli carnitine pathway cai and fix operons by CRP and the CaiF activator. Mol Microbiol 34: 562–575. Chandra, H., Khandelwal, P., Khattri, A., and Banerjee, N. (2008) Type 1 fimbriae of insecticidal bacterium Xenorhabdus nematophila is necessary for growth and colonization of its symbiotic host nematode Steinernema carpocapsiae. Environ Microbiol 10: 1285–1295. Cherepanov, P.P., and Wackernagel, W. (1995) Gene disruption in Escherichia coli: TcR and KmR cassettes with the option of Flp-catalyzed excision of the antibiotic-resistance determinant. Gene 158: 9–14. Choi, K.H., and Schweizer, H.P. (2005) An improved method for rapid generation of unmarked Pseudomonas aeruginosa deletion mutants. BMC Microbiol 5: 30. Ciche, T., and Goffredi, S.K. (2007) General methods to investigate microbial symbioses. In Methods in General and Molecular Microbiology. Reddy, C.R., Beveridge, T.J., Breznak, J.A., Marzluf, G.A., Schmidt, T., and Snyder, L.R. (eds). Washington D.C.: ASM Press, pp. 394–419. Ciche, T.A., and Ensign, J.C. (2003) For the insect pathogen Photorhabdus luminescens, which end of a nematode is out? Appl Environ Microbiol 69: 1890–1897. Ciche, T.A., and Sternberg, P.W. (2007) Postembryonic RNAi in Heterorhabditis bacteriophora: a nematode insect parasite and host for insect pathogenic symbionts. BMC Dev Biol 7: 101. Ciche, T.A., Kim, K.S., Kaufmann-Daszczuk, B., Nguyen, K.C., and Hall, D.H. (2008) Cell Invasion and Matricide during Photorhabdus luminescens Transmission by Heterorhabditis bacteriophora nematodes. Appl Environ Microbiol 74: 2275–2287. Deng, W., Puente, J.L., Gruenheid, S., Li, Y., Vallance, B.A., Vazquez, A., et al. (2004) Dissecting virulence: systematic and functional analyses of a pathogenicity island. Proc Natl Acad Sci USA 101: 3597–3602. Duchaud, E., Rusniok, C., Frangeul, L., Buchrieser, C., Givaudan, A., Taourit, S., et al. (2003) The genome sequence of the entomopathogenic bacterium Photorhabdus luminescens. Nat Biotechnol 21: 1307–1313. Easom, C.A., Joyce, S.A., and Clarke, D.J. (2010) Identification of genes involved in the mutualistic colonization of the nematode Heterorhabditis bacteriophora by the bacterium Photorhabdus luminescens. BMC Microbiol 10: 45. Edwards, R.A., Cao, J., and Schifferli, D.M. (1996) Identification of major and minor chaperone proteins involved in the export of 987P fimbriae. J Bacteriol 178: 3426–3433. Gaudriault, S., Duchaud, E., Lanois, A., Canoy, A.S., Bourot, S., Derose, R., et al. (2006) Whole-genome comparison between Photorhabdus strains to identify genomic regions involved in the specificity of nematode interaction. J Bacteriol 188: 809–814. Gerrard, J.G., Joyce, S.A., Clarke, D.J., Ffrench-Constant, R.H., Nimmo, G.R., Looke, D.F., et al. (2006) Nematode symbiont for Photorhabdus asymbiotica. Emerg Infect Dis 12: 1562–1564. Givaudan, A., Baghdiguian, S., Lanois, A., and Boemare, N. (1995) Swarming and swimming changes concomitant with

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038

1038 V. S. Somvanshi et al. 䊏

phase variation in Xenorhabdus nematophilus. Appl Environ Microbiol 61: 1408–1413. Henderson, I.R., Owen, P., and Nataro, J.P. (1999) Molecular switches – the ON and OFF of bacterial phase variation. Mol Microbiol 33: 919–932. Hoang, T.T., Karkhoff-Schweizer, R.R., Kutchma, A.J., and Schweizer, H.P. (1998) A broad-host-range Flp-FRT recombination system for site-specific excision of chromosomally-located DNA sequences: application for isolation of unmarked Pseudomonas aeruginosa mutants. Gene 212: 77–86. Holden, N.J., and Gally, D.L. (2004) Switches, cross-talk and memory in Escherichia coli adherence. J Med Microbiol 53: 585–593. Huang, L.H., and Syu, W.J. (2008) GrlA of enterohemorrhagic Escherichia coli O157:H7 activates LEE1 by binding to the promoter region. J Microbiol Immunol Infect 41: 9–16. Khan, A.S., and Schifferli, D.M. (1994) A minor 987P protein different from the structural fimbrial subunit is the adhesin. Infect Immun 62: 4233–4243. Kline, K.A., Falker, S., Dahlberg, S., Normark, S., and Henriques-Normark, B. (2009) Bacterial adhesins in hostmicrobe interactions. Cell Host Microbe 5: 580–592. Lane, M.C., Li, X., Pearson, M.M., Simms, A.N., and Mobley, H.L. (2009) Oxygen-limiting conditions enrich for fimbriate cells of uropathogenic Proteus mirabilis and Escherichia coli. J Bacteriol 191: 1382–1392. Lasaro, M.A., Salinger, N., Zhang, J., Wang, Y., Zhong, Z., Goulian, M., and Zhu, J. (2009) F1C fimbriae play an important role in biofilm formation and intestinal colonization by the Escherichia coli commensal strain Nissle 1917. Appl Environ Microbiol 75: 246–251. Li, X., Rasko, D.A., Lockatell, C.V., Johnson, D.E., and Mobley, H.L. (2001) Repression of bacterial motility by a novel fimbrial gene product. EMBO J 20: 4854–4862. Meslet-Cladiere, L.M., Pimenta, A., Duchaud, E., Holland, I.B., and Blight, M.A. (2004) In vivo expression of the mannose-resistant fimbriae of Photorhabdus temperata K122 during insect infection. J Bacteriol 186: 611–622. Miller, W.G., Leveau, J.H., and Lindow, S.E. (2000) Improved gfp and inaZ broad-host-range promoter-probe vectors. Mol Plant Microbe Interact 13: 1243–1250. Moxon, E.R., Rainey, P.B., Nowak, M.A., and Lenski, R.E. (1994) Adaptive evolution of highly mutable loci in pathogenic bacteria. Curr Biol 4: 24–33. Nuccio, S.P., and Baumler, A.J. (2007) Evolution of the chaperone/usher assembly pathway: fimbrial classification goes Greek. Microbiol Mol Biol Rev 71: 551–575. O’Toole, G.A., Pratt, L.A., Watnick, P.I., Newman, D.K., Weaver, V.B., and Kolter, R. (1999) Genetic approaches to study of biofilms. Methods Enzymol 310: 91–109. Ofec, I., Hasty, D.L., and Doyle, R.J. (2003) Bacterial Adhesion to Animal Cells and Tissues. Washington, D.C.: ASM Press. Oshima, K., Toh, H., Ogura, Y., Sasamoto, H., Morita, H., Park, S.H., et al. (2008) Complete genome sequence and comparative analysis of the wild-type commensal Escheri-

chia coli strain SE11 isolated from a healthy adult. DNA Res 15: 375–386. Pearson, M.M., and Mobley, H.L. (2008) Repression of motility during fimbrial expression: identification of 14 mrpJ gene paralogues in Proteus mirabilis. Mol Microbiol 69: 548–558. Pearson, M.M., Sebaihia, M., Churcher, C., Quail, M.A., Seshasayee, A.S., Luscombe, N.M., et al. (2008) Complete genome sequence of uropathogenic Proteus mirabilis, a master of both adherence and motility. J Bacteriol 190: 4027–4037. Posfai, G., Kolisnychenko, V., Bereczki, Z., and Blattner, F.R. (1999) Markerless gene replacement in Escherichia coli stimulated by a double-strand break in the chromosome. Nucleic acids Res 27: 4409–4415. Rendon, M.A., Saldana, Z., Erdem, A.L., Monteiro-Neto, V., Vazquez, A., Kaper, J.B., et al. (2007) Commensal and pathogenic Escherichia coli use a common pilus adherence factor for epithelial cell colonization. Proc Natl Acad Sci USA 104: 10637–10642. Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual, 2nd edn. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press. Sauer, F.G., Mulvey, M.A., Schilling, J.D., Martinez, J.J., and Hultgren, S.J. (2000) Bacterial pili: molecular mechanisms of pathogenesis. Curr Opin Microbiol 3: 65–72. Schwyn, B., and Neilands, J.B. (1987) Universal chemical assay for the detection and determination of siderophores. Anal Biochem 160: 47–56. Townsend, S.M., Kramer, N.E., Edwards, R., Baker, S., Hamlin, N., Simmonds, M., et al. (2001) Salmonella enterica serovar Typhi possesses a unique repertoire of fimbrial gene sequences. Infect Immun 69: 2894–2901. Waterfield, N.R., Ciche, T., and Clarke, D. (2009) Photorhabdus and a host of hosts. Annu Rev Microbiol 63: 557– 574. Weening, E.H., Barker, J.D., Laarakker, M.C., Humphries, A.D., Tsolis, R.M., and Baumler, A.J. (2005) The Salmonella enterica serotype Typhimurium lpf, bcf, stb, stc, std, and sth fimbrial operons are required for intestinal persistence in mice. Infect Immun 73: 3358–3366. van der Woude, M.W., and Baumler, A.J. (2004) Phase and antigenic variation in bacteria. Clin Microbiol Rev 17: 581– 611. Zieg, J., Silverman, M., Hilmen, M., and Simon, M. (1977) Recombinational switch for gene expression. Science 196: 170–172.

Supporting information Additional supporting information may be found in the online version of this article. Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

© 2010 Blackwell Publishing Ltd, Molecular Microbiology, 77, 1021–1038