Signal Recognition Particle-Dependent Inner Membrane Targeting of the PulG Pseudopilin Component of a Type II Secretion System

Abstract

The pseudopilin PulG is an essential component of the pullulanase-specific type II secretion system from Klebsiella oxytoca. PulG is the major subunit of a short, thin-filament pseudopilus, which presumably elongates and retracts in the periplasm, acting as a dynamic piston to promote pullulanase secretion. It has a signal sequence-like N-terminal segment that, according to studies with green and red fluorescent protein chimeras, anchors unassembled PulG in the inner membrane. We analyzed the early steps of PulG inner membrane targeting and insertion in Escherichia coli derivatives defective in different protein targeting and export factors. The β-galactosidase activity in strains producing a PulG-LacZ hybrid protein increased substantially when the dsbA, dsbB, or all sec genes tested except secB were compromised by mutations. To facilitate analysis of native PulG membrane insertion, a leader peptidase cleavage site was engineered downstream from the N-terminal transmembrane segment (PrePulG*). Unprocessed PrePulG* was detected in strains carrying mutations in secA, secY, secE, and secD genes, including some novel alleles of secY and secD. Furthermore, depletion of the Ffh component of the signal recognition particle (SRP) completely abolished PrePulG* processing, without affecting the Sec-dependent export of periplasmic MalE and RbsB proteins. Thus, PulG is cotranslationally targeted to the inner membrane Sec translocase by SRP.

The pullulanase type II secretion system (T2SS) of Klebsiella oxytoca, the Pul secreton, is a prototype of a family of membrane protein complexes dedicated to the transport of folded proteins from the periplasm to the extracellular milieu (49). T2SS machineries share a common origin and architecture and functional properties with systems involved in type IV pilus assembly and retraction, DNA uptake, and natural competence (64). Central to these machineries is their ability to assemble and disassemble filaments that either remain periplasmic and function in protein secretion (pseudopili) or extend well beyond the cell surface and play a role in adhesion, surface motility, or microcolony or biofilm formation (type IV pili) (12). The pseudopili and the closely related class A type IV pili are composed of one major pilin subunit and several minor pilins. Type IVA pilins and T2SS pseudopilins have highly conserved N-terminal regions with properties reminiscent of signal sequences except for their cleavage and N methylation on the N-terminal, cytoplasmic side of the hydrophobic region by a specific prepilin signal peptidase/N-methylase (40). This N-terminal hydrophobic helix leads directly into a 30-residue-long amphipathic α-helical region that is less well conserved and is followed by a globular domain composed of four antiparallel beta-sheets (24). In addition, type IVA pilins have a highly variable loop between 2 disulfide-bonded cysteine residues at their C-terminal ends (12).

The predicted topology of PrePulG in the inner membrane (IM) is that of a type II bitopic membrane protein, with an N-in, C-out orientation. Removal of the positively charged N-terminal region by prepilin peptidase renders PulG competent for assembly. Two assembly factors common to the T2SS and type IV pilus assembly and retraction system, a polytopic IM protein of the PilC family (PulF in the Pul secreton) and an oligomeric ATPase of the PilB (VirB11) family (PulE) (36), probably catalyze filament polymerization. Nevertheless, their direct role remains to be demonstrated in an in vitro system and they could also be involved in pseudopilin precursor export from the cytoplasm.

PulG, the major pseudopilin of the Pul secreton, is assembled into a pilus structure on the cell surface when overproduced (50). Sheering and purification of these pili revealed PulG as their unique component (24), suggesting that minor pilins remain cell envelope associated at the filament base. The crystal structure of residues 25 to 146 of PulG revealed the same basic architecture as that of type IV pilins of Pseudomonas aeruginosa or Neisseria gonorrhoeae (24). Docking of the PulG structure onto the PulG filament, visualized by electron microscopy, enabled a model to be proposed in which the filament is a one-start, left-handed helix with 17 monomers and four helical turns per repeat unit (24).

Despite the wealth of structural data on type IV pili/pseudopili and their subunits, very little is known about the early events involved in pseudopilus biogenesis. Here we analyze the targeting and insertion of the newly synthesized pilin subunits into the IM of Escherichia coli by using a combination of in vivo gene fusion and biochemical approaches. The subcellular location of pilin subunits in the absence or presence of the other Pul secreton factors was analyzed in vivo using PulG fused to green and red fluorescent proteins. Given the high level of structural conservation among pseudopilins and type IV pilins, the conclusions drawn from this study probably apply to this entire protein family.

MATERIALS AND METHODS

Bacterial strains, growth conditions, and genetic techniques.

E. coli strains used in this study are listed in Table 1. Luria-Bertani (LB) broth and agar, tetrazolium-lactose agar, MacConkey maltose agar, or minimal M63 medium was prepared as described previously (32). When appropriate, media were supplemented with maltose (0.4%), lactose (0.4%), arabinose (0.2%), glycerol (0.5%), or glucose (0.2%). Temperature-sensitive phenotypes were tested at 20°C (cold sensitive) or 42°C (heat sensitive) on LB agar. For β-galactosidase and fractionation assays, temperature-sensitive strains were grown under semipermissive conditions (30°C for cold-sensitive and 37°C for heat-sensitive strains), unless otherwise indicated. Antibiotics were used at the following concentrations: tetracycline, 10 μg/ml; chloramphenicol, 25 μg/ml; ampicillin (Amp), 100 μg/ml; kanamycin, 50 μg/ml; spectinomycin, 100 μg/ml, and streptomycin, 100 μg/ml. Generalized transduction with P1vir phage and bacterial conjugation were performed as described previously (32).

TABLE 1.

Escherichia coli strains used in this study

Strain	Genotype	Source or reference
MC4100	(F−) Δlac-169 araD139 thiA rpsL relA motA rbsR	Laboratory collection
MM52	MC4100 secA51(Ts)	35
MM18	MC4100 λmalE-lacZ72-47	22
TY22	MC4100 rpsE ompT::kan zdh::Tn10 secY40	4, 33, 57
AD208	MC4100 ompT::kan zdh::Tn10 secY39	4, 33, 57
JCB656	MC4100 λmalF-lacZ102 dsbA::kan malT(Con)	J. Bardwell
MC4100ΔsecB	MC4100 ΔsecB	P. Delepelaire
CK1637	araD Δ(lac-pro)XIII Rifr NalrargE(Am) mtl-2 recA1(Sms) F′ lacIq1	C. Kumamoto
WAM121	MC4100 ara+ffh::kan attB::ori R6K paraBAD::ffh tet	14, H. Bernstein
PAP7460	MC4100 ΔmalE444 pAPIP501 (F′ lacIq1 ΔlacZM15 proAB+ Tn10)	Laboratory collection
JCB571	MC1000 phoR zih-12::Tn10 dsbA::kan-1	6
IQ85	MC4100 secY24(Ts) zhd-33::Tn10	K. Ito
PS266	MC4100 secE15(Cs)	51
OFM1004	MC4100 secY108	This study
PAP5237	MC4100 dsbA::kan	This study
PAP5250	MC4100 dsbB::kan	This study
OF670	OFM1004 zjb::Tn5 malE18-1	This study
OFM4004	MC4100 secD4004(Cs)	This study
OFM1610	MC4100 secE1610	This study
OFM7004	MC4100 secA7004	This study
PAP9102	MC4100 Δ(attλ-lom)::bla lacIq1gfp-pulG	This study
PAP9121	MC4100 Δ(attλ-lom)::bla lacIq1gfp	This study
PAP7500	MC4100::pulS pulAB pulCO (F′ lacIq1 ΔlacZM15 proAB+ Tn10)	Laboratory collection
PAP7500BG	PAP7500 pulB::km ΔpulG	Laboratory collection

Plasmid construction and molecular biology techniques.

DNA manipulations, plasmid purification, and DNA transformation were performed essentially as described previously (47). PCR amplification was performed using Pfu (Invitrogen) or Phusion (Finnzymes) DNA polymerase in buffers supplied by the manufacturers. Oligonucleotide primers used for PCR and DNA sequence analysis are listed in Table 2. Plasmid pCHAP8075 containing the pulG-lacZ fusion was constructed by cloning the pulG gene, PCR amplified without the stop codon by using primers PulGBamHI and PulGBglII, into the BamHI site of pOF13 (17). Plasmid pCHAP8086 was made by cloning the DNA fragment encoding the prepilin signal sequence of PulG, PCR amplified using primers PulGEco5′ and PulGNco3′, into the EcoRI and NcoI sites of pCHAP7010 (pelB_SP-pulG_25-140His₆) (24). This event replaced the PelB signal peptide with the PulG prepilin signal sequence, followed by a leader peptidase cleavage site. The EcoRI-HindIII fragment containing the pulG_SP-pulG gene was subcloned into a low-copy-number vector, pHSG575 (56), digested with the same enzymes, giving pCHAP8100. Plasmid pCHAP8095 contained a wild-type secE gene, cloned as a PCR fragment generated using primers SecEL and SecER into the EcoRI and BamHI sites of pSU18 (7). Plasmid pCHAP8096 contained a wild-type secD gene, PCR amplified using flanking primers SecDL and SecDR, digested with Xho and HindIII, and cloned into the SalI and HindIII sites of pSU19. The wild-type secF gene was cloned into the EcoRI and BamHI sites of plasmid pSU19 after PCR amplification using primers SecFL and SecFR, giving plasmid pCHAP8092.

TABLE 2.

Oligonucleotides used in this study

Oligonucleotide	Nucleotide sequence (5′-3′)
PulGBamHI	GTAGGATCCTTCCCGATCGTCCAGTTG
PulG-BglII	CCGAGATCTGTGATTGTTCTAATGGAG
PulGEco5′	CGGAATTCGAGTATGTGATTGTTCTAATGGAG
PulGNco3′	GACCATGGCCATCGCCGGCACCACGAGGCTG
SecEL	CCGAATTCCGAAAAGCTAATACGCGTTTC
SecER	CAAGGATCCGAAAACGCCTGAACGACGTAC
SecDL	CACAAGCTTCGAAAGGCACCATGAAGG
SecDR	CCACTCGAGGTTCAACAGTATATTCCTGTG
SecDseq	CGTGCGAAAGAGATTCTGGGTGCGAC
SecD2	GCATCCCGTATACCACTGTTCGTAAAG
SecFL	CAAGAATTCGCGCGTCAAGAAGCTGTC
SecFR	CAAGGATCCGGATTTTCGCAATCTCC
Ngfp-pulG-5	CGAGAATTCAACAACAACCAACGACAGCGCGGATTC
Ngfp-pulG-3	CCGTCGAAGCTTCTATTTCTTCCCGATCGTCCAG
XbaI-RFP-5	CTAGTCTAGAATGGCCTCCTCCGAGGACGTCATCAAGG
HindIIIRFP-3	CCTGAAGCTTTTAGGCGCCGGTGGAGTGGCGG
PulGERI-5	CGAGAATTCATGCAACGACAGCGCGGATTC
GrfpXbaI-3	CTAGTCTAGATTTCTTCCCGATCGTCCAG

Plasmid pCHAP7506 (encoding green fluorescent protein [GFP]-PrePulG) was made by cloning the pulG gene, PCR amplified using oligonucleotides Ngfp-PulG-5 and Ngfp-pulG-3. The PCR product was inserted into the EcoRI and HindIII sites of vector pDSW204 (63). Integration of the pulG-gfp fusion into the attλ site of the chromosome of strain MC4100 via λInCh was performed as described previously (9), resulting in strain PAP9102. pCHAP7506 was able to fully restore pullulanase secretion in a strain lacking pulG, indicating that processing of GFP-PrePulG by PulO prepilin peptidase produces functional PulG. The monomeric red fluorescent protein (mRFP1) fusion vector pCHAP7520 was made by inserting the mRFP1-encoding gene from plasmid pRSetBmRFP1 (11), PCR amplified using oligonucleotide primers XbaI-RFP-5 and HindIIIRFP-3, into pDSW204. The pulG gene, amplified using primers PulGERI-5 and GrfpXbaI-3, was inserted into the EcoRI and XbaI sites of this vector to give pCHAP7516. This plasmid restored to only 20% of wild type the level of pullulanase secretion in a strain lacking pulG, presumably because the RFP domain interferes with PulG assembly and functional PulG is not released from PulG-mRFP by proteolysis (39). All plasmid constructs were verified by DNA sequencing. Other plasmids were pCHAP231, carrying all of the pul secreton genes including pulA (14), and pCHAP710, carrying all of the pul secreton genes except pulA (25).

Mutagenesis and selection of Lac⁺ mutants.

E. coli Lac⁺ mutants expressing the rbsB-lacZ gene fusion were isolated as follows. E. coli strain MC4100 was grown to mid-exponential phase and incubated in the presence of 50 μg/ml of N,N-nitrosoguanidine for 15 min at 37°C. Bacteria were washed twice in M63 salts solution, resuspended in LB medium, and divided into 16 pools, which were grown out overnight at 30°C. The approximate killing rate was 33%. Mutagenized bacteria were transformed with plasmid pOF21 carrying an rbsB-lacZ gene fusion (17), plated on lactose-tetrazolium plates containing ampicillin, and incubated for 48 h at 30°C. Bacteria in pink colonies growing on a white lawn were purified twice on lactose-tetrazolium Amp plates and cured of pOF21 on LB plates containing 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal). White, Amp-sensitive bacteria were purified twice on the same medium and then retransformed with pOF21 to check that the lesions causing the Lac⁺ phenotype were linked to the chromosome. Out of a total of 144 mutants tested in this way, 54 contained chromosomal lesions, suggesting that other, plasmid-linked mutations arose spontaneously during growth on selective plates.

Mutant characterization.

In addition to nine mutations in the dsbA gene, two in mutT, and a number of uncharacterized mutations, the RbsB-LacZ Lac⁺ screen yielded multiple mutations in known sec loci (secA, secDF, secE, and secY) (O. Francetic and C. Kumamoto, unpublished results). Pulse-chase analysis showed that these mutations conferred export defects of both MalE and RbsB. Four of these mutations were characterized in the present study. The mutation conferring the Lac⁺ phenotype in strain OLFM1004 was 90% cotransduced with the Sm^r marker rpsL at min 72, indicating that the lesion is linked tightly to the secY locus. To facilitate its genetic analysis, this mutation was combined with a mutation in the signal sequence of the maltose-binding protein, malE18-1. The resulting double mutant OLF760 (MC4100 zjb::Tn5 malE18-1 secY1004) was completely defective in MalE export and had a Mal⁻ phenotype on MacConkey maltose indicator media. This strain was transformed with the pOFY7, pOFY8, or pOFY9 plasmid, carrying the proximal, middle, or distal third, respectively, of the wild-type secY gene, constructed as described previously (16). In a marker rescue experiment, only the plasmid pOFY7 restored the Mal⁺ phenotype on maltose minimal Amp plates, with a frequency of about 10⁻⁵, corresponding to the rate of homologous recombination. The corresponding fragment from strain OFM1004 was PCR amplified and sequenced to identify a single mutation causing an E108K substitution in the second cytoplasmic loop of SecY.

The lesion in mutant OFM4004, conferring a weak cold-sensitive phenotype, was cotransduced by P1 with the tsx::kan marker and, therefore, is linked to the secDF locus. The Lac⁺ phenotype of strain OFM4004 (pOF21) could be converted to Lac⁻ when cotransformed with the compatible plasmid pCHAP8096 carrying wild-type secD, but not with pCHAP8092 carrying wild-type secF, suggesting that the Lac⁺ phenotype was due to a secD mutation. This was confirmed by DNA sequence analysis of PCR-amplified secD from OFM4004, which indicated a G506D substitution in the fourth transmembrane segment of SecD.

The Lac⁺ mutation in strain OFM1610 was cotransduced with the argE locus and, therefore, is linked to the secE gene. Plasmid pCHAP8095, carrying wild-type secE, restored the original Lac⁻ phenotype. Sequence analysis of the PCR-amplified secE gene from this mutant revealed a single G to A nucleotide change at position −38 relative to the secE start codon. This mutation presumably reduces the level of secE expression. The mutation in strain OFM7004, identified in the same selection, was genetically linked to the leu marker at min 2 of the E. coli chromosome and mapped to the secM-secA operon. The mutation was designated secA7004, although it could be in the secM gene, and was not further characterized.

Protein analysis and enzymatic assays.

Proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on 10%, 12%, or 15% polyacrylamide gels with Tris-glycine and SDS buffers and then transferred electrophoretically to nitrocellulose membranes using a semidry blotting apparatus. Primary antibodies were rabbit antisera raised against purified MalE (J.-M. Betton), PulG, RbsB (17), GFP (Clontech), DsbA (Satish Raina), MalE-PulC, MalE-PulS, Pal (Emmanuelle Bouveret), DjlA (David Clarke), and LacZ (Organon Teknika Corp.). Secondary antibodies were donkey anti-rabbit immunoglobulin G coupled to horseradish peroxidase (Amersham Biosciences). Antibody-treated nitrocellulose membranes were developed by enhanced chemiluminescence (Amersham). β-Galactosidase activity was assayed according to Miller (32). Pullulanase cell surface exposure was assayed as described previously (31).

Pulse-chase and immunoprecipitation analyses.

Cells were grown overnight in M63 medium containing 0.5% glycerol and diluted 1:100 in fresh M63 medium containing 0.5% glycerol or 0.2% glycerol and 0.4% maltose supplemented with a cocktail of 18 amino acids (lacking methionine and cysteine). Cells were grown to an optical density at 600 nm (OD₆₀₀) of 0.4 and pulse-labeled for 15 s with 10 μCi/ml of [³⁵S]methionine, followed by a 1-min chase in the presence of cold methionine (100 μg/ml) and chloramphenicol (170 μg/ml). Samples of 1 ml were withdrawn at the indicated times (see the legend to Fig. 1B) and precipitated with 5% trichloroacetic acid for 30 min on ice. After 10 min of centrifugation in a microcentrifuge at 4°C, pellets were washed twice with 1 ml of cold acetone and air dried. For immunoprecipitation, total extracts were resuspended in 50 μl of 25 mM Tris-HCl (pH 8) buffer containing 1% SDS and 1 mM EDTA and heated at 100°C for 2 min. After cooling to room temperature, 450 μl KI buffer (50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 2% Triton X-100, 1 mM EDTA) was added and samples were centrifuged for 10 min in a microcentrifuge at 4°C. A 200-μl aliquot was taken from each tube and further diluted with 300 μl of ice-cold KI buffer. Anti-MalE antiserum (final dilution, 1:500) was added to each sample, and the samples were incubated overnight on ice. Protein A-coated Sepharose (Amersham Biosciences) was added to each tube and incubated for 20 min on ice with mixing. The samples were centrifuged for 1 min in a microcentrifuge, and supernatants were removed. Pellets were washed twice with 0.5 ml of 50 mM Tris-HCl (pH 8.0), 1 M NaCl, 1% Triton X-100 and once with 10 mM Tris-HCl (pH 8.0). The pellets were resuspended in 50 μl SDS-PAGE sample buffer and heated to 100°C for 2 min to release proteins bound to the beads. The samples were then analyzed by SDS-PAGE and fluorography.

FIG. 1.

Induction of the PulG-LacZ fusion synthesis has no effect on substrates of the Sec pathway. (A) Cells of strains MC4100, MM18, MC4100(pCHAP8075), and MC4100 secY24(Ts) were grown at 30°C in LB medium containing Amp in the absence (−) or presence (+) of 0.2% maltose for 3 h with vigorous shaking. Total cell extracts were analyzed by SDS-PAGE and immunoblotting using anti-LacZ, anti-MalE, anti-RbsB, and anti-PulG antisera, as indicated at the bottom left of each panel (α, anti). The positions of relevant proteins are indicated on the right. (B) Pulse-chase analysis of MalE export in strains MC4100, MM4100 secA(Ts) (MM52) (left 2 lanes), and MC4100(pCHAP8075) (right 2 lanes). Cells were grown at 30°C in minimal maltose medium, pulse-labeled with [³⁵S]methionine for 15 s (P), and chased for 60 s (C). Labeled cell extracts were immunoprecipitated with anti-MalE antibodies. Proteins were analyzed by SDS-PAGE and fluorography. The relevant portion of the resulting fluorogram is shown. The positions of PreMalE and MalE proteins are indicated.

E. coli fractionation and sucrose gradient separation.

Spheroplast and periplasmic fractions were prepared as described previously (44). Bacteria were grown to mid-log phase and total cell and spheroplast fractions were resuspended in SDS sample buffer at a concentration corresponding to 10 OD₆₀₀ units/ml. Periplasmic fractions were precipitated in 10% trichloroacetic acid for 30 min on ice and then pelleted for 10 min in a microcentrifuge at 4°C. After two washes with 1 ml of cold acetone, pellets were air dried and resuspended in the appropriate volume of sample buffer based on the fraction of the total volume concentration relative to that of the spheroplast samples.

Sucrose gradient fractionation of membranes containing PulG or PulG-mRFP was performed as described previously (46). Fractions from the gradients were analyzed by SDS-PAGE on 10% or 12% acrylamide gels, followed by immunoblotting with antibodies against PulG, PulC, PulS, Pal, and DjlA.

Fluorescence microscopy and plasmolysis.

For fluorescence microscopy, strains were grown overnight at 30°C, diluted 1 to 100 in fresh medium, and grown to an OD₆₀₀ of 0.2. Isopropyl β-d-thiogalactoside (IPTG) was then added to a final concentration of 10 or 100 μM and 0.4% maltose was added for strains carrying the pul genes under MalT-dependent promoter control. The cells were incubated for 1 h at 30°C under continuous shaking. Live cells were immobilized on wet agarose-coated glass slides and analyzed using a Zeiss Axioplan 2 fluorescence microscope with a Hamamatsu CCD camera mounted. Images were collected using OpenLab software and processed with Photoshop. Cells were plasmolyzed as described elsewhere (27) prior to examination.

RESULTS

Construction and analysis of a PulG-LacZ hybrid.

Protein fusions to β-galactosidase (LacZ) have been useful tools for the genetic analysis of many biological functions, in particular, protein localization and export. When proteins with signal peptides are fused to LacZ and targeted to extracytoplasmic compartments in E. coli, the LacZ domain becomes enzymatically inactive and the bacteria are phenotypically Lac⁻. Mutations conferring a Lac⁺ phenotype by causing localization of LacZ fused to periplasmic MalE to the cytoplasm contributed to the discovery of the first Sec pathway components, SecA (35) and SecB (26). A similar approach based on LacZ fused to the IM protein MalF led to the identification of a pathway involved in periplasmic disulfide bond formation (30). Other genetic (59) and biochemical (28, 60) approaches identified the signal recognition particle (SRP) as a key component for IM protein biogenesis in E. coli. Recently, a modification of the MalF-LacZ selection produced mutants affected in all known components of the SRP pathway (ffh, ffs, and ftsY) (58). Therefore, we reasoned that the LacZ fusion approach should be sufficiently sensitive to identify factors required for PulG localization.

A full-length pulG gene lacking its stop codon was fused to lacZ lacking its 8 N-terminal codons on a moderate-copy-number plasmid, pCHAP8075 (described in Materials and Methods). The pulG-lacZ fusion was expressed from the pmalE promoter and transcribed in an operon with the downstream lacY and lacA genes. E. coli MC4100 transformed with this plasmid produced pale blue colonies on X-Gal indicator plates and was Lac⁻ on tetrazolium-lactose agar. The manifestly low level of β-galactosidase in this strain could be explained by the targeting of the chimera to the membrane. To ensure that the Lac⁻ phenotype of strain MC4100(pCHAP8075) was due to a low specific activity and not to low-level synthesis of the chimera, pCHAP8075 was introduced into strains carrying mutations in the disulfide bond formation factor genes dsbA and dsbB. The specific β-galactosidase activity of the PulG-LacZ hybrid increased substantially in these strains (Table 3), indicating that disulfide bonds in periplasmically exposed LacZ stabilize the transmembrane conformation of the PulG-LacZ hybrid (55).

TABLE 3.

β-Galactosidase activities of different E. coli strains expressing the pulG-lacZ gene fusion

Strain	Genotype	β-Galactosidase activity [Miller units (±SD)]a		Mutant/wild type ratiob
		−Maltose	+Maltose
MC4100	Wild type	0.2 (±0.1)	0.2 (±0.1)	1
PAP5237	dsbA::kan	14.5 (±5.5)	47.5 (±17.8)	39
PAP5250	dsbB::kan	23.9 (±8.9)	32.2 (±9.6)	150
MC4100 ΔsecB	ΔsecB	ND	0.3 (±0.1)	ND
MM52	secA51(Ts)	5.0 (±2.9)	12.9 (±4.3)	21
MC4100 secY24	secY24(Ts)	7.4 (±1.4)	16.7 (±7.2)	60
OFM1004	secY108	22.2 (±6.8)	46.5 (±3.1)	154
OFM7004	secA7004	1.1 (±0.8)	5.1 (±2.2)	3
OFM4004	secD4004	1.0 (±0.4)	4.9 (±3.3)	6
MC4100 secE(Cs)	secE15(Cs)	2.3 (±1.3)	16.5 (±11.8)	10
OFM1610	secE1610	3.6 (±2.0)	6.5 (±3.1)	16
AD208	secY39(Cs)	ND	30.8 (±2.4)	ND
TY22	secY40(Cs)	ND	27.7 (±4.8)	ND

^aβ-Galactosidase activity of cells grown in the absence(−) or presence (+) of 0.4% maltose in LB growth medium. The mean values from at least three independent experiments are given, with standard deviations indicated in parentheses.

^bβ-Galactosidase activities in mutants normalized against activity in the wild-type strain, all grown without maltose, with values obtained after subtracting the standard deviation from the mean. ND, not determined.

Immunoblot analysis of strain MC4100(pCHAP8075) confirmed the synthesis of a stable maltose-inducible hybrid protein that reacted with both anti-PulG and anti-LacZ antibodies (Fig. 1A). As described previously (22), the induction of MalE-LacZ synthesis in strain MM18 caused pleiotropic defects in protein export, including the accumulation of unprocessed, signal peptide-containing forms of MalE and ribose-binding protein (RbsB). In contrast, maltose-induced production of PulG-LacZ did not cause any export defects (Fig. 1A). Efficient MalE export under these conditions was confirmed by pulse-chase analysis (Fig. 1B). More than 95% of MalE was in the mature form after 15 s of pulse-labeling in both wild-type MC4100 and MC4100(pCHAP8075). Significant precursor accumulation was observed in strain MM52, which carries the secA51(Ts) mutation and was used as a control. These results indicate that the properties of the PulG-LacZ hybrid protein are different from those of MalE-LacZ, suggesting underlying differences in the ways the two chimeras are targeted for export. Indeed, as discussed below, the phenotype of cells producing PulG-LacZ is reminiscent of that of cells producing a LamB-LacZ variant with an abnormally hydrophobic signal peptide (65).

Influence of sec mutations on PulG-LacZ localization.

To identify targeting factors involved in PulG membrane insertion, the PulG-LacZ hybrid was introduced into E. coli MC4100 derivatives with mutations affecting known protein export factors. Specific β-galactosidase activity was used as a measure of PulG-LacZ localization under conditions of basal and maltose-induced pulG-lacZ expression (Table 3). The ΔsecB mutation was without effect, indicating that SecB is not involved in PulG membrane targeting. The secA51(Ts) mutation led to a 20-fold increase in β-galactosidase activity under semipermissive conditions, and secY24(Ts) had an even more pronounced effect. The greatest increase in β-galactosidase activity was observed in the secY108 mutant. The novel secY108 mutant, isolated in a screen for mutants defective in localization of LacZ fused to ribose-binding protein RbsB (Francetic and Kumamoto, unpublished results), has an E108K change in the second cytoplasmic loop of SecY. The secD4004 and secE1610 mutations identified in the same screen had a weaker effect on the export of PulG-LacZ.

The two cold-sensitive mutations secY39 and secY40, shown previously to affect IM protein insertion (1, 34), both led to substantially impaired PulG-LacZ membrane localization at the semipermissive temperature. Since both of these mutations show a synthetic lethal phenotype when combined with deficiencies in the SRP pathway, their rather strong effect, in particular that of secY40, pointed to a possible SRP involvement in PulG targeting. A role for SRP was expected, based on the high level of hydrophobicity of the N-terminal α-helical domain of PulG (Fig. 2) (21).

FIG. 2.

Engineering the leader peptidase cleavage site in PulG. The primary sequence alignment of wild-type PulG and PulG* is shown, with sequence mismatches indicated with a rectangle. The position of the PulG sequence with respect to the cytoplasmic (CYT), inner membrane (IM), and periplasmic (PERI) compartments is shown. Arrows indicate the prepilin peptidase PulO and leader peptidase Lep cleavage sites. The wavy lines indicate the hydrophobic inner membrane (α-1N) and periplasmic amphipathic (α-1C) alpha helical domains. Kyte-Doolittle hydropathy plots of PulG* and wild-type PulG (PulGwt) are shown.

PulG membrane insertion assay.

Several approaches have been used in the biochemical analysis of IM protein insertion (20, 23, 29, 30). Although processing of PrePulG by PulO prepilin peptidase can be used as an indicator of export, only six amino acids are removed, making it difficult to resolve precursor and mature forms (40). Furthermore, PrePulG processing occurs on the cytoplasmic side of the IM and might occur without membrane insertion. To facilitate the analysis of PulG membrane insertion, we created PrePulG* by replacing two amino acids downstream from the transmembrane (TM) helix α-N1 of PulG (Fig. 2) with an artificial leader peptidase cleavage site (pCHAP8086; described in Materials and Methods). In MC4100 F′ lacI^q1(pCHAP8086), the majority of PulG* was found in its processed form, corresponding in size to the periplasmic domain of PulG (residues 25 to 140 of mature PulG), in the soluble, periplasmic fraction (Fig. 3A, wt). In contrast, significant accumulation of the unprocessed PrePulG* was observed in strains MC4100 secY108 and MC4100 secA7004 (Fig. 3A). The level of PrePulG* in these strains correlated well with the levels of export defects of the PulG-LacZ fusion observed in the same mutants (Table 3). Similarly, processing of PrePulG* was defective in MC4100 mutants secE1610 and secD4004 and in secA51(Ts) at 37°C (Fig. 3B and C). Processing of PreMalE was affected in all of the sec mutants.

FIG. 3.

Fractionation analysis of PulG* membrane insertion in wild-type MC4100 (wt) and secA7004, secA51(Ts) [secA (Ts)], secY108, secD4004(Cs), secE1610, secY39(Cs), and secY40(Cs) mutants. All strains are derived from MC4100 and carry the F′ lacI^q1 episome pAPIP501 (Table 1). Bacteria were grown at 30 or 37°C in LB media supplemented with Amp or chloramphenicol to an OD₆₀₀ of 0.8 and subjected to fractionation as described in Materials and Methods. Total cell (T), spheroplast (S), and periplasmic (P) fractions corresponding to 0.05 OD₆₀₀ unit were analyzed with 12% SDS-PAGE and transferred onto nitrocellulose membranes. The membranes were probed with anti-PulG and anti-MalE antibodies. The positions of full-length PrePulG* and processed PulG* and MalE are indicated. A band reacting nonspecifically with the anti-MalE antiserum is labeled PrtX.

The cold-sensitive mutation secY40 (strain TY22) specifically affects the insertion of IM proteins (34), unlike the secY39 allele (strain AD208) that affects the export of both IM and periplasmic (and outer membrane) proteins (1, 34). To determine the extent of the PrePulG* processing and release in the two strains, we introduced an F′ lacI^q1 episome by mating with strain CK1637 F′ lacI^q1. However, the resulting strains showed a growth defect when transformed with plasmid pCHAP8086, and PulG* could not be detected in their extracts. Only when PrePulG* was encoded by the low-copy-number plasmid pCHAP8100 could PulG* insertion be analyzed by fractionation in these mutants. In agreement with the LacZ activity measurements (Table 3), weak accumulation of PrePulG* was observed in both mutant strains. Overlapping bands cross-reacting with the anti-PulG antibodies made it difficult to quantify this defect in PrePulG* processing. As expected, the export of MalE was specifically defective only in strain AD208 F′ lacI^q1, with only very-low-level PreMalE accumulation being observed in strain TY22 F′ lacI^q1 (Fig. 3D).

Strain JS7131 carrying a paraBAD-yidC operon fusion and transformed with plasmid pCHAP8086 was used to test the effect of YidC on PrePulG*. No differences in PrePulG* processing were observed upon removal of arabinose (data not shown), which leads to YidC depletion (48). Immunoblotting revealed a substantial accumulation of PspA after arabinose removal, indicating that YidC levels were reduced (61).

Depletion of the SRP component Ffh completely abolished PrePulG* processing.

To test the involvement of the SRP pathway in PulG export, we introduced plasmid pCHAP8086 into the ffh conditional mutant WAM121 (paraBAD-ffh) (13). When WAM121(pCHAP8086) was grown in LB with 0.2% arabinose, PrePulG* was processed by leader peptidase, albeit less efficiently than in wild-type strains (Fig. 4). However, in cells grown in the presence of glucose (i.e., not expressing ffh), PrePulG* processing was completely abolished, while the processing of MalE and RbsB was unaltered, indicating that the Sec machinery was still functional. Interestingly, the cross-reacting band, migrating at about 29 kDa, detected by anti-MalE antibodies (labeled PrtX in Fig. 4), was also sensitive to SRP depletion. The processing and periplasmic release of DsbA, a known SRP pathway substrate (52), were more defective in WAM121 in the presence of glucose, and the protein was less abundant.

FIG. 4.

Depletion of the SRP component Ffh prevents PrePulG* processing and membrane insertion. Cells of strain WAM121(pCHAP8086) were grown overnight in LB medium supplemented with arabinose. The cells were washed twice with LB medium, diluted 1:50 into fresh LB medium supplemented with arabinose (Ara) or glucose (Glu), and grown to an OD₆₀₀ of 0.8. Total extracts (T) and spheroplast (S) and periplasmic fractions (P) from 0.05 OD₆₀₀ unit of bacteria were separated on 10% and 12% SDS-PAGE gels and transferred to nitrocellulose membranes, which were probed with antisera against PulG, MalE, RbsB, and DsbA. A band reacting nonspecifically with the anti-MalE antiserum is labeled PrtX.

PulG localization analyzed by fluorescence microscopy.

PulG is a relatively abundant component of the Pul secreton and is assembled into periplasmic filaments during pullulanase secretion. The hydrophobic region of PulG is required for SRP-dependent export (this study) as well as for filament assembly and for retention in the membrane in the unassembled state (24, 39). Previous attempts using sucrose gradient fractionation to determine the site where unassembled PulG accumulates gave ambiguous results (37, 42), and the authors of a study with the PulG homolog XcpT proposed that its localization might change upon processing and activation of the secreton (5). We previously described the cellular localization of two other components of the Pul secreton analyzed by fluorescence microscopy of GFP chimeras (10). Here, we used a similar approach to study the localization of PulG in live E. coli cells.

In the absence of Pul secreton factors, GFP-PrePulG was localized at the periphery of MC4100 cells expressing chromosomal gfp-pulG in strain PAP9102 and induced with 100 μM IPTG. Occasional brighter patches were observed, similar to GFP-PulM hybrid proteins seen previously (10). In the presence of the Pul secreton factors encoded by pCHAP710, the hybrid protein was cleaved at the prepilin peptidase recognition site on the cytoplasmic face of the IM, releasing the GFP moiety with a short peptide derived from the N terminus of PrePulG (labeled GFP*) to the cytoplasm (Fig. 5A and B, lanes 4 to 6). Low-level proteolysis observed in the absence of the Pul factors (Fig. 5B, lanes 1 to 3) could be due to one or both of the chromosomally encoded E. coli prepilin peptidases (18). Fluorescence microscopy of plasmolyzed cells producing GFP-PrePulG indicated that the protein was almost entirely located in the inner membrane (Fig. 6A), as shown previously for GFP-PulM (27).

FIG. 5.

Processing and localization of GFP-PrePulG in vivo. (A) Localization of GFP-PrePulG in wild-type E. coli strain MC4100. The left panel shows the expression of pulG-gfp from the chromosomal copy in strain PAP9102 induced with 100 μM IPTG. The center panel shows the processing of GFP-PrePulG in strain PAP9102 carrying plasmid pCHAP710, which encodes the complete Pul secretion system. The right panel shows the localization of the cytoplasmic GFP produced in strain PAP9121. (B) Production of GFP-PrePulG hybrid proteins in strains PAP9102 and PAP9102(pCHAP710) analyzed by immunoblot using anti-GFP antibodies (Clontech). The levels of the GFP-PrePulG hybrid protein induced by increasing concentrations of IPTG in the absence (−, lanes 1, 2 and 3) or presence (+, lanes 4, 5 and 6) of Pul secreton factors are indicated. The migration and mass (kDa) of the molecular size markers are indicated on the left. (C) Membrane localization of PrePulG-mRFP1 in the wild-type E. coli strain MC4100 carrying pCHAP7516 (left, PrePulG-mRFP1) or pCHAP7520 (right, cytoplasmic mRFP1).

FIG. 6.

Inner membrane localization of PulG. GFP-PrePulG was produced from the chromosome in strain PAP9102 under 10 μM IPTG induction (A) or as a C-terminal mRFP fusion from plasmid pCHAP7516 in strain PAP7500BG without (Pul⁻) (B) or with (Pul⁺) (C) the presence of 0.2% maltose to induce production of Pul secretion factors. By using fluorescence microscopy, hybrid proteins were located in the envelope of untreated cells (Control) or specifically in the IM from the OM (Plasm.). The arrowheads indicate plasmolysis bays. Each panel consists of a phase-contrast image (left) and the corresponding fluorescence image (right).

Fusing PulG to mRFP1 (11), which can fold correctly in the E. coli periplasm (27), allowed the effect of other Pul factors on the localization of PulG to be analyzed. The localization pattern of this chimera, encoded by plasmid pCHAP7520, was essentially the same as that of GFP-PrePulG (Fig. 5C and 6B). The maltose-induced production of Pul secreton factors did not change the distribution of PulG-mRFP1 inside cells, which remained IM associated (Fig. 6C). These data resolve previously ambiguous fractionation PulG data. For example, PulG fractionated mainly with the IM during sucrose gradient centrifugation but substantial amounts were also found in the outer membrane (OM) compared to the amounts of other IM proteins such as PulC and DjlA (Fig. 7A). The distribution of PulG-mRFP, as determined by fractionation, was similar to that of PulG, irrespective of whether the Pul secreton factors were present (Fig. 7B). Thus, the prepilin signal sequence of PulG probably anchors it mainly in the IM but substantial amounts (as much as 50% in some experiments, e.g., Fig. 7B, lower panel) might reassociate with the OM when the cells are disrupted.

FIG. 7.

Flotation sucrose gradient fractionation of PulG (A) and PulG-mRFP (B). Membranes obtained by French press disruption of cells of E. coli(pCHAP231) (A) or E. coli(pCHAP7516) with (B, lower panel) or without (B, upper panel) pCHAP710 (encoding all Pul secreton factors) were harvested from the gradients and analyzed by immunoblotting with antibodies against PulC, DjlA, Pal, PulS, (A), and PulG (A and B). The positions of the inner and outer membranes are indicated. Note that PrePulG-mRFP is partially degraded to a form similar in size to PrePulG (B, upper panel) and that this band is probably the faster migrating of the two bands recognized by the antibodies when the bacteria also produced PulG encoded by pCHAP710 (B, lower panel).

DISCUSSION

Properties of the prepilin signal sequence explain the mechanism of PulG export.

On the basis of studies with LacZ chimeras and with derivatives with an artificial leader peptidase cleavage site downstream from the prepilin signal sequence, we conclude that the SRP and Sec pathways are directly involved in the export of the pseudopilin PulG and, by inference, of all pseudopilins and related type IV pilins. PrePulG presumably inserts into the IM according to the positive-inside rule (62), with the positively charged N-terminal region of its prepilin signal sequence and prepilin peptidase cleavage site facing the cytoplasm and the C-terminal globular domain in the periplasm. This orientation is consistent with the fact that fusion of alkaline phosphatase (45) and β-lactamase (39) to the C terminus of full-length or truncated PulG leads to enzymatically active, periplasmically oriented membrane proteins and with the presence in pseudopilin PulK (41) and type IV pilins (12) of an intramolecular disulfide bond whose formation is catalyzed by periplasmic DsbA.

The prepilin signal sequence promotes export though the Sec translocon, anchors the protein in the inner membrane, and promotes subunit assembly into a pseudopilus or type IV pilus filament. The first two functions are independent of the sequence of the prepilin signal sequence, but pilus assembly requires precise interactions between residues in the signal sequences of adjacent protomers that are highly conserved in all members of the pseudopilin/type IV pilin family. The very high hydrophobicity of the prepilin signal sequence (Fig. 2) (13, 21, 60) presumably explains the absolute requirement for SRP for targeting to the Sec translocon. The data presented here rule out the possibility that secreton components, and in particular the PulCEFLM complex (38), are involved in pseudopilin translocation across the IM. Instead, this complex extracts the prepilin signal sequence from the IM and permits its interaction with adjacent prepilin signal sequences in the assembling filament. Arts et al. (3) report essentially identical conclusions in the accompanying paper on the export of the PulG homolog XcpT from Pseudomonas aeruginosa. In the following paragraphs, we discuss specific features of the SRP and Sec requirements for PulG.

SRP.

The role of the SRP pathway in cotranslational IM protein biogenesis is now clearly established. The alternative, YidC-only pathway is used by a small number of proteins (48, 61). Here we present strong evidence for the involvement of SRP but not YidC in PulG targeting. The PulG-LacZ hybrid protein failed to jam the export machinery when overproduced. This is in marked contrast to other LacZ chimeras, such as MalE-LacZ and LamB-LacZ, which are toxic because they jam the Sec export machinery. However, a LamB-LacZ chimera in which the hydrophobicity of the LamB signal peptide had been artificially increased was not toxic because it was routed to the Sec machinery via SRP and was translocated cotranslationally (65). Thus, PulG-LacZ is very similar to this LamB-LacZ variant. Depletion of the SRP component Ffh had the most dramatic effect of any of the conditions tested on PulG export, and the lack of posttranslational processing of PulG with an artificial leader peptidase cleavage site is also consistent with cotranslational PulG membrane insertion.

The Sec translocon.

In vivo studies showed that SRP targets IM proteins to the SecYEGDF translocon for insertion (23, 60), while in vitro cross-linking of nascent IM proteins revealed contacts with Ffh, Fts, SecY, and SecA (43, 52). SecA is required for the insertion of IM proteins with periplasmic loops longer than 30 amino acids (1, 19, 43), while single-spanning IM proteins such as PulG always require SecA, regardless of the length of their periplasmic domain (15). Consistent with this, SecA depletion almost completely abolished PrePulG* processing. Several integral Sec membrane components were also shown to be required, including SecD, which is affected by the novel secD4004 mutation that introduces a charged residue (G506D) in the fourth transmembrane segment, in a region highly conserved in members of the SecD family.

The secY108A mutation exerted a particularly strong effect on PulG insertion. The charge inversion in the second cytoplasmic loop of the SecY E108K variant could affect the interaction with SecA. The cold-sensitive alleles secY39 and secY40 also diminished PulG-LacZ export, consistent with their synthetically lethal effect when combined with defects in the SRP pathway (34). The residues affected by secY39 (R357H) and secY40 (A363T) both map in the fifth cytoplasmic loop of SecY and affect a translocation step subsequent to substrate binding (33). The mutation secY40 appears to interfere with docking of the SRP receptor FtsY to the translocon (2). Export of the aforementioned LamB-LacZ hybrid that was redirected to the SRP pathway (65) was specifically affected only by the secY40 allele, but this was not the case with PulG or with DsbA, another SRP substrate (54). Interestingly, overproduced PulG was specifically degraded in both of these secY mutants, possibly by cytoplasmic proteases Lon and ClpQ (8), whereas the accumulation of MalE was unaffected (data not shown).

Adding a leader peptidase cleavage site to PrePulG (to create PrePulG*) did not change the hydrophobicity of the prepilin signal sequence (Fig. 2), which presumably plays a decisive role in the choice of the targeting pathway. Processing of PrePulG* on the periplasmic side of the hydrophobic TM segment provided a direct kinetic assay for correct membrane insertion. However, pulse-chase analyses showed that PrePulG* is not processed posttranslationally, as observed for the DsbA protein, an SRP substrate with a natural cleavable signal sequence (52, 54). Therefore, we measured the steady-state levels of PulG by fractionation and immunoblotting as sensitive means to probe PulG insertion, especially in rich media that lead to more pronounced protein translocation defects (34). This approach also allowed detection of the weak MalE export defect in the secY40 mutant that was not detected previously by pulse-chase analysis (4, 34). A defect in prepilin peptidase PulO-mediated processing of PulG was also demonstrated in SRP-depleted strains by using a variant of prePulG in which the propeptide was artificially extended with an N-terminal FLAG tag (data not shown). In the accompanying paper, Arts et al. show that processing of native XcpT, a close homologue of PulG, by prepilin peptidase also depends on SRP and the Sec machinery (3).

Localization of intracellular PulG pools.

Previous studies with the Vibrio cholerae PulG homolog EpsG indicated that it clusters at the cell pole, consistent with the facts that type II secretion occurs at the pole in this bacterium and that other secreton components also accumulate there (53). In contrast, GFP-PrePulG and PulG-mRFP were evenly distributed throughout the E. coli inner membrane, consistent with previous data showing similar patterns of GFP-PulM and GFP-PulL distribution (10). The absence of clustering was surprising, since unassembled PulG can be chemically cross-linked into large complexes (39) that suggested the existence of discrete pools of PulG at distinct, putative assembly sites. However, neither PreGFP-PulG nor PulG-mRFP could be cross-linked into similar complexes, possibly because the fluorescent reporter interfered with clustering or with the positioning of amine groups in adjacent chimeras. The GFP-PrePulG chimera was functional, however, when cleaved by prepilin peptidase PulO. Thus, these data do not support the idea that PulG accumulates at specific assembly sites in the E. coli cell envelope, although we cannot exclude the possibility that native PulG might show a different distribution from that of fluorescent reporter-tagged PulG.

The analysis of PulG-mRFP1 indicates accumulation of unassembled PulG mainly or exclusively in the inner membrane. These data are inconsistent with sedimentation sucrose gradient fractionation of membranes derived by disrupting spheroplasts, which indicated OM association of PulG irrespective of the presence of other secreton components (42). Mainly inner membrane association of PulG was indicated here, however, by flotation sucrose gradient fractionation of membranes obtained by disrupting cells in a French pressure cell (Fig. 7A), although large amounts were again found in OM fractions, compared to the amounts of authentic inner membrane proteins. These studies underscore the intrinsic value of in situ localization using fluorescent reporter proteins (27) (Fig. 6), which are less subject to artifacts introduced during cell disruption and fractionation, although the reporter protein might not be entirely neutral. Importantly, none of the data in Fig. 6 and 7 indicate relocalization of PulG upon the expression of other Pul secreton factors. This makes it unlikely that the secreton relocalizes the major pseudopilin to the OM, as was proposed previously (5).