Coordinated Regulation of Expression of Salmonella Pathogenicity Island 1 and Flagellar Type III Secretion Systems by ATP-Dependent ClpXP Protease

Salmonella enterica serovar Typhimurium delivers a variety ofproteins via the Salmonella pathogenicity island 1 (SPI1)-encodedtype III secretion system into host cells, where they elicitseveral physiological changes, including bacterial invasion,macrophage apoptosis, and enteropathogenesis. Once Salmonellahas established a systemic infection, excess macrophage apoptosiswould be detrimental to the pathogen, as it utilizes macrophagesas vectors for systemic dissemination throughout the host. Therefore,SPI1 expression must be restricted to one or a few specificlocations in the host. In the present study, we have demonstratedthat the expression of this complex of genes is repressed bythe ATP-dependent ClpXP protease, which therefore suppressesmacrophage apoptosis. Depletion of ClpXP caused significantincreases in the amounts of two SPI1-encoded transcriptionalregulators, HilC and HilD, leading to the stimulation of hilAinduction and therefore activation of SPI1 expression. Our evidenceshows that ClpXP regulates cellular levels of HilC and HilDvia the control of flagellar gene expression. Subsequent experimentsdemonstrated that the flagellum-related gene product FliZ controlsHilD posttranscriptionally, and this in turn activates HilC.These findings suggest that the ClpXP protease coregulates SPI1-relatedvirulence phenotypes and motility. ClpXP is a member of thestress protein family induced in bacteria exposed to hostileenvironments such as macrophages.

INTRODUCTION

Top

INTRODUCTION

Salmonella enterica serovar Typhimurium is a facultative intracellularpathogen that causes gastroenteritis in humans and systemicdiseases similar to typhoid fever in mice. Specific virulencefactors encoded within the Salmonella pathogenicity islands(SPIs) are required for the development of salmonellosis (20,24, 37, 43). One SPI, SPI1, encodes a type III secretion systemwith a multiprotein secretion apparatus, termed a needle complex,as well as effector proteins that are injected into the hostcell cytoplasm via this apparatus (19, 32, 33, 56, 67). Variousfunctions have been attributed to the SPI1 type III secretionsystem, including actin rearrangement, which causes the bacteriumto be engulfed (9, 18, 47); macrophage apoptosis (7, 26, 30);and enteropathogenesis (65). After invasion of intestinal epithelialcells, Salmonella is able to disseminate to any tissue. It propagateswithin macrophages, a process that requires a second type IIIsecretion system encoded on SPI2. Once Salmonella has establisheda systemic infection, excess macrophage apoptosis would be detrimental.During this systemic phase, the organisms need to repress anddelay the onset of apoptosis to allow sufficient time for themto replicate, escape, and invade new macrophages. Thus, SPI1gene expression must be suppressed during the systemic phaseof infection. We have previously reported that a member of theATP-dependent protease family, Lon, negatively regulates SPI1gene expression (58, 60) and that this down-regulation is quiteimportant for suppressing SPI1-dependent apoptosis sufficientlyto allow time for Salmonella to replicate within the macrophages(57). It has been demonstrated that Lon is also important fordown-regulating SPI1 gene expression in epithelial cells (5).

ATP-dependent proteolysis functions as a precise regulatorymechanism that limits the availability of key enzymes or criticalregulatory proteins controlling gene expression. In additionto Lon, four ATP-dependent proteases, i.e., ClpAP, ClpXP, HslVU,and FtsH, have been found in gram-negative bacteria (for a review,see references 22 and 42). These are collectively called AAA⁺(ATPases associated with diverse cellular activities plus) proteases(46). ClpXP is a bipartite protease responsible for degradingcertain key regulatory proteins and aberrant translation productsbearing the SsrA degradation tag, which is added cotranslationallyto nascent polypeptides when ribosomes stall (16, 21, 23, 61).The ClpP component of ClpXP consists of two stacked heptamericrings, which enclose a central chamber containing the proteolyticactive site. The ClpX component is a hexameric-ring ATPase thatbinds substrate proteins, denatures them, and translocates theunfolded polypeptides into the ClpP degradation chamber (48).

In the present study, we demonstrate that ClpXP negatively regulatesthe expression of SPI1 and hence suppresses macrophage apoptosisand the invasion of epithelial cells. The control of expressionof the SPI1 type III secretion system is complex: several transcriptionalregulators are present within the island. Among these, the masterregulator is HilA, a member of the OmpR/ToxR family of transcriptionalregulators, which can activate expression either directly orindirectly by increasing the expression of another regulator,InvF (2, 31, 37, 39). InvF, in a complex with the chaperoneprotein SicA, induces expression of the sic/sip operon (11).Two homologous proteins, HilC and HilD, both members of theAraC/XylS family of transcriptional regulators, induce hilAexpression (12, 51, 53). HilC and HilD can each individuallybind to the DNA immediately upstream of hilA, and it is believedthat this binding leads to hilA expression (12, 52, 53). Itis further suggested that HilD acts as a direct activator ofhilA (6), as opposed to a derepressor as previously assumed(52, 53). It has also been demonstrated that HilC and HilD eachinduce the expression of the other as well as inducing theirown promoters (40, 47). Recent data have shown that loss ofHilD decreases hilC transcription but that loss of HilC doesnot significantly alter hilD expression (13, 15), suggestingthat HilD may be at the top of the hierarchy of regulation ofSPI1 expression. In addition to these SPI1-encoded regulators,genetic studies have identified many regulatory proteins encodedoutside the island. One such regulator, Lon, has been shownto degrade both HilC and HilD, thus down-regulating SPI1 geneexpression (58).

The structure of the SPI1 type III secretion apparatus is similarto the basal structure of the flagellum (4). Assembly of a flagellumrequires the export of protein subunits from the cytoplasm tothe outer surface of the cell by a mechanism that resemblesthose of secretion systems (4, 27). Furthermore, many componentsof the flagellar export apparatus share a high degree of homologywith those of the type III secretion machinery. These findingssuggest that these two type III machineries have a common evolutionaryorigin. We have previously reported that ClpXP negatively regulatesflagellum biogenesis by controlling the turnover of the FlhD/FlhCcomplex, which functions as a master regulator at the apex ofthe transcription hierarchy of the flagellar regulon (61, 62).

Here, we have studied the mechanism by which ClpXP negativelycontrols SPI1 gene expression. The present results, coupledwith our previous results (61, 62), indicate that ClpXP coregulatesthe expression of two different type III secretion systems,SPI1 related and flagellum related.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions. All strains used in this study are derivatives of S. entericaserovar Typhimurium ${chi}$ 3306 and are listed in Table 1. PlasmidspTKY559 and pTKY562 carry a hilC promoter-lacZ fusion and ahilD promoter-lacZ fusion, respectively (60). Plasmid pTKY651carries a P_A1lacO-1 promoter-hilD fusion (60).

View this table:
[in this window]
[in a new window]

TABLE 1. Bacterial strains used in this study

To construct the fliZ disruption mutants, P22 phages were propagatedon KK1397 (34) to transduce the fliZ::Km mutation. The lysatewas used for infection of S. enterica serovar Typhimurium strains,and the transductants were selected by kanamycin resistance.Bacterial cells were routinely grown in L broth (1% Bacto tryptone[Difco, Detroit, MI], 0.5% Bacto yeast extract [Difco], 0.5%sodium chloride [pH 7.4]) or on L agar. The media were supplementedwith chloramphenicol (25 µg ml^–1), ampicillin (25µg ml^–1), kanamycin (25 µg ml^–1), and/ornalidixic acid (25 µg ml^–1) when necessary. Bacterialcells were grown aerobically in L broth.

DNA techniques. DNA purification, ligation, restriction analysis, PCR amplification,DNA sequencing, and agarose gel electrophoresis were carriedout as previously described (66).

Macrophage infection. Cultured RAW264.7 macrophages were seeded onto a 24-well plateat a density of 2 x 10⁵cells per well and incubated overnightat 37°C. Prior to infection, the macrophages were washedwith Hanks balanced salt solution (HBSS) and challenged withSalmonella strains at a multiplicity of infection of 10. Theplates were centrifuged for 5 min at 500 x g to enhance andsynchronize infection, followed by incubation for 30 min at37°C to permit phagocytosis. The free bacteria were removedby three washes with HBSS. Dulbecco modified Eagle medium (DMEM)containing 10% fetal bovine serum and 100 µg of gentamicinper ml was added, and the cells were incubated for 1.5 h at37°C. The cells were washed with HBSS three times, followedby incubation with DMEM containing 10% fetal bovine serum and10 µg of gentamicin per ml at 37°C.

Detection of histone-associated cytoplasmic DNA fragments after Salmonella-induced macrophage cell death. At 6 h after infection, the macrophages were washed with HBSSand treated with 0.5 ml of lysis buffer for 30 min at room temperature.The lysates were centrifuged at 200 x g for 10 min at 4°C,and a 0.3-ml portion of the supernatant was taken for furtherassay. The samples were diluted 1:100, and the cytoplasmicallylocated histones bound to fragmented DNA were quantified colorimetricallyusing Cell Death Detection ELISA^PLUS (Roche Diagnostics).

Assay for invasion of epithelial cells. Intestine-407 cells were invaded by Salmonella in 24-well tissueculture plates as described previously (58, 59) with some modifications.Bacterial cells were washed with HBSS and used to inoculatemonolayers previously washed with HBSS, at a multiplicity ofinfection of 10. The monolayers were incubated for 2 h at 37°Cand then washed thoroughly with HBSS and lysed with 0.2% TritonX-100 in phosphate-buffered saline to determine the total numberof bacteria associated with the cultured cells. Alternatively,to assess the number of intracellular bacteria, the infectedtissue culture cells were further incubated for 3 h in DMEMcontaining 100 µg of gentamicin per ml to eliminate extracellularbacteria before lysis with the Triton X-100 solution. Bacterialnumbers were determined by plating the lysates on L agar platesafter appropriate dilution.

Collection of whole-cell proteins and secreted proteins. Bacterial cells were grown aerobically in 10 ml of L broth toan optical density at 600 nm of 1.0 at 37°C. To preparewhole-cell proteins, bacterial cells were harvested by centrifuging1 ml of the culture and then suspended in 200 µl of samplebuffer (36). Cell lysates (20 µl) were applied to eachlane of a sodium dodecyl sulfate (SDS)-polyacrylamide gel. Toprepare the proteins secreted into the medium, 7 ml of the sameculture was centrifuged to remove cells. The filtered supernatantwas mixed with prechilled trichloroacetic acid (final concentration,10%), chilled on ice for 15 min, and centrifuged at 10,000 xg for 20 min. The pellet was washed once with acetone and thensolubilized in 70 µl sample buffer (36). The sample (15µl) was loaded on each lane and subjected to immunoblottinganalysis.

Immunoblot analysis. Equivalent numbers of bacterial cells were suspended in samplebuffer (36), boiled for 5 min, and subjected to SDS-10% polyacrylamidegel electrophoresis. Separated proteins on the gels were transferredonto Immobilon-P polyvinylidene difluoride membranes (Millipore,Bedford, MA). Proteins were reacted with rabbit anti-SipB serum(1:12,500), anti-SipC serum (1:12,500), anti-HilA serum (1:12,500),anti-HilC serum (1:12,500), and anti-HilD serum (1:12,500),followed by alkaline phosphatase-conjugated anti-rabbit immunoglobulinG as the secondary antibody. The enzymatic reactions were performedin the presence of 300 µg ml^–1 of nitroblue tetrazolium(Dojin, Kumamoto, Japan) and 150 µg ml^–1 of bromochloroindolylphosphate(Amresco, Solon, OH). Anti-HilA, -HilC, -HilD, and -SipC antiserawere previously established in our laboratory (60).

β-galactosidase assay. The activity of β-galactosidase was determined by the methodof Platt et al. (49). The enzyme units presented here are theaverages from at least three independent assays.

RESULTS

Top

RESULTS

Depletion of the ClpXP protease in S. enterica serovar Typhimurium stimulates the induction of apoptosis in macrophages. In a previous study, we characterized the genomic organizationof the clpP region of S. enterica serovar Typhimurium pathogenicstrain ${chi}$ 3306, clpP-clpX-lon, and constructed an insertional mutationin the clpP gene (66). The insertion of a chloramphenicol resistancecassette into clpP resulted in a polar mutation in clpX, whichtogether with clpP constitutes an operon. However, the insertionevidently did not block the expression of lon preceded by itsown promoter, which is downstream of clpX, because anti-Lonantiserum detected the corresponding protein in ${Delta}$ clpPX mutantcells (66). Disruption of the clpP gene did not affect the cellgrowth rate (62, 66). During characterization of the virulencephenotype of the ${Delta}$ clpPX mutant, we found cytoplasmic histone-associatedDNA fragments, which constitute a marker for apoptotic cells(Fig. 1A), indicating that the mutant induces rapid and excessivemacrophage apoptosis (6 h after infection). Rapid macrophageapoptosis by Salmonella infection is known to be mediated byan SPI1-encoded protein, SipB (26). Although Salmonella alsoinduces SPI2-dependent, SPI1-independent apoptosis of macrophages,the execution of SPI2-dependent apoptosis is considerably delayed(18 to 24 h after infection) and is masked when the SPI1 genesare expressed (44, 64). Another aspect of SPI1-related virulence,the invasion of epithelial cells, was assessed in both the ${Delta}$ clpPXmutant and the parental cells using cultured Intestine-407 cells(Fig. 1B). The level of invasion by ${Delta}$ clpPX was significantlyhigher than that of invasion by the parental clpPX⁺ strain,suggesting that ClpXP is involved in the control of Salmonellainvasiveness.

View larger version (16K):
[in this window]
[in a new window]

FIG. 1. (A) Levels of cytoplasmic nucleosomal fragments in macrophages infected with S. enterica serovar Typhimurium strains ${chi}$ 3306 (clpPX⁺) and CS2007 ( ${Delta}$ clpPX). RAW264.7 cells were infected with bacteria at a multiplicity of infection of 10. At 6 h after infection, the levels of histone-associated cytoplasmic DNA fragments were measured by quantitative enzyme-linked immunosorbent assay using an antihistone antibody as described in Materials and Methods. The data are the means and standard derivations for each strain tested in triplicate. (B) Invasion efficiency of strains ${chi}$ 3306 (clpPX⁺) and CS2007 ( ${Delta}$ clpPX). Cultured Intestine-407 cells were used to assess invasiveness as described in Materials and Methods. The data are the means and standard derivations for each strain tested in triplicate. Invasion efficiency was expressed as the ratio of the number of intracellular bacteria to the number of bacteria adhering to the cultured cells.

ClpXP negatively regulates the expression of the SPI1 regulon. The enhanced ability of ${Delta}$ clpPX to induce macrophage apoptosisand to invade epithelial cells suggested that clpPX deletionresults in increased expression of SPI1 proteins. To examinethis possibility, we initially used immunoblotting to comparethe cellular levels of SipB and SipC, which are known to translocateinto epithelial cells, in ${Delta}$ clpPX cells and the isogenic clpPX⁺ cells. SipB is necessary and sufficient for the inductionof rapid macrophage apoptosis (26). SipC itself is known tocause bundling of actin filaments directly (19). The amountsof SipB and SipC in ${Delta}$ clpPX cells were apparently higher thanthose in the clpPX⁺ cells (Fig. 2A). We have previously identifiedthe SPI1 proteins SipA and SipC by mass spectrometry of secretedproteins separated by SDS-polyacrylamide gel electrophoresis(58). The Coomassie blue-stained gel patterns of the secretedproteins indicate that significantly higher levels of SipA andSipC are secreted in the ${Delta}$ clpPX mutant (Fig. 2C). The 50-kDaband that accumulates in the ${Delta}$ clpPX strain (Fig. 2B) is a flagellinprotein, FliC, as previously identified (62). S. enterica serovarTyphimurium expresses two antigenically distinct flagellinsencoded by fliC and fljB, and the alternative expression ofthese two genes results in an oscillation of phenotype knownas phase variation, which occurs with frequencies ranging from10^–3 to 10^–5 per bacterium per generation (55).Mass spectrometry demonstrated that cultures of strain ${chi}$ 3306(clpPX⁺) include both the FliC and FljB proteins (62). The profileof secreted proteins shows that FliC and FljB are greatly increasedby the disruption of clpPX (Fig. 2C). We then examined the cellularlevels of HilA, a transcriptional activator with a central rolein the regulatory hierarchy of SPI1 gene expression. The resultsshow that the level of HilA is greatly increased in ${Delta}$ clpPX cells.Two SPI1 gene products, HilC and HilD, have been shown to bindto the upstream repressing sequence of hilA to induce its transcription(12, 52, 53). In addition, overproduction of HilC and HilD hasbeen shown to increase hilA expression significantly (13, 40).It is suggested that HilD directly activates hilA transcription(6). Therefore, the amounts of HilC and HilD in ${Delta}$ clpPX cellswere compared with those in the clpPX⁺ cells. The levels ofboth HilC and HilD were significantly increased by disruptionof clpPX, suggesting that the ClpXP protease may down-regulatehilA transcription by a mechanism dependent on both HilC andHilD.

View larger version (27K):
[in this window]
[in a new window]

FIG. 2. Effect of depletion of ClpXP protease on cellular levels of various SPI1 proteins. Bacterial cells of strains ${chi}$ 3306 (clpPX⁺) and CS2007 ( ${Delta}$ clpPX) were used. (A) Immunoblotting analysis of the cellular lysates using anti-SipB, anti-SipC, anti-HilA, anti-HilC, and anti-HilD sera. (B) Coomassie brilliant blue-stained SDS-10% polyacrylamide gel electrophoretic patterns of the same samples used for immunoblotting. Lane M contains the molecular mass standard. (C) Coomassie brilliant blue-stained gel patterns of the proteins secreted into the medium. The proteins were prepared as described in Materials and Methods.

To examine this possibility, the effect of clpPX disruptionon the expression of hilA was examined by use of a chromosomallacZY fusion in strains with ${Delta}$ hilC and/or ${Delta}$ hilD backgrounds. Asshown in Fig. 3A, disruption of clpPX increased hilA transcriptionapproximately sixfold in hilC⁺ hilD⁺ cells. In contrast, thestimulation of hilA expression in ${Delta}$ clpPX cells was completelysuppressed by introducing the ${Delta}$ hilC and/or ${Delta}$ hilD mutation. Theresults of immunoblotting of cell lysates prepared from variousstrains with the mutation backgrounds used for transcriptionalanalysis (Fig. 3A) are shown in Fig. 3B. The apparent levelsof HilA are consistent with the transcriptional analysis results.That is, the amount of HilA in the ${Delta}$ clpPX mutant cells was largerthan that in clpPX⁺ cells, and the accumulation of HilA in ${Delta}$ clpPXcells was suppressed by introducing the hilC and/or hilD mutation.These results strongly suggest that ClpXP regulates hilA transcriptionthrough a pathway dependent on both HilC and HilD.

View larger version (17K):
[in this window]
[in a new window]

FIG. 3. Effect of depletion of ClpXP protease on the expression of the hilA gene in the absence of the hilC and/or hilD gene. (A) The expression levels of a lacZ fusion to the hilA promoter were assayed. The values represent the means and standard deviations for samples tested at least in triplicate. The strains used were CS2110 (clpPX⁺ hilC⁺ hilD⁺), CS2120 ( ${Delta}$ clpPX hilC⁺ hilD⁺), CS2732 (clpPX⁺ ${Delta}$ hilC hilD⁺), CS3319 ( ${Delta}$ clpPX ${Delta}$ hilC hilD⁺), CS2733 (clpPX⁺ hilC⁺ ${Delta}$ hilD), CS3320 ( ${Delta}$ clpPX hilC⁺ ${Delta}$ hilD), CS2815 (clpPX⁺ ${Delta}$ hilC ${Delta}$ hilD), and CS3321 ( ${Delta}$ clpPX ${Delta}$ hilC ${Delta}$ hilD). (B) Immunoblotting analysis of the cellular lysates using anti-HilA serum. The strains used were ${chi}$ 3306 (clpPX⁺ hilC⁺ hilD⁺), CS2007 ( ${Delta}$ clpPX hilC⁺ hilD⁺), CS2724 (clpPX⁺ ${Delta}$ hilC hilD⁺), CS3322 ( ${Delta}$ clpPX ${Delta}$ hilC hilD⁺), CS2725 (clpPX⁺ hilC⁺ ${Delta}$ hilD), CS3325 ( ${Delta}$ clpPX hilC⁺ ${Delta}$ hilD), CS2802 (clpPX⁺ ${Delta}$ hilC ${Delta}$ hilD), CS3328 ( ${Delta}$ clpPX ${Delta}$ hilC ${Delta}$ hilD), and CS2110 ( ${Delta}$ hilA).

We then examined the effect of clpPX disruption on the transcriptionfrom hilC and hilD promoters using lacZ fusion on plasmids.As shown in Fig. 4A, disruption of clpPX increased hilC transcription2.6-fold and hilD transcription 1.9-fold. HilC and HilD eachactivate the expression of the other as well as inducing theirown promoters (39, 47), so we then measured transcription fromthe hilC and hilD promoters on plasmids in hilC- or hilD-disruptedbackgrounds on the chromosome. The results show that the enhancedclpPX disruption effect on the transcription from hilC and hilDpromoters was abolished by introducing either ${Delta}$ hilC or ${Delta}$ hilD mutationson the chromosome. The immunoblotting analyses show that theincreased clpPX disruption effect on cellular levels of HilCand HilD disappeared when either ${Delta}$ hilD or ${Delta}$ hilC was introduced(Fig. 4B). Together, these findings suggest that the both HilCand HilD need to be accumulated to cause a significant stimulationof hilA expression by clpPX disruption.

View larger version (20K):
[in this window]
[in a new window]

FIG. 4. Effect of depletion of ClpXP protease on the expression of the hilC and hilD genes in the absence of hilC or hilD gene. (A) The expression levels of lacZ fusions to hilC and hilD promoters were assayed. The values represent the means and standard derivations for samples tested at least in triplicate. Plasmid pTKY559 (hilC promoter-lacZ fusion) or pTKY562 (hilD promoter-lacZ fusion) was introduced into bacterial strains ${chi}$ 3306 (clpPX⁺ hilC⁺ hilD⁺), CS2007 ( ${Delta}$ clpPX hilC⁺ hilD⁺), CS2724 (clpPX⁺ ${Delta}$ hilC hilD⁺), CS3322 ( ${Delta}$ clpPX ${Delta}$ hilC hilD⁺), CS2725 (clpPX⁺ hilC⁺ ${Delta}$ hilD), and CS3325 ( ${Delta}$ clpPX hilC⁺ ${Delta}$ hilD). The resultant strains were used for determination of β-galactosidase activity. (B) Immunoblotting of the cellular lysates using anti-HilC and anti-HilD sera. The strains used were ${chi}$ 3306, CS2007, CS2724, CS3322, CS2725, and CS3325.

ClpXP regulates SPI1 gene expression via control of flagellar gene expression. How does ClpXP control the levels of HilC and HilD? Our previousfinding that it negatively regulates the expression of the flagellarregulon by controlling the turnover of the master regulator,FlhD/FlhC (62), allowed us to hypothesize that it may regulateHilC and HilD by controlling flagellar gene expression. In S.enterica serovar Typhimurium, the transcription of more than50 genes related to flagellum biogenesis forms a highly orderedcascade divided into three classes, 1, 2, and 3 (35). At theapex of the transcription hierarchy lies the sole operon flhDC.Its products, FlhD and FlhC, act together in a FlhD/FlhC hetero-oligomerto activate the promoter of class 2 genes (28, 38). Our previousstudy demonstrated that ClpXP recognizes and degrades the FlhD/FlhChetero-oligomer, resulting in down-regulation of the flagellarregulon (61). One of seven operons in class 2, fliAZY, encodesan alternative sigma factor, FliA ( ${sigma}$ ²⁸), that is required forthe transcription of class 3 genes, producing FliZ and FliY.Previous studies have reported that the fliA::Tn5 mutation significantlydecreases expression of the hilA-lacZ transcriptional fusionand that expression is restored by a trans-complementing plasmidexpressing fliZ, suggesting that FliZ is required for hilA expression(29, 41). Therefore, ClpXP may possibly control the cellularlevels of the transcriptional regulators HilC and HilD by apathway strongly dependent on FliZ. To examine this possibility,we measured the levels of HilC and HilD in clpPX⁺ and ${Delta}$ clpPXcells after introducing a ${Delta}$ fliZ mutation. The results (Fig. 5)show that the enhancement of cellular HilC and HilD levels byclpPX disruption disappeared in the ${Delta}$ fliZ mutant, suggestingthat the control of HilC and HilD levels by ClpXP depends onthe function of FliZ. Owing to the effect of the ${Delta}$ fliZ mutationon HilC and HilD levels, the enhancement of HilA by clpPX disruptionwas also extinguished. As expected, disruption of the flhDCoperon, encoding the master regulator for expression of theflagellar regulon, blocked the increase of HilC and HilD levelscaused by clpPX disruption. Taking these results together, itis suggested that SPI1 expression is regulated by ClpXP througha mechanism by which FliZ controls the cellular levels of thetranscriptional regulators HilC and HilD.

View larger version (19K):
[in this window]
[in a new window]

FIG. 5. Effects of disruption of fliZ or flhDC on enhancement of cellular levels of SPI1 regulator proteins by depletion of ClpXP. Whole-cell lysates were separated on SDS-10% polyacrylamide gels and then subjected to immunoblotting using anti-HilA, anti-HilC, or anti-HilD serum. The bacterial strains used were ${chi}$ 3306 (clpPX⁺ fliZ⁺ flhDC⁺), CS2007 ( ${Delta}$ clpPX fliZ⁺ flhDC⁺), CS2462 (clpPX⁺ ${Delta}$ fliZ flhDC⁺), CS2464 ( ${Delta}$ clpPX ${Delta}$ fliZ flhDC⁺), CS2609 (clpPX⁺ fliZ⁺ ${Delta}$ flhDC), CS2610 ( ${Delta}$ clpPX fliZ⁺ ${Delta}$ flhDC), CS2802 ( ${Delta}$ hilC ${Delta}$ hilD), and CS3222 ( ${Delta}$ hilA).

Transcriptional analysis of hilC and hilD in fliZ⁺ and ${Delta}$ fliZ cells. To examine whether FliZ controls the transcription of hilC andhilD, we measured the efficiency of transcription from hilCand hilD promoters using lacZ fusions in cells with differentgenetic backgrounds. The results (Fig. 6) show that the ${Delta}$ fliZmutation moderately decreased the transcription of hilD in theclpPX⁺ background, notwithstanding the marked decrease in theamount of HilD protein caused by this mutation (Fig. 5), suggestingthat FliZ may regulate HilD production at a posttranscriptionaland/or posttranslational level.

View larger version (15K):
[in this window]
[in a new window]

FIG. 6. Levels of hilC and hilD transcription in the absence of ClpXP and/or FliZ in cells. The levels of expression of a lacZ fusion to the hilC or hilD promoter in cells with different genetic backgrounds were assayed. Plasmid pTKY559 (hilC promoter-lacZ fusion) or pTKY562 (hilD promoter-lacZ fusion) was introduced into bacterial strains ${chi}$ 3306 (clpPX⁺ fliZ⁺), CS2007 ( ${Delta}$ clpPX fliZ⁺), CS2462 (clpPX⁺ ${Delta}$ fliZ), and CS2464 ( ${Delta}$ clpPX ${Delta}$ fliZ). The resultant stains were used for determination of β-galactosidase activity. The values represent the means and standard deviations for samples tested at least in triplicate.

In contrast, the ${Delta}$ fliZ mutation significantly decreased the transcriptionof hilC in cells with a clpPX⁺ background. It has been demonstratedthat HilC and HilD each significantly activate the expressionof the other when they are overproduced (13, 40). It has alsorecently been reported that loss of HilD decreases hilC transcriptionbut that loss of HilC does not significantly alter hilD expression,so it is proposed that HilD is the apex of regulation of SPI1expression (13, 15). Therefore, the decrease of hilC transcriptioncaused by the ${Delta}$ fliZ mutation in cells with a clpPX⁺ backgroundcould be explained by the decreased amount of HilD.

FliZ modulates HilD at the posttranscriptional level. To examine whether FliZ is involved in the posttranscriptionalcontrol of HilD, we decided to compare the levels of HilD proteinin fliZ⁺ and ${Delta}$ fliZ cells where hilD is expressed under the regulationof the P_A1lacO-1 promoter system. For this purpose, strainsCS2802 (fliZ⁺ ${Delta}$ hilC ${Delta}$ hilD) and CS3329 ( ${Delta}$ fliZ ${Delta}$ hilC ${Delta}$ hilD) were transformedwith the plasmid pTKY651, in which the hilD promoter is replacedby the P_A1lacO-1 promoter. The immunoblotting results in Fig.7 show that HilD was detectable in the absence of IPTG (isopropyl-β-D-thiogalactopyranoside),a condition in which hilD was expressed by read-through fromthe P_A1lacO-1 promoter. The results show that disruption offliZ significantly decreased the cellular level of HilD evenwhen the corresponding gene was expressed by the P_A1lacO-1 promotersystem, suggesting that HilD is controlled posttranscriptionallyby FliZ. To test this inference, we analyzed the in vivo stabilityof HilD. There was no difference in HilD half-life between thefliZ⁺ and ${Delta}$ fliZ cells, suggesting that HilD is not controlledposttranslationally by FliZ (data not shown).

View larger version (19K):
[in this window]
[in a new window]

FIG. 7. Cellular levels of HilD protein expressed by a P_A1lacO-1 promoter-hilD fusion in wild type and FliZ-depleted cells. Plasmid pTKY651 (P_A1lacO-1 promoter-hilD fusion) was introduced into bacterial strains CS2802 (fliZ⁺ ${Delta}$ hilC ${Delta}$ hilD) and CS3329 ( ${Delta}$ fliZ ${Delta}$ hilC ${Delta}$ hilD). Cells of the resultant strains were grown to an optical density at 600 nm of 1.0, collected, lysed, and run on SDS-10% polyacrylamide gels. The separated proteins were transferred to a membrane and then immunostained with anti-HilD antiserum (A). Coomassie brilliant blue-stained gel electrophoretic patterns of the same samples used for immunoblotting are also shown (B). The leftmost lane in panel B contains molecular mass standards.

These findings, coupled with our previous results (61, 62),suggest that the overproduction of FliZ due to accumulationof the FlhD/FlhC master regulator of the flagellar regulon inclpPX-deleted cells results in the accumulation of HilD by posttranscriptionalcontrol and consequently leads to increased SPI1 expression.

DISCUSSION

Top

DISCUSSION

Expression of SPI1 genes is tightly regulated at several stagesin a complex manner by regulators within and outside the island.As summarized in Fig. 8, HilA is the central regulator in theoverall scheme of SPI1 regulation and is known to activate expressionof the prg/org and inv/spa operons. Read-through from inv/spaleads to the activation of sic/spa. InvF, in a complex withthe chaperone protein SicA, also induces expression of the sic/sipoperon, so HilA activates genes encoding all the componentsnecessary for a functional SPI1 type III secretion system (2,11, 12, 15, 37). The products of other genes within SPI1, hilCand hilD, have been shown to activate hilA expression (12, 50,53). HilC and HilD are also capable of activating hilC and hilDexpression independently of each other, and they act in a complexregulatory loop to control hilA expression (13, 50, 53).

View larger version (13K):
[in this window]
[in a new window]

FIG. 8. A model of the coordinated regulation of SPI1 gene expression and flagellar gene expression by ClpXP protease. (A) Regulatory cascade of SPI1 gene expression. HilC and HilD bind directly upstream of the master regulator gene hilA to induce its expression. HilA directly activates the SPI1-encoded prg/org and inv/spa operons by binding just upstream of P_prgH and P_invF-1. The inv/spa transcript reads through the sic/sip operon. Activation of P_invF-1 leads to production of InvF, in a complex with the chaperone protein SicA, and then induces expression of the sic/sip operon. (B) Regulatory cascade of flagellar gene expression. The products, FlhD and FlhC, act together in a FlhD/FlhC complex as the master regulator at the apex of a transcription hierarchy comprising three classes of genes. FliA, which is an alternative sigma factor, ${sigma}$ ²⁸, is required to transcribe the class 3 genes. ClpXP degrades the FlhD/FlhC complex, leading to the down-regulation of flagellar regulon expression. FliZ, which is encoded by the fliAZY operon of the flagellar regulon, activates the expression of hilD at the posttranscriptional level, leading to the stimulation of SPI1 gene expression. ClpXP negatively regulates the expression of SPI1 genes through the repression of flagellar regulon expression. See the text for details and references.

In the present study, we have demonstrated that ClpXP negativelyregulates SPI1 gene expression. We observed that depletion ofClpXP caused a significant increase in the amount of HilC andHilD proteins (Fig. 2), which is associated with increased expressionof all SPI1 genes. Immunoblotting actually revealed that HilAand proteins encoded by genes in the sic/sip operon, SipB andSipC, are greatly increased in the ${Delta}$ clpPX mutant. We hypothesizedthat ClpXP may regulate HilC and HilD through the control ofthe flagellar regulon, since ClpXP has been shown to regulateits expression negatively by degrading the FlhD/FlhC masterregulator, which functions at the apex of the transcriptionhierarchy of the regulon (61, 62). Subsequent experiments showedthat the enhancement of cellular levels of HilC and HilD byclpPX disruption seemed to be caused by an increase in the flagellarprotein FliZ, encoded by the class 2 fliAZY operon. That isto say, disruption of fliZ abolished the increase of HilC andHilD caused by deleting clpPX (Fig. 5). Transcriptional analysisof hilC and hilD in the fliZ⁺ and ${Delta}$ fliZ backgrounds suggestedthat FliZ controls HilC at the transcriptional level and mayregulate HilD production at the posttranscriptional level (Fig.6). Immunoblotting demonstrated that the ${Delta}$ fliZ mutation decreasedthe cellular level of HilD even when hilD was transcribed fromthe P_A1lacO-1 promoter (Fig. 7), suggesting that hilD is controlledposttranscriptionally. Furthermore, it seems unlikely that FliZcontrols HilD at the posttranslational level (our unpublisheddata). Therefore, it can be concluded that FliZ controls HilDat the posttranscriptional level. The function of FliZ is unknown,and its predicted product shows no significant homology to anyknown protein or structural motif.

When the levels of transcription from the hilC and hilD promotersare compared between clpPX⁺ cells and ${Delta}$ clpPX cells in the fliZ⁺background, it is evident that clpPX disruption increases theexpression of both hilC and hilD (Fig. 4A and Fig. 6). It hasbeen demonstrated that HilD is at the top of the hierarchy ofthe SPI1 regulatory loop and has a predominant role, thoughapparently it is not sufficient on its own to activate SPI1(13, 15). Production of HilD leads to transcriptional activationof HilC, and each activates the expression of both (13, 40).Therefore, the increased transcription of hilC and hilD in the ${Delta}$ clpPX cells could be explained by amplification of the regulatoryloop in which FliZ triggered an increase in the amount of HilDvia posttranscriptional control. On the other hand, the stimulationof hilA expression by clpPX disruption was abrogated by introducingthe ${Delta}$ hilC mutation (Fig. 3). This could be because HilD did notaccumulate in the ${Delta}$ hilC ${Delta}$ clpPX double mutant cells (Fig. 4). Asstated above, it is known that HilC and HilD can activate theexpression of hilC and hilD independently of each other as wellas inducing their own promoters (40, 47). Therefore, it is likelythat the moderate effect on HilD production by FliZ needs tobe amplified in the regulatory loop to cause significant stimulationof hilA expression in the ${Delta}$ clpPX mutant cells.

In addition to HilC and HilD, RtsA, which belongs to the AraC/XylSfamily of transcriptional regulators, has been shown to activateexpression of SPI1 genes by binding upstream of hilA to induceits expression (13, 14). HilC, HilD, and RtsA are all also capableof activating expression of hilC, hilD, and rtsA independentlyof each other, constituting a complex feed-forward regulatoryloop that controls hilA expression (13, 14, 15). Therefore,it can be speculated that RtsA acts as an amplifier, leadingto the overexpression of all SPI1 genes in the ${Delta}$ clpPX mutantcells.

SPI1 is a complex regulatory system, and many different signalshave been shown to feed into the network. The current reportsuggests that all the global regulators seem to control hilAexpression in a HilD-dependent manner (15). One such regulatoris the posttranscriptional regulatory protein CsrA, which bindsto the messages of its targets and alters mRNA stability (51).In the ${Delta}$ csrA mutant, the expression of SPI1 genes is greatlyreduced (1). SirA in the BarA/SirA two-component regulatorysystem acts by inducing the expression of two small RNAs, theCsrB and CsrC RNAs, which are antagonistic to CsrA. Overproductionof SirA induces the expression of a hilA-lacZ transcriptionalfusion only when HilD is present (13). Thus, SirA inductionof csrBC prevents CsrA action, indirectly activating hilD expressionposttranscriptionally (17). The RNA chaperone Hfq has recentlybeen recognized as a major posttranscriptional regulator ofbacterial gene expression that participates in numerous regulatorypathways (63). A recent report demonstrates that the ${Delta}$ hfq mutationdrastically reduces Salmonella invasiveness (54). It has alsodemonstrated that ${Delta}$ hfq cells have sevenfold-reduced levels ofhilC, hilD, and rtsA mRNAs, suggesting that Hfq affects signaltransmission further upstream in the SPI1-activating cascade.

HilE is a major negative regulator of SPI1 expression. Deletionof hilE increases expression of hilA only when HilD is present.Although the mechanism of hilE action is not well understood,the results from bacterial two-hybrid studies suggest that HilEbinds directly to HilD, preventing its action (3). hilE is regulatedby several systems that feed into the SPI1 regulation circuit.The systems that negatively regulate SPI1 expression, e.g.,the two-component PhoP/PhoQ and PhoR/PhoB regulatory systemsand FimZY for type 1 fimbrial expression, seem to function primarilythrough HilE (3, 15). Lon protease is a powerful negative regulatorof SPI1 expression (58, 60). Depletion of Lon increases hilAexpression 40-fold and causes a 10-fold increase in invasiveness.Lon has been shown to regulate HilD posttranslationally, specificallyby degradation. Taking these findings together, we suggest thatHilD production is controlled largely at both the posttranscriptionaland posttranslational levels and that this is the key to SPI1regulation. To understand SPI1 regulation completely, the molecularmechanism by which hilD is regulated at the posttranscriptionallevel must be studied in detail.

ClpXP is also known to be a member of the family of stress proteins,which are induced in response to hostile environments (for areview, see reference 45). It has been demonstrated that stressproteins are selectively induced in S. enterica serovar Typhimuriumgrowing within macrophages, where the bacteria are exposed toa variety of bactericidal mechanisms (8). We previously reportedthat hilA expression was repressed in macrophages after phagocytosisbut that this gene was continuously expressed in a S. entericaserovar Typhimurium strain with a defect in Lon, another memberof the stress protein family, growing within macrophages (57).Furthermore, we demonstrated that this derepression of SPI1genes by a Lon-depleted mutant led to rapid and massive macrophageapoptosis by a mechanism involving caspases 1 and 3 (57). Oncea systemic Salmonella infection has been established, excessapoptosis of macrophages, upon which the organism is reliant,would be detrimental to the bacteria. Thus, SPI1 gene expressionmust be controlled in bacteria growing within macrophages tosuppress apoptosis sufficiently to allow time for the bacteriumto replicate, escape, and invade new macrophages, leading tosystemic infection. To ensure this, Salmonella can take advantageof a unique condition in the macrophage to repress SPI1. TheClpXP and Lon proteases may contribute to the down-regulationof SPI1 expression within macrophages. A recent report has demonstratedthat expression of both fliC, encoding a flagellar filament,and fliA, encoding ${sigma}$ ²⁸, is repressed during intracellular growthand that this repression depends on functional ClpXP (10). Theincrease in ClpXP in S. enterica serovar Typhimurium by itsresponse to the hostile environment within macrophages afterphagocytosis would ensure the repression of both SPI1 genesand the flagellar regulon.

ACKNOWLEDGMENTS

We thank T. Tomoyasu, A. Tokumitsu, and A. Suzuki for technicalassistance.

This work was supported by grants-in-aid for scientific research(17390125) and research on priority areas (19041015) from theMinistry for Education, Culture, Sports, Sciences and Technologyof the Japanese Government.

FOOTNOTES

* Corresponding author. Mailing address: Department of Microbiology and Molecular Genetics, Graduate School of Pharmaceutical Sciences, Chiba University, 1-33 Yayoi-cho, Inage-ku, Chiba 263-8522, Japan. Phone: (81) 43-290-2928. Fax: (81) 43-290-2929. E-mail: tomoko-y{at}p.chiba-u.ac.jp

Published ahead of print on 1 February 2008.

REFERENCES

Top