The Clostridium difficile Dlt Pathway Is Controlled by the Extracytoplasmic Function Sigma Factor σ^V in Response to Lysozyme

Abstract

Clostridium difficile (also known as Peptoclostridium difficile) is a major nosocomial pathogen and a leading cause of antibiotic-associated diarrhea throughout the world. Colonization of the intestinal tract is necessary for C. difficile to cause disease. Host-produced antimicrobial proteins (AMPs), such as lysozyme, are present in the intestinal tract and can deter colonization by many bacterial pathogens, and yet C. difficile is able to survive in the colon in the presence of these AMPs. Our prior studies established that the Dlt pathway, which increases the surface charge of the bacterium by addition of d-alanine to teichoic acids, is important for C. difficile resistance to a variety of AMPs. We sought to determine what genetic mechanisms regulate expression of the Dlt pathway. In this study, we show that a dlt null mutant is severely attenuated for growth in lysozyme and that expression of the dltDABC operon is induced in response to lysozyme. Moreover, we found that a mutant lacking the extracytoplasmic function (ECF) sigma factor σ^V does not induce dlt expression in response to lysozyme, indicating that σ^V is required for regulation of lysozyme-dependent d-alanylation of the cell wall. Using reporter gene fusions and 5′ RACE (rapid amplification of cDNA ends) analysis, we identified promoter elements necessary for lysozyme-dependent and lysozyme-independent dlt expression. In addition, we observed that both a sigV mutant and a dlt mutant are more virulent in a hamster model of infection. These findings demonstrate that cell wall d-alanylation in C. difficile is induced by lysozyme in a σ^V-dependent manner and that this pathway impacts virulence in vivo.

INTRODUCTION

Clostridium difficile (Peptoclostridium difficile) causes nearly half a million infections in the United States each year, representing a significant public health threat (1). In order to cause infection, C. difficile must colonize the colon. As an important interface between the host and microbiota, the colon is an environment rich in host innate immune molecules and bacterium-derived antimicrobials made by the indigenous microbiota (2–6). These innate immune molecules and bacterially produced antimicrobials include a variety of cationic antimicrobial peptides (CAMPs), such as lysozyme, LL-37, defensins, and bacteriocins (2, 4, 7–9). Understanding how C. difficile is able to resist killing in this antimicrobial-laden environment could better our understanding of the factors that contribute to the progression of C. difficile infections.

A common mechanism of resistance to CAMPs in many bacteria is the alteration of the cell surface charge (10–12). One mechanism for increasing the surface charge is through the addition of d-alanine (d-Ala) to teichoic acids in the cell wall (10, 12, 13). The addition of d-Ala is mediated by four proteins, DltA, DltB, DltC, and DltD, encoded by the dlt operon (13). The Dlt pathway confers lysozyme resistance to Bacillus subtilis and Enterococcus faecalis (14, 15). Previously, we demonstrated that the d-alanylation of the cell wall via the Dlt pathway is important for resistance of C. difficile to several CAMPs and other antimicrobials, including nisin, gallidermin, polymyxin B, and vancomycin (12).

How the Dlt pathway is regulated in C. difficile is unknown. Expression of dlt increases in C. difficile in the presence of CAMPs (12), but the mechanisms that control this expression remain unidentified. Although a putative DeoR-family regulator (CD2850) is cotranscribed as part of the C. difficile dlt operon, it does not appear to be necessary for dlt expression in vitro (12). The availability of sugars may play a role in regulating dlt expression in C. difficile, as evidenced by a CcpA binding site located within dltD and differential expression of the operon in the presence of glucose (16). In B. subtilis, the dlt operon is regulated by the alternative sigma factor σ^D, the sporulation regulatory protein Spo0A, and the extracytoplasmic function (ECF) sigma factors σ^X and σ^V (15, 17–19). ECF sigma factors are a class of alternative sigma factors broadly involved in functions at the cell surface (20). ECF sigma factors are typically regulated by anti-sigma factors that are located in the cell membrane, which makes ECF sigma factors uniquely suited to regulate genes, such as dlt, that are needed to respond to changes in the cell surface (20). C. difficile encodes orthologs of Spo0A, σ^V (also known as csfV or sigV) and σ^D. Moreover, σ^V is necessary for lysozyme resistance in C. difficile (21). In fact, the C. difficile σ^V anti-sigma factor RsiV binds lysozyme and may serve as a direct lysozyme receptor, as it does in B. subtilis (22). An ortholog of σ^X has not been identified in any sequenced C. difficile isolate, but C. difficile strains encode an additional ECF sigma factor, σ^T (csfT or sigT). Based on the presence of alternative sigma factors in C. difficile that are comparable to those that regulate dlt in B. subtilis, we hypothesized that σ^V, σ^T, or σ^D may regulate the dlt operon of C. difficile in response to CAMPs.

To test this hypothesis, we characterized growth, d-alanylation of the cell wall, and gene expression profiles of dlt, sigV, sigT, and sigD null mutants in the presence of the antimicrobials lysozyme and polymyxin B. In addition, we characterized expression from the dlt promoter to determine regions that are responsible for antimicrobial-dependent expression. Our results demonstrate that σ^V is an important regulator of dlt expression and that σ^V is necessary for controlling d-alanylation of the C. difficile cell wall in response to lysozyme.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1. Escherichia coli strains were grown aerobically in Luria broth (Teknova) at 37°C (23). Cultures were supplemented with 20 μg chloramphenicol ml⁻¹ (Sigma-Aldrich) or 100 μg ampicillin ml⁻¹ (Cayman Chemical Company) as needed. C. difficile strains were grown in brain heart infusion medium supplemented with 2% yeast extract (BHIS; Becton, Dickinson, and Company) or on BHIS agar plates (24) at 37°C in an anaerobic chamber (Coy Laboratory Products) as previously described (25–27). BHIS medium was supplemented with 0.6 to 1.0 mg lysozyme ml⁻¹ (Fisher Scientific), 150 to 200 μg polymyxin B ml⁻¹ (Sigma-Aldrich), 2 μg thiamphenicol ml⁻¹ (Sigma-Aldrich), or 0.5 μg kanamycin ml⁻¹ or 0.5 μg nisin ml⁻¹ (MP Biomedicals) as needed.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Relevant genotype or features	Source, construction, or reference
Strains
E. coli
HB101	F− mcrB mrr hsdS20(rB− mB−) recA13 leuB6 ara-14 proA2 lacY1 galK2 xyl-5 mtl-1 rpsL20	B. Dupuy
MC101	HB101 pRK24	B. Dupuy
MC277	HB101 pRK24 pMC211	33
MC314	HB101 pRK24 pMC235	This study
MC355	HB101 pRK24 pMC286	This study
MC373	HB101 pRK24 pMC316	This study
MC445	HB101 pRK24 pMC358	31
MC463	HB101 pRK24 pMC364	This study
MC464	HB101 pRK24 pMC362	This study
MC466	HB101 pRK24 pMC373	This study
MC468	HB101 pRK24 pMC375	31
MC469	HB101 pRK24 pMC376	This study
MC535	HB101 pRK24 pMC390	This study
MC580	HB101 pRK24 pMC455	This study
MC581	HB101 pRK24 pMC456	This study
MC616	HB101 pRK24 pMC467	This study
MC617	HB101 pRK24 pMC468	This study
MC628	HB101 pRK24 pMC470	This study
MC629	HB101 pRK24 pMC471	This study
MC630	HB101 pRK24 pMC472	This study
MC665	HB101 pRK24 pMC482	This study
MC667	HB101 pRK24 pMC483	This study
MC692	HB101 pRK24 pMC491	This study
MC693	HB101 pRK24 pMC492	This study
MC699	HB101 pRK24 pMC493	This study
MC700	HB101 pRK24 pMC495	This study
MC706	HB101 pRK24 pMC500	This study
MC707	HB101 pRK24 pMC501	This study
C. difficile
630	Clinical isolate	69
630Δerm	Erms derivative of strain 630	N. Minton (70)
JIR8094	Erms derivative of strain 630	C. Ellermeier (21, 71)
TCD20	JIR8094 sigV::ermB	C. Ellermeier (21)
R20291	Clinical isolate	72
MC282	630Δerm pMC211	33
MC319	630Δerm dltD::ermB	This study
MC361	630Δerm sigV::ermB	This study
MC383	630Δerm sigT::ermB	This study
MC448	630Δerm pMC358	31
MC450	MC361 pMC360	This study
MC494	630Δerm pMC364	This study
MC495	630Δerm pMC362	This study
MC497	630Δerm pMC373	This study
MC499	630Δerm pMC375	31
MC500	630Δerm pMC376	This study
MC510	MC361 pMC211	This study
MC512	MC361 pMC358	This study
MC513	MC361 pMC364	This study
MC514	MC361 pMC362	This study
MC515	MC361 pMC373	This study
MC519	MC361 pMC375	This study
MC520	MC361 pMC376	This study
MC551	630Δerm pMC390	This study
MC552	MC361 pMC390	This study
RT1075	630Δerm sigD::ermB	R. Tamayo (45)
MC582	630Δerm pMC455	This study
MC583	630Δerm pMC456	This study
MC584	MC361 pMC455	This study
MC585	MC361 pMC456	This study
MC619	630Δerm pMC467	This study
MC620	630Δerm pMC468	This study
MC632	630Δerm pMC470	This study
MC633	630Δerm pMC471	This study
MC634	630Δerm pMC472	This study
MC635	MC361 pMC470	This study
MC636	MC361 pMC471	This study
MC637	MC361 pMC472	This study
MC668	630Δerm pMC482	This study
MC669	MC361 pMC482	This study
MC682	630Δerm pMC483	This study
MC683	MC361 pMC483	This study
MC695	630Δerm pMC491	This study
MC696	630Δerm pMC492	This study
MC697	MC361 pMC491	This study
MC698	MC361 pMC492	This study
MC701	MC361 pMC493	This study
MC702	MC361 pMC495	This study
MC703	630Δerm pMC493	This study
MC704	630Δerm pMC495	This study
MC714	630Δerm pMC500	This study
MC710	630Δerm pMC501	This study
MC711	MC361 pMC500	This study
MC712	MC361 pMC501	This study
MC744	630Δerm pMC523	This study
MC745	MC361 pMC523	This study
MC746	630Δerm pMC524	This study
MC747	MC361 pMC524	This study
Plasmids
pRK24	Tra+ Mob+; bla tet	73
pCR2.1	bla kan	Invitrogen
pUC19	Cloning vector; bla	74
pCE240	C. difficile TargeTron construct based on pJIR750ai (group II intron, ermB::RAM ltrA); catP	C. Ellermeier (30)
pSMB47	Tn916 integrational vector; Cmr Ermr	75
pMC123	E. coli-C. difficile shuttle vector; bla catP	36
pMC111	pCE240 with dltD-targeted intron	12
pMC211	pMC123 PcprA	33
pMC235	pMC123 with dltD-targeted intron (at nt 367), ermB::RAM ltrA catP	This study
pMC276	pCE240 with sigV-targeted intron	This study
pMC286	pMC123 with sigV-targeted intron (at nt 380), ermB::RAM ltrA catP	This study
pMC312	pCR2.1 with sigT-targeted intron	This study
pMC314	pCE240 with sigT-targeted intron	This study
pMC316	pMC123 with sigT-targeted intron (at nt 537), ermB::RAM ltrA catP	This study
pMC358	pMC123 phoZ	31
pMC360	pMC123 PcprA::sigV	This study
pMC362	pMC123 PdltD200::phoZ	This study
pMC364	pMC123 PdltD100::phoZ	This study
pMC373	pMC123 PdltD300 (630Δerm)::phoZ	This study
pMC375	pMC123 PdltD600 (630Δerm)::phoZ	31
pMC376	pMC123 PdltD600 (R20291)::phoZ	This study
pMC390	pMC123 PdltD112::phoZ	This study
pMC455	pMC123 PdltD119::phoZ	This study
pMC456	pMC123 PdltD170::phoZ	This study
pMC467	pMC123 PdltD25::phoZ	This study
pMC468	pMC123 PdltD50::phoZ	This study
pMC470	pMC123 PdltD140::phoZ	This study
pMC471	pMC123 PdltD150::phoZ	This study
pMC472	pMC123 PdltD160::phoZ	This study
pMC482	pMC123 PdltD130::phoZ	This study
pMC483	pMC123 PdltD75::phoZ	This study
pMC491	pMC123 PdltDT43C::phoZ	This study
pMC492	pMC123 PdltDT51C::phoZ	This study
pMC493	pMC123 PdltD130–75::phoZ	This study
pMC495	pMC123 PdltDG95A::phoZ	This study
pMC500	pMC123 PdltDC93A::phoZ	This study
pMC501	pMC123 PdltDG92A::phoZ	This study
pMC523	pMC123 PdltDA38C::phoZ	This study
pMC534	pMC123 PdltDT48C::phoZ	This study

Strain and plasmid construction.

The oligonucleotides used in this study are listed in Table 2. Primers were designed based on C. difficile strain 630 (GenBank accession number NC_009089.1), unless otherwise specified. Genomic DNA from strain 630Δerm served as the template for PCR amplifications, except where the use of strain R20291 (GenBank accession number NC_013316.1) is noted. PCR, cloning, and plasmid DNA isolation were performed according to standard protocols (25). To create null mutations in C. difficile strain 630Δerm, the group II intron from pCE240 was retargeted using the primers listed in Table 2, as previously described (28–30). To select for TargeTron insertional disruptions, transconjugants were exposed to 5 μg erythromycin ml⁻¹ (Sigma-Aldrich) and 50 μg kanamycin ml⁻¹ (Sigma-Aldrich) to select against E. coli.

TABLE 2.

Oligonucleotides

Primer	Sequencea (5′→3′)	Purpose, source, or referenceb
oMC38	5′-AAAGACGGAGTCACAAGTCACC-3′	dltD qPCR (CD2154) (12)
oMC39	5′-CTGCTTTATACTCGTCACTTCCC-3′	dltD qPCR (CD2154) (12)
oMC44	5′-CTAGCTGCTCCTATGTCTCACATC-3′	rpoC qPCR (CD0067) (12)
oMC45	5′-CCAGTCTCTCCTGGATCAACTA-3′	rpoC qPCR (CD0067) (12)
oMC74	5′-AAAAGCTTTTGCAACCCACGTCGATCGTGAAAAAGTTGTCTTGGTGCGCCCAGATAGGGTG-3′	dltD intron retargeting (12)
oMC75	5′-CAGATTGTACAAATGTGGTGATAACAGATAAGTCGTCTTGTTTAACTTACCTTTCTTTGT-3′	dltD intron retargeting (12)
oMC76	5′-CGCAAGTTTCTAATTTCGGTTACTTTTCGATAGAGGAAAGTGTCT-3′	dltD intron retargeting (12)
oMC193	5′-TGTATAAGGCACTATACTCAGTGG-3′	sigV qPCR (CD1558)
oMC194	5′-ACTCTCCAGTCTCATCTATAAGGTC-3′	sigV qPCR (CD1558)
oMC447	5′-GGCGTAGTATTTTTATTTGGGTTAG-3′	dltD::TargeTron screening
oMC547	5′-TGGATAGGTGGAGAAGTCAGT-3′	tcdA qPCR (CD0663) (33)
oMC548	5′-GCTGTAATGCTTCAGTGGTAGA-3′	tcdA qPCR (CD0663) (33)
oMC703	5′-AAAAGCTTTTGCAACCCACGTCGATCGTGAAAGAGCTTTGGAAGTGCGCCCAGATAGGGTG-3′	sigV (CD1558) intron retargeting
oMC704	5′-CAGATTGTACAAATGTGGTGATAACAGATAAGTCTTGGAAGATAACTTACCTTTCTTTGT-3′	sigV (CD1558) intron retargeting
oMC705	5′-CGCAAGTTTCTAATTTCGGTTGCTCTTCGATAGAGGAAAGTGTCT-3′	sigV (CD1558) intron retargeting
oMC731	5′-GCTACTTCTTCAATCTTTAAATCTTC-3′	sigV::TargeTron screening
oMC800	5′-AAAAGCTTTTGCAACCCACGTCGATCGTGAATCTGTTCTGATTGTGCGCCCAGATAGGGTG-3′	sigT (CD0677) intron retargeting
oMC801	5′-CAGATTGTACAAATGTGGTGATAACAGATAAGTCCTGATTCATAACTTACCTTTCTTTGT-3′	sigT (CD0677) intron retargeting
oMC802	5′-CGCAAGTTTCTAATTTCGGTTACAGATCGATAGAGGAAAGTGTCT-3′	sigT (CD0677) intron retargeting
oMC815	5′-TGGATTCTCTTAAGGAAGAACAATACTTTA-3′	sigT qPCR (CD0677)
oMC816	5′-CCTTAACTTCATCTACTGAATAACCTTCA-3′	sigT qPCR (CD0677)
oMC817	5′-GCGCTACGATTTGCATAGAAGG-3′	sigT::TargeTron screening
oMC818	5′-GCTCATATGATTACCTCCGTGTTTTC-3′	sigT::TargeTron screening
oMC823	5′-GCCGGATCCATTTTCTCTCCTCTAAAAATATTCAAA-3′	Pdlt cloning (31)
oMC826	5′-GCGGAATTCTGATAGTATATAGTTTATATTAGAAAATATAAG-3′	Pdlt300 cloning (630Δerm specific)
oMC827	5′-GCGGAATTCGTTAAAATGTCAAATTATAAGTATGAAAAAG-3′	Pdlt200 cloning
oMC828	5′-GCGGAATTCGTTTTGACGATTTTATTACAATTTTG-3′	Pdlt100 cloning
oMC850	5′-GCGGAATTCTTCTTATATACCATCTGAAATACAGG-3′	Pdlt600 cloning (630Δerm specific) (31)
oMC851	5′-CGCGGATCCGGAGGGAGATTTTACAGGAATG-3′	sigV + RBS cloning
oMC852	5′-GCCTGCAGGTCATTCTTTTTATCCCTACTCTTC-3′	sigV cloning
oMC853	5′-GCGGAATTCTTCTTATATACCATCTGAAATACAAG-3′	Pdlt600 cloning (R20291 specific)
oMC901	5′-CTGAAGCGAAGGCAACTGAA-3′	phoZ qPCR (31)
oMC902	5′-GCTTGCTGTCCGACCAAATA-3′	phoZ qPCR (31)
oMC977	5′-GCGGAATTCGTATCAAAAAAAGTTTTG-3′	Pdlt112 cloning
oMC1023	5′-TTGTTGAATTACTAAGTTCTGATGACCC-3′	Pdlt 5′ RACE (SP1)
oMC1024	5′-TCTCCCTCAAAAGTTCATCAGTTTTAG-3′	Pdlt 5′ RACE (SP2)
oMC1028	5′-GCGGAATTCTGTAACAGTATCAAAAAAAG-3′	Pdlt119 cloning
oMC1029	5′-GCGGAATTCGTGCTAAAAAGAAATTTATTTTTG-3′	Pdlt170 cloning
oMC1067	5′-AATTCTGAATATTTTTAGAGGAGAGAAAATG-3′	Pdlt25 cloning
oMC1068	5′-GATCCATTTTCTCTCCTCTAAAAATATTCAG-3′	Pdlt25 cloning
oMC1069	5′AATTCAATATGATTAATAATAACATAAATTTGAATATTTTTAGAGGAGAGAAAATG-3′	Pdlt50 cloning
oMC1070	5′GATCCATTTTCTCTCCTCTAAAAATATTCAAATTTATGTTATTATTAATCATATTG-3′	Pdlt50 cloning
oMC1071	5′-GCGGAATTCTTTTCTTTTTTTTTACAA-3′	Pdlt140 cloning
oMC1072	5′-GCGGAATTCTTTGGCGTTTTTTTC-3′	Pdlt150 cloning
oMC1073	5′-GCGGAATTCGAAATTTATTTTTGG-3′	Pdlt160 cloning
oMC1079	5′-GCGGAATTCGTAGTTGAATATAC-3′	Pdlt75 cloning
oMC1080	5′-GCGGAATTCTTTTACAATTTTGTAAC-3′	Pdlt130 cloning
oMC1107	5′-GCGGAATTCATTTTCTCTCCTCTCAAAATTGTAATAAAATCGTC-3′	Pdlt130–75 cloning
oMC1108	5′-CAAAAAAAGTTTTAACGATTTTATTAC-3′	SDM of Pdlt (G-95A)
oMC1109	5′-GTAATAAAATCGTTAAAACTTTTTTTG-3′	SDM of Pdlt (G-95A)
oMC1110	5′-CAAAAAAAGTTTTGAAGATTTTATTAC-3′	SDM of Pdlt (C-93A)
oMC1112	5′-CAAAAAAAGTTTTGACAATTTTATTAC-3′	SDM of Pdlt (G-92A)
oMC1114	5′-ACATATCAAAACCAATATGATTAATAATAACA-3′	SDM of Pdlt (T-51C)
oMC1115	5′-TGTTATTATTAATCATATTGGTTTTGATATGT-3′	SDM of Pdlt (T-51C)
oMC1116	5′-CAAAACTAATATGACTAATAATAACATAAATTTG-3′	SDM of Pdlt (T-43C)
oMC1117	5′-CAAATTTATGTTATTATTAGTCATATTAGTTTTG-3′	SDM of Pdlt (T-43C)
oMC1124	5′-GTAATAAAATCTTCAAAACTTTTTTTG-3′	SDM of Pdlt (C-93A)
oMC1125	5′-GTAATAAAATTGTCAAAACTTTTTTTG-3′	SDM of Pdlt (G-92A)
oMC1147	5′-CATATCAAAACTAACATGATTAATAATAAC-3′	SDM of Pdlt (T-48C)
oMC1148	5′-GTTATTATTAATCATGTTAGTTTTGATATG-3′	SDM of Pdlt (T-48C)
oMC1149	5′-CTAATATGATTAATVATAAVATAAATTTG-3′	SDM of Pdlt (A-38C)
oMC1150	5′-CAAATTTATGTTATGATTAATCATATTAG-3′	SDM of Pdlt (A-38C)

^aUnderlined sequences denote restriction sites or intron retarget sites.

^bAbbreviations: qPCR, quantitative PCR; RBS, ribosome binding site; SDM, site-directed mutagenesis.

To generate alkaline phosphatase (AP) reporter gene promoter fusions, regions of various lengths upstream of dltD were PCR amplified from either C. difficile strain 630Δerm or strain R20291 genomic DNA, as noted for the primers listed in Table 2. For site-directed mutagenesis of the promoter region, mutations were generated via splicing by overlap extension (SOEing) PCR using the primers listed in Table 2. These products were independently ligated into the EcoRI/BamHI sites of pMC358 (31) to generate the plasmids listed in Table 2. Plasmids were confirmed by sequencing (Eurofins MWG Operon) and introduced into E. coli strain MC101 by transformation. The resulting E. coli strains were then conjugated to C. difficile strain 630Δerm or MC361, selecting for thiamphenicol resistance, as previously described (12, 32).

To complement the sigV mutant, the sigV coding sequence was cloned into pMC211 to place expression of sigV under the control of the nisin-inducible cpr promoter, as previously described (33, 34). The resulting plasmid (pMC360) was conjugated with MC361 as described above. Strains 630Δerm and MC361 containing the empty pMC211 vector served as controls.

Phase-contrast microscopy.

C. difficile strains were grown in BHIS alone or supplemented with 1 mg lysozyme ml⁻¹ as described above. One milliliter of actively growing culture was removed from the anaerobic chamber, centrifuged at full speed for 1 min, and resuspended in 5 μl supernatant. Two microliters of resuspended pellet was placed on top of a thin layer of 0.7% agarose on a microscope slide. For comparison of the 630Δerm and JIR8094 strains, 250 μl of actively growing C. difficile cultures in BHIS at an optical density at 600 nm (OD₆₀₀) of 0.50 was plated on 70:30 agar (35). After 24 h, growth was scraped from these plates and resuspended in BHIS, and 2 μl was placed on top of a thin layer of 0.7% agarose on a microscope slide. Phase-contrast microscopy was performed using an X100 Ph3 oil-immersion objective on a Nikon Eclipse Ci-L microscope.

qRT-PCR.

Actively growing C. difficile cultures were diluted to an OD₆₀₀ of approximately 0.05 in BHIS alone or with 1.0 mg lysozyme ml⁻¹ or 200 μg polymyxin B ml⁻¹. Cultures were grown to an OD₆₀₀ of 0.5, harvested into cold 1:1 ethanol-acetone, and stored at −80°C. Alternatively, for in vitro toxin expression experiments, 250 μl of actively growing C. difficile cultures in BHIS at an OD₆₀₀ of 0.50 was plated on 70:30 agar. After 12 h, growth from these plates was scraped into cold 1.5:1.5:3 ethanol-acetone-water and stored at −80°C. In addition, cecal contents from animals infected with C. difficile were collected postmortem into cold 1:1 ethanol-acetone and stored at −80°C. RNA was purified and treated with DNase I before cDNA synthesis as previously described (36–38). Fifty micrograms of RNA was used as the template for cDNA generation from in vitro samples, and 200 μg of RNA was used as the template for cDNA generation from cecal samples. The IDT PrimerQuest tool was used to design quantitative reverse transcription-PCR (qRT-PCR) primers (Integrated DNA Technologies, Coralville, IA). Each qRT-PCR was performed in technical triplicate for at least three biological replicates. rpoC served as an internal control transcript to normalize expression for relative quantification. The means and standard errors of the means for the transcriptional ratios of variable and control sets are presented and compared using either a one- or two-way analysis of variance with Dunnett's or Sidak's multiple-comparison test, as indicated.

AP activity assay.

C. difficile strains containing the promoter-reporter gene fusions listed in Table 1 were grown to mid-logarithmic phase (OD₆₀₀, ∼0.5), 1-ml samples were harvested in duplicate, and pelleted cells were stored at −20°C. The samples were analyzed for alkaline phosphatase (AP) activity as previously described (31). Briefly, samples were washed in 0.5 ml wash buffer (10 mM Tris-HCl, pH 8.0, 10 mM MgSO₄) and resuspended in 800 μl of assay buffer (1 M Tris-HCl, pH 8.0, 0.1 M ZnCl₂). Fifty microliters of 0.1% SDS and 50 μl chloroform were added to the samples, which were then vortexed for 15 s. Samples were incubated for 5 min at 37°C and then for 5 min on ice. After rewarming to room temperature, 100 μl of 0.4% pNP (p-nitrophenyl phosphate in 1 M Tris-HCl, pH 8.0) was added to samples in 10-s intervals. Samples were mixed by inversion and incubated at 37°C until the development of yellow color. To stop the reaction, 100 μl of 1 M KH₂PO₄ was added in 10-s intervals and the samples were placed on ice. Developed samples were then centrifuged at 4°C for 5 min, and the supernatant OD₅₅₀ and OD₄₂₀ values were recorded. AP activity was calculated as follows: {[OD₄₂₀ – (1.75 × OD₅₅₀)] × 1,000}/(OD₆₀₀ × volume × time). OD₆₀₀ refers to the absorbance of the culture at 600 nm at the time of sample collection. Volume is the volume of sample analyzed (1 ml). Time is the total reaction time from addition of pNP to the addition of stop buffer. Results are represented as the means of calculated AP activity and standard errors of the means from at least three biological replicates, each performed as technical duplicates. Data were excluded from analysis if technical duplicates varied from each other by greater than 25%. Data were analyzed with a two-way analysis of variance with Dunnett's multiple-comparison test. Data from site-directed mutagenesis constructs were analyzed using the two-tailed Student t test with correction for multiple comparisons by the Holm-Sidak method.

Quantification of d-alanine ester content in teichoic acids.

The amount of d-alanine esters incorporated into teichoic acids of cell walls was quantified as previously described, with minor modifications (12, 39, 40). Cultures were grown anaerobically at 37°C in BHIS or in BHIS supplemented with 0.6 mg lysozyme ml⁻¹ or 150 μg polymyxin B ml⁻¹. Fifty milliliters was harvested by centrifugation at an OD₆₀₀ of 0.5, and cell pellets were stored at −20°C. Cell pellets were washed three times with 1 ml 0.1 M morpholineethanesulfonic acid (MES) (Sigma-Aldrich), pH 6.0, before boiling for 15 min in 0.5 ml 0.2% SDS, 0.1 M MES, pH 6.0, to partially purify cell walls. Pelleted cell walls were then washed four times with 1 ml 0.1 M MES, pH 6.0. The washed and pelleted cell walls were dried on a tabletop vacuum centrifuge heated to 55°C. Total cell wall contents were determined by weighing the dried pellets. To release d-alanine residues, the pellets were resuspended in 0.5 ml 0.1 M sodium pyrophosphate (Sigma-Aldrich), pH 8.3, and incubated at 60°C for 3 h. The samples were then centrifuged and the supernatant was transferred to a fresh tube for use in the quantification assay, as described previously (12). Results are the means and standard errors of the means from at least three biological replicates, each performed as technical duplicates. Data were analyzed using a two-way analysis of variance with Dunnett's multiple-comparison test.

5′ RACE.

RNA was purified from cells collected as described above for qRT-PCR. After DNase I treatment, the RNA was used as a template to generate cDNA using the Roche 5′/3′ RACE (rapid amplification of cDNA ends) second-generation kit, according to the manufacturer's protocol. Primer oMC1023 was used for first-strand synthesis, and oMC1024 and the Roche oligo(T) primer were used for the subsequent PCR amplification step. The resulting cDNA products were purified and either sequenced directly (Eurofins MWG Operon) or cloned into pCR2.1 (Invitrogen TOPO TA cloning kit) before sequencing.

Animal studies.

All animal studies were approved in advance by the Emory University Institutional Animal Care and Use Committee (IUCAC). Female Syrian golden hamsters (Mesocricetus auratus; Charles River Laboratories) were housed individually in sterile cages in an animal biosafety level 2 facility within the Emory University Division of Animal Resources. Hamsters were provided sterile water and rodent feed pellets ad libitum. To induce susceptibility to infection with C. difficile, hamsters were orally gavaged once with clindamycin (30 mg/kg of body weight) 7 days prior to inoculation with C. difficile (41, 42). Hamsters were inoculated by oral gavage with approximately 5,000 C. difficile spores, which were prepared as described previously (33). After preparation, spores were diluted in phosphate-buffered saline (PBS) with 1% bovine serum albumin to prevent clumping of spores and stored at room temperature in glass vials to prevent adhesion to plastic. Prior to plating, aliquots of spores were heated for 20 min at 55°C. Spores were enumerated by plating these heated aliquots on BHIS plus 0.1% taurocholate to induce germination. Spore preparations were heated for 20 min at 55°C prior to inoculating animals. Multiple cohorts of hamsters were tested for each strain of C. difficile (630Δerm, MC319, MC361, JIR8094, and TCD20, or a one-to-one mixture of 630Δerm and MC319) for a total of at least 12 hamsters per strain. A hamster treated with clindamycin, but not inoculated with C. difficile, served as a negative control for each cohort. After inoculation, hamsters were weighed at least once per day and fecal samples were collected daily. Hamsters were monitored for disease symptoms and considered moribund if they either lost ≥15% of their highest body weight or developed symptoms of diarrhea, lethargy, and wet tail. To prevent unnecessary suffering, hamsters meeting either of these criteria were euthanized. Cecal contents were collected at the time of morbidity (postmortem). CFU were enumerated from daily fecal samples and from cecal samples by resuspension in 1× PBS, serial dilution, and plating onto TCCFA agar (43, 44). CFU were enumerated after 48 h of incubation on TCCFA. For samples from animals coinfected with 630Δerm and MC319, samples were plated on both TCCFA and TCCFA with 2 μg/ml erythromycin to distinguish between the strains. These CFU counts were then used to calculate the competitive index (CI) for MC319, using the formula CI = number of MC319 CFU per milliliter over number of 630Δerm CFU per milliliter (in cecal contents) divided by number of MC319 spores per milliliter over number of 630Δerm spores per milliliter (in original inoculum). Differences in CFU counts were analyzed using a one-way analysis of variance with Dunnett's multiple-comparison test, and differences in survival were analyzed using log rank regression.

Statistical analysis.

All statistical analysis was performed using GraphPad Prism version 6.00 for Mac OS X (GraphPad Software, La Jolla, CA, USA).

The locus tags for individual genes mentioned in the text are listed in Table 2.

RESULTS

Impact of sigD, sigT, and sigV disruption on CAMP resistance.

In order to test our hypothesis that Dlt-mediated CAMP resistance is regulated by alternative sigma factors in C. difficile, insertion mutants were generated in sigV, sigD, and sigT in strain 630Δerm using group II intron targeting (28, 45). In previous work, we generated a C. difficile mutant with a nonfunctional dltDABC operon (12), but this mutant was derived from the parent strain JIR8094, which is nonmotile and has a virulence defect (46). Unlike JIR8094, 630Δerm retains the virulence profile of the clinical parent strain, 630, and is therefore a more clinically relevant strain (47–49) The dlt mutation was regenerated in the strain 630Δerm background for these studies. The growth phenotype of these mutants was then assessed in the presence of the antimicrobials lysozyme and polymyxin B (Fig. 1). The strain R20291, a clinical isolate of the epidemic 027 ribotype, was also included in order to assess the antimicrobial sensitivities of this clinically relevant strain. All of the mutants had growth comparable to the parent strain in BHIS; however, the dlt and sigV mutants both had attenuated growth in BHIS supplemented with 1 mg/ml lysozyme (Fig. 1A). The lysozyme-deficient growth phenotype was more pronounced in the dlt mutant than in the sigV mutant. The R20291 strain had a slight growth defect in lysozyme compared to the 630Δerm strain. Growth of the dlt mutant was also attenuated in BHIS supplemented with 200 μg/ml polymyxin B (Fig. 1B). These findings validate earlier studies' finding that d-alanylation by the Dlt pathway is important for CAMP resistance (12) and suggest that σ^V is a candidate regulator of dlt in lysozyme. The attenuated growth of the sigV mutant in lysozyme was complemented by expression of sigV from a plasmid, similar to previous studies (see Fig. S1 in the supplemental material) (21). The sigV mutant did not, however, have a growth defect in polymyxin B (Fig. 1B), demonstrating that σ^V is not necessary for dlt regulation in polymyxin B. In contrast to the phenotype observed for sigV, neither a sigT nor a sigD mutant was attenuated for growth in lysozyme (Fig. 1A). The sigT and sigD mutants were slightly attenuated for growth in polymyxin B during log phase but ultimately achieved the same cell density as the parent strain (Fig. 1B). These results indicate that σ^T and σ^D are not critical for regulation of the dlt operon in C. difficile in response to polymyxin B or lysozyme.

FIG 1.

dlt and sigV mutants have attenuated growth in lysozyme, and a dlt mutant has attenuated growth in polymyxin B. Active cultures of strains 630Δerm and R20291 and dltD (MC319), sigV (MC361), sigT (MC383), and sigD (RT1074) mutants were diluted an to OD₆₀₀ of 0.05 in BHIS supplemented with 1 mg/ml lysozyme (A) or 200 μg/ml polymyxin B (B). All strains grew similarly in BHIS alone, as depicted by the solid black line on each graph. Graphs are representative growth curves from three biological replicates.

dlt and sigV expression is induced by CAMPs.

Based on the similar phenotypes of the dlt and sigV mutants when grown in CAMPs, we further explored σ^V as a potential regulator of the Dlt pathway in response to antimicrobials. We hypothesized that if σ^V regulates dlt expression in response to CAMPs, then expression of sigV and the dlt operon would be simultaneously induced upon exposure to these compounds. Using qRT-PCR, we detected significantly higher dltD expression in 630Δerm and R20291 cells grown in 1.0 mg lysozyme ml⁻¹ or 200 μg polymyxin B ml⁻¹, compared to cells grown in BHIS alone (Fig. 2A). The increase in dltD expression during growth in lysozyme or polymyxin B was more pronounced in the R20291 strain (∼15-fold) than in 630Δerm (∼8-fold). The expression of dltD was greater during growth in lysozyme than in polymyxin B for both R20291 and 630Δerm. But, there was no significant change in dltD expression for the sigV mutant during growth in lysozyme. This result strongly suggests that σ^V is necessary for increased dlt expression in response to lysozyme. Similarly to dltD regulation, we found that sigV expression increases in both R20291 and 630Δerm in response to lysozyme (Fig. 2A), with a higher fold change in R20291 (∼80-fold versus 40-fold, respectively).

FIG 2.

dltD and sigV expression is induced in lysozyme. qRT-PCR analysis of dltD and sigV expression in R20291; 630Δerm; and sigV (MC361), sigT (MC383), and dltD (MC319) mutant strains grown in BHIS supplemented with 1 mg/ml lysozyme (A) or 200 μg/ml polymyxin B (B) as described in Materials and Methods. mRNA levels in 630Δerm and the mutant derivatives of this strain (sigV, sigT, and dltD) are normalized to 630Δerm in BHIS alone. mRNA levels in R20291 are normalized to expression levels in R20291 in BHIS alone. ND, not determined. The sigT mutant was not assessed in polymyxin B. The means and standard errors of the means for three biological replicates are shown. Data were analyzed by a two-way analysis of variance with Sidak's multiple-comparison test. * indicates P < 0.05 compared to the untreated parent strain, unless otherwise noted by a bar between the compared strains.

In contrast, the sigV mutant had a change in dltD expression during growth in polymyxin B similar to that of the parent strain, indicating that a mechanism(s) other than σ^V can regulate dlt in response to polymyxin B (Fig. 2B). In polymyxin B, sigV expression was only marginally higher in the 630Δerm background than in untreated 630Δerm, suggesting that the adaptive response to polymyxin B is not σ^V dependent in this strain (Fig. 2B). In contrast, R20291 induced sigV and dlt transcript more than strain 630Δerm in polymyxin B. These data suggest that R20291 and 630Δerm may have different mechanisms for regulating dlt gene expression during growth in polymyxin B and that σ^V is not a significant regulator of the adaptive response to polymyxin B in the 630Δerm strain. As expected, sigV was not expressed in the sigV mutant under any condition tested (Fig. 2).

dlt and sigV mutants have altered morphology in lysozyme.

Because dltD and sigV expression was increased in lysozyme, and growth of these mutants was also affected by lysozyme, we hypothesized that lysozyme has a greater impact on the cell wall of these mutants than on the parent strain. To test this, we used phase-contrast microscopy to assess the cellular morphology of the dlt and sigV mutants during growth in lysozyme (Fig. 3). Although the sigV and dlt mutants have normal morphology in BHIS medium (Fig. 3A to C), both mutants displayed altered phenotypes in lysozyme compared to the parent strain 630Δerm. In both mutant strains, some of the bacteria took on a curved morphology (Fig. 3F and G). In addition, many of the dlt and sigV mutant cells lysed during growth in lysozyme, and some of the dlt mutant cells appeared elongated. Although more lytic cells were observed in the sigV mutant than the dlt mutant, the dlt mutant grew much more slowly in lysozyme than the sigV strain (Fig. 1A), suggesting that the dlt cells were dying more rapidly. Hence, lysozyme is more effective against C. difficile strains that lack σ^V or cannot incorporate d-alanine into the cell wall.

FIG 3.

dlt and sigV mutants have altered cell morphology in lysozyme. Representative phase-contrast micrographs of 630Δerm, sigV mutant (MC361), dltD mutant (MC319), and R20291 were grown in BHIS alone (A to D) or BHIS supplemented with 1 mg/ml lysozyme (E to H) to mid-log phase. Black arrowheads indicate examples of curved morphology, asterisks indicate examples of lysed cells, and the white arrowhead indicates an example of an elongated cell.

Similarly to the sigV and dlt mutants, R20291 adopted altered morphologies and phenotypes in lysozyme, including curved cell shapes, elongated cells, and apparent cell lysis (Fig. 3H). The more dramatic effect of lysozyme on cell morphology in R20291 than in 630Δerm parallels the slight growth defect that we observed in R20291 in lysozyme (Fig. 1A). These findings indicate that strain R20291 is more affected by lysozyme than is 630Δerm.

d-Alanylation of the cell wall increases upon exposure to CAMPs.

The Dlt pathway is responsible for catalyzing the addition of d-alanine (d-Ala) to teichoic acids in the cell wall of C. difficile (12, 13). To determine if the observed increases in dlt expression affect d-alanylation of the cell wall, we examined the d-Ala content of R20291, 630Δerm, and the sigV and dlt mutants, grown with and without polymyxin B or lysozyme. We calculated the amount of d-Ala esters present in purified cell walls of 630Δerm, R20291, the sigV mutant, and the dlt mutant grown in BHIS alone or in BHIS supplemented with 0.6 mg lysozyme ml⁻¹ or 150 μg polymyxin B ml⁻¹ as described in Materials and Methods (Fig. 4). As previously observed, the dlt mutant had undetectable d-Ala content in the cell wall (12). As expected from dlt expression analyses, the relative d-Ala content in 630Δerm and R20291 was higher in cells exposed to lysozyme or polymyxin B than in cells grown in BHIS alone. d-Ala content was higher in 630Δerm than in R20291 in BHIS and with added lysozyme, suggesting that these strains inherently differ in their ability to d-alanylate teichoic acids. The d-Ala content of the sigV mutant in BHIS alone was similar to that of 630Δerm. The sigV mutant did not have a significantly altered amount of d-Ala when exposed to lysozyme or polymyxin B. These data demonstrate that σ^V is not necessary for basal-level d-alanylation of the cell wall that occurs in the absence of CAMPs, but σ^V is required for increased d-alanylation in the presence of lysozyme.

FIG 4.

A sigV mutant does not increase d-alanine cell wall content in lysozyme. R20291, 630Δerm, sigV (MC361), and dltD (MC319) strains were grown in BHIS alone or in BHIS supplemented with either 0.6 mg/ml lysozyme (Lys) or 150 μg/ml polymyxin B (PmB). Results are presented as the means and standard errors of the means from at least three biological replicates, each performed as technical duplicates. Data were analyzed by a two-way analysis of variance with Dunnett's multiple-comparison test. * indicates P < 0.05 compared to the untreated parent strain, except where indicated by a bar between the compared strains.

Identification of dlt promoter elements.

To evaluate the potential promoter elements necessary for σ^V-dependent and -independent transcription of the dlt operon, we created a series of transcriptional fusions of the predicted dltD promoter to a phoZ (alkaline phosphatase) reporter (31). Segments upstream of the dltD translational start site (TSS) were amplified and ligated to the phoZ reporter gene within a plasmid vector (Fig. 5A). The resultant plasmids were conjugated independently into 630Δerm and the sigV mutant. To assess potential promoter functions of this region, the resultant strains were grown with or without lysozyme and assayed for alkaline phosphatase (AP) activity (Table 3).

FIG 5.

The dlt promoter region. (A) Schematic of the dlt operon and the upstream region used in promoter fusion constructs. The dlt operon consists of four genes, dltDABC. CD2850 encodes a putative DeoR-type regulator, is cotranscribed with the dltDABC operon, and is not required for expression or function of the Dlt pathway (12). CD2855 lies 288 bp upstream of dltD and is not part of the operon (12). Promoter fusion constructs were made with segments of the upstream region included. (B) DNA sequence from strain 630Δerm from 300 bp upstream of the predicted dltD translational start site. Sequence differences in strain R20291 are shown above the sequence. Promoter fusions of the indicated sizes marked by bold nucleotides were created. The transcriptional start sites identified by 5′ RACE analysis are underlined and identified by +1. Identified tandem direct repeat sequences are denoted by italics with black dashed underlines. Identified complementary regions are denoted by black brackets below the sequence. Possible spacing for the −10 and −35 sequences of a putative weak σ^A promoter is marked with gray dashed underlines, and possible spacing for the −10 and −35 sequences of a stronger σ^A promoter and overlapping σ^V-dependent promoter is marked with a solid black underline. Base pairs altered by site-directed mutagenesis are indicated below the sequence. * marks the nucleotides that when mutated abolished promoter activity. The white circle denotes the minimum required for promoter activity, the gray circle denotes the region required for lysozyme-dependent activity, and the black circle denotes the region required for the full level of activity.

TABLE 3.

Alkaline phosphatase activity from Pdlt::phoZ fusions

Reporter fusion	Activity for strain in mediuma
	630Δerm		sigV mutant
	BHIS alone	BHIS + Lysb	BHIS alone	BHIS + Lysc
phoZ	2 ± 0	2 ± 0	2 ± 0	2 ± 0
Pdlt75::phoZ	2 ± 0	2 ± 0	2 ± 0	2 ± 0
Pdlt100::phoZ	17 ± 4	23 ± 3	23 ± 7	19 ± 1
Pdlt112::phoZ	9 ± 1	21 ± 2	8 ± 1	8 ± 2
Pdlt119::phoZ	8 ± 1	26 ± 1	7 ± 0	8 ± 1
Pdlt130::phoZ	30 ± 1	124 ± 10	15 ± 3	36 ± 4
Pdlt140::phoZ	29 ± 7	116 ± 20	22 ± 3	30 ± 7
Pdlt150::phoZ	31 ± 5	122 ± 17	26 ± 5	31 ± 5
Pdlt160::phoZ	58 ± 15	174 ± 31	26 ± 2	36 ± 3
Pdlt170::phoZ	79 ± 4	197 ± 8	69 ± 0	63 ± 15
Pdlt200::phoZ	44 ± 8	144 ± 18	49 ± 6	46 ± 1
Pdlt300::phoZ	32 ± 2	135 ± 10	42 ± 7	38 ± 3
Pdlt600::phoZ	59 ± 6	173 ± 9	49 ± 8	32 ± 4
Pdlt600::phoZ (R20291)	54 ± 8	146 ± 23	48 ± 3	42 ± 14

^a630Δerm and sigV mutant (MC361) with dlt promoter::phoZ fusion plasmids were grown in BHIS alone or with 1 mg/ml lysozyme and assayed for AP activity as described in Materials and Methods. Results are the means of calculated AP units ± standard errors of the means for at least three biological replicates. All biological replicates were performed as technical duplicates.

^bData were analyzed by a two-way analysis of variance with Sidak's multiple-comparison test, comparing data to the same strain and fusion in BHIS. Bold text indicates P < 0.05.

^cData were analyzed by a two-way analysis of variance and Sidak's multiple-comparison test, comparing data to the same fusion in strain 630Δerm grown in lysozyme. Bold text indicates P < 0.05.

Initially, reporter fusions containing the 600-bp region upstream of the dltD start codon from strain 630Δerm or R20291 were assessed for activity in both the 630Δerm and sigV mutant strains (Pdlt₆₀₀::phoZ and Pdlt₆₀₀::phoZ [R20291]) (Table 3). As predicted, this region contains the necessary promoter elements to support transcription, as evidenced by AP activity. Despite multiple nucleotide differences between the Pdlt sequences of the R20291 and 630Δerm strains (Fig. 5B), the AP activities generated from the respective promoter fusions (expressed in the 630Δerm background) were comparable, indicating that these sequence changes do not affect promoter activity. Importantly, these fusions demonstrated lysozyme-dependent induction of activity in 630Δerm, but only lower, constitutive-level activity was observed in the sigV mutant. These results demonstrate that σ^V is required for lysozyme-dependent expression from the dlt promoter.

To determine the minimal sequence required for transcription, we examined activity from increasingly larger portions of sequence, beginning at 25 nucleotides (nt) upstream of the dltD TSS. Segments from 25 to 75 nt upstream of the dltD TSS (Pdlt₂₅::phoZ, Pdlt₅₀::phoZ, and Pdlt₇₅::phoZ fusions) did not generate significant AP activity in the 630Δerm or sigV background, indicating that the 75-bp upstream region is not sufficient for transcription. The Pdlt₁₀₀::phoZ fusion had modest AP activity, demonstrating that the 100-bp region upstream of the dltD TSS is sufficient for transcription. The AP activity of the Pdlt₁₀₀::phoZ fusion was not inducible in lysozyme, which suggests that this region does not contain sequence elements necessary for lysozyme-dependent induction of transcription. Additionally, there were no differences between AP activities of Pdlt₁₀₀::phoZ in the parent strain and in the sigV mutant. Thus, the sequence between 75 and 100 bp upstream of the dltD start codon contains the minimal promoter elements for constitutive, low-level transcription of the operon but is not sufficient for σ^V-dependent transcription.

Sequence analyses of the region revealed two direct repeat sequences spanning from nt −73 to −85 and nt −116 to −128, suggesting that these areas could be involved in regulation. We created constructs using additional nucleotides (Pdlt₁₁₂::phoZ, Pdlt₁₁₉::phoZ, and Pdlt₁₃₀::phoZ) to investigate the function of this region. Lysozyme-inducible AP activity was observed with all three constructs in the parent strain, with the highest constitutive and lysozyme-induced activity found for the Pdlt₁₃₀::phoZ fusion (Table 3). Lower promoter activity was observed with Pdlt₁₁₂::phoZ and Pdlt₁₁₉::phoZ in the parent and sigV strains, suggesting that the direct repeat region may be involved in lysozyme-independent (constitutive) expression of dlt. AP activity from Pdlt₁₁₂::phoZ and Pdlt₁₁₉::phoZ in the sigV mutant was not inducible in lysozyme. Therefore, the segment 112 bp upstream of the dltD TSS contains a sequence necessary for lysozyme-dependent and σ^V-dependent induction of transcription (Fig. 5B).

In addition to the direct repeat sequences mentioned above, two segments of complementary sequence were identified at nt −131 to −139 and nt −157 to −165. Additional reporter fusion constructs were generated to examine these larger segments of the Pdlt upstream region for differences in regulation (Pdlt₁₄₀::phoZ, Pdlt₁₅₀::phoZ, Pdlt₁₆₀::phoZ, and Pdlt₁₇₀::phoZ). The constructs containing promoter segments from −130 to −150 nt upstream of the dltD TSS in the 630Δerm strain demonstrated levels of AP activity similar to each other. The sigV mutant had lower AP activity with all of these constructs when grown in BHIS than when grown in medium containing lysozyme. Higher AP activity was observed for 630Δerm strains expressing the Pdlt₁₆₀::phoZ or Pdlt₁₇₀::phoZ fusion than with the shorter promoter segments. In fact, the Pdlt₁₇₀::phoZ reporter fusion had higher AP activity with or without lysozyme than did fusions containing more upstream sequence (Pdlt₂₀₀::phoZ, Pdlt₃₀₀::phoZ, or Pdlt₆₀₀::phoZ). These data suggest that the region 170 to 200 nt upstream of the dltD TSS may contain elements that negatively affect σ^V-independent promoter activity; however, the factors that contribute to this regulation are not known.

To further characterize the dlt operon promoter elements and identify potential sites of RNA polymerase binding, we performed 5′ RACE analysis on mRNA extracted from 630Δerm grown in the presence of 1 mg/ml lysozyme. This analysis revealed transcriptional start sites at 30 bp and 35 bp upstream of the predicted dltD translational start (Fig. 5B). The location of transcriptional start sites 5 nt apart suggests that two unique, but perhaps overlapping, promoters are involved in dlt transcription. However, these transcriptional start sites and the anticipated −10 and −35 sites (Fig. 5B) are positioned in a segment that was insufficient for reporter expression (Pdlt₇₅::phoZ [Table 3]). Together, these results suggest that RNA polymerase initiates transcription from promoters within the 75 nt upstream of the dltD start codon but that additional upstream sequence is needed to facilitate transcription.

Based on the 5′ RACE results, we predicted that σ^A and/or σ^V −10 promoter elements might be located either 49 to 42 bp or 51 to 45 bp upstream of the dltD TSS (Fig. 5B). We therefore performed site-directed mutagenesis on the nucleotides at positions −43, −51, −92, −93, and −95 bp upstream of the dltD TSS. The AP activities from Pdlt₃₀₀G-95A::phoZ, Pdlt₃₀₀C-93A::phoZ, Pdlt₃₀₀G-92A::phoZ, and Pdlt₃₀₀T-51C::phoZ were all comparable to the activity observed from the native-sequence Pdlt₃₀₀::phoZ construct in both the parent strain and the sigV mutant carrying these constructs during growth in BHIS and lysozyme (Table 4). Therefore, we conclude that T-51, G-92, C-93, and G-95 are not essential for constitutive or lysozyme-dependent expression of dlt. However, the Pdlt₃₀₀T-43C::phoZ fusion had negligible AP activity in the parent strain in either BHIS or lysozyme. Further, the sigV mutant containing the Pdlt₃₀₀T-43C::phoZ construct also lacked expression in BHIS and lysozyme. Hence, nucleotide T-43 is critical for both constitutive and σ^V-dependent, lysozyme-induced transcription of dlt. These results strongly suggest that σ^A- and σ^V-dependent promoters overlap at T-43. Alternatively, it is possible that σ^V-dependent expression in lysozyme is indirect. In that case, overlapping σ^A-dependent promoters would be used for both constitutive and lysozyme-induced expression with lysozyme induction being mediated by a regulatory factor controlled by σ^V.

TABLE 4.

Alkaline phosphatase activity from Pdlt::phoZ fusions with site-directed mutagenesis

Reporter fusion	Activity for strain in mediuma,b
	630Δerm		sigV mutant
	BHIS alone	BHIS + Lys	BHIS alone	BHIS + Lys
Pdlt300::phoZ	32 ± 2	135 ± 10	42 ± 7	38 ± 3
Pdlt300A-38C::phoZ	18 ± 1	102 ± 4	19 ± 3	24 ± 3
Pdlt300T-43C::phoZ	2 ± 0	2 ± 0	2 ± 0	3 ± 0
Pdlt300T-48C::phoZ	2 ± 0	3 ± 0	2 ± 0	3 ± 0
Pdlt300T-51C::phoZ	42 ± 14	145 ± 18	33 ± 5	32 ± 4
Pdlt300G-92A::phoZ	31 ± 3	126 ± 4	27 ± 3	43 ± 7
Pdlt300C-93A::phoZ	31 ± 5	123 ± 9	48 ± 7	54 ± 7
Pdlt300G-95A::phoZ	45 ± 13	149 ± 20	50 ± 7	61 ± 9
Pdlt130::phoZ	30 ± 1	124 ± 10	15 ± 3	36 ± 4
Pdlt130–75::phoZc	2 ± 0	2 ± 0	1 ± 0	2 ± 0

^a630Δerm and sigV mutant (MC361) with dlt promoter::phoZ fusion plasmids were grown in BHIS alone or with 1 mg/ml lysozyme and assayed for AP activity as described in Materials and Methods. Results are the means of calculated AP units ± standard errors of the means for at least three biological replicates. All biological replicates were performed as technical duplicates.

^bThe activities from site-directed mutagenesis constructs were compared to the native-sequence Pdlt₃₀₀::phoZ under the same conditions by Student's two-tailed t test with correction for multiple comparisons by the Holm-Sidak method. Bold text indicates P ≤ 0.05.

^cAP activity compared to activity from the full-length Pdlt₁₃₀::phoZ construct in the same strain and under the same conditions.

In order to test whether separate σ^A- and σ^V-dependent promoters overlap at T-43, we performed site-directed mutagenesis at the nucleotides −48 and −38 upstream of the dlt TSS (Fig. 5B). If distinct −10 promoter elements overlap at T-43, the −48 and −38 nucleotides would be the initial and final nucleotides of these elements, respectively. The Pdlt₃₀₀T-48C::phoZ fusion had negligible AP activity in both the parent strain and the sigV mutant in BHIS or in lysozyme, suggesting that T-48 is necessary for both constitutive and lysozyme-dependent transcription (Table 4). Compared to the native-sequence Pdlt₃₀₀::phoZ construct, AP activity from Pdlt₃₀₀A-38C::phoZ was lower in the parent strain and the sigV mutant in both BHIS and lysozyme. However, the mutation at A-38 did not abolish the induction of activity in lysozyme in the parent strain, indicating that σ^V-dependent transcription was retained. Therefore, A-38 may be important for basal dlt transcription, as well as σ^V-dependent transcription in lysozyme.

σ^V and the Dlt pathway impact C. difficile virulence in vivo.

Because the sigV and dlt mutants are more sensitive to lysozyme (Fig. 1A), we hypothesized that these mutants would be less fit in vivo. In a previous study, Ho et al. demonstrated that a sigV mutant is significantly attenuated in a hamster model of infection (21). However, that study was performed in the JIR8094 strain, which is attenuated for virulence in vivo (46, 50). To determine the relative impacts of the dlt and sigV mutations on virulence, hamster infections were performed using the dlt and sigV isogenic mutants derived from the 630Δerm strain.

Seven days after a single dose of clindamycin, hamsters were gavaged with approximately 5,000 spores of 630Δerm or the dltD or sigV mutant, as described in Materials and Methods. Fecal samples were collected daily, and cecal samples were collected at the point of morbidity (postmortem) to enumerate CFU. Hamsters infected with the sigV mutant reached morbidity significantly faster than those infected with 630Δerm (46.2 h ± 17.9 h for 630Δerm versus 33.2 h ± 6.3 h for sigV mutant), demonstrating that σ^V affects virulence in vivo (Fig. 6A). At the point of morbidity, the ceca of hamsters infected with the sigV mutant contained significantly more CFU than the ceca of those infected with 630Δerm, indicating that this mutant has a growth advantage in the host (Fig. 6B). Hamsters infected with the dlt mutant also reached morbidity significantly earlier than those infected with 630Δerm (46.2 h ± 17.9 h for 630Δerm versus 35.8 h ± 5.0 h for dlt mutant), but hamsters infected with the dlt mutant strain had CFU counts in cecal samples similar to those infected with the parent strain. These data suggest that the lack of d-Ala in the cell wall contributes to increased virulence in vivo but does not provide a growth advantage to the bacterium.

FIG 6.

dlt and sigV mutants are more virulent than the parent strain in a hamster model of infection. Syrian golden hamsters were inoculated with approximately 5,000 spores of 630Δerm (n = 17), dltD mutant (MC319; n = 13), sigV mutant (MC361; n = 12), or a 1:1 mixture of 630Δerm and MC319 (630Δerm versus dltD; n = 12). (A) Kaplan-Meier survival curve depicting time to morbidity. * indicates P ≤ 0.05 by log rank test. The inset table lists the average time to morbidity for each strain ± standard deviation (SD) with bold text indicating P ≤ 0.05 by log rank test. (B) Total number of C. difficile CFU recovered from cecal contents collected postmortem. The dotted line demarcates the limit of detection. The solid black line marks the mean. Numbers of CFU are compared to 630Δerm by a one-way analysis of variance with Dunnett's multiple-comparison test (* indicates P < 0.05). (C) The competitive index (CI) of the dlt mutant for each hamster coinfected with 630Δerm and the dlt mutant is shown. A CI of 1 indicates no fitness advantage. A CI of <1 indicates reduced fitness of the dlt mutant. A CI of >1 indicates increased fitness of the dlt mutant.

To determine if the increased virulence observed with dlt mutant infections could be due to an altered ability of the host to recognize the bacterium (i.e., immune system response to the lipoteichoic acid [LTA] antigen), we performed competitive infections with 1:1 mixtures of 630Δerm and dlt mutant spores. Hamsters coinfected with the mixture of 630Δerm and dlt mutant spores reached morbidity earlier than those infected with 630Δerm alone and at a rate comparable to those infected with the dlt mutant alone (35.4 h ± 5.1 h for coinfection versus 46.2 h ± 17.9 h for 630Δerm and 35.8 h ± 5.0 h for dlt mutant). The dlt mutant therefore remains more virulent than 630Δerm, even when 630Δerm is present. The total number of CFU recovered from the ceca of coinfected hamsters was comparable to the number of CFU recovered from the ceca of hamsters infected with either 630Δerm or the dlt mutant alone. Similar numbers of 630Δerm and dlt mutant CFU were recovered from the ceca of coinfected hamsters (Fig. 6B), and the mean competitive index for the dlt mutant was 1.2 (Fig. 6C), suggesting that neither strain had a significant competitive advantage in vivo.

Because our results for sigV mutant infections differed from results previously obtained in the JIR8094 background (21), we performed an additional experiment using JIR8094 and TCD20 strains (kindly provided by C. Ellermeier), to determine the basis for this variability (see Fig. S6 in the supplemental material). In our hands, animals infected with the JIR8094 strain succumbed to infection 3.7 days later on average than the 630Δerm-infected animals, similar to results obtained by other investigators (47, 51). The animals infected with strain TCD20 (JIR8094 csfV/sigV mutant) presented with symptoms of C. difficile infection (CDI) and became moribund faster than those infected with the JIR8094 parent strain (133.2 ± 75.2 h versus 76.9 ± 12.9 h for TCD20). This is in contrast to the findings of Ho et al. (21), who observed a much longer time to morbidity with the JIR8094 strain and low morbidity with the JIR8094 sigV mutant (TCD20). Thus, both sigV mutant strains caused animals to become moribund more quickly than did the parental strain.

DISCUSSION

Resistance to CAMPs can enable the survival of bacterial pathogens within the host (52, 53). As an intestinal pathogen, C. difficile encounters many CAMPs in the gut, including those produced by the host and indigenous microbiota (2, 4, 7–9, 54, 55). One mechanism that enables C. difficile to resist killing by CAMPs is the altering of cell surface charge via the Dlt pathway, which adds d-alanines to cell wall teichoic acids (12, 13). In this paper, we demonstrate that expression of the Dlt pathway is regulated by the extracytoplasmic function sigma factor σ^V. Moreover, we show that regulation of dlt by σ^V occurs in response to the host-produced CAMP lysozyme and that the incorporation of d-alanine into the cell wall is critical for lysozyme resistance.

The other alternative sigma factors examined, σ^T and σ^D, did not significantly contribute to dlt expression under the conditions tested. Similar to previous findings, we observed that sigT expression increased about 2-fold in the presence of lysozyme (data not shown [30]), implying that σ^T could contribute to lysozyme resistance through a mechanism other than Dlt. The sigT mutant also demonstrated a modest growth delay in polymyxin B (Fig. 1), but σ^T did not appear to influence dlt transcription in polymyxin B (see Fig. S3 in the supplemental material), suggesting that σ^T contributes to polymyxin B resistance through an alternate mechanism. The sigV mutant did not demonstrate a growth defect in polymyxin B (Fig. 1), and the sigV mutant induced dlt expression in polymyxin B similarly to the parent strain (Fig. 2B). Thus, polymyxin B induces dlt expression through a σ^D-, σ^T-, and σ^V-independent mechanism.

Because the ribotype 027 epidemic strains have proven very successful in colonizing and causing disease (56–58), we considered that these strains might have increased resistance to lysozyme. As evidenced by growth assays (Fig. 1), the R20291 strain (027 ribotype) was more sensitive to both lysozyme and polymyxin B than the 630Δerm strain (012 ribotype). Examination of dlt expression in R20291 showed that this strain induced dlt transcription more robustly than 630Δerm in polymyxin B and lysozyme (Fig. 2). But, analyses of d-alanine cell wall content revealed that R20291 incorporated less total d-alanine than did 630Δerm at baseline and in the tested CAMPs. Moreover, R20291 had greater morphological cell changes in lysozyme than did strain 630Δerm (Fig. 3). But, the R20291 strain had significantly more d-alanine incorporation when grown in polymyxin B than in BHIS alone, while no significant change in d-alanine content was observed for 630Δerm in polymyxin B (Fig. 4). The difference in dlt transcription by these strains was not explained by the nucleotide changes in their dlt promoter sequences, as demonstrated with reporter fusions to the R20291 and 630Δerm promoters (see Fig. S2 in the supplemental material). Based on these results, it is likely that R20291 encodes a regulatory factor that influences dlt transcription in response to polymyxin B, which is not present in the 630Δerm strain.

Our results identified the dlt operon as part of the σ^V regulon of C. difficile. In a previous study, Ho et al. identified σ^V as important for lysozyme resistance in C. difficile (21) and identified several σ^V-dependent transcripts, including a peptidoglycan deacetylase, putative exported proteins, an ABC transporter system, and many genes of unknown function, but dlt was not detected. σ^V and other ECF sigma factors have been shown to regulate dlt expression in Bacillus subtilis. C. difficile strain 630 has three identified ECF sigma factors, σ^T, σ^V, and σ^W (30), but σ^W is encoded in only a few strains (59). The R20291 genome encodes multiple sigma factors and putative regulatory proteins that are not present in strain 630. It is possible that in R20291 these regulators, or σ^D, are involved in transcription of dlt in response to other CAMPs or host conditions.

Though likely, these results do not definitively prove that σ^V directly regulates dlt in response to lysozyme. 5′ RACE identified multiple transcriptional start sites, which would be expected if σ^A and σ^V directly mediate RNA polymerase binding from distinct promoters. The transcriptional start sites that we identified are 5 bp apart, which is close enough that two distinct promoters would likely overlap at nt T-43 (Fig. 5B). Site-directed mutagenesis of nucleotides within the predicted −10 promoter elements revealed that a single nucleotide change at −43 or −48 upstream of the dlt TSS was sufficient to abolish lysozyme-dependent and -independent dlt expression (Table 4). In addition, mutagenesis of A-38 decreased both lysozyme-dependent and -independent dlt expression, without abolishing induction of dlt expression in lysozyme. These findings imply that the promoter regions required for σ^A- and σ^V-dependent transcription of dlt overlap. Moreover, a reporter fusion containing the putative −10 and −35 elements (Pdlt₇₅::phoZ) was not sufficient for activity, and full σ^V-dependent transcription was achieved only when additional upstream sequence was included (Pdlt₁₃₀::phoZ). We hypothesize that the tandem repeats contained within this 130-bp region may be important for binding of additional σ^V-dependent regulatory factors. The region that is necessary for a σ^V-dependent lysozyme response (130 to 75 bp upstream of the dltD TSS) is farther upstream than the predicted locations of the promoters, based on the transcriptional start sites identified (Fig. 5). A construct containing only the bp-130-to-75 region had no AP activity, with or without added lysozyme (Table 4). These results indicate that this region does not contain sufficient elements for transcription initiation. Moreover, AP activity peaked with the Pdlt₁₇₀::phoZ construct, which suggests that the regions of complementarity that we identified within this 170-bp region could be involved in secondary structures that impact transcription. Further studies are needed to identify additional factors that bind this region and influence dlt transcription.

Despite the increased sensitivity of the sigV and dlt mutants to lysozyme in vitro (Fig. 1B), both of these mutants demonstrated increased virulence in vivo (Fig. 6A). It is unlikely that this increased virulence is due to increased toxin production, because we observed similar levels of tcdA expression in the two mutants in vitro (see Fig. S4A in the supplemental material). Levels of toxin expression in the cecal contents of infected hamsters at the time of morbidity were also similar between strains, although the cecal contents of hamsters infected with the sigV mutant trended toward higher toxin levels (see Fig. S4B). Given the importance of σ^V and Dlt in lysozyme resistance in vitro, one might expect that the lack of cell wall modification in the dlt and sigV mutants would make the bacteria more susceptible to innate immune clearance. However, the presence of cell wall modifications, while protective against innate immune effectors, is also immunogenic and may increase the host response to the pathogen. Thus, it is possible that the increased virulence of the dlt mutant may be due to an altered host immune response to this mutant. d-Alanylated lipoteichoic acid (LTA) is an epitope for the host receptor Toll-like receptor 2 (TLR2) (60, 61). In most pathogens investigated, the lack of a functional Dlt pathway results in decreased virulence (14, 62–64). But, it is possible that d-alanylation of LTA may be a mechanism by which C. difficile can mask the immunogenic portions of LTA and evade an immune response. Such a mechanism would be similar to how d-alanylation of LTA in Staphylococcus aureus masks antigenic portions of peptidoglycan, resulting in decreased virulence because the immune system can better respond to antigens that are unmasked in the mutant (65). But, in CDI, a more robust immune response leads to greater intestinal injury (55, 66, 67). Because the dlt mutant lacks d-alanylated LTA (Fig. 4), this mutant may elicit a stronger immune response, which causes more severe disease symptoms than does the parent strain. In coinfection experiments, hamsters reached morbidity at a rate comparable to those infected with the dlt mutant alone (Fig. 6A), which would be expected if an enhanced immune response to the mutant leads to increased virulence. However, the parent strain did not have a colonization advantage during coinfection (Fig. 6B and C), as might be expected if the dlt mutant were more readily recognized by the immune system.

σ^V has been established as an important factor for colonization and virulence in E. faecalis, though in E. faecalis a sigV mutant is less virulent than the parent strain and σ^V does not control dlt expression (14). In C. difficile, the sigV mutant retains a baseline level of d-alanylated LTA (Fig. 4) but is unable to induce other σ^V-dependent modifications of the cell surface. The immunogenicity of σ^V-dependent surface modifications is unknown, but our results suggest that σ^V plays a role in host colonization and may affect recognition of the pathogen by the host. Moreover, the increased virulence that we observed for the sigV mutant contradicts an earlier finding of attenuated virulence for a C. difficile sigV mutant (21). This previous study was performed with a sigV mutant in the strain JIR8094 background. JIR8094 colonizes the intestine more slowly than 630Δerm, has lower toxin A and B production (see Fig. S5A in the supplemental material), is nonmotile, has lower expression of flagellar genes, and is overall less virulent (46–48, 50, 68). The virulence defects of this strain explain the shorter average time to morbidity with our parent strain compared to that of Ho et al. (21). Moreover, in our hands, this sigV mutant (TCD20) was more virulent than the parent strain (see Fig. S6). Our results with JIR8094 are more similar to previously published experiments with this strain (47) than the results obtained by Ho et al. (21). Possible reasons for the discrepancies observed between our results and those of Ho et al. (21) may be due to differences in the spore preparation or timing of clindamycin administration, variations in hamster genetics, or differences in the microbiomes of the animals used in these studies.

A number of other questions remain to be answered about the regulation of dlt and the role of σ^V in C. difficile. Does σ^V directly regulate dlt expression? Are the tandem repeats upstream of dlt binding sites for a regulatory factor? What factors regulate dlt expression in response to other triggers, such as polymyxin B? Despite these remaining questions, our finding that σ^V regulates the Dlt pathway in C. difficile in response to lysozyme represents an important insight into the mechanisms that enable C. difficile colonization. These results underscore the complex relationship between mechanisms of antimicrobial resistance and the effects of these modifications on virulence. Mutants in more virulent isolates, such as R20291, may allow for further study of these mechanisms in the mouse model of CDI, which would enable more detailed investigation of the immune response to cell wall modifications. Surviving the innate immune response is a critical step in the process of disease progression and therefore represents a key window of opportunity for therapeutic intervention and prevention of pathogenesis. Identifying ways to increase C. difficile susceptibility to innate immune responses may help extend the utility and efficacy of our current antibiotic therapies.