Sortase-deficient lactobacilli: effect on immunomodulation and gut retention

Abstract

Surface proteins of probiotic microbes, including Lactobacillus acidophilus and Lactobacillus gasseri, are believed to promote retention in the gut and mediate host–bacterial communications. Sortase, an enzyme that covalently couples a subset of extracellular proteins containing an LPXTG motif to the cell surface, is of particular interest in characterizing bacterial adherence and communication with the mucosal immune system. A sortase gene, srtA, was identified in L. acidophilus NCFM (LBA1244) and L. gasseri ATCC 33323 (LGAS_0825). Additionally, eight and six intact sortase-dependent proteins were predicted in L. acidophilus and L. gasseri, respectively. Due to the role of sortase in coupling these proteins to the cell wall, ΔsrtA deletion mutants of L. acidophilus and L. gasseri were created using the upp-based counterselective gene replacement system. Inactivation of sortase did not cause significant alteration in growth or survival in simulated gastrointestinal juices. Meanwhile, both ΔsrtA mutants showed decreased adhesion to porcine mucin in vitro. Murine dendritic cells exposed to the ΔsrtA mutant of L. acidophilus or L. gasseri induced lower levels of pro-inflammatory cytokines TNF-α and IL-12, respectively, compared with the parent strains. In vivo co-colonization of the L. acidophilus ΔsrtA mutant and its parent strain in germ-free 129S6/SvEv mice resulted in a significant one-log reduction of the ΔsrtA mutant population. Additionally, a similar reduction of the ΔsrtA mutant was observed in the caecum. This study shows for the first time that sortase-dependent proteins contribute to gut retention of probiotic microbes in the gastrointestinal tract.

Introduction

Bacterial cell surface proteins have been implicated in the ability of microbes to communicate and interact with their host. Investigation of these proteins aids in understanding the mechanisms underlying health benefits which have been described for commensal and probiotic microbes, including competitive inhibition of pathogens, immunomodulation and maintenance of the gastrointestinal tract (GIT) epithelial barrier (O'Flaherty & Klaenhammer, 2010). In Lactobacillus acidophilus NCFM, deficiency of the cell wall constituent lipoteichoic acid resulted in a shift of pro-inflammatory signals present in a murine colon cancer model such that they returned to a baseline, protective level (Khazaie et al., 2012). Additionally, surface (S)-layer protein A of L. acidophilus has been characterized for its direct interaction with the dendritic cell (DC)-specific ICAM-3-grabbing non-integrin (DC-SIGN) receptor on DCs and subsequent modulation of homeostatic cytokine signals to potentially abrogate pathogen-induced immune signalling (Konstantinov et al., 2008). These reports emphasize the potential contribution of cell surface proteins of probiotic microbes in immunomodulatory roles.

The sortase protein, first identified in Staphylococcus aureus in 1999, covalently couples a subset of surface proteins, referred to as sortase-dependent proteins (SDPs), to the cell wall peptidoglycan of Gram-positive micro-organisms (Mazmanian et al., 1999). Although multiple classes of sortase enzymes exist which function in pili assembly and iron acquisition, the class A sortase (SrtA) is solely responsible for anchoring surface proteins to the cell wall. SrtA recognizes protein targets exported by the secretory pathway that contain a C-terminal LPXTG- or LPXTG-like motif, preceded by a hydrophobic region and positively charged tail. The sortase enzyme then cleaves between the Thr and Gly residues, at which point the sortase and SDP form a complex linked by a thioester acyl bond. This bond is then subjected to nucleophilic attack; the SDP is subsequently linked to lipid II, and incorporated as part of the cell wall biosynthetic process (Hendrickx et al., 2011; Kleerebezem et al., 2010; Maresso & Schneewind, 2008; Spirig et al., 2011).

In S. aureus, the sortase enzyme couples between 18 and 22 SDPs to the cell surface, many of which are associated with virulence and host adhesion (Marraffini et al., 2006). Similarly, SDPs play a role in the virulence of several other Gram-positive pathogenic species such as Listeria monocytogenes (Bierne et al., 2002) and Streptococcus pneumoniae (Paterson & Mitchell, 2006). Meanwhile, the role of SrtA has only recently been studied in species of commensal and probiotic microbes, including Lactobacillus casei BL23 (Muñoz-Provencio et al., 2012), Lactococcus lactis IL1403 (Dieye et al., 2010), Lactobacillus plantarum WCFS1 (Remus et al., 2013) and Lactobacillus salivarius UCC118 (O’Callaghan et al., 2012; van Pijkeren et al., 2006). These studies demonstrated that SrtA and SDPs play pivotal roles in the adhesion to intestinal epithelial cells (IECs), immune modulation of DCs, as well as the expression of mucin genes in Caco-2 IECs. To date, a role for SrtA in gut retention has not been demonstrated for probiotic lactic acid bacteria.

In the current study, SrtA was examined in the context of a widely used commercial probiotic, L. acidophilus, and the commensal GIT and genital tract inhabitant, Lactobacillus gasseri. The genomes of both L. acidophilus NCFM and L. gasseri ATCC 33323 have been sequenced, enabling targeted gene deletion and phenotypic analyses (Altermann et al., 2005; Azcarate-Peril et al., 2008). Each of the two genomes encodes one SrtA, along with eight and six intact SDPs predicted in L. acidophilus and L. gasseri, respectively. A sortase gene deletion (ΔsrtA) mutant of each strain was generated and assessed with regard to key probiotic traits in vitro, including survival in gastrointestinal juices, adhesion capacity and immune signalling. Furthermore, in vivo gut retention of the L. acidophilus ΔsrtA mutant was evaluated in a germ-free mouse model. Co-colonization of both ΔsrtA and parent strains in the mice resulted in a significant reduction of the mutant population compared with the parent, indicating that SDPs are involved in the gut retention of probiotic microbes in the GIT.

Methods

Bacterial strains and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1. Lactobacillus strains were propagated in deMan, Rogosa and Sharpe (MRS) medium (Becton Dickinson [BD]) at 37 °C statically under aerobic conditions. Escherichia coli strains were maintained in brain heart infusion medium (BHI) (BD) at 37 °C aerobically with shaking. Plating was performed on solid medium with the addition of 1.5 % (w/v) agar to either MRS or BHI medium. Antibiotics including erythromycin (Em, 2–5 µg ml⁻¹ for lactobacilli or 150 µg ml⁻¹ for E. coli) (Fisher Scientific), kanamycin (Kn, 40 µg ml⁻¹) (Fisher), chloramphenicol (Cm, 2–5 µg ml⁻¹) (Fisher), rifampicin (Rif, 100–250 µg ml⁻¹) (Sigma Aldrich) and streptomycin (Str, 500–1000 µg ml⁻¹) (Sigma) were aseptically added to sterilized MRS or BHI broth or agar medium where indicated.

Table 1. Strains and plasmids used in this study.

Strain or plasmid	Genotype or characteristics	Reference
E. coli
EC101	repA integrated in chromosome, Knr, ΔlacZ, host for pORI-based plasmids	Law et al. (1995)
NCK1911	Host for pTRK935, a pORI-based plasmid used for counterselective gene replacement	Goh et al. (2009)
NCK2231	pTRK1059 host	This study
NCK2255	pTRK1066 host	This study
L. acidophilus
NCK56	NCFM strain, human intestinal isolate	Barefoot & Klaenhammer (1983)
NCK1909	NCFM Δupp; parent/background strain for L. acidophilus deletion mutants	Goh et al. (2009)
NCK1910	NCK1909 harbouring pTRK669 helper plasmid, host for upp-based counterselective gene replacement	Goh et al. (2009)
NCK2232	NCK1909 ΔsrtA	This study
NCK2272	NCK2232 with stable mutation conferring StrR	This study
NCK2300	NCK1909 with stable mutation conferring RifR	Goh & Klaenhammer (2014)
L. gasseri
NCK2253	ATCC 33323 Δupp; parent/background strain for L. gasseri deletion mutants	Selle et al. (2014)
NCK2254	NCK2253 harbouring pTRK669 helper plasmid, host for upp-based counterselective gene replacement	Selle et al. (2014)
NCK2256	NCK2253 ΔsrtA	This study
Plasmids
pTRK669	Ori (pWV01), Cmr, RepA+, thermosensitive	Russell & Klaenhammer (2001)
pTRK935	pORI-based integration vector used in counterselective gene replacement, P-upp, lacZ', EmR	Goh et al. (2009)
pTRK1059	pTRK935 containing deletion construct for LBA1244 (srtA) cloned into BamHI/SacI sites, EmR	This study
pTRK1066	pTRK935 containing deletion construct for LGAS_0825 (srtA) cloned into BamHI/SacI sites, EmR	This study

DNA manipulation strategies.

The DNA manipulation and cloning methodology used to construct ΔsrtA mutants followed those previously described (Goh et al., 2009). All primers were synthesized by Integrated DNA Technologies (Table 2). PCR amplifications for gene deletions were generated using the Expand High Fidelity Plus PCR system (Roche Molecular Biochemicals) and were digested for cloning purposes using BamHI and SacI (Roche). Ligation was performed using NEB Quick Ligase (New England BioLabs). Chemically competent E. coli cells (Hanahan, 1985) were used for cloning of recombinant plasmids and the plasmids were electrotransformed into L. acidophilus or L. gasseri as previously described (Walker et al., 1996). PCR screening for deletion mutants was performed using Choice-Taq Blue DNA polymerase (Denville Scientific).

Table 2. Primers used in this study.

Primer target	Primer sequence (5′–3′)*†
LBA1244
Upstream	F: GTAATAGGATCCCTCACTTGACTATGAGATTACTG
R: CGTAGTAACACTACTCTTTTGTTT
Downstream	F: AAAAGAGTAGTGTTACTACGGCAACAGATGAAAATTTGAAGG
R: TAAAGTAGAGCTCGGCATATTTAACTTCATCGATTCC
Gene deletion screening and sequencing	F: GGCGAATTTGTTACAATGG
	R: GTATAGGCAACTCCAGCACC
LGAS_0825
Upstream	F: GTAATAGGATCCGATAACAGGTTATCGTGCCACCG
R: GCAGCAATTCGAACTATCCA
Downstream	F: TGGATAGTTCGAATTGCTGCGCCTGCAACAGATAAGAATTTA
R: TAAAGTAGAGCTCCCTGAAATGTATGGTGATCAG
Gene deletion screening and sequencing	F: CAATATGCCATTGGCTGAG
	R: GCCCGACACATTGTATAAG

^*Forward and reverse primers are denoted by F and R, respectively.

^†Underlined sequences denote restriction sites.

Construction of sortase-deficient derivatives of L. acidophilus NCFM and L. gasseri ATCC 33323.

An in-frame deletion of the SrtA-encoding gene, srtA, in L. acidophilus NCFM (LBA1244) or L. gasseri ATCC 33323 (LGAS_0825) was constructed using a upp-counterselective gene replacement system (Goh et al., 2009; Selle et al., 2014). Artemis software (Rutherford et al., 2000) was used to visualize the specific chromosomal region associated with each of the srtA genes, and primers (Table 2) were designed to amplify homologous regions of approximately 600 bp upstream and downstream of the deletion target. Once PCR-amplified, the two flanking regions were then linked together using splicing by overlapping extension PCR (SOE-PCR) (Horton et al., 1989). The counterselective integration vector, pTRK935 (Table 1), and the SOE-PCR product were digested with BamHI and SacI, ligated and transformed into chemically competent E. coli EC101 cells. Inserts of recombinant plasmids obtained from positive clones were sequenced by Davis Sequencing to confirm nucleotide sequence fidelity. These recombinant plasmids, designated pTRK1059 and pTRK1066 (Table 1), were electrotransformed into their respective Lactobacillus background hosts, L. acidophilus NCK1910 and L. gasseri NCK2254 (Table 1). Transformants were isolated based on their resistance to both Em and Cm (2 µg ml⁻¹ each) conferred by the presence of pTRK669 and either pTRK1059 or pTRK1066, respectively. These transformants were then subcultured three times in MRS broth with 2 µg Em ml⁻¹ in a 42 °C water bath to cure the pTRK669 helper plasmid and facilitate selection of integrants resulting from homologous recombination of either pTRK1059 or pTRK1066 into the L. acidophilus or L. gasseri chromosome, respectively. Chromosomal integration rendered the cells Cm-sensitive (Cm^S), Em-resistant (Em^R) and 5-fluorouracil (5-FU; Sigma)-sensitive (5-FU^S), and the integrants were isolated by replica plating on MRS agar with 2 µg Em ml⁻¹ or 5 µg Cm ml⁻¹. Selected integrants with Em^R and Cm^S phenotypes were subcultured three times in MRS broth without antibiotic at 24 h intervals at 37 °C. Following the third passage, cultures were diluted and plated on glucose semi-defined medium (Kimmel & Roberts, 1998) supplemented with 100 µg 5-FU ml⁻¹ (Goh et al., 2009) to select for 5-FU^R colonies in which plasmid excision had occurred, resulting in reversion to either the WT or ΔsrtA genotype. Colonies with the ΔsrtA genotype were screened by PCR using primers flanking the deletion region (Table 2). Deletions were confirmed by DNA sequencing and the ΔsrtA mutants of L. acidophilus and L. gasseri were designated NCK2232 and NCK2256, respectively (Table 1).

Growth and survival assays.

L. acidophilus NCK2232 and L. gasseri NCK2256 were evaluated for growth in MRS and survival in simulated gastric juices (SGJs) and small intestinal juices (SIJs) (Charteris et al., 1998; Frece et al., 2005; Goh & Klaenhammer, 2010) and compared to the parent strains L. acidophilus NCK1909 and L. gasseri NCK2253, respectively. For growth studies, the growth in MRS broth (2 % inoculum from a stationary phase culture) was monitored for 12 h at OD₆₀₀ and viable cells (c.f.u. ml⁻¹) were enumerated by plating on MRS. For survival assays, 1 ml of a 16 h MRS-grown overnight culture was spun down, washed twice in PBS (pH 7.4; Invitrogen) and resuspended in 1 ml of sterile water. This cell suspension (0.2 ml) was combined with 1 ml of SGJ (0.5 % NaCl, 3 g pepsin l⁻¹, pH 2.0) or 1 ml of simulated SIJ (0.5 % NaCl, 1 g pancreatin l⁻¹ and 3 g Oxgall l⁻¹, pH 8.0) and incubated at 37 °C. Survival was measured over the course of 1 h in SGJ and 4 h in SIJ as determined by plating on MRS agar (Goh & Klaenhammer, 2010). Pepsin and NaCl were both obtained from Fischer Scientific; pancreatin is a product of MP Biomedicals and Oxgall was obtained from Difco.

Mucin and Caco-2 adherence assays.

Adherence to porcine mucin (Sigma) and Caco-2 adenocarcinoma cells (ATCC HTB-37) was performed as described previously (Goh et al., 2009). Briefly, type III porcine gastric mucin (Sigma) was immobilized in 96-well flat bottom microtitre plates for 18 h at 4 °C. Excess mucin was washed with PBS and blocked with 2 % BSA (Invitrogen) at 4 °C. Wells were washed twice with PBS to remove excess BSA. An overnight culture was washed in PBS and adjusted to an OD₆₀₀ of 0.6. The bacterial suspension (100 µl) was added to each well and incubated for 1 h at 37 °C. Adherent cells were recovered following treatment of the wells with 0.05 % (v/v) Triton X-100 (Fisher) and plating on MRS. For Caco-2 adherence assays, 21 day Caco-2 monolayers (between passages 26 and 45) were incubated (37 °C) with bacterial strains resuspended in PBS at 1×10⁸ c.f.u. ml⁻¹ for 1.5 h. Following incubation, Caco-2 monolayers were washed five times with PBS and disrupted with 0.05 % Triton X-100. Adherent bacterial cells were enumerated on MRS.

DC assays.

Bacterial strains were co-incubated with marrow-derived murine immature DCs and supernatants from the DCs were evaluated for the production of four cytokines (TNF-α, IL-6, IL-10 and IL-12) using sandwich ELISA according to the manufacturer’s instructions (Qiagen). DCs from BALB/c mice (Astarte-Biologics) were stored in liquid nitrogen until use. On the day of the experiment, DCs were thawed, transferred to a sterile 50 ml conical tube containing 100 µg of DNase I (Stem Cell Technologies) and resuspended in 25 ml of RPMI 1640 medium (Invitrogen) supplemented with 10 % (v/v) FBS (Invitrogen). Cells were collected by centrifugation (200 g, 15 min, room temperature) and the cell pellet was again treated with 100 µg of DNase I and resuspended in 25 ml of RPMI with 10 % FBS. After centrifugation, 100 µg of DNase I was added and the cells were resuspended in 5 ml of the same medium. Following enumeration of viable cells using Trypan Blue staining (Sigma), DCs were diluted to a final concentration of approximately 1×10⁶ DCs ml⁻¹ in RPMI with 10 % FBS and 100 µg Str ml⁻¹. Aliquots (100 µl) of the standardized DC suspension were transferred into wells of 96-well polypropylene plates (Sigma). These plates containing DCs were then incubated at 37 °C in 5 % CO₂ while bacterial strains were prepared.

Stationary phase cultures of L. acidophilus NCK1909 and NCK2232 (ΔsrtA), and L. gasseri NCK2253 and NCK2256 (ΔsrtA) grown in MRS broth were collected (10 000 g, 1 min, room temperature) and washed once with PBS. Supernatants were discarded and each strain was resuspended in PBS to a final OD₆₀₀ corresponding to 1×10⁸ c.f.u. ml⁻¹, centrifuged (10 000 g, 1 min) and each cell pellet was resuspended in 1 ml of RPMI with 10 % FBS. A standardized bacterial suspension (200 µl) was added to the DCs in the 96-well plate so that the final ratio of bacterial to DCs was approximately 10 : 1. Co-incubation was carried out for 24 h (37 °C, 5 % CO₂). Supernatants were collected after centrifugation (1735 g, 10 min) of the 96-well plates and stored at −80 °C until assayed using ELISA according to the manufacturer's instructions (Qiagen).

Selection of naturally antibiotic-resistant strain of L. acidophilus NCK2232.

A Str-resistant (Str^R) derivative of the L. acidophilus ΔsrtA mutant (NCK2232) was selected on MRS agar supplemented with 1000 µg ml⁻¹ of Str and designated NCK2272 (Table 1). A Rif-resistant (Rif^R) derivative of L. acidophilus NCK1909, designated NCK2300, which is stable and does not display altered growth characteristics, was used as the control parent strain (Goh & Klaenhammer, 2014) (Table 1). L. acidophilus NCK2272 was evaluated for growth in MRS broth and the growth profile resembled that of the parental strain (NCK2232). Briefly, overnight culture of each strain was inoculated at 1 % into triplicate wells containing 200 µl aliquots of MRS medium in a 96-well plate. Growth was monitored over 24 h at 37 °C at OD₆₀₀ in a microtitre plate reader (FLUOROstar OPTIMA). Each strain was represented in triplicate and the entire experiment was performed on three separate occasions. Additionally, NCK2272 was evaluated for its ability to maintain the Str^R marker without antibiotic pressure in MRS broth over the course of 28 days of continuous subculturing. Plating of the subcultures on both MRS and MRS with Str was performed daily for the first 7 days, and subsequently on days 14, 21 and 28.

In vivo colonization studies of L. acidophilus parent and ΔsrtA mutant strains in a germ-free mouse model.

Germ-free 129S6/SvEv mice used in the colonization experiments were from breeding colonies maintained in the North Carolina State University Gnotobiotic Core of the Center for Gastrointestinal Biology and Disease. Mice were maintained in cages in germ-free flexible film isolators housed in a room with cycles of 12 h of light and darkness, and were provided access to a standard diet (Prolab RMH 3500, LabDiet) and water ad libitum. Germ-free status is periodically evaluated, no less frequently than once a month, by culturing faecal samples in thioglycollate broth, on blood agar and Sabouraud agar. In addition, prior to colonization experiments, the mice were also verified germ-free by culturing of faecal samples aerobically and anaerobically on plate count agar and MRS agar. Animal use protocols were approved by the Institutional Animal Care and Use Committee of North Carolina State University.

Mono-colonization experiments with L. acidophilus WT (NCK56) or ΔsrtA mutant (NCK2232).

L. acidophilus NCK56 and NCK2232 were propagated overnight in MRS broth (37 °C) and harvested at stationary phase (16 h of growth). Bacteria were harvested by centrifugation (1735 g, 10 min, room temperature), washed once in PBS and resuspended in PBS to an OD₆₀₀ corresponding to 5×10⁹ c.f.u. ml⁻¹ for each strain.

For mono-colonization studies with either NCK56 or NCK2232, germ-free 129S6/SvEv mice (n = 6; 13–14 weeks old; males and females) were gavaged with 1×10⁹ cells in 200 µl PBS (200 µl per mouse) for two consecutive days (representing days −1 and 0). Faecal samples were collected periodically over the course of 4 weeks. Each faecal sample was weighed, resuspended in 1 ml of PBS, diluted and plated on MRS for enumeration. Viable cell counts were expressed in c.f.u. g⁻¹ of faecal samples.

Co-colonization experiments with L. acidophilus parent (NCK2300) and ΔsrtA mutant (NCK2272).

L. acidophilus NCK2300 (Rif^R) and NCK2272 (ΔsrtA, Str^R) were propagated in MRS broth (37 °C) for 16 h, harvested by centrifugation (1735 g, 10 min, room temperature), washed once in PBS, and resuspended in PBS to an OD₆₀₀ corresponding to 5×10⁶ c.f.u. ml⁻¹ for each strain. Equal volumes of L. acidophilus NCK2300 and NCK2272 were combined in a 1 : 1 ratio for administration to mice.

Five (three female, two male), 17-week old germ-free 129S6/SvEv mice were gavaged once with 200 µl of the above bacterial suspension so that each mouse received a total of ca. 1×10⁶ c.f.u. ml⁻¹ containing 5×10⁵ c.f.u. of each strain. Faecal samples were collected approximately 72 h following the gavage (day 3) and on days 5, 7, 10, 14, 21 and 24 thereafter. Faecal pellets were resuspended in 1 ml of PBS, diluted and cell counts of L. acidophilus NCK2300 and NCK2272 (c.f.u. g⁻¹) were differentially enumerated on MRS with 100 µg Rif ml⁻¹ and 500 µg Str ml⁻¹, respectively. In addition, mice were euthanized on day 24 and the caeca and small intestines were excised for enumeration of NCK2300 and NCK2272 in the GIT. To recover bacterial cells from the tissue samples, the entire organ (caecum or small intestine) from each of the five mice was homogenized in 10 ml of PBS, diluted and then plated on the corresponding selective media.

Results

Sortase and SDPs in L. acidophilus and L. gasseri

The genomes of both L. acidophilus NCFM and L. gasseri ATCC 33323 contain a single class A housekeeping sortase gene, LBA1244 and LGAS_0825, respectively. Each strain possesses 12 potential SDPs, as identified using the LAB secretome database and a domain search for the LPXTG cell wall anchoring motif (Table 3) (Zhou et al., 2010). Each target was then further assessed for its potential functionality based on the presence of a signal peptide as determined using InterProScan (http://www.ebi.ac.uk/interpro/interproscan.html) and SignalP 4.1 (http://www.cbs.dtu.dk/services/SignalP/) (Petersen et al., 2011). In total, eight and six intact SDPs (containing both signal peptide and LPXTG motif) were identified in L. acidophilus NCFM and L. gasseri ATCC 33323, respectively (Table 3). Of the eight SDPs in L. acidophilus, four were predicted as mucus-binding (mub) proteins or mub protein precursors (LBA1019, LBA1392, LBA1652 and LBA1709). Similarly, in L. gasseri, two of the six intact SDPs were annotated as adhesion exoproteins (LGAS_0942 and LGAS_0943), each containing varying numbers of mub domain repeats (Table 3) (Azcarate-Peril et al., 2008).

Table 3. SDPs identified in the genomes of L. acidophilus NCFM and L. gasseri ATCC 33323.

Predicted function	ORF designation	Protein length (no. of aa)	Signal peptide (+/−)*	Signal peptide cleavage site†
L. acidophilus NCFM
Putative mucus-binding protein	LBA1018	346	−
Putative mucus-binding protein	LBA1019	2650	+	VHA/DEINI
Mucus-binding protein precursor	LBA1392	4326	+	VHA/ENIDN
Mucus-binding protein precursor	LBA1652	1174	+	PVKA/TSS
Mucus-binding protein precursor	LBA1709	1208	+	VQA/DSVE
Surface protein	LBA1611	2539	−
Surface protein	LBA1633	1659	−
Surface protein	LBA1634	1924	+	QA/ATEEE
Putative fibrinogen-binding protein	LBA1496	991	−
Hypothetical protein	LBA0036	435	+	TYA/ANLSD
Putative membrane protein	LBA1740	1376	+	VNA/DEMT§
Hypothetical protein	LBA1793	438	+	HA/DKGST
L. gasseri ATCC 33323
Adhesion exoprotein‡	LGAS_0045	3692	−
Adhesion exoprotein‡	LGAS_0143	2823	−
Hypothetical protein	LGAS_0146	967	+	VQA/ASTN
Hypothetical protein	LGAS_0383	476	−
Adhesion exoprotein‡	LGAS_0410	2457	−
Hypothetical protein	LGAS_0866	268	+	AFA/ATTDS
Adhesion exoprotein‡	LGAS_0942	2833	+	GVAKA/DTV
Adhesion exoprotein	LGAS_0943	979	+	TAQA/DSVN
5′-nucleotidase/2′,3′-cyclic phosphodiesterase- related esterase	LGAS_1067	765	+	GSVKA/DEV
Hypothetical protein	LGAS_1663	2449	+	TTVQA/ASA
Possible cell surface protein	LGAS_1671	2552	−
Possible cell surface protein	LGAS_1725	1993	−

^*The presence of a signal peptide was determined using InterProScan (http://www.ebi.ac.uk/Tools/pfa/iprscan/).

^†Signal peptides for L. acidophilus (this study) and L. gasseri (Kleerebezem et al., 2010) were determined using SignalP 4.1 (http://www.cbs.dtu.dk/services/SignalP/). The predicted cleavage sites (indicated by the backslash) for both microbes were also determined using SignalP 4.1 (this study).

^‡Contain variable numbers of mucus-binding (mub) domains. LGAS_0943 was also blasted using blastp; however, no mub domain was found in this particular adhesion exoprotein.

^§Signal peptide of YSIRK origin was detected by InterProScan; however, this signal peptide was not detected within the cut-off restraints (D = 0.500) of SignalP 4.1. The sequence listed is the most likely cleavage site determined using SignalP 4.1 (D = 0.450).

Construction of sortase-deficient mutants

Sortase deletion mutants of L. acidophilus NCFM and L. gasseri ATCC 33323 were generated using upp-based counterselection gene replacement systems (Goh et al., 2009; Selle et al., 2014). The ΔsrtA mutant of each strain was confirmed by PCR and DNA sequencing using primers flanking the srtA deletion target (Table 2). The resulting ΔsrtA isogenic mutants of L. acidophilus and L. gasseri were designated NCK2232 and NCK2256, respectively (Table 1). Both mutants contain in-frame deletions of approximately 90 % of the sortase gene. Due to the deficiency of SrtA, these strains are also expected to lack the surface display of predicted SDPs (Dieye et al., 2010; Remus et al., 2013). When both ΔsrtA mutants were examined for their growth characteristics in culture medium (MRS) and survival in SGJ and SIJ, neither L. acidophilus NCK2232 nor L. gasseri NCK2256 showed growth impairment or increased sensitivity to stress, when compared to their respective parent strains NCK1909 and NCK2253 (data not shown).

Adhesion of ΔsrtA mutants to porcine mucin and IECs in vitro

To determine the effect of sortase deletion on the adhesive capacity of L. acidophilus NCK2232 and L. gasseri NCK2256, the mutants and their respective parent strains were incubated with immobilized porcine mucin and Caco-2 epithelial monolayers. Both NCK2232 and NCK2256 mutants showed reduced adhesion to porcine mucin (a reduction of 27.08±5.52 % and 36.90±10.11 %, respectively) compared with the parental strains (Fig. 1). For adherence to IECs using the Caco-2 model, neither of the ΔsrtA mutants demonstrated differences in adherence capacity as compared with their parent strains (Fig. 1). These results indicate that the deficiency of sortase in L. acidophilus and L. gasseri does not impact adherence in vitro to this particular cell line under the experimental conditions. The observation that mucin adherence was impacted points to the specific binding capacity of some of the SDPs, likely to be mub proteins and adhesion exoproteins, to mucin and perhaps specific cellular components or receptors not presented on Caco-2 cells.

Fig. 1.

Relative adherence of stationary phase ΔsrtA mutants of L. acidophilus NCFM (NCK2232, dark grey bars) and L. gasseri ATCC 33323 (NCK2256, light grey bars) to (a) porcine mucin and (b) Caco-2 cells. Relative adherence was determined as the percentage adherence of the ΔsrtA mutants as compared with that of the parent strain (black bars; standardized to 100 %) by plating and enumeration of adhered cells. Results shown are the median of three independent experiments (each with four technical replicates) with error bars representing 1 sd. Asterisks indicate a significant reduction in adherence (P<0.05) as determined by the Student t-test.

Sortase mutants elicit anti-inflammatory signals from murine immature DCs

Immune cells residing in the GIT communicate with the microbiota and elicit specific responses through cytokine production. The interaction of microbial cell surface proteins with the GIT is crucial in the understanding of microbe–host signalling. To this end, the production of IL-6, IL-10, IL-12 and TNF-α was quantified from the supernatants of murine DCs co-incubated with either of the ΔsrtA mutants or their respective parent strains (Fig. 2). DCs exposed to the ΔsrtA mutant of L. acidophilus (NCK2232) showed a marked decrease in production of the pro-inflammatory cytokine TNF-α as compared with the parent strain (NCK1909) (Fig. 2d). In addition, DCs incubated with the ΔsrtA mutant of L. gasseri (NCK2256) produced significantly less IL-12 than did the parent L. gasseri strain (NCK2253) (Fig. 2c). No significant changes in IL-10 (Fig. 2a) or IL-6 (Fig. 2b) production were observed between the DCs co-incubated with ΔsrtA mutant of either L. acidophilus or L. gasseri and the parent strains. Clear differences in the cytokine profiles of DCs were observed between the parent strains of L. acidophilus and L. gasseri. Notably, cytokine production, with the exception of IL-12, was significantly higher in the supernatants of DCs incubated with L. acidophilus than with L. gasseri (Fig. 2). These results point to the cell surface structure of L. acidophilus, potentially attributed by the presence of the S-layer (typically not found in L. gasseri), which may stimulate cells of the immune system to a greater degree.

Fig. 2.

Cytokine expression of murine DCs exposed to both the parent (grey bars) and ΔsrtA mutants (hashed white bars) of both L. acidophilus NCFM and L. gasseri ATCC 33323. Actual concentrations of (a) IL-10, (b) IL-6, (c) IL-12 and (d) TNF-α from supernatants of DCs incubated with the parent strain or ΔsrtA mutant as measured by ELISA are shown. The bacterial cell to DC ratio was targeted at 10 : 1. The data shown are the mean cytokine concentration from three independent ELISAs containing two technical replicates each. Error bars represent 1 sd, and statistically significant changes are indicated by either a single or double asterisk denoting P<0.05 or P<0.01, respectively, as determined by the Student t-test.

L. acidophilus SrtA impacts gut retention in vivo

Prior to co-colonization of germ-free mice with a 1 : 1 ratio of the parent and ΔsrtA derivative of L. acidophilus, both strains were individually examined for their ability to persist in the GIT of germ-free 129S6/SvEv mice. When mono-associated, both strains were found to persist at similar levels for a period of 28 days (Fig. 3), and thus competitive co-colonization was pursued to further evaluate the effect of sortase deficiency on the gut persistence of L. acidophilus.

Fig. 3.

Enumeration of L. acidophilus NCK56 WT strain and L. acidophilus NCK2232 (ΔsrtA) when mono-colonized in 129S6/SvEv germ-free mice. Each data point represents the mean bacterial cell count (c.f.u. g⁻¹) from six mice faecal samples and the error bars represent 1 sd.

In order to selectively enumerate the L. acidophilus parent (NCK1909) and ΔsrtA mutant (NCK2232) when co-colonized in 129S6/SvEv mice, the parent and mutant strains were marked with spontaneous resistance to Rif and Str, respectively. A Str^R derivative of NCK2232, designated NCK2272 (Table 1), was generated using natural selection. L. acidophilus NCK2272 was consecutively subcultured for 28 days without selection and the Str^R phenotype was stable over this time period (data not shown). This strain, in combination with a previously isolated Rif^R derivative of the parental NCK1909, designated NCK2300 (Table 1), were utilized in a competitive model of colonization using 129S6/SvEv germ-free mice (Goh & Klaenhammer, 2014). L. acidophilus NCK2300 and NCK2272 were combined in a 1 : 1 ratio and introduced into mice through intragastric gavage. Faecal pellet collection started 72 h following gavage, and on days 5, 7, 10, 14, 21 and 24 thereafter for bacterial enumeration (Fig. 4a). Beginning on day 5 and continuing through to day 24, the NCK2272 mutant showed a statistically significant (approximately 1 log reduction, P<0.05) impairment in its ability to colonize the murine GIT compared with the parent NCK2300 (Fig. 4b). The relative levels of NCK2300 and NCK2272 were also enumerated from caecal and small intestinal tissues following euthanasia on day 24. Similarly, the NCK2272 mutant was recovered from caecal samples at a statistically significant lower level (P<0.05) than NCK2300, and mirrored the lower level of colonization observed during the previous 24 days of faecal collection (Fig. 4c). Although statistically insignificant, NCK2272 was enumerated at lower levels than NCK2300 from the small intestinal tissues. Collectively, these results clearly demonstrated a role for SrtA in the gut retention of L. acidophilus.

Fig. 4.

Co-colonization experiment in germ-free 129S6/SvEv mice. (a) Experimental workflow used to assess the persistence of sortase-deficient L. acidophilus (NCK2272, Str^R) as compared to the parent strain (NCK2300, Rif^R) in a 1 : 1 competitive co-colonization. Naturally antibiotic-resistant strains of each bacterium were combined in a 1 : 1 ratio (day 0). Five germ-free 129S6/SvEv mice were gavaged with the bacterial mixture and stool samples were collected for bacterial enumeration for 24 days on differential selective media. (b) Enumeration of each strain (c.f.u. g⁻¹) was performed by antibiotic selective plating. An asterisk (*) indicates a statistically significant, as determined by the Student t-test (P<0.05), difference between NCK2300 and NCK2272 (ΔsrtA). Each data point represents the mean bacterial cell count (c.f.u. g-1) from five mice faecal samples and the error bars represent 1 SD. (c) Point plot indicating c.f.u. g⁻¹ of NCK2300 (diamonds) and NCK2272 (squares) recovered on antibiotic selective media from both the caecum and small intestine tissue samples following euthanasia on day 24. Horizontal bar represents the mean. An asterisk (*) indicates a statistically significant (P<0.05) difference between the parent and the ΔsrtA mutant.

Discussion

Lactic acid bacteria have a long history of safe consumption, and a subset of these microbes with proposed probiotic properties have been studied for their health-promoting benefits. A current survey of the literature indicates increased interest in understanding the cell surface molecules associated with probiotic action (Bron et al., 2012). The roles of sortase and SDPs have yet to be investigated in L. acidophilus, a commercial probiotic, or the human commensal L. gasseri. In this work, the role of sortase was examined in relation to desirable probiotic phenotypes, such as survival in simulated GIT juices, adherence to mucin and IECs, immunomodulatory capacity and persistence and colonization in a murine model. For both L. acidophilus and L. gasseri, sortase deficiency did not alter survival in simulated GIT juices, which is consistent with previous observations in L. plantarum WCFS1 (Remus et al., 2013). In addition, the ability to adhere to Caco-2 IECs was not affected by the ΔsrtA mutations. On the contrary, in L. salivarius UCC118, the absence of sortase significantly reduced adherence to Caco-2 cells (68 %, P = 0.007) (van Pijkeren et al., 2006). Interestingly, both ΔsrtA mutants of L. acidophilus and L. gasseri exhibited a decreased ability to adhere to porcine mucin, likely due to the loss of SDPs classified as mub proteins and adhesion exoproteins containing mub domains. The difference in the overall mucin composition of the in vitro models used may explain why the deficiency in SDPs did not consistently affect adhesion to both mucin and Caco-2 cells. Mucin 2 (MUC2) is the predominant secreted mucin in the intestinal tract, and Caco-2 cells have recently been shown to exhibit low MUC2 mRNA levels. Rather, Caco-2 cells express high levels of MUC5AC, a mucin type predominantly found in the stomach, pancreas and hepatobiliary system (Bu et al., 2011). We hypothesized that Caco-2 cells may lack specific receptors for the SDPs represented in the strains examined, or other bacterial surface and cellular proteins may compensate binding to epithelial cells when SDPs are not displayed on L. acidophilus and L. gasseri.

It was also observed that DCs co-incubated with L. acidophilus produce greater amounts of IL-6, IL-10 and TNF-α than those co-incubated with L. gasseri. L. acidophilus is a member of the acidophilus complex A homology group of lactobacilli, which are characterized by the presence of a thick, electrostatically associated, outer shell of proteins known as the S-layers. DNA hybridization with specific probes to S-layer protein genes slpA and slpB of L. acidophilus revealed the presence of S-layer genes in other members of the acidophilus complex A, i.e. Lactobacillus crispatus, Lactobacillus amylovorus and Lactobacillus gallinarum (Boot et al., 1996). However, L. gasseri ATCC 33323 does not produce an S-layer. Taxonomically, L. gasseri is grouped in acidophilus complex B with other lactobacilli such as Lactobacillus johnsonii, which also does not express an S-layer. Components of the S-layer of L. acidophilus, specifically SlpA, have been shown to act as a specific ligand for the DC receptor, DC-SIGN (Konstantinov et al., 2008). This particular difference in surface topology is thought to be crucial in the immunomodulatory nature of the strains examined. Meanwhile, although deletion of the srtA gene did not significantly alter the in vitro immunostimulation capacity of L. acidophilus and L. gasseri, both ΔsrtA mutants showed differential immunomodulatory profiles of TNF-α and IL-12, respectively. This may be due to the specific immune signalling properties of the different SDPs expressed by both species. While in general the SrtA protein is conserved amongst lactobacilli, the number and diversity of SDPs varies greatly. For example, lactobacilli encode from six to 37 SDPs depending on the species (Call & Klaenhammer, 2013). In addition, orthologues of the majority of SDPs encoded in L. acidophilus NCFM and L. gasseri ATCC 33323 were observed only among strains of their respective species, highlighting the niche-specific and adaptive nature of surface proteins in commensal and probiotic lactobacilli.

Previous work with a sortase-deficient strain of L. plantarum WCFS1 found no effect on gut retention; however, this was determined using only two mice and in a conventional murine model (Remus et al., 2013). Persistence of the sortase-deficient strain of L. acidophilus NCK2232 was evaluated in a germ-free mouse model. Prior to the co-colonization experiment with the ΔsrtA mutant and the parent strain, each strain was mono-colonized in 129S6/SvEv mice. The ΔsrtA derivative of L. acidophilus NCK2232 persisted for 28 days at levels of approximately 1×10⁹ c.f.u. g⁻¹, mirroring the persistence capacity of the WT strain. Subsequently, competitive co-colonization of the ΔsrtA derivative was pursued, and modelled after challenge studies typically used to assess the ability of probiotic microbes to competitively exclude pathogens. In vivo co-colonization of the L. acidophilus ΔsrtA mutant and the parent strain in germ-free 129S6/SvEv mice resulted in a significant reduction of the ΔsrtA mutant population starting at day 5 through day 24. A similar reduction of the ΔsrtA mutant was also observed in the caecum on day 24, demonstrating a role for SrtA in the competitive persistence of L. acidophilus in the gut environment.

In summary, this work has shown that strains deficient in SrtA impact mucin binding, while leaving the capacity of these strains to adhere to IECs unaffected in vitro. The removal of SrtA was shown to play a role in modulating the immune response of DCs in both L. acidophilus and L. gasseri. When the ability of sortase-deficient L. acidophilus to persist in the GIT of germ-free mice was examined in vivo, the mutant demonstrated an impaired ability to persist in the GIT in the co-presence of the parent population. This study further reinforces the need to closely examine the cell surface display of probiotic lactobacilli, as they have shown to be important components of probiotic functionality, including GIT retention and immunomodulation in the host.

Abbreviations:

BHI medium

brain heart infusion medium

DC

dendritic cell

GIT

gastrointestinal tract

IEC

intestinal epithelial cell

MRS medium

deMan, Rogosa and Sharpe medium

SDP

sortase-dependent protein

SGJ

simulated gastric juice

SIJ

simulated small intestinal juice