A Natural Mouse Model for Neisseria Colonization.

Commensals are important for the proper functioning of multicellular organisms. How a commensal establishes persistent colonization of its host is little understood. Studies of this aspect of microbe-host interactions are impeded by the absence of an animal model. We have developed a natural small animal model for identifying host and commensal determinants of colonization and of the elusive process of persistence. Our system couples a commensal bacterium of wild mice, Neisseria musculi, with the laboratory mouse. The pairing of a mouse commensal with its natural host circumvents issues of host restriction. Studies are performed in the absence of antibiotics, hormones, invasive procedures, or genetic manipulation of the host. A single dose of N. musculi, administered orally, leads to long-term colonization of the oral cavity and gut. All mice are healthy. Susceptibility to colonization is determined by host genetics and innate immunity. For N. musculi, colonization requires the type IV pilus. Reagents and powerful tools are readily available for manipulating the laboratory mouse, allowing easy dissection of host determinants controlling colonization resistance. N. musculi is genetically related to human-dwelling commensal and pathogenic Neisseria and encodes host interaction factors and vaccine antigens of pathogenic Neisseria Our system provides a natural approach for studying Neisseria-host interactions and is potentially useful for vaccine efficacy studies.

INTRODUCTION

Commensals (i.e., microbiota) play a critical role in the physiology of multicellular organisms. They are required for the homeostasis of many bodily processes, and they participate in gut and immune system development and prevent pathogen colonization. Perturbations in these microbial communities are strongly linked to obesity, inflammatory bowel disease, diabetes, and autoimmunity (1–7).

The mechanisms underlying host and commensal determinants of persistent colonization are little understood. The majority of commensals cannot be cultured or manipulated genetically (8–10). Because of host restriction barriers, few animal models provide a natural setting for probing commensal-host interactions. Neisseria, a genus of Gram-negative betaproteobacteria, provides an opportunity to develop a natural small animal model for this purpose.

The genus Neisseria contains a large number of genetically related species (11). The vast majority of these are commensals of hosts ranging from rodents, canids, and bovines to nonhuman primates and humans (12–17). Neisseria gonorrhoeae and Neisseria meningitidis are the only two species that cause disease. These pathogens, which infect only humans, also behave like commensals in that they have a tendency to colonize asymptomatically (18–20). Commensal Neisseria spp. are little studied, and there are no small animal models for colonization. Several mouse models have been developed for pathogenic Neisseria infection, but due to the strict tropism of N. gonorrhoeae and N. meningitidis for humans, they are necessarily heterologous systems that require invasive procedures; antibiotics; hormones; direct administration of human homologous proteins, such as transferrin; and/or the use of transgenes expressing human proteins (21–24).

We recently isolated a new species of commensal Neisseria, Neisseria musculi, from the oral cavity (OC) of healthy wild mice (17). N. musculi is easily cultured and manipulated in vitro and is genetically related to other Neisseria. With the aim of developing a small animal model for Neisseria colonization, we determined whether N. musculi could be paired with inbred laboratory mice. We report that N. musculi colonizes the oral cavity and gut of laboratory mice for at least 1 year without causing disease. Long-term colonization is achieved with a single oral dose. Using this model, we discovered that permissiveness to N. musculi colonization is strongly influenced by host genetics and by innate, but not adaptive, immunity. For N. musculi, colonization requires its type IV pilus (Tfp). Finally, we present evidence that N. musculi encodes homologs of host interaction factors and vaccine antigens found in pathogenic Neisseria spp. and that it expresses one of the vaccine targets, capsular polysaccharide. We discuss the power of our natural small animal model to broaden our knowledge of commensal and pathogenic Neisseria biology and of host components that restrict/permit colonization.

RESULTS

N. musculi colonizes the oral cavities and guts of mice in a mouse strain-specific manner.

We isolated N. musculi from the oral cavity of a wild mouse, Mus musculus domesticus (17). Our repeated attempts to culture Neisseria from the oral cavities of inbred mice from Jackson Laboratory and Taconic were unsuccessful. Since inbred laboratory mice do not harbor Neisseria, this provided an opportunity to test the susceptibility of these animals to N. musculi colonization.

The Collaborative Cross (CC) is a powerful new tool in mouse genetics that allows the linkage of alleles with phenotypic traits (25). We tested N. musculi on selected CC founder strains. These strains include 5 conventional, widely used inbred strains and 3 wild-derived inbred strains from distinct M. musculus subspecies (Table 1). The wild-derived strains are CAST, wild mice trapped in Thailand belonging to a distinct subspecies, Mus musculus castaneous; PWK, trapped in the Czech Republic and belonging to the subspecies Mus musculus musculus; and WSB/EiJ (WSB), trapped in Maryland, USA, and belonging to M. musculus domesticus. The conventional inbred strains are chimeras with varying degrees of genetic relatedness to CAST, PWK, and WSB, although their genetic origins are overwhelmingly M. musculus domesticus (25).

TABLE 1

Susceptibility of Collaborative Cross founder strains to colonization by N. musculi

Strain	No. colonized/inoculated (%)a	P valueb
CAST/EiJ	35/40 (87)
A/J	26/28 (92)	NSc
C57BL/6J	12/23 (52)	<0.006c
NOD/LtJ	0/4 (0)	<0.0004c
NZO/HILtJ	0/4 (0)	<0.0004c
PWK/PhJ	0/9 (0)	<105c
WSB/EiJ	0/9 (0)	<105c
129S1/SvImJ	0/4 (0)	<0.0004c
MyD88−/−	21/21 (100)	<0.0014
RAG-1−/−	3/14 (21)	NSd

Mice were scored for the presence of N. musculi in the oral cavity and fecal pellet each week for 3 months.

χ2 with Yates correction for small numbers and Bonferroni for multiple pairwise comparisons. NS, not significant.

Compared to CAST.

Compared to WT BL/6.

The mouse inoculation protocol is shown in Fig. S1 in the supplemental material. Prior to inoculation, the presence of Neisseria in these animals was determined by plating OC and fecal pellet (FP) samples on selective agar. The mice used in this study have always been culture negative. The next day, AP2365, a naturally occurring rifampin-resistant (Rifr) rough variant of N. musculi, was gently pipetted into the OCs of the animals, and N. musculi counts in OCs and FPs were determined weekly for 3 months by plating samples on selective agar. CAST/EiJ (CAST) and A/J mice were very susceptible to colonization (Table 1): the OCs and FPs of 35/40 (87%) CAST mice and 26/28 (92%) A/J mice were continuously culture positive. C57BL/6J (B6) mice (12/23; 52%) were partially resistant to colonization. In contrast, NOD, NZO, PWK, WSB, and 129S1 mice were highly resistant at the same infectious dose by the same route of inoculation.

A representative colonization experiment, involving inoculation of 10 CAST mice, is shown in Fig. 1. The number of N. musculi CFU in the OC and FP quickly reached a plateau and remained steady thereafter, indicating the commensal had adapted to these niches and that replication and turnover had reached equilibrium. Generally, when N. musculi was cultured from the OC, it was also recovered from the FP. Samples reisolated from colonized mice were N. musculi, as judged by multilocus sequence typing of 51 ribosomal genes (rMLST) of 10 OC and 10 FP colonies recovered from CAST mice at 5 weeks postinoculation (data not shown). All N. musculi bacteria reisolated from the OC and FP had the rough colony phenotype, like the inoculation strain.

FIG 1

N. musculi colonizes the oral cavity (A) and gut (B) of CAST mice and different sections of the gastrointestinal tract (C). The samples in panels A and B are from the same experiment; each mouse was assigned a unique color. The samples in panel C are taken from 3-month-colonized CAST mice from a different experiment. The plots indicate geometric means with geometric standard deviations (SD). LOD, limit of detection; Int, intestine.

Two colonized CAST mice were followed long term. N. musculi was continuously recovered from their OCs and FPs for 52 weeks (see Fig. S2 in the supplemental material). Colonized B6 mice yielded similarly high N. musculi counts weekly for 52 weeks (data not shown).

Throughout our studies, all inoculated and uninoculated mice remained healthy: none lost weight, and all maintained healthy coats and normal activity. At necropsy (performed by D. Beselson, Director, University of Arizona Animal Care Facility), the organs of the 52-week-colonized mice resembled those of healthy mice.

N. musculi was not cultured from the peripheral blood of 4 CAST and 4 A/J mice 4 h or 28 days postinoculation. Although this experiment did not address whether N. musculi enters the bloodstream, the result suggests the commensal did not survive at this site.

Taken together, these results demonstrate that the susceptibility of a mouse to N. musculi colonization is strongly influenced by its genetic background. In susceptible mouse strains, N. musculi easily colonizes the OC and gut and persists in these niches for lengthy periods without causing disease.

N. musculi colonizes the entire gastrointestinal tract in mice.

We examined the location of N. musculi in the gastrointestinal tracts of 3-month-colonized CAST mice. The stomachs, small intestines, large intestines, and ceca of necropsied animals were flushed with sterile saline to remove the luminal content, and the tissues were homogenized and plated on selective agar. N. musculi was recovered from all sampled sections of the gut (Fig. 1C). It was impossible to sample organs from the same animal on successive days or to determine whether N. musculi populations in these organs were self-sustaining. However, the large numbers of N. musculi bacteria recovered from tissue-associated gut samples long after inoculation strongly suggest the commensal was not simply in transit from the OC.

To determine whether N. musculi could be horizontally transmitted, we cohoused 2 colonized CAST mice with 3 naive CAST or B6 mice for 12 weeks. None of the uninoculated mice became colonized. To determine whether the endogenous flora influenced colonization, we cohoused 4 B6 and 4 CAST mice for 12 weeks before inoculation. This did not alter the colonization susceptibility of either mouse. Moreover, CAST and B6 mice bred in house or purchased from the Jackson Laboratory were always colonized at the same frequency. Although these experiments involved small numbers of mice, the evidence suggests that the preexisting flora did not play a significant role in determining colonization susceptibility; final evidence awaits fecal transplant studies. To determine whether in vivo passage of N. musculi would increase its colonization efficiency, we inoculated 4 naive B6 mice with N. musculi isolated from the OC of a persistently colonized CAST mouse. This in vivo passage did not alter N. musculi colonization efficiency. Taken together, these results suggest that neither housing conditions nor the endogenous microbiota is a significant roadblock to N. musculi colonization.

Innate immunity determines susceptibility to N. musculi colonization.

The partial resistance of B6 mice to N. musculi colonization (Table 1) provided an opportunity to investigate the role of the immune system in determining colonization susceptibility. N. musculi was assayed in two strains of immunodeficient B6 mice: B6-MyD88−/− mice, which lack the MyD88 adaptor that mediates signaling through many Toll-like receptors, and B6-Rag-1−/− mice, which lack T and B cells and cannot mount an adaptive immune response (Table 1). Strikingly, MyD88−/− mice were exquisitely susceptible to N. musculi colonization, unlike the B6 parental strain (21/21 MyD88−/− mice colonized versus 12/23 B6 mice; P < 0.001). MyD88−/− mice also had higher N. musculi burdens than the parental wild-type (WT) strain (P = 0.0026) (Fig. 2). In contrast, Rag-1−/− mice were no more susceptible than WT B6 mice. These results indicate that the innate, but not the adaptive, immune system is a major determinant of N. musculi colonization. The increased numbers of N. musculi bacteria recovered from MyD88−/− mice compared to WT B6 mice suggest that the innate response plays an ongoing role in controlling N. musculi numbers.

FIG 2

MyD88−/− mice have higher N. musculi burdens than parental B6 mice. N. musculi CFU in oral swabs taken from B6 and MyD88−/− mice are shown (n = 9 or 10 mice/group). The bars indicate means with SD. Significance was determined using Student's t test on the average burden per strain over the lifetime of the experiment. The data are representative of 2 independent experiments.

N. musculi colonization requires the type IV pilus.

To test the usefulness of our model for studying commensal determinants of colonization, we focused our attention on the Tfp. All Neisseria species have a complete set of Tfp biogenesis genes (16, 17, 26). In the case of pathogenic Neisseria, Tfp is implicated in promoting colonization, based on experiments using cultured human cells and a limited number of human challenge studies (27–31). The function of Tfp has never been tested in a natural animal model.

For this experiment, a nonpiliated mutant of N. musculi, ΔpilE, was constructed by deleting the gene encoding the Tfp fiber subunit; a complemented strain, ΔpilE::pilEWT-C10, was also constructed. The piliation statuses of the ΔpilE and complemented strains were validated by several methods. Unlike the WT and complemented strains, ΔpilE did not produce pilE mRNA, as judged by reverse transcription (RT)-PCR (see Fig. S3 in the supplemental material). N. musculi ΔpilE exhibited phenotypes characteristic of nonpiliated mutants: it was defective in DNA transformation (see Table S1 in the supplemental material) and attached less well to surfaces (Fig. 3). These results indicate N. musculi ΔpilE does not produce the Tfp fiber. Finally, the growth of ΔpilE was examined. The WT, ΔpilE, and complemented strains grew equally well (see Fig. S4 in the supplemental material). The slightly lower optical density at 600 nm (OD600) of ΔpilE cultures was not statistically different at any time point; it likely reflects the slight tendency of ΔpilE cells to aggregate in liquid culture.

FIG 3

N. musculi ΔpilE is defective in attachment (A) and biofilm formation (B). ΔpilE::ΔpilE, complemented strain. (B) Statistical analysis was performed in GraphPad Prism 7 by one-way analysis of variance (ANOVA) with Tukey's multiple-comparison test. ***, P < 0.001; *, P < 0.05. No significant difference was detected between the WT and complemented strains.

N. musculi ΔpilE was defective in colonizing the OC and gut in CAST and B6 mice compared to the WT and complemented strains (Fig. 4; see Table S3 in the supplemental material) (P < 0.0001 for WT versus ΔpilE for both CAST and B6 mice). The few OC and FP reisolates were N. musculi, as judged by pilE sequencing (primers NP246F and NP246R2), and their mutated pilE locus was unaltered (data not shown). The complemented ΔpilE::pilEWT-C10 strain colonized the OC and FP of CAST mice like WT N. musculi (OC, P = 0.3316; FP, P = 0.9916). In B6 mice, the colonization behavior of the complemented strain did not fully revert to that of WT N. musculi, even after taking into account the partial colonization resistance of B6 mice (OC, P = 0.0013; FP, P = 0.0008). We cannot explain this behavior. During the transformation/recombination process that inserted the WT pilE sequence into the ΔpilE strain, a mutation may have occurred elsewhere in the genome that affected the colonization behavior of the complemented strain. The transformation/recombination process may have had a polar effect on a gene immediately downstream of the complemented pilE site. In the annotated N. musculi genome, pilE is at the end of the contig; the identity of the downstream gene is unknown. Other explanations are also possible, but we note that this colonization behavior of the complemented strain is observed only in the B6 genetic background and not in the CAST background. Finally, we note that in these experiments, the rough variant of N. musculi (WT, ΔpilE, and complemented strains) was used, and all reisolated N. musculi strains exhibited the rough colony phenotype.

FIG 4

N. musculi ΔpilE is defective in colonizing the oral cavity (A) and gut (B) of CAST and B6 mice. AP2365 ΔpilE::pilEWT-C10, pilE complemented strain. Each N. musculi strain was assayed in 10 mice. Oral-swab and fecal samples from the same mouse were assigned the same color.

N. musculi encodes host interaction factors and vaccine candidates of human-dwelling Neisseria.

Finally, we determined whether N. musculi could be used to model human-dwelling species of Neisseria. To date, Neisseria colonization studies have focused almost exclusively on the two pathogens N. gonorrhoeae and N. meningitidis. Great efforts have been made to identify host interaction factors with the goal of identifying vaccine antigens capable of stimulating protective immune responses. Chief among these vaccine development efforts has been the use of reverse vaccinology to identify genome-derived Neisseria antigens (GNAs) in N. meningitidis (32). Currently, similar work is being conducted to identify vaccine antigens in N. gonorrhoeae (33, 34).

Many homologs of pathogenic Neisseria host interaction factors and candidate vaccine antigens, including GNAs, were found in N. musculi and human-dwelling commensal Neisseria strains (Tables 2 and 3). Two GNAs with high identity and query coverage values were GNA1220, a membrane protein of unknown function containing a stomatin-like domain; and GNA33, a membrane-associated lytic transglycosylase required for cell separation (35, 36). N. meningitidis GNA1946 and the N. gonorrhoeae ortholog, NGO2139, which are methionine-binding subunits of ABC transporters, both retrieved the same N. musculi ortholog with greater than 75% identity and 95% query coverage. GNA1946 and NGO2139 (MetQ) induce the production of serum bactericidal antibodies (32, 37). N. musculi also has a homolog for N. meningitidis LpdA, a high-molecular-weight protein, P64k, that is very immunogenic and is used frequently as a carrier protein for weaker immunogens (38, 39).

TABLE 2

Putative orthologs of protective antigens encoded in the AP2031 genome

Protein query	Query accession no.	Query species	Ortholog
			N. musculi		N. polysaccharea		N. lactamica		N. cinerea		N. subflava		N. oralis		N. mucosa		N. elongata		N. bacilliformis
			% id.a	% qc.b	% id.	% qc.	% id.	% qc.	% id.	% qc.	% id.	% qc.	% id.	% qc.	% id.	% qc.	% id.	% qc.	% id.	% qc.
LctP	CBA04244	N. meningitidis	25	95	25	95	98	100	92	100	92	100	92	100	94	100	82	99	40	42
LpdA	CAA57206	N. meningitidis	74	100	74	100	85	100	84	100	85	100	84	100	84	100	71	100	72	100
GNA1030c	NP_274064	N. meningitidis	48	23	29	79	95	88	90	100	74	100	86	100	91	86	65	88	62	100
GNA1220	NP_274245	N. meningitidis	81	99	81	99	93	100	93	100	34	31	82	98	84	100	75	99	78	96
GNA1946	NP_274940	N. meningitidis	77	95	77	95	88	100	85	100	79	97	82	97	82	98	80	96	67	96
GNA2091c	NP_275079	N. meningitidis	63	79	63	79	88	100	80	100	86	79	67	99	85	80	63	79	65	79
GNA33	NP_273099	N. meningitidis	77	85	77	85	91	100	87	100	78	100	73	99	77	100	74	85	67	94
NadAc,d	NP_274986	N. meningitidis	43	15	43	15	27	30	67	28	NDe	NDe	38	15	69	46	48	7	35	27
PorA P1c,d	NP_273150	N. meningitidis	53	97	53	97	86	100	85	100	71	98	59	98	71	100	61	99	56	99
ExbB	NP_274732	N. meningitidis	68	99	68	99	96	100	93	99	73	99	73	99	76	99	58	99	57	98
GNA992	NP_274028	N. meningitidis	48	23	48	23	91	94	39	79	62	23	64	14	65	29	63	67	64	18
GNA2001	NP_274993	N. meningitidis	68	65	68	65	65	100	60	100	60	97	56	100	56	97	79	98	76	57
GNA1870 (fHbp)c,d	NP_274866	N. meningitidis	30	38	30	38	34	61	91	100	39	83	31	62	38	82	28	89	29	64
NspA	NP_273705	N. meningitidis	42	86	42	86	84	87	48	86	45	86	29	36	40	43	47	86	45	86
TBP2	CAA55541	N. meningitidis	25	13	25	13	74	100	35	98	26	8	28	17	31	12	26	10	28	47
TbpA	AAF81744	N. meningitidis	30	54	30	54	94	100	75	100	26	58	28	61	30	71	31	67	32	54
GNA2132 (NHBA)c,d	NP_275117	N. meningitidis	NDe	ND	ND	ND	75	100	40	40	32	25	34	9	35	51	32	29	28	34
GNA1162	NP_274189	N. meningitidis	33	40	33	40	95	100	88	100	48	13	35	14	54	99	48	35	52	10
PilC1	YP_207232	N. gonorrhoeae	36	68	36	68	46	100	40	89	40	65	36	67	38	66	24	65	26	31
PilQ	YP_207267	N. gonorrhoeae	56	100	56	100	81	100	77	100	61	100	57	99	58	100	49	97	49	97
AniA	YP_208345	N. gonorrhoeae	79	79	79	79	93	79	92	79	87	79	80	79	89	79	80	79	78	79
OpaD	YP_208563	N. gonorrhoeae	32	69	32	69	66	100	29	86	32	85	30	19	23	16	32	89	29	85
OpcA	CAB45007	N. gonorrhoeae	28	16	28	16	40	64	41	83	25	76	20	76	31	15	26	82	23	38
LptD	YP_208748	N. gonorrhoeae	60	98	60	98	91	98	81	100	63	98	61	98	61	98	55	100	55	92
BamA	YP_208831	N. gonorrhoeae	75	100	75	100	90	100	94	100	82	100	80	100	82	100	71	100	69	100
TamA	YP_208979	N. gonorrhoeae	68	90	68	90	95	100	83	98	77	89	74	88	75	88	63	89	62	89
NGO2054	YP_209073	N. gonorrhoeae	64	76	64	76	94	100	78	100	62	100	57	81	60	57	54	100	62	74
NGO2139 (MetQ)	YP_209148	N. gonorrhoeae	78	95	78	95	90	100	78	100	82	97	86	97	85	98	82	96	77	82

%id., percent identity. Boldface indicates >50% identity.

%qc., percent query coverage. Italics indicate >75% query coverage.

Component of rMenB-OMV vaccine (Novartis).

Component of the Bexsero and Trumemba vaccines (GlaxoSmithKline/Novartis and Pfizer).

ND, significant similarity not detected.

TABLE 3

Orthologs of N. meningitidis capsule synthesis, transport, and translocation proteins and presence of selected capsule transcripts in N. musculi

Protein query	Query accession no.	Query species	Maximum identity (%)a	Query coverage (%)b	Genome annotation	mRNAc
CssA	WP_002233375.1	N. meningitidis	72	96	UDP-N-acetylglucosamine-2-epimerase	+
CssB	WP_002233374.1	N. meningitidis	87	99	UDP-N-acetyl-d-mannosamine dehydrogenase	ND
CssC	CCP19843.1	N. meningitidis	28	13	Capsule polymerase	ND
CtrA	NP_273135	N. meningitidis	56	93	Capsule transport complex	+
CtrB	NP_273136	N. meningitidis	67	91	Capsule transport complex	ND
CtrC	NP_273137	N. meningitidis	73	100	Capsule transport complex	ND
CtrD	NP_273138	N. meningitidis	84	98	Capsule transport complex	ND
CtrE	NP_273145	N. meningitidis	60	93	Capsule translocation	+
CtrF	NP_273146	N. meningitidis	61	99	Capsule translocation	+

Boldface, sequence identity > 50%.

Italics, query coverage > 75%.

+, positive; ND, not determined.

We also conducted BLAST searches using three β-barrel-containing outer membrane proteins as queries: N. meningitidis NspA, a factor H ligand, and N. gonorrhoeae adhesins OpaD and OpcA. (The N. meningitidis OpcA ortholog is a lectin capable of interacting with vitronectin [40, 41]). NspA and OpaD retrieved the same N. musculi homolog (NspA, 42% identity, 86% query coverage; OpaD, 32% identity, 69% query coverage). The bulk of shared identity in these proteins localized to the β-barrel strands. OpcA did not have a significant hit.

Capsular polysaccharide is a target of several meningococcal vaccines (42). BLAST searches using N. meningitidis capsule proteins showed that N. musculi has genes for capsule biosynthesis, transport, and translocation proteins (Table 3). With the exception of the putative capsule polymerase (CssC), all the capsule-related proteins have high sequence homology with their N. meningitidis orthologs (≥56% identity and ≥75% query coverage).

N. musculi expresses a polysaccharide capsule.

We determined whether N. musculi produces a capsule by using two biochemical tests, India ink and alcian blue staining, that are widely used to detect capsulated organisms (43). After India ink treatment, N. musculi cells (smooth and rough variants) were surrounded by a clear halo against a dark background, which is indicative of capsulated organisms (Fig. 5A). India Ink stained cells of capsulated N. meningitidis 8013 similarly, but not those of the unencapsulated N. meningitidis FAM2 (Fig. 5A). To further confirm that the refractile zone of these cells corresponds to capsular polysaccharide, we stained extracts from the cells with alcian blue (44). The results indicated that a high-molecular-weight alcian blue-reactive smear was present in the capsulated N. meningitidis 8013 and N. musculi AP2365 smooth and rough variants but not in the unencapsulated N. meningitidis FAM2 (Fig. 5B).

FIG 5

N. musculi (Nmus) produces a capsule. (A) India ink staining of N. musculi smooth (S) and rough (R) strains, capsulated (Cps+) N. meningitidis (Nme) strain 8013, and unencapsulated (Cps−) strain FAM2. The cells were counterstained with crystal violet. (B) Alcian blue staining of lysates of these strains separated by SDS-6% PAGE.

Finally, we determined whether capsule genes are transcribed in N. musculi by means of RT-PCR of selected capsule biosynthesis, transport, and translocation genes (see Fig. S5 and Table S2 in the supplemental material). Transcripts for ccsA, ctrA, ctrE, and ctrF were detected using this method (Table 3). Taken together, these results indicate that the capsule genes in N. musculi are expressed.

DISCUSSION

We have developed a genetically tractable small animal model for identifying host and microbial determinants of colonization and persistence. The system pairs the laboratory mouse with a commensal of wild mice, N. musculi, which is closely related to human-dwelling species of Neisseria (17) (Tables 2 and 3) (see below). The protocol does not require antibiotics, hormones, invasive procedures, or genetic manipulation of the animal. A single oral dose of N. musculi results in long-lasting colonization of the oral cavity and gut of the mouse (see Fig. S2 in the supplemental material). All animals were healthy throughout the study.

Using this model, we showed that host genetics and innate immunity strongly control susceptibility to colonization by N. musculi (Table 1). Reagents and tools are readily available for the laboratory mouse, allowing easy dissection of host components that restrict/permit N. musculi colonization. The mice in this study were founder strains of the Collaborative Cross, a powerful new tool that can be used to link genetic traits with biological phenotypes. As susceptibility of these strains to N. musculi colonization ranges from sensitive to highly resistant, the Collaborative Cross will allow us to identify host alleles that determine colonization resistance.

We also used the model to examine the role of the N. musculi type IV pilus in colonization. All Neisseria strains and many bacteria belonging to other genera express Tfp. In the cases of the pathogens N. meningitidis and N. gonorrhoeae, cell culture experiments and a small number of human challenge studies strongly imply a role for the Tfp in colonization (27–31). Here, we corroborate these findings, providing in vivo proof that the N. musculi Tfp is an important colonization determinant (Fig. 4). In vitro studies have identified other, more subtle activities of the N. gonorrhoeae Tfp, including the reprogramming of the host transcriptional profile and activation of immune signaling pathways (45, 46). Our model provides the first opportunity to identify the in vivo endpoints of these activities, as well as the functions of host interaction factors held in common between N. musculi and human commensal Neisseria (Tables 2 and 3).

The animal models currently in use to study N. gonorrhoeae and N. meningitidis are heterologous systems that pair a mouse with a human-specific pathogen (21–23). These approaches limit the ability to utilize the full breadth of mouse genetic techniques available for studying host determinants of persistent colonization. Although N. musculi does not cause disease, it does encode many pathogenic Neisseria host interaction factors and candidate vaccine antigens (Tables 2 and 3). Indeed, N. musculi expresses one of these candidate vaccine antigens, capsular polysaccharide (Fig. 5). Our model will be a useful tool for characterizing the in vivo functions of these host interaction factors and is potentially useful for evaluating vaccine candidates for the pathogens.

The fact that N. musculi colonizes the gastrointestinal tract of laboratory mice should not be a surprise. N. musculi colonizes the oral cavity of wild mice and is detected in their guts (17), and Neisseria species have been detected in animal feces (15). Human niches for Neisseria are generally assumed to be the nasopharynx (N. meningitidis, N. gonorrhoeae, and commensal species), genital tract (N. gonorrhoeae and occasionally N. meningitidis), and rectum (N. gonorrhoeae), but to our knowledge studies have not been done to determine the presence of Neisseria in the human gut, either by direct culture or molecular species identification. The large numbers of N. musculi organisms recovered from the oral cavities and guts of mice over a long period indicate the organism is able to adapt to a variety of environments within the animal. Taken together, these observations suggest that Neisseria is a more successful and adaptable organism than had previously been suspected.

There is currently a great interest in the microbiome. In spite of the large number of papers on the subject, little is known about how changes in the microbiome are brought about. Conspicuous by their absence are data concerning the acquisition of a new commensal in the presence of an existing microbiota or after antibiotic treatment. Our model opens the door to these investigations.

MATERIALS AND METHODS

Generation of the rifampin-resistant N. musculi strain.

AP2365, a naturally occurring Rifr rough variant of the N. musculi type strain (17), was isolated by plating AP2031 (AP2031T) on GCB (Becton Dickinson) agar containing rifampin (50 mg/liter).

Mouse strains.

All inbred mouse strains and Collaborative Cross parental strains were obtained from the Jackson Laboratory (Bar Harbor, ME). All animal protocols were approved by the University of Arizona IACUC.

Mouse inoculation protocol.

Mice were rested in the University of Arizona mouse facility for 2 weeks before inoculation. The inoculation protocol is shown in Fig. S1 in the supplemental material. To determine the presence of Neisseria species in the indigenous flora of the animals, the oral cavities of the mice were swabbed using the BD BBL CultureSwab Plus Transport System (Fisher Scientific); the swabs were suspended in GCB medium base (Becton Dickinson) plus Kellogg's supplements I and II (27), and dilutions of the suspensions were plated on GCB agar containing vancomycin (2 mg/liter) and trimethoprim (3 mg/liter). Bacteria on the plates were counted after incubation for 48 h at 37°C, 5% CO2. Neisseria has never been recovered from mice before inoculation. Fecal pellets of the mice were suspended and processed similarly. On the day of inoculation, AP2365 was swabbed from an agar plate and resuspended in phosphate-buffered saline (PBS) at an OD600 of 2.0. Inbred mice were manually restrained, and 50 μl of the bacterial suspension was pipetted into the oral cavity. The oral cavities of the inoculated mice were swabbed weekly or biweekly. Swab suspensions in GCB medium base (Becton Dickinson) were plated on GCB agar containing rifampin (40 mg/liter), and the plates were incubated for 48 h at 37°C and 5% CO2.

Verification of N. musculi in oral-swab suspensions.

Samples from each colony growing on GCB-rifampin agar were used for verification of N. musculi as described previously. Briefly, internal transcribed spacer (ITS) primers specific to sequences that are highly conserved among Neisseria species were used for colony PCR (17). The ITS sequences of sample isolates were compared to that of the type strain, AP2031, for species validation and were found to be identical.

Construction of N. musculi ΔpilE and its complemented strain.

Table S2 in the supplemental material lists the primers used for strain construction. In AP2365 ΔpilE, the pilE open reading frame was replaced with a kanamycin resistance cassette. Primers IM011F and IM012R, containing flanking sequences for the pilE gene in N. musculi AP2031T. were used to amplify the Kan resistance cassette from plasmid pNBNeiKan (17) (synthesized by GenScript). The amplified DNA was purified and transformed into WT N. musculi AP2031T by spot or liquid transformation as described previously, and transformants were selected on GCB agar containing Kellogg's supplements I and II (27) and Kan (50 mg/liter). The ΔpilE::kan locus in AP2031T was transferred to the rifampin-resistant N. musculi strain AP2365 as follows. Primers NP246F and NP246R2 were used to amplify ΔpilE::kan from AP2031T, and the amplified DNA was cloned into pGEMT (Promega). The recombinant plasmid DNA was introduced into AP2365 by spot transformation. Transformants were selected on GCB agar containing supplements I and II and Kan (50 mg/liter). The ΔpilE::kan locus in AP2365 was confirmed by Sanger sequencing of PCR products generated with primers NP246F and NP246R2.

The complemented strains AP2365ΔpilE::pilEWT-C10 and AP2365ΔpilE::pilEWT-C21, independent clones, were constructed as follows. Primers IM013 and IM014 were used to amplify the chloramphenicol (Cm) resistance cassette from plasmid pLES94 (47). Primers MR485 and MR486 were used to amplify the WT pilE locus in AP2031T. The Cm PCR product was digested with Pacl and EcoRV (New England BioLabs), and the pilE PCR product was digested with Afel and l KpnI (New England BioLabs). The two digested DNAs were ligated into similarly digested pUC19 (New England BioLabs) using T4 ligase (New England BioLabs). Primers IM0015 and IM0016 were used to amplify the pilE::cm region in the recombinant plasmid, and the amplified DNA was cloned into pGEMT (Promega). DNA from the resulting plasmid was electroporated into AP2365 ΔpilE::kan to replace the mutated pilE locus. Transformants were selected and maintained on GCB agar containing supplements I and II and chloramphenicol (2.5 mg/liter). The pilE loci in the complemented strains were confirmed by Sanger sequencing of the PCR products generated with primers NP246F and NP246R2.

Transformation assays.

DNA transformations were performed as described previously (17). Briefly, the recipient strains AP2365, AP2365ΔpilE, and AP2365ΔpilE::pilEWT-C10 were grown for 16 h at 37°C on GCB agar containing supplements I and II and the appropriate selective antibiotic(s). Bacterial cells were suspended in GCB broth containing MgSO4 (5 mM). Thirty microliters of each suspension, previously diluted to an OD600 of 1.5, was added to 0.2 ml of liquid GCB containing MgSO4 (5 mM) and 1 μg of chromosomal DNA from N. musculi strain AP2093, a naturally occurring isolate whose rpsL gene contains a point mutation conferring resistance to streptomycin (17). Following incubation at 37°C for 20 min, the bacteria were added to 2 ml of liquid GCB containing supplements I and II and incubated at 37°C and 5% CO2 for 4 h. Transformants were enumerated by plating cells onto GCB agar containing supplements I and II and streptomycin (100 μg/ml), and the total input bacteria were enumerated by plating an equal volume on supplemented GCB agar without antibiotics.

RNA extraction, cDNA synthesis, and RT-PCR.

Bacterial cells were grown to mid-log phase in GCB broth containing supplements I and II, and total RNA was extracted using TRIzol (Invitrogen) according to the manufacturer's instructions. Contaminating DNA was removed using DNA-free (Ambion). The quality and amount of RNA were determined by spectrophotometry (NanoDrop; Thermo Scientific). For RT-PCR, 1,000 μg of RNA was used to generate the first strand, using Moloney murine leukemia virus (MMLV) reverse transcriptase (Promega) according to the manufacturer's instructions. This was followed by a PCR using GoTaq green master mix (Promega). N. musculi pilE was amplified using primers MR489 and MR490. N. musculi 16S rRNA was amplified using primers MR493 and MR494. N. musculi ctrA was amplified using primers IM017 and MR018. N. musculi cssA was amplified using primers IM019 and IM020. N. musculi ctrE was amplified using primers IM021 and IM022. N. musculi ctrF was amplified using primers IM023 and IM024. The primer sequences are listed in Table S2 in the supplemental material.

Growth curves.

Bacterial cells were grown for 16 h at 37°C and 5% CO2 on GCB agar containing supplements I and II and the appropriate selective antibiotics. Cells were scraped from the plates, suspended in supplemented GCB, and diluted to an OD600 of 0.05; 2 ml of each bacterial sample was added to 60-mm dishes and incubated at 37°C and 5% CO2. Bacterial density was measured every 2 h for 10 h using a Beckman Coulter (Brea, CA) DU730 spectrophotometer. The cell density at each time point was expressed by subtracting the OD600 value at time zero from the OD600 value at the time of collection.

Adherence assay.

A static biofilm assay adapted from that of Merritt et al. (48) was used to measure adherence. Briefly, 2 ml supplemented liquid GCB was added to each well of a 6-well dish (Corning), and 1 × 107 CFU of N. musculi WT, ΔpilE, or the complemented strain was introduced into the wells. The plates were incubated at 37°C and 5% CO2 for 16 h. Each well was gently washed 3 times with 1 ml sterile PBS. Any residual wash buffer was forcibly shaken from the plate to remove all planktonic bacteria. One milliliter of 0.1% crystal violet was added to each well, and the plate was incubated for 30 min at room temperature. The excess dye was removed, and all the wells were washed with 10 ml of PBS. Retained crystal violet was solubilized by the addition of 1 ml of 30% glacial acetic acid, and the OD550 was measured on a Beckman Coulter (Brea, CA) DU730 spectrophotometer. Three fields were imaged before the initial washes and after crystal violet staining. The results were representative of three independent experiments performed in technical triplicate.

BLAST searches.

Tblastn searches were conducted using TBLASTN 2.7.1+ (49, 50). Protein query sequences from N. meningitidis and N. gonorrhoeae were used to search N. musculi strain AP2031's genome sequence (PubMLST identifier [ID], 29520 [51]). Many N. meningitidis queries used for the analysis were retrieved from the Protegen protective antigen database (52). The accession numbers for the commensal human-dwelling Neisseria genome data used for BLAST searches were as follows: Neisseria polysaccharea ATCC 43768, NZ_ADBE00000000; Neisseria lactamica 02-06, NC_014752; Neisseria cinerea ATCC 14685, NZ_ACDY00000000; Neisseria subflava NJ9703, NZ_ACEO00000000; Neisseria oralis CCUG 26878, PubMLST ID 19091; Neisseria mucosa ATCC 25996, NZ_ACDX00000000; Neisseria elongata ATCC 29315, NZ_CP007726; Neisseria bacilliformis ATCC BAA-1200, NZ_AFAY00000000.

India ink stain and light microscopy.

N. meningitidis capsulated strain 8013 and unencapsulated strain FAM2 and N. musculi AP2365T were suspended in India ink (BD Diagnostic) and spread as thin films on a microscope slide (43). After the films were allowed to air dry, the bacteria were counterstained with crystal violet (Gibson) for 1 min. The slides were gently rinsed with water and examined under a light microscope at ×100 magnification.

Capsule extraction.

Capsule was extracted as described previously (44). N. meningitidis capsulated strain 8013 and unencapsulated strain FAM2 and N. musculi AP2365 rough and smooth variants were grown on GCB agar for 17 to 18 h at 37°C and 5% CO2. Cells were suspended in PBS to an OD600 of 0.8, and 1 ml of the suspension was pelleted by centrifugation (10,000 × g; 5°C) for 2 min. The pelleted cells were resuspended in 0.5 ml PBS and incubated at 55°C for 30 min to allow release of capsular material. The bacteria were pelleted again, and the supernatants were concentrated 10-fold in an Amicon Ultra centrifuge filter with a 10,000-molecular-weight cutoff. Capsular material was separated by SDS-6% PAGE, stained with the cationic dye alcian blue (0.125% alcian blue in 40% ethanol-5% acetic acid; Sigma) for 2 h, and destained overnight in 40% ethanol-5% acetic acid.