Identification of Actinobacillus suis Genes Essential for the Colonization of the Upper Respiratory Tract of Swine

Abstract

Actinobacillus suis has emerged as an important opportunistic pathogen of high-health-status swine. A colonization challenge method was developed, and using PCR-based signature-tagged transposon mutagenesis, 13 genes belonging to 9 different functional classes were identified that were necessary for A. suis colonization of the upper respiratory tract of swine.

Over the last two decades, Actinbacillus suis has emerged as an important pathogen of high-health-status (HHS) swine (30, 38). Although A. suis can persist as a commensal in the upper respiratory tract (URT), under conditions which are poorly understood, it can metastasize and cause septicemia, enteritis, abortion, arthritis, mastitis, pericarditis, myocarditis, and erysipelas-like cutaneous lesions (30). Little is known about the virulence factors of A. suis, but it is hypothesized that homologs of those found in the close relative, Actinobacillus pleuropneumoniae (10, 30), including RTX toxins, urease, and iron-regulated outer membrane proteins might play a role in pathogenesis. Despite the fact that there seem to be many shared virulence determinants, the spectrum of diseases that A. suis can cause and the range of hosts it can affect are different from those of A. pleuropneumoniae, so there are clearly other factors involved (10, 24, 42). In order to begin to understand the mechanism of the switch from benign commensal to invasive pathogen, we set out to first identify the factors of A. suis that are required for colonization of upper respiratory tract of swine, using signature-tagged transposon mutagenesis (STM).

A spontaneous nalidixic acid-resistant (Nal^r) variant of the virulent O2/K2 A. suis strain H91-0380 was used in this study (48, 62). Escherichia coli SM10 λpir (47) was used as the donor strain in the conjugation experiments to deliver the tagged transposons. The A. suis and E. coli strains were routinely grown on blood agar or brain-heart infusion (BHI) plates at 37°C in an atmosphere of 5% CO₂. As required, antibiotics were added to the following final concentrations: kanamycin (Kan), 50 μg/ml; Nal, 25 μg/ml; ampicillin (Amp), 100 μg/ml; bacitracin (Bac), 10 μg/ml; crystal violet (Cv), 1 μg/ml; and chloramphenicol (Chl), 5 μg/ml. For mating experiments, A. suis and E. coli strains were grown in BHI broth with the appropriate antibiotics.

For this work, a new PCR STM system was developed using pLOF/Km (28, 56). To add the oligonucleotide tags, pLOF/Km was partially digested with NotI, blunt-ended, and recircularized, and pLOF/KmdeltaNotI (in which the NotI site upstream of the MluI, SfiI, NotI polylinker was inactivated) was obtained. Twenty-two double-stranded tags (synthesized as 21-mers and their reverse complements; Table 1) were cloned into the KpnI site of pUC18/NotI that had been blunt ended. The NotI fragments of the different tags were ligated with NotI-restricted pLOF/KmdeltaNotI and electrotransformed into E. coli DH5α λpir; the arrangement of all of the different tagged plasmids was confirmed by PCR. The 22 pLOF/Km NotI tags (designated pLOF/Tag1, -2, -3, etc.) were then electroporated into E. coli SM10 λpir and maintained at −70°C in 30% glycerol. E. coli SM10 λpir cells carrying the different pLOF/Tag plasmids were used as donors for the conjugal transposition into A. suis H91-0380Nal^r. Briefly, overnight cultures of the donor and recipient were mixed (200 μl and 600 μl or 100 μl and 700 μl, respectively), harvested by centrifugation, washed with 10 mM MgSO₄, centrifuged, suspended in 50 μl of 10 mM of MgSO₄, and applied as spots onto 0.22-μm filters (Millipore Corporation, Bedford, MA) that had been placed onto nonselective BHI plates with 100 μM IPTG (isopropyl-β-d-thiogalactopyranoside). The plates were incubated at 37°C for 6 to 8 h in 5% CO₂. The mating mixtures were washed off the filters with 5 ml of PBS, pelleted, and suspended in 100 μl of PBS and plated on BHI-Nal/Kan plates. The transconjugants were then “gridded” on blood agar, BHI-Nal/Kan, and BHI-Amp (to identify pLOF/Km cointegrates) plates. Nal^r Kan^r Amp^s A. suis transconjugants were grown overnight in BHI-Nal/Kan broth, frozen in 30% glycerol, and stored in individual vials at −70°C. A total of ∼1,100 transconjugants with 22 different tags were obtained from more than 40 independent conjugation experiments. For each screen, input pools of 18 different transconjugants that had been grown overnight, harvested, and suspended in sterile PBS (∼5 × 10⁹ CFU/ml) were assembled. Five hundred μl/nostril was delivered in drops during the inhalation phase to two 3- to 4-week-old piglets (both male and female, Yorkshire) from the high-health-status herd at the Arkell Swine Research Station, Ontario, Canada. In preliminary experiments, it was determined that A. suis could be recovered from nasal cavities in fairly large numbers, although occasionally the recovery was low. Greater numbers of A. suis cells were always detected from pharyngeal swabs as well as in tonsillar tissues. Forty-eight hours postinfection, the piglets were euthanized and the anterior nasal cavity and pharynx were swabbed. The snout was then cut off at the level of the 3rd premolar, and swabs of cranial nasal cavity and posterior nasal cavity were obtained. Finally, the lingual and palatine tonsils were dissected out. The swabs were streaked on “super-selective” BHI (SSBHI) plates containing nalidixic acid, kanamycin, bacitracin, and crystal violet. The tissues were finely minced and then homogenized in glass tissue grinders in 3 ml of PBS, and 100 μl of each homogenate was likewise plated. All animal procedures were approved by the University of Guelph Animal Care Committee and were in accordance with the guidelines of the Canadian Council for Animal Care.

TABLE 1.

Tag oligonucleotides used in this study

Tag no.	Nucleotide sequencea
1	5′-GTACCGCGCTTAAACGTTCAG-3′
2	5′-GTACCGCGCTTAAATAGCCTG-3′
3	5′-GTACCGCGCTTAAAAGTCTCG-3′
4	5′-GTACCGCGCTTAATAACGTGG-3′
5	5′-GTACCGCGCTTAAACTGGTAG-3′
7	5′-GTACCGCGCTTAATGTAACCG-3′
9	5′-GTACCGCGCTTAATAGGCAAG-3′
10	5′-GTACCGCGCTTAACAATCGTG-3′
11	5′-GTACCGCGCTTAACGCAAAAG-3′
12	5′-GTACCGCGCTTAATCAAGACG-3′
13	5′-GTACCGCGCTTAACTAGTAGG-3′
14	5′-GTACCGCGCTTAACAGCTTAG-3′
15	5′-GTACCGCGCTTAACACTGATG-3′
16	5′-GTACCGCGCTTAAGCACTATG-3′
17	5′-GTACCGCGCTTAATGCACATG-3′
18	5′-GTACCGCGCTTAAGTCGAATG-3′
21	5′-GTACCGCGCTTAACTTGACAG-3′
22	5′-GTACCGCGCTTAAAGACTAGG-3′
23	5′-GTACCGCGCTTAATCGATGAG-3′
25	5′-GTACCGCGCTTAAACCTTGTG-3′
26	5′-GTACCGCGCTTAAGTACGAAG-3′
28	5′-GTACCGCGCTTAATTATGGGG-3′

^aThe consensus 5′ ends have higher ΔGs to optimize the PCRs. The variable 3′ ends of each of the oligonucleotide tags that permit amplification of the specific DNA sequences are indicated in bold.

The mutants recovered on SSBHI plates were harvested in PBS and pooled. Approximately 150 to 200 μl of each cell suspension was pelleted and suspended in 300 μl TE-PCR buffer (10 mM Tris-HCl, pH 7.4, 0.1 mM EDTA), boiled for ∼10 min, and centrifuged for 1 min, and 10 μl of each supernatant was used in 50-μl PCRs (28). The PCR mixtures contained a final concentration of 1× PCR buffer (Invitrogen, Life Technologies, Carlsbad, Calif.), 1.5 mM MgCl₂, 200 μM deoxynucleoside triphosphates (dNTPs), 1 μM of a unique Tag primer, 1 μM of the universal KanR1 primer, and 2.5 U of Taq DNA polymerase (Invitrogen, Life Technologies). The PCR conditions were as follows: 1 min at 95°C, followed by 30 cycles of 95°C for 30 s, 55°C for 1 min, and 72°C for 1 min, with a final extension step of 7 min at 72°C, yielding a PCR product of 500 bp. The resultant fragments were visualized by ethidium bromide staining following electrophoresis in a 1.5% agarose gel.

Two piglets were used to screen each input pool. Mutants that were lost in the first screen were combined and tested in a second screen, again in two animals. Strains that were eliminated from the URT of both pigs in two independent screens (i.e., four animals) were analyzed further by direct sequencing of the inverse PCR products of the regions flanking the transposon as described previously (15), except primer SO69 (5′-TACTCCTGATGATGCATGG-3′) was used in place of TEF69. The sequences were compared to entries in the nonredundant public databases at the National Center for Biotechnology Information (http://www.ncbi.nlm.gov/BLAST) using the BlastX and BlastN programs (3) and assigned to cluster of orthologous groups of proteins (COG) functional categories published at http:/www.ncbi.nlm.nih.gov/COG/.

Putative colonization-defective mutants were further characterized by competing individual mutants with the wild-type strain, A. suis H91-0380 Nal^r carrying pFP12 (34). The competitive index (CI) was calculated as the output ratio of mutant to wild-type bacteria divided by the input ratio of mutant to wild-type bacteria (61). For in vivo studies, the mutant and wild-type strains were grown separately overnight in 20 ml of BHI broth with appropriate antibiotics and the next day the bacteria were pelleted and suspended in PBS to the same optical density at 600 nm (OD₆₀₀) (∼2.8). Two piglets were inoculated intranasally with 2× 500 μl of a 1:1 mixture of the mutant and wild type. In order to calculate the input ratio, 10-fold dilutions of each were plated to determine the exact number of CFU. The pigs were sacrificed 48 h postinfection, and samples were collected and processed as described above. The colonies recovered from lingual and palatine tonsils were used to calculate the output ratio. To determine the in vitro CI, mutant and wild type A. suis strains were first grown on blood agar and then harvested in sterile BHI broth and the suspensions were adjusted to the same OD₆₀₀. Ten-fold dilutions of the strains were plated to calculate the exact input ratio, and 0.2 ml of each strain was used to inoculate 20 ml of BHI/Nal broth. The cultures were grown for ∼7 h (stationary phase), and then dilutions of the mixed culture were plated on BHI and BHI-Kan plates. The number of CFU obtained on BHI-Kan plates (mutant) was subtracted from the number obtained on BHI plates (mutant and wild type) to calculate the number of wild-type A. suis cells. The in vitro CI was calculated as described above.

Of 770 mutants screened for their ability to colonize the URT of piglets, 97 independent mutants (40 rRNA gene mutants and 57 independent mutants of 30 different genes) were found to be colonization deficient. This rate of loss of ST mutants was higher (12.6%) than has been reported in many other studies (2 to 5%) (e.g., references 26 and 45). It is speculated that competition with commensal organisms (streptococci, enterococci, and staphylococci were detected) together with a relatively small number of colonization sites available in tonsil might be a reason for this high loss.

Although transposition by Tn10 is considered “random” (27), the consensus sequence 5′-NGCTNAGCN-3′ has been shown to be a preferred substrate sequence recognized by the Tn10 transposase (8). In this STM screen, 40 independent colonization defective mutants were identified with Tn10 insertions in rRNA genes (Table 2). The targets for most of these insertions in the 16S rRNA genes were 5′-AGCTAAGCA-3′ and 5′CGCACTGTT-3′; in the 23S rRNA genes, Tn10 integration was guided by the sequence 5′-CACTAAGCT-3′. Fuller et al. (15) also found a large number of transposon insertions in the rRNA genes in the related organism, A. pleuropneumoniae. In the present study, multiple independent insertions were also seen in cpxR (2), hns (5), pomA (2), aceE (3), lpdA (2), nifU (3), nrdA (3), guaA (2), tRNA-Tyr (3), PM1911 (6), and PM1778 (7). Taken together, these data suggest that although Tn10 has proven to be useful in the study of Actinobacillus spp. (10, 15), it alone cannot be used to do a saturating screen.

TABLE 2.

Putative colonization-essential genes of Actinobacillus suis

Mutanta	Nearest homolog species	% Similarity (span)b	Product	Putative function	CIin vivoc	CIin vitroc
Transcription (COG 1D K)
2-43 (+1)d	cpxR, Pasteurella multocida	60 (166)	CpxR	Response regulator to stress, pilus biogenesis	0.28	0.83
3-39	hupA, Pasteurella multocida	79 (72)	HupA	Global transcriptional regulator, DNA-binding protein Huα	0.29	0.66
3-31 (+4)	hns, Pasteurella multocida	54 (106)	H-NS	DNA-binding, global regulator	0.18	4
15-8	YP0456, Yersinia pestis	63 (65)	Putative right-origin DNA-binding protein	Global regulator	0.35	0.37
Outer membrane, cell wall biogenesis (COG ID M)
23-58 (+1)	pomA, Mannheimia haemolytica	77f (378)	Outer membrane protein A	Attachment, structural integrity	0.07	0.5
1-28	pfhaB1, Pasteurella multocida	96 (31 nt)	Putative filamentous hemagglutinin	Attachment	0.17	0.17
22-100	lcbB, Neisseria meningitidis	34 (64)	Putative bifunctional polymerase	Cell envelope biogenesis	0.66	0.66
Amino acid metabolism and transport (COG ID E)
4-4	VCA0160, Vibrio cholerae	73 (38)	Tryptophan-specific transport protein	Channeling of tryptophan into cell membrane	1	1.5
4-21	lmo0845 (∼yxjH), Listeria monocytogenes (∼Bacillus subtilis)	54 (211)	YxjH protein	Methionine biosynthesis	0.12	1.6
Energy production and conversion (COG ID C)
2-50	atpG, Haemophilus influenzae	67 (227)	ATP synthase gamma chain	Pump H+ ions, synthesize ATP, maintain electrophysiology of the cell, protect cells from oxidative and acidic stress	0.67	2.1
3-26	atpC, Pasteurella multocida	76 (30)	ATP synthase epsilon chain	Pump H+ ions, synthesize ATP, maintain electrophysiology of the cell, protect cells from oxidative and acidic stress	NTe	0.31
18-18 (+2)d	aceE, Haemophilus influenzae	88 (118)	Pyruvate dehydrogenase, E1 component	Catalyze oxidative decarboxylation of pyruvate to acetyl coenzyme A	NT	0.02
18-15 (+1)d	lpdA, Haemophilus influenzae	95 (147)	Dihydrolipoamide dehydrogenase	Catalyze oxidative decarboxylation of pyruvate to acetyl coenzyme A	0.41	0.64
22-47 (+2)d	nifU/iscU, Haemophilus influenzae	87 (125)	IscU	Fe-S assembly, iron homeostasis in eukaryotes	0.001	0.5
Lipid metabolism (COG ID I)
3-52	est, Pasteurella multocida	42 (124)	Esterase	Lipolytic enzyme, hydrolyzes lung surfactant	NT	0.66
Carbohydrate transport and metabolism (COG ID G)
26-75	ptsH, Pasteurella multocida	89 (19)	PtsH	Phosphocarrier protein, transport of sugars	NT	0.48
Nucleotide metabolism (COG ID F)
23-48	thyA, Haemophilus influenzae	74 (81)	Thymidylate synthetase	Thymine nucleotide metabolism	0.12	0.33
18-21 (+2)d	nrdA, Rickettsia conorii	35 (281)	NrdA, ribonucleoside-diphosphate reductase alpha chain	Catalyzes conversion of ribonucleotides to corresponding deoxyribonucleotides	NT	0.36
26-80 (+1)d	guaA, Pasteurella multocida	100 (81)	GMP synthase	Purine biosynthesis	0.3	1.6
Translation, ribosomal structure, and processing (COG ID J)
18-8	tRNA-Gly3, Haemophilus influenzae	97 (175 nt)	tRNA-glycine	Addition of glycine amino acid to the polypeptide chain	0.02	1.2
23-27 (+2)d	tRNA-Tyr, Haemophilus influenzae	97 (175 nt)	tRNA-tyrosine	Addition of tyrosine amino acid during protein synthesis	0.15	0.33
22-29	tufB, Haemophilus influenzae	96 (175 nt)	Elongation factor TuB	Processing factors for protein synthesis	NT	0.28
16-2	tufA, Pasteurella multocida	75 (136)	Elongation factor TuA	Processing factors for protein synthesis	NT	0.45
15-26	rpl9, Haemophilus influenzae	49 (81)	50S ribosomal protein L9	Proper folding and configuration of rRNA	NT	1.7
23-37	rpl32, Haemophilus influenzae	94 (38)	50S ribosomal protein L32	Proper folding and configuration of rRNA	NT	0.34
3-44 (+18)	HirrnC23S, Haemophilus influenzae	83 (164 nt)	23S rRNA	Protein synthesis	NT	2.0
3-45 (+20)	rrnA 16S, Actinobacillus equuli	99 (261 nt)	16S rRNA	Protein synthesis	2.3	0.81
Function unknown COG ID S
4-2 (+5)d	PM1911, Pasteurella multocida	63 (158)	Unknown	Conserved hypothetical (putative metal-binding protein)	0.01	0.93
3-54 (+6)d	PM1778, Pasteurella multocida	64 (73)	Unknown	Unknown	0.4	0.66
17-21	PM1568, Pasteurella multocida	75 (68)	Unknown	Unknown	1	1.1
3-17	PM0682, Pasteurella multocida	92 (28)	Unknown	Unknown	1	1.92
17-66	No significant match in the database		Unknown	Unknown	NT	1.3

^aMutants lost from both animals in two independent screens.

^bPercent similarity determined using BLASTX (span in amino acids, except tRNA and rRNA genes indicated in bp).

^cCalculated as (mutant CFU/wild-type CFU recovered)/(mutant CFU/wild-type CFU_input).

^dNumber of independent insertions in the same reading frame.

^eNT, not tested.

^fBased on sequence of entire gene (1,122 bp).

In this STM screen, nine mutants that carried mini-Tn10 insertions in the genes encoding transcriptional regulatory proteins were lost. Among these, two independent mutants harbored Tn10 integration in a homolog of cpxR, a gene which encodes the response regulator protein of the cpxA-R two-component regulatory system. The CpxR mutant, 2-43, grew slightly less well than the wild type in vitro (CI_{in vitro} = 0.83) and was markedly attenuated in vivo (CI_{in vivo} = 0.28). In uropathogenic E. coli, the CpxA-R system is involved in the P pilus biogenesis and is thought to regulate the expression of other virulence factors, including hemolysin and cytotoxic necrotizing factor (22). Attenuation of the cpxR ST mutant was likely due to effects on the expression of more than one gene, but further work will be required to identify the CpxR regulon in A. suis.

Global regulatory proteins, such as HU, integration host factor (IHF), and H-NS (histone-like proteins), and their functional equivalents in other genera are known to bind DNA and play an important role in gene regulation (36). Several A. suis mutants of hns, hupA, and a protein carrying a helix-turn-helix (HTH) sequence were colonization defective in initial screens. Although the hns mutant 3-31 appeared to grow much better in vitro (CI_{in vitro} = 4.0), it was greatly attenuated in vivo (CI_{in vivo} = 0.18). The reason(s) for the more robust growth in vitro is not immediately obvious; however, in E. coli, it is known that although H-NS acts as a positive regulator in vivo, it behaves as a negative regulator in vitro (36). Several studies have concluded that this protein is a transcriptional repressor and a translation inhibitor (43, 52) and growth of the wild type in vitro might have been repressed with respect to the A. suis hns mutant. H-NS plays a vital role in cellular response to changes in the environment and controls the expression of many different genes in E. coli, Salmonella enterica serovar Typhimurium, and Shigella spp. (20, 63).

Another colonization-deficient mutant carried an insertion in a hupA gene homolog. In competition assays, it grew somewhat more slowly in vitro (CI_{in vitro} = 0.66) but was more clearly attenuated in vivo (CI_{in vivo} = 0.29). HupA mutants are asynchronous in DNA replication initiation and have been associated with attenuated colonization of S. enterica serovar Typhimurium in chicken alimentary tract (60).

A mutant with insertion in a gene whose product is 63% similar to a putative DNA-binding protein of Yersinia pestis with an HTH motif also was unable to colonize the URT. Many proteins with HTH motifs bind to the major groove of the DNA helix (36) and some, such as the VirB protein of Shigella flexneri, are key regulators of virulence gene expression (7). Given the low and comparable CI_{in vitro} and CI_{in vivo} (0.37 and 0.35), loss of this mutant may, however, have been a function of slow growth.

In this screen, mutants with insertions in three different genes related to the cell surface components (COG ID-M) were found to be colonization deficient. The first of these genes was an ortholog of the pomA gene of Mannheimia hemolytica (ompA gene in E. coli). Two independent colonization-deficient mutants carried mini-Tn10 inserted at the 33rd bp of the ompA_A.s. gene, immediately upstream of the sequence GGCTTAGCA. Growth of the ompA mutant, 23-58, was somewhat attenuated in vitro (CI_{in vitro} = 0.5) and more markedly attenuated in vivo (CI_{in vivo} = 0.07). The ompA gene of A. suis was cloned, and based on the entire gene sequence, OmpA_A.s. was found to share 70% identity and 77% similarity with PomA of the bovine respiratory pathogen M. hemolytica. The OmpA protein of E. coli, which is the best characterized of these outer membrane proteins, is known to be vital for structural integrity of the cell membrane and for the generation of normal cell morphology. In addition, OmpA_E.c. recognizes many ligands on eukaryotic target molecules, including glycoprotein motifs on human brain microvascular endothelial cells and is implicated in the invasion of these cells (39). The OmpA homolog of nontypeable Haemophilus influenzae (NTHi), OmpP5, plays an important role in the initiation of infection by mediating binding to the nasopharyngeal mucin-sialic acid oligosaccharides (41) and attachment to the epithelial cells of eustachian tubes (32). As well, attenuation due to a mutation in an OmpP5 homolog has also been observed in an STM screen of A. pleuropneumoniae (15). Further studies will be needed to determine whether the inability of the A. suis ompA mutant is due to a failure to bind to the tissues of the URT or an inability to survive the environment of the URT.

Another mutant that did not survive within porcine nasopharyngeal tissues had a mutation in a gene that shared 96% identity at the nucleotide level (but only 31 nucleotides) with the extreme 5′ end of the filamentous hemagglutinin (FHA) gene, pfhaB1, of Pasteurella multocida. The role of FHA in attachment of Bordetella pertussis to respiratory cilia has been extensively documented (58). In an STM screen of P. multocida, Fuller et al. (14) found an fha mutant that was markedly attenuated. In view of the fact that the region of homology was very short and there was no difference between the competitive indices in vitro and in vivo (0.17), caution is needed in speculating on its role in A. suis colonization at this point.

A mutant of lcbB, whose product might play a role in the biosynthesis of cell envelope carbohydrates, was also found to be colonization defective. In STM and other in vitro and in vivo infection studies (23, 46), mutants bearing transposon insertions in polysaccharide biosynthesis clusters have been reported to be unable to survive in the host. Alhough there have been several studies reporting the involvement of polysaccharides in adherence and colonization of members of the family Pasteurellaceae (23), given the limited homology, caution is also needed in interpreting the role of the lcbB mutant.

Two genes involved in amino acid transport and biosynthesis were identified in this screen. The first of these genes is a homolog of the lmo0845 gene product of Listeria monocytogenes, which in turn, is a homolog of yxjH of Bacillus subtilis. In B. subtilis, this gene appears to function in sulfate assimilation and methionine biosynthesis and is a part of the S-box regulon (5). In this study, the growth of wild type A. suis was better in vivo compared to the yxjH mutant, while the mutant grew slightly better in vitro. A second mutant with an insertion in a gene encoding tryptophan-specific transport protein homolog grew equally well as the wild type when competed individually in vivo, so its role in colonization remains unclear.

Ten mutants with mutations in energy production and conversions (COG ID-C) were identified in the preliminary screens. Five transconjugants were lost during in vivo selection due to disruption of the pyruvate dehydrogenase (PDH) enzyme operon. In these strains, there were insertions in sequences that had high homology with the pyruvate dehydrogenase subunits, E1 (aceE) and E3 (lpdA) genes of H. influenzae and P. multocida respectively. The PDH cluster catalyzes the irreversible oxidative decarboxylation of pyruvate to acetyl coenzyme A, linking glycolysis with the tricarboxylic acid cycle (16). Its role in pathogenesis has been demonstrated in a number of species. For example, lpdA mutants of Streptococcus pneumoniae and Pseudomonas aeruginosa are avirulent in mice and rats, respectively (13, 50). In S. enterica serovar Typhimurium, the expression of lpdA is induced in response to the bactericidal proteins of neutrophils (40). In A. pleuropneumoniae, lpdA has also been identified as a virulence associated gene (15). Given the low and comparable CI_{in vitro} and CI_{in vivo} of the lpdA mutant (0.64 and 0.41) and very low CI_{in vitro} of the aceE mutant (0.02), it is possible that mutation of PDH genes may have caused a global effect on the cell physiology, leading to the clearance of A. suis ST strains.

Disruption of the atp operon also gave rise to colonization-defective A. suis mutants. In strain 3-26, Tn10 inserted in a gene homolog of atpC, while another attenuated mutant harbored the transposon in atpG, whose product is the gamma chain of ATP synthase. In in vivo competition assays, the wild-type A. suis was recovered in 15-fold-greater numbers than the atpG mutant, while in vitro, the mutant cells outnumbered the wild type (CI_{in vitro} = 2.1). ATP synthase gene products function in energy production and are crucial for maintaining the electrical potential of the cell membrane and resistance to acid stress (44). Their role in virulence and survival has been shown in S. enterica serovar Typhimurium (17, 59), Listeria monoctogenes (44), P. multocida (14), and A. pleuropneumoniae (15). We speculate that the atp mutants of A. suis could not maintain cytoplasmic buffering in the face of oxidative and osmotic stresses in the respiratory tissue environments, though the reason for the robust growth of the atpG mutant and somewhat weaker growth of the atpC mutant in vitro is not obvious.

Three colonization-deficient mutants had Tn10 insertions in the iscU gene, which is involved in iron-sulfur (Fe-S) assembly, an essential constituent of many apo-enzymes. As well, the isc operon has a critical role in sensing intracellular iron and oxidant levels (57). Agar and coworkers have shown that the IscU protein provides a scaffold for IscS-directed assembly of clusters, which are then used for the maturation of apo-Fe-S proteins (1). In the current study, an iscU mutant was found to be severely attenuated in vivo (CI = 0.001).

The est and ptsH gene products are known to have a role in virulence (6, 18, 64). In this study, A. suis mutants carrying Tn10 within the est and ptsH genes were eliminated; however, their attenuation was not verified individually and their in vitro CIs were less than 1.

Mutants with mutations of three nucleotide metabolism genes (COG ID-F) were eliminated from the URT in initial screens. A 10-fold reduction in growth of a guaA mutant was seen in the in vivo competition assay, while in BHI medium, the mutant outcompeted the parent strain. The enzyme encoded by guaA, GMP synthase, is involved in the terminal steps of the purine biosynthetic pathway, where it converts XMP to GMP (33). Attenuation due to mutation in the guaAB operon has been reported in members of Enterobacteriaceae (33), as well as in P. multocida and A. pleuropneumoniae (14, 46), and a number of these auxotrophic mutants have been proposed for use as live attenuated vaccines.

A thyA mutant was also found to be colonization deficient, but its growth in vitro was also much less than that of the parental strain. thyA-deficient strains are unable to produce dTTP for DNA synthesis and can grow on minimal medium only in the presence of thymine or thymidine (2). The thyA gene is reported to be critical for the survival under these conditions, while deficiency of many other metabolic genes may only cause biostasis (2). In Shigella flexneri, mutation of the thyA gene leads to greatly reduced virulence (11).

Three independent nrdA mutants were colonization defective. The nrdA gene encodes ribonucleotide reductase, which catalyzes conversion of ribonucleotides to the corresponding deoxyribonucleotides for DNA synthesis and repair (9) and is important for survival under oxidative stress (49). In BHI medium, the mutant grew less well than the wild type (CI_{in vitro} = 0.36) and since its in vivo survival was not compared with the wild type, its role in colonization of A. suis has not been confirmed.

Gene products related to translation processing (COG ID-J) are fundamental for growth and stability of an organism and are needed in high concentrations (25). As rapid growth is a prerequisite for colonization (31, 55), proteins must be synthesized at a high rate. In addition, there are reports of further functions of tRNA genes in peptidoglycan synthesis (53), posttranslational addition of amino acids to proteins (29), and interaction with the host (21). In this study, a number of colonization-deficient mutants had Tn10 insertions in genes related to protein synthesis, including those for tRNA-glycine 3 and tRNA-tyrosine 1. In competition assays, the tRNA-Gly3 mutant was recovered in much lower numbers from the host (CI_{in vivo} = 0.02), while it grew equally well as the wild type in vitro (CI_{in vitro} = 1.2). In contrast, the tRNA-Tyr1 mutant was equally attenuated in both in vivo and in vitro competition assays. In an STM study of A. pleuropneumoniae, mutants of tRNA genes, tRNA-Leu and tRNA-Glu were found to be attenuated (15). The elongation factors (TuA and TuB) and ribosomal protein genes were also identified in this screen; however, their attenuation was not confirmed separately in vivo.

In these experiments, a large number of ST mutants with insertions in the 16S and 23S rRNA genes appeared to be colonization deficient. The E. coli genome carries seven rRNA (rrn) operons, P. multocida has six, and A. suis possesses at least four 16S rRNA operons (based on Southern hybridization using a 16S rRNA gene probe; data not shown). Asai and coworkers showed that E. coli requires all of the operons for optimal adaptation to changing physiological conditions (4). These operons are under the control of growth rate-dependent promoters, and expression of RPs and translation factors are linked to the cellular concentration of rRNA (55). These reports indicate a selective advantage for the growing organism, and deletion of rRNA gene might have adverse effect, especially under in vivo conditions. However, when the 16S rRNA mutant 3-45 was competed with the wild type, it was recovered at a higher rate from the host than the wild type (CI_{in vivo} = 2.3), despite the fact that its CI_{in vitro} was somewhat less than 1. Although it is hard to reconcile this finding, when CIs are done with single mutants such results are not uncommon (e.g., references 19 and 46). Also surprisingly, the CI_{in vitro} of the 23S rRNA mutant 3-44 was 2.0. The loss of a very large number of A. suis rRNA mutants with inserts in several different sites suggests the need for further analysis of the role of ribosomal genes in pathogenesis.

Although the genome sequencing of many pathogens has been completed, the functions of a large fraction (∼ 40%) of the predicted genes are unknown (COG ID-S). In STM screens, one or more hypothetical genes are often reported as having a role in virulence (37). In the current study, mutants of four hypothetical genes were identified as colonization defective (Table 2). Of these four mutants, one (PM1911) was markedly attenuated in vivo (CI_{in vivo} = 0.01). Although the CI_{in vivo} of mutant 17-66, which did not share any similarity with any known gene was not determined, its CI_{in vitro} was greater than 1.

In this study, four of the mutants grew as well as or better than the wild type when competed individually in vivo, these mutants do not appear to be essential for A. suis colonization (i.e., those bearing Tn10 insertions in the tryptophan transport gene homolog, two unknowns, and in a 16SrRNA gene); and three had similar in vivo and in vitro CIs (i.e., lcbB, pfhaB1, and a global regulator gene) (Table 2). These mutants were, however, eliminated in both animals in two screens (i.e., in four animals) and, except for the unknowns, their role in the pathogenesis is known in other pathogens (7, 23). The fact that CI_{in vivo} values do not reflect the results of the initial screen is reported in many STM studies (e.g., references 19 and 46), which points to a weakness of the STM method.

In the present study, mutants carrying insertions in the metabolic genes estA, ptsH, and nrdA genes and the mutants with Tn10 in the genes associated in the translation process, such as elongation factor genes tufA and tufB, and ribosomal protein genes rpl9 and rpl32, were also lost. These are all known to be highly expressed genes that play an important role in the survival and virulence of several different pathogens (12, 35, 51, 65), but they were not competed with the wild type individually, so the nature of their colonization defect remains to be confirmed. Although the CI_{in vitro} of tufA, tufB, and rpl32 was less than 1, the rpl9 mutant was clearly not attenuated in vitro.

To our knowledge, this is the first report of transposon mutagenesis of A. suis. For these experiments, 22 uniquely tagged Tn10 transposons that could be amplified by PCR were constructed. Although the problem of “hot spots” is acknowledged, these transposons should be useful for the study of other Pasteurellaceae. As well, a nasopharyngeal colonization challenge system was developed that, again, should be useful for the study of a variety of organisms. Using this challenge system, we were able to demonstrate for the first time that the principal colonization sites of A. suis in swine were likely the nasopharynx and lingual and palatine tonsils.

In the current STM screen, 28 genes belonging to 9 different COGs were identified that appeared to be essential for A. suis colonization; 13 were later confirmed to be colonization defective in CI experiments. The homologs of many of these genes, e.g., genes of the atp and PTS complexes, ompA, pfhaB1, lcbB, hupA, hns, tufA, tRNA genes, lpdA, and guaA have also been identified in genome-wide screens of other mucosal organisms, and some of these have been well established as virulence genes of important pathogens (14, 15, 22, 46, 54). The response regulator cpxR was identified for the first time in an STM screen. In addition to cpxR, nrdA, and nifU, four unknown genes and the ribosomal genes rpl9 and rpl32 were identified for the first time in a genome-wide virulence screen. Experiments to identify the receptors for, and/or the precise function of, some of these colonization factors are currently under way.