Signature tagged mutagenesis in the functional genetic analysis of gastrointestinal pathogens

Abstract

Signature tagged mutagenesis is a genetic approach that was developed to identify novel bacterial virulence factors. It is a negative selection method in which unique identification tags allow analysis of pools of mutants in mixed populations. The approach is particularly well suited to functional genetic analysis of the gastrointestinal phase of infection in foodborne pathogens and has the capacity to guide the development of novel vaccines and therapeutics. In this review we outline the technical principles underpinning signature-tagged mutagenesis as well as novel sequencing-based approaches for transposon mutant identification such as TraDIS (transposon directed insertion-site sequencing). We also provide an analysis of screens that have been performed in gastrointestinal pathogens which are a global health concern (Escherichia coli, Listeria monocytogenes, Helicobacter pylori, Vibrio cholerae and Salmonella enterica). The identification of key virulence loci through the use of signature tagged mutagenesis in mice and relevant larger animal models is discussed.

Introduction

Pathogenic bacteria have developed a myriad of molecular mechanisms to invade and establish infection within the host. Elucidating the mechanisms of pathogenesis associated with infectious agents is essential in order to design rational drug treatments and vaccination strategies. Since virulence genes are either directly or indirectly associated with various stages of the infection process, their inactivation may lead to complete loss of virulence or a decrease in the level of infection. One of the main tools to identify virulence factors is to use gene disruption strategies, such as random transposon mutagenesis and to analyze the individual mutants that carry a single gene deletion. An alternative method to the slow and time-consuming analysis of individual mutants is provided by signature tagged mutagenesis (STM).¹

STM is based on random transposon mutagenesis, but allows parallel analysis of mutant strains of the pathogenic bacteria of interest (Fig. 1). In STM each mutant is tagged with a unique DNA sequence in such away that all tags can be co-amplified from the DNA of a mixed population of mutants by a single PCR.^1,2 The original tag consisted of a short DNA sequence of 40 bp that was flanked by two invariant arms of 20 bp. The region between the variable and invariable region contained a restriction enzyme site that could be used to release the arms from the central regions following amplification and labeling therefore allowing for tag specific probes to be generated. The original method to detect the signature tags was by dot blot but over the years there have been many methods for detection of signature tags (PCR, polymorphic tag-length transposon mutagenesis). STM was initially developed to identify virulence genes in Salmonella enterica serovar typhimurium but has subsequently been used in many screens in other bacterial species as well as in the yeast Saccharomyces cerevisiae, the fungus Cryptococcus neoformans and the parasite Toxoplasma gondii.^1-10

Figure 1. (A) Overview of how DNA tags are transformed into the transposon plasmid and used to create a bank of individually tagged mutants. (B) Differently tagged mutants are pooled in the input pool (IP) and grown overnight these are then used to infect the animal model of interest (mouse) and mutants that survive the infection are recovered from the organs of interest and used as the output pool (OP). (C) Detection of IP and OP pools can be performed by different methods the original technique was dot blots. The DNA tags are used to create a hybridization probe for IP and OP and the results are compared to detect non-colonizing mutants. A newer method of detecting non-colonizing mutants is by PCR. The DNA tags are amplified by using a primer for the plasmid and a tag-specific primer. Once again the IP and OP are compared to identify non-colonizing mutants.

For any bacterial pathogen there are several critical parameters that must be followed in order to ensure an efficient STM screen in vivo. First, the transposon chosen should insert randomly into the chromosome, a property which varies depending on the transposon. It was previously noted that the Tn917 transposon system used in Listeria monocytogenes has tendency for hot-spots, while this is not the case with a recently developed marnier transposon system pJZ037.^11-14 Second, the pool size must be determined with respect to the inoculum dose (normally this ranges between 48–96 mutants per pool). Finally the route of administration and the infecting dose must be established and the best time frame for when to evaluate a possible attenuation in virulence.

The main advantages of this system compared with other classical gene inactivation methods (targeted or random) is that STM is a negative selection screen allowing the discovery of virulence genes without prior knowledge of their nature or function.¹⁵ Secondly, as a large number of mutants can be screened in tandem (up to 96), this method is in principal much faster and more exhaustive in identifying virulence factors compared with standard transposon systems. The disadvantage of this system is that it is limited to finding non-essential genes (i.e., genes not required for growth in broth). Furthermore some DNA tags are unable to be amplified from chromosomal DNA of bacteria after recovery from the animal host. This reason for this is not known but it can result in loss of reproducibility and the identification of false negatives. This can be overcome by re-organizing the candidates in a new pool that is then re-screened in the animal. While this may be time-consuming it reduces the number of false-negative candidates. Another drawback encountered with STM is the issue of bias, revealed by the fact that previously published screens have repeatedly found the same candidates.¹⁵ The bias present in a STM screen could be due to many different factors such as; the choice of animal model, mode of infection and target organ. This redundancy may also be due to the fact that transposons do not insert absolutely at random. Finally, in vivo STM screens do not discriminate between genes that are essential for growth in cellular models to those that are dispensable.

Despite these experimental difficulties STM has been successfully applied to many different pathogens as well as pathogenic fungi, S. cerevisiae and toxoplasmasa.^2,3,5-10,15 In this review we will focus on STM screens associated with gastrointestinal (GI) infections (Table 1). GI illness is a global public health concern and in the developed world it is usually self-limiting but has a considerable economic burden.^16-18 In the developing world GI infections, particularly diarrheal disease, are still the third most common cause of death in children under 5 years of age.¹⁹ An example of such a disease is Salmonella enterica serovar Typhi infection (typhoid fever), which results in more than 2 million infections a year leading to approximately 200,000 deaths.²⁰

Table 1. Overview of STM studies in GI bacteria.

Bacteria	Animal Model	Inoculation route	Pool size	No of attenuated mutants	Reference
L. monocytogenes	Mouse: Liver, spleen	IP	96	2000	96
	Mouse: Brain	IV	48	2000	15
	Mouse: Liver, spleen	IP	96	2000	100
	Mouse: Liver	IV	48	4176	4
E. Coli	Calf: Feces, Colon	Oral	95	570	47
	Calf: Feces	Oral	95	2850	46
C. rodentium	Mouse: Colon	Oral	48	576	49,50
	Mouse: Colon	Oral	48	576	48
H. pylori	Gerbil: Stomach	Intra-gastric	24	960	26
C. jejuni	Chicken: Ceca	Oral	14	700	59
	Chicken: Cecal	Oral	82	246	61
S. enterica	Pig: Ilea mucosa	Oral	95	1045	81
	Calf: Ilea, Liver	Oral	36	180	110
	Spleen, MLN
	Pig: MLN	IN	48	960	111
	Chicken: Spleen	Oral	12	1152	112
	Calf: Ilea	Oral	95	190	80
	Chicken: Ceca	Oral	95	285	80
	Pig: MLN	Oral	45	45	113
	Mouse; Spleen	IP	1000	1000	86
V. cholerae	Mouse: Small intestine	Oral	48	1100	29
V. cholerae	Mouse: Small intestine	Oral	96	9600	37

IP, intraperitoneal; IV, intravenous; IT, intratracheal; IN, intranasal; MLN, mesenteric lymph nodes

Bacterial pathogens have developed several intricate systems to evade detection by the immune response and to circumnavigate the stresses they may encounter in the GI tract (pH, osmotic stress, bile and acid stress).^21-23 Following ingestion, the first physical stress encountered by the bacterium is the low pH of the stomach (pH 2), followed by the increased osmolarity of the upper small intestine (equivalent to 0.3 M NaCl) and in the duodenum, the antimicrobial activity of the biological detergent bile (1 L of bile is produced in the liver, stored interdigestively in the gall bladder and secreted into the duodenum each day).²⁴ STM is an important tool to aid our understanding of how bacterial pathogens can survive such physical and chemical barriers and cause both localized and systemic infection in the human host.

Helicobacter pylori

H. pylori represents an important pathogen that colonizes the gastric mucosa and results in an acute inflammatory response by damaging the gastric epithelium. Inflammation can progress to several disease states, ranging from superficial gastritis, chronic atrophic gastritis, peptide ulceration to associated lymphoma and gastric cancer.²⁵H. pylori has been designated a class I carcinogen by the World Health Organization (WHO). STM provides an important tool to understand the genes necessary for specific adaption of the pathogen to the gastric mucosa.²⁶ Using an STM approach Kavermann and colleagues screened 960 independent transposon mutants for their ability to colonize the gerbil stomach 21 d post-infection. In total they identified 47 genes that are essential for gastric colonization.²⁶ A large proportion of the genes crucial for colonization consisted of genes associated with bacterial chemotaxis, motility, adhesions and LPS production.²⁶ One interesting outcome from this study was the identification of a secreted collagenase vital for colonization. The gene hp0169 has sequence homology with PrtC from Porphyromonas gingivalis, which functions as a Ca²⁺ dependent metallo-protease with collagenolytic activity. The mutant strain produced half as much proteolytic activity as the wild-type and was unable to degrade collagen.²⁶ Collagen type I and III are important components of the extracellular matrix of the stomach epithelium. Furthermore type I collagen is present in the area around gastric ulcers and is important for the process of ulcer healing.²⁷ Kavermann and colleagues suggest that the secretion of a collagen degrading enzyme by H. pylori could be responsible for the persistence of gastric or duodenal ulcerogenesis and the delayed healing process.

Vibrio cholerae

Cholera is an acute diarrheal disease which is characterized by discharge of voluminous rice water stool caused by toxigenic Vibrio cholerae strains.²⁸ The pathogen enters the host through the oral route of infection, transits the gastric acid barrier of the stomach and colonizes the small intestine. Once established within this niche the bacteria begin to produce the cholera toxin (CT), which is responsible for the diarrheal disease that is characteristic of cholera.²⁸ The ensuing diarrheal disease can lead to death by dehydration within hours of infection. V. cholerae O1 and O139 serogroups producing cholera toxin (CT) are mainly responsible for cholera outbreaks that can cause havoc in highly populated regions in Asia, Africa and Latin America.²⁸ V. cholerae O1 is further divided into El Tor and classical biotypes. The ongoing pandemic of cholera that started in 1961 is caused by O1 El Tor biotype.²⁸

An STM bank was constructed in V. cholerae El Tor strain C6709–1 to identify genes essential for colonization in an infant mouse model of infection.²⁹ In total 1,100 mutants were screened in pools of 48 mutants and after 21–24 h the small intestines were removed and used to identify non-colonizing mutants. The initial screen identified 51 strains attenuated for colonization and these were further tested in an infant mouse competition assay to confirm this phenotype. Of these 51 strains only 23 showed a significant colonization defect.²⁹ Five of these strains corresponded to genes present on the toxin co-regulated pilus (tcp) operon which has previously been shown to be critical for colonization.^30-32

The remaining isolates had mutations in genes involved in LPS, biotin or purine biosynthesis genes. There were 4 mutations in the LPS biosynthetic gene rfbL. While it is known that alterations in LPS cause reduction in both virulence and colonization that may not be the only reason why this mutant is defective in colonization.^33,34 The rfbL gene is located in a 20 kb region containing a number of genes involved in O-antigen formation and a mutation in this region also appears to cause translocation arrest of TcpA which is important for colonization of the infant mouse model.^35,36 There were several mutations in genes involved in purine biosynthesis; purD, purH and purK and one in biotin biosynthesis (bioB). These mutants were auxotrophs in minimal media and were only able to grow when supplemented with purine or biotin. This indicates that purine and biotin is limiting in the infant mouse intestine.

A second, more comprehensive STM screen was performed in V. cholerae El Tor strain N16961. This strain has enhanced colonization capacity compared with the C6709–1 strain.³⁷ Indeed Merrell and colleagues reproducibly recovered 10-fold more bacteria using the N16961 strain compared with the C6709–1 strain in the infant mouse infection model allowing them to carry out a much larger screen using 96 individually tagged mutants per pool. In total there were 9600 mutants screened for attenuation in infant mice and 1026 mutants showed a reduced capability to infect mice. These 1026 mutants were re-assembled into 22 secondary pools of 30–50 STM strains and re-screened in the same suckling mouse model of infection. This revealed 251 strains that were still unable to colonize the small intestines. Only the site of transposon insertion was identified in 164 strains as the rest of the mutations were plasmid integrations and would require cloning to detect the site of integration. To determine how these mutations may affect survival in the host, virulence-attenuated pools (VAP) were constructed and screened for acid tolerance (as an in vivo-associated stress). Using this method they identified nine mutants that were essential for both colonization and survival under acid stress conditions.

One such mutant identified from this screen is a homolog of gshB, which encodes the enzyme glutathione synthase (GSH) and catalyzes the final step of glutathione synthesis. A previous publication demonstrated that a mutation in gshB in Rhizobium leguminosarum was unable to regulate intracellular pH leading to accumulation of intracellular K⁺. However this screen was the first to link reduced colonization and decreased survival in organic acids to a gshB mutation in V. cholerae. An additional virulence factor shown to be important in colonization and acid tolerance response was a HepA homolog. HepA was originally identified in E. coli as a protein that co-purified with RNA polymerase and inhibited the binding of sigma 70.³⁸ A mutation in E. coli in this gene leads to increased sensitivity to UV damage.³⁸ It is thought that acid stress causes DNA damage and if HepA is required for synthesis of the DNA repair enzymes it would establish a link between the two functions. Both of these mutations resulted in a 1000 fold decrease in colonization signifying the importance of both of these genes in the infant mouse model.³⁷

An interesting finding from this screen is that over 20% of the genes identified in the screen play a role in energy metabolism. This indicates that the environment in the small intestine of the suckling mouse is nutrient limited. Therefore to survive within such an environment V. cholerae must actively employ multiple pathways for energy source acquisition and survival.

Escherichia coli

E. coli is a common member of the commensal bacteria of the large intestine, however certain strains have the ability to cause disease. Enterohemorrhagic E. coli (EHEC) is known to cause disease in humans associated with diarrhea and hemorrhagic colitis.^39-41 EHEC serotype O157:H7 has emerged as a major cause of severe diarrhea worldwide and EHEC is the leading predecessor to pediatric acute renal failure in many countries.^42,43 Healthy ruminants are the principal reservoir of EHEC and human infections occur by ingestion of contaminated meat or dairy products contaminated with ruminant feces.^44,45 To increase the knowledge of EHEC factors associated with bovine colonization a STM mutant bank was created in O157:H7 background.⁴⁶ A total of 1900 mutants were screened by oral inoculation of 10–14-day old calves with recovery of the output pool 5 days post-infection. 79 mutants were identified to be absent or poorly represented in the output pool. All the genes were grouped according to their function and consisted of genes involved in TTSS, surface structure, O-islands, regulatory genes, genes involved in central intermediary metabolism and hypothetical genes. Thirteen transposons were inserted into genes on the locus of enterocyte effacement (LEE). LEE encodes a TTSS required for formation of attaching and effacing lesions on the intestinal epithelia. Their data demonstrated that the structural component of TTSS escC plays a vital role in colonization of the calves as this mutant was highly attenuated following oral infection of calves. This was the first time that the structural components of the TTSS were implicated in colonization of the calf intestine.⁴⁶ Furthermore this work demonstrated that colonization of the bovine intestines requires multiple elements not associated with LEE. Their evidence suggests that a novel fimbrial locus (z2199-z2206) plays an important function in intestinal colonization.⁴⁶ This was the first comprehensive study to elucidate the genes required by E. coli O157:H7 for infection of bovine intestine and this new information can be used to facilitate the development of strategies that may aid in the control of EHEC in the ruminant reservoir.

Another STM study investigated the ability of a EHEC O26:H^- strain to colonize the bovine intestine.⁴⁷ This strain was analyzed as it was believed that this non-O157:H7 strain may colonize the bovine intestine by a different mechanism.⁴⁷ In total 570 mutants were examined in six calves and 84 mutants were absent or poorly represented in the output pools at 5 days post-infection.⁴⁷ As with the 0157:H7 strain LEE mutants were identified as playing a role in infection and colonization of the intestine.⁴⁷ One interesting finding was the role of fimbriae in EHEC O26:H^- colonization. Fimbriae are cell surface appendages widely used by bacteria for attachment to host tissues during infection. However enhanced expression of type I fimbriae due to a fimE mutation was detrimental for the persistence of EHEC O26:H^- in vivo.⁴⁷ It is thought this may be due to enhanced triggering of innate immune responses in the host leading to more rapid clearance of the mutant from the intestinal tract.⁴⁷

Van Diemen and colleagues also demonstrated that certain cytotoxins play a role in colonization in EHEC O26:H^{- 47}. The STM screen in EHEC O26:H^- demonstrated that an insertion in exhA which encodes an enterohemolysin significantly attenuates virulence.⁴⁷ Furthermore a mutation in pssA was also identified in this STM screen. PssA encodes a serine protease that is cytotoxic to VERO cells and belongs to the SPATE (serine protease autotransporters of Enterobacteriaceae) group of proteins whose role in pathogenesis in not fully elucidated. This discovery is the first direct evidence demonstrating a role in for SPATE proteins in the intestinal colonization of calves by EHEC.⁴⁷

Citrobacter rodentium

The mouse enteric pathogen Citrobacter rodentium, like its human counterpart, enteropathogenic E. coli, causes attaching and effacing (A/E) lesions in the intestinal epithelium of its host. This phenotype requires virulence factors encoded by the LEE pathogenicity island. An STM screen was created in C. rodentium and analyzed in two different strains of mice; C3H/HeJ and C57BL/6.^48-50 The C3H/HeJ strains of mice are highly susceptible to infection due to a lack of an innate immune response to LPS and are therefore colonized more rapidly and to a higher degree. This results in more extensive colonic hyperplasia and higher mortality rates.⁵¹ In contrast C57BL/6 mice are more resistant to C. rodentium infection and disease and screens in this model may uncover additional genetic factors necessary for full virulence in a resistant host background.⁵¹

In the initial screen using the C3H/HeJ mice a total of 576 mutants were analyzed and 14 mutants were identified as being defective in their ability to colonize the descending colon at 5 d post-infection.^49,50 These affected genes included cfcH which encodes a type I pilus biogenesis nucleotide-binding protein homolog.⁵⁰ Further analysis of this region identified that this was part of a 12 gene cluster that has identity at the amino acid level to type IV class B pilus biogenesis genes of ETEC, EPEC and V. cholerae. These type IV pili share some structural, biochemical and functional features in common.⁵² They have been grouped into two classes, with class B pilins being associated with intestinal infections.⁵³ Mundy and colleagues suggest that disruption of cfcH is likely to prevent the assembly and/or secretion of CFC pili.⁵⁰ This would result in C. rodentium cells being unable to adhere to colonic epithelia and therefore unable to establish an infection.⁵⁰ Another gene identified in this screen encodes a novel type III secreted protein, EspI, which is encoded outside the LEE region and is present in the sequenced A/E EHEC and EPEC pathogens.⁴⁹ The function of this gene has not been elucidated but it has been shown to play an important role in both bacterial colonization of colonic epithelium of infected mice and induction of hyperplasia in the colonic epithelium of infected mice.⁴⁹

The second STM screen performed in C. rodentium utilized the C57BL/6 mouse model. The study examined 576 mutants of which 19 were attenuated for survival at 5–7 d post-infection.⁴⁸ Several insertions corresponded to previously identified virulence genes, including the gene cluster cfc and the espI. However one interesting finding was an insertion in the gene encoding a putative translocation effector of A/E pathogens, NleB.^48,54 An nleB deletion mutant was constructed and tested for its ability to colonize the mouse. As with the transposon mutant the nleB deletion mutant was outcompeted by the wild-type in mixed infections. In addition in single infections it was also shown to be essential for colonization and virulence.⁴⁸ NleB is also present in the EHEC O157:H7 strain indicating that C. rodentium is an invaluable small animal model to represent other A/E pathogens and to test the role of new effector proteins in disease.⁴⁸

Campylobacter jejuni

Campylobacter jejuni is the most common bacterial cause of food-borne disease in the developed world, with estimates that it infects 1 out 100 individuals in the United States and United Kingdom.^55,56 In the developed world, campylobacteriosis is common in neonates and young adults resulting in mild bloody diarrhea, abdominal cramps and the presence of fecal leukocytes.^57,58 Although the vast majority of cases are self-limiting, campylobacter can cause severe post-infection complications, such as bacteraemia and polyneuropathies such as Guillain-Barre´ and Miller-Fisher syndrome.⁵⁹C. jejuni usually infects the avian gastrointestinal (GI) tract particularly of chickens.⁶⁰ During the slaughtering process, the GI contents may contaminate the meat products and ingestion and handling of contaminated meats are a main cause of sporadic cases of C. jejuni disease.⁵⁵ STM has been used in both early and late chicken models of infection with varying degrees of success.

The first STM screen in C. jejuni analyzed cecal colonization of chicks in a 1-d old chick model of commensalism.⁶¹ In total 1550 C. jejuni mutants were screened of which 29 were attenuated for colonization representing 22 different genes required for wild type levels of infection.⁶¹ The vast majority of the mutants (17) exhibited a non-motile phenotype or displayed altered flagellar motility.⁶¹ It was previously known that motility of C. jejuni is required for wild-type levels of cecal colonization therefore validating the efficacy of the screen.^62,63 Of the remaining mutants two were of particular interest Cj0019c and Cj0020c later designated docB and docA (determination of chick colonization) respectively.

The docB gene had homology to numerous bacterial methyl-accepting chemotaxis proteins (MCP). In other bacteria these proteins detect specific environmental components and transduce signals to the flagellar motor to alter the direction of motility.^61,64,65 The docB mutant was shown to have a 100-fold lower level of colonization in all intestinal organs compared with wild type. The second gene docA shows significant homology to bacterial perisplasmic cytochrome c peroxidises which bind two c-type heme compounds to convert hydrogen peroxide to water without generating other reactive oxygen intermediates.^61,66-68 This mutant was able to colonize all intestinal organs from day 1 to day 4 but after day 4 the level of colonization decreased or remained level. Finally at day 7 the level of colonization was 10–100-fold less than wild type. These results suggest that these genes are not required for specific tissue tropisms but instead are necessary for wild-type colonization of the entire GI tract but it is not yet known if these genes play a role in pathogenesis.

It is believed in the field that chickens are colonized by C. jejuni at around 2-weeks of age.⁶⁹ A model of colonization in longer-term studies using 2-week old birds with an established gut flora is considered a better reflection of natural conditions. Therefore another screen was performed in C. jejuni in a 2-week old chick model.⁵⁹ An STM library was generated in three strains of C. jejuni using a modified marnier transposon system. In each STM library there was a high-frequency of random loss of colonization-proficient mutants from pools and experimental variation during mixed infections.⁷¹ This was not observed when a similar STM screen was conducted in Salmonella enterica serovar Typhiumurium using the same model of 2-week old chicken colonization indicating that these findings may be a consequence of the nature of C. jejuni colonization.⁷⁰ The primary reason for this is thought to be population bottlenecks during transit through the GI tract prior to establishment of bacterial colonies in the ceca indicating that STM may not be suitable for studying long-term colonization in C. jejuni.⁵⁹ In an attempt to overcome this problem the authors developed another method of screening known as wild-type isogenic tagged strains (WITS). As with STM each strain only differs from each other by a 40-bp DNA tag inserted into a pseudogene. The relative proportion of these strains in a population can be analyzed by quantitative RT-PCR (qRT-PCR). As with STM there was extensive variation between experiments demonstrating that simple colonization models are not appropriate for this organism.⁷¹

Salmonella enterica

S. enterica is a facultative intracellular pathogen of animals and humans. There are more than 2,400 serovars which can be divided into three broad classes based on their host specificity.⁷² S. enterica Typhimurium causes gastroenteritis in humans, but in mice it leads to systemic infection after oral infection. In mice the bacterium spreads to cells of the reticulo-endothelial (RE) system and therefore it replicates human typhoid fever.⁷³S. typhimurium was the first bacterium used in an STM screen.¹ It was chosen for proof-in-principle of STM because of its excellent genetic systems and well validated animal models.^1,73 The original STM screen identified a novel pathogenicity island, SPI-2 ⁷⁴. SPI-2 was only identified due to its role in disseminated infection. SPI-2 encodes a type three secretion system (TTSS) which is required for intracellular replication and systemic infection.⁷⁴ Furthermore, SPI-2 genes have been shown to be involved in the survival of Salmonella within the macrophages and play a role in avoidance of NADPH oxidase-dependent killing.^75-77 Recently SPI-2 mutants have been included in live attenuated vaccines in S. typhi and S. typhimurium indicating how STM has lead from proof-of-principle all the way to potential clinical applications.^78,79

As stated earlier S. typhimurium can colonize a range of different hosts and STM has been used to try and elucidate both species specific virulence factors and common colonization factors to allow a better understanding of how this bacteria is able to infect such a wide variety of different niches. A previous mini-Tn5Km2 STM bank was used to screen mutants for attenuated virulence in both calves and chicks.⁸⁰ S. typhimurium infection in calves results in enterocolitis followed by systemic infection, while infection of 2-week old chicks results in asymptomatic cecal colonization. The STM screen in the calves recovered mutants 3–5 days after infection from the ileal mucosa while in the chicks the mutants were recovered from homogenized ceca at 4 d post-infection.⁸⁰ In total 1045 mutants were screened in both hosts. Of the screened mutants 75 were associated with attenuation in the calves, 61 were associated with attenuation in chicks alone and 52 mutants were attenuated in both species.⁸⁰

A large proportion (n = 40) of the mutated genes were within Salmonella pathogenicity islands (SPIs 1–5). All of the mutants with a transposon insertion in SPI-1 or SP1–2 were deficient for colonization of the calf model but only 3/32 of these SPI mutants resulted in poor colonization of the chick ceca.⁸⁰ This suggests that S. typhimurium is much less dependent on TTSS-1 and TTSS-2 to colonize the intestines of chicks compared with calves.⁸⁰ Furthermore SPI-4 was found to be required for colonization of calf ileum but not for cecal colonization of chickens.⁸⁰ Several genes required for production of LPS were identified by this in vivo screen as being attenuated in both calf and chick colonization models.⁸⁰ LPS is widely considered to play a role in protecting the bacteria against host defense mechanisms such as bile salts, gastric acidity and phagocytes. The precise role of LPS in Salmonella virulence is not yet known but it clearly has a major function in colonization of the intestinal sites of different hosts. Of the genes associated with attenuation within the chick model several mutants had transposon insertions in genes required for production of six different fimbriae.⁸⁰ Fimbriae are used by bacteria to adhere to each other as well as host surfaces, and this data indicates that fimbriae are important for colonization within the intestine of the chicken.

The same STM mutant library was used to identify novel genes associated with virulence in the intestinal colonization of pigs.⁸¹ This screen identified 119 mutants attenuated for virulence in the porcine model of infection. Of these 119 mutants, the transposon insertion site of 79 had been identified in the previous screen.⁸⁰ The remaining 40 transposon mutants were associated only with attenuation in the pig model of infection.⁸¹ One of the mutants identified had a transposon insertion in safA gene. This gene is part of an operon (safABCD) which is located within the Salmonella enterica centisome 7 genomic island (SCI) also known as SPI-6. The safA gene encodes Salmonella-specific putative atypical fimbriae. As this mutant was not attenuated in either the chick or calf model of infection this may reflect the ability of the bacteria to express different adhesions to exploit different animal hosts.⁸¹ Several genes under the control of the PhoPQ two-component system (TCS) were identified in this screen. The Salmonella PhoPQ TCS is known to regulate a number of genes that play a role in growth in low Mg²⁺, resistance to antimicrobial peptides, bile salts and acid pH. This raises the possibility that this system may play a role in colonization of pigs by Salmonella.⁸¹

In recent years STM has been improved by modifying the mutagenizing transposon to include a T7 RNA polymerase promoter (P_T7) that is used to generate a unique transcript for each mutant from the genomic sequence adjacent to the mutation. This modification makes exogenous unique sequence tags unnecessary. Relative abundance of the input and output P_T7 transcripts are monitored using an ORF microarray.^82-85 This method has been applied to Salmonella using a lamba-red recombination method that includes features to minimize polarity and to construct targeted deletion mutants. The P_T7 was added to the cassette inserted during mutagenesis and positioned to produce a gene-specific transcript from the genomic sequence adjacent to the insertion.⁸⁶ In total there was a pool of over 1,000 gene-targeted Salmonella mutants injected into BALB/C mice and the spleen analyzed 2 d post infection. A threshold of a 2-fold change was designated as having decreased fitness compared with input pool.⁸⁶ In total 120 mutants were shown to have decreased fitness, among these were mutants associated with TTSS encoded by SPI-2 and associated effector genes, and genes for cell wall biosynthesis all known to play a role during infection.⁸⁶

There were 15 mutants previously unknown to have a role during systemic infection in BALB/c mice identified using this method. These included all five candidate sRNA, a supernumerary tRNA (LeuX) and nine protein coding genes.⁸⁶ Mutations in leuX, encoding tRNA-Leu, were previously known to reduce the expression of Type I fimbriae and reduce bladder epithelial invasion and intracellular proliferation of Uropathogenic E. coli.^87,88 This was the first time a leuX mutant was demonstrated to have a role during Salmonella infection.⁸⁶ One advantage of this method is being able to identify mutants that may enhance survival during infection, exemplified by a mutant in oxyS. OxyS is a member of the OxyR regulon expressed during oxidative stress⁸⁶ and encodes a sRNA that regulates over 40 genes in E. coli.^89,90 Indeed this study is one of the first to uncover a distinct phenotype for small non-coding RNA mutants during Salmonella infection.⁸⁶

The advantages of this modified STM system is that it permits easy generation and confirmation of specific deletion mutants, it will minimize the effects of population bottlenecks on screening, will allow analysis in relevant animal models and will permit more appropriate doses to be efficiently used to identify Salmonella mutants with altered fitness in vivo.⁸⁶ Furthermore it is the first step toward a more complete description of Salmonella genes involved in systemic infection, in particular genes that may have a milder phenotype that are difficult to detect by the older STM methods.⁸⁶

Listeria monocytogenes

Listeria monocytogenes is a Gram-positive food-borne pathogen responsible for life-threatening infections in humans and animals. It is a facultative intracellular pathogen capable of entering a wide variety of host cells, including epithelial cells, hepatocytes, fibroblasts, endothelial cells and macrophages.^91-94 Infection occurs in step-wise manner consisting of entry into the host, lysis of the phagosomal vacuole, multiplication in the cytosol and direct cell to cell spread using actin based motility.⁹⁵ Each step is dependent on virulence factors which are located in a cluster of genes encoding a regulatory protein (PrfA), a phosphatidylinositol specific phospholipase C (PlcA), the hemolysin listeriolysin A (LLO), a metalloprotease (Mpl), an actin recruiting protein (ActA) and a lecithinase (PlcB).⁹⁵ A second locus encodes two proteins involved with invasion, InlA and InlB. Expression of these virulence genes are controlled by the pleiotropic regulator PrfA.

A STM approach was implemented in the L. monocytogenes EGDe background using a Tn917 derivative transposon. Mutants were screened in vivo for reduced colonization of the spleen and liver at 72 h post-infection.⁹⁶ From this screen the response regulator VirR was identified. VirR (Virulence Regulator) has high homology to the OmpR-PhoB family of regulators.⁹⁶ It is part of a seven gene operon, which also contains the sensor kinase, VirS. The VirRS two-component system (TCS) is novel in that unlike most TCS the constituent genes are not adjacent to each other but are separated by three other loci.⁹⁶ The results obtained from this study demonstrated that the ΔvirR strain had a reduced ability to grow and multiply within both liver and spleen even after 24 h indicating a crucial role for VirR in the establishment of successful L. monocytogenes infection.⁹⁶

This STM screen also identified another novel virulence factor in L. monocytogenes designated FbpA. This gene has strong homology to atypical fibronectin-binding proteins such as PavA of Streptococcus pneumoniae, Fpb54 of S. pyogenes and FbpA of S. gordonii.^97-99 Fibronectin is a dimeric glyocoprotein that has a critical role in eukaryotic cellular processes such as adhesion, migration and differentiation.¹⁰⁰ However, many bacteria such as S. pneumoniae and Staphylococcus aureus utilize fibronectin to facilitate their internalization into epithelial cells.¹⁰¹ Mutation of fbpA in L. monocytogenes resulted in a 100-fold decrease in bacterial counts in the intestine and liver in orally infected mice when compared with the wild-type strain after 72 h.¹⁰⁰ Furthermore, there was a 10-fold decrease in bacterial counts in the mesenteric lymph between mutant and wild-type but no difference in the spleenic bacterial counts.¹⁰⁰ This data indicated that FbpA is involved in the hepatic phases of listeriosis and represents a novel virulence factor.

Another STM screen was performed in L. monocytogenes using aTn1545 transposon-based approach.^4,102 As with the previous screens this study looks at genes that were attenuated for virulence in the liver post-infection.¹⁰² This work identified an agrA homolog in L. monocytogenes that plays a role in bacterial virulence. AgrA is a response regulator that is part of a TCS that has been extensively studied in S. aureus and shown to control virulence factor expression.^103,104 A mutant in agrA in L. monocytogenes EGDe background had a 10-fold higher LD₅₀ compared with wild-type in the mouse infection model.¹⁰² While this mutant only resulted in a slight attenuation in vivo it influences the production of several secreted factors, including LLO.¹⁰²

As well as having tropism for the liver and spleen L. monocytogenes also has the ability to infect the brain and cause meningitis, meningoencephalitis and abscess formation. The STM bank created by Autret et al. was used to identify genes that are involved in passage across the blood-brain barrier.⁴ This is the only STM screen in L. monocytogenes that examined the brain as the target organ. They identified 10 mutants attenuated for virulence in the brain. They tested four of the mutants for their ability to colonize the brain, liver and spleen and found that the genes were needed specifically to infect the brain and not for growth within the host.⁴ This demonstrates that certain virulence genes are responsible for different tropisms within the host.

Detection of Transposon Insertion Sites by Second Generation Sequencing

Some limitations associated with the STM approach include the workload involved in constructing and screening large libraries and the costs associated with screening in animals (particularly if screens are performed in ruminants). To overcome these limitations, new high-throughput sequencing-based approaches have been developed that permit the simultaneous assignment of insertion-site and fitness score for mutants screened in pools.¹⁰⁵ Transposon-directed insertion site sequencing (TraDIS) is one such approach which exploits Illumina sequencing to obtain the sequence flanking each transposon insertion in large mutant pools.¹⁰⁶ The use of second generation sequencing provides high levels of coverage of mutant banks and provides a numerical measure of the extent to which mutants are selected in vivo. TraDIS prevents the need to construct and array uniquely tagged mutants and to sub-clone and sequence attenuating mutants, thereby substantially saving time and cost. Furthermore the approach can be applied to very large mutant pools and could potentially reduce the numbers of animals used in large transposon screens. The study that first developed the TraDIS approach utilized the method for in vitro analysis of bile tolerance in S. Typhi.¹⁰⁶

A similar massively parallel sequencing approach was recently applied to assign genotype and fitness scores to E. coli O157:H7 mutants previously screened in calves.^46,107 Of the 1,805 mutants screened, parallel sequencing assigned insertion site and fitness scores for 1,645 mutants which represented insertions in 855 different genes. This represented 91.1% of mutants analyzed while the previous STM bank only identified insertion sites in 4.2% of mutants.¹⁰⁷ Furthermore the STM screen in O157:H7 identified 13 attenuating mutations in LEE genes but sequencing identified 54 insertions in the LEE region which corresponded to 21 different genes.¹⁰⁷ Analysis of the STM bank of EHEC O26:H^- demonstrated a role for cytotoxins (EhxA and PssA) during pathogenesis but these genes were not identified in the EHEC O157:H7 screen.^46,47 Parallel sequencing revealed that several such mutants were represented in the library and were generally negatively selected in calves.¹⁰⁷

A similar massively parallel sequencing approach called INSeq (insertion sequencing) was developed independently to analyze transposon insertion sites in the gut commensal organism Bacteroides thetaiotaomicron.¹⁰⁸ The authors utilized this approach to analyze genes required for colonization of conventional and germ-free or mono-colonized gnotobiotic mice. The work revealed how colonization by Bacteroides is influenced by existing populations in the gut and competition for key nutrients in this environment.

Conclusions

STM is a powerful genetic tool that allows identification of genes that are important for different facets of pathogenesis and is well suited for analysis of elements required for gut colonization and localized pathogenesis. Recent technical advances in the screening, choice and identification of negative selection screens have broadened its applicability and versatility. One such technical advance is the development of a positive STM screen.¹⁰⁹ This was used to screen for patho-adaptive Pseudomonas aeruginosa mutants promoting survival in the cystic fibrosis lung.¹⁰⁹ This novel approach could be applied to other pathogens that enhance fitness in the host through patho-adaptive mutations and could provide a basis for a more comprehensive understanding of chronic infectious disease.¹⁰⁹

Overall the STM tool is an important method for better understanding the behavior of microbes in the gut and other environments and in conjunction with other genome wide techniques (microarray technology, in vivo expression technology, TraDIS) can be used to fully understand the multi-faceted nature of bacterial pathogenesis. It is expected that the results from these STM screens can be used to help develop vaccines or drugs to prevent additional infections and decrease the economic burden associated with such infections.