Logo of iaiPermissionsJournals.ASM.orgJournalIAI ArticleJournal InfoAuthorsReviewers
Infect Immun. Jul 2002; 70(7): 3404–3412.
PMCID: PMC128117

Identification of Escherichia coli Genes That Are Specifically Expressed in a Murine Model of Septicemic Infection


Identification and characterization of bacterial genes that are induced during the disease process are important in understanding the molecular mechanism of disease and can be useful in designing antimicrobial drugs to control the disease. The identification of in vivo induced (ivi) genes of an Escherichia coli septicemia strain by using antibiotic-based in vivo expression technology is described. Bacterial clones resistant to chloramphenicol in vivo were recovered from the livers of infected mice. Most of the ivi clones were sensitive to chloramphenicol when grown in vitro. Using reverse transcription-PCR, it was demonstrated that selected ivi clones expressed cat in the livers of infected mice but not during in vitro growth. A total of 750 colonies were recovered after three successive rounds of in vivo selection, and 168 isolated ivi clones were sequenced. The sequence analysis revealed that 37 clones encoded hypothetical proteins found in E. coli K-12, whereas 10 clones contained genes that had no significant homology to DNA sequences in GenBank. Two clones were found to contain transposon-related functions. Other clones contained genes required for amino acid metabolism, anaerobic respiration, DNA repair, the heat shock response, and the cellular repressor of the SOS response. In addition, one clone contained the aerobactin biosynthesis gene iucA. Mutations were introduced in to seven of the identified ivi genes. An in vivo mouse challenge-competition assay was used to determine if the mutants were attenuated. The results suggested that these ivi genes were important for survival in vivo, and three of the seven mutant ivi clones were required for successful infection of mice.

Escherichia coli is a common gram-negative, enteric organism that also is an important etiological agent causing bloodstream infections in humans, particularly nosocomial septicemia (22, 37, 43). E. coli has been reported to be responsible for more than one-quarter of all septicemic infections (11). Individuals who are in a high-risk category for disease are the elderly and hospitalized patients. In the elderly, the fatality rate due to E. coli septicemia is 6.9% (2). E. coli strains expressing the K1 antigen also are important extraintestinal pathogens and are associated with neonatal meningitis. The incidence of E. coli K1-induced meningitis is 0.1 per 1,000 live births (7), and fatality rates range from 25 to 40% (40).

One approach to develop new treatments for E. coli infections is to identify and characterize genes that are expressed in vivo during disease and consequently those that are essential for the survival of bacteria in the infected host. During septicemia, E. coli is exposed to a unique set of growth conditions and clearance mechanisms. A number of growth conditions that affect the expression of virulence-associated E. coli genes, such as iron and nutrition starvation, oxygen tension, pH, and stress are well documented (29). These in vitro growth techniques mimic individual growth conditions present in the infected host but do not replicate the highly complex environment that a pathogen encounters in the infected host. In the complex host-pathogen interaction, pathogenic bacteria also must evade the host immune system in a nutrition-depleted environment. These hostile environments require that a successful pathogen must express specific sets of genes in response to the growth milieu. Most pathogenic bacteria respond to the interaction with eukaryotic cells by expressing various virulence factors, including adhesins (9, 17, 18), toxins (6), and iron uptake systems (45). In recent years, techniques that could be used to identify bacterial genes that are uniquely expressed during infection have been developed (15, 26, 27). In vivo expression technology (IVET) has been used as a positive selection tool for in vivo-induced (ivi) genes. This technique was originally developed for Salmonella enterica serovar Typhimurium and has been useful in the identification of previously uncharacterized genes (26). The original technique relied on complementation of an auxotrophy during selection in a murine infection model. Auxotrophic strains survived only if a cloned chromosomal DNA fragment possessing in vivo promoter activity promoted expression of a promoterless auxotrophic gene integrated into the bacterial chromosome. Using a similar technique, the auxotrophic marker was successfully replaced by a promoterless chloramphenicol acetyltransferase (cat) gene to isolate ivi genes of S. enterica serovar Typhimurium (27). In addition, other molecular techniques, such as signature-tagged mutagenesis to negatively select for tagged mutant genes required for pathogenesis (15), in vivo-induced antigen technology (IVIAT) selecting for ivi genes by screening expression libraries with a patient's serum after absorbing with cells of the pathogen grown in vitro (13), and a promoter trap technique that relies on genetic recombination as a reporter of gene expression (24), are being used to identify in vivo-expressed genes. These techniques have been successfully used to identify ivi genes of Yersinia enterocolitica (46), S. enterica serovar Typhi (38), Pseudomonas aeruginosa (42), Actinobacillus pleuropneumoniae (10), Staphylococcus aureus (24), and Pasteurella multocida (16).

In this study we sought to identify genes of an E. coli strain that causes nosocomial septicemia that were expressed exclusively during disease. To fulfill this objective, we developed a variation of the original IVET procedure. The procedure employs a nonchromosomal integration system with a low-copy-number plasmid for large-scale screening of chromosomal fragment::cat fusions to isolate ivi genes. The ivi genes identified included genes similar to uncharacterized open reading frames (ORFs) of E. coli K-12, genes that have no similarity to E. coli K-12, and genes that have no homologues in the public databases of DNA and protein sequences.


Bacterial strains, plasmids, and culture media

E. coli strain i484 was isolated from a human with septicemia. It is sensitive to all antibiotics tested. It has the serotype O25:H, autoagglutinating. Its derivatives and other E. coli strains used in this study are listed in Table Table1.1. Luria-Bertani (LB) medium was routinely used to grow E. coli strains either in liquid culture or on agar plates at 37°C. Final concentrations of the antibiotics in growth media were as follows: ampicillin, 100 μg/ml; chloramphenicol, 25 μg/ml; and streptomycin, 20 μg/ml. To induce iron limitation in minimal medium and LB agar plates, 2,2′-dipyridyl (ICN Biomedicals, Inc., Irvine, Calif.) was used at concentrations ranging from 25 to 150 μM.

Bacterial strains and plasmids

Construction of genomic library.

A genomic library of E. coli i484 was constructed in plasmid pKK232-8, a promoter selection vector (Pharmacia, Piscataway, N.J.). Chromosomal DNA was isolated from E. coli strain i484 by a procedure employing cetyltrimethylammonium bromide (1) and was partially digested with restriction endonuclease Sau3AI. DNA fragments of 0.5 to 2 kb were ligated into the BamHI site of the vector pKK232-8. After transformation of the recombinant DNA into E. coli strain HB101, a total of 4 × 104 colonies were collected by plating on LB agar containing ampicillin. Random clones were picked, and the average insert size of the genomic library was shown to be approximately 1 kb. This represented an eightfold redundancy, based on the size of the E. coli K-12 genome. Subsequently, plasmid DNA was prepared from the recombinant strain and electroporated into the streptomycin-resistant mutant of i484, strain i659.

ivi gene selection in mouse septicemia model.

Six- to 8-week-old female ICR mice (Harlan Dawley, Indianapolis, Ind.) were challenged by intraperitoneal injection with cells from the genomic library. Bacterial cells in exponential growth were harvested and resuspended in phosphate-buffered saline. Based on preliminary data (see Results) the challenge dose was set at 108 CFU. Immediately after challenge of a group of five mice, mice were injected intraperitoneally with 500 μg of chloramphenicol. A total of five doses of antibiotic were administered at 4-h intervals. At 24 h postchallenge, mice were euthanized with CO2. The livers were isolated, homogenized in 10 ml of sterile H2O, plated on LB agar plates containing 20 μg of streptomycin per ml, and incubated overnight at 37°C. Bacterial colonies were then replica plated onto LB agar plates containing 100 μg of ampicillin per ml and on plates with 100 μg of ampicillin per ml and 20 μg of chloramphenicol per ml. Subsequently, plated clones that were sensitive to chloramphenicol were pooled and subjected to two more rounds of in vivo selection.

To determine the effects of mutations in the ivi genes of strain i484 in vivo, a mouse competition infection model was used. A mixture containing an equal number of cells of the mutant and wild-type strains was used to challenge mice. Twenty-four hours postchallenge, bacteria were recovered from the livers of challenged mice and plated on LB agar. The wild-type strain was resistant to streptomycin, whereas the mutant encoded resistance to ampicillin. Survival of each bacterial strain was scored on LB agar plates containing either ampicillin or streptomycin.

Cycle sequencing and DNA sequence analysis.

Double-stranded plasmid DNA was prepared using the Qiagen (Valencia, Calif.) plasmid kit. A sequencing primer, 5′-GAAAATCTCGTCGAAGCTC-3′ (Ocat1), that read into the flanking cloned DNA fragment was designed within cat. DNA sequencing was performed using the Taq DyeDeoxy terminator cycle sequencing kit (Perkin-Elmer, Foster City, Calif.). Sequencing reactions were analyzed using an automated DNA sequencer (model ABI-373; Applied Biosystems, Foster City, Calif.). DNA and protein sequences were analyzed using the Wisconsin package version 9.1 (Genetic Computer Group, Inc., Madison, Wis.). Homology searches and sequence comparisons were performed with sequences in GenBank, EMBL, and unfinished genome databases through National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov). Sequences that had no homologue in the GenBank were analyzed using GeneMark (http://dixie.biology.gatech.edu/GeneMark/gmhmm2_prok.cgi) to predict the 5′ ends of the gene (25).

Isolation of RNA and RT-PCR.

RNA was isolated from the livers of infected and noninfected mice and from E. coli cells grown in LB medium by using the RNeasy total RNA kit (Qiagen Inc., Valencia, Calif.). To isolate RNA from mouse liver, the liver was removed, immediately frozen in liquid nitrogen, and ground to fine powder with a mortar and pestle. Approximately 50 mg of ground tissue containing 3 × 104 bacterial cells or 109 E. coli cells grown in LB medium were used to isolate total RNA. The cat gene transcript was then detected by reverse transcription-PCR (RT-PCR). Two primers were designed, 5′-TTCTGCCGACATGGAAGCCATCAC-3′ (RP1) and 5′-CCTATAACCAGACCGTTCAGCTGG-3′ (FP1), that were specific to cat and that amplified a 516-bp internal fragment of cat mRNA. cDNA synthesis and PCR amplification were performed using Tth DNA polymerase (Boehringer Mannheim, Indianapolis, Ind.). RT was performed in the presence of 0.9 mM MnCl2 at 66°C for 20 min. After cDNA synthesis, the forward primer and 1.5 mM MgCl2 were added to the reaction mix, followed by PCR amplification. PCR amplification was for 36 cycles as follows: denaturation at 95°C for 2 min, annealing at 60°C for 1 min, and elongation at 72°C for 1.5 min. At the end of the 36th cycle, reaction mixtures were left at 72°C for another 5.5 min. Ten microliters of reaction mixture was loaded on a 0.8% agarose gel and subjected to electrophoresis. To confirm that the amplified fragment indeed was a product derived from RNA and not the result of the amplification of the genomic copy of the cat gene, the mRNA preparations were subjected to degradation with DNase-free RNase. This was followed by PCR amplification.

Site-directed mutagenesis.

A PCR-based method was used to introduce mutations into seven in vivo-expressed (ivi) genes. Using DNA sequence information on the cloned ivi gene, oligonucleotide primers that were specific for the 3′ end of each cloned DNA fragment were designed. The primers were designed so that frameshift mutations resulting in a premature stop codon were introduced in the ivi sequence, or specific stop codons were designed within the primers (Table (Table2).2). A second forward primer (5′-CACACCGCATTATGGTGCACTCTCAGTACAATCTGC-3′) that was adjacent to the cloning site in the vector pKK232-8 was designed. Prior to PCR amplification, 200 pmol of each primer was phosphorylated using 20 U of T4 polynucleotide kinase (Gibco/BRL, Grand Island, N.Y.), 1 mM ATP, and 6 μl of forward reaction buffer (350 mM Tris-HCl [pH 7.6], 50 mM MgCl2, 500 mM KCl, 5 mM 2-mercaptoethanol) in a total volume of 30 μl. Reaction mixtures were incubated at 37°C for 30 min, followed by heat inactivation of T4 polynucleotide kinase at 65°C for 15 min. The ivi gene fragments were then amplified by PCR using Pwo DNA polymerase (Boehringer Mannheim). Amplification by Pwo DNA polymerase results in blunt-ended amplified fragments. The conditions for PCR included one cycle of denaturation at 95°C for 2 min followed by 30 cycles of denaturation at 95°C for 15 s, annealing at 60°C for 30 s, and polymerization at 72°C for 1 min during the first 10 cycles, for 1 min 20 s during the following 10 cycles, and for 1 min 40 s for the next 5 cycles, with a 2-min elongation time for the last 5 cycles. At the end of the reaction, the mixture was incubated at 72°C for 5 min. The amplified DNA fragments were ligated into the SmaI site of suicide vector pGP704 (30), and the mutations were introduced into i484 as described previously (12). Suicide vector pGP704 contains the RP4 mob site and R6K origin of replication, which requires the π protein to function. The π protein is provided in trans in the vector-containing strain by λpir. The recombinant suicide vector was electroporated into E. coli strain SM10 λpir. Conjugation was carried out on LB agar plates between recipient strain i659 and donor strain E. coli SM10 λpir carrying recombinant pGP704. Transconjugants were selected for resistance to ampicillin and streptomycin. The isolated transconjugants were due to the integration of the recombinant pGP704 by homologous recombination between the target ivi gene on the chromosome and the mutated copy of the gene on recombinant pGP704. The recombination event resulted in production of a truncated target ivi gene and a second mutated copy of the ivi gene. Chromosomal integration was confirmed by Southern blot hybridization using SmaI-linearized wild-type plasmid pGP704 as a probe labeled with 32P.

Primers used for site-directed mutagenesis


Selection of ivi genes in a mouse model of septicemia.

The experimental approach to identify bacterial genes that were specifically induced during E. coli septicemia employed a promoter detection cloning vector and an in vivo antibiotic selection mouse infection model. E. coli strain i484, which is a clinical isolate from a patient with septicemia, was used as the infectious agent. A genomic library was created in the plasmid pKK232-8 (4). pKK232-8 contains a promoterless chloramphenicol acetyltransferase (cat) gene, which was used as a gene reporter. Prior to in vivo selection, challenge doses and antibiotic treatment conditions were selected to optimize the selection process. First it was necessary to select a challenge dose that would result in disease but not in rapid death of the challenged mice. An intraperitoneal challenge dose that resulted in survival of all mice at 24 h postchallenge and in death by 48 to 72 h was sought. Doses of 107, 108, 109, and 1010 CFU were used to challenge mice. At the lowest challenge dose, no significant sign of disease was observed within 72 h. In addition, bacteria were not recovered from the livers of these animals. In contrast, the higher doses (109 and 1010 CFU) resulted in severe disease within an hour of intraperitoneal challenge and caused death of all of the challenged mice within 6 h postchallenge. With a challenge dose of 108 CFU per mouse, signs of disease appeared within 5 h postchallenge but all mice remained alive at 24 h postchallenge; 50% of the mice died in 48 h. This dosage was used for all subsequent challenge experiments.

Since the selection model being developed sought to identify clones that became resistant to chloramphenicol in vivo, it was necessary to determine a chloramphenicol dosing that would protect mice after challenge with chloramphenicol-sensitive E. coli. Mice were challenged intraperitoneally with i484 (2 × 108 CFU) and were given intraperitoneal injections of 0.5 mg of chloramphenicol per mouse at the time of challenge and every 4 h. After five doses of chloramphenicol, the E. coli strain was completely cleared from the livers of challenged mice. This antibiotic treatment protocol was used in the subsequent in vivo selection challenges.

As described in Materials and Methods, a genomic library of E. coli i484 was created in the vector pKK232-8. The average insert size was 1 kb. DNA fragments from E. coli i484 possessing functional promoters were expected to promote resistance to chloramphenicol. The library was screened in vitro prior to passage through the mouse selection model for sensitivity to chloramphenicol. Approximately 96% of the clones contained inserts that did not express the reporter gene for chloramphenicol acetyltransferase and thus were sensitive to chloramphenicol. The library was used to challenge mice that were subsequently treated with chloramphenicol to select for clones expressing the reporter gene in the infected host. It was reasoned that treatment with chloramphenicol would enrich for library clones that contained promoters that were active in vivo. The mice were euthanized at 24 h postchallenge, and 1.6 × 104 E. coli colonies were recovered from mouse liver. In vitro growth on plates containing either streptomycin or streptomycin plus chloramphenicol showed that 18% were resistant to chloramphenicol. This represented a 4.5-fold increase of in vitro chloramphenicol-resistant colonies. Colonies sensitive to chloramphenicol in vitro were presumed to contain genes expressed in vivo only or those that escaped selection by chloramphenicol. The chloramphenicol-sensitive colonies were pooled and subjected to a second round of screening. Pooled bacterial colonies were used to infect mice, followed by the chloramphenicol treatment. After the second round of enrichment, 99.96% of the colonies were sensitive to chloramphenicol in vitro. The in vitro chloramphenicol-sensitive colonies were subjected to a third round of in vivo selection under the same conditions, and a total of 750 bacterial colonies were recovered. All of the isolated colonies after final selection were sensitive in vitro to chloramphenicol (25 μg/ml).

Demonstration of in vivo specific induction of ivi clones.

Seven hundred fifty of the ivi colonies were obtained after three cycles of challenge in chloramphenicol-treated mice. To confirm that the chloramphenicol-treated mouse model indeed selected for clones that expressed cat in vivo, the bacterial cells recovered after the third round of ivi selection were pooled and used to challenge a group of chloramphenicol-treated mice. A second group of chloramphenicol-treated mice were challenged with the chloramphenicol-sensitive parental E. coli strain i484 containing the plasmid pKK232-8. After 24 h postchallenge, the parental strain was completely cleared from mouse livers, whereas 6 × 106 bacterial cells were recovered from the livers of mice challenged with the mixture of ivi clones. These results are consistent with the conclusion that promoters were present upstream of cat in some of the clones that were active during growth in mouse liver. It was considered possible that some or all of the clones simply represented cells that grew poorly or not at all in the liver and therefore were transiently resistant to chloramphenicol. To confirm that the apparent resistance to chloramphenicol in vivo was due to expression of the reporter gene for chloramphenicol acetyltransferase, an RT-PCR method was used to detect the presence of cat-specific transcripts. RNA was prepared from the livers of mice challenged with different ivi clones. RNA also was obtained from cells grown in vitro. In addition, total RNA was isolated from in vitro-grown E. coli strain i484 containing pKK232-8, from liver from a mouse challenged with E. coli strain i484, and from uninfected mouse liver. Using these RNA preparations, RT-PCR amplification was performed using primers that amplified a 516-bp fragment of cat. The 516-bp fragment was detected only from mouse livers infected with ivi clones (Fig. (Fig.1).1). All other RNA preparations, including in vitro-grown ivi clones, did not yield any amplification of this DNA fragment, confirming that the isolated ivi genes were exclusively induced during the process of disease in host tissues and not during in vitro growth. Subsequently, RNA preparations were subjected to digestion with DNase-free RNase followed by PCR amplification to verify that the amplification product was due to RNA and not a genomic copy of cat. No PCR amplification was observed when RNA was digested with RNase, demonstrating that amplification of the 516-bp fragment by ivi clones grown in mice was the result of reverse transcription and amplification of cat-specific mRNA (Fig. (Fig.1b).1b). When DNase was used to digest any DNA in the RNA preparations, the 516-bp fragment was still produced by the ivi clones grown in vivo. These results further confirm the conclusion that isolated ivi clones were exclusively expressed in infected host tissues and not during in vitro growth in LB broth.

FIG. 1.
(a) RT-PCR was performed to detect the presence of cat mRNA in ivi strains i673, i675, i681, i682, i683, i684, i686, and i689 (lanes 1 to 8, respectively); RNA was isolated from infected mice livers. Lane 9, RNA isolated from the liver of a mouse challenged ...

In vitro analysis of ivi clones.

Although the 750 isolated ivi colonies were sensitive to 25 μg of chloramphenicol per ml in vitro, it was speculated that the isolated ivi genes may vary in promoter strength. Therefore, the MICs of chloramphenicol may not be the same for all of the isolated ivi clones when they are grown in vitro. The MIC for E. coli strain i484 containing pKK232-8 is 5 μg/ml. To determine if basal levels of the promoter::cat fusions might vary, the ivi clones were plated on LB agar plates containing 5, 10, 15, 20, or 25 μg of chloramphenicol per ml. While all of the clones were sensitive to chloramphenicol at 25 μg/ml, 6% of the clones were resistant to chloramphenicol at 20 μg/ml. The number of resistant clones increased to 9, 18, and 34% when the concentration of chloramphenicol was 15, 10, or 5 μg/ml, respectively. Thus, some of the clones contained promoters that were expressed at low levels in vitro.

We expected that some of the ivi clones would be iron regulated. Therefore, the ivi clones were grown on minimal medium containing various concentration (25 to 150 μM) of 2,2′-dipyridyl and 25 μg of chloramphenicol per ml. Two clones that expressed cat during iron-limiting growth were identified. The remaining colonies (748) remained sensitive to chloramphenicol when subjected to these growth conditions and likely did not contain promoter::cat fusions that were controlled by iron limitation.

Sequence analysis of ivi clones.

A total of 168 ivi cloned inserts were sequenced, and the nucleotide sequences and predicted amino acid sequences were analyzed. Of the 168 clones that were sequenced, 40 were found once, 37 were found twice, and 18 were found three times, for a total of 95 unique sequences. Sequence analysis revealed that 53 isolated ivi clones were previously characterized genes of E. coli K-12 (Table (Table3).3). Eighteen of the isolated ivi clones contained genes of E. coli K-12 that had been assigned putative functions based on sequence similarities to other known genes. In addition, 14 ivi clones were hypothetical ORFs of E. coli with no assigned functions. Among the isolated ivi genes, 10 clones were unique in that they had no homologue in GenBank and databases of unfinished genomes; these sequences were submitted to GenBank (Table (Table4).4). These clones were analyzed using GeneMark version 2.4, a gene identification and ribosomal binding site prediction program. Shine-Dalgarno-like sequences were identified in eight of the unique ivi clones, followed by an initiation codon. The putative ribosomal binding sites were in the correct orientation to express the reporter cat gene. Putative ribosomal binding sites were not identified in two of the unique ivi clones.

List of E. coli genes identified to be in vivo induced
In vivo-induced E. coli genes submitted to GenBank

In vivo analysis of ivi mutants.

Since the ivi genes of E. coli strain i484 were induced exclusively during infection of mice, they may encode factors required for survival in the infected host. Therefore, mutations in such genes could affect the survival of the clones in the hostile environment of the host. We examined this hypothesis by introducing mutations by site-directed mutagenesis in seven selected ivi genes. These mutants were then used in the competitive infection model. For this model, a mutant was mixed with an equal number (108) of wild-type i659 cells and used to challenge a group of 10 mice. Chloramphenicol selection was not used. At 24 h postchallenge, the mice were euthanized and the concentrations of the mutant and the parent were determined by plating on separate antibiotic-containing plates (streptomycin for the parent and ampicillin for the mutant). The ivi genes selected for mutagenesis were a cloned gene with no significant similarity to known sequences, an ORF of unknown function (o761), ycjF (a gene that encodes a hypothetical protein), one gene with similarity to serine/threonine protein phosphatase (ORF V), and three genes of known function (aroA, bglG, and clpB). The results of the challenges are shown in Table Table5.5. With the exception of clpB (Δivi816) and aroA (Δivi819), all mutants tested were attenuated.

Comparisons of mutant and wild-type growth in the mouse competition model of septicemia


During septicemic infections, bacterial pathogens such as E. coli encounter unique environments to which the pathogen responds by the coordinated expression of regulatory, metabolic, and virulence genes (9, 29). Expression of these genes in the infected host tissues contributes to the survival and successful colonization of these habitats, which otherwise would be too hostile for commensal organisms. Identification of these uniquely expressed genes is important both in understanding the infection process and for the development of new approaches to control the pathogen. For example, the identification of some genes could lead to novel discoveries of new antibiotics. During the last few years, various techniques that have been used to identify bacterial genes that are induced only during the infection process have been reported (13, 15, 24, 27). IVET, which was described by Mahan et al. (26), was developed by using an auxotrophic complementation scheme. In its original form, IVET employed an S. enterica serovar Typhimurium purA mutant to select for ivi genes. In this technique, the promoterless auxotrophic gene purA was fused with a reporter lacZ gene. DNA fragments from a pathogenic test strain were inserted in front of purA, followed by homologous recombination into the genome of an S. enterica serovar Typhimurium purA deletion mutant. These cells were then used to challenge mice. If a promoter was in the cloned DNA fragment and if the promoter was active in vivo, it would drive the expression of purA and consequently complement the auxotrophy. Since purA mutants are avirulent, only those clones expressing purA are viable in vivo. The lacZ reporter was then used to measure gene expression. With IVET, ivi genes of S. enterica serovar Typhimurium and P. aeruginosa have been identified (39, 42). Since the original demonstration that IVET effectively identified ivi genes of S. enterica serovar Typhimurium, several other molecular techniques have been developed that could be used to identify ivi genes. These techniques include signature-tagged mutagenesis, subtractive hybridization, resolvase-dependent detection, and the use of patients serum antibodies that are absorbed with the cells of an in vitro-grown pathogen followed by screening of an expression library. By using these techniques, ivi genes of Streptococcus pneumoniae, Streptococcus gordonii, A. pleuropneumoniae, S. aureus, and Mycobacterium avium (10, 21, 24, 31, 32) have been isolated.

In this report we have described a variation of the IVET procedure to isolate E. coli genes that were induced in an infected host during bacterial septicemia. This method utilizes a promoter trap vector, pKK232-8, which carries a promoterless antibiotic resistance gene (for chloramphenicol acetyltransferase). In this ivi gene selection scheme, transcriptionally active cloned DNA fragments were selected in an antibiotic-treated mouse infection model. The fact that the modified IVET procedure identified several previously known virulence-associated genes, such as aroA, clpA, fepE, and iucA (Table (Table3),3), helped to validate the approach. A similar method has been successfully used to identify ivi genes in S. enterica serovar Typhi and Y. enterocolitica (38, 46). Another difference between the original IVET procedure and the one described here is that it was not necessary to integrate the hybrid plasmid into the chromosome. This simplifies the DNA sequencing of the cloned fragment.

The screening method used in this study selected for genes that were expressed during in vivo growth. To identify those expressed exclusively in vivo, a combination of in vivo and in vitro selection using chloramphenicol was used. By selecting in vivo chloramphenicol-resistant but in vitro chloramphenicol-sensitive clones, genes that were constitutively expressed regardless of growth conditions were eliminated. Initially, after in vivo selection, the pool of recovered bacterial clones was screened for in vitro resistance or sensitivity to chloramphenicol. Isolated bacterial cells were replica plated on LB agar either with chloramphenicol or with ampicillin. All of the clones expressing the reporter gene in vitro were excluded, while the remaining clones were subjected to additional in vivo selection. After three successive rounds of in vivo and in vitro selection, only genes that were believed to be exclusively expressed in vivo were identified. This conclusion was confirmed by demonstrating that the reporter gene transcript (cat) was found only in mice infected with one of the clones and not when cells were grown in LB medium.

Although the ivi genes identified here were expressed in infected mouse liver, it is possible that specific sets of ivi genes may be induced in other host tissues. The method we described could readily be adopted to identify genes induced in other tissues. Since chromosomal integration was not required in this procedure, analysis of the cloned sequence is simple, because no subcloning steps are required. In addition, the induction of the reporter gene can easily be detected in infected tissues by RT-PCR with this modified IVET procedure.

With this modified IVET procedure using the mouse model of septicemic infection, 750 ivi-specific colonies were isolated and 168 have been sequenced. The E. coli ivi genes identified can be grouped as follows: (i) genes that encode enzymes required for biosynthetic pathways, (ii) genes induced in response to DNA damage as part of the SOS response, (iii) unclassified ORFs with possible functions assigned based on some degree of homology to known genes, (iv) hypothetical ORFs of unknown function, and (v) ORFs bearing no known homology to any sequences in the public databases. Among the biosynthetic genes, aroA, which is required for aromatic amino acid biosynthesis, is well characterized. Mutations in aroA have been shown to result in attenuation in S. enterica serovar Typhi, S. enterica serovar Typhimurium, and Neisseria gonorrhoeae (5, 8, 19). In addition, an aromatic biosynthetic mutant of the fish pathogen Aeromonas salmonicida has been reported to be attenuated (41). The fact that aroA was detected provides supportive evidence that the procedure indeed could be used to identify essential genes expressed in vivo. In this study, the aroA mutation created in E. coli i484 was not significantly attenuated. The low levels of attenuation of E. coli strain i484 ΔaroA must have been due to the limitations of the competitive assay, in which a mixture of biosynthetic mutant and wild-type strains was used. Since chorismate is the precursor for enterobactin biosynthesis (14), the aroA mutant, which is unable to synthesize chorismate, could utilize enterobactin synthesized by the wild-type strain used for coinfection.

Another ivi biosynthetic gene, metK, encodes S-adenosylmethionine synthetase. MetK plays an important role in cellular metabolism. In E. coli and S. enterica serovar Typhimurium it acts as a corepressor of genes encoding methionine biosynthetic enzymes (33). E. coli mutants impaired in S-adenosylmethionine synthetase have been shown to cause constitutive expression of the heat shock protein LysU (28). In addition to metK, E. coli also has a second gene for S-adenosylmethionine synthetase, metX (34). Both metK and metX are differentially regulated. metX is expressed in rich medium while metK is specifically induced in minimal medium (34). This observation further suggests the ivi nature of metK, which may be induced in the nutritionally depleted environment of mouse liver during septicemic infection, but when an ivi metK clone was grown on rich LB agar it could not induce the reporter cat gene.

Two of the identified E. coli ivi ORFs, o758 and f443, share 99 and 100% similarity with ATP-dependent proteases ClpB and HslV, respectively. Clp belongs to a highly conserved family of ATP-dependent proteases. This protease degrades unstable proteins and consequently modulates gene expression (44). For Bacillus subtilis it has been shown that the Clp protease is important in modulating stress responses and in modulating stationary-phase responses (23). No significant loss of virulence by the clpB mutant of E. coli strain i484 was observed in the in vivo mouse challenge-competition assay. The other protease, Hsl, is a heat shock protein that has been reported to play an important role in reversing SOS-mediated inhibition of cell division. Hsl degrades SulA, an SOS response cell division inhibitor (20). The Hsl-mediated degradation of SulA has been shown in vitro as well as in E. coli cells grown in LB medium (35, 36). It could be that the selective expression of ATP-dependent proteases in vivo is important for the survival of pathogenic E. coli in mouse liver.

Although previously known genes were identified as ivi genes, they were unable to express a reporter (cat) gene in vitro on rich medium that contained 25 μg of chloramphenicol per ml but could express the reporter gene in the livers of chloramphenicol-treated mice. The presence of cat gene transcripts in the livers of mice challenged with clones carrying bgl, aroA, or clp genes suggested that growth conditions found in mouse liver selectively induce these genes whereas in vitro growth on rich media simply inhibits the induction of these genes. It could be that specific growth conditions may be required to express these ivi genes in vitro, such as growth on medium with limiting nutrients. In fact, replicating certain in vivo growth conditions, such as iron and nutrition starvation, oxygen tension, and pH (29) has resulted in the identification of virulence-associated genes of bacterial pathogens. However, due to insufficient knowledge of the growth milieu that a pathogen encounters in the infected host, it is presently not possible to express all of the ivi genes in vitro. Regardless of the in vitro expression of the ivi genes, functional characterization of ivi genes will help elucidate their importance in pathogenesis and will provide better understanding of the highly complex nature of host-pathogen interaction. Since in vivo bacterial growth in mouse liver presents bacteria with unique sets of challenges, the expression of known ivi genes during septicemia might play important roles in enhancing in vivo bacterial survival and pathogenicity.

While identifying known genes involved in pathogenesis helps us to understand the process, the most interesting results are from the identification of sequences of unknown function. Thirty-two E. coli ORFs of unknown function were identified, although 18 of these ORFs have been assigned putative functions based on DNA or amino acid sequence similarity. Fourteen other ORFs, while being known based on previous sequencing projects, have no known functions. Ten ivi clones had no homology to sequences in the public databases, including E. coli K-12 and E. coli O157:H7. These clones could encode functions that are required to cause septicemia. One of these clones, ivi817, was mutagenized, and the virulence of the mutant was tested in the mouse competitive model. The mutant was highly attenuated (~28-fold), demonstrating the importance of the gene during septicemia. Similar results were obtained using a mutant of strain ivi815, carrying a hypothetical ORF of E. coli with unknown function. This mutant was attenuated ~81-fold. Mutants with mutations in ORF o761 and ORF V also were unable to survive effectively against the wild-type strain in an infection model. These results indicate that the uncharacterized ivi clones are important for bacterial virulence or in vivo survival. Therefore, this method has facilitated the identification of novel virulence-associated factors that may play crucial roles in E. coli pathogenesis. Future experiments to characterize these genes will truly reflect the importance of these gene products in causing disease and may provide novel targets to treat or control E. coli infection.

In summary, the isolated known ivi genes reported here encode various functions, from transport of macromolecules to metabolism, from DNA repair and replication to cell division, and from amino acid synthesis to polypeptide biosynthesis and degradation. Therefore, we believe that the ivi DNA sequences of clones that are collectively termed ORFs may encode many physiological functions to enable the pathogen to compete effectively in the tissues of infected hosts. Furthermore, DNA sequences of ivi clones that do not have homologues in the DNA databases are of utmost interest, as they may be required to cause septicemia. The IVET method we described here identified potentially new virulence factors that may be required for survival in liver tissues of the infected animal. The method can be used to identify tissue-specific ivi genes of pathogenic bacteria. The recovery of pathogens from the challenged animal at different time points and from various infected organs may also enhance our understanding of the infection process. The technique used allowed us to identify known sequences of unknown functions as well as totally novel sequences. The identification of these ORFs should lead to studies that will further enhance our understanding of disease and the growth and survival of pathogens in various habitats. Finally, among the ivi clones selected for further studies, when mutations were created in the cloned sequences, they served to attenuate mutants except for the clpB and aroA mutants. Coupled with the demonstration that the IVET technique does select for genes expressed exclusively in vivo, based on the results of RT-PCR, we believe that this tool will be important in furthering our understanding of disease and disease control.


We thank Richard Wilson, Pennsylvania State University, for serotyping E. coli strain i484.

This research work was supported by a grant from Johnson and Johnson, Corporate Office of Science and Technology.


Editor: J. T. Barbieri


1. Ausubel, F. M., R. Brent, R. E. Kingston, D. D. Moor, J. G. Seidman, J. A. Smith, and K. Struhl. 1997. Current protocols in molecular biology. p. 2.4.1-2.4.2. John Wiley and Sons, New York, N.Y.
2. Baine, W. B., W. Yu, and J. P. Summe. 2001. The epidemiology of hospitalization of elderly Americans for septicemia or bacteremia in 1991-1998: application of medical claim data. Ann. Epidemiol. 11:118-126. [PubMed]
3. Boyer, H. W., and D. Rouland-Dussiox. 1969. A complementation analysis of the restriction and modification of DNA in Escherichia coli. J. Mol. Biol. 41:459-472. [PubMed]
4. Brosius, J. 1984. Plasmid vector for the selection of promoters. Gene 27:151-160. [PubMed]
5. Chamberlain, L. M., R. Strugnell, G. Dougan, C. E. Hormaeche, and R. Demarco de Hormaeche. 1993. Neisseria gonorrhoeae strain MS11 harbouring a mutation in gene aroA is attenuated and immunogenic. Microb. Pathog. 15:51-63. [PubMed]
6. Coburn, J., and J. M. Leong. 2000. Arresting features of bacterial toxins. Science 290:287-288. [PubMed]
7. De Louvois, J. 1994. Acute bacterial meningitis in the newborn. J. Antimicrob. Chemother. 34(Suppl. A):61-73. [PubMed]
8. Dilts, D. A., I. Riesenfeld-Orn, J. P.Fulginiti, E. Ekwall, C. Granert, J. Nonenmacher, R. N. Brey, S. J. Cryz, K. Karlsson, K. Bergman, T. Thompson, B. Hu, A. H. Bruckner, and A. A. Lindberg. 2000. Phase I clinical trial of aroA aroD and aroA aroD htrA attenuated S. typhi vaccines; effect of formulation on safety and immunogenicity. Vaccine 18:1473-1484. [PubMed]
9. Finlay, B. B., and S. Falkow. 1989. Common themes in microbial pathogenicity. Microbiol. Rev. 53:210-230. [PMC free article] [PubMed]
10. Fuller, T. E., R. J. Shea, B. J. Thacker, and M. H. Mulks. 1999. Identification of in vivo induced genes in Actinobacillus pleuropneumoniae. Microb. Pathog. 27:311-327. [PubMed]
11. Geerdes, H. F., D. Ziegler, H. Lode, M. Hund, A. Loehr, W. Fangmann, and J. Wagner. 1992. Septicemia in 980 patients at a university hospital in Berlin: prospective studies during 4 selected years between 1979 and 1989. Clin. Infect. Dis. 15:991-1002. [PubMed]
12. Hamada, N., K. Watanabe, C. Sasakawa, M. Yoshikawa, F. Yoshimura, and T. Umemoto. 1994. Construction and characterization of a fimA mutant of Porphyromonas gingivalis. Infect. Immun. 62:1696-1704. [PMC free article] [PubMed]
13. Handfield, M., L. J. Brady, A. Progulske-Fox, and J. D. Hillman. 2000. IVIAT: a novel method to identify microbial genes expressed specifically during human infection. Trends Microbiol. 8:336-339. [PubMed]
14. Hantash, F. M., M. Ammerlaan, and C. F. Earhart. 1997. Enterobactin synthase polypeptides of Escherichia coli are present in an osmotic-shock-sensitive cytoplasmic locality. Microbiology 143:147-156. [PubMed]
15. Hensel, M., J. E. Shea, C. Gleeson, M. D. Jones, E. Galton, and D. W. Holden. 1995. Simultaneous identification of bacterial virulence genes by negative selection. Science 269:400-403. [PubMed]
16. Hunt, M. L., D. J. Boucher, J. D. Boyce, and B. Adler. 2001. In vivo expressed genes of Pasteurella multocida. Infect. Immun. 69:3004-3012. [PMC free article] [PubMed]
17. Isaacson, R. E. 1977. K99 surface antigen of Escherichia coli: purification and partial characterization. Infect. Immun. 15:272-279. [PMC free article] [PubMed]
18. Isaacson, R. E., and P. Richter. 1981. Escherichia coli 987P pilus: purification and partial characterization. J. Bacteriol. 146:784-789. [PMC free article] [PubMed]
19. Izhar, M., L. DeSilva., H. S. Joysey, and C. E. Hormaeche. 1990. Moderate immunodeficiency does not increase susceptibility to Salmonella typhimurium aroA live vaccines in mice. Infect. Immun. 58:2258-2261. [PMC free article] [PubMed]
20. Khattar, M. M. 1997. Overexpression of the hslVU operon suppresses SOS-mediated inhibition of the cell division in Escherichia coli. FEBS Lett. 414:402-404. [PubMed]
21. Kili, A. O., M. C. Herzberg, M. W. Meyer, X. Zhao, and L. Tao. 1999. Streptococcal reporter gene-fusion vector for identification of in vivo expressed genes. Plasmid 42:67-72. [PubMed]
22. Kreger, B. E., D. Craven, P. Carling, and W. McCabe. 1980. Gram-negative bacterium. III. Reassessment of etiology, epidemiology and ecology in 612 patients. Am. J. Med. 68:332-343. [PubMed]
23. Kruger, E., U. Volker, and M. Hecker. 1994. Stress induction of Bacillus subtilis and its involvement in stress tolerance. J. Bacteriol. 176:3360-3367. [PMC free article] [PubMed]
24. Lowe, A. M., D. T. Beattie, and R. L. Deresiewicz. 1998. Identification of novel staphylococcal virulence genes by in vivo expression technology. Mol. Microbiol. 27:967-976. [PubMed]
25. Lukashin, A. V., and M. Borodovsky. 1998. GeneMark.hmm: new solution for gene finding. Nucleic Acids Res. 26:1107-1115. [PMC free article] [PubMed]
26. Mahan, M. J., J. M. Slauch, and J. J. Mekalanos. 1993. Selection of bacterial virulence genes that are specifically induced in host tissues. Science 259:686-688. [PubMed]
27. Mahan, M. J., J. W. Tobias, J. M. Slauch, P. C. Hanna, R. J. Collier, and J. J. Mekalanos. 1995. Antibiotic-based selection for bacterial genes that are specifically induced during infection of a host. Proc. Natl. Acad. Sci. USA 92:669-673. [PMC free article] [PubMed]
28. Matthews, R. J., and F. C. Neidhardt. 1988. Abnormal induction of heat shock proteins in an Escherichia coli mutant deficient in adenosylmethionine synthesis activity. J. Bacteriol. 170:1582-1588. [PMC free article] [PubMed]
29. Mekalanos, J. J. 1992. Environmental signals controlling expression of virulence determinants in bacteria. J. Bacteriol. 174:1-7. [PMC free article] [PubMed]
30. Miller, V. L., and J. J. Mekalanos. 1988. A novel suicide vector and its use in construction of mutagenesis: osmoregulation of outer membrane proteins and virulence determinants in Vibrio cholerae ToxR. J. Bacteriol. 170:2575-2583. [PMC free article] [PubMed]
31. Plum, G., and J. E. Clark-Curtiss. 1994. Induction of Mycobacterium avium gene expression following phagocytosis by human macrophages. Infect. Immun. 62:476-483. [PMC free article] [PubMed]
32. Polissi, A., A. Pontiggia, G. Feger, M. Altieri, H. Mottl, L. Ferrari, and D. Simon. 1998. Large-scale identification of virulence genes from Streptococcus pneumoniae. Infect. Immun. 66:5620-5629. [PMC free article] [PubMed]
33. Satischandran, C., J. C. Taylor, and G. D. Markham. 1990. Novel Escherichia coli K-12 mutants impaired in S-adenosylmethionine synthesis. J. Bacteriol. 172:4489-4496. [PMC free article] [PubMed]
34. Satischandran, C., J. C. Taylor, and G. D. Markham. 1993. Isozymes of S-adenosylmethionine synthetase are encoded by tandemly duplicated genes in Escherichia coli. Mol. Microbiol. 9:835-846. [PubMed]
35. Seong, I. S., J. Y. Oh, J. W. Lee, K. Tanaka, and C. H. Chung. 2000. The HslU ATPase acts as a molecular chaperone in prevention of aggregation of SulA, an inhibitor of cell division in Escherichia coli. FEBS Lett. 477:224-228. [PubMed]
36. Seong, I. S., J. Y. Oh, S. J. Yoo, J. H. Seol, and C. H. Chung. 1999. ATP-dependent degradation of SulA, a cell division inhibitor, by the HslVU protease in Escherichia coli. FEBS Lett. 459:211-214. [PubMed]
37. Shires, G., and P. Dineen. 1982. Sepsis following burns, trauma, and intraabdominal infections. Arch. Intern. Med. 142:2012-2022. [PubMed]
38. Staendner, L. H., and M. Rohde. 1995. Identification of Salmonella typhi promoters activated by invasion of eukaryotic cells. Mol. Microbiol. 18:891-902. [PubMed]
39. Stanley, T. L., C. D. Ellermeier, and J. M. Slauch. 2000. Tissue-specific gene expression identifies a gene in the lysogenic phage Gifsy-1 that affects Salmonella enterica serovar Typhimurium survival in Peyer's patches. J. Bacteriol. 182:4406-4414. [PMC free article] [PubMed]
40. Unhanand, M., M. M. Mustafa, G. H. McCracken, Jr., and J. D. Nelson. 1993. Gram-negative enteric bacillary meningitis: a twenty years experience. J. Pediatr. 122:15-21. [PubMed]
41. Vaughan, L. M., P. R. Smith, and T. J. Foster. 1993. Aromatic-dependent mutant of fish pathogen Aeromonas salmonicida is attenuated in fish and is effective as a live vaccine against the salmonid disease furunculosis. Infect. Immun. 61:2172-2181. [PMC free article] [PubMed]
42. Wang, J., A. Mushegian, S. Lory, and S. Jin. 1996. Large scale isolation of candidate virulence genes of Pseudomonas aeruginosa by in vivo selection. Proc. Natl. Acad. Sci. USA 93:10430-10439. [PMC free article] [PubMed]
43. Weinstine, M. P., L. B. Reller, J. R. Murphy, and K. A. Lichtenstein. 1983. The clinical significance of positive blood culture: a comprehensive analysis of 500 episodes of bacteremia and fungemia in adults. I. Laboratory and epidemiologic observations. Rev. Infect. Dis. 5:35-53. [PubMed]
44. Wickner, S., M. R. Maurizi, and S. Gottesman. 1999. Posttranscriptional quality control: folding, refolding, and degrading proteins. Science 286:1888-1893. [PubMed]
45. Williams, P. H., and N. H. Carbonetti. 1986. Iron, siderophores, and the pursuit of virulence: independence of the aerobactin and enterobactin iron uptake systems in Escherichia coli. Infect. Immun. 51:942-947. [PMC free article] [PubMed]
46. Young, G. M., and V. L. Miller. 1997. Identification of novel chromosomal loci affecting Yersinia enterocolitica pathogenesis. Mol. Microbiol. 25:319-328. [PubMed]

Articles from Infection and Immunity are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...