Chapter 2.8. IVET and IVIAT technologies for the discovery of novel antigens


Chin Kai Ling, Phua Kia Kien

Art work
A simple diagnostic test to identify children who need antibiotic treatment 
could save hundreds of thousands lives in poor areas of the world.
I believe in health care as a human right.
Paul Farmer


Microbial pathogenesis in humans is multifaceted and dynamic, and is constantly evolving within the host. Following entry into the host, the pathogen has to survive in a hostile environment because the host is unlikely to remain passive when its own survival is threatened. Consequently, the circulatory system of the host carries killer cells, phagocytic cells, and complementary antibody molecules directed against bacteria, which it recognizes as foreign invaders. An attack on a pathogen results in microbial debris, which the host’s immune system processes and uses to build an adaptive immune defense, making it more difficult for the bacteria to colonize the host. Thus, during the course of an infection, many pathogen genes may be induced or repressed at different times to ensure survival and proliferation in the host. Therefore, the expression of pathogen virulence determinants is unlikely to be observed wholly under in vitro growth conditions in the laboratory, because it is impossible to mimic all the complex and changing environmental stimuli that occur at the site of infection (in vivo). Because it is unethical to infect live animals with pathogenic strains of bacteria, a method has to be devised to circumvent the in vivo requirement without subjecting experimental animals to pain and suffering.

Recent advances in DNA recombination technology have provided two alternative ways to identify in vivo-induced (IVI) genes that are specifically upregulated in vivo and are always associated with the in vivo survival and pathogenicity of the pathogen. In vivo expression technology (IVET) and in vivo-induced antigen technology (IVIAT) were established to study in a more realistic manner the dynamic phenomenon occurring between hosts and pathogens during infection. A deeper understanding of host–pathogen molecular interactions facilitates the discovery of new molecular biomarkers that are more relevant to a disease and improved its diagnosis, prognosis, and treatment.

In vivo expression technology (IVET)

IVET is used to identify bacterial virulence genes that are expressed in a live host with an intact immune system, and which eventually help the bacteria to colonize this niche environment [1]. In other words, this approach is used to identify gene products that are turned off when the pathogen is outside the host but are turned on during infection of the host. Two major components are required for IVET, namely, a bacterial strain that has a mutated gene, and a plasmid that carries the promoter trap [2]. This gene-trapping cassette consists of a promoterless gene and a transcriptionally linked reporter gene (rep) (Figure 1). The mutated strain is not able to grow without an active promoter that is randomly cloned into an exon [3]. The transcriptional fusion of an active promoter with the reporter gene induces gene expression in a specific niche environment. Many variations of IVET have been reported that depend on the reporter gene being used. Examples include auxotrophy-based selection [4], antibiotic resistance-based selection [5], and system-specific selection for gene expression during a particular stage of infection [6] (Figure 1). Angelichio and Camilli (2002) reported that using IVET, many unique IVI genes were induced during infections in gram-negative bacteria, gram-positive bacteria, and fungi, and some were essential for virulence. The advantage of IVET is the strength of the positive selection strategy, which helps to isolate the genes that are constitutively and highly expressed in the test environment [2]. The IVI genes are detected using blue/white screening on X-gal gel; the process depends on the lac operon. The lac gene in the plasmid encodes LacZ protein (Lac+) and forms a functional β-galactosidase enzyme that metabolizes X-gal, a colorless modified galactose sugar, to form 5-bromo-4-chloro-indoxyl, which is then spontaneously oxidized to an insoluble blue pigment, 5,5ʹ-dibromo-4,4ʹ-dichloro-indigo. Upon insertion of an active IVI gene into the plasmid, bacterial cells lose their ability to produce LacZ (Lac-) and cannot hydrolyze the sugar; hence, white colonies are formed on the plate [7].

Because IVET only identifies genes that are highly expressed during growth in the host environment, the identification of weakly or transiently expressed genes may be hindered. This disadvantage of IVET screening is circumvented by a modification known as recombinase-based in vivo expression technology (RIVET). Commonly, a resolvase gene such as tnpR that is involved in DNA recombination is used as the reporter gene. A gene cassette that contains an antibiotic resistance gene (e.g., the gene that encodes resistance to tetracycline) flanked by two resolvase recognition sequences (res1) is integrated into the bacterial chromosome at a neutral site in the genome (Figure 1). Once an active promoter has been introduced into the plasmid, the promoter fuses with and directs the transcription of tnpR, resulting in resolvase activity; the antibiotic resistance gene is excised, creating an antibiotic-sensitive phenotype [8]. Using antibiotic (tetracycline) sensitivity testing, resolved plasmids containing promoters that are active in vitro and are sensitive to antibiotics are discarded, whereas unresolved plasmids that are resistant to antibiotics form bacterial colonies on the plate. The unresolved plasmids are then used to infect a suitable host. Active promoters that interact with the host are screened to identify the clones containing IVI genes, which are resolved in the in vivo environment. This step is rather laborious because resolved clones lose positive selection bias, and antibiotic sensitivity screening via replica plating is required. Livny and Friedman (2004) tried to address this issue by creating a selectable in vivo expression technology (SIVET) whereby another antibiotic resistance gene is fused (e.g., the gene that encodes resistance to chloramphenicol, cat) beside the normal gene cassette (res1-tet-res1) to form a culture-selectable gene cassette (res1-tet-res1-cat). This chloramphenicol gene is disrupted by the normal gene cassette. Thus, during the antibiotic screening process for the IVI genes, unresolved strains remain tetracycline-resistant and chloramphenicol-sensitive, whereas resolved strains are sensitive to tetracycline and resistant to chloramphenicol [9]. RIVET has been successfully used to identify temporally activated IVI genes during infection in Enterococcus faecalis [10], Sinorhizobium meliloti [11], Escherichia coli, and Burkholderia tropica [12]. However, owing to the sensitivity of RIVET, bacteria with genes expressed at moderate or high levels during in vitro growth, which are nevertheless important for in vivo infection, will also have resolvase activity, and will therefore losing their antibiotic resistance and will not be identified [13].

In vivo expression technology–immunoprecipitation (IVET-IP) was created to overcome the disadvantages of IVET and RIVET. It is based on a reporter gene that comprises a fusion of the c-Myc gene with the gene that encodes the bacterial outer membrane protein. c-Myc-tagged cells are obtained by immunoprecipitation with anti-c-Myc antibody. This does not cause selective pressure on the identified cells because all the genes that are highly or weakly expressed can be identified, owing to the high sensitivity of the immunoprecipitation method. In a study using IVET-IP, purified c-Myc-tagged cells were screened using a microarray system, and 173 IVI genes were identified in Bordetella bronchiseptica infection [14]. These genes encode proteins that are important in pathogen metabolism, regulation of gene expression via transcription and translation, and interactions with the host environment, such as the export or secretion of membrane proteins. The researchers also found that 17 of the 173 IVI genes are unique to B. bronchiseptica, and have the potential to serve as specific biomarkers for the disease.

Figure 1. Schematic overview of the five main in vivo expression technology (IVET) selection strategies. Reporter gene fusion libraries are made by ligating random genomic fragments (designated gene X’) into the IVET vector of choice, which is then transformed into Escherichia coli for generation of a random library. The resulting plasmid is then extracted and transformed into bacteria for chromosome integration. The auxotrophy-based selections contain the promoterless egf gene, which encodes an essential growth factor, and surviving strains contain IVI-purA fusions owing to the requirements of auxotrophy complementation. Antibiotic resistance-based selections contain the promoterless AbR gene, and surviving strains contain IVI-cat fusions owing to the requirements of antibiotic resistance. System specific-based selections contain the promoterless recombinase gene (rec), and surviving strains contain IVI-hly fusions owing to the requirements of Hly expression to escape phagosomes. Recombinase-based in vivo expression technology (RIVET) involves the reporter gene, tnpR, whose protein product excises a resolvase substrate cassette (res1-tet-res1). Prescreening is required to remove strains harboring gene fusions that are active in vitro. Tetracycline-sensitive strains contain IVI-tnpR fusions because excision and loss of the antibiotic marker is recovered via replica plating [1]. In vivo expression technology–immunoprecipitation (IVET-IP) involves a promoterless c-Myc-tagged brkA gene, and surviving strains expressing IVI-brkA fusions with a c-Myc tag peptide on the bacterial surface are immunoprecipitated with anti-c-Myc antibody [14]. The IVI genes that are induced during infection in the host can be recovered by screening for Lac- colonies (white). (Modified from [1, 14]).

In Vivo-Induced Antigen Technology (IVIAT)

IVIAT is an immunoscreening technique that helps to identify bacterial antigens that are specifically expressed inside the host (in vivo) during an infection, but not when the bacteria are grown outside the host (in vitro) under standard laboratory conditions, or when the genes are expressed significantly higher in vivo than in vitro [15]. The method allows direct identification of microbial proteins without the need for an animal model [16]. Numerous studies have reported the presence of IVI proteins in bacteria such as Mycobacterium tuberculosis [17], E. coli [18], and Salmonella enterica serovar Typhi [19]; parasites such as Toxoplasma gondii [20, 21]; and fungi such as Candida albicans [22]. This shows that IVIAT is applicable to prokaryotic, eukaryotic, and multi-cellular pathogens. These IVI proteins are involved in the pathogenesis of the disease and adaptation of the pathogen in the in vivo environment of the host, and could therefore be exploited as potential biomarkers for the disease.

Generally, IVIAT facilitates the efficient screening of a whole genome without prior knowledge of the basic genetics of the pathogen [23]. This is because random protein expression libraries are generated from the pathogen’s DNA/cDNA fragments, and identification of the clones that produce the recombinant proteins is based on antigen–antibody reactivity. In contrast, IVET requires selection of a promoterless reporter gene during construction of the promoter trap to identify specific IVI genes. Thus, IVIAT is useful for the study of new emerging pathogens where no reliable source of genetic information is available for the pathogen. Furthermore, the expression libraries can be constructed using multiple strains of the pathogen. Inclusion of unique genes from each individual strain is important because multiple-virulence strains can coexist in a single disease [23].

The expression library can be constructed using either expression vectors or a bacteriophage lambda-based expression system, as illustrated in Figure 2. A major concern regarding IVIAT is the efficiency of transformation and the variability of protein expression in the E. coli host expression system, owing to codon bias, restriction-site limitations (if the library is generated using restriction digestion), difficulty with membrane protein expression, and the toxicity of overexpressed proteins [23]. Bacteriophages have certain advantages over plasmids in expressing foreign genes during library construction such as the efficiency of packaging (transduction) and infecting E. coli (transfection), the ease of amplification, and the simplicity of plaque screening. However, plasmids show higher protein expression levels and are easier to purify [24]. The insertion size of the DNA should be between 500 and 1,500 bases, because larger inserts do not express well.

To address variations in individual immune response, sera are pooled from individuals infected via different routes of transmission, or affected by different courses of infection or clinical outcomes, and used as probes (the source of antibodies) against the generated protein expression library to identify the antigenic proteins expressed during infection [25]. IVIAT also permits the identification of antigens expressed at various stages of infection. Studies conducted by Amerizadeh et al. (2013a; 2013b) to identify IVI antigens of the parasite Toxoplasma gondii showed that different IVI proteins were expressed during the acute [21] and chronic [26] stages of toxoplasmosis. Although T. gondii-infected patients are commonly asymptomatic and the disease can be resolved without treatment, eradication of the parasite is important because it may cause severe mental problems, especially in newborns and immunocompromised patients. Pre-adsorption of the pooled sera against in vitro-grown bacteria, namely the pathogen of interest and E. coli, is carried out to enrich antibodies against IVI antigens (Figure 2, Step 2). This helps to eliminate antibodies that bind to the in vitro-grown pathogen and cross-reactive antibodies, which bind to proteins in the E. coli expression host. However, this step may be a major disadvantage of IVIAT because antibodies that are reactive to both in vivo- and in vitro-expressed pathogen proteins are removed, thereby giving a false negative association with these virulence-associated antigens [15].

Theoretically, when probing the expression library with enriched antibodies against IVI antigens (Figure 2, Step 3), recombinant proteins that form an immunoprecipitate with the antibodies are considered IVI antigens. However, the antibody-binding activity of the IVI antigens is a marker of protein expression in vivo, and does not reflect the protective efficacy of the host’s humoral immune response. Furthermore, patients might have been exposed to a variety of organisms and infections throughout their life that share antigen cross-reactivity with the current bacteria [23], and the antibody-binding activity may be due to immune responses to closely related antigens [27]. Thus, the product of an IVI gene may not be immunogenic (protective efficacy). It is therefore advisable that the genome of the pathogen should be compared with the genomes of other pathogens to rule out cross-reactivity. The false positive result that may arise owing to this phenomenon is subsequently confirmed in Step 5 (Figure 2).

Although the use of bacteriophage lambda has increased the efficiency of generating primary libraries, plasmids have major advantages over bacteriophage vectors during the insertion stage (Figure 2, Step 4). First, the large size of the bacteriophage typically complicates the restriction mapping and DNA sequencing of the identified genes. Second, proteins are expressed at a lower level in the bacteriophage and are difficult to purify. The problem facing bacteriophage propagation is that after infection of an E. coli cell, foreign protein can only be made for about 50 minutes before the host cell lyses. Therefore, a lower yield of expressed protein is obtained from the crude cell lysate [24]. Therefore, subcloning to transfer the DNA fragments from the bacteriophage to a plasmid using an in vivo excision technique is required for the identification of the gene using a DNA sequencing method, and subsequent production of sufficient protein for confirmatory analysis [28]. As described by Amerizadeh et al. (2013), this excision technique is carried out by mixing the selected phage clone with E. coli XLBlue MRFʹ cells and an ExAssist helper phage is used for in vivo excision of the phagemid from the bacteriophage. The phagemids are mixed with SOLR cells, from a non-suppressing E. coli strain for phagemid colony amplification [26].

Many studies conducted using techniques such as real-time reverse transcription polymerase chain reaction (RT-PCR) [15], immunofluorescent detection in infected human tissues utilizing monoclonal antibodies raised in mice, rabbits, or guinea pigs [29], and microarray analysis [30] have confirmed the involvement of the IVI genes identified by IVIAT. In a study conducted by Amerizadeh et al. (2013) on acute toxoplasmosis, 29 IVI genes were identified from each IgM and IgG immunoglobulin class. Quantitative analysis using RT-PCR to measure protein expression levels showed that 20 of the IgM-detected genes and 11 of the IgG-detected genes were upregulated in vivo compared with in vitro. Some of the proteins were hypothetical with unknown functions; others were membrane proteins, or virulence proteins important for T. gondii pathogenesis processes such as motility, attachment, invasion, and signal transduction, which could induce high levels of antibody during an infection. This shows that IVIAT is a useful technique for the identification of immunogenic proteins, which are expressed specifically during in vivo host infections, and that virulence proteins are better biomarkers for the future diagnosis of acute toxoplasmosis.

Table 1. Comparison of in vivo expression technology (IVET) with in vivo-induced antigen technology (IVIAT) [1, 23, 25]

Figure 2. Schematic diagram of in vivo-induced antigen technology (IVIAT) for Salmonella Typhi. Step 1: The expression library of the S. Typhi DNA/cDNA is generated either using expression vector (pET30a, b and c) or a bacteriophage. The expression vectors with inserts are transformed into expression host cells such as Escherichia coli strain BL21 (DE3). Proteins are expressed in the agar plate by isopropyl β-D-1-thiogalactopyranoside (IPTG) induction, and a copy of the colonies is transferred onto a nitrocellulose membrane. The membrane is exposed to chloroform vapor for 15 s to lyse the E. coli cells and release the proteins. The DNA/cDNA fragments that have been transducted into the bacteriophage are transfected into the dH5α E. coli system. Propagation of the bacteriophage in the host results in E. coli cell lysis and formation of a plaque. A copy of the plaque is transferred to a nitrocellulose membrane pre-saturated with IPTG, which is incubated overnight for protein expression. Step 2: Sera from infected patients are pooled and pre-adsorbed with cells (whole cells or cell lysate) of laboratory-grown (in vitro) S. Typhi and E. coli to enrich antibodies against IVI antigens. Step 3: Expressed proteins bound to the nitrocellulose membranes are probed with the pre-adsorbed serum and IVI antigens are detected by western blotting. Step 4: The reactive colonies contain plasmids, whereas the reactive plaques contain the recombinant bacteriophage vectors. These bacteriophages are converted to recombinant plasmids using an in vivo excision technique. The plasmids are then purified and their DNA sequenced to identify the gene sequences, which encode the targeted IVI antigens. Step 5: Large-scale protein expression is carried out to obtain the antigens in high quantity, and used to verify the IVI antigens expressed by the pathogen during infection. (Modified from Rollins et al. (2005), Handfield et al. (2000b), and Amerizadeh et al. (2013b) [23, 25, 26] [21, 23, 25]).


IVET and IVIAT are gene fusion techniques that provide an alternative means of studying complex phenotypes. They facilitate the identification of microbial in vivo-induced genes and proteins expressed uniquely during infection of the host. This could improve our understanding of the mechanisms behind bacterial colonization in a host or even in a specific niche, in the case of IVET. Despite their limitations, these techniques remain useful for identifying the genes/proteins expressed in vivo that are important for the virulence and survival of the pathogen in the host. In time, these technologies will become the platform of choice for new IVI biomarker discovery..


  1. Angelichio MJ, Camilli A (2002) In vivo expression technology. Infect Immun 70:6518-6523.
  2. Rediers H, Rainey PB, Vanderleyden J, De Mot R (2005) Unraveling the secret lives of bacteria: use of in vivo expression technology and differential fluorescence induction promoter traps as tools for exploring niche-specific gene expression. Microbiol Mol Biol Rev 69:217-261.
  3. Springer PS (2000) Gene traps: tools for plant development and genomics. Plant Cell 12:1007-1020.
  4. Handfield M, Lehoux DE, Sanschagrin F, Mahan MJ, Woods DE, Levesque RC (2000a) In vivo-induced genes in Pseudomonas aeruginosa. Infect Immun 68:2359-2362.
  5. Mahan MJ, Tobias JW, Slauch JM, Hanna PC, Collier RJ, Mekalanos JJ (1995) Antibiotic-based selection for bacterial genes that are specifically induced during infection of a host. Proc Natl Acad Sci U S A 92:669-673.
  6. Gahan CG, Hill C (2000) The use of listeriolysin to identify in vivo induced genes in the gram-positive intracellular pathogen Listeria monocytogenes. Mol Microbiol 36:498-507.
  7. Padmanabhan S, Banerjee S, Mandi N (2011) Screening of Bacterial Recombinants: Strategies and Preventing False Positives. In: Molecular Cloning - Selected Applications in Medicine and Biology. Brown GG (ed.). InTech. doi: 10.5772/22140.
  8. Merighi M, Ellermeier CD, Slauch JM, Gunn JS (2005) Resolvase-in vivo expression technology analysis of the Salmonella enterica serovar Typhimurium PhoP and PmrA regulons in BALB/c mice. J Bacteriol 187:7407-7416.
  9. Livny J, Friedman DI (2004) Characterizing spontaneous induction of Stx encoding phages using a selectable reporter system. Mol Microbiol 51:1691-1704.
  10. Frank KL, Barnes AM, Grindle SM, Manias DA, Schlievert PM, Dunny GM (2012) Use of recombinase-based in vivo expression technology to characterize Enterococcus faecalis gene expression during infection identifies in vivo-expressed antisense RNAs and implicates the protease Eep in pathogenesis. Infect Immun 80:539-549.
  11. Gao M, Teplitski M (2008) RIVET-a tool for in vivo analysis of symbiotically relevant gene expression in Sinorhizobium meliloti. Mol Plant Microbe Interact 21:162-170.
  12. Lozano MJ, Salas ME, Giusti Mde L, Martini MC, López JL, Salto I, Del Papa MF, Pistorio M, Lagares A (2013) Novel tnpR-based transposable promoter traps suitable for RIVET studies in different gram-negative bacteria. J Microbiol Methods 93:9-11.
  13. Osorio CG, Crawford JA, Michalski J, Martinez-Wilson H, Kaper JB, Camilli A (2005) Second-generation recombination-based in vivo expression technology for large-scale screening for Vibrio cholerae genes induced during infection of the mouse small intestine. Infect Immun 73:972-980.
  14. Abe H, Kamitani S, Fukui-Miyazaki A, Shinzawa N, Nakamura K, Horiguchi Y (2015) Detection of genes expressed in Bordetella bronchiseptica colonizing rat trachea by in vivo expressed-tag immunoprecipitation method. Microbiol Immunol 59:249-261.
  15. Gu H, Zhu H, Lu C (2009) Use of in vivo-induced antigen technology (IVIAT) for the identification of Streptococcus suis serotype 2 in vivo-induced bacterial protein antigens. BMC Microbiol 9:201.
  16. An R, Grewal PS (2012) Comparative study of differential gene expression in closely related bacterial species by comparative hybridization. Methods Mol Biol 815:103-119.
  17. Li T, Gao ZY, Wang HL, Feng EL, Chen ZD, Li XY, Huang LY, Huang CF (2005) [Screen in vivo induced gene of Mycobacterium tuberculosis by IVIAT]. Yi Chuan Xue Bao 32:111-117.
  18.  John M, Kudva IT, Griffin RW, Dodson AW, McManus B, Krastins B, Sarracino D, Progulske-Fox A, Hillman JD, Handfield M, Tarr PI, Calderwood SB (2005) Use of in vivo-induced antigen technology for identification of Escherichia coli O157:H7 proteins expressed during human infection. Infect Immun 73:2665-2679.
  19. Charles RC, Sultana T, Alam MM, Yu Y, Wu-Freeman Y, Bufano MK, Rollins SM, Tsai L, Harris JB, LaRocque RC, Leung DT, Brooks WA, Nga TV, Dongol S, Basnyat B, Calderwood SB, Farrar J, Khanam F, Gunn JS, Qadri F, Baker S, Ryan ET (2013) Identification of immunogenic Salmonella enterica serotype Typhi antigens expressed in chronic biliary carriers of S. Typhi in Kathmandu, Nepal. PLoS Negl Trop Dis 7:e2335.
  20. Tao Q, Xiao J, Wang Y, Fang K, Li N, Hu M, Zhou Y, Zhao J (2014) Identification of genes expressed during Toxoplasma gondii infection by in vivo-induced antigen technology (IVIAT) with positive porcine sera. J Parasitol 100:470-479.
  21. Amerizadeh A, Khoo BY, Teh AY, Golkar M, Abdul Karim IZ, Osman S, Yunus MH, Noordin R (2013a) Identification and real-time expression analysis of selected Toxoplasma gondii in-vivo induced antigens recognized by IgG and IgM in sera of acute toxoplasmosis patients. BMC Infect Dis 13:287.
  22. Cheng S, Clancy CJ, Checkley MA, Handfield M, Hillman JD, Progulske-Fox A, Lewin AS, Fidel PL, Nguyen MH (2003) Identification of Candida albicans genes induced during thrush offers insight into pathogenesis. Mol Microbiol 48:1275-1288.
  23. Rollins SM, Peppercorn A, Hang L, Hillman JD, Calderwood SB, Handfield M, Ryan ET (2005) In vivo induced antigen technology (IVIAT). Cell Microbiol 7:1-9.
  24.  Christensen AC (2001) Bacteriophage lambda-based expression vectors. Mol Biotechnol 17:219-224.
  25. Handfield M, Brady LJ, Progulske-Fox A, Hillman JD (2000b) IVIAT: a novel method to identify microbial genes expressed specifically during human infections. Trends Microbiol 8:336-339.
  26. Amerizadeh A, Idris ZM, Khoo BY, Kotresha D, Yunus MH, Karim IZ, Saadatnia G, Teh AY, Noordin R (2013b) Identification of Toxoplasma gondii in-vivo induced antigens by cDNA library immunoscreening with chronic toxoplasmosis sera. Microb Pathog 54:60-66.
  27. Rollins SM, Peppercorn A, Young JS, Drysdale M, Baresch A, Bikowski MV, Ashford DA, Quinn CP, Handfield M, Hillman JD, Lyons CR, Koehler TM, Calderwood SB, Ryan ET (2008) Application of in vivo induced antigen technology (IVIAT) to Bacillus anthracis. PLoS One 3:e1824.
  28. Short JM, Fernandez JM, Sorge JA, Huse WD (1988) Lambda ZAP: a bacteriophage lambda expression vector with in vivo excision properties. Nucleic Acids Res 16:7583-7600.
  29. Handfield M, Seifert T, Hillman JD (2003) In vivo expression of bacterial genes during human infections. Methods Mol Med 71:225-242.
  30. Larocque RC, Harris JB, Dziejman M, Li X, Khan AI, Faruque AS, Faruque SM, Nair GB, Ryan ET, Qadri F, Mekalanos JJ, Calderwood SB (2005) Transcriptional profiling of Vibrio cholerae recovered directly from patient specimens during early and late stages of human infection. Infect Immun 73:4488-4493.