Applied Microbiology and Biotechnology

, Volume 93, Issue 6, pp 2291–2300

Strategies to overexpress enterotoxigenic Escherichia coli (ETEC) colonization factors for the construction of oral whole-cell inactivated ETEC vaccine candidates


    • Gothenburg University Vaccine Research Institute (GUVAX) and Department of Microbiology and Immunology, The Sahlgrenska AcademyUniversity of Gothenburg
  • Ann-Mari Svennerholm
    • Gothenburg University Vaccine Research Institute (GUVAX) and Department of Microbiology and Immunology, The Sahlgrenska AcademyUniversity of Gothenburg

DOI: 10.1007/s00253-012-3930-6

Cite this article as:
Tobias, J. & Svennerholm, A. Appl Microbiol Biotechnol (2012) 93: 2291. doi:10.1007/s00253-012-3930-6


Enterotoxigenic Escherichia coli (ETEC) is an important cause of diarrheal disease and deaths among children in developing countries and the major cause of traveler's diarrhea (TD). Since surface protein colonization factors (CFs) of ETEC are important for pathogenicity and immune protection is mainly mediated by locally produced IgA antibodies in the gut, much effort has focused on the development of an oral CF-based vaccine. The most extensively studied ETEC candidate vaccine is the rCTB-CF ETEC vaccine, containing recombinantly produced cholera B subunit and the most commonly encountered ETEC CFs on the surface of whole inactivated bacteria. Initial clinical trials with this vaccine showed significant immune responses against the key antigens in different age groups in Bangladesh and Egypt and protection against more severe TD in Western travelers. However, when tested in a phase-III trial in Egyptian infants, the protective efficacy of the vaccine was found to be low, indicating the need to improve the immunogenicity of the vaccine, e.g., by increasing the levels of the protective antigens. This review describes different strategies for the construction of recombinant nontoxigenic E. coli and Vibrio cholerae candidate vaccine strains over-expressing higher amounts of ETEC CFs than clinical ETEC isolates selected to produce high levels of the respective CF, e.g., those ETEC strains which have been used in the rCTB-CF ETEC vaccine. Several different expression vectors containing the genes responsible for the expression and assembly of the examined CFs, all downstream of the powerful tac promoter, which could be maintained either with or without antibiotic selection, were constructed. Expression from the tac promoter was under the control of the lacIq repressor present on the plasmids. Following induction with isopropyl-β-d-thiogalactopyranoside, candidate vaccine strains over-expressing single CFs, unnatural combinations of two CFs, and also hybrid forms of ETEC CFs were produced. Specific monoclonal antibodies against the major subunits of the examined CF were used to quantify the amount of the surface-expressed CF by a dot-blot assay and inhibition ELISA. Oral immunization with formalin- or phenol-inactivated recombinant bacteria over-expressing the CFs was found to induce significantly higher antibody responses compared to immunization with the previously used vaccine strains. We therefore conclude that our constructs may be useful as candidate strains in an oral whole-cell inactivated CF ETEC vaccine.


DiarrheaETECColonization factorsRecombinant over-expressionNontoxigenic strainsNonantibiotic selection markerOral vaccine


Enterotoxigenic Escherichia coli (ETEC) is the most common cause of diarrhea in the developing world, resulting in ca 20% of all diarrheal episodes in children and in 20–50% of all diarrheas among travelers to these areas (Wolf 1997; Wennerås and Erling 2004; Qadri et al. 2005). Following ingestion of contaminated food or water, ETEC bacteria colonize the small-bowel mucosa with the help of surface structures called colonization factors (CFs) (Gaastra and Svennerholm 1996; Qadri et al. 2005) that bind to specific glycoprotein or glycolipid receptors on host epithelial cells (Jansson et al. 2006, 2009). Upon infection with ETEC, watery diarrhea may result from the release and actions of either or both of a heat-labile (LT) or heat-stable toxin (ST) (Sánchez and Holmgren 2005). LT is structurally, functionally, and antigenically very similar to cholera toxin (CT), whereas ST is a nonrelated and nonimmunogenic molecule.

E. coli strains of more than 100 different O groups have been identified, although some O groups are more prevalent than others, but there are also large geographic differences (Wolf 1997). Protective immunity of E. coli O antibodies has been demonstrated in experimental studies but only against ETEC expressing homologous O groups (Svennerholm 2011). Hence a vaccine based on E. coli O groups would be very complicated. So far, more than 23 CFs have been recognized on human ETEC isolates. These surface protein CFs are key virulence factors of ETEC, which promote colonization of the bacteria in the small intestine, allowing close proximity of the expressed toxins to the intestinal epithelium. ETEC CFs are mainly fimbrial or fibrillar surface proteins, although some do not conform to this general pattern (Gaastra and Svennerholm 1996; Qadri et al. 2005). Among the latter CFs is CS6, which is nonfimbrial in structure and does not appear to protrude from the cell surface (Gaastra and Svennerholm 1996; Tobias et al. 2008a). Different CFs have been found on ETEC in varying frequencies in different geographic areas, during different seasons, and in different categories of patients (Qadri et al. 2005). Of the wide range of CFs, the most commonly present on diarrheagenic strains are CFA/I, CS1, CS2, CS3, CS4, CS5, CS6, and in some studies, also CS7, CS14, CS17, and CS21 (Gaastra and Svennerholm 1996; Qadri et al. 2005). Several of these CFs may be expressed on the same bacteria. For example, CS1 or CS2 is generally expressed together with CS3, whereas some strains express CS3 alone. Similarly, CS4 or CS5 is generally expressed together with CS6, whereas there are many strains that express only CS6 (Gaastra and Svennerholm 1996). CFA/I has not been found to be naturally coexpressed with either of CS1–CS6 (Gaastra and Svennerholm 1996; Qadri et al. 2005).

Molecular and structural studies have shown that fimbrial CFs, such as CFA/I, CS1, CS2, and CS4 are rigid fimbrial rods with a diameter of 7 nm which consist of structural or major subunit proteins of 15–30 kDa. Each fimbriae may contain 100–1000 of such subunits (Gaastra and Svennerholm 1996; Li et al. 2009). CS3 and CS5 have fibrillar or fimbrial flexible morphologies with a diameter of 2–6 nm, respectively (Gaastra and Svennerholm 1996). It has been shown that native expression of ETEC CFs is optimally achieved at 37 °C (Viboud et al. 1993; Cassels and Wolf 1995) and in CFA medium (Tobias et al. 2008b), although some CFs such as CS5, CS7, CS12, CS14, and CS17 require addition of bile salts (Cassels and Wolf 1995; Tobias et al. 2010b).

Studies have also shown high level of genetic relatedness among several ETEC CFs, particularly in the N-terminal region of the major subunit proteins as well as in the tip protein (Gasstra and Svennerholm 1996; Li et al. 2009). One group of such related fimbriae, the so-called CFA/I family, consists of eight antigenically and structurally related members, CFA/I, CS1, CS2, CS4, CS14, CS17, CS19, and PCFO71 (Gaastra and Svennerholm 1996; Anantha et al. 2004; Tobias et al. 2010b). Additional prominent group of ETEC CFs is the CS5 family, consisting of CS5, CS7, CS18, and CS20 (Gaastra and Svennerholm 1996; Qadri et al. 2005).

Several of ETEC CFs, e.g., CFA/I, CS1, CS2, and CS4, are organized in operons consisting of two genes which encode structural major and minor (or tip) subunits and two genes which encode a chaperon and an usher protein; the latter genes are responsible for folding, transport, and correct assembly of the subunits to form the complete structures of the fimbriae (Table 1; Gaastra and Svennerholm 1996). On the other hand, the operon of the fibrillar CS5 contains six genes, which encode the major subunit, two minor pillin subunits, two chaperons (each specific for one of the minor pillin subunits), and an usher protein (Duthy et al. 2002). Molecular studies of CS6 have shown that its operon also contains genes encoding a chaperon (cssC) and an usher (cssD); but unlike other CFs, the operon contains two genes which encode two major structural subunits (cssA and cssB) (Wolf et al. 1997). The operon of CS3 consists of a cluster of genes, i.e., CstA-H, containing one or two nearly identical major subunits, a chaperon, and an usher protein (Yakhchali and Maning 1997).
Table 1

Structural and assembly-encoding genes in the operons of different ETEC CFs


Structural and assembly-encoding genes

Accession number

Major subunit/s

Minor subunit/s























csfD, csfE

csfF, csfB




cssA, cssB

Not identified




CFs are usually plasmid-born, unlike CS2 which is located on the ETEC chromosome (Caron and Scott 1990; Perez-Casal et al. 1990). Expression of most CFs is regulated by various positive regulators, such as CfaR (or CfaD) which regulates expression of CFA/I (Caron and Scott 1990); Rns, of CS1 and CS2 (Caron et al. 1989); CsfR, of CS4 (Cassels and Wolf 1995); and CsvR, of CS5 (de Haan et al. 1991). Rns and CfaR have been shown to activate transcription of CF operons by acting on the transcription suppression mediated by the global regulator H-NS (Jordi et al. 1994). Unlike for other ETEC CFs, a regulator for expression of CS6 has not been identified. It was initially concluded by Caron and Scott (1990) that CS3 expression is not under positive regulation. However, it was later on shown that expression of CS3 in Vibrio cholerae is under regulation (Favre et al. 2006), suggesting that CS3 expression, after all, might be regulated also in ETEC.

Opinions differ regarding the contribution of the major and the minor (tip) protein subunits of the fimbriae in the CFA/I family to the pathogenicity of the bacteria and their roles as protective antigens. The tip proteins have the ability to bind to receptors on different classes of erythrocytes, although the nature of the receptors and their physiological relevance has yet to be demonstrated (Li et al. 2007, 2009). At the same time, we have shown that the major subunit of the CFA/I fimbriae is able to bind to a number of sugar structures present in the human small intestine and that this binding is independent of the tip protein (Jansson et al. 2006). Furthermore, we could previously show that monoclonal antibodies against the N-terminal region of the major subunit of CFA/I could inhibit the binding of CFA/I and CS4 expressing ETEC to isolated human jejunal enterocytes, as well as afford passive protection against a challenge with such bacteria in rabbits (Rudin et al. 1996).

Based on the proven efficacy of the CF and LT antigens, we conclude that an ETEC vaccine should most likely contain CFs present on the most prevalent ETEC pathogens, particularly on strains producing LT+ST or ST alone and a nontoxic LT antigen to provide broad spectrum protection (Svennerholm and Tobias 2008; Svennerholm and Glenn 2010). Thus, cholera B subunit, which cross-reacts significantly with ETEC LT, has been shown to provide highly significant protection against LT and LT/ST ETEC both in travelers and in ETEC endemic areas. Strong evidence for a protective effect of CFs are found in experimental studies in animals (as reviewed by Svennerholm 2011), but even more important studies in challenged volunteers (Levine et al. 1979) and in a birth cohort in a highly ETEC endemic area that has reinfections with ETEC expressing homologous CFs are very rare at variance with reinfections with ETEC expressing nonrelated CFs (Qadri et al. 2007). Based on these findings and on the high proportion of ETEC expressing LT and/or the most prevalent CFs, we hypothesize that a multivalent ETEC vaccine containing CFA/I, CS1–CS6, and an LT toxoid may provide protection against ca 80% of ETEC strains worldwide (Svennerholm and Tobias 2008; Isidean et al. 2011). The vaccine should also provide strong mucosal immune responses against the key protective antigens locally in the small intestine (Svennerholm and Tobias 2008). Different strategies have been taken to develop a vaccine containing ETEC CFs and a toxoid and that stimulates protective immune responses.

The most extensively tested of different vaccines is an ETEC vaccine consisting of five inactivated E. coli strains expressing CFA/I, CS1, CS2, CS3, CS4, and CS5, mixed with rCTB. Clinical trials have shown that this oral rCTB-CF ETEC vaccine induced significant immune responses against both CTB and the CFs present in the vaccine in most immunized Swedish volunteers as well as adults and children in Egypt and Bangladesh (Svennerholm and Tobias 2008). The vaccine also provided significant protection against diarrhea that was sufficiently severe to interfere with the daily activity in US travelers going to Mexico and Guatemala (Sack et al. 2007). However, in an active surveillance study in children 6–18 months old in Egypt, the vaccine did not afford significant protection (Walker et al. 2007). Possible reasons for the low efficacy in infants might be the comparatively low antibody responses induced against the CF antigens in this age group (Hall et al. 2001; Savarino et al. 2002) and that most cases in the children were mild. Thus, it is well known for many children vaccines, e.g., rotavirus (Vesikari 1999), that protective efficacy against mild diarrhea is considerably lower than that against more severe disease. One possibility of enhancing vaccine efficacy may be to simply increase the dosage of CF-expressing bacteria. However, this would probably also increase the reactogenicity and, at least, lead to adverse effects in young infants, since a high dose of E. coli bacteria, even in the form of E. coli K12 placebo, induced vomiting in the youngest age groups of children in Bangladesh (Qadri et al. 2006).

Based on the lack of protective efficacy of the rCTB-CF ETEC vaccine observed in the study in children in Egypt, studies to improve the immunogenicity of the vaccine have been initiated. These include attempts to increase the amounts of protective antigens in the vaccine, particularly the amount of the CFs on the bacterial surface, by recombinant technology. Here, we review the strategies which we have applied to achieve this goal.

Recombinant over-expression of ETEC CFs, with antibiotic selection marker

Over-expression of single ETEC CFs on E. coli

To examine the feasibility of over-expressing ETEC CFs, CFA/I being one of the most prevalent ETEC CFs was first chosen. A DNA fragment carrying the entire operon of CFA/I (cfaA, cfaB, cfaC, cfaE) (Table 1) from a wild-type ETEC reference strain (Table 2) was amplified and digested to harbor restriction sites for HindIII and Eco31I, at the 5′ end (Tobias et al. 2008b). This fragment was ligated with a tac-containing expression vector, pAF-tac (Frisk et al. 2001; Sadeghi et al. 2002), harboring EcoRI and HindIII compatible ends and also containing the ampicillin-encoding gene bla, resulting in pJT-CFA/I-AMP in which the CFA/I operon is downstream of the promoter (Fig. 1). The expression of CFA/I by the constructed plasmid was examined in a nontoxigenic E. coli strain (TOP10) background, i.e., TOP10-CFA/I-Amp (Table 2), after an addition of isopropyl-β-d-thiogalactopyranoside (IPTG) as an inducer. Using a dot-blot assay with a specific monoclonal antibody (MAb) against CFA/I (López-Vidal et al. 1988; Tobias et al. 2008b), the strain was found to express up to 16-fold higher levels of CFA/I compared to the ETEC reference strain which had previously been used as CFA/I-expressing strain in the rCTB-CF ETEC vaccine (Table 2). The reference strain had been selected as a strain expressing the highest amount of CFA/I on its surface among a large number of clinical isolates tested. Considerably higher (5-fold) expression level of CFA/I by the recombinant strain than the reference strain was also observed using inhibition ELISA, in which only surface expression of CFs is determined (López-Vidal et al. 1988) (Fig. 2a). The high amounts of CFA/I on the surface of the recombinant strain was confirmed by means of immuno-electron microscopy, showing long CFA/I fimbriae with large number of immuno-gold particles (Tobias et al. 2008b).
Table 2

List of reference and constructed recombinant strains


Relevant characteristic


Reference strains

 CFA/I reference strain

Expressing CFA/I

ICDDRB, Dhaka, Bangladesh

 CS2 reference strain

Expressing CS2 (+CS3)

ICDDRB, Dhaka, Bangladesh

 CS4 and CS6 reference strain

Expressing CS4 (+CS6)

McConnell, et al. 1988

 CS5 reference strain

Expressing CS5 (+CS6)

McConnell, et al. 1988

Recombinant strains



Tobias et al. 2008b



Tobias et al. 2010a



Tobias et al. 2010a



Tobias et al. 2010a



Tobias et al. 2010a



Tobias et al. 2008a, 2010a


pJT-CFA/I-Amp, pJT-CS2-Cm

Tobias et al. 2010a


pJT-CFA/I-Cm, pJT-CS4-Amp

Tobias et al. 2010a


pJT-CFA/I-Cm, pJT-CS5-Amp

Tobias et al. 2010a


pJT-CFA/I-Cm, pJT-CS6-Amp

Tobias et al. 2010a



Tobias et al. 2010b



Tobias et al. 2010b



Tobias et al. 2008b

 V. choleraethyA)-CFA/I


Tobias et al. 2008b



Tobias et al. 2011
Fig. 1

The general concept of constructing plasmid vectors for over-expression of different ETEC CFs. All CF operons are located downstream of the strong tac promoter, which is under the regulation by lacIq. For simplicity, and based on the homology between the operons of CFA/I, CS2, and CS4, the specific genes within the operon of these CFs (as shown in Table 1) are indicated collectively in one plasmid vector. The plasmid vectors for over-expression of CS5 and CS6 which are not homologous are shown separately. The selection markers are either genes encoding antibiotic resistance, i.e., bla (ampicillin) or cat (chloramphenicol), or a nonantibiotic marker, i.e., thyA (thymidilate synthetase)
Fig. 2

Inhibition ELISA results, showing over-expression of (A) CFA/I on the TOP10-CFA/I-Amp (A) and of CS6 on TOP10-CS6-Amp (B) recombinant strains (black square) compared with the corresponding CFA/I and CS6 reference strains (white square). The dashed lines show the reciprocal titers on the x-axes for each strain giving 50% inhibition of binding of the specific MAb to CFA/I (A) and CS6 (B), respectively, to corresponding solid phase antigen. The graphs represent the average data of several experiments, and the error bars indicate SE

The successful approach of over-expressing CFA/I on E. coli was further applied to clone and overexpress additional ETEC CFs on E. coli strain TOP10 (Tobias et al. 2010a). The expression vector pAF-tac was used to express CS4, CS5, and CS6, while the expression vector pMT-tac was used to express CS2. Fragments containing the entire operons of CS2 (cotA, cotB, cotC, cotD), CS4 (csaA, csaB, csaC, csaE), CS5 (csfA, csfB, csfC, csfE, csfF, csfD), and CS6 (cssA, cssB, cssC, cssD) (Table 1) were amplified and cloned into the expression vectors resulting in pJT-CS2-Cm, pJT-CS4-Amp, pJT-CS5-Amp, and pJT-CS6-Amp (Fig. 1), respectively (Tobias et al. 2010a). The described plasmids were electroporated into the E. coli TOP10 strain resulting in the recombinant TOP10 E. coli strains, TOP10-CS4-Amp, TOP10-CS5-Amp, and TOP10-CS6-Amp (Table 2). In dot-blot assays using MAbs against CS2, CS4, and CS6, the cloned CFs were expressed in at least 8-fold higher amounts when compared with the corresponding reference strains which had previously been used in the rCTB-CF ETEC vaccine (Table 2). Inhibition ELISA assays also indicated that all recombinant strains expressed at least 4-fold more of the respective CF, the CS6 strain almost 20-fold more, than those found on the corresponding reference strains (Tobias et al. 2010a) (Fig. 2b).

To examine the stability of the plasmids and whether the recombinant strains stably express the colonization factors, colony blot assay was applied (Tobias et al. 2010a). Examining a hundred colonies of the recombinant bacteria, which have been subcultured continuously for up to 100 generations, we could show that all the colonies express the examined CF.

Coexpression of unnatural combinations of ETEC CFs on E. coli

To examine the possibility of coexpressing two main ETEC CFs, which are not naturally coexpressed on ETEC, the additional plasmid pJT-CFA/I-Cm was constructed (Tobias et al. 2010a). For coexpression of CFA/I and CS2, the plasmids pJT-CFA/I-Amp and pJT-CS2-Cm were used; while for the coexpression of CFA/I with CS4, CS5, or CS6, pMT-CFA/I-Cm was used along with pJT-CS4-Amp, pJT-CS5-Amp, and pJT-CS6-Amp, respectively. To achieve coexpression, two plasmids were electroporated into E. coli TOP10, resulting in recombinant strains coexpressing either CFA/I+CS4, CFA/I+CS5, or CFA/I+CS6 (Table 2) (Tobias et al. 2010a).

Inhibition ELISA analyses showed that both coexpressed CFs on each recombinant strain were expressed in 4–20-fold higher amounts when compared with the CFs expressed by corresponding reference strains. Furthermore, the levels of the CFs on double-expressing strains were not significantly different from the levels found on the recombinant strains expressing each CF alone.

The immunogenicity of one of the coexpressing strain, i.e., the TOP10-CFA/I-CS2 strain, over-expressing CFA/I and CS2 was analyzed. Oral immunization of mice with 5 × 108 formalin-inactivated recombinant bacteria and 7.5 μg CT as adjuvant per dose, in two rounds 2 weeks apart, induced significantly higher anti-CFA/I and anti-CS2 serum IgG+IgM titers, as well as IgA levels in fecal extracts, as compared to in mice similarly immunized with the same number of inactivated bacteria of the corresponding reference strains (Tobias et al. 2010a).

Over-expression of hybrid ETEC CFA/I+CS2 on E. coli

Intra-subclass conservation of the tip protein among the CFA/I family of CFs, e.g., CFA/I and CS2, has been reported, and in addition, the major subunits of these CFs are also highly conserved particularly at the amino terminal end which is important for assembly of the fimbrial structure (Anantha et al. 2004; Li et al. 2009). Interaction between the amino terminus of the major subunit and the tip protein is also essential for subsequent efficient initiation and assembly of fimbriae (Sakellaris et al. 1999). It has been shown that genes responsible for the assembly of CS1 and CS2 can reciprocally interact and cross-complement each other and form the typical fimbriae on the surface (Froehlich et al. 1995). Since the tip proteins within the CFA/I family are highly conserved and immunologically cross-reactive, a rationale in which the tip proteins from a number of different CFs are included in a single vaccine, i.e., a CfaE (the tip protein of CFA/I)-based subunit vaccine has been suggested (Poole et al. 2007; Anantha et al. 2004; Li et al. 2009). Generation of two or three strains that express different hybrid fimbriae with multiple CF antigen specificities could be used in an oral ETEC vaccine that would provide adequately broad protection against ETEC expressing different CFs. This approach could also offer an alternative solution to the problem of reactogenicity of high doses of E. coli bacteria in infants (Qadri et al. 2006).

To examine the feasibility of constructing such hybrid fimbriae, we first examined whether the conserved amino terminus of major subunit proteins in the CFA/I family of fimbriae would allow the substitution of one major subunit with another. A hybrid operon substituting the major subunit of CFA/I, cfaB, with the corresponding gene, cotA, from the CS2 operon in the plasmid pJT-CFA/I(∆CfaB)-CotA (hybrid I) was constructed. This plasmid was thereafter introduced into TOP10, i.e., TOP10-CFA/I(∆CfaB)-CotA (hybrid-I; Table 1) (Tobias et al. 2010b). Expression of the CFA/I(∆CfaB)-CotA fimbriae was examined using a combination of different immunological assays. When the recombinant strain was tested in an agglutination assay, the anti-CS2, but not anti-CFA/I MAb, gave rise to visible agglutination, indicating that the assembled fimbriae harbored the major subunit of CS2. To confirm that the hybrid I fimbriae, CFA/I(∆CfaB)-CotA has the tip protein of CFA/I, hemagglutination analyses were performed. In this assay, ETEC strains expressing CFA/I agglutinate human type A erythrocytes and CS2 positive strains agglutinate bovine blood cells through the respective tip protein (Anantha et al. 2004; Li et al. 2009); the recombinant strain agglutinated human group A red but not bovine red blood cells, indicating that the hybrid I fimbriae contain the tip protein of CFA/I. Quantification of the level of hybrid-I expression on TOP10-CFA/I(∆CfaB)-CotA in dot-blot and inhibition ELISA showed comparable signals or inhibitory titers for the recombinant strain and the CS2 reference ETEC. After showing that the major subunit of CS2 can be assembled and expressed using the CFA/I machinery and tip proteins, we examined whether the CFA/I system could be used to generate hybrid fimbriae that, in addition to the CFA/I tip protein, also contain the major subunits of both CFA/I and CS2. The plasmid pJT-CFA/I-CotA was constructed and introduced to TOP10, i.e., TOP10-CFA/I-CotA (hybrid-II; Table 2). Bacterial agglutination assays showed that the recombinant strain TOP10-CFA/I-CotA gave rise to visible aggregates with both anti-CFA/I and anti-CS2 MAbs, indicating that the bacteria expressed the major subunits of CFA/I and also CS2. Hemagglutination confirmed that the hybrid II fimbriae expressed the tip protein of CFA/I, not of CS2 (Tobias et al. 2010b). As tested by dot-blot and inhibition ELISA assays, elevated expression levels of both CFA/I and CS2 were found on the hybrid II CFA/I-CotA strain when compared with the corresponding reference strains (Tobias et al. 2010b). To further confirm the surface expression of the hybrid II fimbriae CFA/I-CotA, immunoelectron microscopy was applied. Using the specific anti-CFA/I and anti-CS2 MAbs and gold particles with different sizes for detection of each CF, the strain TOP10-CFA/I-CotA was shown that the recombinant strain expressed was intact fimbriae containing both CFA/I and CS2 on its surface (Tobias et al. 2010b). The presence of both major subunits on the hybrid II fimbriae was confirmed in a sandwich (capture) ELISA assay showing that after capturing the purified hybrid II fimbriae with anti-CFA/I MAb on the solid phase, polyclonal antibodies against both CFA/I and CS2 could bind to the hybrid fimbriae. The same results were also observed when anti-CS2 MAb was used to capture the hybrid II fimbriae. We could exclude that the hybrid was not a result of an aggregation of the two CFs, since mixture of purified CFA/I and CS2 did not give any positive signals in the sandwich ELISAs used.

The immunogenicity of formalin-inactivated TOP10-CFA/I-CotA (i.e. hybrid II) and each of the corresponding ETEC reference strains expressing CFA/I and CS2 was also examined in intragastrically immunized mice, given two rounds 2 weeks apart with two times 5 × 108 formalin-killed bacteria together with CT. The serum anti-CFA/I and anti-CS2 IgG+IgM as well as fecal IgA antibody titers induced by the hybrid II fimbriae strain were significantly higher against both CFs than those induced by the corresponding reference strains. (Tobias et al. 2010b).

Comparative analyses within the CFA/I family has shown that CS2 is defined as the out-group root in the phylogenetic tree (Anantha et al. 2004). By showing that CFA/I fimbriae can also contain the major subunit of CS2, as shown in hybrid II, we believe that construction of hybrid fimbriae containing the major subunits of other CFs within the CFA/I family, which are phylogenetically closer to CFA/I than CS2, may be achievable.

Recombinant over-expression of ETEC CFs with nonantibiotic selection marker

The risks of using antibiotic resistance markers in gene cloning are potential proliferation of clinically important antibiotic resistance genes and possible traces of antibiotic residues left in vaccine preparations. Additionally, in order to remove the possibility of horizontal spread of genes encoding antibiotic resistance in the environment, such markers should be avoided in vaccine strains. Therefore, expression vectors with nonantibiotic selection markers were constructed and used for expression of ETEC CFs.

Over-expression of CFA/I with nonantibiotic selection marker on E. coli

The ampicillin resistance gene (bla), in the pJT-CFA/I-AMP vector was replaced with thyA as a positive selective marker, resulting in the plasmid pJT-CFA/I-ThyA (Fig. 1). We also generated a thyA mutant of E. coli TOP10 strain and used it as a recipient for the plasmid pJT- CFA/I-ThyA to construct the recombinant strain TOP10-CFA/I-ThyA (Table 2). Growing this strain in a medium devoid of antibiotics, and under inducing conditions, the expression of CFA/I on TOP10-CFA/I-ThyA was comparable to that on TOP10-CFA/I-Amp and significantly higher than that seen in the CFA/I reference strain (Tobias et al. 2008b).

Immunization of mice with the recombinant strain TOP10-CFA/I-ThyA induced significantly higher mean serum IgA as well as IgG+IgM titers against CFA/I compared to that in mice given with the same number of bacteria of the reference strain. These antibody titers were shown to be of similar magnitudes as compared with those in mice immunized with TOP10-CFA/I-Amp.

Over-expression of CS6 with nonantibiotic selection marker on E. coli

Based on the high prevalence of CS6 on clinical isolates, there is a considerable interest in using CS6 alone or in combination with other antigens in an ETEC vaccine (Sack et al. 2007; Svennerholm and Tobias 2008; Tobias et al. 2010a). Therefore, a DNA fragment carrying the structural genes (cssA, cssB, cssC, cssE) for CS6, from a wild-type ETEC strain with surface expression of CS6, was successfully amplified by PCR and cloned to construct the plasmid pJT-CS6-ThyA (Fig. 1) (Tobias et al. 2011) which was electroporated into a thymine-deficient and nontoxigenic E. coli C600(∆thyA) strain (Tobias et al. 2011). Under inducing conditions, the recombinant C600(∆thyA)-CS6 strain expressed more than 10-fold higher levels of CS6 compared to the CS6 reference strain (Table 2), which had previously been used in the CF-CTB-ETEC vaccine.

Inactivation of the recombinant CS6-expressing bacteria, i.e., C600(∆thyA)-CS6, with formaldehyde as used for inactivation of bacteria expressing other ETEC CFs, was found to kill the CS6 positive bacteria but resulted in a complete loss of detectable CS6 antigen (Tobias et al. 2011). In contrast, treating the bacteria with a certain range of phenol concentrations not only killed the bacteria but also preserved the CS6 antigen. The immunogenicity of the strain C600(∆thyA)-CS6 was evaluated in mice orally immunized with phenol-inactivated bacteria of this strain and a corresponding number of bacteria of the reference strain. All immunized mice responded with production of serum IgG+IgM and fecal IgA antibodies against CS6 on an average of more than 60-fold higher serum antibody titres, and 75-fold higher levels of intestinal IgA antibodies were observed in mice immunized with C600(∆thyA)-CS6 bacteria than those given the reference strain. (Tobias et al. 2011).

Over-expression of CFA/I with nonantibiotic selection marker on V. cholerae

The possibility of using V. cholerae as a delivery vector for heterologous antigens, from E. coli and Shigella, has been examined earlier (Butterton et al. 1997). The construction of a V. cholerae strain that expresses one or more of the major CFs of ETEC may have the potential to protect against both ETEC and V. cholerae. In addition, the presence of CFA/I on V. cholerae may promote binding of the bacteria to the intestinal mucosa, resulting in a more efficient induction of intestinal immune responses. Furthermore, the toxicity of V. cholerae lipopolysaccharide (LPS) is lower than that of E. coli LPS, which may render recombinant V. cholerae strains less reactogenic. Therefore, a thyA-dependent V. cholerae strain, i.e., V. choleraethyA) (Carlin and Lebens, unpublished) as a recipient for plasmid pJT-CFA/I-ThyA was used. Dot-blot analysis showed a strong surface expression of CFA/I on the recombinant CFA/I over-expressing V. cholerae strain V. choleraethyA)-CFA/I (Table 1) (Tobias et al. 2008b). Using the specific anti-CFA/I MAb for immunogold labeling, the recombinant strain was shown to express intact CFA/I fimbriae on the surface with gold particles along the fimbriae (Tobias et al. 2008b). We also examined the immunogenicity of V. cholerae expressing CFA/I in mice given two oral doses of 109 formalin-killed V. choleraethyA)-CFA/I with CT adjuvant and found strong IgA as well as IgG+M responses in serum against CFA/I.


The most extensively studied ETEC vaccine, rCTB-CF ETEC, although with conferring protection against ETEC diarrhea interfering with the daily activities of travelers, did not provide significant protection in young children in a developing country. Furthermore, in studies of the vaccine in Bangladeshi and Egyptian infants, increased frequency of vomiting was observed when the children were given an adult dose of the vaccine. However, such adverse effects were not found when using a 4-fold lower, but less CF immunogenic, dose of the bacteria. Our approach has, therefore, been to try to improve the efficacy of inactivated oral ETEC vaccines by constructing different types of recombinant strains over-expressing ETEC CFs which can induce increased anti-CF immune responses with decreased number of bacteria. Preclinical studies have shown that such CF over-expressing strains can induce significantly higher anti-CF immune responses than previously used vaccine strains. Furthermore, a prototype vaccine containing an inactivated E. coli strain stably over-expressing CFA/I under nonantibiotic selection conditions has recently been tested for safety and immunogenicity in a phase-I clinical trial in Swedish volunteers. The results showed that the prototype vaccine was safe and induced significant intestinal (fecal) immune responses to CFA/I in most of the vaccinees (Lundgren et al. 2011). Based on these promising results, a new tetravalent vaccine containing four different E. coli strains individually over-expressing CFA/I, CS3, CS5, and CS6, under nonantibiotic selection conditions, together with an LT toxoid will be evaluated in a phase-I clinical trial within short.

As both LT and CF immunity have been shown to be strongly protective against ETEC strains expressing corresponding antigens and that CFA/I, CS3, CS5, and CS6 are considered to be among the most prevalent ETEC CFs (Wolf 1997; Qadri et al. 2005; Isidean et al 2011), we believe that a tetravalent vaccine containing recombinant strains over-expressing these CFs together with an LT toxoid could provide broad protection against ETEC infections. Furthermore, since cross-reactive immunity to ETEC CFs within the CFA/I, e.g., against CS4 and CS14, and CS5 family, i.e., CS7, has been shown (Qadri et al. 2006), such a vaccine may protect against ETEC expressing CFA/I and CS5-related CFs. The risk of CF replacement, if such a tetravalent vaccine is used in vaccination programs, should be considered. Although natural ETEC infection has been shown to induce strong protection against reinfection with ETEC expressing homologous CFs, there has been no clear shift in CF profile in highly ETEC endemic areas, since ETEC expressing CFA/I and CS1–CS6 still predominate worldwide (Qadri et al. 2005; Isidean et al. 2011). However, before initiating large clinical trials of new candidate vaccines, a detailed evaluation of possible CF and O antigen replacement during prolonged periods in different settings should be valuable. Such studies could be undertaken based on retrospective data, e.g, in Egypt, Mexico, and Bangladesh where ETEC infections including their toxin and CF profiles, have been followed during prolonged periods.

The described recombinant strains over-expressing unnatural combinations of two ETEC CFs or a hybrid consisting of major subunits and the tip protein of ETEC CFs may have additional advantages over strains over-expressing single ETEC CFs by allowing usage of decreased number of recombinant bacteria in the vaccine formulation.

Although the possibility of using non-E. coli bacterial vectors such as attenuated Shigellae or Salmonellae (Ranallo et al. 2005; Ziethlow et al. 2008) for expression of ETEC CFs has been described, the ability of such bacteria to invade epithelial cells may limit their usefulness to induce mucosal immune responses on the intestinal surface. Hence, noninvasive bacterial vectors such as nonvirulent V. cholerae and E. coli may be more efficient in inducing local immune responses in the gut. The described selective adherence of V. cholerae to the M cells of the gastrointestinal tract (Owen et al. 1986) and the lower LPS reactogenicity of vibrios than of E. coli bacteria make attenuated V. cholerae strains expressing ETEC CFs attractive alternative candidates as vectors for ETEC antigens in a whole cell vaccine.

Thus, the recombinant E. coli or V. cholerae strains over-expressing the major ETEC CFs as single, double, or hybrid fimbriae on two or three vector strains may be attractive candidate strains for use in an oral-inactivated human ETEC vaccine. If such strains are supplemented with a recombinantly produced LT toxoid, and if available also an immunogenic ST toxoid, a vaccine containing comparatively few bacteria that provides broad protective coverage against ETEC infections worldwide and in different populations may be achieved.

Although epidemiological studies have shown that there is an acquired immunity against ETEC (López-Vidal et al. 1990; Qadri et al. 2005), many epidemiological studies have shown that diarrhea caused by ETEC is still one of the main causes of morbidity and mortality among infants and young children in the developing world. We therefore think that although immunity against ETEC is achieved over time, a vaccine which could protect young children in highly ETEC endemic areas during the initial years of life would have an important impact on the health of this important target group. Similarly, ETEC is still the most prevalent cause of diarrhea in travelers to ETEC endemic areas, and an effective ETEC vaccine may have an important impact on their well-being during visits to such areas.


The work described in this review was supported by grants from the Marianne and Marcus Wallenberg Foundation, the Swedish Research Council (VR), the Sahlgrenska University Hospital (ALF), and the Swedish Agency for Research and Economic Cooperation (Sida-Sarec).

Copyright information

© Springer-Verlag 2012