candidate vaccine strain using a chromoprotein-based synthetic reporter operon. The strategy consists of the four steps: (1) characterization of natural bacterial isolates to choose the candidate; (2) optimization of the bacterial cell transformation protocol; (3) development of an amilCP-expressing reporter operon; and (4) genome integration of the reporter operon expressing an immuno-stimulating protein with simultaneous deletion of the
2. Identification and Initial Characterization of Natural V. cholerae Isolates
In the first step of our strategy for creating a candidate vaccine strain, we used several methods to identify and characterize two natural nontoxigenic
V. cholerae strains, designated strains 31 and 41. According to the MALDI-TOF mass spectrometry results obtained with a MALDI BioTyper, strains 31 and 41 were identified as
Vibrio albensis (with identification scores equal to 2.01 and 1.88, respectively).
V. albensis is the name for species comprising non-O1 and non-O139
V. cholerae strains
[20][21]. However, the identification scores allow us only to reliably conclude that both strains belong to the
Vibrio genus. Although microscopic examination and sugar utilization tests also indicated that our strains have
Vibrio-like features, these results were not enough to identify the strains. To further increase the identification reliability, we sequenced the genomes of our strains on the Ion S5XL platform (Thermo Scientific, Waltham, MA, USA). 16S RNA sequences extracted from the assembled genomes suggested that both strains belonged to the genus
Vibrio. Next, a genome-based identification of strains was performed using the maximum average nucleotide identity (ANI) obtained from paired comparisons with all the non-redundant genomes available in the MIGA and the NCBI Genome databases using the NCBI Prok tool on the Microbial Genomes Atlas web server
[22]. The genomes of strains 31 and 41 were most similar to those of
V. cholerae LMA3984 (ANI 98.95%) and
V. cholerae NZ (ANI 98.77%), respectively. Then, an MLST analysis performed with the MLST 2.0 web server showed that both strains belong to
V. cholerae. Strain 31 has the sequence type 760, while strain 41 has a new combination of alleles, and the nearest sequence type is 1317. Thus, genetic analysis strongly suggested that our strains belong to the species
V. cholerae.
Next, draft genome sequences were used to search for the presence of CRISPR/Cas systems using CRISPRminer
[23]. We found CRISPR/Cas components of type IF systems with spacers against various
V. cholerae phages in both strains. Consistently, only fragments of prophages were present in the strains’ genomes. Moreover, as indicated in the strains’ passports, they are resistant to various phages. Therefore, the data suggest that the identified CRISPR/Cas systems are active.
Then, we searched for virulence genes and genes that provide resistance to known antibiotics. In the case of strain 31, 148 factors with varying degrees of integrity of the reading frame were found. In the case of strain 41, 140 factors were found. According to the BLAST search results, the strains have some genes encoding accessory toxins. Most of these genes are inactivated by indels, while some of them seem to be active. Therefore, the ORF of the hlyA gene encoding haemolysin in strain 41 has no errors, which is consistent with the formation of large α-haemolysis zones. In contrast, the hlyA gene in strain 31 was inactivated by indels, and accordingly, the colonies formed weak β-haemolysis zones. We concluded that strain 31 is safer and used it in subsequent experiments.
3. Optimization of V. cholerae Transformation
Highly efficient transformation is essential for effective genetic engineering of a chosen bacterial strain. In the second step of our strategy, we optimized the transformation of
V. cholerae strain 31. Therefore, we constructed the reporter plasmid pTZ57R-amilCP encoding the amilCP purple chromoprotein (CP) and having an
E. coli pUC origin (
Figure 1a) that can be maintained in
Vibrio species
[12][24]. Using the protocol for
V. cholerae transformation by electroporation as described in
[25], we obtained 10
1–10
2 CFU per 1 μg of plasmid (
Figure 1b, condition 1). However, this transformation efficiency was not enough to modify the
V. cholerae genome with an integrative construct. We hypothesize that the low transformation efficiency may be caused by DNA degradation. In the protocol
[25], the cells are used in the logarithmic growth phase, when
V. cholerae secretes two extracellular DNases, Dns and Xds. Dns endonuclease is responsible for plasmid degradation
[26], while Xds endonuclease is responsible for the degradation of linear DNA
[27]. The genome of the
V. cholerae 31 strain contains complete sequences for genes encoding both DNases. Dns is repressed in the stationary phase by a quorum sensing-dependent mechanism
[26][28]. Therefore, we used a stationary phase culture for transformation. This allowed us to increase the transformation efficiency significantly to approximately 10
5–10
6 CFU per 1 μg of the plasmid (
Figure 1b, condition 2). Using cells grown to the stationary phase on minimal M9 medium further increased the efficiency of
V. cholerae transformation to 10
6–10
7 CFU per 1 μg of plasmid (
Figure 1b condition 3). Thus, we optimized the conditions for the high-efficiency transformation of natural
V. cholerae strains by electroporation.
Figure 1. Optimization of V. cholerae electrotransformation. (a) Scheme of the episomal reporter plasmid pTZ57R-amilCP. PlacZ: promoter of the LacZ α-peptide, amilCP: the gene encoding the purple chromoprotein. The curved arrows denote promoters; (b) colonies of V. cholerae transformed with pTZ57R-amilCP; (c) stationary-phase liquid cultures of V. cholerae strains that were nontransformed (K-) or transformed with the plasmid pTZ57R-amilCP; and (d) improvement of the protocol for V. cholerae transformation by electroporation. CFU: colony-forming unit. The conditions for transformation are described in the text. The data are presented as the mean (n = 3) ± SD. Statistical significance was calculated using Student’s two-tailed t-test for comparing two independent means. *** indicates p < 0.001.
4. Optimization of Synthetic Reporter Operon Expression
In the third step, we optimized the expression of the amilCP-based reporter operon. The promoter dramatically affects heterologous protein expression in
V. cholerae both in vitro and in vivo
[29][30][31]. Previously, the
V. cholerae lacZ locus was used as a safe site for integrating genetic constructs
[32][33], and the
lacZ promoter was utilized to express heterologous proteins
[34]. Therefore, we created the integration construct pCI-amilCP for insertion of amilCP into the
lacZ locus (
Figure 2a). This construct harboured the
amilCP-lacZ synthetic operon under the control of the natural
lacZ promoter. Integration flanks were obtained from the other
V. cholerae strain, P-19241. Since they differ in sequence from the
V. cholerae strain 31
lacZ locus, we identified the part of the
lacZ locus that came from the integration plasmid. The integration flanks were 3 kb long; flanks of such length provide high integration efficiency
[35]. After pCI-amilCP transformation, none of the grown colonies had colour imparted by
amilCP. We hypothesized that the amilCP expression level from the integrated construct was not sufficient to colourize the colonies.
Figure 2. Integration of the synthetic reporter operon into the lacZ locus. (a) Schematic depiction of the integration reporter plasmid pCI-amilCP. The curved arrows denote promoters; (b) schematic depiction of the integrative pALAL plasmid carrying the synthetic reporter operon containing the genes for β-lactamase (AmpR), E. coli galactopermease (lacY), chromoprotein (amilCP), and V. cholerae β-galactosidase (lacZ); (c) colour of cell pellets of V. cholerae transformed with the integrative plasmid pALAL. An untransformed V. cholerae culture (K-) was used as a negative control. V. cholerae transformed with the plasmid pTZ57R-amilCP was used as a positive control. (d) Design of PCR to check the correct integration of the reporter operon. The positions of primers are marked with arrows. If the construct is integrated at the correct genomic position, the primer pair VC-17/VC-18 should yield a 3903 bp fragment, and the VC-19/VC-20 pair should yield a 3182 bp fragment; (e) PCR analysis of the edited V. cholerae colonies. PCR fragments were separated in a 1.5% agarose gel in the presence of EtBr. Genomic DNA from nontransformed strain 31 was used as a negative control (K-). M: 1 kb DNA marker, 1–15: colonies used in the analysis. An asterisk (*) indicates the non-specific band.
Next, we tried to use
lacZ as a reporter gene. The genome of strain 31 harbours a complete β-galactosidase gene without deleterious mutations. Accordingly, the ONPG test confirmed the presence of β-galactosidase activity in the cell lysates. As expected for
V. cholerae strains, the genome of our strain has no gene for high-affinity 5-Bromo-4-chloro-3-indoyl-beta-D-galactopyranoside (X-Gal) transport, such as LacY galactopermease. Unexpectedly, strain 31 was able to utilize X-Gal. The colonies developed colour slowly within 48 h, suggesting the presence of some low-activity transporters for X-Gal. The possible transporters for X-Gal can be ABC transporters encoded by the
mgl operon and involved in methylgalactoside and galactose transport in gram-negative bacteria
[36][37]. Previously, the
E. coli lacYZ operon was used as a strong reporter for in vivo identification of infection-inducing genes in pathogenic
V. cholerae strains
[38]. Therefore, we took advantage of active β-galactosidase and added the
E. coli lacY gene to the synthetic reporter construct (
Figure 2b). Genes were separated by spacers from the
V. cholerae S10 operon of the ribosomal proteins. We reasoned that S10 spacers should provide a high level of mRNA translation. We expected that colonies expressing the improved reporter operon would develop colour more rapidly. Indeed, in
E. coli, the
lacY-containing synthetic reporter causes
E. coli colonies to develop more intense colouration. However, the improved reporter operon did not enhance
V. cholerae transformant colouration intensity on X-Gal plates (data not shown). We hypothesize that X-Gal transport is not the rate-limiting process of colour development in our strain.
To grow only transformed colonies, we added the β-lactamase gene into the synthetic reporter operon, yielding the pALAL integrative plasmid (Figure 2b). The β-lactamase gene should mediate resistance to ampicillin in transformed cells that are otherwise sensitive to it. V. cholerae transformation with the pALAL plasmid provided 101–102 colonies per 1 μg of the plasmid on ampicillin-supplemented agar plates. The cell pellets of several randomly picked transformed colonies 2, 3, and 5 clearly had a darker colour than control non-transformed cells, suggesting a low level of amilCP accumulation (Figure 2c). PCR with pairs of the primers VC-17/VC-18 and VC-19/VC-20 verified the integration of the reporter operon into the target lacZ locus in these and other recombinant V. cholerae colonies (Figure 2d,e). Sequencing of PCR products further confirmed the correct integration of constructs.
Moreover, we also tried to transform V. cholerae with linear DNA. Therefore, we used a 9082 bp integration cassette amplified from the pALAL plasmid with the primers VC05 and VC09. As a result, we obtained only one ampicillin-resistant colony (# 1). This colony had no purple colour. However, the cell pellet from this colony had a darker colour than that in the negative control (Figure 2c), suggesting a low level of amilCP accumulation. PCR confirmed correct integration of the construct in the lacZ locus (Figure 2e). Unexpectedly, sequencing revealed the integration of a 516 bp fragment of the p15A plasmid ori next to the left flank of the integrated construct. We hypothesize that the PCR fragment was destroyed upon transformation (apparently, by the Xds exonuclease). Seemly, the pALAL plasmid copurified with the PCR product was integrated into the genome instead of the PCR product. Therefore, we may conclude that our strain can be transformed only by circular plasmids.
Our results suggest that the native
lacZ promoter is relatively weak and cannot provide sufficient expression levels of reporter proteins to visualize transformed colonies. Therefore, we used the stronger promoter of
recA, which highly and stably expresses the housekeeping gene in
V. cholerae [31]. We assembled two plasmids bearing a synthetic operon consisting of the amilCP and chloramphenicol acetyltransferase genes under the control of the
recA promoter and terminator regions. The plasmid pCI-RACR-0.5 bears short
recA flanks that are 0.5 kb long (
Figure 3a) and is supposed to be episomal, while the pCI-RACR-3.0 plasmid bears 3 kb-long flanks (
Figure 3b) and is expected to be integrated into the
recA locus. As expected, pCI-RACR-0.5 was not integrated into the
recA locus. This was confirmed by PCR (
Figure 3c,d) and the presence of the episomal plasmid in cell cultures (
Figure 3e). The
V. cholerae colonies, as in the case of pTZ57R-amilCP, were coloured purple.
V. cholerae colonies transformed with pCI-RACR-3.0 exhibited a purple centre (
Figure 3f), and their pellets were also clearly purple in colour (
Figure 3g). PCR confirmed the correct integration of pCI-RACR-3.0 into the
recA locus (
Figure 3h). Next, we found that
V. cholerae colonies grown on brain heart infusion (BHI) medium for 72 h developed a more intense colour. Therefore, we optimized the expression of the synthetic operon to identify transformed
V. cholerae colonies simply by visual inspection.
Figure 3. Integration of the synthetic reporter operon into the recA locus. (a,b) Schematic depiction of the pCI-RACR reporter constructs. The curved arrows denote promoters; (c) scheme of the experiment for PCR verification of construct integration. The positions of primers are marked with arrows; (d) PCR check for the integration of pCI-RACR-0.5. If the construct was integrated into the correct genomic position, the VC-43/VC-44 pair amplified a fragment of 3157 bp (lane 1), the VC-43/VC-47 pair amplified a fragment of 3330 bp (lane 2), the VC-43/VC-34 pair amplified a fragment of 3824 bp (lane 3), the VC-45/VC-46 pair amplified a fragment of 3250 bp (lane 4), and the VC-35/VC-46 pair amplified a fragment of 3929 bp (lane 5). (e) Plasmid preparations from E. coli (lane 1) or V. cholerae colonies transformed with pCI-RACR-0.5 (lanes 2 and 3); (f) colour of V. cholerae colonies transformed with the pALAL or pCI-RACR-3.0 integrative plasmid; (g) colour of cell pellets of V. cholerae transformed with the integrative pCI-CR-3.0. An untransformed V. cholerae culture (K-) was used as a negative control; (h) PCR check for the integration of pCI-RACR-3.0. Untransformed V. cholerae strain 31 (K-) was used as a negative control. The primers used were the same as those described in part (c) of the figure.
6. Construction of a Candidate V. cholerae Vaccine Strain
In the final step, we proceeded to construct a candidate vaccine strain. Since the cholera toxin β-subunit (CtxB) is safe and provides protective properties for vaccines, e.g., for the widely used rBS-WC (Dukoral)
[40][41][42], we sought to convert our strain to a CtxB producer. To achieve this goal, we assembled constructs to integrate a synthetic operon consisting of
ctxB,
amilCP, and
cat into the
recA locus. However, in all the constructs purified from violet
E. coli colonies resistant to Cm, the
ctxB ORF was damaged, leading to inactive protein synthesis. We hypothesize that high levels of the cholera toxin produced from the strong constitutive promoter are highly toxic to
E. coli. Therefore, the surviving colonies harboured the assembled constructs with
ctxB mutations. We reasoned that to assemble the construct correctly, we needed to use a promoter that is weak or inactive in
E. coli and active in
V. cholerae. The promoter of the cholera toxin operon is very weak in
E. coli [43]. At the same time, it is highly active in
V. cholerae and directly regulated by two transcription factors, toxT
[44] and toxR
[45]. Indeed, the construct (
Figure 5a) was assembled in
E. coli correctly, suggesting that the β-subunit was expressed at nontoxic levels. The obtained integration construct was transformed into
V. cholerae. The colony with the most intense violet colouration was further analysed. PCR confirmed the correct integration of the
ctxB-containing operon into the
recA locus (
Figure 5b). The recombinant strain produced the cholera toxin β-subunit at 6.64 µg/mL in the cell lysate and about 1.6 µg/mL in the culture media (
Figure 5c). The productivity of our strain is close to one of the best CTB-producing
V. cholerae recombinant strains M7922-C1, which produces β-subunit at 3.17 ± 1.69 µg/mL
[46]. The genetic stability test suggests the high stability of the integrated
ctxB-expressing operon (
Figure 5d). Thus, we constructed a novel
V. cholerae candidate vaccine strain designated rVCH-31.1 suitable for further preclinical studies.
Figure 5. Construction of a candidate V. cholerae vaccine strain expressing cholera toxin β-subunit. (a) Schematic depiction of the pCI-RCCACR-3.0 integrative construct bearing a synthetic operon with ctxB, the amilCP reporter, and the cat marker genes. The curved arrows denote promoters; (b) PCR check for the integration of pCI-RCCACR-3.0. The positions of the primers are marked with arrows. In the case of correct genomic integration of the construct, the VC-43/VC-44 primer pair amplified a fragment of 3157 bp (lane 1), the VC-43/VC-39 primer pair amplified a fragment of 3368 bp (lane 2), the VC-43/VC-41 primer pair amplified a fragment of 3759 bp (lane 3), the VC-45/VC-46 primer pair amplified a fragment of 3250 bp (lane 4), and the VC-35/VC-46 primer pair amplified a fragment of 3929 bp (lane 5). M: DNA molecular weight marker; (c) results of GM1-ELISA of CtxB production by the rVCH-31.1 strain. The data are presented as the mean (n = 3) ± SD. The negative controls were the original V. cholerae nontoxigenic strain (K-1) or recombinant strain edited with the pCI-RACR-3.0 plasmid (K-2). ‘<Curve’ indicates that the signal was below the calibration curve; (d) quality control of the V. cholerae candidate vaccine strain by the genetic stability test. The data are presented as the mean (n = 5) ± SD. NS: nonsignificant differences according to one-way ANOVA.