Essential genes from Arctic bacteria used to construct stable, temperature-sensitive bacterial vaccines

Abstract

All bacteria share a set of evolutionarily conserved essential genes that encode products that are required for viability. The great diversity of environments that bacteria inhabit, including environments at extreme temperatures, place adaptive pressure on essential genes. We sought to use this evolutionary diversity of essential genes to engineer bacterial pathogens to be stably temperature-sensitive, and thus useful as live vaccines. We isolated essential genes from bacteria found in the Arctic and substituted them for their counterparts into pathogens of mammals. We found that substitution of nine different essential genes from psychrophilic (cold-loving) bacteria into mammalian pathogenic bacteria resulted in strains that died below their normal-temperature growth limits. Substitution of three different psychrophilic gene orthologs of ligA, which encode NAD-dependent DNA ligase, resulted in bacterial strains that died at 33, 35, and 37 °C. One ligA gene was shown to render Francisella tularensis, Salmonella enterica, and Mycobacterium smegmatis temperature-sensitive, demonstrating that this gene functions in both Gram-negative and Gram-positive lineage bacteria. Three temperature-sensitive F. tularensis strains were shown to induce protective immunity after vaccination at a cool body site. About half of the genes that could be tested were unable to mutate to temperature-resistant forms at detectable levels. These results show that psychrophilic essential genes can be used to create a unique class of bacterial temperature-sensitive vaccines for important human pathogens, such as S. enterica and Mycobacterium tuberculosis.

All organisms have a group of genes that are indispensible for their viability. Members of the domain Bacteria have a minimal set of about 100 essential genes that they all share, with a few exceptions. Historically, the effort to define essential genes relied on the painstaking process of attempting to delete an individual gene with and without a genetic complement, and showing that a deletion was only possible when another copy of the gene was present. Recently the availability of highly efficient transposon mutagenesis techniques, together with microarray analysis, has accelerated the discovery of candidate essential genes (1–5). Furthermore, the explosion in the number of sequenced bacterial genomes has augmented the identification of probable essential genes by allowing the identification of highly conserved genes that are retained in almost all bacteria (6).

The amino acid sequences of the products of essential genes are highly conserved across all of the bacteria, including homologs found in extremophiles, such as thermophiles (heat-loving) and psychrophiles (cold-loving). Although it is generally assumed that psychrophiles descended from warm-loving relatives, there may have been cold environments on the Earth that harbored psychrophiles for as long as 2 billion years (7). Hence, psychrophilic proteins, including essential proteins, have probably been adapting to cold environments for millions to billions of years. The observed increased heat sensitivity of psychrophilic enzymes may be a result of the changes that lead to enhanced activity at lower temperatures or may be because of changes that were introduced by genetic drift in the absence of selective pressure for thermal stability (8–10); both factors may contribute to the thermal lability of many psychrophilic proteins. Overall, any change that leads to lower enzyme stability, such as fewer salt bridges between protein domains, can be found in psychrophilic enzymes.

The introduction of mutations that make an essential gene product temperature-sensitive (TS) renders the host organism TS. This approach has been used for decades to create TS viruses that are often used as live vaccines (11). Typically, a virus is repeatedly passed in eggs or tissue culture at low temperature, and after several passages a TS form is isolated and used as a vaccine. Prominent examples of TS viral vaccines include the Sabin polio vaccine (12) and the FluMist influenza vaccine (13, 14). Although there are live vaccines for bacterial diseases, most notably the antituberculosis (bacillus Calmette–Guérin) vaccine, these are usually generated by attenuating the virulence of a pathogen by passaging in vitro or by some form of mutagenesis.

Most of the gene substitutions performed in this work were done in the Gram-negative bacterium Francisella tularensis subsp. novicida (F. novicida). This strain of F. tularensis is highly virulent in mice but is essentially avirulent in humans. However, most F. tularensis strains are highly infectious to humans and cause the potentially lethal, zoonotic disease tularemia (15). F. tularensis is a facultative intracellular pathogen that can grow in several cell types, and is frequently found within cells of the monocytic cell lineage in infected animals. Although antibody plays a role in immune protection against F. tularensis, it is thought that cell-mediated immunity plays the dominant role in the immune clearance of this organism (16).

In this work we have used the substitution of essential genes from psychrophiles into mesophilic (“moderate temperature-loving”) pathogens to create stable, TS strains of bacterial pathogens. We show that these strains can grow in cool sites in the body but fail to disseminate to internal organs. One TS strain was shown to provide protection against a subsequent challenge by a virulent form of the pathogen.

Results

We chose to use the pathogenic mesophile F. novicida as the host for essential genes from the psychrophile Colwellia psychrerythraea. The genomic DNA of these Gram-negative bacteria have similar G+C contents (17, 18) of 32.5% (Fn) and 38.0% (Cp), and we thought that this similarity should facilitate heterologous protein expression. The maximum growth temperature of F. novicida is about 45 °C and that for C. psychrerythraea is about 19 °C.

To maximize the chances that the foreign psychrophilic gene would express identically to the homolog that it replaced, we integrated the psychrophilic genes into the F. novicida chromosome so that transcription was driven by the host promoter, and translation of the mRNA was directed by the host ribosome binding site (Fig S1 A–C). About 60% of the attempts to substitute psychrophilic essential genes into F. novicida resulted in strains with a TS phenotype. The precise success rate is difficult to assess, as false-negative results were common and difficult to prove. In some cases the C. psychrerythraea gene integrated into the F. novicida chromosome in such a way as to form a hybrid between the psychrophilic and mesophilic homologs (Fig. S1 D and E). The conservation of essential genes and the close G+C content of the DNA of the two bacteria made these events more likely. The temperature inactivation phenotype imparted on F. novicida varied with the nine different genes and three orthologs of ligA (Table 1).

Table 1.

Psychrophilic genes that induce TS phenotype on Mesophiles

Gene symbol*	Temp (°C)†	Mutation rate‡	Product function
ligASf	33	4.0 × 10−6	DNA ligase
ligACp	35	<1.2 × 10−10	DNA ligase
ligAPh	37	<1.5 × 10−10	DNA ligase
hemCCp	37	<2.5 × 10−10	Heme biosynthesis
pyrGCp	37	8.5 × 10−8 §	CTP synthetase
dnaKCp	38	<3.2 × 10−10	Molecular chaperone
murGCp	38	2.6 × 10−4	Muramic acid metabol.
dnaKSf	39	<3.1 × 10−10	Molecular chaperone
fmtCp	41	1.8 × 10−9	Methionyl-tRNA formyltransferase
ftsZCp	42	ND	Cell division protein
cmkCp	43	ND	Cytidylate kinase
tyrSCp	44	ND	Aminoacyl-tRNA synth.

ND, Not determined. The restrictive temperature of these strains was too close to that of the wild-type strain to allow an accurate determination of the mutation rate.

*Subscripts indicate bacterial source of psychrophilic gene: CP: C. psychrerythraea; PH: Psuedoalteromonas haloplanktis; SF: Shewanella frigidimarina.

^†Indicates restrictive temperature which is the lowest temperature at which no growth is observed.

^‡Rate at which mutants arise that are resistant to temperatures 3 °C higher than the restrictive temperature. Note that strains carrying ligA_Cp do not generate temperature-resistant strains at 37 °C at a detectable level.

^§Mutations that were examined (8/8) were outside of the pyrG_Cp gene.

TS proteins engineered for research purposes often differ from the wild-type form by a single amino acid, and these can revert to temperature-resistance by a change in the variant amino acid or through an amino acid change elsewhere in the protein. Because marine psychrophilic bacteria have been adapting to the cold for at least several million years (7), we reasoned that some essential genes encoding TS products would not be able to mutate to a variant expressing a temperature-resistant product without changing several codons. Supporting this idea, we found that of the nine TS strains that we could test, five were unable to generate temperature-resistant forms at detectable levels (Table 1). Some strains readily yielded temperature-resistant forms, suggesting that many alternative amino acid changes in the psychrophilic gene could lead to temperature resistance. This phenomenon is illustrated for ligA_Sf, which was able to change to a temperature-resistant form by several different amino acid changes in the N-terminal adenylation domain (8) (Fig. S2). However, for eight independent temperature-resistant mutants of a strain carrying pyrG_Cp, we were unable to detect any changes to the pyrG_Cp gene, suggesting that one or more extragenic suppressor mutations was responsible for the temperature-resistant phenotype. Thus, it is possible that pyrG_Cp too is unable to mutate to temperature resistance at a detectable level.

Most of the TS F. novicida strains harboring psychrophilic essential genes grew identically to the wild-type strain in broth cultures (Fig. 1 A–C). However, as F. novicida is a facultative intracellular pathogen, the more important and stringent phenotype is the growth rate in infected macrophages. When the F. novicida strains with substitution of ligA_Cp, ligA_Ph, and dnaK_Cp were used to infect the J774 macrophage-like cell line, we found that the F. novicida carrying the psychrophilic ligA genes grew like the wild-type strain at permissive temperatures, but that the F. novicida strain carrying dnaK_Cp grew poorly (Fig. 2 D–F). The decline in the number of viable TS F. novicida in the macrophages after the shift to the restrictive temperature is a result of the phenotype of the bacterium and not because of death of the macrophages. Microscopic examination of the J774 cells infected with the TS F. novicida strains and reacted with a viability stain showed that they were healthy before and after the shift to the higher temperature. However, J774 cells infected with the wild-type F. novicida strain showed progressively fewer viable cells as the intracellular growth experiment continued (Fig. S3).

Fig. 1.

Growth of F. novicida strains carrying TS essential genes. (A–C) Growth of strains in bacteriological broth before and after shift to restrictive temperatures. The apparent increase in growth rate of the hybrid strain seen in C was observed in all three repetitions of this experiment. (D–F) Growth of TS strains in J774 macrophage-like cells before and after a shift to restrictive temperature. The F. novicida strain harboring the TS dnaK always grew poorly in J774 cells at all temperatures tested. Error bars (not visible in some panels) indicate SEMs.

Fig. 2.

Protective immunity induced by TS F. novicida strains. (A and B) Three strains were tested for their ability to induce a protective immune response, as measured by a reduction in the bacterial burden in the spleen and lungs following a challenge infection. All three strains provided a statistically significant level of protection when compared with a mock vaccination with saline. Vaccinations were via a s.c. injection at the base of the tail, and challenges were via an intranasal instillation of wild-type F. novicida. For A and B, P values were calculated with a two-tailed, unpaired t test (n = 3). (C) Protective effect of F. novicida-ligA_Ph as measured by organ burden vs. time after challenge, as compared with a mock vaccination with saline. Saline vaccinated mice were killed on day 3 postchallenge. On day 24, the lungs of six of six mice yielded no detectable F. novicida (limit of detection, 100 CFU per organ). At the same time-point, one of six spleens had detectable F. novicida. P values for data plotted for days 1 to 3 were calculated using two-way ANOVA (n = 4). (D) Survival curve of vaccinated vs. saline-injected mice. Any mouse showing significant morbidity was killed, and none were allowed to die of the effects of the infection. (Insert) Weight change of vaccinated vs. saline-injected mice following challenge with wild-type F. novicida. P values for the survival curve and the weight changes calculated using the Mantel-Cox and the two-way ANOVA test (n = 4 for days 1–3), respectively. Data-points for days 6 and 14, n = 3; for day 24, n = 6.

The temperature of the human body is dependent on a number of variables, but there is a significant difference between the temperature of the skin (∼32–36 °C) and the body core (∼37–40 °C) (19, 20). TS viral vaccines, such as FluMist, take advantage of the anatomical differences in human body temperatures (11, 13, 14). One of the possible uses of a TS pathogenic bacterial strain is as a vaccine that is able to grow in cool tissue near the body surface but unable to proliferate in the warmer body core. To test if our TS F. novicida strains had this property, we injected TS strains s.c. at the base of the tail in rats and mice. Rats are able to support high body burdens of F. novicida without any apparent morbidity (21), and thus allowed injection of larger numbers of the wild-type strain and strains with higher restrictive temperatures. Following a s.c. injection in rat or mouse tails, those TS F. novicida strains with restrictive temperatures at or below 37 °C failed to appear in the spleen, whereas those with restrictive temperatures above 37 °C could be detected in the spleen (Table 2). All strains could be easily found at the site of infection 3 d after injection. Similar results were seen when a TS F. novicida strain was injected into the fleshy part of the mouse ear (Fig S4). In contrast, wild-type F. novicida appeared in the spleen in large numbers, indicating that it spread from the skin site of injection. These results show that the distribution of TS F. novicida in a mammal is determined by the restrictive temperature of the bacterium, and strains with lower restrictive temperatures persist at a cool site in the body.

Table 2.

Dissemination of TS F. novicida strains from tail to spleen in Lewis rats and BALB/c mice

	Rats*		Mice†
F. novicida strain	CFU/tail	CFU/spleen	CFU/tail (SD)	CFU/spleen (SD)
Wild type (45 °C)	104/7 × 103	4 × 106/2 × 106	1 × 105 (105)	2 × 104 (2 × 105)
ligACp(35 °C)	5 × 102/3 × 102	<102/<102‡	2 × 103 (104)	<102
ligAPh(37 °C)	3 × 102/2 × 102	<102/<102	2 × 104 (104)	<102
dnaKCp(38 °C)	2 × 104/8 × 104	5 × 102/<102	3 × 103 (2 × 103)	5 × 102 (9 × 102)
fmtCp(41 °C)	5 × 103/2 × 103	4 × 105/2.1x105	Not determined	Not determined

*Two rats injected, and individual organ burdens listed.

^†Average from three mice (SD = SD), each injected with about 10⁵ CFU (except wild type, 35 CFU). Organ burdens were determined on day 3 postinfection.

^‡The limit of detection is 100 CFU.

The persistence of TS F. novicida strains at skin sites suggests that they may be able to induce a protective immunity. In the mouse, F. novicida is extraordinarily infectious and lethal, with the LD₅₀ below 10 CFU (22). To test if TS F. novicida strains can induce a protective immune response, we vaccinated mice s.c. at the base of the tail and challenged mice with an intranasal infection 21 d after vaccination. In one experiment, we examined the immunity induced by three TS strains compared with a mock vaccination. Using organ bacterial burden as a measure of protection, we found that TS F. novicida strains carrying ligA_Cp, ligA_Ph, or dnaK_Cp all provided protection against a challenge with wild-type F. novicida (Fig. 2 A and B). Although the F. novicida strain carrying dnaK_Cp gave the most dissemination, it grew the poorest in macrophages (Fig. 1), and induced the least amount of immunoprotection. These results suggest that dissemination does not necessarily enhance immune stimulation. Because F. novicida-ligA_Ph appeared to provide the best protection, we used it in a second, larger vaccine-protection study. Once again, s.c. vaccination with this strain provided clear protection against a lethal intranasal challenge (Fig. 2 C and D), as evidenced by the lower bacterial organ burdens, morbidity, and weight loss afforded to vaccinated mice.

To test if a psychrophilic ligA could function in other Gram-negative bacteria, we introduced the ligA_Cp gene into a Salmonella enterica strain that had an inactivated chromosomal ligA (23). The resulting hybrid strain was TS (Fig. 3A), indicating that the ligA_Cp gene can function in an enteric Gram-negative bacterium, and is thus likely to function in this large group of bacteria that include numerous important pathogens, such as Salmonella thyphi, Escherichia coli, and Yersinia pestis (Fig S5). We also introduced a codon-optimized version of ligA_Cp into Mycobacterium smegmatis, a research surrogate for Mycobacterium tuberculosis. A codon optimized version of ligA_Cp was thought to be necessary because the ligA_Cp gene is 40% G+C and the M. smegmatis ligA gene is 70% G+C. After deleting most of the M. smegmatis ligA gene (24), we found that the resulting strain was TS (Fig. 3B), indicating that the ligA_Cp gene product can function in a Gram-positive lineage bacterium, providing the potential for constructing a stable TS M. tuberculosis strain that can be used as a vaccine.

Fig. 3.

Growth dependence on ligA_Cp renders Salmonella and Mycobacterium TS. (A) S. enterica ser. Thyphimurium with phage Mu insertion in the chromosomal ligA gene showed growth dependence on a plasmid-borne copy of ligA_Cp, and was TS. Half of the data-points removed for clarity. (B) M. smegmatis with a deletion of 54% of its ligA gene is supported for growth by a plasmid-borne copy of codon-optimized ligA_Cp.

Discussion

In this study we have exploited the conservation of function of bacterial essential genes and the diversity of the temperature stability of these gene products imposed by environmental adaptation. Furthermore, we have taken advantage of the extensive evolution of the psychrophilic essential genes to identify some which are unable to change to a temperature-resistant form through simple mutational changes. Our work focused primarily on essential genes from one psychrophile, C. psychrerythraea, but our testing of a few genes from other psychrophiles yielded some that functioned in the mesophile F. novicida and rendered it TS. Among the hundreds of psychrophilic bacterial species, there are undoubtedly numerous essential gene products that are inactivated at temperatures that are relevant to biotechnology applications. Given our finding that about one-half of the essential genes are unable to mutate to temperature resistance, it is likely that a high proportion of newly found TS essential genes will also be stably TS. Furthermore, the functioning of the product of ligA_Cp in Francisella, Salmonella, and Mycobacterium suggest that some psychrophilic essential genes will function in a wide variety of human pathogenic bacteria, representing both Gram-negative and Gram-positive species and including bacteria with genomes of widely different G+C contents.

We have highlighted in this work the possibility of using psychrophilic essential genes for constructing live, TS vaccines. Live bacterial vaccines are often considered the most efficacious for protecting against diseases that require cell-mediated immunity for clearance. However, the technology described in this work may have applications in other areas, and the type of psychrophilic essential gene used could be tailored to a particular need. Inactivation of a DNA repair enzyme, like DNA ligase, results in bacterial cells that are quickly and irreversibly inactivated. When grown in large bioreactors, such strains could be easily killed at temperatures that would not damage antigenic molecules that might make up the components of a killed whole-cell or subunit vaccine. TS essential genes could be used in any bacterium whenever there is a need to use heat to kill the bacterium at a moderate temperature.

The current effort to study highly infectious select-agent pathogens is hampered by the need for expensive biocontainment facilities. The availability of TS strains that are fully competent at expressing virulence genes at 35 °C, but relatively harmless to humans because of a TS lesion that prevents growth at 37 °C, would allow most of the biochemical and genetic analysis of dangerous pathogens to be performed without the need for full physical containment.

Materials and Methods

Bacterial broth growth was done in tryptic soy broth with 0.1% cysteine (TSBC) for F. novicida, M7H9 broth with 0.05% Tween-80 for M. smegmatis, or in Luria-Bertani broth for S. enterica. F. novicida mutants were created through the use of overlap PCR to join the psychrophilic homolog gene of interest (18, 25) with the F. novicida flanking region. The fused PCR product was ligated to an erythromycin resistant–sucrose hydrolase (sacB) cassette (26) and integrated into the F. novicida chromosome. Following counter selection with sucrose, erythromycin-sensitive colonies were identified and tested for a TS phenotype. Salmonella and Mycobacterium TS mutants were created through plasmid complementation of chromosomal insertion or deletion mutants (23, 24). F. novicida and S. enterica growth curves were performed in the BioTek EL808 plate reader using Gen5 software; M. smegmatis growth was measured with a Klett-Summerson colorimeter. Twelve-week-old female BALB/C mice were used for infection experiments, and they were handled in accordance with the guidelines of the Canadian Council on Animal Care. For vaccination, anesthetized mice were injected using a tuberculin syringe s.c. at the base of the tail. Intranasal challenge was via an instillation of 6 μL of bacteria into each nostril; for the experiments in Fig. 2 A and B, this is equivalent to 88 CFU, and for Fig. 2 C and D this volume contained 30 CFU. Bacterial organ burden was determined by agar plating of organ homogenates on TSBC agar plates. Intramacrophage growth was assessed by plating infected macrophage lysates on agar medium, as described previously (27) and in SI Materials and Methods. The DNA sequences of integrated psychrophilic genes have been submitted to GenBank and have the accession numbers: ligA_Sf: HM003397; ligA_Cp: HM003389; ligA_Ph: HM003396; hem_Cp: HM003395; pyrG_Cp: HM003399; dnaK_Cp: HM003391; murG_Cp: HM003398; dnaK_Sf: HM003392; fmt_Cp: HM003393; ftsZ_Cp: HM003394; cmk_Cp: HM003390; tyrS_Cp: HM003400. Additional methods are presented in SI Materials and Methods.