Yersinia pestis Caf1 Protein: Effect of Sequence Polymorphism on Intrinsic Disorder Propensity, Serological Cross-Reactivity and Cross-Protectivity of Isoforms

Yersinia pestis Caf1 is a multifunctional protein responsible for antiphagocytic activity and is a key protective antigen. It is generally conserved between globally distributed Y. pestis strains, but Y. pestis subsp. microtus biovar caucasica strains circulating within populations of common voles in Georgia and Armenia were reported to carry a single substitution of alanine to serine. We investigated polymorphism of the Caf1 sequences among other Y. pestis subsp. microtus strains, which have a limited virulence in guinea pigs and in humans. Sequencing of caf1 genes from 119 Y. pestis strains belonging to different biovars within subsp. microtus showed that the Caf1 proteins exist in three isoforms, the global type Caf1NT1 (Ala48 Phe117), type Caf1NT2 (Ser48 Phe117) found in Transcaucasian-highland and Pre-Araks natural plague foci #4–7, and a novel Caf1NT3 type (Ala48 Val117) endemic in Dagestan-highland natural plague focus #39. Both minor types are the progenies of the global isoform. In this report, Caf1 polymorphism was analyzed by comparing predicted intrinsic disorder propensities and potential protein-protein interactivities of the three Caf1 isoforms. The analysis revealed that these properties of Caf1 protein are minimally affected by its polymorphism. All protein isoforms could be equally detected by an immunochromatography test for plague at the lowest protein concentration tested (1.0 ng/mL), which is the detection limit. When compared to the classic Caf1NT1 isoform, the endemic Caf1NT2 or Caf1NT3 had lower immunoreactivity in ELISA and lower indices of self- and cross-protection. Despite a visible reduction in cross-protection between all Caf1 isoforms, our data suggest that polymorphism in the caf1 gene may not allow the carriers of Caf1NT2 or Caf1NT3 variants escaping from the Caf1NT1-mediated immunity to plague in the case of a low-dose flea-borne infection.


Introduction
The outbreaks, epidemics and pandemics of human plague are caused by Yersinia pestis subsp. pestis strains that possess universal hypervirulence for a wide range of mammals and are ubiquitously distributed [1][2][3][4]. The representatives of a more ancestral subsp. microtus are endemic within populations of some voles (Microtus spp.) and cause only rare sporadic diseases [1,5] with no human-to-human transmission [5]. Strains of both subspecies can make a proteinaceous capsule first described by Alexandre Yersin [6]. This antiphagocytic capsule [7] is the main component of plague vaccines [8][9][10][11] and is the most important target for laboratory diagnosis of plague [12].
Capsule biogenesis is implemented by a conserved chaperone/usher pathway [13]. Caf1 structural subunit is encoded by a 510-nucleotide caf1 gene. The precursor protein contains 170 amino acids. The typical cleavage site is located between Ala21 and Ala22 residues [14].
Recently it was shown that a single nucleotide substitution found in the bv. caucasica strains Pestoides F [15] and G8786 [16] resulted in Ala48 ! Ser48 substitution, while mutations in Y. pestis E1979001 (bv. antiqua) and F1991016 (bv. orientalis) resulted only in truncation down to 147 and 130 amino-acid residues giving, most likely, non-functional peptides. More recently, sequencing of caf1 gene from 41 subsp. microtus strains isolated from voles and their fleas in Georgia and Armenia indicated that all of them had the same Ala48 ! Ser48 substitution (caf1 NT2, accession no. EF165977), while the strains isolated from gerbils and susliks of the same region carried the gene with a canonic sequence (caf1 NT1, accession no. EF165976) [17].
In this study, we provide the first evidence that the allele type NT2 (Ser48 Phe117) is unique to the Transcaucasian-highland and Pre-Araks natural plague foci, while a novel NT3 type (Ala48 Val117) is endemic in Dagestan-highland natural plague focus. This is supported by sequencing data on the caf1 genes from 119 strains of Y. pestis belonging to seven out of eight biovars of subsp. microtus [4]. Our computational analysis revealed that the Caf1 isoforms found in Y. pestis endemic strains should not make them significantly different in terms of pathogenicity. To test this hypothesis, we evaluated serologic cross-reactivity and cross-protection of the Caf1 isoforms. All the three isoforms could be equally detected by immunochromatography test for plague. When compared to the classic Caf1 NT1 , endemic Caf1 NT2 and Caf1 NT3 had lower immunoreactivity in ELISA and lower indices of self-and cross-immunity. However, although a notable reduction in the cross-protection was observed between all isoforms, the polymorphisms in the caf1 gene may not provide for Caf1 NT2 or Caf1 NT3 Y. pestis strains the possibility to escape from the Caf1 NT1 -mediated plague immunity in the case of a low-dose flea-borne infection.

Materials and Methods
Bacterial strains and culture conditions Y. pestis intraspecies classification used in this study corresponds to the International Codex of Bacterial Nomenclature [4,18,19]. In this study, we used a total of 119 strains of Y. pestis representing seven out of eight belonging to subsp. microtus biovars, such as caucasica (78), altaica (17), qinghaiensis (2), xilingolensis (3), hissarica (4), talassica (4), and ulegeica (11) [4], as well as belonging to the main subsp. pestis a vaccine strain EV line NIIEG (bv. orientalis) and a wild type strain 231 (bv. antique). Characteristics of the strains used for testing of serologic cross-reactivity and cross-protection are shown in Table 1. In addition, avirulent bacteria, Pgm − vaccine strain EV, as well as strains C-376pCD1 − and C-824pCD1 − , depleted of the low-calcium-response virulence plasmid were used for all Y. pestis-derived Caf1 preparations.
Bacteria were grown at 28°C for 48 h on brain heart infusion (BHI; HiMedia Laboratories) supplemented with 2% agar at pH 7.2. For Caf1 isolation and purification, bacteria were grown at 37°C in New Brunswick Scientific fermenters with working volumes up to 10 L of liquid aerated media. Growth medium was BHI supplemented with 0.5% yeast extract (Difco). Acidity and oxygen levels were controlled with a specified pO 2 value >10%. Biomasses were harvested by centrifugation after 48 h.
All handling of samples containing live wild-type Y. pestis isolates was performed in a select agent authorized BSL3 facility under protocols approved by the State Research Center for Applied Microbiology and Biotechnology Institutional Biosafety Committee.

Sequencing of caf1 genes
The nucleotide sequence of each caf1 gene was determined by direct sequencing of the PCR fragment obtained after amplification of the part of caf1 operon of the corresponding strain. The primers caf1-F (5'-GAATTTGTTCGTGGATTGGA-3') and caf1-R (5'-TTAAAGGAG GGCATAATAGC-3'), both flanking the caf1 gene, were located within the caf1A and YPMT1.85 (similar to a fragment of integrase) genes, respectively, and were used for both fragment amplification and direct sequencing. Determined sequences of the caf1 genes were deposited to the GenBank (accession numbers KP641181.1-KP641299.1) and compared to reported sequences of this gene in other Y. pestis strains.).

Isolation and purification of Caf1 isoforms
Avirulent bacteria, Pgm − vaccine strain EV, as well as strains C-376pCD1 − and C-824pCD1 − , depleted of the low-calcium response virulence plasmid were used for all Y. pestis-derived Caf1 preparations. Cell-free Caf1 was extracted directly from the supernatants of Y. pestis broth cultures and purified by chromatography. Clarified Caf1 supernatant was slowly mixed with 4 M ammonia sulfate (AS) solution to achieve a 1 M final concentration, and precipitate was collected by centrifugation at 15000 × g for 30 min at 4°C. A HiPrep Phenyl FF (High Sub) 16/10 column (GE Healthcare) was used for the initial chromatography step. Prior to sample loading, the column was equilibrated with 5 column volumes of a 20 mM Tris buffer, supplemented with 1 M AS (pH 8.0). Clarified supernatant was loaded and subjected to the following steps: 4 column volumes 1 M AS + 20 mM Tris, pH 8.0; 100%-0% AS + 20 mM Tris, pH 8.0 over ten column volumes; hold with 0% AS buffer for 4 column volumes. Ten mL fractions were collected for each elution step. All fractions were analyzed using 12.5% SDS-PAGE. The fractions containing protein Caf1 were combined and concentrated by a second passage through a hydrophobic interaction chromatography (HIC) column. Protein was concentrated by a single step elution by 20 mM Tris, pH 8.0. The fractions were desalted on a XK 26/30 column (GE Healthcare) packed with Toyopearl HW-40F chromatographic media (Tosoh Bioscience) and pre-equilibrated with 20 mM Tris, pH 8.0. Ten mL fractions were collected, and purified Caf1 protein was concentrated using Millipore YM-10 membrane for a subsequent storage at -70°C until used.

Ethics Statement
All protocols for animal experiments were approved by the State Research Center for Applied Microbiology and Biotechnology Bioethics Committee (Permit No: VP-2015/2) and were performed in compliance with the NIH Animal Welfare Insurance #A5476-01 issued on 02/07/ 2007, and the European Union guidelines and regulations on handling, care and protection of Laboratory Animals (http://ec.europa.eu/environment/chemicals/lab_animals/home_en.htm).

Mice
Seven week old female BALB/c mice were purchased from Laboratory Animals Breeding Center (Shemyakin and Ovchinnikov Institute of Bioorganic Chemistry, Russia), housed in polycarbonate cages, and maintained in light-controlled (lights on from 7:00 to 19:00) BSL3 room at the State Research Center for Applied Microbiology and Biotechnology. The temperature and the humidity of the animal room were maintained at 22°C ± 2°C and 50% ± 10%, respectively. Mice were given tap water and mouse mixed fodder PK-120 (Laboratorkorm, Russia) ad libitum throughout the study. The number of mouse for experiments used the minimum number of the necessity. The mice were divided into all groups randomly. In this study, we have used humane endpoints for the infected animals. According to the animal protocol the mice should be euthanized in the animal survival studies, when they became either of the following: lethargic, dehydrated, moribund, unable to rise, non-responsive to touch, or lost more than 10% body mass. Humane euthanasia, CO 2 exposure (anesthesia) using compressed CO 2 gas followed by cervical dislocation has been used by well-trained individuals. We have monitored the health condition of the animals at least twice a day. There was no unexpected death during the entire set of experiments.

Animal immunization
Mice were randomly divided into four groups (n = 96 per group) and vaccinated subcutaneously (s.c.) with 10 μg of each Caf1 isoform in 0.1 mL PBS (pH 7.2) adsorbed (1:10, w/w) to the vehicle, aluminum hydroxide gel colloidal suspension (Sigma, USA), or only with the vehicle in PBS as a negative control (placebo). After 30 days, the animals were boosted with an identical dose of the same antigen into the same inoculation site.

Serologic cross-reactivity
Immunochemical specificity of Caf1 isoforms was assessed with the immunochromatographic rapid diagnostic test for plague that utilizes the anti-Y. pestis Caf1 NT1 monoclonal antibody F19 (State Research Center for Applied Microbiology and Biotechnology, Russia).
Blood samples were obtained under anesthesia with CO 2 gas by retro-orbital route by welltrained individuals. Antibody titers were determined by indirect ELISA a day before and 43 days after the second Caf1 immunization individually in five randomly selected animals from each group of 96 mice immunized with one of the Caf1 isoforms, and the mean titer was calculated. Microtiter plates (Greiner Bio-One, Austria) were coated with 100 ng/well of Caf1 in 0.1 M sodium bicarbonate buffer (pH 9.6) overnight at 4°C. Non-specific binding was blocked with 3% gelatin from cold water fish skin (Sigma) in 0.01 M PBS, pH 7.2. Test sera were added using 2.5-fold serial dilutions in 0.01 M PBS buffer containing 0.05% tween-20 (PBST) and incubated for 2 h at 37°C. After four washes with 0.01 M PBST, 100 μl of sheep anti-mouse IgG conjugated to horseradish peroxidase (GE Healthcare) at a dilution of 1:4000 was added for 1.5 h at 37°C. The plates were washed with PBST and 100 μl of 0.01% o-phenylendiamine-H 2 O 2 was added to each well. The reaction was stopped by the addition of 100 μl of 1 M H 2 SO 4 per well, and OD was read at 450 nm using EVOLIS Twin Plus System (BIO-RAD, USA). The titer of antibodies was estimated as the maximum dilution of serum giving an OD reading that exceeded the background by 0.1. Background values were obtained from serum samples collected from the animals injected with the vehicle alone.

Cross-protection
The ability of an antigen isoform to protect an animal from death after administration of a high dose of a virulent strain producing a different isoform of the same antigen, designated Immunity Index (II) was calculated as the ratio: where LD 50imm is LD 50 for animals immunized with an antigen under the study; LD 50veh is LD 50 for vehicle-treated animals.
To estimate LD 50 , 45 days after the booster dose, four groups of mice were infected with virulent strains producing either of the Caf1 isoforms (See Table 1). Each group was subdivided into three subgroups of 32 mice that were challenged with 10-fold dilutions of virulent strains of Y. pestis (231 (NT1), C-376 (NT2), and C-824 (NT3)), (a. 2 × 10 3 to 2 LD 50 ; eight mice for a dose) subcutaneously (in the interior thigh). Animals that succumbed to infection were sacrificed and examined bacteriologically to verify that infection was the cause of death. The remaining animals were observed for three weeks. The animals that survived were humanely euthanized.

Statistical methods
Data on ELISA were expressed as means ±SEM (standard error of the mean). The LD 50 and a 95% confidence intervals of the virulent strains for immunized and naïve animals were calculated using the Kärber method [53]. Mortality timeframes were recorded, and the mean life to death time span was calculated for each treatment group. Comparison of the survival curves was performed using Log-rank (Mantel-Cox) test. A P value below 0.05 was considered to be significant.

Comparison of the Caf1 antigen sequence heterogeneity
Sequencing of caf1 genes from 119 Y. pestis strains belonging to different biovars within subsp. microtus showed that the Caf1 proteins possess three isoforms, the global allele type NT1 (Ala48 Phe117), NT2 type (Ser48 Phe117) peculiar to Transcaucasian highland and Pre-Araks natural plague foci, and a novel NT3 type (Ala48 Val117) endemic for Dagestan-highland natural plague focus. Fig 1 represents the results of multiple sequence alignment of Caf1 isoforms analyzed in this study and shows that the Caf1 NT2 found in Transcaucasian highland and Pre-Araks natural plague foci is different from the major allele isoform Caf1 NT1 by having an Ala48!Ser48 substitution, whereas Caf1 NT3 protein isolated from the Dagestan-highland natural plague focus has a Phe117!Val117 substitution.

Intrinsic disorder
To understand if found Caf1 polymorphism has an effect on structural and functional properties of this protein, we evaluated the disorder propensities of Caf1 NT1 , Caf1 NT2 , and Caf1 NT3 isoforms and analyzed the effect of corresponding amino acid substitutions on potential disorder-based binding sites. Results of these analyses are summarized in Fig 2 that compares the disorder profiles obtained for the Caf1 isoforms by PONDR1 VSL2 (Fig 2A), PONDR-FIT (Fig 2A) and PONDR1 VLXT algorithms (Fig 2B). This analysis revealed that, although Caf1 is predicted to be mostly ordered protein, it has several disordered regions. Curiously, both substitutions found in Caf1 (the Ala48!Ser48 in Caf1 NT2 and the Phe117!Val117 in Caf1 NT3 ) cause noticeable increase in the local intrinsic disorder propensity of the short regions surrounding the corresponding substitutions. Importantly, Fig 2 clearly shows that although the effects of these substitutions on the intrinsic disorder propensities of the Caf1 isoforms are not very strong, there is a reasonable agreement between the results obtained by the three computational tools.
Next, we evaluated the presence of potential disorder-based binding regions in various Caf1 isoforms using the ANCHOR algorithm [51,52] that utilizes the following criteria: (i) residues of potential disorder-based region belong to a long disordered segment and are not a part of a globular domain; (ii) residues of such a region are not able to form enough favorable contacts with its own local sequential neighbors to fold; (iii) these potential binding residues can form enough favorable interactions with globular proteins upon binding [51,52]. This algorithm also filters out potential disorder-based regions shorter than six residues. Fig 2B represents the results of this analysis and shows that the Caf1 NT1 and Caf1 NT2 have a very short potential binding site (region 140-142) which was filtered out by the algorithm because of its small size. On the other hand, the Phe117!Val117 substitution found in Caf1 NT3 causes an extension of this site to 4 residues (140-144). Since this length is below the length threshold utilized by ANCHOR algorithm, this potential binding site was also filtered out. Fig 2B also shows that there is some delocalization in the effects of the Phe117!Val117 substitution on the disorder propensity and on the disorder-based binding potential.

Isolation and purification of Caf1 isoforms
Hydrophobic chromatography allowed for isolation of highly purified Caf1 NT1 Caf1 NT2 isoforms as peaks at 350 mM of the ammonium sulphate gradient, while the Caf1 NT3 elution was spread between 600 and 50 mM under the same elution conditions.

Serologic cross-reactivity
Serologic cross-reactivity has been tested by both immunochromatography and ELISA techniques. The detection limit of immunochromatography was 1.0 ng/mL with the range of 1.0 ng-1.0 μg/mL. Results obtained were identical for all the three isoforms, indicating that all of them could be readily detected by the antibody used in this method. Fig 3 shows that, after both steps of immunization, the antibody titers estimated by ELISA were 4-7 times higher in the animals immunized by the Caf1 NT1 isoform, regardless of the antigen adsorbed on the plates. The antibody response to immunization with Caf1 NT3 isoform was the lowest. All the three isoforms absorbed to the plastic were recognized by the sera of immunized animals.
Interestingly, the booster immunizations resulted in a somewhat lower antibody response.
numbers of corresponding LD 50 . The NT1 isoform that has been traditionally used in commercial and experimental vaccines proved to be the most protective one. This isoform was 100% protective in mice challenged with 2000 LD 50 of the strain 231 (100%), 85% when C-376 strain was used, and 40% mice were protected when infected with the C-824 strain (Fig 4). Thus, immunization with NT1 isoform partly protected against infection induced by NT1 and NT2 carriers even at a high dose. Vaccination with NT2 protected from infection induced by NT2  and NT3, but was less effective against NT1 carriers. At the dose equal to 20 LD 50 of any of the three strains tested, all animals vaccinated with Caf1 NT1 survived. This data suggests that vaccination with the NT1 isoform of Caf1 provides a better protection against all the three Y. pestis variants. All the vehicle-treated mice died within a week after infection regardless of the bacterial stain (data not shown). Vaccination with the Caf1 NT1 isoform produced a level of protection that exceeded 5334 times that of control (vehicle-treated) mice. The other two isoforms were less potent resulting in only a 1264-to 2000-fold increase in self-resistance. Thus, calculated indices of immunity are in agreement with the data obtained from the survival curves, suggesting that Caf1 NT1 vaccination is superior compared to the other two isoforms.

Discussion
Recently, it has been demonstrated that some of Y. pestis strains isolated from Georgian and Armenian natural plague foci had the NT2 allele of caf1 gene differing from the typical NT1 allele by a single nucleotide replacement causing a substitution of alanine by serine (Ala48 ! Ser48) [15][16][17]. Our investigations have proven that this replacement is characteristic of all bv. caucasica strains from Transcaucasian-highland plague foci # 4-6 and bv. caucasica isolates from Pre-Araks natural plague focus # 7. It is possible that similar strains are circulating in neighboring territories of Turkey and Iran. We also found a novel NT3 allele (Ala48 Val117) endemic for Dagestan-highland natural plague focus # 39. But the classic Caf1 NT1 isoform (Ala48 Phe117) remains the major type within the rest of Y. pestis.
In silico analysis of amino-acid sequence of Caf1 NT1 performed in several laboratories by different methods aimed at location of B-and T-cell epitopes generated conflicting data [54][55][56][57]. Information from the laboratory of Dr. D.N. Rao seems to be the most reliable, as it was confirmed by immunoassays and in vivo experiments [54].
Our computational analysis revealed that the Ala48!Ser48 and Phe117!Val117 substitutions found in the Caf1 NT2 and Caf1 NT3 isoforms, respectively, have some effects on the local intrinsic disorder propensities of these proteins. Both substitutions cause noticeable increase in the intrinsic disorder propensity, but only latter is expected to have some effect on the disorder-based interactivity of this protein. Even in this case, mutation-induced increase in interactivity is minimal, although behavior of the isoform during hydrophobic chromatography is different: the Caf1 NT3 elution is spread between 600 and 50 mM of the ammonium sulfate gradient, while Caf1 NT1 and Caf1 NT2 isoforms are peaking at 350 mM.
It has been shown that Caf1 NT1 isoform of capsular protein protects bacteria from phagocytosis [58] and exhausts the complement system by selective activation of C`2 and C`4 components thus preventing complement-mediated opsonization of bacteria [59]. On the other hand, it is one of immunodominant antigens responsible for protective immunity against plague [8,60,61]. Accordingly, it has been used as the major molecular target for immunodiagnostics, and as a principal component of the vaccines [8,[60][61][62][63]. Important questions of whether Caf1 NT1 -mediated protective immunity can be circumvented by the strains that carry NT2 and NT3 alleles of caf1 gene, and whether current immunoassays can detect Caf1 NT2 -and Caf1 NT3producing bacteria have not been addressed before.
In our cross-protection studies, the level of immune response achieved was described in terms of the immunity index. This index represents the difference in challenge dose required to cause death in immunized versus naïve animals ( Table 2). Similar data were obtained from the analysis of survival curves (Fig 4). Taking into account that infected fleas can contain up to 5000 Y. pestis CFU, but the median number of transmitted bacteria is 82 CFU [2], we can predict that the immunity induced by Caf1 NT1 isoform is sufficient for protection from a low-dose flea-borne infection caused not only by a strain with the same isoform, but also by isolates producing Caf1 NT2 and Caf1 NT3 variants. However, this low dose cross-protection may be insufficient in case of infection with high doses of strains producing Caf1 NT2 or Caf1 NT3 .
The antibody titers positively correlated to the immunity index (Fig 5). For instance, the index of immunity in the Caf1 NT1 immunized group was higher than that in the other groups, and these mice had higher antibody titers. In contrast, mice immunized with Caf1 NT2 or Caf1 NT3 had lower immunity index and lower immunoreactivity in ELISA.
All the Y. pestis Caf1 protein isoforms tested demonstrated a strong serological cross-reactivity as judged by immunochromatography, suggesting that the monoclonal antibody used in this assay is specific to a shared epitope. It may be reasonable to suggest including the strains with Caf1 endemic isoforms for validation of newly developed immune tests targeting this protein.
The antibody response to the immunizations for all three Caf1 isoforms seems to be lower after the second immunization. We can speculate that this phenomenon may be due to the different terms of antibody-titer measurement after the first and second immunizations, i.e. post-immunization-exposure increase from the four to six weeks was sufficient for onset of depletion of humoral immune response. Another supposition is that the antibody titers measured by ELISA went down after the booster immunizations might be due to the action of regulatory cells. Noteworthy, they were higher after the first vaccination with the NT1 isoform, suggesting that this standard vaccine protein has a superior immunogenicity compared to the minor isoforms. However, after the boost, this difference substantially disappeared, indicating that NT2 and NT3 proteins can also be used for a vaccine, given that immunization is performed in two subsequent injections. These two seemingly less immunogenic isoforms may in fact prove to be better vaccines and induce a longer lasting immunity to Y. pestis due to a lower capacity to induce regulatory cells [64]. To test these possibilities, more experiments have to be performed.

Conclusions
The main question addressed in this study was whether specific immunity induced by Y. pestis Caf1 NT1 vaccine can be protective against Y. pestis strains harboring NT2 or NT3 allele types. Our data clearly demonstrate that animals vaccinated with Caf1 NT1 acquire immunity to all the three bacterial strains tested. Therefore, polymorphism in the caf1 cannot be considered a sufficient instrument that would allow Y. pestis an escape from Caf1 NT1 -mediated anti-plague immunity in the case of a low-dose flea-borne infection.