Ethics and biosafety

This study was approved by the Ethics Committee of the Faculty of Tropical Medicine, Mahidol University (MUTM 2015-002-01, MUTM 2018-039-02 and MUTM-EXMPT 2020–005). Written informed consent was obtained from all subjects enrolled in our previous studies [5,35]. Work with live B. pseudomallei was performed at a biosafety level (BSL-3) following the protocol that was approved by the Institutional Biosafety Committee of the Faculty of Tropical Medicine, Mahidol University (TM-2020-005).

Animal experiments were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The University of Nevada, Reno Institutional Animal Care and Use Committee (IACUC) or University of Georgia IACUC approved the experiments.

DNA extraction and whole genome sequencing (WGS) of B. pseudomallei isolates

B. pseudomallei isolates were collected from primary clinical specimens obtained from 1,283 melioidosis patients who participated in the DORIM study between July 2015 and December 2018 [5]. The genomic DNA was extracted from B. pseudomallei using a QIAamp DNA mini kit (Qiagen) [43]. Whole genome sequencing was carried out using a 150-base-read library on the Illumina HiSeq2000 system with a 100-cycle paired-end run. The sequences were assembled and annotated sequences using Velvet v.1.2.10 [44] and obtained from our previous study, which deposited in European Nucleotide Archive (ENA) under study accession number PRJEB25606 and PRJEB35787 [43]. The bacterial strains used in this study are listed in S1 Table.

Analysis of hcp1 variants and other T6SS1-related genes

Variations of hcp1, tssB, tssC and tssE sequences of clinical B. pseudomallei isolates were identified from assembly data using the blastn function with default parameters (word size: 11, Match: 2, Mismatch: 3, Existence: 5, Extension: 2) in CLC genomic workbench version 20 (Qiagen). B. pseudomallei K96243 was used as a wild-type. Nucleic acid sequences of each allele were translated into amino acid sequences using standard genetic code. Representative nucleic acid or amino acid sequences were aligned using CLUSTAL W for multiple alignments [45] in the MEGA 11 [46]. The evolutionary analysis was performed using the Maximum Likelihood method with 1,000 replicates of bootstrap analysis. Tamura-Nei model and Jones-Taylor-Thornton (JTT) models were used for substitution analysis of nucleic acid and amino acid sequences, respectively [47,48]. The hcp1 variants identified by WGS analysis were verified. We performed Sanger sequencing to confirm the DNA sequences of three variant alleles, including allele 6, allele 7, and allele 8 using c_d_bpss1498-wt_F and c_d_bpss1498-wt_R primer pair (S2 Table). PCR products were purified with ExoSAP-IT (Applied Biosystem) before DNA sequencing.

Production of recombinant Hcp1 (rHcp1) proteins

hcp1 genes were amplified from genomic DNA of B. pseudomallei K96243 for hcp1^wt, DR1235 for hcp1^{allele 8} (Hcp1^{variant A}) and DR0089 for hcp1 ^{allele 6} (Hcp1^{variant B}) with primers listed in S2 Table. PCR products were cloned into the pGEM-T Easy vector (Promega). hcp1 genes were subcloned into the NcoI and XhoI restriction sites of expression vector pET28a to produce recombinant Hcp1 proteins (rHcp1^wt, rHcp1^{variant A} or rHcp1^{variant B}) with an N-terminal 6×His-tag. The insert hcp1 sequences in pGEM-T Easy and pET28a vectors were verified by DNA sequencing. The recombinant Hcp1 expression constructs were transformed into Escherichia coli BL21DE (Invitrogen). Expression of the recombinant His-tagged Hcp1 proteins was induced using isopropyl β-D-1-thiogalactopyranoside (IPTG) and proteins were purified by affinity chromatography using HisPur Ni-NTA Resin (Thermo Scientific). Coomassie blue staining of purified rHcp1^wt, rHcp1^{variant A} and rHcp1^{variant B} via SDS-PAGE showed single bands of protein with molecular weights between 19–21 kDa (S1A Fig). Western blot analysis of these proteins using an anti-His antibody (Sigma; H1029) showed the same banding patterns, indicating that the purified recombinant Hcp1 proteins were of high quality (S1B Fig), suitable for further characterization. For the ELISpot assay, endotoxin was removed from the purified proteins using Pierce high-capacity endotoxin removal resin (Thermo Scientific). Endotoxin levels were determined using a Pierce LAL chromogenic endotoxin Quantitation kit (Thermo Scientific).

Hcp1 antibody production

To obtain polyclonal antibodies directed against Hcp1^{variant A}, purified His-tagged rHcp1^{variant A} (5 μg/dose) formulated in tissue culture-grade PBS (pH 7.2; Gibco) with Alhydrogel 2% (250 μg/dose; Brenntag) and CpG (10 μg/dose; ODN 2006; Invivogen) was administered to female C57BL/6 mice (n = 6 mice per group; Charles River Laboratories) subcutaneously on days 0, 21, and 35. Terminal bleeds were conducted 1 week after the final dose. Polyclonal mouse serum generated against recombinant Burkholderia mallei Hcp1 (rHcp1) as part of previous studies was available for use [32,38]. Serum was stored at -80°C until required for use.

The rHcp1-specific monoclonal antibody #3 (mAb-H1-3) was generated by fusing splenocytes obtained from a BALB/c mouse immunized with rHcp1 [32], with Sp2/mIL6 cells (ATCC CRL 2016). The fused cells were plated in methylcellulose medium containing hypoxanthine, aminopterin, and thymidine using a ClonaCell HY kit per the manufacturer’s specifications (Stemcell Technologies). Hybridomas secreting antibodies specific to Hcp1 were identified by ELISA.

Enzyme-linked immunosorbent assay (ELISA)

ELISA was used to determine IgG antibody levels against rHcp1^wt, rHcp1^{variant A} and rHcp1^{variant B} in the plasma of melioidosis patients and healthy donors from Northeast Thailand, who were randomly selected from the previous study [5], as previously described [32]. EDTA plasma samples were collected from melioidosis patients on the day bacterial culture results were reported positive for B. pseudomallei [5]. ELISA was performed in duplicate using 50 μl of 2.5 μg/ml rHcp1^wt, rHcp1^{variant A} or rHcp1^{variant B} protein. Primary antibodies (human plasma, mAb H1-3, anti-rHcp1 [32,38] or anti-rHcp1^{variant A}) and secondary antibodies (anti-human IgG conjugated with a horseradish peroxide (HPR; Dako; P0214) for human antibody detection and anti-mouse immunoglobulins conjugated with HRP (Dako; P0260) for mouse antibody detection) were diluted 2,000-fold before used. Mouse monoclonal anti-polyhistidine antibody (Sigma; H1029) was diluted 3,000-fold before use. A positive result was determined using an OD 450 nm cut-off value of 1.165, as previously evaluated by Receiver Operating Characteristic (ROC) for the diagnosis of melioidosis in the Thai population [32].

Western blot analysis

Purified rHcp1^wt, rHcp1^{variant A} or rHcp1^{variant B} was mixed with Laemmli sample buffer at a concentration of 300 μg/ml and heated to 95°C for 10 min. Five microliters of each sample were loaded into 15% acrylamide gel electrophoresis and then transferred to the PVDF membrane by electrotransfer. Membranes were blocked with 5% skim milk for 1 h and then reacted with 1:2,000 dilution of rHcp1-specific mAb H1-3, anti-rHcp1 [32,38] or anti-rHcp1^{variant A} polyclonal in PBS at room temperature for 1 h. The membranes were then probed with a 1:5,000 dilution of HRP-conjugated rabbit anti-mouse antibody (Dako; P0260) in PBS at room temperature for 1 h. The immunoblots were then developed with a 3,3’-diaminobenzidine (DAB) substrate.

Peripheral blood mononuclear cell (PBMC) isolation

PBMCs were collected from melioidosis patients and healthy donors in Northeast Thailand, randomly selected from the previous study [35]. In brief, PBMCs were isolated from 15 ml of heparinized blood by density gradient centrifugation in Lymphoprep (Axis Shield) using a Sepmate tube (STEMCELL technologies) [35]. Heparinized blood was diluted with an equal volume of RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS) before PBMC isolation. The purified PBMCs were resuspended in FBS supplemented with 10% dimethyl sulfoxide (DMSO) for storage in liquid nitrogen until use.

IFN-γ ELISpot assays

Cellular immune responses to rHcp1^wt, rHcp1^{variant A} and rHcp1^{variant B} were measured using IFN- γ ELISpot assays as previously described [35]. Frozen PBMC were rapidly thawed and gently diluted into a pre-warmed RPMI medium containing 10% FBS and Benzonase. The cells were then washed twice with RPMI supplemented with 10% FBS. PBMCs were resuspended in CTL-TEST medium supplemented with 1% L-glutamine and incubated at 37°C with 5% CO₂ for 1 h before viability was determined. 2 ×10⁵ PBMCs were stimulated with 25 μg/ml of rHcp1^wt, rHcp1^{variant A} or rHcp1^{variant B} in 96-well plates pre-coated with anti-human IFN-γ antibody at 37°C with 5% CO₂ for 24 h. Phytohemagglutinin (PHA) was used as a positive control. IFN-γ secretion was detected using a Human IFN-γ single-color enzymatic ELISpot kit (Cellular Technology Ltd.). The ELISpot plates were processed and developed per the manufacturer’s instructions. Plates were imaged using an ImmunoSpot S6 Micro analyzer and IFN-γ-secreting T cells were quantitated using the ImmunoSpot v5.1 professional DC smart count software (Cellular Technology Ltd.). Experiments were performed in duplicate.

Epitope prediction

The MHC-I and -II binding predictions of Hcp1^wt, Hcp1^{variant A} and Hcp1^{variant B} were conducted using the Immune Epitope Database (IEDB) analysis resource with NetMHCpan 4.1 EL and NetMHCIIpan 4.1 EL methods [49]. The MHC alleles were selected based on common HLA types in the Thai population, including HLA-A*11:01, -B*46:01, -C*01:02, -DRB1*12:02, -DQA1*01:01 and -DQB1*05:02 [50]. The epitope lengths for MHC-I and -II were set at 9–10 and 15 amino acids, respectively.

Cell culture and multinucleated giant cell (MNGC) formation assays

MNGC formation induced by B. pseudomallei was performed using A549 cells essentially as previously described [51]. Briefly, A549 cells at 1.5 × 10⁴ in 96-well plates were infected with B. pseudomallei at a multiplicity of infection (MOI) 50 and incubated at 37°C with 5% CO₂ for 2 h. Extracellular bacteria were removed by aspiration and any remaining extracellular bacteria were killed with 250 μg/ml kanamycin. Infected cells were incubated for a further 8 h then fixed with 4% paraformaldehyde in PBS for 30 min and stained with Giemsa as previously described [52]. A MNGC was defined as a single cell with three or more nuclei. The percentage of MNGC formation efficiency was calculated by dividing the number of nuclei within multinucleated cells by the total number of nuclei × 100. The average number of nuclei in a MNGC was determined to represent the MNGC size. Experiments were performed in triplicate and four independent experiments were conducted. The bar graphs show the mean of percentage of MNGC formation efficiency or average number of nuclei from four independent experiments with error bars representing the standard deviation (SD).

hcp1 gene expression

One hundred microliters of overnight culture of B. pseudomallei in LB broth was inoculated into fresh 2 ml of RPMI 1640 medium and then incubated at 37°C with shaking at 200 rpm for 5 h. hcp1 gene expression was stimulated by 200 μM glutathione (Sigma) for 2 h. One milliliter of bacterial culture was collected for RNA extraction using RNeasy mini kit with RNAprotect Bacteria reagent (Qiagen). DNA contamination was treated by RNase-Free DNase as described by the manufacturer (Qiagen) and verified by no PCR product of dnaK gene.

hcp1 gene expression was determined using iTaq Universal SYBR Green One-Step kit (Bio-rad). Expression of dnaK gene was used for normalization (ΔCt) [40]. The normalized hcp1 gene expression levels were compared between conditions with and without glutathione stimulation (ΔΔCt). Relative gene expression was calculated by 2^-ΔΔCt. The primers for hcp1 and dnaK genes expression are shown in S2 Table.

Genetic manipulation of B. pseudomallei

To generate a hcp1 deletion mutant, bpss1498 from B. pseudomallei strain K96243 was deleted by allelic replacement using established methods [53]. In brief, 219 bp upstream and 462 bp downstream fragments of hcp1 gene were amplified and ligated together to generate a knockout fragment that was then cloned into pGEM-T Easy vector before being subcloned into pEXKm5 at NotI and EcoRI restriction cut sites. The resulting construct was transferred to B. pseudomallei K96243 by conjugation with E. coli RHO3. The mutation by allelic replacement was selected for by sucrose induction at room temperature. The hcp1 mutant, referred to B. pseudomallei K96243Δhcp1, was confirmed by PCR and DNA sequencing.

hcp1^{allele 6} from B. pseudomallei DR1235 for Hcp1^{variant A} and hcp1^{allele 7} from B. pseudomallei DR0089 for Hcp1^{variant B} were introduced into B. pseudomallei K96243Δhcp1 by allelic replacement as described above. The insertion fragment was designed by insertion of hcp1 variants between upstream and downstream of the deletion fragment. The insertion fragments were synthesized by Gene Universal Inc (DE, USA) and cloned into pEXKm5. The insertion vectors were transferred to B. pseudomallei K96243Δhcp1 by conjugation with E. coli RHO3. The allelic replacement was induced by a culture of the bacteria on a sucrose medium. Two hcp1 mutants with variant genes referred to B. pseudomallei K96243Δhcp1::hcp1^{allele 6} and B. pseudomallei K96243Δhcp1::hcp1^{allele 7} were confirmed by PCR and DNA sequencing. The primers for the genetic manipulation of B. pseudomallei are shown in S2 Table.

Protein crystallization

The initial crystallization and screening of rHcp1^{variant B} was performed using an Oryx8 automated crystallization (Douglas Instrument). The protein concentration was adjusted to 10 mg/ml in PBS pH 7.2 and crystallized using the sitting drop vapor diffusion method across 288 conditions, employing Crystal Screen Cryo (Hampton Research; HR2-133), JCSG-plus (Molecular Dimensions; MD1-37), and Morpheus (Molecular Dimensions; MD1-46) at 18°C. The crystals of the rHcp1^{variant B} were grown in condition consisting of 1.5 M ammonium sulfate, 3.75% v/v 2-propanol, and 25% glycerol. The crystals were frozen in liquid nitrogen for storage and diffraction experiments. The resulting crystals were of high quality and diffracted to 1.58 Å resolution.

Diffraction data collection and crystal structure analysis

Diffraction data sets were collected at the TPS05a beamline at National Synchrotron Radiation Research Center (NSRRC, Hsinchu, Taiwan) with a Rayonix MX300HS CCD detector at the temperature of 100 K and a wavelength of 0.999840 Å. The diffraction data were processed using autoPROC [54]. Data collection and processing statistics are shown in S3 Table.

All computations were carried out using programs from the CCP4 suite, unless otherwise stated [55,56]. The initial molecular replacement solution was obtained using Phaser and the Hcp1^wt monomer (PDB ID 3wx6)[30] as a template. The structure was refined with REFMAC [57] alternating with manual model correction in COOT [58]. Hcp1^{variant B} structure has been deposited with the PDB (PDB ID 8Z7K). The details of the structure analysis are described in S1 Text.

Statistical analysis

The Wilcoxon matched-pairs signed-ranks test was used to analyze humoral and cellular responses to the various rHcp1 proteins assessed in this study. The Student’s t-test was used to test the differences in MNGC formation efficiency among B. pseudomallei containing different Hcp1 types, as well as differences in hcp1 gene expression. ANOVA was performed to evaluate differences in intracellular survival among bacterial strains. P ≤ 0.05 was considered a statistically significant difference. The statistical analysis was performed using STATA version 13.0.

Variation of B. pseudomallei Hcp1 in the International Nucleotide Sequence Database Collaboration (INSDC)

We initially searched for variations of B. pseudomallei Hcp1 amino acid sequence that were available in the public database using protein-protein BLAST (blastp) using WP_004525344, identical to Hcp1 of B. pseudomallei K96243, as query sequence. Nine reference amino acid sequences from RefSeq (NCBI Reference Sequence Database) were found to have 100% coverage with the Hcp1^wt and were annotated from B. pseudomallei (Fig 1A). From the International Nucleotide Sequence Database Collaboration (INSDC), B. pseudomallei Hcp1 amino acid sequences were annotated for 3,154 records on March 4^th, 2024. The evolution of Hcp1 was separated into three clades (Fig 1B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1.

(A) Amino acid alignment and (B) phylogenetic tree based on amino acid sequences of 9 Hcp1 types of B. pseudomallei recorded in the RefSeq: NCBI Reference Sequence Database on March 4^th, 2024, generated using the Maximum Likelihood method and JTT matrix-based model with 1,000 bootstraps. Bootstrap values are shown in red text. The branch lengths are presented under each branch.

https://doi.org/10.1371/journal.pntd.0012758.g001

Clade 1 consisted of 5 Hcp1 types, encompassing a total of 2,936 records. The prominent type, WP_004525344, accounted for 2,929 records, found from both clinical and environmental isolates worldwide (2,194 from clinical sources, 629 from environmental sources, 7 from laboratory stock, and 99 from unknown sources). The remaining Hcp1 types in clade 1 included WP_131116732 (2 records from clinical isolates in Sri Lanka), WP_038768460 (1 record from an environmental isolate in Australia), WP_208807185 (2 records from clinical isolates from India), and WP_119637682 (2 records from clinical isolates in USA).

Clade 2 comprised 3 Hcp1 types: WP_041194730 (3 records from 2 environmental isolates in Australia), WP_038792666 (5 records from 3 clinical isolates from Australia, Thailand and the USA, as well as 2 environmental isolates from Australia), and WP_038782671 (1 record from an environmental isolate in Australia).

Interestingly, 6.6% (209/3,154) of the Hcp1 sequences were found in clade 3, represented by WP_004555009 (59 records from clinical sources and 150 records from environmental sources). These were mainly distributed in Thailand and sporadically found in Lao PDR, Vietnam, India and Australia, indicating distinct evolutionary divergence from the other clades.

Genetic variation of hcp1 in clinical B. pseudomallei isolates

We next examined the Hcp1 diversity in clinical isolates of B. pseudomallei collected from patients admitted at nine different hospitals in Northeast Thailand (S1 Table). In this study, we aimed to examine the variation of B. pseudomallei Hcp1 by retrieving the full length of hcp1 gene from the assembly data. Our goal was to identify all known Hcp1 types annotated in a public database and potentially discover new Hcp1 variants. The analysis identified 8 distinct hcp1 alleles across 1,283 primary clinical isolates (S2 Fig). Among these isolates, the majority (1,208/1,283 isolates, 94.2%) contained a hcp1 that was identical to that of B. pseudomallei K96243 (designated as hcp1^wt or hcp1^{allele 1}). In 50 B. pseudomallei isolates (50/1,283 isolates, 3.9%), there were four hcp1 alleles (hcp1^{allele 2}, hcp1^{allele 3}, hcp1^{allele 4} and hcp1^{allele 5}) with more than 99.6–99.8% identity to hcp1^wt, including synonymous single nucleotide polymorphisms (SNPs). The remaining isolates harbored three additional alleles hcp1^{allele 6} (hcp1^{variant B}; N = 1), hcp1^{allele 7} (N = 14) and hcp1^{allele 8} (N = 10) with identities of 97.5%, 87.1% and 87.3%, respectively, with non-synonymous SNPs compared to hcp1^wt. The hcp1^{allele 6–8} were validated by DNA sequencing. Notably, hcp1^{allele 7} and hcp1^{allele 8} (hcp1^{variant A}) were on a branch separated from other alleles.

Amino acid sequence variation of Hcp1 in clinical isolates of B. pseudomallei

The 8 hcp1 alleles from clinical isolates of B. pseudomallei were translated to only three Hcp1 amino acid sequence types (S3 Fig) out of nine types in the INSDC database (Fig 1). The majority of clinical isolates (1,258/1,283 isolates, 98.1%) harbored Hcp1 sequences identical to the reference strain K96243 (WP_004525344) and are representative of wt (allele 1) and alleles 2–5 shown in S2 Fig We identified two other variants of Hcp1 in the 25 remaining B. pseudomallei isolates. The first Hcp1 variant, designated Hcp1^{variant A} (WP_004555009), was observed in 24 B. pseudomallei isolates (1.9%). The Hcp1^{variant A} was translated from 2 nucleic acid sequences (hcp1^{allele 7} and hcp1^{allele 8} in S2 Fig). Compared to Hcp1^wt, Hcp1^{variant A} had 81.1% amino acid identity (137/169) with two amino acid deletions at positions 96 and 98 in multiple alignments (S3 Fig). Likewise, in hcp1^{allele 7} and hcp1^{allele 8}, the evolutionary analysis demonstrated that Hcp1^{variant A} (WP_004555009) was discrete from other Hcp1 proteins (S3 Fig). The second variant, Hcp1^{variant B} (WP_038792666) was found in only one clinical isolate (DR0089). Compared to Hcp1^wt, Hcp1^{variant B} showed 95.9% identity (163/170) and harbored an amino acid insertion at position 97 of the alignment (S3 Fig).

Additionally, we found that B. pseudomallei isolates harboring a Hcp1^{variant A} or Hcp1^{variant B} were dispersed throughout Northeast Thailand. Twenty-five B. pseudomallei isolates with variants of Hcp1 were cultured from patients admitted to 7 of 9 hospitals in this study. Twenty-four isolates were cultured from Thai patients who resided in 7 provinces in Northeast Thailand and another one isolate was cultured from a Cambodian patient (S1 Table).

Genomic comparison of B. pseudomallei isolates with variants of Hcp1

Whole genome sequencing analysis showed that B. pseudomallei isolates with variant Hcp1 had diverse genetic backgrounds. A total of 223 known sequence types (STs) by multiple locus typing (MLST) analysis were assigned to 1,283 B. pseudomallei isolates in this study (S1 Table) [43]. Twenty-five B. pseudomallei isolates with a variant of Hcp1 were assigned to 17 known STs by MLST analysis and were distributed into 13 of 101 lineages by PopPUNK analysis (S1 Table) [43]. Of these, 9 isolates were distributed in major lineages (lineage 1 to 3) [43].

We further analyzed the whole genome sequences for variation in the T6SS1 gene cluster bpss1493 (BPS_RS26870)–bpss1511 (BPS_RS26960) of B. pseudomallei strains that had a variant of Hcp1. Variation in TssC, a T6SS contractile sheath protein encoded by bpss1497 (BPS_RS26890), was associated with Hcp1 variation. Six amino acid sequence types of TssC were identified in 1,283 B. pseudomallei isolates in this study (S4 Fig and S4 Table). We observed two conserved amino acid substitutions, V457A and V486I, in 3 types of TssC (TssC^{variant C}, TssC^{variant D} and TssC^{variant E}) from the 24 B. pseudomallei isolates that produced Hcp1^{variant A} (S1 Table). These two amino acid substitutions in TssC were not observed in other clinical isolates of B. pseudomallei in this study. B. pseudomallei containing hcp1^{variant B} contained TssC^wt, similar to B. pseudomallei K96243 (S1 Table).

B. pseudomallei isolates containing hcp1^{variant A} were also associated with amino acid substitutions in TssB (a T6SS contractile sheath small subunit) and TssE (a T6SS baseplate subunit) (S1 and S4 Tables and S5 and S6 Figs). Of these 24 isolates, 23 isolates had a conserved amino acid change, T74A, in TssB (TssB^{variant D}; S5 Fig). However, the T74A change in TssB that was found in 3 types of TssB, including TssB^{variant D}, TssB^{variant E} and TssB^{variant F} was not specific to B. pseudomallei strains harboring the Hcp1^{variant A}. This change was also detected in 165 B. pseudomallei isolates with Hcp1^wt or Hcp1^{variant B}. Similar to TssB, 22 B. pseudomallei isolates containing hcp1^{variant A} had conserved amino acid changes, E27A and H32R, in TssE (TssE^{variant F}, TssE^{variant G} and TssE^{variant I}; S1 Table and S6 Fig) but both non-synonymous SNPs were also detected in 35 B. pseudomallei isolates with Hcp1^wt.

Antigenicity of Hcp1

A recent study investigated the impact of a variant hcp1 gene found in environmental B. pseudomallei isolates and its role in virulence in an animal model and found that isolates carrying the variant hcp1 gene (containing hcp1^{variant A}; ST93 and ST60) exhibited a reduced level of virulence compared to strains with hcp1^wt (ST58 and ST176) [42]. It is possible that the different amino acid compositions of Hcp1 proteins may significantly influence their biological properties.

We predicted the MHC-I and -II binding epitopes of Hcp1^wt, Hcp1^{variant A} and Hcp1^{variant B} using IEDB analysis resource. The prediction showed 963, 951, and 969 MHC-I binding epitopes for Hcp1^wt, Hcp1^{variant A}, and Hcp1^{variant B}, respectively. Additionally, 310, 306, and 312 epitopes were predicted to bind with MHC-II for Hcp1^wt, Hcp1^{variant A}, and Hcp1^{variant B}, respectively. We next assessed the cross-reactivity among antibodies against the Hcp1 proteins identified in this study. We performed ELISA and Western blot analyses using three purified rHcp1 proteins, rHcp1^wt, rHcp1^{variant A} and rHcp1^{variant B}. We utilized various antibodies targeting different Hcp1 types, including mouse monoclonal anti-polyhistidine antibody (mAb-6×His), mouse monoclonal antibody specific for Hcp1^wt (mAb H1-3), mouse serum raised against B. mallei rHcp1 by immunization (anti-rHcp1) [32,38] and mouse serum raised against rHcp1^{variant A} by immunization (anti-rHcp1^{variant A}). The reactivity of mAb-6×His demonstrated that the binding affinity of three rHcp1 proteins was comparable when coated on the ELISA plate (Fig 2A and S5 Table). Anti-rHcp1 was used for comparison since Hcp1 from B. mallei and B. pseudomallei K96243 are >99% identical and have only one amino acid difference [59] at position 148 of alignment in S3 Fig.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Antigenic variation of recombinant Hcp1 proteins.

(A) Reactivity of mouse anti-polyhistidine (mAb-6×His), mouse monoclonal antibody against rHcp1^wt (mAb H1-3), mouse sera obtained from B. mallei rHcp1 immunization (anti-rHcp1) and mouse sera obtained from rHcp1^{variant A} immunization (anti-rHcp1^{variant A}) against rHcp1^wt (blue bar), rHcp1^{variant A} (red bar) and rHcp1^{variant B} (green bar). The experiment was performed in duplicate. The bar graph represents mean of OD values and error bar shows SD. (B) Reactivity of human sera from healthy donors, melioidosis patients infected with B. pseudomallei containing hcp1^wt, melioidosis patients infected with B. pseudomallei containing hcp1^{variant A} and melioidosis patient infected with B. pseudomallei containing hcp1^{variant B} against rHcp1^wt (blue dot), rHcp1^{variant A} (red square) and rHcp1^{variant B} (green triangle). The data points represent the mean of duplicate OD values. The black line represents the median of each group. The orange dotted line represents the cut-off value. * P ≤ 0.05, *** P ≤ 0.001. (C) IFN-γ secretion from PBMC of 10 healthy donors and 23 melioidosis patients infected with B. pseudomallei containing hcp1^wt after stimulated with PHA (purple diamond), rHcp1^wt (blue dot), rHcp1^{variant A} (red square) and rHcp1^{variant B} (green triangle). The data points present the mean of duplicate wells. The error bars represented 95% CI. (D) Heat map of IFN-γ secretion from PBMC of 10 healthy donors and 23 melioidosis patients infected with B. pseudomallei containing hcp1^wt after stimulated with PHA, rHcp1^wt, rHcp1^{variant A} and rHcp1^{variant B}. The data points present the mean of duplicate wells.

https://doi.org/10.1371/journal.pntd.0012758.g002

We determined the cross-reactivity of the Hcp1 proteins with mAb H1-3 that recognizes the native form of Hcp1^wt by ELISA. As shown in Fig 2A, mAb H1-3 reacted with rHcp1^wt but not rHcp1^{variant A} or rHcp1^{variant B}, suggesting that mAb H1-3 recognized the epitope affected by amino acid substitutions in both variants, which alter the native form of Hcp1^wt. We next examined the reactivity of rHcp1^wt, rHcp1^{variant A} and rHcp1^{variant B} with anti-rHcp1 polyclonal serum. Anti-rHcp1 recognizes all three types of rHcp1 with strong reactivity toward rHcp1^wt and rHcp1^{variant B} but showed approximately half the reactivity toward rHcp1^{variant A} by ELISA (Fig 2A and S5 Table) and showed only a faint band in Western blot analysis (S7A Fig). Additionally, the decreasing reactivity of anti-rHcp1 with rHcp1^{variant A} was consistently observed in various dilutions of anti-rHcp1 (S8A Fig and S6 Table). Conversely, mice immunized with rHcp1^{variant A} produced antibodies that cross-reacted toward all three Hcp1 proteins at similar levels by both ELISA and Western blot analysis (Figs 2A and S7B). Furthermore, we observed comparable levels of anti-rHcp1^{variant A} activity toward all three types of rHcp1 even when the anti-serum was diluted to 1:10,000 (S8B Fig and S7 Table).

Antibody responses against Hcp1 variants in melioidosis patients

To assess the impact of Hcp1 variation on the humoral immune responses associated with melioidosis patients, we performed ELISAs to determine the reactivity of IgG antibodies against rHcp1^wt, rHcp1^{variant A} and rHcp1^{variant B}. Plasma samples were collected from healthy donors and three groups of melioidosis patients who were infected with B. pseudomallei strains containing hcp1^wt, hcp1^{variant A} or hcp1^{variant B} (Fig 2B and Tables 1 and S8). Plasma samples were obtained from melioidosis patients within 24 h after diagnosis based on the culture method. The median time from admission to blood collection of melioidosis patients infected with B. pseudomallei containing hcp1^wt was 4 days (IQR 3–5), which was not significantly different from the patients infected with B. pseudomallei containing hcp1^{variant A}, with a median time of 3 days (IQR 2.5–4; P = 0.1890, Mann-Whitney U test). In contrast, blood from the patient infected with B. pseudomallei containing hcp1^{variant B} was collected after 9 days of admission. An OD value of 1.165, previously evaluated by ROC for diagnosis of melioidosis in the Thai population [32], was used as the threshold for positive specific antibody responses.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Reactivity of plasma from melioidosis patients against the three recombinant B. pseudomallei Hcp1 proteins.

https://doi.org/10.1371/journal.pntd.0012758.t001

Antibodies from healthy donors had very low reactivity towards rHcp1^wt, rHcp1^{variant A} and rHcp1^{variant B}, with OD 450 values ranging from 0.004–0.352 (Fig 2B and S8 Table). In contrast, melioidosis patients infected with B. pseudomallei containing hcp1^wt showed the highest median levels of IgG antibodies specifically targeting rHcp1^wt, but their antibody levels against rHcp1^{variant A} and rHcp1^{variant B} were lower (P < 0.001 and P = 0.004, respectively; Fig 2B). When an OD 450 value of 1.165 was used as a cut-off to determine a positive result, 16 of 22 (72.7%) melioidosis patients infected with B. pseudomallei containing hcp1^wt were detected (Table 1). Among these patients, 9 (56.3%) and 13 (81.3%) of the patients had antibodies that cross-reacted with rHcp1^{variant A} and rHcp1^{variant B} at lower levels (S8 Table). However, 3 of 16 (18.8%) melioidosis cases had negative reactivities with either rHcp1^{variant A} or rHcp1^{variant B} (S8 Table). Only 6 of 22 patients (27.3%) in the melioidosis patient group were infected with B. pseudomallei strains that produced Hcp1^wt and had negative Hcp1-specific IgG levels for all three rHcp1 proteins (S8 Table).

The reactivity of IgG antibodies from melioidosis patients infected with B. pseudomallei containing hcp1^{variant A} showed no significant difference when tested against rHcp1^wt, rHcp1^{variant A}, and rHcp1^{variant B} (rHcp1^wt versus and rHcp1^{variant A}, P = 0.061; rHcp1^{variant A} versus rHcp1^{variant B}, P = 0.077; and rHcp1^wt versus rHcp1^{variant B}, P = 0.119) (Fig 2B). Reactivity of IgG antibodies against rHcp1^{variant A} was detected in 10 of 24 (41.7%) melioidosis patients infected with B. pseudomallei containing hcp1^{variant A} (S8 Table). Among them, 80.0% (8 of 10) and 70.0% (7 of 10) showed cross-reactivity with rHcp1^wt and rHcp1^{variant B} at higher levels than rHcp1^{variant A}. Only two melioidosis cases produced specific IgG antibodies that reacted with rHcp1^{variant A} alone. Among 24 melioidosis patients infected with B. pseudomallei containing hcp1^{variant A}, 14 (58.3%) were negative for rHcp1^{variant A}-specific antibodies. Of these, 4 of 14 melioidosis patients (28.6%) produced antibodies against rHcp1^wt and rHcp1^{variant B} but did not react to rHcp1^{variant A}. Therefore, 10 of 24 patients (41.7%) in this group had antibody levels against all three recombinant Hcp1 proteins that were below the cut-off value.

The melioidosis patient infected with B. pseudomallei DR0089 that containing hcp1^{variant B} did not have detectable reactivity against purified rHcp1^wt, rHcp1^{variant A} and rHcp1^{variant B} at the time of enrollment or within 24 h after a positive culture result report.

Cellular immune responses against variants of Hcp1

To examine cellular immune responses specific for Hcp1^wt, Hcp1^{variant A} and Hcp1^{variant B}, IFN-γ ELISpot assays were conducted. PBMCs isolated from 23 melioidosis patients infected with B. pseudomallei strains that contained hcp1^wt were stimulated with rHcp1^wt, rHcp1^{variant A} or rHcp1^{variant B}. PBMCs from 10 healthy donors were used as controls. As expected, PBMCs from both melioidosis patients and healthy donors were responsive to stimulation with PHA, a positive control antigen (S9 Table). Although the purified rHcp1^wt, rHcp1^{variant A} and rHcp1^{variant B} contained residual endotoxin levels at 8.84, 12.21 and 47.20 EU/ml, respectively, this did not appear to affect the immune response. The healthy donors from Northeast Thailand, an endemic area of melioidosis, may have been previously exposed to B. pseudomallei during daily life, potentially developing low-level T cell responses to rHcp1 proteins, as previously reported [35]. IFN-γ secretion by PBMCs from healthy donors showed no correlation with endotoxin levels (S9 Table), suggesting that endotoxin contamination did not influence the observed IFN- γ secretion, consistent with previous findings [60, 61]. Moreover, PBMCs from melioidosis patients exhibited IFN-γ-secreting T cell responses when stimulated with all three rHcp1 proteins, with rHcp1^{variant B} inducing significantly IFN-γ secretion than rHcp1^wt and rHcp1^{variant A} (P = 0.0054 and 0.0002; Fig 2C and 2D). The rHcp1^{variant A} stimulated IFN-γ secretion from PBMCs at a level similar to that of rHcp1^wt (P = 0.330).

Multinucleated giant cell formation (MNGC) by clinical isolates of B. pseudomallei

MNGC formation is a unique phenotype associated with B. pseudomallei infections, and it is believed that these structures play a role in enabling the bacteria to evade host immune responses [62]. Previous studies have shown that Hcp1 is essential for MNGC formation by B. pseudomallei in murine macrophages [29]. To investigate a potential role for Hcp1 variation in MNGC formation, A549 cells were infected with B. pseudomallei at MOI 50 for 10 h and then visualized with Giemsa staining. The clinical B. pseudomallei isolates producing variant Hcp1 (DR1235 for Hcp1^{variant A} and DR0089 for Hcp1^{variant B}) were compared with the reference strain K96243 (produces Hcp1^wt) for their abilities to stimulate MNGC formation (Fig 3). While the three isolates survived at similar levels in A549 cells (S9A Fig and S10 Table), their ability to stimulate MNGC formation varied (Fig 3A) in both percentage of MNGC formation and average number of nuclei in MNGC (P < 0.001 for all comparisons; Fig 3B and 3C and S11 Table). As expected, B. pseudomallei K96243 could spread from cell to cell and stimulate the formation of enlarged MNGCs following infection. In contrast, B. pseudomallei DR1235 induced a lower number of MNGCs with a lower number of nuclei. For B. pseudomallei DR0089, an expansion of the cytoplasm of the infected cells was observed, but MNGC formation was infrequently detected at 10 h of infection.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Multinucleated giant cell formation efficiency of clinical B. pseudomallei isolates with variant hcp1 genes in A549 cells.

(A) Multinucleated giant cell formation in A549 cells infected with strains K96243 (Hcp1^wt), DR1235 (Hcp1^{variant A}) and DR0089 (Hcp1^{variant B}) at MOI 50 for 10 h. The bar scale represents 20 μM length. MNGC formation efficiency was calculated by determining the percentage of a MNGC formation (B) and the average number of nuclei in each MNGC (C). The bar graphs show the mean of three independent experiments. The error bars represent the standard deviation (SD).

https://doi.org/10.1371/journal.pntd.0012758.g003

Genomic diversity of B. pseudomallei influences MNGC formation

It is possible that the MNGC formation associated with clinical isolates of B. pseudomallei may be influenced by other genes related to intracellular survival and/or transcriptional regulation of T6SS1 genes rather than this being a direct effect of Hcp1 variation. We, therefore, examined the levels of hcp1 expression by isolates harboring the three different Hcp1 variants. Since glutathione stimulates hcp1 gene expression in vitro, simulating the intracellular phase of infection [63], we determined hcp1 expression by B. pseudomallei K93243, DR1235 and DR0089 under glutathione induction. The glutathione-induced expression of hcp1 in B. pseudomallei isolates DR1235 (Hcp1^{variant A}) and DR0089 (Hcp1^{variant B}) was significantly lower than in B. pseudomallei K96243 (P = 0.023 and 0.010, respectively; S9B Fig and S12 Table). The low hcp1 expression (S9B Fig) and MNGC formation efficiency (Fig 3B and 3C and S11 Table) suggest that the level of hcp1 gene expression may influence the efficiency of MNGC formation in A549 cells.

Next, we selected additional clinical isolates of B. pseudomallei expressing Hcp1^wt and Hcp1^{variant A} to compare their abilities to stimulate MNGC formation in A549 cells. Fourteen isolates from each group, representing different multi-locus sequence types (STs), were included for MNGC formation analysis (S13 Table). We observed individual variations in MNGC formation efficiency among clinical B. pseudomallei isolates. However, the group expressing Hcp1^{variant A} (with 13 STs) showed lower levels of MNGC formation compared to the group expressing Hcp1^wt (with 11 STs) (P = 0.027; S9C Fig and S13 Table). However, a group of clinical B. pseudomallei isolates containing hcp1^{variant A} showed varying abilities to form MNGC, suggesting that factors beyond the hcp1 allele, potentially by the different genetic backgrounds of B. pseudomallei isolates, may affect gene expression and the ability to stimulate MNGC formation.

Multinucleated giant cell formation (MNGC) efficiency of B. pseudomallei K96243 with hcp1 variants

Expression of different hcp1 alleles by clinical isolates of B. pseudomallei may affect MNGC formation efficiency. To address the possibility that different genetic backgrounds might influence MNGC formation efficiency, we performed mutagenesis experiments to generate derivative strains of B. pseudomallei K96243 that express Hcp1^{variant A} or Hcp1^{variant B}. To facilitate these studies, we first constructed a hcp1 deletion mutant in B. pseudomallei K96243 (K96243Δhcp1) and then used this mutant to generate K96243Δhcp1::hcp1^{allele 8} (expresses variant A) and K96243Δhcp1::hcp1^{allele 6} (expresses variant B). These two strains were then tested for their ability to survive in A549 cells (S10A Fig and S14 Table), levels of hcp1 gene expression (S10B Fig and S15 Table) and MNGC formation ability (Fig 4) in comparison to K96243 and K96243Δhcp1. All four strains tested, including the wild type K96243, K96243Δhcp1, K96243Δhcp1::hcp1^{allele 8} and K96243Δhcp1::hcp1^{allele 6}, were able to replicate in A549 cells at similar levels (P > 0.05 for all time points; S10A Fig). In addition, hcp1 expression levels under glutathione stimulation by K96243Δhcp1::hcp1^{allele 8} and K96243Δhcp1::hcp1^{allele 6} were comparable to wild type K96243 (P > 0.05 for all comparisons; S10B Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Multinucleated giant cells (MNGC) formation efficiency of genetically manipulated B. pseudomallei K96243 with variant hcp1 genes in A549 cells.

(A) MNGC of genetically manipulated B. pseudomallei isolates (strain K96243, K96243Δhcp1, K96243Δhcp1::hcp1^{allele 6} and K96243Δhcp1:: hcp1^{allele 7}) after infection at MOI 50 for 10 h. The bar scale represents 20 μM length. MNGC formation efficiency is presented by the percentage of MNGC formation (B) and average nuclei in MNGC (C). The bar graphs show the mean of three independent experiments. The error bars represent the standard deviation (SD).

https://doi.org/10.1371/journal.pntd.0012758.g004

Four different B. pseudomallei strains (K96243, K96243Δhcp1, K96243Δhcp1::hcp1^{allele 8} and K96243Δhcp1::hcp1^{allele 6}) with the same genetic background except for the hcp1 gene, were evaluated for MNGC formation efficacy after infecting A459 cells for 10 h (Fig 4 and S16 Table). Consistent with previous reports [29,30], we observed that the loss of hcp1 results in an inability to stimulate MNGC formation, although the bacteria could still survive and multiply within the cells. Expression of the Hcp1^{variant A} in B. pseudomallei K96243Δhcp1::hcp1^{allele 8}, restored MNGC formation to levels comparable to the wild-type K96243 (P = 0.131 for % MNGC formation and P = 0.337 for average nuclei in MNGC). In contrast, expression of Hcp1^{variant B} in B. pseudomallei K96243Δhcp1::hcp1^{allele 6} resulted in a slightly increased MNGC formation efficiency compared to K96243Δhcp1 (P < 0.0001 for both % MNGC formation and average nuclei in MNGC). However, this strain expressing Hcp1^{variant B} exhibited significantly decreased levels of cell-to-cell spread compared to the wild-type K96243 (P < 0.0001 for both % MNGC formation and average nuclei in MNGC).

Structural analysis of Hcp1^{variant B}

Based on the observation that strains expressing Hcp1^{variant B} did not induce MNGC formation, we hypothesized that this amino acid substitution might affect protein structure and the formation of the tube-like structure of the T6SS1. Therefore, we conducted protein crystallization experiments to examine the structure of Hcp1^{variant B}.

Optimized crystallization conditions allowed us to obtain better quality crystals of Hcp1^{variant B} (PDB ID 8z7k, resolution of 1.58 Å) than those used to determine the Hcp1^wt structure (PDB ID 3wx6, 2.7 Å) [30]. The new data made it possible: to realize that the crystals are partially disordered and to understand their overall organization, to build and refine a sensible periodic representation of the structure, to re-solve and analyze the structure of 3wx6 and to prove that it is isomorphous to ours, and suffers from the same crystal pathology which was wrongly diagnosed as twinning.

Since the 3wx6 structure had been solved, the structure of the T6SS tubes was determined for Vibrio cholerae by cryogenic electron microscopy (cryo-EM), PDB ID 5ojq [64]. The structure revealed a head-to-tail stacking of Hcp1 hexamers in the tubes, meaning that the dodecamers observed in both 3xw6 and our structure are crystallographic artifacts. Interestingly, the head-to-tail tubes of Hcp1 had already been observed in a crystal structure of Hcp1 from Pseudomonas aeruginosa (1y12) [65] long ago and superposed reasonably well on the tubes reported in the above cryo-EM study, with the main difference being a several-degree twist present in the EM which would not be possible in a crystal structure. The comparison with the above structures allowed us to analyze the possible effect of the mutations on the T6SS tube formation, as illustrated in Fig 5A–5D. Namely, mutations H90Y, R91S, T94A, T96K and T96_T97insE are located either closer to or inside the loop P92-T102 (WT numbering), which is essential for inter-hexamer contacts (Fig 5C), and mutation Q14T is located at the outer surface of the hexamer (Fig 5D) and can be important for the interaction with the sheath proteins. The mutated residues T14, Y90 and S91 are well-defined in our structure (Fig 5C and 5D). Interestingly, Q14 and R91 are conserved in V. cholerae and P. aeruginosa (Fig 5E). Additionally, the structure analysis comparing our X-ray structure of Hcp1 with a cryo-EM structure of T6SS sheath/tube complex in V. cholerae suggested that the amino acid substitution of T167K located near the C-terminus of protein did not affect the structure.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Structural analysis of B. pseudomallei Hcp1^{variant B}.

The superposition of the T6SS tubes, comprising Hcp1 and sheath proteins, of B. pseudomallei and V. cholerae (5ojq) showed an overall structural similarity and identified intermolecular interfaces that may be impacted by mutations associated with Hcp1^{variant B}. X-ray structure of Hcp1^{variant B} (orange, yellow and green) was superposed onto Hcp1^wt (3wx6; light blue and grey). AlphaFold2 prediction of B. pseudomallei sheath protein (magenta) was superposed onto sheath (cyan) proteins in the cryo-EM model of T6SS tubes from V. cholerae (5ojq). (A) Two superposed subunits of B. pseudomallei Hcp1^{variant B} (orange and green) highlight a head-to-tail stacking of the Hcp1 hexamers in the T6SS tubes and a twist of the adjacent hexamers. (B) The superposed hexamer of B. pseudomallei Hcp1^{variant B} (orange and yellow) is formed by symmetry related subunits from the crystal structure and superposition highlights its strong structural conservation. (C) and (D) are zoom-ins of (A) and (B), respectively. (C) Four mutations and an insertion between the residues 90 and 97 are likely to have a strong effect on the conformation of the loops involved in the interactions between hexamers, while (D) the mutation Q14T affects the interaction of the Hcp1 hexamer with the sheath. Mutated residues in (C) and (D) are shown in sticks and the adjacent V. cholerae Hcp1 subunits in (B) are shown in light blue. (E) The amino acid sequence alignment of B. pseudomallei Hcp1^wt (3wx6) and B. pseudomallei Hcp1^{variant B} (8z7k) from, V. cholerae Hcp1 (5ojq) and P. aeruginosa Hcp1 (1y12). Highlighted regions indicate unmodelled amino acids in Hcp1^{variant B} crystal structure with mutations shown in red and conserved amino acids shown in pink.

https://doi.org/10.1371/journal.pntd.0012758.g005

In conclusion, the structural study confirmed the preservation of the hexameric structure in Hcp1^{variant B}. This, however, required resolving the 3wx6 structure and verifying that it shared similar crystal pathology with our structure. The possible effect of the mutations on T6SS tube formations in Hcp1^{variant B} was analyzed by comparing our X-ray structure of Hcp1 with a cryo-EM structure of T6SS sheath/tube complex in V. cholerae. The amino acid substitutions in Hcp1^{variant B} may have a negative effect on T6SS tube formation and, as a result, reduce its effectiveness in MNGC formation.

Funding

This research project was supported by Mahidol University to ST. NC and TEW were supported by the US National Institutes of Health U01AI115520. PJB and MNB were supported by Defense Threat Reduction Agency contract HDTRA1-18-C-0062. CC was funded by the Wellcome International Intermediate Fellowship (216457/Z/19/Z) and the Sanger International Fellowship. This research was funded in part by the Wellcome Trust [220211] to NC. For the purpose of Open Access, the author has applied a CC BY public copyright license to any Author Accepted Manuscript version arising from this submission. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.