Development of a groEL gene–based species-specific multiplex polymerase chain reaction assay for simultaneous detection of Vibrio cholerae, Vibrio parahaemolyticus and Vibrio vulnificus

Authors


Correspondence

In-Soo Kong, Department of Biotechnology, Pukyong National University, Busan 608-737, Korea. E-mail: iskong@pknu.ac.kr

Abstract

Aims

To develop an effective multiplex polymerase chain reaction (PCR) for the simultaneous detection of three important Vibrio species, Vibrio cholerae (Vc), V. parahaemolyticus (Vp) and V. vulnificus (Vv) using the groEL gene, a potential phylogenetic marker.

Methods and Results

Three species-specific primer sets were designed to target Vc, Vp and Vv. A total of 131 Vibrio and non-Vibrio strains were used to determine the specificity and sensitivity of primers. The primers produced specific PCR fragments from all target species strains and did not cross-react with other Vibrio and non-Vibrio species. This PCR method showed good efficiency in detecting coexisting target species in the same sample with a detection limit of 100 pg of Vc, Vp and Vv from mixed purified DNA. Detection of three target species was also possible from artificially inoculated shellfish, flounder and sea water.

Conclusions

The groEL gene is a potential marker for accurate simultaneous detection of Vc, Vp and Vv and could be used to detect these species in environmental and clinical samples.

Significance and Impact of the Study

This newly developed multiplex PCR is a useful and cost-effective method that is applicable in a disease-outbreak prediction system and may provide an effective tool for both the epidemiologist and ecologist.

Introduction

Vibrio species are naturally diverse bacteria that inhabit aquatic environments and marine animals as symbionts and commensals (Izumiya et al. 2011). Among more than 70 identified Vibrio species, Vibrio cholerae, V. parahaemolyticus and V. vulnificus are of major concern as they are pathogenic to animals, including humans (Oliver 1989; Thompson et al. 2004; Tracz et al. 2007). All three species are commonly associated with sea water, sediment, shellfish and the intestinal contents of fish (Wong et al. 2012). Under optimum conditions, these species are incorporated in high quantities by aquatic organisms, especially by filter feeders such as mussels, clams and oysters, which concentrate these bacteria in their muscle due to their filter-feeding habit. Infection by V. cholerae, V. parahaemolyticus and V. vulnificus occurs through ingestion of contaminated seafood or exposure to aquatic environments (Oliver 2006; Ottaviani et al. 2009). V. cholerae is the etiological agent of cholera, an acute dehydrating diarrhoea that occurs in epidemic form throughout the world, particularly in developing countries (Faruque and Nair 2002), whereas the non-O1 and non-O139 strains are involved in sporadic infection. V. parahaemolyticus is a leading pathogen that causes seafood-borne gastroenteritis worldwide, including developed countries such as the United States and Japan (Nair et al. 2007). Another pathogen, V. vulnificus, causes gastroenteritis, septicaemia and severe wound infection with a high mortality in susceptible persons (Jones and Oliver 2009). Moreover, both V. parahaemolyticus and V. vulnificus strains can cause diseases in aquatic organisms, including economically important fish and shrimp (Thompson et al. 2004).

Various polymerase chain reaction (PCR)–based methods have been reported for Vibrio species identification. These methods include real-time PCR, micro-arrays and multiplex PCR. However, the first two detection methods are costly due to the requirement for expensive instruments, whereas the multiplex PCR method that detects single or multiple species targets is effective. Multiplex PCR has been proven to provide rapid and highly sensitive methods for the specific detection of micro-organisms (Fan et al. 2008) and can be easily performed in diagnostic laboratories. Many virulence genes, such as omp, ctx, zot, ace, tcp, rtx, sto and hly in V. cholerae; tdh, trh and toxR in V. parahaemolyticus; and vvh, viuB and toxR in V. vulnificus, have been targeted for species-specific detection by uniplex or multiplex PCR (Lalitha et al. 2008; Neogi et al. 2010; Teh et al. 2010). According to Neogi et al. (2010), to accurately detect particular individual species, it is critical to address unresolved complications such as the precise differentiation of V. parahaemolyticus from closely related species, the simultaneous detection of all target species in a sample and the coexistence of V. cholerae, V. parahaemolyticus and V. vulnificus in costal environments, diseased animals, seafood or aquaculture (Gopal et al. 2005; Mahmud et al. 2008).

In the environment, Vibrio species can exchange genetic elements such as virulence genes or undergo deletion of a particular gene (Izumiya et al. 2011), resulting in the possibility of obtaining false-positive or false-negative results when targeting virulence genes. To overcome this problem, new PCR methods have been developed targeting housekeeping genes like pho, amiB, dnaJ, gyrB, rpoA and rpoB (Thompson et al. 2005; Nhung et al. 2007). The groEL gene encodes the chaperonin GroEL (synonyms are Cpn 60, GroL, Hsp 60, and Mop A), which plays an essential role in the control of cellular stress and is also a powerful phylogenetic marker (Junick and Blaut 2012). The superiority of the groEL gene compared to 16S rRNA and 23S rRNA has already been reported in the detection of Vibrio species (Nishibuchi 2006; Yushan et al. 2010). This gene has been shown to be a suitable marker for the successful detection of many bacteria including Vibrio species (Kim et al. 2010, 2012; Yushan et al. 2010; Hossain et al. 2011). Therefore, the aim of this study was to develop a suitable multiplex PCR method using the groEL gene that can be used for accurate simultaneous species-specific detection of V. cholerae, V. parahaemolyticus and V. vulnificus.

Materials and methods

Bacterial culture and DNA extraction

A total of 131 bacterial strains were used in this study (Table 1), including 9 V. cholerae strains, 70 V. parahaemolyticus strains and 11 V. vulnificus strains. All Vibrio species were cultured in brain heart infusion (BHI; BD, Franklin Lakes, NJ, USA) broth with 2·5% sodium chloride, while the other bacterial strains were cultured in Luria–Bertani (LB, USB, Cleveland, OH, USA) or BHI broth. Genomic DNA of all Vibrio and non-Vibrio strains purified and identified by 16S rRNA in our previous study (Hossain et al. 2011) was used as template DNA. Template DNA was also extracted from the target Vibrio species by the simple boiling method as described by Kim et al. (2008).

Table 1. Strains used in this study
 Micro-organismsSource or referenceUniplex PCRMultiplex PCR
VcVpVv
  1. L, laboratory collection; C, clinical strain; E, environmental strain; ATCC, American Type Culture Collection, USA; KCCM, Korean Culture Center of Microorganisms, Korea, KCTC, Korean Collection for Type Cultures, Korea.

  2. +, only amplification product of 418 bp for Vibrio cholerae, 644 bp for V. parahaemolyticus and 192 bp for Vibrio vulnificus; -, no amplification products.

  3. Vc, Vp and Vv represent V. cholerae, V. parahaemolyticus and V. vulnificus, respectively.

1 Vibrio aestuarianus KCCM 40863
2 V. alginolyticus KCTC 2472, 3 E
3 V. anguillarum KCTC 2711, J-O-2, J-O-3, YT, NB10
4 V. campbellii KCCM 41986
5 V. cholerae KCCM 41626, KCTC 2715, 2 C, 5 E+ +
6 V. cincinnatiensis KCTC 2733
7 V. damsella E
8 V. diazotrophicus KCCM 41606
9 V. fluvialis ATCC 33809
10 V. furnissii KCTC 2731, E
11 V. harveyi KCCM 40866
12 V. hollisae KCCM 41680
13 V. logei KCTC 2721
14 V. mediterranei KCCM 40867
15 V. metschinikovii KCTC 2736
16 V. mimicus ATCC 33653
17 V. natriegens KCCM 40868
18 V. navarrensis KCCM 41682
19 V. nereis KCCM 41667
20 V. ordalii KCCM 41669
21 V. parahaemolyticus KCCM 41664, KCCM 11965, KCTC 2471, 37 E, 30 C + +
22 V. proteolyticus KCTC 2730
23 V. tubiashii KCTC 2728
24 V. vulnificus KCCM 41665, KCTC 2962, KCTC 2980, KCTC 2981,KCTC 2982, KCTC 2983, KCTC 2985, KCTC 2986, KCTC 2987, 2 E

+

+

+

+

+

+

25 Aeromonas hydrophila KCTC 2358
26 Escherichia coli L, E
27 Edwardsiella tarda KCTC 12267, E
28 E. ictaluri ATCC 33202
29 Enterobacter cloacae E
30 Klebsiella oxytoca E
31 K. pneumoniae E
32 Salmonella typhi E
33 Shigella flexneri E
34 S. sonei E

Oligonucleotide primers

All available sequences of the groEL gene among Vibrio and non-Vibrio species were downloaded from GenBank (http://www.ncbi.nlm.nih.gov/Genbank/), and sequences were aligned using the ClustalW program. Species-specific conserved regions of the groEL gene for each of the three target species were identified, and specific primers were designed (Fig. 1). Based on the mismatches between groEL gene sequences of each target species and those of other Vibrio species, primers groVc1: 5′–GATCTTGACTGGCGGTGTTGTG–3′ and groVc2: 5′–GTCACCCACCAGAGAAGAGAGT–3′ for V. cholerae, groVp1: 5′–GTCAGGCTAAGCGCGTAAGCA–3′ and groVp2: 5′–GCATGCCTGCGCTTTCTTTTTG–3′ for V. parahaemolyticus and groVv1: 5′–GTTCGCGCTGGTGAAGGTTCA–3′ and groVv2: 5′–TGGCATACCAGAGTCTTTCTGTG–3′ for V. vulnificus were designed for the specific amplification of 418, 644 and 192 bp fragments, respectively.

Figure 1.

Comparison of primer sequences used for V. cholerae, V. parahaemolyticus and V. vulnificus with those of other Vibrio species. VC: Vibrio cholerae, VP: V. parahaemolyticus, VV: V. vulnificus, VM: V. mimicus, VFu: V. furnissii, VAl: V. alginolyticus, VH: V. harveyi, VO: V. ordalii, VCo: V. coralliilyticus, VFi: V. fischeri, Van: V. angustum, VMet: V. metschinikovii, VAng: V. anguillarum, VSp: V. splendidus and VSh: V. shiloi. Identical nucleotide sequences are indicated by dots.

Multiplex PCR assay and its efficiency test

Polymerase chain reaction conditions were optimized using 50 μl reaction mixture for each tube containing 1 μl of DNA template, 10× PCR buffer containing MgCl2, 0·2 mmol l−1 of dNTP, 0·6 U of Taq polymerase (Takara Bio, Otsu, Shiga, Japan) and variable concentrations of each primer set. The final concentration of each primer set was standardized to obtain proper intensity for each amplicon. PCR conditions were optimized as follows: initial denaturation of 5 min at 94°C followed by 30 cycles each having denaturation for 30 s at 94°C, annealing for 30 s at 69°C and extension for 30 s at 72°C, and final extension step for 5 min at 72°C in a PCR thermal cycler (2720 Thermal cycler; Applied Biosystems, Carlsbad, CA, USA). The PCR products were subjected to 1% agarose gel electrophoresis. The PCR protocol was verified with all strains belonging to the target as well as nontarget species (Table 1). The specificity and sensitivity test of the multiplex PCR was performed according to Hossain et al. (2011). The efficiency of multiplex PCR was also checked using variable DNA concentrations in mixed conditions. DNA representing 106 CFU of V. cholerae was mixed with a DNA mixture of V. parahaemolyticus and V. vulnificus representing 105, 104 and 103 CFU of each bacterium. Similarly, DNA representing 106 CFU of V. parahaemolyticus or V. vulnificus was mixed with DNA representing a similarly lower number of two other species. The multiplex PCR was carried out under optimal conditions.

Evaluation of multiplex PCR in shellfish homogenates and flounder

To test the applicability of this multiplex PCR method for accurate identification of three targeted species from shellfish, tissue homogenates of oyster (Crassostrea gigas), blood clam (Tegillaria granosa), thick shell mussel (Mytilus coruscus) and Manila clam (Tapes philippinarum) were used in a spiking test as described by Kumar et al. (2006). Briefly, 100 μl of one, two or three target species (V. cholerae: 6 × 107 CFU ml−1, V. parahaemolyticus: 9 × 106 CFU ml−1 and V. vulnificus: 10 × 106 CFUml−1) was added to shellfish homogenates (15 ml) and mixed by vortexing. One millilitre of the spiked shellfish homogenates were then transferred to tryptone broth (5% tryptone + 2% NaCl) for enrichment at 37°C for 5 h. Total DNA was extracted from tissues using a DNA extraction kit (NucleoGen Biotech, Siheung, Korea), and multiplex PCR amplification was performed maintaining optimized concentrations of reagents and temperature cycling parameters.

Twenty healthy flounders (Paralichthys olivaceus) were reared in aerated plastic containers (15 l) and divided into five groups. Fish from four groups were intraperitoneally injected with V. cholerae, V. parahaemolyticus and V. vulnificus separately and in combination with the same infectious dose used to infect shellfish tissues, and the last group of fish was kept as a noninoculated control. The internal organs (gill, liver, spleen, intestine and kidney) of inoculated and noninoculated fish were collected at 48 h postinfection. Total DNA was extracted from tissues using a DNA extraction kit, and multiplex PCR amplification was performed.

Detection in artificially inoculated sea water

Sterilized sea water (~300 ml) was incubated at 37°C for 24 h after artificial inoculation with V. cholerae (6 × 104 CFU ml−1), V. parahaemolyticus (9 × 103 CFUml−1) and V. vulnificus (1 × 104 CFU ml−1) separately and in combination. The inoculated sea water and sea water containing infected flounder were collected, and bacterial chromosomal DNA was extracted following the method of Hossain et al. (2011).

Results

Uniplex and duplex PCR amplification

Three sets of primers designed to target three Vibrio species were used in uniplex and duplex PCRs to amplify 418 bp for V. cholerae, 644 bp for V. parahaemolyticus and 192 bp for V. vulnificus (Fig. 2a). Distinguishable amplicons were produced when duplex PCR was performed using DNA template mixtures of any two of the three target species (Fig. 2a).

Figure 2.

Agarose (1%) gel electrophoresis of PCR products of V. cholerae (Vc), V. parahaemolyticus (Vp) and V. vulnificus (Vv) amplified during standardization of multiplex PCR (a) and specificity testing using Vibrio and non-Vibrio species (b). (a) Lanes 1 and 9, 100-bp DNA ladder; lane 2, Vc; lane 3, Vp; lane 4, Vv; lane 5, Vc and Vp; lane 6, Vc and Vv; lane 7, Vp and Vv; lane 8, Vc, Vp and Vv. (b) Lanes 9 and 26: 100-bp DNA ladder; lanes 1–8, 10–25 and 27–34: V. aestuarianus, V. alginolyticus, V. anguillarum, V. campbellii, V. cholerae, V. cincinnatiensis, V. damsella, V. diazotrophicus, V. fluvialis, V. furnissii, V. harveyi, V. hollisae, V. logei, V. mediterranei, V. metschinikovii, V. mimicus, V. natriegens, V. navarrensis, V. nereis, V. ordalii, V. parahaemolyticus, V. proteolyticus, V. tubiashii, V. vulnificus, Aeromonas hydrophila, Edwardsiell tarda, Enterobacter cloacae, Escherichia coli, Klebsiella oxytoca, K. pneumoniae, Salmonella typhi and Shigella flexneri, respectively.

Multiplex PCR

PCR amplification of mixed genomic DNA from three species with each set of primers produced a single DNA fragment of the expected molecular weight (Fig. 2a). This suggests that the primers specific to the individual pathogens used in this study and would not generate false positives in the PCR. Multiplex PCR enabled simultaneous amplification of all three targets with comparable band intensities using the PCR cycling parameters.

Specificity of V. cholerae, V. parahaemolyticus and V. vulnificus primers

The newly developed multiplex PCR produced amplicons of the expected sizes; that is, 418 bp for V. cholerae, 644 bp for V. parahaemolyticus and 192 bp for V. vulnificus (Fig. 2b). Furthermore, the PCR products were sufficiently different in size to be distinguishable by agarose gel electrophoresis. Specific amplicons were also produced from all strains of the three species (Table 1). No amplified products were obtained with other Vibrio and non-Vibrio enteric species used in this study (Fig. 2b). The results demonstrated that the primers groVc1–groVc2, groVp1–groVp2 and groVv1–groVv2 are sufficiently specific to V. cholerae, V. parahaemolyticus and V. vulnificus, respectively.

Evaluation of the detection limit and efficiency of multiplex PCR

The multiplex PCR worked efficiently with DNA templates from each species individually as well as in combination with three species (Fig. 3). There was a qualitative decrease in amplicon intensity with decreasing DNA concentration. The detection limit for genomic DNA in the uniplex PCR was 100 pg for all three species (Fig. 3a). The detection limit of mixed genomic DNA in multiplex PCR was also 100 pg for all three Vibrio species (Fig. 3a). When cell lysate was used to determine the detection limit, the sensitivities of groEL primers for V. cholerae, V. parahaemolyticus and V. vulnificus were 140, 130 and 50 CFU, respectively (data not presented). When variable DNA concentrations were used, specific amplicons with conspicuous band intensities were produced even with 100-fold differences in cell numbers among different species, for example, 105 CFU V. cholerae along with 103 CFU of each of V. parahaemolyticus and V. vulnificus per tube (Fig. 3b). However, a 1000-fold difference in cell density resulted in the generation of bands representing only the species with the greater cell density.

Figure 3.

Sensitivity for the detection of V. cholerae, V. parahaemolyticus and V. vulnificus (a). 10-fold serial dilution (1 μg, 100 ng, 10 ng, 1 ng, 100 pg, 10 pg, 1 pg and 0·1 pg) of purified chromosomal DNA of cultured cells; Lanes 9, 18 and 27: 100-bp DNA ladder; lanes 1–8, V. cholerae; lanes 10–17, V. vulnificus; lanes 19–26, V. parahaemolyticus and lanes 28–35, mixed chromosomal DNA of the three species. Efficacy of multiplex PCR for detection of targets using variable DNA template concentrations (b). Vc, Vp and Vv represent V. cholerae, V. parahaemolyticus and V. vulnificus, respectively. Lane 1, 100-bp DNA ladder; Lanes 2, 3 and 4, 105 CFU of V. cholerae mixed with 10-, 100- and 1000-fold fewer CFUs of the other two species, respectively. A similar strategy was followed for lanes 5–7 and lanes 8–10 with 105 CFU of V. parahaemolyticus and V. vulnificus, respectively; lane 11, positive control.

Detection of V. cholerae, V. parahaemolyticus and V. vulnificus from artificially inoculated shellfish homogenates, flounder and sea water

Target Vibrio species were detected by multiplex PCR from all artificially inoculated shellfish homogenates. Manila clam homogenates were inoculated with one, two or three target species, and bands were visualized by gel electrophoresis after multiplex PCR using extracted DNA as the template (Fig. 4a). Similar results were observed for all shellfish species investigated (data not presented). DNA extracted from all flounder organs tested positive for the three target Vibrio species by multiplex PCR (Fig. 4b). This newly developed multiplex PCR detected V. cholerae, V. parahaemolyticus and V. vulnificus from all inoculated sea water samples and from sea water containing infected fish (Fig. 4b).

Figure 4.

Detection of V. cholerae, V. parahaemolyticus and V. vulnificus in inoculated samples by multiplex PCR. Vc, Vp and Vv represent V. cholerae, V. parahaemolyticus and V. vulnificus, respectively. (a) DNA extracted from Manila clam homogenates (Tapes philippinarum) infected with Vc (lane 2), Vp (lane 3), Vv (lane 4), Vc + Vp (lane 5), Vc + Vv (lane 6), Vp + Vv (lane 7) and Vc + Vp + Vv (lane 8) after 5-h enrichment; lane 1, 100-bp DNA ladder; lane 9, positive control. (b) Sea water inoculated with Vc (lane 2), Vp (lane 3), Vv (lane 4) and all three species (lane 5); sea water containing infected fish inoculated with: lane 6, Vc; lane 7, Vp; lane 8, Vv and lane 9, all three species; lanes 1 and 10, 100-bp DNA marker; Detection in organs of flounders inoculated with Vc, Vp and Vv together (lanes 11, 12, 13 and 14; gill, liver, kidney and intestine, respectively).

Discussion

Food-borne pathogens pose a significant threat to public health, leading to a substantial economic burden in many countries (Chen et al. 2012). For this reason, the availability of rapid, sensitive and specific diagnostic methods for the detection of disease-causing pathogens is important. Molecular diagnosis protocols have provided effective methods for the diagnosis of bacterial agents because they permit specific and sensitive detection (Gonzalez et al. 2004). Several PCR-based detection methods for the detection of V. cholerae, V. parahaemolyticus and V. vulnificus have been developed. The use of virulence genes as identification markers may be of significance because their existence may be linked to pathogenesis. However, when applied to environmental samples, there is a potential risk of misidentification because such genes might transfer among bacteria. Neogi et al. (2010) reported that among closely related Vibrio species, horizontal transfer of toxigenic genes can equip the nontoxigenic strains with epidemic potential. Therefore, it is important to conduct surveillance on the total population (both toxigenic and nontoxigenic) of these three target species. A suitable phylogenetic marker is necessary for the detection of all strains of a particular species. The groEL gene, which has been established as a good marker for species-specific detection of various bacteria including V. anguillarum and V. parahaemolyticus, was used for the simultaneous detection of V. cholerae, V. parahaemolyticus and V. vulnificus in this study. It is also important to detect pathogenic strain of particular species. Many virulence genes have been used as a target marker for the specific detection of pathogenic strains. But, during disease outbreak or screening of samples, the first choice is to detect particular pathogen at species level instead of strain level. So, our developed multiplex PCR method is suitable in this context, and if necessary, the pathogenic strains can be confirmed using virulence marker.

Pinto et al. (2005) developed a collagenase-targeted multiplex PCR with high specificity to detect Vibrio species, but they did not include V. vulnificus, an organism of public health concern in their assay. The PCR detection assay developed by Nhung et al. (2007) was specific only to pathogenic Vibrios and failed to identify nonpathogenic Vibrios. Teh et al. (2010) also developed a multiplex PCR assay using gyrB and pntA genes to detect pathogenic and nonpathogenic Vibrio species, but they did not confirm its efficacy in a mixed population. Nhung et al. (2007) proposed a dnaJ gene–based multiplex PCR but did not verify their method with a mixed population; the differences among amplicon sizes were also problematic. Tarr et al. (2007) targeted sodB, flaE and hsp genes to detect V. cholerae, V. parahaemolyticus and V. vulnificus, but they did not verify with mixed populations. According to Neogi et al. (2010), the method developed by Bauer and Rørvik (2007) failed to differentiate V. parahaemolyticus from V. alginolyticus, and the method developed by Grim et al. (2009) could not differentiate V. cholerae from V. mimicus. Neogi et al. (2010) used toxR and vvhA genes for the detection of V. cholerae, V. parahaemolyticus and V. vulnificus by multiplex PCR and successfully detected these three species. They performed both specificity and sensitivity tests using a mixed population but used only pond water to confirm the efficiency of the developed method, instead of sea water or shellfish.

The multiplex PCR developed in this study successfully detected V. cholerae, V. parahaemolyticus and V. vulnificus without false-positive results from nontarget species. The species-specific primer sets also produced amplicons of various sizes that were easily distinguishable by electrophoresis. The detection levels of both uniplex and multiplex PCR assays from mixed purified genomic DNA of the three target species were similar. Efficiency was also good in a mixed population similar to that described by Neogi et al. (2010). We used sea water, shellfish and flounder as samples for artificial infection to confirm the efficiency and accuracy of three primer sets. Wong et al. (2012) mentioned that detection of bacteria in food by PCR is often hindered by the presence of inhibitors. Enrichment procedures can minimize the interference of the PCR inhibitors and increase the concentration of the target micro-organisms. Enrichment for 5–12 h was applied during the detection of Vibrio species from fish, fishery products, shellfish and water (Hossain et al. 2011; Jeyasekaran et al. 2011; Malayil et al. 2011; Wong et al. 2012). In this study, we extracted DNA from inoculated shellfish homogenates and sea water before and after enrichment for 5 h. Detection was possible in sea water samples before and after enrichment, but weak amplicons were observed in shellfish samples before enrichment. Enrichment resulted in detection with a strong signal, indicating that the multiplex PCR assay was able to detect V. cholerae, V. parahaemolyticus and V. vulnificus from the artificially inoculated samples.

In conclusion, the multiplex PCR assay developed in this study is highly sensitive and specific to the simultaneous detection of V. cholerae, V. parahaemolyticus and V. vulnificus. Further evaluation of this newly developed method using environmental and clinical samples is necessary to verify its detection efficacy, which will ultimately assist the epidemiologists, physicians and ecologists to investigate these three important Vibrio species.

Acknowledgements

This research was supported by a project grant (YSG-RE0701) from Yeongnam Sea Grant, Korea. MTH wishes to thank National Institute for International Education (NIIED), Korea for financial support.

Ancillary