Molecular epidemiology and evolutionary histories of human coronavirus OC43 and HKU1 among patients with upper respiratory tract infections in Kuala Lumpur, Malaysia

A total of 2,060 consenting outpatient adults presented with symptoms of acute URTI were recruited at the Primary Care Clinics of University Malaya Medical Centre in Kuala Lumpur, Malaysia between March 2012 and February 2013. Prior to collection of nasopharyngeal swabs, demographic data such as age, gender and ethnicity were obtained. In addition, the severities of symptoms (sneezing, nasal discharge, nasal congestion, headache, sore throat, voice hoarseness, muscle ache and cough) were graded based on previously reported criteria [18‐21]. The scoring scheme used had been validated earlier on the adult populations with common cold [19]. The nasopharyngeal swabs were transferred to the laboratory in universal transport media and stored in −80 °C.

Total nucleic acids were extracted from nasopharyngeal swabs using the magnetic beads-based protocols implemented in the NucliSENS easyMAG automated nucleic acid extraction system (BioMérieux, USA) [22, 23]. Specimens were screened for the presence of respiratory viruses using the xTAG Respiratory Virus Panel FAST multiplex RT-PCR assay (Luminex Molecular Diagnostics, USA) which can detect HCoV-OC43, HCoV-HKU1 and other respiratory viruses and subtypes [24].

RNA from nasopharyngeal swabs positive for HCoV-OC43 and HCoV-HKU1 was reverse transcribed into cDNA using SuperScript III kit (Invitrogen, USA) with random hexamers (Applied Biosystems, USA). The partial S gene (S1 domain) [HCoV-OC43; 848 bp (24,030-24,865) and HCoV-HKU1; 897 bp (23,300-24,196)], complete N gene [HCoV-OC43; 1,482 bp (28,997-30, 478) and HCoV-HKU1; 1,458 bp (28,241-29,688)] and partial 1a (nsp3) gene [HCoV-OC43; 1,161 bp (6,168- 7,328) and HCoV-HKU1; 1,115 bp (6,472-7,586)] were amplified either by single or nested PCR, using 10 μM of the newly designed or previously described primers listed in Table 1. The PCR mixture (25 μl) contained cDNA, PCR buffer (10 mM Tris–HCl, 50 mM KCl, 3 mM MgCl, 0.01 % gelatin), 100 μM (each) deoxynucleoside triphosphates, Hi-Spec Additive and 4u/μl BIO-X-ACT Short DNA polymerase (BioLine, USA). The cycling conditions were as follows: initial denaturation at 95 °C for 5 min followed by 40 cycles of 94 °C for 1 min, 54.5 °C for 1 min, 72 °C for 1 min and a final extension at 72 °C for 10 min, performed in a C1000 Touch automated thermal cycler (Bio-Rad, USA). Nested/semi-nested PCR was conducted for each genetic region if necessary, under the same cycling conditions at 30 cycles. Purified PCR products were sequenced using the ABI PRISM 3730XL DNA Analyzer (Applied Biosystems, USA). The nucleotide sequences were codon-aligned with previously described complete and partial HCoV-OC43 and HCoV-HKU1 reference sequences retrieved from GenBank [11, 25‐32].

Maximum clade credibility (MCC) trees for the partial S (S1 domain), complete N and partial 1a (nsp3) genes were reconstructed in BEAST (version 1.7) [27, 33, 34]. MCC trees were generated using a relaxed molecular clock, assuming uncorrelated lognormal distribution under the general time-reversible nucleotide substitution model with a proportion of invariant sites (GTR + I) and a constant coalescent tree model. The Markov chain Monte Carlo (MCMC) run was set at 3 × 10⁶ steps long sampled every 10,000 state. The trees were annotated using Tree Annotator program included in the BEAST package, after a 10 % burn-in, and visualized in FigureTree (http://​tree.​bio.​ed.​ac.​uk/​software/​Figuretree/​). Neighbor joining (NJ) trees for the partial S (S1 domain), complete N and partial 1a (nsp3) genes were also reconstructed, using Kimura 2-parameter model in MEGA 5.1 [35]. The reliability of the branching order was evaluated by bootstrap analysis of 1000 replicates. In addition, to explore the genetic relatedness between HCoV-OC43 and HCoV-HKU1 genotypes, the pairwise genetic distances among sequences of the S gene were estimated. Inter- and intra-genotype nucleotide distances were estimated by the bootstrap analysis with 1000 replicates using MEGA 5.1. Such analysis has not been done for the N and the 1a genes because those regions were highly conserved across genotypes [10, 11, 32]. To test for the presence of recombination in HCoV-OC43, the S gene was subjected to pairwise distance-based bootscanning analysis using SimPlot version 3.5 [10, 36]. Established reference genomes for HCoV-OC43 genotype A (ATCC VR-759), B (87309 Belgium 2003), and C (HK04-01) were used as putative parental lineages, with a sliding window and step size of 160 bp and 20 bp, respectively. In addition, MaxChi recombination test [37] was performed in the Recombination Detection Program (RDP) version 4.0 [38]. In RDP the highest acceptable p value (the probability that sequences could share high identities in potentially recombinant regions by chance alone) was set at 0.05, with the standard multiple comparisons corrected using the sequential Bonferroni method with 1,000 permutations [39].

The origin and divergence time (in calendar year) of HCoV-OC43 and HCoV-HKU1 genotypes were estimated using the MCMC approach as implemented in BEAST. Analyses were performed under the relaxed molecular clock with GTR + I nucleotide substitution models and constant-size and exponential demographic models. The MCMC analysis was computed at 3 × 10⁶ states sampled every 10,000 steps. The mean divergence time and the 95 % highest posterior density (HPD) regions were estimated, with the best-fitting models were selected by Bayes factor inference using marginal likelihood analysis implemented in Tracer (version 1.5) [33]. The evolutionary rate for S gene of betacoronaviruses (6.1 × 10⁻⁴ substitutions/site/year) reported previously was used for analysis [36].

The association of HCoV-OC43 and HCoV-HKU1 infections with specific acute URTI symptoms and its severity (none, moderate and severe) as well as demographic data were evaluated using the Fisher’s exact test/Chi-square test carried out in the statistical package for the social sciences (SPSS, version 16; IBM Corp).

During the 12-month study period (March 2012 to February 2013), all nasopharyngeal swab specimens from 2,060 patients collected from Kuala Lumpur, Malaysia were screened for the presence of HCoV-OC43 and HCoV-HKU1 using multiplex RT-PCR method, in which a total of 48 (2.4 %) subjects were found positive for betacoronavirus. HCoV-OC43 and HCoV-HKU1 was detected in 26/2060 (1.3 %) and 22/2060 (1.1 %) patients, respectively, while no HCoV-OC43/HCoV-HKU1 co-infection was observed. Age, gender and ethnicity of the patients were summarized in Table 2. The median age of subjects infected with HCoV-OC43 and HCoV-HKU1 was 53.0 and 48.5, respectively. Both HCoV-OC43 and HCoV-HKU1 were co-circulating throughout the year, although lower numbers of HCoV-OC43 were detected between October 2012 and January 2013 while no HCoV-HKU1 was detected during these months (Fig. 1).

The partial S (S1 domain), complete N and partial 1a (nsp3) genes of 23 HCoV-OC43 isolates were successfully sequenced, while another three xTAG-positive HCoV-OC43 isolates could not be amplified, probably due to low viral copy number in these specimens. Based on the phylogenetic analysis of the S gene, one subject (1/23, 4.3 %) was grouped with HCoV-OC43 genotype B reference sequences while another subject (1/23, 4.3 %) was grouped with HCoV-OC43 genotype D sequences. The remaining 21 isolates formed two phylogenetically discrete clades that were distinct from other previously established genotypes A, B, C, D (genotype D is a recombinant lineage that is not readily distinguished from genotype C in the S and N phylogenetic trees) and E [11, 32] (Fig. 2 and Additional file 1: Figure S1). Of the 21 isolates, ten isolates have formed a cluster with other recently reported isolates from Japan, Thailand and China [31, 32] supported by the posterior probability value of 1.0 and bootstrap value of 36 % at the internal tree node of the MCC and NJ trees, respectively with intra-group pairwise genetic distance of 0.003 ± 0.001. These isolates were provisionally designated as novel lineage 1. Spatial structure was observed within novel lineage 1, with an isolate from China sampled in year 2008 located at the base of the phylogeny. Moreover, another eleven HCoV-OC43 isolates have formed a second distinct cluster supported by significant posterior probability and bootstrap values at the internal tree node (1.0 and 98 %, respectively) and intra-group pairwise genetic distance of 0.004 ± 0.001. The cluster contained Malaysian and Chinese isolates [32] only, and was denoted as novel lineage 2. Based on the phylogenetic inference of the conserved N gene, only one subject was grouped with the genotype B reference in concordance with the S gene (Additional file 2: Figure S2). Unlike the phylogenetic inference of the S gene, the remaining 22 isolates were seen intermingled with each other forming a single cluster together with isolates indicated as novel lineages 1 and 2 in the S gene, in addition to one genotype D strain. It is however important to note that the tree resolution was poor, due primarily to the lack of the N gene reference sequences in the public database. On the other hand, phylogenetic analysis of the 1a (nsp3) gene (Additional file 3: Figure S3) revealed that all except genotype A could not be differentiated clearly within this region, due mainly to the low genetic diversity between genotypes. The limited number of 1a reference sequences available in the public database could have also resulted in a poor 1a tree topology. In addition, phylogenetic trees of previously described complete and partial S gene sequences as well as partial 1a (nsp3) and complete RdRp gene sequences were reconstructed to further confirm the reliability of the partial S1 and nsp3 for identification of HCoV-OC43 genotypes (Additional file 4: Figure S4 and Additional file 5: Figure S5).

To assess the diversity between HCoV-OC43 genotypes, inter-genotype pairwise genetic distance was estimated for the S gene, listed in Table 3. Using the oldest genotype as reference i.e. genotype A, genetic variation between genotype A and genotypes B to E was 2.2–2.7 %. Genetic distance between novel lineages 1 and 2 compared to genotype A was 3.2 % and 3.1 %, respectively, higher than that of other established genotypes. Taken together, the distinct inter-genotype genetic variations of the two novel lineages 1 and 2 against other previously established genotypes corroborated with the MCC inference (Fig. 2) in which both lineages formed distinct phylogenetic topologies.

On the other hand, phylogenetic analysis of 22 HCoV-HKU1 S and N genes indicated the predominance of HCoV-HKU1 genotype B (72.7 %, 16/22), followed by HCoV-HKU1 genotype A (27.3 %, 6/22) (Fig. 3, Additional file 6: Figure S6 and Additional file 7: Figure S7). Interestingly, the S and N genes of HCoV-HKU1 were equally informative for genotype assignment, while genotypes A, B and C were less distinctive based on the 1a gene phylogenetic analysis due to the high genetic conservation within this region (Additional file 8: Figure S8). Inter-genotype genetic diversity among HCoV-HKU1 genotypes showed that genotype A was more genetically diverse than genotypes B and C based on the genetic data of the S gene (Table 3). The difference in genetic distance between genotype A and genotypes B and C was 15.2–15.7 %, while the difference in genetic distance between genotypes B and C was 1.3 %.

Evidence of possible recombination was observed in the S gene of novel lineage 1, involving genotypes B and C (Fig. 4). All isolates within novel lineage 1 showed similar recombination structures (representative isolates from Malaysia (12MYKL0208), Japan (Niigata.JPN/11-764), Thailand (CU-H967_2009) and China (892A/08) were shown). Similarly, sign of possible recombination was noticed within novel lineage 2 (Fig. 4). All Malaysian and Chinese isolates showed similar recombination structures in the S gene involving genotypes A and B (12MYKL0002, 12MYKL0760 and 12689/12 representative sequences were shown). Moreover, using the aforementioned putative parental and representative strains, MaxChi analysis of the novel lineages 1 and 2 isolates supported the hypothesis of recombination in the S gene (p < 0.05) (Additional file 9: Figure S9). Taken together, the emergence of novel lineage 1 and novel lineage 2 in these Asian countries was likely to be driven by natural recombination events.

The divergence times of HCoV-OC43 and HCoV-HKU1 were estimated using the coalescent-based Bayesian relaxed molecular clock under the constant and exponential tree models (Fig. 2 and Fig. 3; Table 4). The newly estimated mean evolutionary rate for the S gene of HCoV-OC43 was 7.2 (5.0 – 9.3) × 10⁻⁴ substitutions/site/year. On the other hand, the evolutionary rate for the S gene of HCoV-HKU1 was newly estimated at 6.2 (4.2–7.8) × 10⁻⁴ substitutions/site/year. These estimates were comparable to previous findings of 6.1–6.7 × 10⁻⁴ substitutions/site/year for the S gene reported elsewhere [11].

Based on these evolutionary estimates of the S gene, the common ancestor of HCoV-OC43 was dated back to the 1950s. Divergence time of genotype A was dated back to early 1960s, followed by genotype B around 1990s. Interestingly, genotypes C, D, E, and novel lineages 1 and 2 were all traced back to the 2000s (Fig. 2). Moreover, the common ancestor of HCoV-HKU1 was traced back to early 1950s, as estimated from the S gene. Subsequently, HCoV-HKU1 continued to diverge further into distinctive genotypes (A-C). Genotype A was dated to the late 1990 and genotypes B and C were both traced back to early 2000s (Fig. 3). Bayes factor analysis showed insignificant differences (Bayes factor <3.0) between the constant and exponential coalescent models of demographic analysis. Divergence times generated using the exponential tree model were slightly (but not significantly) different from those estimated using the constant coalescent model (Table 4). Of note, HCoV-OC43 and HCoV-HKU1 genotype assignments were less distinctive within the N and 1a genes (as compared to the S gene); these regions were therefore deemed unsuitable for divergence time estimations in this study.

The study was approved by the University of Malaya Medical Ethics Committee (MEC890.1). Standard, multilingual consent forms allowed by the Medical Ethics Committee were used. Written consents were obtained from all study participants.

Not applicable.

HCoV-OC43 and HCoV-HKU1 nucleotide sequences generated in the study are available in GenBank under the accession numbers KR055512-KR055644.