Enterovirus D68 Infections Associated with Severe Respiratory Illness in Elderly Patients and Emergence of a Novel Clade in Hong Kong

Despite the recent emergence of enterovirus D68 (EV-D68), its clinical impact on adult population is less well defined. To better define the epidemiology of EV-D68, 6,800 nasopharyngeal aspirates (NPAs) from 2010–2014 were subject to EV-D68 detection by RT-PCR and sequencing of 5′UTR and partial VP1. EV-D68 was detected in 30 (0.44%) NPAs from 22 children and 8 adults/elderlies. Sixteen patients (including five elderly) (53%) had pneumonia and 13 (43%) patients were complicated by small airway disease exacerbation. Phylogenetic analysis of VP1, 2C and 3D regions showed four distinct lineages of EV-D68, clade A1, A2, B1 and B3, with adults/elderlies exclusively infected by clade A2. The potentially new clade, B3, has emerged in 2014, while strains closely related to recently emerged B1 strains in the United States were also detected as early as 2011 in Hong Kong. The four lineages possessed distinct aa sequence patterns in BC and DE loops. Amino acid residues 97 and 140, within BC and DE-surface loops of VP1 respectively, were under potential positive selection. EV-D68 infections in Hong Kong usually peak in spring/summer, though with a delayed autumn/winter peak in 2011. This report suggests that EV-D68 may cause severe respiratory illness in adults/elderlies with underlying co-morbidities.

The virus was first isolated from throat swab samples of children with bronchiolitis and pneumonia in California in 1962 and referred as "Fermon virus" 24 . Rhinovirus 87 (RV87) and EV-D68 are now considered as the same virus with both RV and EV features (genetically related to other EVs but more temperature-and acid-labile) 25,26 . EV-D68 mainly affects infants, children and adolescents, while infections in adults were less commonly reported. Clinical manifestations of EV-D68 infections mainly resemble influenza-like illness, but life-threatening infections were also reported 16,[27][28][29][30][31] . Neurological involvement, ranging from encephalomyelitis to acute flaccid paralysis, was rare, although a recent report highlighted possible association with acute flaccid myelitis 16,32,33 . Based on sequences of the viral capsid gene VP1, EV-D68 strains are often categorized into 3 distinct genetic groups, clade A to C, in addition to the original Fermon lineage 29,33 .
Subsequent to its first isolation, EV-D68 infections have only been rarely reported in the following decades. In US, only 26 cases of EV-D68 infections were detected by passive EV surveillance from 1970 to 2005 34 . However, increasing reports of EV-D68 infections have recently been noted in various countries from Africa, America, Asia and Europe [27][28][29]31,[35][36][37] . Since August 2014, more than 1000 cases of EV-D68 have been reported in the United States (US), with at least 14 potentially fatal cases 31 . In Hong Kong, unlike EV-A71 infection which is a notifiable disease, the prevalence of EV-D68 was largely unknown. However, a fatal case of EV-D68 was recently reported in a 10-year-old boy who presented with respiratory symptoms and complicated by encephalitis (http://www. chp.gov.hk/en/view_content/36405.html). To better define the disease impact and clinical spectrum of EV-D68, we examined the clinical and molecular epidemiology of EV-D68 among hospitalized patients with suspected respiratory virus infections in Hong Kong. The partial VP1, 2C and 3D genes of EV-D68 strains were sequenced to study the genetic diversity and evolutionary dynamics.

Clinical characteristics of patients with EV-D68 infections.
A bimodal age distribution was observed among the 30 patients (male: female = 19:11) with EV-D68 infections (Table 1), which included 22 children (< 18 years), one 37-year-old adult and seven elderlies (> 60 years). Of the seven elderlies, two resided in elderly homes (patients 19 and 23). A 15-year-old adolescent (patient 6) and a 37-year-old adult (patient 13) were also institutionalized. Close contacts of three patients were noted to have recent respiratory illnesses, including siblings of patients 3 and 4, and domestic helper of patient 28. One elderly (patient 20) had recent travel history to mainland China. No epidemiological linkage was identified among the 30 patients.
Twenty-two patients had underlying diseases, such as respiratory diseases and allergies. Most patients presented with acute respiratory illness. Sixteen patients (53%), including five elderlies, presented with pneumonia which represented the most common diagnosis. One elderly (patient 23) with pneumonia was complicated by congestive heart failure. Various radiological abnormalities were observed, including perihilar haziness, focal haziness, infiltrates and consolidation. Three patients (10%) presented with acute bronchiolitis or bronchitis. Exacerbations of small airway diseases were common, including asthmatic attacks (n = 13), febrile wheeze (n = 2) and exacerbation of chronic obstructive pulmonary disease (n = 2). Five patients (17%) presented with upper respiratory tract infection (URTI), including a 63-year-old elderly (patient 21) who was complicated by acute coronary syndrome and shock. Another 2-year-old girl (patient 2) with acute bronchiolitis was complicated by intestinal obstruction. All patients recovered. Except for a 6-year-old girl (patient 29) with influenza C virus co-infection, no respiratory viruses or bacteria were identified from other patients.

Molecular epidemiology of EV-D68 circulating in Hong
Kong. Phylogenetic analysis of partial VP1, 2C and 3D regions showed that the 30 EV-D68 strains fell into two major clades, clades A and B, corresponding to the two of the three EV-D68 clades described previously ( Fig. 2 and Supplementary Fig. 1) 33,35 . Previous studies have also revealed further subgroups, such as clade B1 and B2 among strains recently emerged in US 30,33,38 . Upon VP1 analysis, 15 strains belonged to clade B, some of which being closely related to recent B1 strains from US. Five strains detected in 2014 form a potentially new clade, B3, together with strains from Taiwan and mainland China. The other 15 strains belonged to clade A, which can be further divided into two subclades, A1 (n = 2) and A2 (n = 13). Analysis of 2C and 3D sequences showed similar clustering ( Supplementary Fig. 1). The four circulating lineages, clade A1, A2, B1 and B3, shared ≥ 91.8% aa identities to each other in partial VP1 and displayed distinct amino acid polymorphisms in VP1, 2C and 3D (Supplementary Table 2 and Supplementary Fig. 2).
The four clades appeared to dominate in different years. While both clade A1 and B1 were detected in 2010, clade A1 was not detected in subsequent years. Both clade A2 and B1 were co-circulating in 2011 and 2012. However, clade A2 and B was the only circulating lineage in 2013 and 2014 respectively. Moreover, A2 strains circulating in 2013 were clustered together, as with B3 strains in 2014 (Fig. 2), suggesting epidemics from specific lineages. The B3 strains were only distantly related to recent B1 strains in US, suggesting the recent emergence of this novel subclade. Interestingly, adults/elderlies infected by EV-D68 were exclusively associated with clade A2, including five closely related strains from elderlies in 2013. Closely related B3 strains were also noted in six children in 2014.
Scientific RepoRts | 6:25147 | DOI: 10.1038/srep25147 Detection of positive selection in VP1 genes. To assess adaptive evolution of EV-D68, d N /d S ratios (ω ) in VP1 gene across the 30 strains were calculated on codon-by-codon basis. The overall ω was 0.077, with most residues having ω < 1, indicating purifying selection (Fig. 3). Nevertheless, residues 97 and 140 (within BC and DE-surface loops respectively) had ω > 1 but without statistical significance, indicating possible functional constraints during evolution. The dominant aa compositions at BC and DE-surface loops are shown in Fig. 4.

Discussion
Our study represents the first to demonstrate EV-D68 as a possible cause of severe respiratory illness in adults or elderlies with underlying co-morbidities. In contrast to paediatric population, the role of EV-D68 in adult infections was less clear with only scarce reports. In the Netherlands, EV-D68-related respiratory illness has been described in adults but without clinical details 27,29 . Two reports from China have identified 13 and two adults respectively with URTIs due to EV-D68 28,39 . A recent report from Denmark identified EV-D68 infections in two adults with URTIs and one adult co-infected by RV with fever, cough and breathing difficulties 30 . In this study, only one elderly presented with URTI alone, while another elderly with URTI was complicated by acute coronary syndrome. The other five elderlies presented with pneumonia, with two having respiratory or cardiac complications. Another 37-year-old adult also presented with pneumonia. Our findings suggested that EV-D68 can cause severe infections in adults or elderlies with underlying diseases. The EV-D68 infections in adults or elderlies may be underdiagnosed. The conserved 5′ UTR is a common target for initial RT-PCR detection of EVs, while specific diagnosis of EV-D68 usually relies on specific RT-PCR assays or VP1 sequencing. In particular, a number of EV-D68 specific diagnostic tools based on real-time PCR assays have been developed recently [40][41][42] . Although commercially available multiplex PCR assays may include EV-D68, they may lack sensitivity and specificity, and may not differentiate EV-D68 from RVs 42 . Moreover, EV-D68 is seldom included in diagnostic panels for respiratory viruses, while most surveillance programs for EVs focus on stool samples. Inclusion of EV-D68 in routine respiratory virus panels may also help better assess its clinical and public health significance.
The present study also represents the first to describe the epidemiology of EV-D68 infections in Hong Kong. The steady annual incidence observed suggested that EV-D68 has been circulating for at least several years. Unlike EV-A71 with seasonal peaks during summer/fall, as many as 50% of EV-D68 infections may fall outside usual EV peak seasons 34,43,44 . EV-D68 appears to peak during spring/summer in our population, while a delayed autumn/winter peak was observed in 2011. Therefore, factors other than temperature may also play a role in the seasonality of EV-D68 infection. Lower respiratory illness was common among both elderlies and children, with pneumonia being the most common diagnosis. Many patients were also complicated by exacerbation of  small airway disease, in line with the disproportionate presence of asthma among cases in US 45 . Interestingly, a 2-year-old girl (patient 1) had hand-foot-mouth disease which has not been previously reported for EV-D68, although the possibility of co-infection by other EVs cannot be excluded. However, given the limited number of positive samples in this study, further epidemiology studies with inclusion of more cases are required to assess the seasonality, epidemiology and disease impact of EV-D68 in our population. Our results revealed four lineages of EV-D68, A1, A2, B1 and B3, having circulated in our population and the potential emergence of a new subclade B3 in 2014. Moreover, adults/elderlies were infected exclusively by clade A2, suggesting that this lineage may be emerging in our elderly population. The recent emergence of EV-D68  in different countries may be due to viral evolution of strains/lineages with different antigenicity, with major bifurcation of currently circulating EV-D68 strains dated back to around 1945 33,35,39 . The aa sequence patterns in BC and DE-surface loops, which determine the antigenic epitopes, may correlate with lineage classification 35 . The four lineages in this study also possessed distinct aa sequence patterns in BC and DE loops ( Supplementary  Fig. 2a). Moreover, most B strains circulating in 2014 formed a new separate subclade, B3, only distantly related to recent B1 sand B2 trains from US. Other potential B3 strains were also detected in Beijing and Taiwan in 2014 (Fig. 2). On the other hand, some of our B strains detected as early as 2011 were closely related to B1 strains from US. Our findings suggested that the recent epidemic in US may have originated from strains circulating in Asia several years ago, while a new lineage, B3, has only recently emerged in Hong Kong and neighbouring regions. More sequence data are needed to understand the genetic changes that may have driven the emergence of new lineages of EV-D68.
Different EVs are known to utilize difference receptors for cellular entry, which may explain their diverse clinical manifestations and tissue tropisms. In contrast to most EVs with enteric tropism, EV-D68 is seldom detected from stool of infected patients. Although EV-D68 possessed some properties of RVs, it does not utilize the receptors of RVs, namely intercellular adhesion molecule 1 (ICAM-1) or low-density lipoprotein receptor (LDL-R) 46,47 . Instead, it shares the same receptor, decay accelerating factor (DAF), with EV-D70 and several echovirus serotypes 25,48,49 . Interestingly, both EV-D68 and EV-D70 belong to the species EV-D, although EV-D70 is mainly associated with acute hemorrhagic conjunctivitis 19 . In addition, EV-D68 may use sialic acids as receptor, with preferential binding α 2-6-linked sialic acids (α 2-6 SAs) than to α 2-3 SAs, which may suggest higher affinity for upper respiratory epithelium 47,50 . Since our results suggest that lower respiratory illness can be common in EV-D68 infections, further studies are required to investigate if some strains or lineages may possess higher affinity for the lower airway. Patients and microbiological methods. All NPAs were collected from hospitalized patients in two regional hospitals in Hong Kong during a five-year period (January 2010 to December 2014), and were tested negative for influenza A and B viruses, parainfluenza viruses types 1, 2 and 3, respiratory syncytial virus, adenovirus, human metapneumovirus, human coronaviruses and human bocavirus 2 .

Methods
RT-PCR for detection of EV-D68 from NPAs. RNA extraction and RT-PCR for EVs were performed using previously described protocols with modifications and primers targeted to the 5′ untranslated region (5′ UTR) as shown in Supplementary Table 1 2,51 . Both strands of PCR products were sequenced twice with an ABI 3130xl DNA Analyzer (Applied Biosystems), using the PCR primers. Nucleotide sequences were compared to those of known EVs with sequences available in GenBank. Positive samples with sequences belonging to EVs were subject to EV-D68 detection by RT-PCR and sequencing of VP1 gene using EV-D68 specific primers. Samples containing EV-D68 upon VP1 gene sequencing were further subject to 2C and 3D gene sequencing.
RT-PCR and sequencing of VP1, 2C and 3D gene regions. The partial VP1, 2C and 3D gene regions of EV-D68 strains detected from NPAs were amplified and sequenced using primers shown in Supplementary Table 1 and previously described protocols with modifications 2,52 . Phylogenetic trees of each region were constructed using maximum likelihood (ML) method in MEGA6, with bootstrap analysis of 1000 replicates.
Selective pressure analysis. The number of synonymous substitutions per synonymous site, d S , and non-synonymous substitutions per non-synonymous site, d N , in VP1 were calculated using Nei-Gojobori method (Jukes-Cantor) in MEGA5. Sites under positive selection were inferred using single-likelihood ancestor counting (SLAC) and fixed effects likelihood (FEL) methods as implemented in DataMonkey server (http://www.datamonkey.org). The overall ω (d N /d S ) value was calculated according to NJ trees under the TrN93 substitution model. Positive selection for a site was considered to be statistically significant if P-value was < 0.1. A mixed-effects model of evolution (MEME) was further used to identify positively selected sites under episodic diversifying selection in particular positions among different clades within a phylogenetic tree even when positive selection is not evident across the entire tree. The relative residue abundance within BC and DE-surface exposed loops were depicted using WebLogo.