Abstract
Alterations of the receptor-binding properties of swine influenza A viruses (SIVs) during their isolation in embryonated chicken eggs have not been well studied. In this study, the receptor-binding properties of classical H1 SIVs isolated solely in eggs or Madin–Darby canine kidney (MDCK) cells were examined. Sequencing analysis revealed substitutions of D190V/N or D225G in the haemagglutinin (HA) proteins in egg isolates, whereas MDCK isolates retained HA genes identical to those of the original viruses present in the clinical samples. Egg isolates with substitution of either D190V/N or D225G had increased haemagglutinating activity for mouse and sheep erythrocytes, but reduced activity for rabbit erythrocytes. Additionally, egg isolates with D225G had increased haemagglutination activity for chicken erythrocytes. A direct binding assay using a sialyl glycopolymer that possessed either a 5-N-acetylneuraminic acid (Neu5Ac) α2,6galactose (Gal) or a Neu5Acα2,3Gal linkage revealed that the egg isolates used in this study showed higher binding activity to the Neu5Acα2,3Gal receptor than MDCK isolates. Increased binding activity of the egg isolates to the Neu5Acα2,3Gal receptor was also confirmed by haemagglutination assay with resialylated chicken erythrocytes by Galβ1,3/4GlcNAcα2,3-sialyltransferase. These observations were reinforced by flow-cytometric and N-glycan analyses of the erythrocytes. The α2,3-linked sialic acids were expressed predominantly on the surface of mouse and sheep erythrocytes. Chicken erythrocytes expressed Neu5Acα2,3Gal more abundantly than Neu5Acα2,6Gal, and rabbit erythrocytes expressed both 5-N-glycolylneuraminic acid (Neu5Gc) α2,6Gal and Neu5Acα2,6Gal. Our results demonstrate clearly that classical H1 SIVs undergo alterations in receptor-binding activity associated with an amino acid substitution in the HA protein during isolation and propagation in embryonated chicken eggs.
INTRODUCTION
Swine influenza viruses (SIVs) have been recognized as an important pathogen in swine production and a public-health concern worldwide (Olsen et al., 2006a). Three subtypes, H1N1, H1N2 and H3N2, have been predominantly circulating in pig populations over several decades (Van Reeth, 2007). However, these subtypes have genetically distinctive characters in different geographical locations, and complicated genetic diversity has thus been demonstrated even among the same subtypes (Brown et al., 1998; Campitelli et al., 1997; Guan et al., 1996; Karasin et al., 2006; Olsen et al., 2006b; Saito et al., 2008; Takemae et al., 2008; Vincent et al., 2008; Webby et al., 2000; Yu et al., 2009). Because of the segmented nature of the genome, co-infection with more than two influenza viruses of different origins in a single host allows viruses to exchange their genes, resulting in this genetic diversity (Olsen et al., 2006a).
Pigs are known to be susceptible to both human and avian influenza viruses (Kida et al., 1994). Influenza viruses bind to the sialic acid (SA) linked to galactose (Gal) on host cells through the receptor-binding site of the haemagglutinin (HA) protein (Lamb & Krug, 2001). Human viruses bind preferentially to SA linked to Gal by an α2,6 linkage (SAα2,6Gal), whereas avian viruses bind preferentially to SAα2,3Gal (Rogers & D'Souza, 1989; Rogers & Paulson, 1983). It has been demonstrated that both the SAα2,3Gal and SAα2,6Gal receptors are distributed on the pig tracheal epithelium (Ito et al., 1998). This observation substantiates the hypothesis that pigs can be a mixing vessel of human and avian influenza viruses. In addition, SIVs with the classical swine H1 gene have been reported to bind to both 5-N-acetylneuraminic acid (Neu5Ac) α2,3Gal and Neu5Acα2,6Gal (Matrosovich et al., 2000).
Two substrates, embryonated chicken eggs and Madin–Darby canine kidney (MDCK) cells, have generally been used for the isolation and propagation of influenza viruses over the years (Lamb & Krug, 2001). Human viruses isolated and grown in eggs, however, change their receptor-binding properties, which is known as egg adaptation (Katz et al., 1987). Numerous studies have been carried out to elucidate the mechanism of egg adaptation in human H1N1 and H3N2 influenza viruses (Katz et al., 1987; Robertson et al., 1987). Several passages of a human influenza virus in the allantoic cavity of eggs induce amino acid substitutions within the receptor-binding site on the HA1 molecule, resulting in acquisition of specificity to SAα2,3Gal receptors (Gambaryan et al., 1999; Ito et al., 1997b). Such alteration of binding properties was suggested to arise under selective pressure by the receptors on the cells in which the virus replicates; epithelial cells of the allantoic cavity possess only receptors with the α2,3 linkage, whereas MDCK cells possess receptors with both the α2,3 and α2,6 linkages (Ito et al., 1997b).
Unlike human influenza A viruses, alteration of the receptor-binding specificities of SIVs after isolation and passages in embryonated eggs has not been well documented. A recent study indicated that MDCK-grown SIVs with classical swine H1 genes displayed affinity to sialyl glycopolymers with α2,6 linkage, but not to those with α2,3 linkage, suggesting that egg-adapted SIVs altered their receptor-binding properties (Gambaryan et al., 2005). The present study was designed to clarify the effects of substrate for the isolation of SIVs on their growth characteristics and receptor-binding specificities. We investigated the amino acid changes on the HA1 molecule associated with isolation and passages in eggs by comparing putative amino acid sequences obtained from clinical specimen-derived, egg-grown and MDCK-grown SIVs of the H1N1 and H1N2 subtypes isolated in 2008 in Thailand and Japan, respectively. The receptor-binding properties of egg-grown and MDCK-grown viruses were compared by their haemagglutinating activities with erythrocytes from different animal species. A receptor-binding assay using sialyl glycopolymer and a haemagglutination assay using enzymically resialylated chicken erythrocytes were also performed. SA linkages on the erythrocyte surfaces were examined by N-glycan and flow-cytometric analyses. The results provide significant information on the receptor-binding properties of SIVs, information that substantiates the egg adaptation of SIVs and enables us to suggest a preferential substrate for isolating SIVs.
RESULTS AND DISCUSSION
Virus isolation and amino acid substitutions observed in egg isolates
SIV of the H1N1 subtype, designated A/swine/Ratchaburi/NIAH101942/2008 (Rat101942), was isolated from a nasal swab collected from a clinically healthy, 12-week-old fattening pig in both MDCK cells and the allantoic cavity of embryonated eggs. After two passages in each substrate, the supernatant of the infected cells showed HA activity at a titre of 8, whereas infectious allantoic fluid showed HA activity at a titre of 4, with 1 % guinea pig red blood cells (RBCs). The MDCK isolate of A/swine/Tochigi/1/2008 (H1N2) (Tochigi1) was obtained in this study from the lung homogenate of a 22-week-old pig in Japan that showed coughing and a high body temperature (Yoneyama et al., 2010). Harvested viruses from MDCK cells and from the amniotic fluid of Tochigi1 after two passages in each substrate showed HA titres of 4 and 16 with 1 % chicken RBCs, respectively. However, Tochigi1 did not grow in allantoic fluid in earlier passages (Yoneyama et al., 2010). The passage histories of Rat101942 and Tochigi1 analysed in this study appear in Table 1⇓. Phylogenetic analysis of HA gene sequences of both viruses obtained by direct sequencing of the nasal swabs revealed that they belonged to the classical swine H1 lineage (data not shown). N1 and other internal genes of Rat101942 resided in the Eurasian avian-like swine lineage (data not shown), as did previously characterized Thai H1N1 isolates (Takemae et al., 2008).
Comparison of growth characteristics and nucleotide and amino acid sequences
na, Not applicable.
Coding sequences of the MDCK isolates of Rat101942 (MD5) were identical in all eight gene segments to the sequences obtained from PCR products amplified directly from the specimen. This was also the case with the MDCK isolate of Tochigi1 (MD2) in the HA, neuraminidase (NA) and nucleoprotein (NP) genes. Sequence comparison between MD5 and Al3 of Rat101942 revealed substitutions in the deduced amino acid sequences at D190V (H3 numbering; Nobusawa et al., 1991) in the HA1 protein and K99R in the NP protein (Table 1⇑). Double peaks of A and G at nt 665 in HA1, resulting in a mixed population (D and G) at position 225 in the HA1 region, were seen in Am2 of Tochigi1. Thus, MDCK isolates appear to represent viruses originally existing in the specimens. The amino acids at positions 190 and 225 in HA1 are known to be located at the receptor-binding site and the left edge of the receptor-binding pocket of the HA protein (Gamblin et al., 2004; Nobusawa et al., 1991), respectively (Fig. 1⇓).
Three-dimensional structure of the globular head of the HA protein, based on A/swine/Iowa/30 (H1N1), from front (a) and lateral (b) views. The locations of aa 190 and 225 are illustrated by space-fill modelling. The receptor-binding site is shaded.
It was interesting to note that, although valine predominated at position 190 in Al3 of Rat101942, a smaller peak of A at nt 560 in HA1, resulting in a non-synonymous substitution of D at position 190, was observed in the chromatography of the sequence analysis. To examine the extent of each variant of Rat101942 at the position, 50 clones encoding the HA1 regions from each sample (nasal swab, MD5 and Al3) were analysed by using a TOPO TA cloning kit for sequencing (Invitrogen) according to the manufacturer's instructions. Forty-seven of 50 clones (94 %) had Val at position 190 in HA1 of Al3. No variations were found in the clones derived from the nasal swab and MD5 (Table 1⇑).
Comparison of growth ability and receptor-binding properties between MDCK isolates and egg isolates
The observation that egg isolates have an amino acid substitution in the vicinity of the receptor-binding site in the HA1 region prompted us to examine whether such amino acid substitutions are associated with alteration(s) in the receptor-binding properties and/or growth abilities of the egg isolates.
Rat101942 strains were subjected to further plaque purification in MDCK cells three times after MD5 and limiting dilutions four times in eggs after Al3 to eliminate the heterogeneity of the viruses before subjecting them to analyses of their receptor-binding specificity. MD9 or MD10, which were propagated on MDCK cells after plaque purification, were confirmed to be identical in the HA gene segment to the original virus. By the first limiting dilution, two kinds of variant at the same position, D190N and D190V, were obtained (Al4 of Rat101942). Clones possessing valine and asparagine at position 190 after final limiting dilution followed by propagation were designated Al8-1 (D190V) and Al8-2 (D190N), respectively (Table 1⇑). After establishing the clone, MD9 of Rat101942 was passaged in the allantoic cavity of eggs, and the egg isolates (Al8-1, Al8-2) were passaged in MDCK cells to examine whether the substitutions at positions 190 in the HA1 and 99 in the NP were reversible in opposite environments. No changes at position 190 in the HA1 or position 99 in the NP were observed in either MD9 passaged in eggs five times (MD9Al5) or the egg isolates passaged five times on MDCK cells (Al8-1MD5 and Al8-2MD5).
Tochigi1 MD2 was also propagated further twice in MDCK cells (MD4), and Am2 in the allantoic cavity three times to obtain Am2Al3 (Table 1⇑). MD4 was confirmed to possess HA, NA and NP genes identical to those seen in the lung samples. Am2Al3 (D225G) showed a single peak of G at nt 665 in HA1 in the chromatography of the sequence analysis. Except for that residue, the identity of the HA, NA and NP genes with Am2 was confirmed in Am2Al3.
Comparison of 50 % egg infective dose (EID50) ml−1 and 50 % tissue culture infective dose (TCID50) ml−1 values on MDCK cells demonstrated that the Tochigi1 egg isolates developed growth ability in the allantoic cavity of embryonated eggs (Table 1⇑). The virus titre in the eggs of Tochigi1 Am2Al3 was significantly higher than the TCID50 ml−1 value (Student's t-test, P<0.05). The virus titre in the eggs of Tochigi1 MD4 was comparable to that in the MDCK cells. This suggests that the Tochigi1 egg isolates were more adapted in their ability to replicate in the allantoic cavity of embryonated eggs. Due to egg adaptation, there was increased growth ability of Tochigi1 Am2 in embryonated eggs associated with D225G in the HA molecule. In contrast, there was no change in the growth abilities of Rat101942 in the allantoic cavity (Table 1⇑). All of the viruses obtained showed higher titres in the allantoic cavity than in MDCK cells (Student's t-test for MD10 and Al8-2, P<0.05; Welch's t-test for Al8-1, P<0.05). The mechanism underlying the difference in growth ability in embryonated eggs remains to be elucidated.
Evidence that viruses derived from different substrates differ in their receptor specificities was obtained by haemagglutination tests with erythrocytes from different animals (Table 2⇓). MDCK and egg isolates agglutinated rabbit, mouse and sheep erythrocytes differently. MDCK isolates (MD10 of Rat101942 and MD4 of Tochigi1) agglutinated rabbit erythrocytes with HA titres of 256 and 8, respectively. However, HA titres of the egg isolates (Al8-1 and Al8-2 of Rat101942 and Am2Al3 of Tochigi1) with rabbit erythrocytes were <1. Although the MDCK isolates did not agglutinate mouse or sheep erythrocytes, all egg isolates did (Table 2⇓). Between Al8-1 and Al8-2 of Rat101942, no apparent differences were observed. In addition, Am2Al3 of Tochigi1 had an 8-fold increase in haemagglutination activity with chicken erythrocytes compared with that of MD4 (Table 2⇓). These results revealed that the egg isolates had increased haemagglutinating activity with mouse and sheep erythrocytes and reduced activity with rabbit erythrocytes after egg adaptation.
Haemagglutinating activity with 1 % erythrocytes from different animals
HA titres of each virus were standardized to 64 by 1 % guinea pig RBCs. Amino acid substitutions in HA1 are indicated in parentheses. <1 indicates that no titres were observed with the undiluted viruses.
A direct binding assay using sialyl glycopolymers demonstrated clearly that mutations at either position 190 or 225 in HA1 of egg isolates enhanced their Neu5Acα2,3Gal specificity (Fig. 2⇓). MD10 of Rat101942, possessing D at position 190 in HA1, bound significantly to Neu5Acα2,6Gal, but not to Neu5Acα2,3Gal (Student's or Welch's t-test was applied according to preliminary tests of equality of variances for concentrations from 1.6 to 100 ng per well, P<0.01) (Fig. 2a⇓). Increased binding to Neu5Acα2,3Gal was demonstrated clearly with the egg-adapted isolates of Rat101942, although they still bound preferentially to Neu5Acα2,6Gal, as judged by the absorbance values from the assay (Fig. 2b, c⇓). The binding patterns obtained were similar to that of the D225G mutation in HA1 of Tochigi1 (Fig. 2d, e⇓). MD4 (225D) of Tochigi1 bound significantly to Neu5Acα2,6Gal (Student's or Welch's t-test for concentrations except for 0 and 3.1 ng per well, P<0.05) (Fig. 2d⇓), whereas Am2Al3 (225G) clearly increased the binding for Neu5Acα2,3Gal (Fig. 2e⇓). These results showed that Tochigi1 from a clinical sample could be grown successfully in the amniotic cavity, but not in the allantoic cavity. Amniotic membrane cells, which possess receptors with both α2,3 and α2,6 linkages (Ito et al., 1997b), appeared to allow Tochigi1 original virus with α2,6 specificity to replicate in the amniotic cavity, whereas a minority of the Tochigi1 virus with the D225G substitution eventually grew to replace the original population by utilizing α2,3 receptors. No substitutions except for D225G were found in Tochigi1 Am2. Our previous study revealed that Japanese H1 SIVs possessing 225D from clinical specimens did not grow well in the allantoic cavity (Saito et al., 2008). A phenomenon similar to that seen in the SIVs examined is well-known in human influenza epidemic viruses. Human H3N2 viruses passaged in the amniotic cavity several times showed a substantial increase in α2,3 affinity, resulting from an amino acid substitution (L226Q) in the receptor-binding site in the HA1 molecule (Ito et al., 1997b). In addition, human H1 egg-adapted variants with either 190D→N or 225D→G/N in the HA1 molecule always had increased affinity for Neu5Acα2,3Gal-containing receptors (Gambaryan et al., 1999).
Direct binding assay using sialyl glycopolymers for Rat101942 and Tochigi1. The receptor-binding activities of MD10 (a), Al8-1 (b) and Al8-2 (c) of Rat101942, and MD4 (d) and Am2Al3 (e) of Tochigi1 are shown. □ and ○ represent mean±sd absorbances (n=3) of Neu5Acα2,3Gal and Neu5Acα2,6Gal, respectively. Asterisks indicate significant differences (*P<0.05; **P<0.01) between the absorbances for each concentration of sialyl glycopolymers by Student's or Welch's t-test.
The results described above provided us with another perspective on residue 225 in the HA of classical H1 SIVs. Most of the triple-reassortant H1 SIVs in North America, as well as the H1N2 SIVs in Japan, have retained 225D since 1999 and 2003 (Choi et al., 2002; Karasin et al., 2000; Saito et al., 2008; Yassine et al., 2009), respectively, whereas most of the classical SIVs circulating before the late 1990s had 225G (Olsen et al., 2000; other data from GenBank). It has been suspected that the classical swine H1 protein retained 225G due to egg adaptation (Gambaryan et al., 2005), coinciding with the fact that many laboratories used embryonated eggs for SIV isolation before the 1990s (Kida et al., 1994; Olsen et al., 2000). However, Rat101942 MD10 in this study and some of the triple-reassortant H1 viruses isolated from MDCK cells, such as A/swine/MN/22860-T/2001 (H1N2), retained 225G (Choi et al., 2002). It is therefore evident that not all D225G substitutions observed in recent SIVs are due to egg adaptation.
MDCK isolates (MD10 of Rat101942 and MD4 of Tochigi1) appeared to possess an ability to bind to Neu5Acα2,3Gal, because the binding of MDCK isolates to Neu5Acα2,3Gal increased slightly at sialyl glycopolymer concentrations of 12.5–100 ng per well when twice the usual amount of MDCK isolates was used in the direct binding assay (data not shown), suggesting that binding of the MDCK isolates to Neu5Acα2,3Gal can be observed in the assay when sufficient viruses against the receptors exist. This was clearly supported by a haemagglutination assay using resialylated chicken red blood cells (R-CRBCs) (Fig. 3⇓). MDCK isolates of both Rat101942 and Tochigi1 agglutinated the Neu5Acα2,3Gal R-CRBCs, although the titres were lower than those for the native CRBCs and the Neu5Acα2,6Gal R-CRBCs. All of the egg isolates were shown to bind both receptors as well as those of native CRBCs (Fig. 3⇓), coinciding with the result in the direct binding assay.
Haemagglutinating activity of Rat101942 and Tochigi1 with native CRBCs and R-CRBCs with Neu5Acα2,3Gal or Neu5Acα2,6Gal.
SA expression on RBCs
The relative amounts of SAα2,3Gal and SAα2,6Gal linkages on the chicken, sheep, mouse and rabbit erythrocyte surfaces, which showed different HA reactivity between MDCK and egg isolates, reinforced the results described above (Fig. 4⇓). Chicken erythrocytes were labelled with lectins binding to both SAα2,3Gal and SAα2,6Gal linkages, as reported previously (Ito et al., 1997a). Sheep and mouse erythrocytes were shown to contain the SAα2,3Gal linkage; the presence of the SAα2,6Gal linkage was not demonstrated by this assay. In contrast, on rabbit erythrocytes, the SAα2,6Gal linkage was shown to exist in a lower amount and SAα2,3Gal linkage was not demonstrated.
Relative amounts of SAα2,3Gal and SAα2,6Gal on the surface of chicken, sheep, mouse and rabbit erythrocytes. Cell numbers are plotted against log10(fluorescence intensity) of cells incubated without lectins (negative control) or with DIG-labelled lectins from Maackia amurensis (specific for SAα2,3Gal) or from Sambucus nigra (specific for SAα2,6Gal).
N-glycan analysis characterized N-linked glycans expressed on the surfaces of rabbit and chicken erythrocytes. Three and four peaks were eluted by diethylaminoethyl columns in rabbit and chicken erythrocytes, respectively [see Supplementary Fig. S1(a), available in JGV Online]. Peaks 1, 2 and 3 observed in both erythrocyte types were identified as neutral, monosialylated and disialylated N-glycans, respectively. Trisialylated glycan (peak 4) was only observed in chicken erythrocytes. The molar ratios of each N-glycan calculated from the peak areas were 83.5, 11.0 and 5.5 % in rabbit erythrocytes, respectively, and 28.9, 23.0, 26.5 and 21.6 % in chicken erythrocytes, respectively. The N-glycan structure analysis of each peak obtained by octadecyl silica (ODS) columns [Supplementary Fig. S1(b–e)] identified 21 kinds of N-glycan, including 18 neutral glycans, one monosialylated glycan and two disialylated glycans, on rabbit erythrocytes. Twenty-seven glycans, including ten neutral glycans, seven monosialylated glycans, eight disialylated glycans and two trisialylated glycans, were also identified on chicken erythrocytes. Each structure identified is shown in Table 3⇓. Both Neu5Ac and 5-N-glycolylneuraminic acid (Neu5Gc) were found on rabbit erythrocytes, whilst, in contrast, only Neu5Ac was found on chicken erythrocytes (Fig. 5⇓; Table 3⇓). The most abundant SA species on rabbit erythrocytes was Neu5Gcα2,6Gal (8.2 %), which was about four times more abundant than Neu5Acα2,6Gal (2.0 %; Fig. 5⇓). On chicken erythrocytes, Neu5Acα2,3Gal (37.9 %) was expressed more predominantly than Neu5Acα2,6Gal (17.4 %; Fig. 5⇓). The total amounts of N-glycans were 18.6 and 8.4 pmol (mg dried rabbit and chicken erythrocytes)−1, respectively.
Molar ratio of acidic N-glycans to total N-glycans on the surface of rabbit and chicken erythrocytes. Each amount of acidic N-glycans [pmol (mg dried erythrocytes)−1] is shown on the bars.
Structures and relative quantities of the neutral, mono-, di- and trisialyl pyridylamino (PA) oligosaccharides derived from rabbit and chicken erythrocytes
The egg-adapted SIVs in this study did not agglutinate rabbit erythrocytes (Table 2⇑), which contain both Neu5Ac-and Neu5Gcα2,6Gal molecules (Fig. 5⇑), whereas the MDCK isolates did. However, the egg-adapted isolates agglutinated Neu5Acα2,6Gal R-CRBCs (Fig. 3⇑). There were four times more Neu5Gcα2,6Gal molecules than Neu5Acα2,6Gal molecules on rabbit erythrocytes (Fig. 5⇑). Neu5Gc molecules were shown to be expressed abundantly on pig tracheal epithelia (Suzuki et al., 1997). Some of the avian-like H1 and human-like H3 swine viruses, as well as the human H1 and H3 viruses, have been suggested to recognize Neu5Gcα2,6Gal molecules as their receptors (Higa et al., 1985; Ito et al., 1997a). Those observations suggest the potential of Neu5Gcα2,6Gal as a receptor for classical H1 SIVs. The mutation at either residue 190 or 225 of HA1 during egg adaptation might reduce the binding ability with Neu5Gcα2,6Gal receptors, so that the egg-adapted SIVs fail to agglutinate rabbit erythrocytes through Neu5Gcα2,6Gal receptors. Further studies are needed to investigate the role of Neu5Gcα2,6Gal as a receptor molecule for classical SIVs.
As has been pointed out by many researchers (Olsen et al., 2006a), because pigs can act as a mixing vessel of influenza viruses, it is crucial to monitor influenza viruses in these animals to prevent emergence of a possible pandemic virus (Webster et al., 1992). Cases of swine-to-human transmission of influenza virus have been reported sporadically worldwide many times in the past (Myers et al., 2007). At least 11 sporadic human cases by triple-reassortant swine H1N1(2) viruses were confirmed from December 2005 to February 2009 in the USA (Shinde et al., 2009). Pandemic (H1N1) 2009 viruses that possess the HA, PB2, PB1, PA, NP and NS genes from triple-reassortant H1 SIVs and the remaining two genes, encoding the NA and matrix (M) proteins, from Eurasian avian-like SIVs have been a threat to humans since the infections of two patients were first identified in April 2009 in the USA (CDC, 2009). Also, highly pathogenic H5N1 avian influenza viruses still remain an important public-health concern in the world (WHO, 2009). H5N1 viruses transmitted from poultry to pigs have been isolated sporadically in China (Shi et al., 2008) and Indonesia (Takano et al., 2009). It is crucial to evaluate the receptor specificity of influenza virus isolated from the pig population in order to assess transmission ability among humans, predicting pandemic ability. However, most of the swine H5N1 viruses in the above-mentioned studies were isolated in embryonated eggs, and thus could be subjected to egg adaptation by which their ability to bind α2,6 linkages is altered. Even in a natural setting, it has been demonstrated that the H1 avian virus altered its receptor-binding property in pigs soon after the epizootic of 1979 in the European swine population (Matrosovich et al., 2000). Choice of an adequate substrate to isolate SIVs has therefore become increasingly important for the detailed characterization of influenza viruses isolated from pigs. Although a limited number of strains were examined in the present study, the results provide evidence that the MDCK cell is a more desirable substrate than embryonated chicken egg for isolation of influenza viruses from pigs.
METHODS
Sample collection and virus isolation.
Forty nasal swabs were collected from apparently healthy pigs, including 20 sows (1–2 years old), ten 12-week-old fattening pigs and ten 9-week-old piglets, at a farm in Ratchaburi province, Thailand, in January 2008. Each swab was placed into a 15 ml tube containing 2 ml transport medium [minimal essential medium (MEM; Invitrogen) including penicillin (1000 units ml−1), streptomycin (1000 μg ml−1), fungizone (25 μg ml−1) and 0.5 % BSA, 0.01 M HEPES]. The swabs were transported to NIAH, Thailand, on ice. Each swab was divided into three portions. Two of the three portions were used for virus isolation after filtering with a 0.45 μm pore size filter (Millipore): one was inoculated into the allantoic cavities of 10-day-old embryonated chicken eggs and incubated for 48 h at 37 °C, and the other was inoculated onto a monolayer of MDCK cells for 45 min at 37 °C for viral adsorption to the cells. After adsorption, MEM without fetal calf serum containing 1 μg acetylated trypsin ml−1 was added. The inoculated cells were incubated for more than 48 h at 37 °C in 5 % CO2 until the cytopathogenic effect appeared. Haemagglutinating agents arising from the same swab in both an embryonated egg and MDCK cells were identified as H1N1 by the PCR method using specific primers designed in a previous study (Takemae et al., 2008). They were designated A/swine/Ratchaburi/NIAH101942/2008 (H1N1) (Rat101942). The last portion was subjected to viral RNA extraction by using an RNeasy mini kit (Qiagen), followed by RT-PCR and direct sequencing of the PCR products. A lung specimen from a pig, which yielded A/swine/Tochigi/1/2009 (H1N2) (Tochigi1) (Yoneyama et al., 2010), was subjected to virus isolation in MDCK cells at NIAH, Japan. After being ground in the transport medium (10 %, w/v), the specimen was inoculated into MDCK cells as described above. Tochigi1 isolated and passaged twice in the amniotic cavity of embryonated chicken eggs at Tochigi Central Livestock Hygiene Service Center was subjected to further analysis at NIAH, Japan. Rat101942 and Tochigi1 were subsequently passaged in MDCK cells or the allantoic cavity (Table 1⇑).
Characterization of viruses and phylogenetic analysis.
RT-PCR, direct sequencing of the PCR products and phylogenetic analysis of the viruses obtained in this study were carried out as described previously (Takemae et al., 2008). The TCID50 ml−1 values on MDCK cells and the EID50 ml−1 values at 37 °C of the viruses used were determined.
Haemagglutination assay with erythrocytes of different animal species.
One per cent erythrocytes of guinea pig, chicken, goose, rabbit, rat, mouse and sheep in PBS (pH 7.4) were used. When comparing the haemagglutination activities of each virus with different species, the quantities of each virus were standardized with 1 % guinea pig erythrocytes to give HA titres of 64. Viruses with HA titres of <64 were concentrated by using an Amicon Ultra centrifugal filter with an MWCO 10 000 membrane (Millipore), then diluted with PBS to obtain HA titres of 64 with 1 % guinea pig erythrocytes. Titres were judged after 30 min incubation with chicken and goose erythrocytes and 1 h incubation with the other mammalian erythrocytes at room temperature (RT).
Direct binding assay.
One hundred microlitres of serial dilutions of sialyl glycopolymers containing Neu5Ac linked to Gal by either an α2,3 or an α2,6 linkage in PBS were added to the wells of 96-flat-well microplates (Costar 2504; Corning) to give a final concentration range of 0–1000 ng ml−1 (0–100 ng per well). The plates containing sialyl glycopolymers were irradiated with UV light at 254 nm for 10 min and then washed three times with 300 μl PBS. The plates were blocked with PBS Milk Blocking Solution (BioFX Laboratories) for 30 min at RT. After washing three times with PBS containing 0.1 % Tween 20 (PBST), 50 μl virus-containing PBS with HA titres of 32 with 1 % guinea pig erythrocytes was applied to the wells and incubated at 4 °C for 1 h. Virus-containing PBS was prepared to reduce the materials binding non-specifically to sialyl glycopolymers as follows: allantoic fluid or culture supernatant containing viruses (MD10, Al8-1 and Al8-2 of Rat101942 and MD4 of Tochigi1) was concentrated by using an Amicon Ultra centrifugal filter, then diluted with PBS. Am2Al3 of Tochigi1 was partially purified by ultracentrifugation (28 000 r.p.m. for 2 h in an SW28 rotor) through 25 and 75 % sucrose, then diluted with PBS, because their non-specific binding materials in allantoic fluid could not be removed by using the Amicon Ultra centrifugal filter. After adsorption of the viruses to sialyl glycopolymers, the plates were fixed with 100 μl 3.7 % formaldehyde in each well for 30 min at RT. Allantoic fluid or culture supernatant without viruses was used as a negative control after filtration by an Amicon Ultra centrifugal filter. After five washes with PBST, 50 μl anti-A/duck/Bavaria/2/1977 (H1N1) rabbit hyperimmune serum for Rat101942 strains and anti-A/swine/Iowa/15/1930 (H1N1) chicken hyperimmune serum for Tochigi1 at a dilution of 1 : 6400 with PBST containing 0.5 % BSA were added to respective wells. These two different antibodies, prepared in NIAH, Japan, were chosen by reactivity using the haemagglutination inhibition or ELISA tests (data not shown). However, because of different affinities of the antibodies to the viruses used, it was not possible to compare the receptor-binding affinities of Rat101942 and Tochigi1. The plates were incubated for 1 h at 37 °C. Next, after washing five times again with PBST, 50 μl peroxidase-conjugated AffiniPure Goat Anti-Rabbit IgG or peroxidase-conjugated AffiniPure Goat Anti-Chicken IgY (IgG) (both from Jackson ImmunoResearch Laboratories) diluted at 1 : 64000 with PBST containing 0.5 % BSA was applied to the respective wells. After incubation for 1 h at 37 °C, 100 μl tetramethylbenzidine substrate (BioFX Laboratories) was added to the wells and the plates were incubated for 30 min at RT. Reactions were stopped with 100 μl 450 nm Liquid Stop Solution (BioFX Laboratories). Absorbances were measured at 450 nm.
Resialylation of CRBCs.
CRBCs were modified enzymically to contain exclusively either Neu5Acα2,3Gal or Neu5Acα2,6Gal. Briefly, 100 μl aliquots of 10 % CRBCs suspended with PBS(+) containing 0.9 mM CaCl2 and 0.5 mM MgCl2, pH 7.4, were treated with 50 mU sialidase from Vibrio cholerae (Roche), hydrolysing N- or O-acyl SAs in α2,3, α2,6 and α2,8 bonds with glycoconjugates for 2 h at 37 °C. Removal of SA from native CRBCs was confirmed by the absence of haemagglutination activity with the SIVs used. Desialylated CRBCs were washed twice with cold PBS, then divided into two aliquots and resuspended in 50 μl PBS or PBS(+) containing 1 % BSA, respectively. Cytidine 5′-monophospho-N-acetylneuraminic acid (1.5 mM; Sigma-Aldrich) with either 2.5 mU Galβ1,4GlcNAcα2,6-sialyltransferase (Calbiochem) or 1 mU Galβ1,3/4GlcNAcα2,3-sialyltransferase (Calbiochem) was added into respective desialylated CRBCs, then incubated at 37 °C for 4 h or overnight, respectively. After three washes with cold PBS, the resialylated CRBCs were suspended with 500 μl PBS at a final concentration of 1 % (v/v) and then used for the haemagglutination assay. The HA titres were judged after 1 h incubation at 4 °C.
Flow-cytometric analysis of SA linkages on erythrocytes.
Relative amounts of SAα2,3Gal and SAα2,6Gal on the surface of the erythrocytes of chicken, sheep, mouse and rabbit were detected by using a digoxigenin (DIG) glycan differentiation kit (Roche). Briefly, approximately 5×106 erythrocytes were washed twice with cold PBS containing 10 mM glycine, followed by once with buffer 1 (50 mM Tris/HCl, 0.15 M NaCl, 1 mM MgCl2, 1 mM MnCl2, 1 mM CaCl2, pH 7.5). The washed erythrocytes were incubated with blocking solution supplied with the DIG kit for more than 30 min at RT. DIG-labelled lectins, SNA (Sambucus nigra agglutinin) specific for Neu5Ac/Neu5Gcα2,6Gal residues or MAA (Maackia amurensis agglutinin) specific for Neu5Ac/Neu5Gcα2,3Gal residues, were incubated at a final concentration of 1 μg ml−1 with anti-DIG antibodies conjugated to fluorescein (Roche) diluted 200-fold in buffer 1 for 1 h at RT. Erythrocytes in blocking solution were suspended in the lectin–antibody solution for 1 h at 4 °C. Control experiments were performed without lectins. The fluorescence intensity of 10 000 erythrocytes from each animal was analysed on a flow cytometry EPICS XL system 2 (Beckman Coulter).
N-glycan analysis of rabbit and chicken erythrocytes.
Approximately 10 ml fresh blood obtained from rabbits and chickens was haemolysed in 10 vols 0.2 % acetic acid, followed by three washes with saline by centrifugation at 30 000 g for 15 min and then drying in a vacuum overnight. Lipids of dried cells of rabbit (0.16 g) and chicken (0.55 g) were extracted in a series of ethanol, chloroform/methanol and acetone (Sriwilaijaroen et al., 2009). Subsequently, N-linked glycans of each erythrocyte were analysed by multi-dimensional HPLC mapping and matrix-assisted laser desorption–ionization time-of-flight mass spectrometric (MALDI-TOF-MS) techniques as described previously (Sriwilaijaroen et al., 2009; Takahashi & Kato, 2003).
Acknowledgments
This work was supported by the programme of the Founding Research Center for Emerging and Reemerging Infectious Diseases commissioned by the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan.