Avian influenza H9N2 virus isolated from air samples in LPMs in Jiangxi, China

In this study, 46 sample sites in ten different LPMs were selected in Nanchang city, Jiangxi province, China. These LPMs were distributed in four districts (Xinjian, Donghu, Xihu, and Qingshanhu) (Fig. 1) of the city. Jingdong Bird and Flower market in Qingshanhu district was a large wholesale market; the others were all retail markets. Air samples were collected monthly from April 2014 to March 2015 via special air pumps (SQC-1000, Yancheng Tianyue Instrument & Meter Co., Ltd., Jiangsu, China). This equipment is widely applied in collecting aerial particulates and aerosols in China and uses a collection bottle (Qingdao Junray Intelligent Instrument Co., Ltd.) to collect air samples by passing through a transport medium. In each month, 10 air samples were collected in Jingdong Bird and Flower market, and 6 air samples were collected in each of the other retail LPMs. For retail markets, first, the wind direction in the market was determined. The poultry stall was chosen as the study center; 2 air samples were collected on the left and right of the stall at 3 m from the stall; and three other air samples were collected at 3 m, 5 m and 7 m from the stall in the downwind direction; another air sample was collected in the upwind direction at 1 m distance from the stall (Fig. 2a). For Jingdong Bird and Flower wholesale market, five air samples were collected in the east, south, west, and north and the center of the LPM. Samples were also collected 1 m, 3 m, 5 m and 7 m from the outside entrance (Fig. 2b). Each air sample was collected for 1.5 h, then stored in MEM (Gibco), aliquoted into 3 tubes and stored at −80 °C. To avoid excess bubbles, no BSA was included in the transport medium during sample collection; instead, BSA was added immediately after collection. All samples were transported on ice.

RNA was extracted with Qiagen RNeasy Mini kit, and influenza A was detected by RRT-PCR with primer and probe sets recommended by WHO (FluA-Forward: GACCRATCCTGTCACCTCTGAC; FluA-Reverse: GGGCATTYTGGACAAAKCGTCTACG;

FluA-probe: TGCAGTCCTCGCTCACTGGGCACG). If the aerosol samples were Flu A-positive, they were treated with 6 antibiotics (Penicillin: 2 × 10⁶ U/L, Streptomycin: 200 mg/L, Polymyxin: 2 × 10⁶ U/L, Gentamicin: 250 mg/L, Nystatin: 0.5 × 10⁶ U/L, Levofloxacin Hydrochloride: 600 mg/L) for 1 h, then virus isolation was conducted by inoculating the treated samples into 9- or 10-day-old specific-pathogen-free embryonated eggs. After incubation at 37 °C for 48 h, the allantoic liquid of the eggs was harvested and tested by Flu A fast diagnostic kits (Colloidal Gold Antibody Conjugates, Beijing NBGen Bio-technology Limited Company). If the result was positive, the allantoic liquid was aliquoted and stored at −80 °C.

Viral RNA was extracted from harvested allantoic liquids using the Qiagen RNeasy Mini kit. Invitrogen SuperScript III One-step RT-PCR platinum Taq HiFi kit was used to amplify the RNA with one-step RT-PCR primers as follows: influenza primer A, GGGGGGAGCAAAAGCAGG; influenza primer B, GGGGGGAGCGAAAGCAGG influenza primer C, CGGGTTATTAGTAGAAACAAGG [14]. PCR products were purified by QIAGEN MinElute Reaction Cleanup Kit. One ng of DNA product was processed for NGS (next generation sequencing) sample preparation with the Illumina Nextera XT DNA Sample Preparation Kit (96 Samples), and sequencing was performed using a MiSeq v2 kit (500 cycles) to produce 2 × 250 paired-end reads (Illumina, San Diego, CA, USA) [15]. After automated cluster generation with MiSeq, the sequencing files were processed and genomic sequence reads were obtained. The full genome sequence was assembled by CLC platform.

Alignment of each segment was performed with the Mega (Version 6.05) software package; the phylogenetic trees of all the 8 segments were conducted by using neighbor-joining method with Kimura two-parameter distances. The bootstrap value was 1000.

In total, 807 air samples were collected from 46 stalls in 10 different LPMs, from 4 districts in Nanchang city (Fig. 1) between April 2014 and March 2015. Of the 807 air samples, 275 samples (34.1%) were tested to be Flu A positive based on RRT-PCR results. Further statistical analysis suggested that air samples collected 3 m from poultry stalls were significantly more likely to test positive for Flu A than air samples collected at 5 and 7 m in the retail LPM (χ² = 39.792, p < 0.01), while no difference was detected between the positive rate of air samples collected at 5 and 7 m (χ² = 3.030, p > 0.05). In the wholesale LPM, the Flu A positive rates of air samples collected at 5 and 7 m were not significantly different (χ² = 3.155, p > 0.05). The control samples were all negative for Flu A. One virus was successfully isolated from these Flu A RRT-PCR positive samples by inoculating samples into Specific Pathogen-Free (SPF) embryonated chicken eggs, and the HA titer of this virus was 256. This virus isolation positive sample was collected from Jingdong Bird and Flower market in Qingshanhu district, Nanchang city on July 31, 2014. Full genome sequencing of the virus showed it was a subtype-H9N2 virus and was termed as A/environment/Jiangxi/14737/2014 (H9N2) (Abbreviated as JX14737 here after). The sequences of all eight segments of this virus were submitted to the Global Initiative on Sharing All Influenza Data (GISAID) with accession numbers from EPI858159 to EPI858166.

To understand the genetic relationships between this aerosolized H9N2 virus and previously published H9N2 viruses, the homology of JX14737 with other viruses was analyzed with BLAST in GenBank, and it was found that JX14737 shared high nucleotide homology with the H9N2 AIVs from Jiangxi, Guangdong, Zhejiang, Anhui and Shanghai in China. The phylogenetic analysis of the HA and NA genes of the H9N2-subtype virus indicated that both HA and NA genes (Fig. 3a and b) belonged to a Eurasian avian lineage, and the virus fell into the G57 genotype cluster, which belonged to the Y280 lineage. The phylogenetic analysis of six internal genes of the H9N2-subtype virus showed that all of them fell into the G57 genotype cluster (Additional file 1A-F).

A mutation at Q226L (H3 numbering) was detected in the HA protein, which indicated that the virus was prone to binding to α2,6-sialic acid receptor; this human-like receptor binding preference increased the risk of human infection of AIVs [16]. We downloaded 273 HA genes from GISAID of H9N2-subtype viruses that had been collected from different parts of the word between April 2014 and March 2015. After comparing JX14737 with these H9N2 sequences from GISAID, we found an unusual mutation in position 137 of the HA gene. Of these 273 sequences, most had K/T/S/N in position 137, with only 3 isolates showing A as in JX14737. Position 137 K was associated with avian-like receptor binding preference [17], but the significance of 137A had not been determined yet. The I292 V mutation in PB2 was associated with increased polymerase activity in 293 T cells, and the A588 V mutation enhanced polymerase activity, viral replication capability in mammalian and avian cells, and virulence in mice [18]. The hemagglutinin cleavage site of JX14737 possessed only one basic amino acid (PSRSSR↓GL) between HA1 and HA2, which indicated that this virus was a low pathogenic AIV in poultry (Table 1).