Use of Sequence-Independent, Single-Primer-Amplification (SISPA) for rapid detection, identification, and characterization of avian RNA viruses

Abstract

Current technologies with next generation sequencing have revolutionized metagenomics analysis of clinical samples. To achieve the non-selective amplification and recovery of low abundance genetic sequences, a simplified Sequence-Independent, Single-Primer Amplification (SISPA) technique in combination with MiSeq platform was applied to target negative- and positive-sense single-stranded RNA viral sequences. This method allowed successful sequence assembly of full or near full length avian influenza virus (AIV), infectious bronchitis virus (IBV), and Newcastle disease virus (NDV) viral genome. Moreover, SISPA analysis applied to unknown clinical cases of mixed viral infections produced genome assemblies comprising 98% NDV and 99% of IBV genomes. Complete or near complete virus genome sequence was obtained with titers at or above 10^4.5 EID₅₀/ml (50% embryo infectious dose), and virus identification could be detected with titers at or above 10³ EID₅₀/ml. Taken together, these studies demonstrate a simple template enrichment protocol for rapid detection and accurate characterization of avian RNA viruses.

Highlights

• •A simple, random priming technique was optimized to target viral RNA genomes.
• •This technique allows characterization of multiple viruses in single reaction.
• •Complete or near complete genome sequence with titers at or above 10^4.5 EID₅₀/ml.
• •The detection limit with viral titers at or above 10³ EID₅₀/ml.

1. Introduction

RNA viruses are a large genetically-diverse group of infectious agents whose evolutionary diversity is driven by adaptability to its host (Sanjuán et al., 2010). Characterizing the genomes by sequence-based methods is the most widely used approach to determine RNA virus diversity and thereby determine the relationships between isolates within the population. Furthermore, to fully understand the mechanisms of viral adaptation and evolution, and to perform molecular diagnosis of emerging and re-emerging viral infections there is an increasing need to produce full length viral genomes sequences.

Utilization of next generation sequencing (NGS) in molecular epidemiological analysis of outbreak strains facilitates the rapid and accurate identification of etiologic agents and can be used to provide information on the origin of viruses (Gilchrist et al., 2015). In the absence of targeted sequencing protocols, as well as during outbreak situations, the random amplification of purified nucleic acids in combination with NGS can be a useful tool for monitoring viral evolution or diagnosis. Although most molecular detection methods are unable to detect mixed viral infections, NGS allows for both target dependent and target independent sequencing of viral genomes and provides an opportunity to detect the genome of multiple viruses simultaneously (Thorburn et al., 2015). However, the samples used in such studies are often available in limited quantities of virus.

Sequence-Independent, Single-Primer Amplification (SISPA) is one of the random priming methods develop by Reyes and Kim (1991) that allows enrichment of the viral genome in only a few steps (Djikeng et al., 2008). There have been several modifications of the SISPA method since its first implementation, including random-PCR (rPCR) (Froussard, 1993). This combines first-step reverse transcription followed by denaturation, annealing and amplification by using random hexamers tagged with a known sequence which is then used as a primer binding extension sequence. It was developed to yield double-stranded cDNA in sufficient abundance for cloning and then sequencing (Froussard, 1993, Zou et al., 2003). DePew et al. (2013) further modified the SISPA protocol to identify cultivable viruses from a single plaque using 454 and HiSeq Illumina platform without the need of a cloning step.

Previous studies have combined random priming approaches with NGS for identification of a novel mink astrovirus (Blomström et al., 2010), the partial sequencing of a novel paramyxovirus in penguins (Miller et al., 2010b), influenza viruses (Afonso, 2007), metagenomics analysis of Dengue virus infected mosquitoes (Bishop-Lilly et al., 2010), and viruses in human stool samples (Victoria et al., 2009). Moreover, the NGS approach has recently been applied for viral genome sequencing of archived serum samples from the oldest HIV-1 positive group genomes, resulting in identification of “Patient 0” (Worobey et al., 2016).

Single-stranded positive-sense RNA (ssRNA+) and negative sense RNA (ssRNA) viruses represent a large group of avian viral agents, including West Nile virus (WNV), avian metapneumovirus (aMPV), infectious bronchitis virus (IBV), Newcastle disease virus (NDV), and avian influenza virus (AIV) (Swayne, 2013). Among RNA viruses, AIV of Orthomyxoviridae family and NDV of Paramyxoviridae family are considered as the most devastating poultry diseases owing to their highly pathogenic forms as well as worldwide distribution and economic implications (Swayne, 2008; Miller et al., 2010a; Alexander et al., 2012). IBV of the Coronaviridae family also causes significant morbidity and economic losses to the poultry industry affecting primarily the respiratory tract, and can demonstrate a wide range of tissues tropism, including the renal and reproductive systems (Bande et al., 2016). Combining target independent genome amplification with the high sensitivity of NGS provides the opportunity not only for detection of these viral pathogens, but can also simultaneously provide complete genomic sequences to allow further genetic characterization. Nonetheless, the feasibility of applying these techniques to diagnosis and research of pathogens requires further evaluation and optimization study.

A simplified, sequence-independent technique of directional amplification in combination with NGS was used in this study. The protocol was tested against purified virus stocks and clinical samples. Viral genome sequences were determined from both negative- and positive-sense single stranded RNA viruses. The limit of detection necessary for virus identification using SISPA–NGS and metagenomics approaches was also determined.

2. Material and methods

2.1. Viruses

Viruses used in these studies included two highly pathogenic AIV, A/duck/Vietnam/NCVD-672/2011(H5N1) (Dk/Vn), and A/turkey/Minnesota/15–12582-1/2015(H5N2) (Tk/Mn), two NDV isolates, A/duck/Vn/Long bien/78/02 (Dk/Vn/NDV) and Lasota B1 vaccine strain (LaS), and a single IBV strain, Ark99. The IBV (Ark99 strain) was kindly provided by Jackwood (University of Georgia, Athens, USA). All viruses were propagated in 9–11 day of embryonating specific pathogen free chicken eggs. Following three days of growth, the allantoic fluid was harvested for RNA extraction. Clinical samples, consisting of oral swabs, were collected from flocks of chickens in Jordan. The samples were kept under BSL3 facility in −80 °C until used. The SISPA-NGS method was applied to determine viral presence in clinical samples. Furthermore, clinical samples were used for comparison to purified virus stocks.

2.2. RNA isolation

Total RNA extraction from pure virus stocks and clinical samples was performed using RNeasy Mini Kit (QIAGEN, Valencia, USA) according to manufacturer's instruction. Optimal quality of the extracted RNA was verified by obtaining the OD₂₆₀/OD₂₈₀ values using NanoDrop (Thermo Scientific, Wilmington, USA).

2.3. Viral RNA quantification

For AIV, quantitative real time RT-PCR (qRRT-PCR) was performed as previously described (Kapczynski et al., 2013, Spackman et al., 2002). Briefly, qRRT-PCR reactions targeting the influenza virus M gene was conducted using AgPath-ID one-step RT-PCR Kit (Ambion, Austin, TX) and the ABI 7500 Fast Real-Time PCR system (Applied Biosystem, Calsbad, CA). The RT step conditions for reactions were 10 min at 45 °C and 95 °C for 10 min. The cycling conditions were 45 cycles of 15 s, 95 °C; 45 s, 60 °C. For virus quantification, a standard curve was established with RNA extracted from dilutions of the same titrated stock of the viruses. Ct (cycle threshold) values of each viral dilution were plotted against viral titers. The resulting standard curve had a high correlation coefficient (r² > 0.99), and it was used to convert Ct values to EID₅₀/ml.

For NDV, qRRT-PCR targeting the NDV M gene was conducted as described previously (Wise et al., 2004). The RT step was 30 min at 50 °C, followed by 15 min at 95 °C. The cycling conditions consisted of 40 cycles of 10 s of denaturation at 94 °C, 30 s of annealing at 52 °C, and extension at 72 °C for 10 s. For virus quantification, a standard curve was established with RNA extracted from dilutions of the titrated stock of the virus qRRT-PCR was conducted using AgPath-ID one-step RT-PCR Kit (Ambion) and the ABI 7500 Fast Real-Time PCR system (Applied Biosystems, Calsbad, CA).

For IBV, qRRT-PCR for IBV detection was performed as described previously (Callison et al., 2006). The reaction was conducted at 50 °C for 30 min; 95 °C for 15 min; 40 cycles of 94 °C for 1 s followed by 60 °C for 60 s. For virus quantification, a standard curve was established with RNA extracted from dilutions of the titrated stock of the virus. RRT-PCR was conducted using AgPath-ID one-step RT-PCR Kit (Ambion) and the ABI 7500 Fast Real-Time PCR system (Applied Biosystems, Calsbad, CA).

2.4. Random priming (RP)-mediated SISPA

As shown in Fig. 1, first-strand cDNA was synthesized in a 20-μl reaction mixture with 5 μl of viral nucleic acids from each sample, 100 pmol of primer K-8N (GACCATCTAGCGACCTCCACNNNNNNNN) modified from (Rosseel et al., 2013), SuperScript IV Reverse Transcriptase (ThermoFisher scientific), and dNTPs (10 μM) following the manufacturer's instructions. To convert the first-strand cDNA into double-stranded cDNA, 20 μl of the first-strand cDNA was heated to 95 °C for 3 min and then cooled to 4 °C in the presence of 10 pmol of primer K-8N, and 10 μM dNTPs in 1× Klenow reaction buffer (NEB). Afterwards, 1 uL of Klenow fragment were added and incubated at 37 °C for 60 min (final volume, 25 μl). After conversion into dsDNA by Klenow polymerase (NEB), the products were purified using Agencourt AMPure XP beads (Beckman Coulter). The purified dsDNA of the Klenow reaction was subsequently used as a template for PCR amplification.

Fig. 1.

Overview of the SISPA strategy.

Sequence-independent PCR amplification was conducted with 5 μl of the double-stranded cDNA template in a final reaction volume of 50 μl, which contained 1× Phusion HF buffer, 200 μM deoxynucleoside triphosphate (dNTP), 10 μM primer K (GACCATCTAGCGACCTCCAC), and 0.5 U Phusion DNA polymerase (NEB). The PCR cycling was performed as follows: 98 °C for 30 s, followed by 35 cycles of 98 °C for 10 s, 55 °C for 30 s, and 72 °C for 1 min, with a final extension at 72 °C for 10 min. PCR products were purified using Agencourt AMPure XP beads (Beckman Coulter). For quantification of the ds cDNA, the Qubit dsDNA HS assay (Invitrogen) was performed according to the manufacturer's instruction.

2.5. Sequence-dependent approach for AIV genome sequencing

One step RT-PCR was conducted with 5 μl and 20 μl of RNA template in a final reaction volume of 50 μl using OneTaq® One-Step RT-PCR Kit (NEB) with the primers Optil-F1, GTTACGCGCCAGCAAAAGCAGG, Optil-F2, GTTACGCGCCAGCGAAAG CAGG, and Optil-R1, GTTACGCGCCAGTAGAAACAAGG. The PCR cycling was performed as follows: 95 °C for 2 min, 42 °C for 60 min, 94 °C for 2 min, 5 cycles of 94 °C for 30 s, 44 °C for 30 s, and 68 °C for 3.5 min, followed by 26 cycles of 94 °C for 30 s, 57 °C for 30 s, 68 °C for 3.5 min with a final extension at 68 °C for 10 min (Zhou et al., 2009).

2.6. Genome sequencing

The Nextera XT DNA Sample Preparation Kit (Illumina, San Diego, CA, USA) and 0.2 ng/μl (1 ng total) of ds cDNA were used in this study to generate multiplexed paired-end sequencing libraries, according to the manufacturer's instructions. The dsDNA was fragmented and tagged with adapters by Nextera XT transposase. The Nextera XT transposome fragmented PCR amplicons with added adaptor sequences enabled a 12-cycle PCR amplification to append additional unique dual index (i7 and i5) sequences at the end of each fragmented DNA for cluster formation. PCR fragments were purified on Agencourt AMpure XP beads (Beckman Coulter). Fragments were analyzed on a High Sensitivity DNA Chip on the Bioanalyzer (Agilent Technologies) before loading on the sequencing chip. Briefly, the libraries were adjusted to 4 nM concentration and equal volumes of 5 μl of each library were pooled. The pool was denatured with NaOH (0.2 N final concentration) and further diluted to 10 pM. Control library (5% PhiX library, Illumina, USA) was added to the pool. The library pool was loaded in the flow cell of the 500 cycle MiSeq Reagent Kit v2 (Illumina, USA). The barcoded multiplexed library sequencing (2  ×  250 bp) was performed on an Illumina MiSeq platform (Illumina, USA).

2.7. Sequence analysis

The de novo and directed assembly of genome sequences were performed using the Geneious 9.1.2 (http://www.geneious.com, Kearse et al., 2012) and GALAXY software (Afgan et al., 2016). The reads were directly mapped to the reference genomes (or all segments of reference genome): A/chicken/Vietnam/NCVD-A937/2011(H5N1) (8 segments, GenBank accession no. KP097848, KP097869, KP097890, KP097925, KP097946, KP097981, KP098002, KP098023); A/turkey/Minnesota/15-12582-1/2015(H5N2) (8 segments, GenBank accession no. KX351776- KX351783), A/chicken/Israel/554/2005(H9N2) (8 segments, GenBank accession no. EF492279, EF492227, EF492377, EF492336, EF492348, EF492308, EF492406, EF492250), Infectious bronchitis virus strain Arkansas Vaccine (GenBank accession no. GQ504721.2), Infectious bronchitis virus serotype H120, complete genome (GenBank accession no. JF950510.1, Newcastle disease virus strain JSG0210 (GenBank accession no. JF340367.1), and Newcastle disease virus strain go/CH/GD-QY/1997 (GenBank accession no. KJ782375.1). All sequence sets were analyzed by the Kraken Metagenomics classifier (Wood and Salzberg, 2014) installed at Illumina BaseSpace app using MiniKraken database version 20141208 (https://basespace.illumina.com/apps).

3. Results

3.1. Whole genome sequencing of pure virus stock

An optimized SISPA protocol was tested in this study to amplify the viral genome of AIV, NDV and IBV using viral preparation from allantoic fluid. The workflow for the SISPA–NGS technique is presented in Fig. 1. This modified SISPA protocol allowed successfully assembly of sequences into full or near full-length viral genomes representing negative- and positive-sense single stranded RNA ( Table 1). The genome coverage for the viral stock with the titer at or above 10⁶ EID₅₀/ml range from 99.8% for AIV to 100% for NDV Lasota B1 vaccine strain, and IBV Ark99 (Table 1). The percentage of reads assembled to the reference genomes was greater than 40% for all the viruses tested (Fig. S1). Although the average depth differed between the viruses tested, the number of reads were high and sufficient to perform polymorphism analysis (Table 1). The mean read depth metric performed with Geneious 9.1.2. for AIVs, IBV and NDV is shown in Fig. S2 and Fig. S3, respectively.

Table 1.

Quantitation of specific virus reads and sequence coverage in virus stocks.

Virus	Virus titer (EID50/ml)	Gene segments	Coverage (%)	Mean depth of coverage	SD	Assembled reads/total paired readsa	Q30
							(%)
AIV H5N2	106	PB2	100	12,180.5	12,784.2	112,856	96.5
A/turkey/Minnesota/15–012582-1/2015		PB1	98.9b	4685.6	4207.6	46,495	96.5
		PA	100	7385.4	7727.8	79,192	87.4
		HA	100	14,846.5	12,410.4	102,233	96.4
		NP	100	1917.1	1125.9	20,574	75.9
		NA	100	13,563.2	11,913.8	92,618	94
		M	100	8656.1	6618.8	37,642	93.7
		NS	100	3800.7	2176.8	14,017	97
		Genome				505,627/ 718,922	N/A
AIV H5N1	105.9	PB2	98.1c	200.4	92.7	3007	95.8
A/duck/Vn/NCVD-672/2011		PB1	100	62.7	46.3	972	96.3
		PA	100	91.4	63.5	1296	94.9
		HA	100	1330.9	817.2	15,448	67.4
		NP	100	409.6	460.4	4233	86.7
		NA	98.7d	780.2	640.8	7302	95.7
		M	100	1378.6	1061.9	9142	95.7
		NS	100	149.9	79.6	758	95.8
		Genome				163,661/ 421,418	N/A
NDV	107	NP	100	625.6	203.9
Lasota vaccine strain		P	100	1898.2	1025.1
		M	100	4280.4	2819.7
		F	100	3513.5	1078.1
		HN	100	23,703.1	20,025.1
		L	100	6408.3	3946.8
		Genome	100	7151	10,584.4	388,022/ 847,390	94.8
	106.1	N	100	24,060	8087.3
		5b	100	9269.7	3608.6
IBV		5a	100	2272.4	990.8
Ark99		M	100	490.6	114
		3c	100	597.9	49.9
		3b	100	951.8	214.6
		3a	100	1767.1	160.7
		S	100	1281.7	958.4
		ORF1ab	100	6382.4	15,046.8
		Genome	100	6503.4	13,883.2	576,259/ 772,020	96.6

SD, standard deviation; N/A, not applicable.

^aThe numbers of assembled sequencing reads to a reference genome.

^bGaps at nucleotide position 588–613.

^cGaps at nucleotide position 1–44.

^dGaps at nucleotide position 1–17.

3.2. The genome coverage of clinical samples

Next, the modified SISPA technique was tested against clinical field samples. The genome coverage of clinical samples with virus titers greater than 10⁶ EID₅₀/ml was similar to the genome coverage obtained after sequencing the viral stock and ranged from 98.1% for AIV, 94.9–98.1% for NDVs, to 98.8% for IBV ( Table 2). Moreover, near full length NDV chicken/Jordan/15/2004 and IBV Jordan/Mass/15/2004 viral genomes were obtained from a mixed infection clinical sample (Table 2). However, the mean depth of coverage decreased as compared to the virus stock sequencing, and was very low especially for the NP gene of NDV, and NS gene of AIV (Table 2). The percentage of reads assembled to the reference genomes was below 20% for clinical samples tested which was much lower as compared to the virus stock (Fig. S1), most likely due to overwhelming competition with non-specific or host genome sequences. The distribution of the reads classified by Kraken for mixed infection clinical sample is shown in Fig. 2. Out of 304,125 total reads, 40,329 reads were classified to the domain level, 35,744 to the family level, and 35,908 to the species level. An analysis of the classified reads reveals that 52.8% of reads were classified into virus domain. Newcastle disease virus and avian coronavirus represented 19% and 8% of total reads classified to the virus domain, and 11.26% and 4.99% of total reads classified to the species level, respectively.

Table 2.

Quantitation of specific virus reads and coverage in the clinical samples.

Virus	Virus titer (EID50/ml)	Gene segment	Coverage (%)	Mean depth of coverage	SD	Assembled reads/total paired readsa	Q30
							(%)
AIV H9N2	107.7	PB2	100	349	241	4879	93.7
A/chicken/Jordan/13/2003		PB1	100	100	263.7	139.8	94.4
		PA	100	71.6	27.6	909	94.2
		HA	100	611.9	473.2	5959	94.1
		NP	100	46.2	32.7	1024	94.4
		NA	100	1092.4	981.7	9617	94.5
		M	100	545.4	425.2	3214	93.4
		NS	85.4b	3.1	2	17	90.5
		Genome				29,217/ 530,908	N/A
NDV	107.1	NP	82.1	2.7	2.0
chicken/Jordan/17/2004		P	100	21.3	13.4
		M	99.3	16.7	8.0
		F	100	33.2	15.8
		HN	100	19.5	7.2
		L	99.7	21.3	19.6
		Genome	94.9c	18.1	16.3	1378/ 534,158	92.1
NDVf chicken/Jordan/15/2004	107.2	NP	90.1	2.1	2.6
		P	100	100	34.1
		M	100	536.1	528
		F	100	48.8	17.3
		HN	100	281.6	280.6
		L	99.98	60	50.9
		Genome	98.1d	874.3	3063	114,992/ 608,250	93.4
IBVf Jordan/Mass/15/2004	106.2	N	100	176	43.9
		5b	100	100	69.1
		5a	100	45.9	6.9
		M	100	24	5.1
		3c	100	40.5	7.7
		3b	100	73.9	14.9
		3a	100	118	7.3
		S	100	68	61.7
		ORF1ab	98.5	23.7	31.7
		Genome	98.8e	2468	6728	118,763/ 608,250	98

SD, standard deviation; N/A, not applicable.

^aThe numbers of assembled sequencing reads to a reference genome.

^bGaps at nucleotide position 1–65; 603–763; 809–838.

^cGaps at nucleotide position 1–196; 1148–1237;1419–1750; 3085–3316; 15,103– 15,190.

^dGaps at nucleotide position 1–979; 14,939– 14,979.

^eGaps at nucleotide position 1–302; 15,059– 15,063; 15,113– 15,213; 27,622– 27,641.

^fMix viral infections obtained in a single reaction from clinical sample.

Fig. 2.

Taxonomic distribution of a mixed infection clinical sample of Newcastle disease virus and avian coronavirus classified by Kraken metagenomics tool.

The NDV and IBV viral sequences obtained from clinical samples were submitted to GenBank under accession numbers: NDV chicken/Jordan/17/2004, GenBank accession no. KY212126, NDV chicken/Jordan/15/2004, GenBank accession no. KY212127, IBV Jordan/Mass/15/2004, GenBank accession no. KY273667.

3.3. Limit of detection

The second goal of the study was to assess the feasibility of virus identification using SISPA–NGS and to get an estimate of its limit of detection for diagnosis of viral infections. For this purpose, two AIV isolates (Dk/Vn H5N1 and AIV A/chicken/Jordan/13/2003 (H9N2) (Ck/JO)), two NDV isolates (Dk/Vn/NDV and LaS), and one IBV isolate (Ark99) were selected. Extracts of the dilution series of viral RNA from each strain was used (Table S1). The results demonstrate that initial viral titers greater than 10^4.5 EID₅₀/ml allowed the generation from 85.5% to 99.4% of genome sequence ( Fig. 3). We were able to successfully assemble the sequences of AIV ( Table 3), NDV ( Table. 4) and IBV ( Table 5) into full or near full length of genomes. Detailed statistics on the NDV and IBV genomes assembly are provided in Supplementary Tables S2 and S3, respectively. Although the genome coverage of virus with titers below 10^3.0 EID₅₀/ml was decreased, it was still possible to detect AIV and NDV viral genomes using Kraken metagenomics approach. For example, with the AIV clinical sample (10^2.5 EID₅₀/ml), out of 105,544 total reads classified, 49.0% were classified to the domain level, 23.6% to the family level, and 24.3% to the species level. Influenza A virus represented 100% of ssRNA viruses, 23% of reads classified to the virus level and 0.6% of all reads classified. For NDV with the viral titer of 10^2.5 EID₅₀/ml out of 117,062 total reads classified, 53.9% reads were categorized into domain level, 14.7% into family level, and 19.8% into species level. NDV represented 96% of reads classified to ssRNA viruses, 4.0% of reads classified to the virus level and 0.9% of all reads classified. IBV, was not detected when the virus titer was below 10³ EID₅₀/ml. The limit of detection for IBV by the Kraken metagenomics approach used lies at 10^3.1 EID₅₀/ml. In this case, out of 34,254 reads classified, 25.6% were categorized into domain level, 22.3% into family level, and 18.8% into species level. Avian coronavirus represented 98% of positive ssRNA viruses and 2% of viruses.

Fig. 3.

Genome coverage comparison by EID₅₀/ml for the viruses tested.

Table 3.

Sensitivity of SISPA – next generation sequencing for avian influenza viruses.

Virus	Virus titer (EID50/ml)	Ct Value	Gene segment	Read mappeda	Genome coverage	Mean depth of coveragec	Contig (s) by de novo assemblyd	Detection by metagenomicse
					(%)b
AIV H5N1	104.9	26.4	PB2	898	90.8	36.8
A/duck/Vn/NCVD-672/2011			PB1	755	66.3	83.6
			PA	60	78	3.7
			HA	202	89.3	16.6
			NP	104	99.1	10.7
			NA	101	84.8	12.8
			M	96	96.9	10.3
			NS	118	79.1	19.4
			Genome		85.5		Y	Y
	103.9	30.1	PB2	39	42.1	2.1
			PB1	7	0	0
			PA	185	81.2	24.2
			HA	6	0	0
			NP	88	56.2	1.9
			NA	17	44.1	1.8
			M	0	0	0
			NS	18	39	1.4
			Genome		32.8		N	Y
	102.9	33.5	Genome		10.4		N	N
AIV H9N2	104.5	28.7	PB2	1894	100	179.5
A/chicken/Jordan/13/2003			PB1	737	97.7	74.6
			PA	269	100	29.4
			HA	1631	94.9	225.5
			NP	188	97.6	28.4
			NA	2154	92.9	358.7
			M	908	100	212.2
			NS	12	91.4	2.9
			Genome		96.8		Y	Y
	103.5	30.8	PB2	532	82.6	47.4
			PB1	5940	97.2	244.9
			PA	254	66.7	18.1
			HA	28	36.3	2.7
			NP	35	55.2	3.8
			NA	355	81.1	44.6
			M	228	57.8	40.6
			NS	2	0	0
			Genome		59.6		N	Y
	102.5	34.3	Genome		12.9		N	Y

^aThe numbers of assembled sequencing reads to a reference genome.

^bThe breadth of coverage of a target genome (%).

^cThe average number of times that each base in the reference is covered by aligned reads.

^dThe formation of contigs in de novo genome assembly.

^eDetection of AIV viral genomes using Kraken metagenomics approach.

Table 4.

Sensitivity of SISPA – next generation sequencing for Newcastle disease viruses.

Virus	Virus titer (EID50/ml)	Ct Value	Total paired reads	Read mappeda	Genome coverageb	Mean depth of coveragec	Contig (s) by de novo assemblyd	Detection by metagenomicse
NDV A/duck/Vn/Long bien/78/02	104.5	25.02	494,904	8431	97.5	82.3	Y	Y
	103.5	28.34	348,868	3636	56.8	41.1	N	Y
	102.5	31.46	344,538	1307	13.7	15.8	N	Y
	101.5	34.45	565,646	N/A	N/A	N/A	N	N
NDV Lasota	107	18.5	847,390	388,022	100	7151	Y	Y
	106	21.4	766,972	369,177	99.6	6273.2	Y	Y
	105	24.7	854,782	4623	94.5	72.1	Y	Y
	104	27.5	1,061,980	320	42.2	4.4	N	Y
	103	30.5	1,036,364	32	N/A	N/A	N	N
	102	33.7	1,028,688	N/A	N/A	N/A	N	N

N/A, not applicable.

^aThe numbers of assembled sequencing reads to a reference genome.

^bThe breadth of coverage of a target genome (%).

^cThe average number of times that each base in the reference is covered by aligned read.

^dThe formation of contigs in de novo genome assembly.

^eDetection of IBV viral genomes using Kraken metagenomics approach.

Table 5.

Sensitivity of SISPA – next generation sequencing for infectious bronchitis virus.

Virus	Virus titer (EID50/ml)	Ct Value	Total paired reads	Read mappeda	Genome coverageb	Mean depth of coveragec	Contig (s) by de novo assemblyd	Detection by metagenomicse
IBV Ark99	106.1	19.5	772,020	576,259	100	6503.4	Y	Y
	105.1	22.5	898,634	192,805	97.3	1813.1	Y	Y
	104.1	25.8	336,150	4339	73.3	29.6	Y	Y
	103.1	28.8	259,626	222	15.4	1.3	N	Y
	102.1	32.4	275,462	N/A	N/A	N/A	N	N

N/A, not applicable.

^aThe numbers of assembled sequencing reads to a reference genome.

^bThe breadth of coverage of a target genome.

^cThe average number of times that each base in the reference is covered by aligned read.

^dThe formation of contigs in de novo genome assembly.

^eDetection of IBV viral genomes using Kraken metagenomics approach.

3.4. Comparison of sequence-dependent and sequence independent approaches for AIVs

Finally, the sequence-dependent sequencing and SISPA technique were compared using AIV. For this purpose, Dk/Vn H5N1 and Ck/JO H9N2 viruses were used in the experiment (Table S4). Viral particle titers greater than 10⁶ EID₅₀/ml allowed generation of the complete genome of all 8 segments with high depth of coverage, with a range from 52,928 reads/bp for M gene to 544.9 reads for NA gene. Since there was no overwhelming competition with other non-specific sequences as observed in sequence-independent approach, the depth of coverage in sequence-dependent sequencing was higher than that of SISPA–NGS method. However, below 10⁶ viral particles per RT-PCR reaction, assembled sequences of AIVs were missing nearly 50% of one of the genes (the coverage of PB1 segment of Dk/Vn H5N1 was 54.2% and NA gene of A/chicken/Jordan/13/2003 was 52.9%). The genome coverage of each segment was substantially decreased in samples with the titer below 10⁵ viral particles per RT-PCR reaction, especially for gene longer than 1.7 kb, such as PB2, PB1, PA, and HA (Table S4). We were not able to amplify and sequence the samples with titers below 10⁴ EID₅₀/ml (data not shown), suggesting that it has lower limit of detection than SISPA-NGS method.

4. Discussion

Whole-genome sequencing provides an unsurpassed level of genetic information for virus discovery, diagnosis, characterization and genotypic classification that would not be available by using classical methodologies. Moreover, NGS technologies provide opportunities to screen for pathogens in outbreak situations from known and novel or unexpected pathogens in a single reaction. Thus, developing the simple, universal and fast amplification protocol that yields double-stranded cDNA in sufficient abundance can facilitate and expedite the whole–genome sequencing and might be a useful procedure for use not only in research but also in routine clinical diagnosis. So far, several research groups have optimized and adapted modifications of SISPA method of viral genomes and coupled it to the construction of a plasmid library generating full or near full length viral genomes (Allander et al., 2001, Djikeng et al., 2008). Palmenberg et al. (2009) have successfully applied the SISPA method to obtain the complete genome sequence of human RNA viruses, including the full length genome sequences of a total of 78 rhinovirus isolates available at the American Type Culture Collection (ATCC). In turn, Rosseel et al. (2012) applied the DNase SISPA method to detect Schmallenberg virus in infected field samples. Other groups have also used the DNase- SISPA technique for the detection of bovine and human viruses (Allander et al., 2001, Djikeng and Spiro, 2009), for the identification of a novel human coronavirus (van der Hoek et al., 2004) and for viral discovery in the plasma of HIV infected patients (Jones et al., 2005).

In this study, we present a simplified sequence-independent technique of directional amplification, in combination with illumina MiSeq NGS Platform that permitted us to generate the viral genome sequences of AIV, NDV and IBV. The main advantage of this modified SISPA protocol is its simplicity and possibility to detect multiple RNA viruses in single reaction without the need to propagate the viruses, thereby greatly reducing time and costs necessary to characterize viral genomes. It was possible to extract and amplify RNA in 5 h using modified SISPA protocol and identify the viral genomes in 3 days (including 49 h Illumina Miseq run). The protocol presented here, do not require a pre-processing step, duplication of amplification, or pooling of samples (Depew et al., 2013). We applied a denaturation step before conversion of cDNA into dsDNA by Klenow polymerase and size selection after the Klenow reaction and PCR amplification, which improved the yield of double-stranded cDNA and increased the average size of fragments. The PCR product obtained after amplification can be directly processed to generate multiplexed paired-end sequencing libraries. We have demonstrated that the method is efficient, easy to perform, and needs modest virus titers to detect and obtain complete or nearly complete genome sequence. It is predominantly useful for obtaining genome sequences from RNA viruses, uncharacterized RNA viruses, or investigating complex clinical samples (such as mixed infections in single reaction) as no prior sequence information is needed. As shown in Table 2 and Fig. 2, the SISPA method in combination with NGS can be a useful random priming method for a molecular detection of mixed viral infections for disease diagnosis in a single reaction.

Not unexpectedly, the detection limit was dependent on viral load in the samples. Based on these findings, a minimum initial virus titer of 10^4.5 EID₅₀/ml was needed to obtain the full or nearly full genome sequence as the titer of the virus correlated positively with the genome coverage. Secondly, for diagnostic detection, identification of already known agents required a minimum titer of 10³ EID₅₀/ml. We also demonstrate that non-specific bacterial or host genes in clinical samples can greatly reduce the efficiency of viral genome amplification and sequencing. Metagenomics analysis of samples that contain less than 10^3.0 EID₅₀/ml of the viruses was possible, but could vary greatly and result in low reproducibility if the sample contains a high amount of non-viral or host genes. In this case a pre-processing step in the sample preparation, such as DNase treatment and filtration, could improve the efficacy of viral amplification and sequencing.

The SISPA technique was further compared with sequence-dependent amplification for AIV (Zhou et al., 2009). The sequence-dependent AIV sequencing allowed us to generate complete genome sequence when the sample contained at least 10⁶ EID₅₀/ml of virus, and the depth of coverage was higher than that obtained from SISPA amplification. The limitation of this method is its relatively higher limit of detection and the need of primer sequences at the 5′ and 3′ ends of the genes. Comparison of sequence-dependent and sequence-independent sequencing approaches for AIV showed that both of the methods can result in complete genome sequences when the virus titer is higher than 10^5.0 EID₅₀/ml. In this case, the method of choice depends on the purpose of the study (screening, virus discovery, diagnosis, SNP analysis, etc.). Below 10^5.0 EID₅₀/ml, sequence-independent amplification can result in lower depth of coverage, but better overall coverage for large segments. In addition, both methods showed uneven distribution of sequence reads between segments. Optimization of PCR condition could be helpful to improve this limitation by extending the random oligomer sequence (Rosseel et al., 2013), using different 5′ defined tag sequences (Zhou et al., 2009), and using augmented PCR cycling for long segments.

In conclusion, NGS approaches are now leading to a more in-depth analysis of virus genetic diversity and thereby increase our overall knowledge of viral epidemiology and evolutionary dynamics what is crucial for understanding viral ecology. In this study, the simplified SISPA protocol in combination with NGS efficiently generated full length RNA viral genomes of AIV, NDV and IBV. It will be useful to detect the emergence of novel virus strains and provide information on their genetic features, enabling a better diagnosis and disease control.

Appendix A. Supplementary material

Supplementary material