建立禽流感病毒株變異群的次世代定序分析流程-備妥未來病毒偵測與流行潛力趨增之前瞻預警

Establish an Analysis Framework for Detecting Sub-populations of Avian Influenza Virus Variants by Next Generation Sequencing for Better Preparedness and Early Warning of Viral Variants with Increasing Epidemic Potential through Virological Surveillance

周柏豪(Po-Hao Chou)

指導教授 ： 金傳春

國立臺灣大學/公共衛生學院/流行病學與預防醫學研究所/碩士(2016年)

https://doi.org/10.6342/NTU201601031

摘要

禽流行性感冒（簡稱禽流感）病毒防治已視為重要議題。許多嚴重的禽流感疫情在台灣家禽養殖場爆發，不僅造成鉅大經濟損失，也加劇暴露民眾感染禽流感病毒的風險。事實上，以傳統桑格定序法(Sanger sequencing)所建立的禽流感病毒監測系統在台灣已實施多年，依然無法完善地盡早預警與防治禽流感疫情。近年，次世代定序(next generation sequencing, NGS) - 此新穎的DNA定序方法已被廣用於許多病毒的偵測，若啟用於偵測禽流感病毒，將能比傳統桑格定序法提供更多不同次病毒群的些微序列變化資訊；運用此新穎法至田野流行病學所得禽流感病毒，並研發次世代定序後的大量序列數據之一系列分析流程，遂顯得格外重要。因此，本研究的目標為: (1)研發並建立次世代定序分析流程法，以檢測與評估禽流感病毒的細微核甘酸與胺基酸突變；(2)在混合H5與H6不同亞型的禽流感病毒中，以此新分析流程法探尋其可能具有流行病學重要性的特殊病毒核甘酸變異模式，如找出兩不同次病毒群含有致病分子標幟(即基因重組的前兆)；(3)評估其未來應用於禽流感病毒偵測系統的適用性；及(4)進行公共衛生風險評估。

針對第一、二目標，做法上是使用2013年台灣家鴨的三禽流感病毒株，各被鑑定為H5N2(DV10955病毒株)、H6N1(DV11011病毒株)和H6N1與H5N2兩亞型混合(DV11018病毒株，並非混合DV11011與 DV10955)，以進行次世代定序，並研究分析產出的高通量病毒序列資料。為了評估檢測禽流感病毒中細微的核甘酸與胺基酸突變與兩不同次病毒群是否含有不同亞型禽流感病毒特殊位點的核甘酸變異百分比；先以統計軟體R研發新分析流程法，包括排序、移除定序實驗中曾用引子(primers)與聚合酵素鏈反應(polymerase chain reaction, PCR)可能造成的偏差干擾後，一方面分析禽流感病毒每位點的核甘酸與胺基酸突變百分比；另方面也假定次世代定序中產生的定序誤差發生次數符合卜瓦松分佈(Poisson distribution)，並利用卜瓦松累積分布函數檢定此次世代定序過程可能產生的定序誤差而致其影響最小化。最後，仔細深入分析各病毒核苷酸與胺基酸位點間的差異在11種不同的病毒蛋白之變化，並比較此三株不同亞型病毒在次病毒群的核甘酸變異於混合型與單一亞型之異處。此外，為了以次世代定序建立禽流感病毒偵測系統的預警機制並評估其在公共衛生的風險，特別將2013年三株鴨流感病毒中細微的胺基酸突變和過去相同亞型的病毒株(18株H5N2, 52株H6N1)比較，找出過去未曾在台灣出現過的胺基酸突變，而後將這些胺基酸突變和2015年禽流感大疫情63株台灣雞地方性流行H5N2禽流感病毒與clade 2.3.4.4 H5病毒 (H5N2、H5N8、H5N3)的胺基酸進行比較，期望找到潛藏在2013年病毒株中為過去未見的細微胺基酸突變卻又出現在引發2015年台灣禽流感大疫情的病毒序列而有預警性。在公共衛生風險評估上，是依據美國疾管與預防中心提供可能會提高禽流感病毒在哺乳類感染或致病力的66個胺基酸突變，期能於病毒檢體中尋找曾以弱勢存在卻能提高病毒對於哺乳類感染或致病力的胺基酸突變。

研究結果顯示三株鴨流感核苷酸位點遺傳變異百分比的平均值均小於4%。然而，H5N2 DV10955病毒株核甘酸與胺基酸變異的95百分位數卻高於其他兩株病毒；顯示此H5N2 DV10955病毒株可能具有較高的基因突變潛能與引發流行的風險。另方面，混合亞型的禽流感病毒一向被認為是其基因重組的前兆，研究中發現以次世代定序新分析法也可偵測並辨認混合亞型禽流感病毒特殊位點的核苷酸變異百分比。另自此三株病毒中，發現PB1和PB1-F2兩病毒基因之核苷酸變異百分比明顯高於病毒的其他基因外，在病毒各位點的核甘酸比較上，也發現此三株病毒基因均在PB1的核甘酸位點100到1300區域有較高的變異。另外，共發現36個基因突變不僅以弱勢[0.3%-14.6%]存在於2013年此三株病毒，另也出現在引發2015年台灣禽流感大疫情的病毒序列中[1%-90.6%]，且此且此三株2013年鴨流感病毒中具有三個共同的細微突變，如M2的胺基酸位點13上天冬醯胺(Asparagine)突變為羥丁胺酸(threonine)。同時，也在病毒中HA切割位上發現一些已證實會在感染雞隻後具較高致死力(lethality)的細微突變 [谷胺酸(Glutamate)突變為離胺酸(lysine)]，促使演化為高致病性病毒。

公共衛生風險評估的結果顯示，2013年的臺灣三株鴨流感病毒(H5N2 DV10955、H6N1 DV11011、H6N1與H5N2兩亞型混合DV11018)中確實存在一些已證實會增加病毒對哺乳類致病力的細微胺基酸突變，包括：(1) PB2位點627分別發現了0.28%、0.13% 與0.18%的賴胺酸(lysine)突變，(2) PB2位點701各有0.31%、0.51%及0.13%的天冬醯胺(Asparagine)突變；此外兩H5鴨病毒在HA 222位點各有0.05%及0%的白胺酸(Leucine)突變，與224位點各有0.06%及0.35%的絲胺酸(Serine)突變。最重要的是這些結果證實可以藉由次世代定序協助預測病毒進化的未來走向與下一次可能發生禽流感疫情的風險，同時也可以結合相關的禽流感病毒研究成果進行公共衛生風險評估。

綜言之，本研究致力研發一套合適分析禽流感病毒次世代定序資料的方法，並以新角度研究台灣禽流感病毒的核甘酸與胺基酸變異。結果顯示可以次世代定序法:追蹤台灣禽場的禽流感病毒間或許具有共同的微觀核甘酸與胺基酸變異；並檢視特殊位點、高致病性及公共衛生風險相關胺基酸的隨時間變化，來評估禽流感病毒增加流行潛力之分子標幟，以提早執行防治措施：更重要的是分析2003年12月台灣雞流感H5N2病毒在無菌雞胚傳代以模擬流行情形，也發現除了HA切割位之外，另伴隨PB1較其他7段基因有最多的核甘酸變異，顯示不僅HA還須考慮其他基因在雞流感病毒的致病角色。未來希望以此新開發的方法分析更多台灣田野的禽流感病毒次世代定序序列，整合微觀的分子變異與巨觀的病毒變異群群體變異，將徹底明瞭禽流感病毒的動態演變趨勢，藉以推動最佳防治策略。

關鍵字

禽流感病毒 ； 病毒監測 ； 次世代定序 ； 病毒基因變異分析 ； 流行潛力趨增 ； 臺灣

並列摘要

Prevention of avian influenza viruses (AIVs) has become an important issue in recent years. Several severe outbreaks of AIVs in poultry farms of Taiwan leading to tremendous economic loss and increasing human risk of acquiring infection with AIVs. In fact, virological surveillance system has relied on traditional method using Sanger sequencing, but it has still struggled to guide early warning for successful preventing outbreaks of AIVs. Recently, next-generation sequencing (NGS), a novel method of DNA sequencing, has been widely applied to virological surveillance for many viruses, especially providing more information on viral subtitle sequence variations of different sub-populations of AIVs than Sanger sequencing. Applying this novel approach to AIV isolates obtained from field epidemiology and developing a series of workflow to analyze massive NGS sequence data becomes crucially important. Therefore, the specific aims of this study were: (1) to establish a new analysis method of NGS data for monitoring minor mutations of nucleotides and amino acids that might be possibly present in AIVs, (2) to identify special patterns in variations of nucleotides with epidemiological significance that would be existed in mixed or single subtypes AIVs, thus providing early clues for viral genetic reassortment or recombination; (3) to evaluate its application to future surveillance system of AIVs after best optimization of NGS approach; and (4) to search for possibilities for public health risk assessment using viral population data from NGS.

For the 1st and 2nd aims, the three duck isolates of AIVs obtained from 9- to 11-day embryonated chicken eggs through virological surveillance at a live-poultry market in Taiwan in 2013 were used for getting NGS data. They were identified as H5N2 (ID#DV10955), H6N1 (ID#DV11011), and mixed subtypes of H5N2 and H6N1 (ID#DV11018, not mixing DV11011 with DV10955), respectively. To evaluate minor variations of nucleotides and amino acids of AIVs as well as changing patterns in minor mutations of nucleotides in mixed subtypes of AIVs; a new analysis workflow [such as mapping the sequencing reads, removing the bias caused by the primers and polymerase chain reaction (PCR) process, and analyzing the percentage of variations in nucleotides and amino acids at each position of AIV] was firstly developed by the R software. Subsequently, a statistical test assuming the frequencies of sequencing errors in NGS process following the Poisson distribution and thus using its cumulative distribution function to examine and minimize inaccuracy of sequencing errors existed in the NGS data. Finally, similarities and differences in variations in nucleotides and amino acids of each of the 11 viral proteins of the three studied duck AIVs were analyzed. For the 3rd and 4th aims, to establish an early warning mechanism for AIVs surveillance system using the NGS approach, the minor mutations of amino acids in the 2013 Taiwan three duck influenza virus were compared with their same subtype of AIVs before 2013 (18 strains of H5N2, 52 strains of H6N1) to find out the novel mutations of amino acids that had not appeared in the past Taiwan AIV. These novel mutations of amino acids identified in our studied the three AIVs in 2013 were then compared with the 63 strains of enzootic H5N2 AIVs and clades 2.3.4.4 H5 AIVs (H5N2, H5N8, H5N3) to search for these minor mutations that had never been present in the past but were hidden in the AIVs in 2013 did also occur in strains of AIV that caused large outbreaks in 2015 with better predictability for early warning. Public health risk assessment was based on the 66 mutations of amino acids that may increase the virulence and infectivity of mammalian hosts listed by the Centers for Disease Control and Prevention in the United States (US-CDC), to find out whether the minor mutations of amino acids with increasing virulence and infectivity to mammalian might be present in the studied 2013 AIVs.

The results showed that the means of nucleotide variations in the three AIV isolates were lower than 4%. However, the 95th percentile of nucleotide variations of Dk-H5N2-#DV10955 was higher than other two AIV isolates, indicating that certain positions of this H5N2 virus contained higher variations of nucleotide. It implies that the H5N2 had higher potential in genetic variations and risk for leading to outbreaks. On the other hand, the NGS data was capable to differentiate the mixed subtypes of H5N2 and H6N1 from single subtype of AIV by examining the nucleotide variations at specific positions. Such mixed subtypes have been thought with potentiality of gene reassortment or genetic recombination resulting in a novel influenza virus with higher pathogenicity is an important virological surveillance target of AIVs. Among the nucleotide variations of the tested AIVs’ 11 viral proteins, PB1 and PB1-F2 (encoded from PB1 gene) ranked the highest, particularly in the common regions at positions of 100-1300 nucleotides of PB1 presented with highly genetic variations than other viral genes were identified in all the three tested AIV isolates. In addition, the 36 minor mutations of amino acids that were not only hiddenly present in the three 2013 studied AIVs [0.3%-14.6%] but also can be identified in the 2015 Taiwan AIVs [1%-90.6%] during outbreaks. Furthermore, three common minor mutations were found in the 2013 three duck influenza viruses, including a mutation from asparagine (D) to threonine (T) at position 13 of M2. Besides, the minor mutations - the lysine (K) at the HA cleavage site were observed in the two H5 related studied viruses exhibiting greater lethality in chickens.

Public health risk assessment shows that the minor mutations of amino acids with higher infectivity and pathogenicity to mammals indeed did exist in the 2013 Taiwan three duck influenza virus (H5N2 DV10955, H6N1 DV11011, H6N1 and H5N2 mixing two subtypes DV11018), including: (1) 0.28%, 0.13% and 0.18% mutations of lysine (K) at position 627 of the PB2; (2) 0.31%, 0.51% and 0.13% the mutation of asparagine (N) at position 701 of the PB2. In addition, the two duck H5 viruses had 0.05% and 0% mutations of leucine (L) at position 222 of the HA, and 0.06% and 0.35% mutations of serine (S) at position 224 of the HA, respectively. Most importantly, these results supported that the NGS can help not only to predict further viral evolution of AIVs and the risk for next possible outbreak, but also to assess public health risk after integrating with related scientific results on molecular signatures of AIVs.

In conclusion, this study develop a framework to analyze the NGS data of AIVs and to provide a novel approach to study the genetic variations of AIVs obtained from surveillance and field epidemiology. Our results suggested that NGS approach can: (1) detect the common minor variations being present in different AIV isolates at the micro level in Taiwan field, (2) provide information on minor mutations of amino acids with epidemiological significance that might increase viral pathogenicity and epidemic potentiality of AIVs for early prevention; and (3) offer scientific evidence on changing of those important amino acids over time for better public health risk assessment. Most importantly, serial passages of Taiwan 2003 chicken H5N2 LPAIVs in specific pathogen free eggs found that amino acid changes at the HA cleavage site accompanied with the highest diversities of PB1 nucleotides than other gene segments, implying the significant role of PB1 in addition to HA in viral pathogenicity in chicken influenza viruses. Future studies need to increase more field specimens for NGS data of AIVs. To fully understand the dynamic changes and evolutionary trend in AIVs, obtaining minor mutations at micro-level as well as viral quasispecies population changes at macro-level can certainly help formulate the best prevention and control strategies of AIVs.

並列關鍵字

Avian Influenza Virus ； Virological surveillance ； Next Generation Sequencing ； Genetic variation analysis ； Increasing epidemic potential ； Taiwan

參考文獻

Chang-Chun, Lee. 2006. Molecular Epidemiology of Avian Influenza Virus in A Live Bird Market in Taiwan during 2005-2006, National Taiwan University.

Aly, M. M., A. Arafa, and M. K. Hassan. 2008. "Epidemiological findings of outbreaks of disease caused by highly pathogenic H5N1 avian influenza virus in poultry in Egypt during 2006." Avian Dis no. 52 (2):269-77. doi: 10.1637/8166-103007-Reg.1.

Aw, P. P., P. F. de Sessions, A. Wilm, L. T. Hoang, N. Nagarajan, O. M. Sessions, and M. L. Hibberd. 2014. "Next-generation whole genome sequencing of dengue virus." Methods Mol Biol no. 1138:175-95. doi: 10.1007/978-1-4939-0348-1_12.

Barrero, P. R., M. Viegas, L. E. Valinotto, and A. S. Mistchenko. 2011. "Genetic and phylogenetic analyses of influenza A H1N1pdm virus in Buenos Aires, Argentina." J Virol no. 85 (2):1058-66. doi: 10.1128/JVI.00936-10.

Bennett, S. 2004. "Solexa Ltd." Pharmacogenomics no. 5 (4):433-8. doi: 10.1517/14622416.5.4.433.