B型肝炎病毒在不同生物尺度下演化的模式與改變速度

Abstract

許多研究指出B型肝炎病毒(HBV)的分歧度與分化時間無顯著相關。若是經由母子垂直感染得到慢性B型肝炎的患者，會有一持久的免疫耐受期(immune tolerance phase)。我們提出此時期的免疫耐受性(immune tolerance)可能是造成分歧度與分化時間無顯著相關性的原因。免疫清除期所觀察到大部分HBV的遺傳變異不會傳至下一代，或在下一代免疫耐受期時被具高複製能力的病毒株所競爭淘汰。綜合以上所述，在一定時間內，病患間HBV的遺傳距離會和垂直感染的次數呈負相關。換言之，垂直感染次數越多，則病毒歧異度越低。為了驗證「重置假說」，我藉由一個三代八口、垂直感染B型肝炎的帶原家族進行研究。每一位病患的檢體都利用聚合酶連鎖反應增幅出HBV全長序列進行克隆。克隆完成後透過PCR 確認，並選取12個正確植入的克隆進行定序。我先推估出HBV在宿主間兩兩間平均核甘酸分歧度(average pairwise nucleotide divergence),並利用線性回歸比較其與分化時間(time of separation)以及母子垂直感染的次數(numbers of mother-to-infant transmission, NMITs)的相關性。結果顯示核苷酸分歧度與分化時間結果顯示兩者間無顯著相關，但卻與NMITs有顯著負相關(R=0.49 , P=0.004)。進一步分析顯示負相關主來自於氨基酸變換而非沉默取代 (R=0.33, P=0.072)。我們也藉由次世代定序的方法(NGS)以求更精準地解釋HBV演化的動態。我選取12支檢體建庫，並利用Illumina-Solexa平台進行HBV全基因組定序。NGS資料庫的分析也顯示核甘酸分歧度與分化時間之間不具有顯著相關性，但卻與NMITs有顯著負相關 (R=0.52, P=10-7)。若將HBV分為抗原決定位(epitope)與非抗原決定位(non-epitope)兩個區域，可發現HBV基因組全長、抗原決定位區域以及非抗原決定位區域皆與NMITs具有顯著負相關性；然而，僅有非抗原決定位區域與分化時間具有顯著正相關。該結果顯示非抗原決定位區域上多是中性突變(neutral mutation)，與NMITs間的負相關是由於遺傳上「搭便車」(genetic hitchhiking)的現象。相反的，免疫汰選較強的抗原決定位區域上產生的突變乃是造因於免疫廓清期的免疫汰選。接著，我分別計算兩組祖母-母親-孫女支系(lineage)上固定的核苷酸差異(fixed nucleotide difference)數量。與聚合酶比較，發現位在表面抗原上抗原決定位區域內的氨基酸變換較顯著。除此之外，我也在兩組支系上發現反轉突變(reversed mutation)與大量的新突變(de novo mutation)。病毒在不同宿主間對免疫辨識產生快速的適應性，顯示宿主的免疫系統在慢性HBV感染具有一很強的汰選力。我們認為抗原決定位區域與NMITs具有顯著負相關性可以支持我們所提出的重置假說，亦即，經過母子垂直感染後，病毒株會重置為與野生型較相近的序列。因為大部分的遺傳變異在母子垂直感染後會消失，因此大量的新突變並對長期的HBV演化歷程造成的影響十分微弱。

The lack of correlation between time of separation and divergence of hepatitis B virus (HBV) has been reported in many previous studies. I hypothesize that this lack of correlation is partly due to limited host immune response to HBV during immune tolerance phase if acquired perinatally. Most variation observed at relative late stage of infection, e.g. after immune clearance phase, within HBV donors would not be transmitted or be outcompeted by viral strains with high replicative ability in recipients during immune tolerance phase. In other words, after mother-to-infant transmission, the viral quasi-species will “reset” to wild-type like sequences and most of mutations accumulated in previous host are lost. The “reset” hypothesis predicts that, for a given time, genetic distance of HBV between individuals would be negatively correlated with the number of mother-to-infant transmission. To test “reset” hypothesis, I selected a family of three generations including eight individuals acquired hepatitis B virus (HBV) infection via perinatal transmission. For each patient, serial serum samples were collected and full-length HBV sequences were recovered and cloned into vectors. 10-12 colonies were picked for PCR check and subjected to sequencing. I first estimated average pairwise evolutionary distances between hosts and then correlated it to time of divergence and number of mother-to infant transmissions (NMITs) by linear regression. Interestingly, I found the correlation between evolutionary distance and time of separation is very weak but significantly negative to NMITs. Further analysis revealed this negative correlation is mainly due to amino-acid alternation instead of silent changes (R=0.33, P=0.072). I also applied next generation sequencing (NGS) technology to study dynamics of HBV evolution with better resolution. I constructed 12 libraries and whole-genome sequencing was carried out by Illumina GAIIx paired-end platform. Analysis based on NGS datasets revealed lack of correlation between nucleotide divergence and time of separation but showed significantly negative correlation with NMITs (R=0.52, P<10-7). Partition of HBV genome showed this negative correlation with NMITs was found not only in full alignment, but also in epitope and non-epitope regions. However, only non-epitope regions are positively correlated to time of divergence, suggesting mutations occurred in this region is neutral or near-neutral and negative correlation with transmission is through genetic hitchhiking. On the other hand, epitope region is under strong immune selection during immune clearance phase. Numbers of fixed nucleotide differences along two grandmother-mother-granddaughter lineages were calculated. Compared to polymerase, I found significantly more amino-acid alternations occurred in epitope regions on surface protein. Besides, we found several reversed mutations in both lineages and a large amount of de novo mutations (7 in a total19 reversed mutations). These rapid adaptations to immune recognition in different hosts suggest host immune acts as a strong selection force during the course of chronic HBV infection. I found a negative correlation between epitope region and NMITs support my reset hypothesis: Viral quasi-species reset to wild-type sequences after mother-to-infant transmission. Consequently, high proportion of de novo mutations will not contribute to long-term HBV evolution because most of genetic variations are lost after transmission.