情人节后送给老婆的花

LOL，job opening直接送到我的信箱了

正是中国病毒流行的时候，这个位置恰好是研究病毒和免疫系统的，还是这里的外系不认识的同事送来的。。。要不然我也不会打开看了。除了病毒和免疫我要补课，要求的试验技术我都能胜任，好多还是我的强项呢。。。就是病毒技术，似乎暗示不重要：你可以在病毒领域得到很好的训练。。。

大城市哦，又是同一个大学系统，pension还可以继续。得回家问问领导。反正明年五月以后我这里还不知道有没有funding呢。就看她舍不舍得后院的两颗桃树两颗枣树了。。。大城市房子贵，可以住apartment, 钓鱼种菜就免了。。。生活好像没有现在的乐趣多了。。。

（二0二0二月七日）

数钱——w-2表到了，今年能拿回来一千多

嘿嘿，每年小秘把税相关的表格收集齐了，就不管了，只催着填税表或追问能拿回来多少钱，够不够出去吃一顿。这不，昨天最后一张w-2到了，今天就问有多少能拿回来的。只好先毛估估：少交了一千多，儿子有两千五百的education credit，最后能拿回来1000出头。。。

今年不用追加IRA了。一来今年放进退休账户的钱根据去年的数字做了调整，保证marginal税率就是12%；二来口袋里也没钱，正忙着给儿子凑大学第三年的钱呢。。。给小秘说：别怕，万一不够，咱就找小丫头借，然后说没钱还她，看她怎么办。。。哈哈

 
今晚只请小朋友吃饭。都说他们离父母远了，不能和爹娘一起过中国年，叔叔阿姨请他们来。。。可一个武汉肺炎吓得只好取消了。想想他们都回学校三个礼拜了，该安全了。。。校园至今只有一个疑似病例，还不是大学学生，就请他们过来吃晚饭。。。可惜有一个今晚有事，只能下次等我挖了野菜包饺子，请他爹娘一起来。。。老同学朋友都不错，虽然比我挣得多多了，从不嫌弃我们穷博士后家的野菜和野生鱼。。。

做多了，做多了，剩的还可以再开一个party

呵呵，让两个能吃的小伙子带了点回去，剩下的我和领导三天不用开火了。。。烤红薯烤排骨烤虾，还有一锅剩饭和水果。。。还好牛排给吃光了。

领导说，明天不用做饭，来家坛视察，看我网上打架。。。我说我有很多花衣服，你找得出来我让你喝瓶我的啤酒。。。晚上被拷问了很久：谁是小秘？！。。。我说我是穷人啊，老婆当小秘用的。他们富人才小秘当老婆用。

（二0二0二月八日）

简单科普一下DNA和RNA

RNA和DNA分子结构上很相似的，上面带的碱基都差不到。碱基是四种，RNA里面是A，G，C，U；DNA里面是A，G，C，T。RNA里面的U在DNA里面变成了T。DNA或RNA都可以形成双链的，靠的是这些碱基配对，就是A:T或者G:C。如果出现RNA/DNA的杂交双链，那就是A:U了，但G:C还是不变。

RNA和DNA的另一个差别就是Backbone（骨架上）。骨架上的糖单元，一个多一个氧原子（RNA)，DNA比RNA少一个氧原子。因为骨架很相似，所以，我们写DNA或RNA序列时，不写骨架单元的。如写DNA，就会写成 5’GTACATTCGGAA，写RNA序列，就是5’AGUCUUCTTGUAA。。。看到U你就知道是RNA了。上面那个5’是表示序列的顺序。因为可以写成从左到右，也可以从右到左，有一个5’或3’，就不会出错了。这个很重要的：在写两条链配对的时候，一条是从5’到3’；另一条就是从3’到5’。两条DNA配对后就是这么写的：

很多人常常不写5’或3’，所以，就统一规定：不写的一律按5’到3’处理。这就是当年考研究生时的一个考题：序列AGU的配对序列是哪个：UCA还是ACU。。。

能形成双链是它们能作为遗传物质的根本需要：一条链可以作为模板复制出另一条。细胞一分为二，DNA就必须复制出第二份，才能分家。所以，DNA或RNA聚合酶就是干这事。

冠状病毒颗粒里面就只有一条链的RNA。等RNA进了宿主细胞，RNA聚合酶就先复制另一条对应的RNA。这个复制出来的RNA链是病毒RNA的配对链，不能包装成病毒，而是作为模板来产生千万条可作包装用的病毒RNA。。。

艾滋病病毒也是RNA。但它是先复制成RNA-DNA杂交双链，然后再合成DNA-DNA双链。这个DNA-DNA双链能整合到人的DNA里面，然后再合成RNA，包装成病毒。

RNA聚合酶和DNA聚合酶比较，一个特性就是总体来讲RNA聚合酶的fidelity不如DNA聚合酶好。出现这个差别的原因是RNA聚合酶只有合成的功能，就是把核苷酸按配对原则一个一个加上去。但配对出错了，它没有效正的功能。在用模板合成第二条链时，A：T/U或者G:C的配对规则遵守得越好，突变的可能就越低。对病毒来讲突变多一些是好事--突变越多，找到新宿主的可能性越多，还能逃避宿主免疫系统的追杀，有利它这个品种在多变的环境里的生存。。。病毒从一个宿主跳到另一个新宿主，就是靠这个突变来完成的。

（二0二0年二月十一日）

停药

今天是这个冬天最冷的一天，早晨-5F，从停车场走到办公室，鼻毛都冻硬了。。。没办法，今天还得早到：今天是老鼠卵巢癌模型的最后一天。昨天最后一次给药后，今天作最后的检测。本来都是中午作的，可今天动物房要调整，我们的检测仪器需要搬动。为了保持数据一直，决定在搬动前把试验作完。。。老板知道我喜欢走路去动物房，校园里走十分钟。他说：要不我明天早点来上班，先给你一个ride...

注射给药的老鼠三天七天的结果已经知道了，检测不到肿瘤；主要是口服的还得看看是不是到14天会缩小很多或完全消失：一个模型用的抗药性细胞，知道有难度；另一个七天的时候就缩小了90%。

昨天老板通知说，接受往临床推进的公司传来好消息：他们用普通乳腺癌细胞（不抗药）作的老鼠试验，每7天给一次药（注射），三个礼拜后，肿瘤手测（老式的用尺量，哈）消失。。。我们大笑：还手测呢。。。我们早就用分子标记的办法多少年了。。。我们的方式灵敏度高多了，也合适卵巢癌子宫癌等等埋藏在身体深处的肿瘤。不然就得杀了老鼠才能测量了。

公司不用我们标记的细胞，也是花钱让别的公司作的，估计也是希望第三方验证我们的结果。他们的结果倒是无意帮了我们的忙：我们一直怀疑用药后癌细胞突然大量死亡是不是可以启动身体里的免疫系统，然后对没有死亡的癌细胞发起攻击，不需要用药也能清除。。。肿瘤的老鼠模型都得用免疫系统不健全的动物，不然会排斥人的细胞，这个难题我们一直没有克服：找老鼠肿瘤细胞来替代人癌细胞的模型一直不成功；humanized老鼠有太贵。。。他们这种七天给一次药，至少是支持我们的假设的。

肿瘤的免疫方向是目前癌症研究最热门的地方。我们一直找不到切入点。给老板开玩笑：再搞点经费，我就可以在你这里干到退休了。。。

终于对退休有底了。。。

哈哈，w-2表到齐了，开始准备手填税表。明天和领导shopping的路上一起去图书馆拿说明书，从头到尾读一遍，就可以自己填了。。。

前两天就把可能拿回来的refund毛估估了一下。今天趁领导给父母打电话的时间段，整理一下去年两老的开销数字，为退休做准备。去年是儿子全年在大学生活，两老“自立”的一年，全部数字加加减减，两老去年基本生活开销是35000。。。

LOL，不错不错，后年儿子大学毕业我们就可以退休了。今天给领导说：太冷了，就不买花了，好不好。再说拿着花走到车边，几分钟花就冻死了，不如明天买打折的。

钱这样才能省出来啊。。。

（二0二0年二月十四日）

情人节后送给老婆的花

懒虫不肯起床，只好我亲自下厨。。。蛋花也是花啊。。。

下午去买下礼拜两老的吃喝。店里情人节一过，花就打折卖。Sam's的一把花巨大，还便宜一半，把领导乐开了花。我买其它的东西，她就不啰嗦了，随便。。。趁机买两版牛排，一版烟熏三纹鱼。。。酒还没喝完，下次再悄悄买。

动物实验动物房

我没在P3或P4级的实验室干过。但凭biosafety的逻辑也能理解。P3-P4的东西必须是只能进，出必须消毒处理后才可以的。就是我们P2的生物垃圾，也是需要高温高压后才能扔进普通垃圾的。但20年前确实没这么严格。

动物房分级的。这个是因为不同的动物能对付的环境不一样。我们做肿瘤模型的老鼠，免疫系统是不完整的。有完整免疫系统的老鼠会排斥人的癌细胞。因此，这些老鼠不能接触病毒细菌等，所以他们是在最干净的实验室里面。所有用具食品都是消毒处理后才能进我们的老鼠房间。

P3-P4级实验室需要的动物不应该在普通的动物房，这应该是biosafety的常识。P3-P4实验室需要动物，应该有自己特殊的动物房，在P4的设施之内。动物实验完毕，应该是无害化处理，就是高温高压消毒的。

但不要以为做烈性病毒或细菌研究的都是P4。很多研究是把病毒或细菌的的基因克隆出来，就是没有毒性的基因片段之后，给非P3-P4实验室用的。也要记住很多病毒细菌不属于P3-P4级别。我们常用的大肠杆菌是人体里面就有的；我们实验用的人工病毒是不具备繁殖能力的，所以这些P2实验室就可以作。美国有严格的biosafety分类。

 

（二0二0年二月十五日）

今年夏天的全家度假计划搞好了

根据前几年孩子们能接受的模式，女儿飞过去，我们两老开车过去，这样就省了女儿时间，也在当地有自己的车。今年和他们约好时间地点后，他们就象往年一样，把计划的事情交给老爸老妈。老妈是甩手掌柜，所以，就只能我照着google地图，找好玩的地方了。

Niagara Fall，一天
Letchworth State Park，Corning Glass Museum，一天
Watkins Glen State Park，一两个winery，一天
Robert H. Treman State Park，Buttermilk Falls State Park，Taughannock Falls State Park（就看waterfall,少走路)，一天
Cascadilla Gorge Trail，Cornell University校园，一天
Cayuga湖环湖绕一圈，看几个Winery，一天

一天机动，上面那个要没看完，或找到新的有趣活动，就保留这一天。

给儿子说好了，参观winery老爸是要喝酒的，开车的任务就交给他了。。。反正他还没到喝酒年龄，只能看着我们喝。

（二0二0年二月十五日）

今晚拿啥下酒？

走了四mile，天气不错。礼拜四还是今年冬天最冷的日子，-5F；现在已经是41F了，明天据说是50F。领导在蒸馒头，数数13个。。。酒友说，晚上喝一瓶吧。我说好。翻翻冰箱，临时也来不及准备，干脆就来个酸豆角夹馒头算了。给领导说，一个芹菜炒豆腐干，一个酸豆角炒肉末，加辣。。。

希望她明天还有馒头当早饭。。。

 

呵呵，老了，一瓶酒下肚，啥痛都没了

还说不学老美，不用cheese下红酒呢，结果，帮酒友修完了马桶，他拎着两瓶酒，一块cheese，上门来了。我家领导的13个馒头，最后剩了五个。给酒友领导说：每次我们爷们高兴喝两杯，都是你们领导们嘀咕啰嗦。。。下次把家里的马桶水龙头再搞坏几个，我们酒就可以随便喝了。。。两瓶酒，总比找人上门修理便宜吧。。。

LOL，一个教授一个博士后，居然拆了马桶，装回去还漏水。。。拆了两次装了两次，最后让酒友领导这几天注意点，漏水是不是好了。不行下次再拆了装，再找个借口喝两瓶。。。

爷们的手，都是老茧接老茧，拆了马桶就端酒杯哦。。。

（二0二0年二月十六日）

 

中药西药的差别

传统中药是凭经验来的。除了来自于自然界，也不需要分清到具体的化学成分（如分子结构），分子水平的作用机理。。。正是这种原因，现代科学尤其生物学和化学的知识没有被广泛用来提升它，很大程度上局限了它发展，也容易被人用来欺骗病人。。。简单的如箐篙素，其实中药都差不多失传了，原因是它不能用传统的煮中药方式，它在高温下会因为化学结构的改变而失去活性。。。屠呦呦先生也是试了很多提取方式，发现乙醚提取有效--乙醚可以低温挥发，保持了箐篙素的化学结构。提取成功后，后面的过程都是西医的研究方式，搞清作用的原理，也搞清了高温失活的化学原因，为进一步找类似药物提供了理论基础。

西医利用现代生物学化学知识，搞清楚人体细胞里分子之间的相互作用机制，针对不同的疾病在分子水平上进行干预达到治疗目的。虽然很多疾病还没有找到合适的药物，但前景是诱人的。很多人以为西药不会“头疼医脚”，那是错误的理解。西医还没有发展到搞清身体里的所有疾病的分子机制，或知道了机制但还没有找到合适的药物。作为科学，西医对不能治疗的疾病是承认的，如对付病毒除了疫苗抗体，目前就没有针对性的特效药。。。但发展的趋势可以从过去100年看出来。西医不仅仅靠天然的药物，也能根据天然的化合物结构化学合成和找出更好更有效的药物；不仅仅是小分子化合物，大分子的biologics，如对付过敏的针剂，肿瘤免疫疗法里的抗体等等。

中药的经验决定了它可能对一个病人有效，但对下一个有没有借鉴，这个预见性就差了。西医在一步一步进步，预见性也是从很差，到越来越好。根本原因就是从分子水平上来推理，失败了可以有根有据地找原因。给你们说一个例子：很多人听说了西药clinic trial失败的，这个不假。但同样的药物，后面的数据分析往往会发现看起来失败了的trial，可能在一个小的亚型病人里成功了。。。为什么呢？因为这个小分型是建立在新的分子基础上的认知。。。举个我熟悉的例子：乳腺癌分三个亚型，我们作的这个亚型最大，约占病人的70%。如果一个针对我们这亚型的药，有效率是50%，那么在一个所有乳腺癌病人参与clinic trail里，有效率就只有35%。。。这个逻辑如果被一个只占总病人的10%的亚型来分析：对这个亚型有50%的效果的药物，在总体病人的clinic trial数据里，就只有5%了，这很可能就在数据分析的误差范围内，会被认为失败了。。。

西医的进步就体现在分子水平的研究对病人分型会越来越细，如果每个亚型都找到对应的candidate药物，在clinic trial时就会挑选这个亚型的病人，而不是经验性质的乳腺癌/肝炎等等的所有病人。。。这就是目前的趋势。就是我们，也在考虑如果上clinic trial怎么从分子水平来给病人分型，然后挑选出合适的人选了。因为我们发现我们的candidate药物不仅仅需要我们研究的这个亚型的标志性分子，还有另外几个分子也是需要的，换句话说，就是我们研究的这个亚型可能还能细分成不同的亚亚型，以前不知道，现在知道了。。。所以，分子水平的诊断分型，然后找合适的药物，就成了未来生物医药研究的前景。。。

（二0二0年二月十七日）

 

人应该属于社会性的动物，必要的社会圈子还是要的

我这两年把自己封闭了一些，主要是家里事情多，同时钱上面因为每年要贡儿子三万多，比较紧，很多华人的活动我都选择了不去。不过，小范围的活动还是保持着的。邻居三五聚聚，老朋友来来往往，小朋友来热闹一下。。。把家里空巢的日子点缀得还不象老人院。。。

感觉被人需要是最励志的。空巢了事情不多，时不时被邻居朋友叫一下，伸个手帮过忙最有意思。其实来往的都是知根知底的，忙完了往往就多了个理由聚一起聊天喝酒喝茶。到了不敢多吃的年龄，谁也不在乎吃啥喝啥，就是social一下。。。

领导总问我退休了有啥计划。我其实还没有具体计划，得等退休的时候数着钱才能计划。搬家不搬家？啥时候退休？有没有钱留给儿女？每次讨论这个话题我俩都是在憧憬中结束：哪里不错呢，我还有一个小时候的朋友或大学同学在那里；要不咱搬到那谁住的小区去？或者等儿女定下来不走了，咱搬过去凑热闹。。。

完全孤独地离群索居，我俩不行。

 

我来解读一下美英澳三国科学家关于武汉病毒不是人造病毒的证据

1。有两个突变被怀疑是人为（一个是与病毒进入人细胞的部位有关；还有一个我记得是复旦（可能记错）那边的一位指出的，有没有比复旦这位更早的，我不清楚）。。但这两个被怀疑的突变都在其它病毒里找到了，也就是自然界存在的。

2。与病毒进入人细胞的的功能有关的突变。这个突变使病毒颗粒更容易和人细胞表面的ACE2蛋白结合，因此进入人细胞更容易。但：A。这个突变其实和以前作的一些研究的结论不完全重合，就是没有完全采纳以前有结论的突变；B。病毒重组的一些实验室常用载体的DNA痕迹不存在；C。这个突变其实不仅仅和进入人细胞有关，和另外几种动物的细胞也有关。因为那些动物的ACE2上和病毒结合的位点和人的ACE2一样。这就是说病毒可能还经过了其它动物。

3。复旦（可能记错）那边的一位指出的突变, 禽流感病毒里就有，功能还不完全清楚。一般来说，这个突变似乎与禽流感的毒性加大有关。但武汉病毒里的这个突变其实不仅仅是加了这个禽流感里面的蛋白酶的切点，还多加了一个氨基酸。这个多余的氨基酸应该是影响蛋白质三维结构的，可能与病毒表面蛋白的糖基化有关，但需要证实。也就是说，如果人工改造病毒，干吗多加这个功能还不确定的氨基酸。

4。病毒的起源推测：蝙蝠的RaTG3病毒和武汉病毒的同源性高达96%，可能性最大；马来西亚非法出口广东的Malayan pangolins携带的一种冠状病毒虽然同源性不如上面这个RaTG3高，但它带有武汉病毒里和ACE2结合需要的全部六个氨基酸，也是怀疑对象。。。

5。到底是在动物身上先突变好了再传到人，或传到人后再突变，两种可能目前没法区分。但从不同病人身上分离出来的病毒的DNA序列都很接近，表明这次的武汉病毒是同一个源头。（也就是中国武汉病毒所发表的DNA序列已经被别的实验室分离的病毒证实）

结论：目前的DNA序列比较不支持武汉病毒是人工构建的，被怀疑的突变更象天然产生和自然筛选的。

******************************************************************************

病毒的起源，包括阴谋论，我大体上分三种：

1。病毒从野生动物到人。这个不算阴谋论。它的问题是：找到的最接近的蝙蝠病毒与武汉病毒是96%同源，所以，comment sense就是蝙蝠和人中间还有一个宿主。华南农大有报道说穿山甲上找到99%的同源病毒，要被证实，这个理论会被大多数人接受。但华南农大没有后续的报道。。。找到中间宿主会强有力地支持这个理论。

2。人造病毒，然后不小心泄露或投毒。

3。天然病毒，实验室因研究需要分离保存后，不小心泄露或投毒。

我个人认为，该文的证据基本排除了上面的第二种可能。但第三种可能，就是天然病毒被实验室分离后，再人为泄露（包括有意或无意），没法排除。所以第一种和第三种可能会被长期争执，直到找到中间宿主。。。

（二0二0年二月十八日）

附原文：

The Proximal Origin of SARS-CoV-2
Kristian G. Andersen1,2*, Andrew Rambaut3, W. Ian Lipkin4, Edward C. Holmes5 & Robert F. Garry6,7

1Department of Immunology and Microbiology, The Scripps Research Institute, La Jolla, CA 92037, USA.

2Scripps Research Translational Institute, La Jolla, CA 92037, USA.

3Institute of Evolutionary Biology, University of Edinburgh, Edinburgh, UK.

4Center for Infection and Immunity, Mailman School of Public Health of Columbia University, New York, New York, USA.

5Marie Bashir Institute for Infectious Diseases and Biosecurity, School of Life and Environmental Sciences and School of Medical Sciences, The University of Sydney, Sydney, Australia.

6Tulane University, School of Medicine, Department of Microbiology and Immunology, New Orleans, LA, USA.

7Zalgen Labs, LCC, Germantown, MD, USA.

*Corresponding author:

Kristian G. Andersen
Department of Immunology and Microbiology,
The Scripps Research Institute,
La Jolla, CA 92037,
USA.

Since the first reports of a novel pneumonia (COVID-19) in Wuhan city, Hubei province, China there has been considerable discussion and uncertainty over the origin of the causative virus, SARS-CoV-2. Infections with SARS-CoV-2 are now widespread in China, with cases in every province. As of 14 February 2020, 64,473 such cases have been confirmed, with 1,384 deaths attributed to the virus. These official case numbers are likely an underestimate because of limited reporting of mild and asymptomatic cases, and the virus is clearly capable of efficient human-to-human transmission. Based on the possibility of spread to countries with weaker healthcare systems, the World Health Organization has declared the COVID-19 outbreak a Public Health Emergency of International Concern (PHEIC). There are currently neither vaccines nor specific treatments for this disease.

SARS-CoV-2 is the seventh member of the Coronaviridae known to infect humans. Three of these viruses, SARS CoV-1, MERS, and SARS-CoV-2, can cause severe disease; four, HKU1, NL63, OC43 and 229E, are associated with mild respiratory symptoms. Herein, we review what can be deduced about the origin and early evolution of SARS-CoV-2 from the comparative analysis of available genome sequence data. In particular, we offer a perspective on the notable features in the SARS-CoV-2 genome and discuss scenarios by which these features could have arisen. Importantly, this analysis provides evidence that SARS-CoV-2 is not a laboratory construct nor a purposefully manipulated virus.

The genomic comparison of both alpha- and betacoronaviruses (family Coronaviridae ) described below identifies two notable features of the SARS-CoV-2 genome: (i) based on structural modelling and early biochemical experiments, SARS-CoV-2 appears to be optimized for binding to the human ACE2 receptor; (ii) the highly variable spike (S) protein of SARS-CoV-2 has a polybasic (furin) cleavage site at the S1 and S2 boundary via the insertion of twelve nucleotides. Additionally, this event led to the acquisition of three predicted O-linked glycans around the polybasic cleavage site.

Mutations in the receptor binding domain of SARS-CoV-2
The receptor binding domain (RBD) in the spike protein of SARS-CoV and SARS-related coronaviruses is the most variable part of the virus genome. Six residues in the RBD appear to be critical for binding to the human ACE2 receptor and determining host range1. Using coordinates based on the Urbani strain of SARS-CoV, they are Y442, L472, N479, D480, T487, and Y4911. The corresponding residues in SARS-CoV-2 are L455, F486, Q493, S494, N501, and Y505. Five of these six residues are mutated in SARS-CoV-2 compared to its most closely related virus, RaTG13 sampled from a Rhinolophus affinis bat, to which it is ~96% identical2 (Figure 1a). Based on modeling1 and biochemical experiments3,4, SARS-CoV-2 seems to have an RBD that may bind with high affinity to ACE2 from human, non-human primate, ferret, pig, and cat, as well as other species with high receptor homology1. In contrast, SARS-CoV-2 may bind less efficiently to ACE2 in other species associated with SARS-like viruses, including rodents and civets1.

The phenylalanine (F) at residue 486 in the SARS-CoV-2 S protein corresponds to L472 in the SARS-CoV Urbani strain. Notably, in SARS-CoV cell culture experiments the L472 mutates to phenylalanine (L472F)5, which is predicted to be optimal for binding of the SARS-CoV RBD to the human ACE2 receptor6. However, a phenylalanine in this position is also present in several SARS-like CoVs from bats (Figure 1a). While these analyses suggest that SARS-CoV-2 may be capable of binding the human ACE2 receptor with high affinity, the interaction is not predicted to be optimal1. Additionally, several of the key residues in the RBD of SARS-CoV-2 are different to those previously described as optimal for human ACE2 receptor binding6. In contrast to these computational predictions, recent binding studies indicate that SARS-CoV-2 binds with high affinity to human ACE27. Thus the SARS-CoV-2 spike appears to be the result of selection on human or human-like ACE2 permitting another optimal binding solution to arise. This is strong evidence that SARS-CoV-2 is not the product of genetic engineering.

Polybasic cleavage site and O-linked glycans
The second notable feature of SARS-CoV-2 is a predicted polybasic cleavage site (RRAR) in the spike protein at the junction of S1 and S2, the two subunits of the spike protein (Figure 1b)8,9. In addition to two basic arginines and an alanine at the cleavage site, a leading proline is also inserted; thus, the fully inserted sequence is PRRA (Figure 1b). The strong turn created by the proline insertion is predicted to result in the addition of O-linked glycans to S673, T678, and S686 that flank the polybasic cleavage site. A polybasic cleavage site has not previously been observed in related lineage B betacoronaviruses and is a unique feature of SARS-CoV-2. Some human betacoronaviruses, including HCoV-HKU1 (lineage A), have polybasic cleavage sites, as well as predicted O-linked glycans near the S1/S2 cleavage site.

While the functional consequence of the polybasic cleavage site in SARS-CoV-2 is unknown, experiments with SARS-CoV have shown that engineering such a site at the S1/S2 junction enhances cell–cell fusion but does not affect virus entry10. Polybasic cleavage sites allow effective cleavage by furin and other proteases, and can be acquired at the junction of the two subunits of the haemagglutinin (HA) protein of avian influenza viruses in conditions that select for rapid virus replication and transmission (e.g. highly dense chicken populations). HA serves a similar function in cell-cell fusion and viral entry as the coronavirus S protein. Acquisition of a polybasic cleavage site in HA, by either insertion or recombination, converts low pathogenicity avian influenza viruses into highly pathogenic forms11-13. The acquisition of polybasic cleavage sites by the influenza virus HA has also been observed after repeated forced passage in cell culture or through animals14,15. Similarly, an avirulent isolate of Newcastle Disease virus became highly pathogenic during serial passage in chickens by incremental acquisition of a polybasic cleavage site at the junction of its fusion protein subunits16. The potential function of the three predicted O-linked glycans is less clear, but they could create a “mucin-like domain” that would shield potential epitopes or key residues on the SARS-CoV-2 spike protein. Biochemical analyses or structural studies are required to determine whether or not the predicted O-linked glycan sites are utilized.

 

figure
figure2718×1487 394 KB
 

Figure 1. (a) Mutations in contact residues of the SARS-CoV-2 spike protein. The spike protein of SARS-CoV-2 (top) was aligned against the most closely related SARS-like CoVs and SARS-CoV-1. Key residues in the spike protein that make contact to the ACE2 receptor are marked with blue boxes in both SARS-CoV-2 and the SARS-CoV Urbani strain. ( b) Acquisition of polybasic cleavage site and O-linked glycans. The polybasic cleavage site is marked in grey with the three adjacent predicted O-linked glycans in blue. Both the polybasic cleavage site and O-linked glycans are unique to SARS-CoV-2 and not previously seen in lineage B betacoronaviruses. Sequences shown are from NCBI GenBank, accession numbers MN908947, MN996532, AY278741, KY417146, MK211376. The pangolin coronavirus sequences are a consensus generated from SRR10168377 and SRR10168378 (NCBI BioProject PRJNA573298)18,19.

Theories of SARS-CoV-2 origins
It is improbable that SARS-CoV-2 emerged through laboratory manipulation of an existing SARS-related coronavirus. As noted above, the RBD of SARS-CoV-2 is optimized for human ACE2 receptor binding with an efficient binding solution different to that which would have been predicted. Further, if genetic manipulation had been performed, one would expect that one of the several reverse genetic systems available for betacoronaviruses would have been used. However, this is not the case as the genetic data shows that SARS-CoV-2 is not derived from any previously used virus backbone17. Instead, we propose two scenarios that can plausibly explain the origin of SARS-CoV-2: (i) natural selection in a non-human animal host prior to zoonotic transfer, and (ii) natural selection in humans following zoonotic transfer. We also discuss whether selection during passage in culture could have given rise to the same observed features.

Selection in an animal host. As many of the early cases of COVID-19 were linked to the Huanan seafood and wildlife market in Wuhan, it is possible that an animal source was present at this location. Given the similarity of SARS-CoV-2 to bat SARS-like CoVs, particularly RaTG13, it is plausible that bats serve as reservoir hosts for SARS-CoV-2. It is important, however, to note that previous outbreaks of betacoronaviruses in humans involved direct exposure to animals other than bats, including civets (SARS) and camels (MERS), that carry viruses that are genetically very similar to SARS-CoV-1 or MERS-CoV, respectively. By analogy, viruses closely related to SARS-Cov-2 may be circulating in one or more animal species. Initial analyses indicate that Malayan pangolins ( Manis javanica ) illegally imported into Guangdong province contain a CoV that is similar to SARS-CoV-218,19. Although the bat virus RaTG13 remains the closest relative to SARS-CoV-2 across the whole genome, the Malayan pangolin CoV is identical to SARS-CoV-2 at all six key RBD residues (Figure 1). However, no pangolin CoV has yet been identified that is sufficiently similar to SARS-CoV-2 across its entire genome to support direct human infection. In addition, the pangolin CoV does not carry a polybasic cleavage site insertion. For a precursor virus to acquire the polybasic cleavage site and mutations in the spike protein suitable for human ACE2 receptor binding, an animal host would likely have to have a high population density – to allow natural selection to proceed efficiently – and an ACE2 gene that is similar to the human orthologue. Further characterization of CoVs in pangolins and other animals that may harbour SARS-CoV-like viruses should be a public health priority.

Cryptic adaptation to humans. It is also possible that a progenitor to SARS-CoV-2 jumped from a non-human animal to humans, with the genomic features described above acquired through adaptation during subsequent human-to-human transmission. We surmise that once these adaptations were acquired (either together or in series) it would enable the outbreak to take-off, producing a sufficiently large and unusual cluster of pneumonia cases to trigger the surveillance system that ultimately detected it.

All SARS-CoV-2 genomes sequenced so far have the well adapted RBD and the polybasic cleavage site, and are thus derived from a common ancestor that had these features. The presence of an RBD in pangolins that is very similar to the one in SARS-CoV-2 means that this was likely already present in the virus that jumped to humans, even if we don’t yet have the exact non-human progenitor virus. This leaves the polybasic cleavage site insertion to occur during human-to-human transmission. Following the example of the influenza A virus HA gene, a specific insertion or recombination event is required to enable the emergence of SARS-CoV-2 as an epidemic pathogen.

Estimates of the timing of the most recent common ancestor (tMRCA) of SARS-CoV-2 using currently available genome sequence data point to virus emergence in late November to early December 201920,21, compatible with the earliest retrospectively confirmed cases22. Hence, this scenario presumes a period of unrecognised transmission in humans between the initial zoonotic transfer event and the acquisition of the polybasic cleavage site. Sufficient opportunity could occur if there had been many prior zoonotic events producing short chains of human-to-human transmission (so-called ‘stuttering chains’) over an extended period. This is essentially the situation for MERS-CoV in the Arabian Peninsula where all the human cases are the result of repeated jumps of the virus from dromedary camels, producing single infections or short chains of transmission that eventually resolve. To date, after 2,499 cases over 8 years, no human adaptation has emerged that has allowed MERS-CoV to take hold in the human population.

How could we test whether cryptic spread of SARS-CoV-2 enabled human adaptation? Metagenomic studies of banked serum samples could provide important information, but given the relatively short period of viremia it may be impossible to detect low level SARS-CoV-2 circulation in historical samples. Retrospective serological studies potentially could be informative and a few such studies have already been conducted. One found that animal importation traders had a 13% seropositivity to coronaviruses23, while another noted that 3% residents of a village in Southern China were seropositive to these viruses24. Interestingly, 200 residents of Wuhan did not show coronavirus seroreactivity. Critically, however, these studies could not have distinguished whether positive serological responses were due to a prior infection with SARS-CoV-1 or -2. Further retrospective serological studies should be conducted to determine the extent of prior human exposure to betacoronaviruses in different geographic areas, particularly using assays that can distinguish among multiple betacoronaviruses.

Selection during passage. Basic research involving passage of bat SARS-like coronaviruses in cell culture and/or animal models have been ongoing in BSL-2 for many years in multiple laboratories across the world25-28. There are also documented instances of the laboratory acquisition of SARS-CoV-1 by laboratory personnel working under BSL-2 containment29,30. We must therefore consider the possibility of a deliberate or inadvertent release of SARS-CoV-2. In theory, it is possible that SARS-CoV-2 acquired the observed RBD mutations site during adaptation to passage in cell culture, as has been observed in studies with SARS-CoV5 as well as MERS-CoV31. However, the acquisition of the polybasic cleavage site or O-linked glycans - if functional - argues against this scenario. New polybasic cleavage sites have only been observed after prolonged passaging of low pathogenicity avian influenza virus in cell culture or animals. Furthermore, the generation of SARS-CoV-2 by cell culture or animal passage would have required prior isolation of a progenitor virus with a very high genetic similarity. Subsequent generation of a polybasic cleavage site would have then required an intense program of passage in cell culture or animals with ACE-2 receptor similar to humans (e.g. ferrets). It is also questionable whether generation of the O-linked glycans would have occurred on cell culture passage, as such mutations typically suggest the involvement of an immune system, that is not present in vitro .

Conclusions
In the midst of the global COVID-19 public health emergency it is reasonable to wonder why the origins of the epidemic matter. A detailed understanding of how an animal virus jumped species boundaries to infect humans so productively will help in the prevention of future zoonotic events. For example, if SARS-CoV-2 pre-adapted in another animal species then we are at risk of future re-emergence events even if the current epidemic is controlled. In contrast, if the adaptive process we describe occurred in humans, then even if we have repeated zoonotic transfers they are unlikely to take-off unless the same series of mutations occurs. In addition, identifying the closest animal relatives of SARS-CoV-2 will greatly assist studies of virus function. Indeed, the availability of the RaTG13 bat sequence facilitated the comparative genomic analysis performed here, helping to reveal the key mutations in the RBD as well as the polybasic cleavage site insertion.

The genomic features described here may in part explain the infectiousness and transmissibility of SARS-CoV-2 in humans. Although genomic evidence does not support the idea that SARS-CoV-2 is a laboratory construct, it is currently impossible to prove or disprove the other theories of its origin described here, and it is unclear whether future data will help resolve this issue. Identifying the immediate non-human animal source and obtaining virus sequences from it would be the most definitive way of revealing virus origins. In addition, it would be helpful to obtain more genetic and functional data about the virus, including experimental studies of receptor binding and the role of the polybasic cleavage site and predicted O-linked glycans. The identification of a potential intermediate host of SARS-CoV-2, as well as the sequencing of very early cases including those not connected to the Wuhan market, would similarly be highly informative. Irrespective of how SARS-CoV-2 originated, the ongoing surveillance of pneumonia in humans and other animals is clearly of utmost importance.

Acknowledgements
We thank all those who have contributed SARS-CoV-2 genome sequences to the GISAID database (https://www.gisaid.org/ 25) and contributed analyses and ideas to Virological.org 16 (http://virological.org/ 4). We thank the Wellcome Trust for supporting this work. ECH is supported by an ARC Australian Laureate Fellowship (FL170100022). KGA is supported by NIH grant 1U19AI135995-01. AR is supported by the Wellcome Trust (Collaborators Award 206298/Z/17/Z – ARTIC network) and the European Research Council (grant agreement no. 725422 – ReservoirDOCS).