内容来源：https://www.quantamagazine.org/the-ancient-weapons-active-in-your-immune-system-today-20260415/

内容总结：

古老武器在今日免疫系统中依然活跃：细菌与病毒的亿万年军备竞赛塑造人类防线

在自然界永无休止的进化军备竞赛中，一场持续数十亿年的激战正在微观世界上演：细菌与感染它们的病毒（噬菌体）彼此攻防，不断升级武器库。科学家近期发现，这场古老战争留下的“遗产”至今仍深植于人类细胞之中，构成了我们抵御病原体的第一道免疫防线的核心部件。

过去十年间，多项突破性研究揭示，人类先天免疫系统的关键组件竟与细菌的抗病毒“武器”同源，且功能机制高度保守。哈佛医学院微生物学家菲利普·克兰祖施（Philip Kranzusch）表示：“宿主与病毒相互作用的规则在数十亿年间未曾改变，这令人难以置信。”他的研究团队最早发现人类免疫核心蛋白cGAS-STING在细菌中亦存在类似物，该成果使其荣获2026年美国国家科学院分子生物学奖。

两大发现浪潮引爆领域
这一领域的突破始于两波关键发现。2018年，以色列魏茨曼科学研究所的罗腾·索雷克（Rotem Sorek）团队通过计算预测与实验筛选，在细菌基因组“防御岛”区域发现了数百种全新的抗噬菌体系统。几乎同时，2019年起的研究浪潮证实，其中部分细菌防御机制同样存在于动植物细胞（包括人类细胞）中，且工作原理相同。

“这两股趋势的结合真正引爆了该领域：大量新型噬菌体防御系统被发现，且它们与人类免疫直接相关，能从机制上为我们揭示免疫奥秘。”克兰祖施解释道。

从细菌到人类：cGAS-STING通路的跨越
2013年，生物化学家陈志坚（James Chen）团队阐明了人类细胞感知异常胞质DNA并启动免疫应答的cGAS-STING通路，该成果于2024年获拉斯克奖。然而，该通路在人类中的进化起源一度成谜。

克兰祖施与合作者通过结构生物学研究发现，人类cGAS蛋白与细菌中功能类似的酶在DNA序列上已无相似性，但二者产生信号分子cGAMP的关键三维结构却亿万年保持不变。“结构是功能的关键。蛋白质结构不能变，但序列可以改变以应对病毒的干扰。”现任职于科罗拉多大学博尔德分校的亚伦·怀特利（Aaron Whiteley）指出。

后续研究进一步证实，细菌中的cGAS和STING同源蛋白同样执行抗病毒防御功能，且机制与人类细胞如出一辙。“所有线索终于串联起来，”克兰祖施说，“这打破了‘免疫蛋白不应古老’的旧有观念。”

攻防博弈与医学启示
细菌与噬菌体间的攻防博弈精巧而激烈。例如，为对抗细菌的cGAS-STING系统，噬菌体会进化出“分子海绵”吸收信号分子；细菌则相应演化出“潘诺普特斯”（Panoptes）系统释放诱饵信号，误导噬菌体并确保真实警报送达。类似地，人类免疫蛋白viperin、gasdermin等均被发现存在于细菌中，并采用相同作用机制。

这些发现开辟了探索人类先天免疫的新路径。索雷克指出：“在细菌中开展研究远比在人类中容易。我们工作的重要影响之一，正是能够利用细菌来研究高等生物的免疫机制。”目前，科学家已开始利用已知的数百种细菌免疫系统反向预测人类等真核生物中未知的防御机制，并成功在珊瑚、昆虫等动物基因组中找到功能相同的细菌来源抗病毒蛋白。

进化熔炉与基因“借用”
细菌群体犹如一个高速进化的“分子武器熔炉”，不断锻造出新防御系统。尽管已发现数百种系统，但单个细菌通常只携带约十余种，因其具有成本且可能引发意外自我毁灭。有趣的是，一些在细菌中相对罕见的机制（如cGAS-STING）却在真核生物中成为核心武器，而CRISPR等常见系统却未被人类继承。

进化生物学家尤金·库宁（Eugene Koonin）认为，这或许与基因组织结构有关：细菌的防御基因常成簇排列（操纵子），易于整体水平转移；而真核生物基因组结构复杂，操纵子易被破坏且难以重新获取。此外，部分细菌防御组件在进入真核生物后可能演化出新功能，如某些限制性内切酶变身为表观遗传修饰的关键酶。

更近期的进化事件显示，真核生物持续从细菌和噬菌体“借用”基因。例如，数百万年前，果蝇从内共生细菌的噬菌体中获得了一个毒素基因，并将其转化为对抗寄生蜂的武器。加州大学伯克利分校生物学家诺亚·怀特曼（Noah Whiteman）指出：“从进化迅速的生物那里借用基因是一种高效策略，这比自行演化工具快得多。”

未来展望
如今，科学家已清晰认识到，细菌群落是一个持续锻造新型分子武器的进化熔炉，其中蕴藏着成百上千种防御系统。从单细胞真核生物到动植物，不同谱系在进化史上反复借用、改造并丢失这些武器。在这口永恒沸腾的熔炉中，从微生物到动植物的所有进化战争规则，仍在不断书写。正如克兰祖施所言：“前方仍有大量工作待完成，而这正是如此多实验室投身其中的原因。”这些古老而精妙的免疫机制，不仅揭示了生命防御的深层历史，更为未来新型医学疗法和生物技术工具的开发提供了广阔前景。

中文翻译：

今日活跃在你免疫系统中的远古武器

引言

进化军备竞赛在自然界中无处不在——两个物种相互对抗，各自为取得优势而不断演化出新的或更精良的武器。其中最古老、最激烈的战斗之一，已在细菌与感染它们的病毒之间持续了数十亿年。这场不断升级的战争，筛选出了能设计新方法入侵细菌细胞的噬菌体病毒（或称“噬菌体”），反过来也筛选出了能设计新方法抵御噬菌体的细菌。在试图智取对方的过程中，每个物种都会尝试一切手段以保持领先一步。

近年来，研究人员有了一个惊人的发现：细菌用来防御噬菌体的一些“机器”，几乎原封不动地存在于我们自己的细胞中。根据过去十年间的数十项发现，细胞与病毒之间的交战规则早在数十亿年前就已确立，并且至今仍在很大程度上定义着我们应对感染的“第一响应者”——先天免疫系统——如何保护我们免受病毒和细菌的侵害。

“看到宿主与病毒相互作用的规则在数十亿年间保持不变，这真是一件令人难以消化的事情，”哈佛医学院的微生物学家菲利普·克兰祖施说。他是最早发现人类免疫系统的一个关键组成部分也存在于细菌中的研究人员之一。

他说，与我们这些长寿的多细胞真核生物相比，病毒的进化速度“高得离谱”。确实，为了跟上微生物的快速进化步伐，动物已经进化出了诸如抗体这样的针对性防御机制，能在新型病毒入侵时适应它们。然而，奇怪的是，我们的前线免疫防御似乎与细菌共享许多抗病毒工具。“为什么这些规则会如此固定不变？”克兰祖施问道。

最近的两波发现浪潮打开了这个领域的大门。首先，在2018年，研究人员报告了多种新颖的细菌抗病毒防御系统，如今其数量已达数百种。第二波浪潮始于2019年左右，研究表明其中一些细菌机制存在于植物和动物细胞中，包括我们自己的细胞——而且它们的工作方式与在那些远古祖先中一样。

“这两件事结合在一起，才真正让这个领域爆发式发展：有很多新的噬菌体防御系统，而且，嘿，这些系统直接相关、意义重大，并且从机制上能告诉我们关于人类免疫的信息，”克兰祖施说。他因在该领域的研究，将于2026年4月26日在美国国家科学院第163届年会上获得分子生物学奖。

这些以及随后的其他发现，揭示了一片尚未被探索的人类先天免疫图景——这片图景可能带来新的医疗方法和生物技术工具，就像细菌免疫系统CRISPR-Cas的发现使基因编辑成为可能一样。

“有很多工作要做，而且做这些工作真的非常重要，”克兰祖施说，“这就是为什么这么多实验室都投身于此。”

防御岛

直到过去十年之前，生物学家只知道细菌防御病毒的两种方式。在20世纪50年代，他们发现了限制性内切酶修饰系统，这是细菌的一种蛋白质，能在特定序列处切割入侵病毒的DNA。这些酶最终成为实验室的主力工具，推动了基因工程的革命。随后，在21世纪初，研究人员描述了CRISPR系统，它也能识别并切割特定的DNA序列，如今被广泛用于基因编辑。

大约在CRISPR被发现的同时，微生物免疫学家罗特姆·索雷克正在劳伦斯伯克利国家实验室做博士后，研究细菌基因组学。“我立刻想到，这是我遇到过的最有趣的生物学现象之一，”他回忆道。然而，他也清楚地意识到，大多数细菌并没有CRISPR系统。那么，他想知道，它们是如何对抗噬菌体的呢？

因为限制性内切酶修饰系统和CRISPR都催生了重要的生物技术工具，“我认为如果我们发现更多的免疫机制，可能会很有用，”索雷克说，“这就是动机的来源。”

几年后，在以色列魏茨曼科学研究所建立了自己的实验室后，索雷克和他的团队观察到，包括限制性内切酶修饰系统和CRISPR阵列在内的大型免疫基因群，往往聚集在细菌基因组的同一区域。他和其他实验室观察到，这些“防御岛”或“基因组岛”内功能未知的基因，可能代表着新的抗噬菌体机制。

为了验证这一点，索雷克的团队开发了一种计算流程来发现更多的防御岛。然后，他通过实验将每个预测的防御系统与多种噬菌体进行对抗，在培养皿中设置细菌与噬菌体的对决。2018年，他的团队证明，这些防御岛中的许多未知基因确实起到了多种抗噬菌体防御系统的作用。

通过这种计算预测加实验筛选的方法，几个研究团队发现了数百种新的先天免疫系统：一场名副其实的“噬菌体防御大爆发”，克兰祖施说。（就在本月，2026年4月，两个研究团队使用机器学习识别了数十万个候选抗噬菌体防御系统，并且已经验证了其中的几十个。）

“这太令人兴奋了，因为它表明存在着一大片无人知晓的新型防御系统领域，”斯坦福大学的计算生物学家亚历克斯·高说，他在数千个预测的细菌防御系统中，实验验证了29个具有抗噬菌体活性。“它就隐藏在我们眼前这一大堆数据之下。”

不久之后，人类先天免疫中一个关键通路的发现，将与这场噬菌体防御机制的爆发相碰撞，揭示出植物和动物免疫中一片广阔、未被探索的领域。

STING循环

一个细胞如何知道自己被病毒入侵了？一个迹象是细胞质中出现了不该存在的DNA——这是感染或严重压力的信号。

2013年，生物化学家陈志坚（James Chen）及其在德克萨斯大学西南医学中心的团队发现了人类细胞在检测到这种错位DNA时如何产生先天免疫反应。当cGAS酶感知到细胞质中的DNA时，它会产生一种叫做环状GMP-AMP（cGAMP）的次级分子。然后，cGAMP激活STING蛋白，从而引发一系列信号级联反应，最终开启免疫基因。这些免疫基因启动炎症反应——这是抵御入侵者的第一道防线。

STING蛋白及其激活炎症的下游过程早在2008年就被描述了。但当时没人知道STING是如何被激活的。陈志坚的工作——描述了一条名为cGAS-STING的新的人类先天免疫通路——填补了这一空白。2024年，陈志坚因这一突破获得了拉斯克奖。

“陈的论文非常出色——堪称即时经典，”克兰祖施说，“它让其他一切都变得合理了。”

几乎是所有。唯一不太合理的一点是，这个关键的cGAS-STING机制是如何在人类中进化出来的。“很难理解这是怎么发生的，”克兰祖施说。

为了探究其进化历史，他在其他物种中寻找能产生cGAMP的其他蛋白质。当他找到一种——一种能制造cGAMP的细菌酶，由哈佛医学院约翰·梅卡拉诺斯实验室于2012年描述——当时在加州大学伯克利分校做博士后的克兰祖施致力于理解其分子结构。

这种细菌蛋白与人类cGAS的DNA序列并不相同。因此，当他发现这两种蛋白质的形状和结构几乎一模一样时，他感到非常惊讶。“我一直记得那一天，”他回忆道，“因为我跑进我导师的办公室说，‘你不会相信这个的。’”

2016年，当克兰祖施在哈佛医学院建立自己的实验室时，他和梅卡拉诺斯，以及博士后亚伦·怀特利和本·莫尔豪斯，继续进行研究。他们对多种细菌cGAS样酶（都能产生cGAMP或相关的二核苷酸信号）进行的精细结构研究表明，这些酶看起来也像人类cGAS。他们在2019年和2020年的一系列论文中发表了他们的发现。

这些发现非常了不起。经过数十亿年的进化，人类cGAS及其细菌对应物在DNA甚至氨基酸序列水平上已无法识别出亲缘关系。怀特利（现在在科罗拉多大学博尔德分校领导自己的实验室）说，在一个长达300个氨基酸的序列中，细菌和人类的cGAS蛋白只有五到六个氨基酸相同。然而，产生二核苷酸cGAMP信号的那部分蛋白质的形状，在这漫长的时间里结构上一直保持不变。

“结构，这是功能的关键，”怀特利说，“蛋白质的结构不能改变。但它可以改变序列，以试图破坏病毒试图对抗该系统的所有方式。”

结构相似性的发现已经足够令人惊讶了。但这些细菌酶，是否像人类cGAS一样，能防御外来病毒呢？是的。同年，索雷克的团队证明，细菌cGAS酶确实起到了抗噬菌体防御的作用。

两个团队最终也在细菌中发现了STING。当他们确认细菌STING在免疫防御中的功能与人类STING非常相似时，“所有的点真的开始连接起来了，”克兰祖施说，“然后我们就有了完整的图景。”

这项工作中最困难的部分既不是解析蛋白质结构，也不是测试它们的功能。“该领域的教条是免疫蛋白不应该很古老。而cGAS-STING通路最难克服的就是这一点，”克兰祖施说。

“我们知道它是保守的，执行着相同的功能，并且在数十亿年的进化中得以保留，”他继续说，“随着数据变得压倒性，这打破了该领域所有其他发现的障碍。”

角色阵容

自从发现细菌与人类cGAS-STING之间不可思议的对应关系以来，对细菌防御岛的计算分析已经预测了数百种不同的先天免疫机制。有些机制切割病毒DNA或RNA转录本来杀死病毒；另一些则终止新病毒DNA的复制。“相当多的这些防御系统结果都是自杀性的，”美国国立卫生研究院的进化生物学家尤金·库宁说，他在2011年发表了一篇关于防御岛的基础性论文。也就是说，它们导致细胞自我毁灭，从而防止病毒在群体中进一步传播。

但数十亿年来，噬菌体已经进化出巧妙的应对措施来规避此类防御。例如，针对细菌的cGAS-STING，噬菌体会部署一些分子来“海绵式吸收”连接传感器（cGAS）和效应器（STING）的环状二核苷酸信号（cGAMP）。这有效地短路并克服了防御。

作为反击，细菌进化出一种名为Panoptes的机制——由怀特利、莫尔豪斯及其同事在2025年首次描述——它持续产生与cGAS产生的信号相似但不完全相同的cGAMP信号。入侵的噬菌体于是吸收了这些cGAMP诱饵，使得真正的信号能够到达其目标（STING）并触发细胞自我毁灭。

这个伎俩之所以有效，仅仅是因为cGAS和Panoptes产生的二核苷酸足够不同，细胞能够区分它们，同时又足够相似，以至于噬菌体无法分辨。这是一种危险的平衡——很可能经常出错。

“这就是细菌的神奇之处，”怀特利说，“它们复制得如此之快，以至于可以尝试很多无效的方法，只为找到那少数有效的方法。”

另一个令人着迷的攻防例子可以在一种消耗NAD的细菌防御机制中找到。NAD是一种必需的辅因子，作为电子载体，每秒钟都在润滑细胞中数百万生化反应的齿轮。通过快速摧毁所有细胞内的NAD，细菌使生化反应停滞，阻止病毒复制。但索雷克的团队发现，噬菌体也不甘示弱，已经进化出重建NAD并逃避这种细菌防御的方法。

还有viperin，一种人类蛋白质，能制造修饰过的核苷酸，迅速终止病毒复制；其作用机制在2018年被破译。不久之后，曾在索雷克实验室做博士后、现在巴黎巴斯德研究所领导自己研究小组的奥德·伯恩海姆，在细菌中发现了viperin的同源物。她还证明它们的工作方式与人类viperin相同。

Gasdermins是存在于细胞质中的免疫蛋白，当感知到感染时，通过在细胞膜上穿孔来杀死细胞。人类gasdermins的机制在2015年被描述。2022年，曾在索雷克实验室接受博士后培训、现在魏茨曼研究所领导自己研究小组的塔尼塔·韦因发现，gasdermins在细菌中的作用方式与在人类中相同。

研究人员最初通过将现有的人类免疫知识映射到细菌基因组上，在该领域取得了进展。现在他们正在做相反的事情：研究这数百种新的细菌免疫系统是否可以用来预测人类和其他真核生物中仍然未知的机制。

这种预测框架已经结出果实。例如，索雷克的团队发现了一种细菌蛋白质，在感知到感染时，会耗尽细胞中的ATP（分子能量），从而阻止病毒复制。他们后来在动物（包括珊瑚和昆虫，但非人类）的基因组中发现了它。在活体组织中进行测试时，珊瑚和昆虫的蛋白质工作方式与在细菌中完全相同。

“这是一个非常有力的启示，因为在细菌中进行研究比在人类中进行研究要容易得多，”索雷克说，“我们研究最重要的影响之一，就是这种利用细菌来研究高等生物免疫的能力。”

进化熔炉

尽管有数百种细菌防御系统，但大多数细菌只有大约十几种。它们本身代价高昂，并且总是带有意外触发自我毁灭的风险，这限制了任何一个细菌所能拥有的数量。有些机制，如限制性内切酶修饰系统，很常见，而另一些则相对罕见。CRISPR存在于大约40%的原核生物基因组中；viperins仅存在于大约0.5%中。

库宁说，原核生物中许多最常见的免疫机制并未被真核生物继承，而一些相对罕见的机制却被继承并“蓬勃发展”。问题是：为什么？为什么我们的细胞没有CRISPR？而为什么cGAS-STING这种在细菌中相对罕见的免疫机制，却成为了我们武器库中的核心工具？

在某些情况下，细菌防御机制可能在大约20亿年前被真核生物获得，当时一个古菌细胞首次吞噬了一个细菌细胞——后者最终定居下来成为线粒体细胞器——并播下了真核生物谱系的种子。其他机制可能后来通过水平基因转移获得，这是细菌常用的一种交换DNA片段的方式，在真核生物中发生频率较低。

“获得本身并不是什么大问题，”库宁说，“（罕见的防御机制）如何以及为何取代了最常见的原核生物防御机制——这更有趣，当然，也不完全清楚。”

一种可能性是，在细菌中，防御机制通常由几个遗传成分组成，这些成分组织成小的阵列或操纵子，并一起被调控。限制性内切酶修饰系统、毒素-抗毒素基因、CRISPR和Cas、cGAS和STING——每一个都是由在细菌基因组中彼此相邻的基因组成的系统。这使得细菌很容易共享整个工具包。

但在真核生物中，由于我们的基因组组织和调控方式更为复杂，基因的操纵子组织常常被打乱。“一旦你失去了一个操纵子，你就不太可能通过水平转移重新获得它，”库宁说，“一旦被打乱，它就实际上消失了。”

当匹兹堡大学的泰拉·莱文和她实验室的博士后爱德华·卡尔伯森调查了广泛真核生物中的cGAS和STING蛋白时，他们发现其中一个或另一个缺失的情况相当普遍。这就引出了一个问题：当另一个不存在时，剩下的那一个在做什么？

“这些组件要么不存在，要么不是以我预期的组合存在——它们可能在做什么？”莱文说，“这是我们在该领域这部分反复遇到的问题。”

有可能这些组件进化出了全新的功能。例如，十多年前，进化生物学家L. Aravind表明，一些限制性内切酶修饰系统成为了真核生物表观遗传学中的必需酶，它们在染色质的特定位置放置（或移除）表观遗传标记。2025年，他的团队表明，动物发育中必需的信号分子Wnt蛋白，也起源于细菌冲突系统。

这样，细菌就成为一种“创客空间”，用于加速的进化实验和创新，产生的新颖性随后被播撒到整个生命界。这种播撒不仅仅是遥远的过去：在更近的进化时期，真核生物一直在继续向细菌和噬菌体借用、获取甚至窃取。

几百万年前，野生果蝇 Drosophila ananassae 水平获得了一个能切割DNA的病毒毒素基因，很可能来自其内共生细菌的噬菌体。2025年，伯克利的生物学家诺亚·怀特曼和他的团队发现，果蝇使用这种噬菌体毒素不是为了杀死病毒，而是为了毒杀在果蝇幼虫体内产卵的寄生蜂。令人惊奇的是，果蝇设法运用这种强大的新武器，却没有毒害自己的细胞。

与可以轻松与邻居交换基因并以极快速度进化的细菌不同，多细胞生物受限于有性生殖——这意味着进化以较慢的速度进行。通过利用细菌和病毒武器的快速进化，昆虫和其他真核生物可以加速它们自身的进化过程，以领先敌人一步。

“当你从另一个进化非常迅速的生物那里借用一个基因时，这是一个很好的策略，因为你可以以比你制造它们更快的速度借用工具，”怀特曼说。

现在很清楚，细菌群落就像一个熔炉，新的分子武器在其中被锻造并不断进化。这个微生物武器库包含数百种，也许数千种不同的防御系统。

从单细胞真核生物到植物和动物的多样化真核生物谱系，在进化过程中反复地从这个武器库中借用、改造和丢失防御机制。在这个熔炉中，所有进化战争——微生物的、动物的、植物的——规则仍在继续书写。

英文来源：

The Ancient Weapons Active in Your Immune System Today
Introduction
Evolutionary arms races — where one species is pitted against another, driving the evolution of new or more sophisticated weapons as each tries to gain the upper hand — are ubiquitous in nature. One of the oldest and fiercest battles has been waged for billions of years between bacteria and the viruses that infect them. This escalating warfare has selected for bacteriophage viruses (or “phages”) that devise new ways to invade bacterial cells and, in turn, for bacteria that devise new ways to fend phages off. In their attempts to outmaneuver one another, each species will try anything to stay one step ahead.
In recent years researchers have come upon a surprising finding: Some of the machinery that bacteria use to defend against phages exists, almost unchanged, in our own cells. According to dozens of discoveries made over the past decade, the rules of engagement between cells and viruses were written billions of years ago and still largely define how our innate immune system, the first responder to infection, defends us against viruses and bacteria today.
“Seeing that the rules of host-virus interactions are unchanged over billions of years is a really hard thing to digest,” said Philip Kranzusch, a microbiologist at Harvard Medical School who was one of the first researchers to discover that a key component of human immunity also exists in bacteria.
The pace of evolution for viruses is “insanely high,” he said, compared to that of long-lived, multicellular eukaryotes like ourselves. Indeed, to keep up with the rapid pace of microbial evolution, animals have evolved targeted defenses such as antibodies that adapt to novel viruses as they invade. And yet, strangely enough, our frontline immune defenses seem to share many antiviral tools with bacteria. “Why would the rules be so fixed?” Kranzusch said.
Two recent waves of discovery broke this field open. First, in 2018, researchers reported a variety of novel bacterial defense systems against viruses, which now number in the hundreds. The second wave, starting around 2019, showed that some of these bacterial mechanisms exist in plant and animal cells, including our own — and that they still work the same way they did in those distant ancestors.
“Those two things combining together is what really made the field explode: There’s lots of new phage defense systems, and, hey, those systems are directly related and relevant and mechanistically going to tell us information about human immunity,” said Kranzusch, who will receive the National Academy of Sciences Award in Molecular Biology for his research in the field at their 163rd annual meeting on April 26, 2026.
These and other discoveries that followed reveal an unexplored landscape of human innate immunity — one that could lead to new medical treatments and biotechnological tools, much as the discovery of the bacterial immune system CRISPR-Cas made genome editing possible.
“There’s a lot to do, and it’s really important to do it,” Kranzusch said, “which is why so many labs have jumped in on it.”
Defense Islands
Until the last decade, biologists knew of only two ways that bacteria defend against viruses. In the 1950s, they discovered restriction-modification enzymes, which are bacterial proteins that cut DNA from invading viruses at specific sequences. These enzymes eventually became laboratory workhorses that enabled the revolution in genetic engineering. Then, in the early 2000s, researchers described CRISPR systems, which also recognize and cut specific DNA sequences and are now widely used for genome editing.
Around the time CRISPR was being discovered, Rotem Sorek, a microbial immunologist, was working on bacterial genomics as a postdoc at Lawrence Berkeley National Lab. “I immediately thought that that’s one of the most interesting things in biology I ever bumped into,” he recalled. However, it also became clear to him that most bacteria don’t have CRISPR. In that case, he wondered, how do they fight phages?
Because both restriction modification enzymes and CRISPR had led to important biotechnological tools, “I thought that if we discovered more immune mechanisms, it might be useful,” Sorek said. “That was the source of the motivation.”
A few years later, after he established his own lab at the Weizmann Institute of Science in Israel, Sorek and his team observed that big constellations of immune genes, including restriction-modification enzymes and CRISPR arrays, tended to cluster together in the same region of bacterial genomes. He and other labs observed that genes of unknown function within these “defense islands” or “genomic islands” could potentially represent novel anti-phage mechanisms.
To find out, Sorek’s team developed a computational pipeline to discover more defense islands. Then he experimentally pitted each predicted defense system against a variety of phages, setting up bacteria-versus-phage duels in petri dishes. In 2018, his team showed that many of the unknown genes in these defense islands did, in fact, function as a variety of anti-phage defense systems.
Through this approach of computational prediction followed by experimental screening, several research teams discovered hundreds of new systems of innate immunity: a veritable “explosion of phage defense,” Kranzusch said. (Just this month, in April 2026, two research teams used machine learning to identify hundreds of thousands of candidate anti-phage defense systems, and they have already validated dozens of them.)
“It was so exciting because it suggested that there was just this big landscape of new types of defense systems that no one knew about,” said Alex Gao, a computational biologist at Stanford University who has experimentally verified anti-phage activity in 29 out of thousands of predicted bacterial defenses. “It was hiding under our noses in this big vat of data.”
Soon, the discovery of a critical pathway in human innate immunity would collide with this explosion of phage defense mechanisms, revealing a vast, unexplored territory in plant and animal immunity.
STING Cycle
How can a cell tell that it has been invaded by a virus? One sign is the presence of DNA where it doesn’t belong in the cytoplasm — a sign of infection or severe stress.
In 2013, the biochemist Zhijian “James” Chen and his team at the University of Texas Southwestern Medical Center discovered how human cells generate an innate immune response when they detect this kind of misplaced DNA. When the enzyme cGAS senses DNA in the cytoplasm, it produces a secondary molecule called cyclic GMP-AMP (cGAMP). Then cGAMP activates the STING protein, which sets off a signaling cascade that ultimately turns on immune genes. These immune genes kick off an inflammatory response — the first line of defense against an invader.
The STING protein, and the downstream process by which it activates inflammation, were described back in 2008. But no one knew how STING was activated. Chen’s work — describing a new human innate immune pathway called cGAS-STING — filled in the gap. In 2024, Chen received the Lasker Award for this breakthrough.
“Chen’s paper was brilliant — an instant classic,” Kranzusch said. “It made everything else make sense.”
Almost everything. One thing that didn’t quite make sense was how the essential cGAS-STING mechanism could have evolved in humans. “It’s hard to understand how that would come to be,” Kranzusch said.
To dig into its evolutionary history, he looked for other proteins that produce cGAMP in other species. When he found one — a bacterial enzyme that makes cGAMP, described by John Mekalanos’ lab at Harvard Medical School in 2012 — Kranzusch, who at the time was a postdoctoral fellow at the University of California, Berkeley, worked to understand its molecular structure.
The bacterial protein did not share a DNA sequence with the human cGAS. So it was a huge surprise when he found that the two proteins were virtually identical in shape and structure. “I always remember that day,” he recalled, “because I ran into my adviser’s office and I’m like, ‘You’re not going to believe this.’”
In 2016, when Kranzusch launched his own lab at Harvard Medical School, he and Mekalanos, along with postdoctoral fellows Aaron Whiteley and Ben Morehouse, continued the search. Their careful structural work on a variety of bacterial cGAS-like enzymes, all of which produce cGAMP or related dinucleotide signals, showed that these enzymes look just like human cGAS too. They published their findings in a series of papers in 2019 and 2020.
The findings were remarkable. After billions of years of evolution, human cGAS and its bacterial equivalents were not recognizably related at the level of DNA or even amino acid sequence. Out of a 300-amino-acid-long sequence, the bacterial and human cGAS proteins share only five or six amino acids, said Whiteley, who now runs his own lab at the University of Colorado, Boulder. Yet the shape of the part of the protein that produces the dinucleotide cGAMP signal has remained structurally unchanged for all that time.
“The structure, that’s key to function,” Whiteley said. “The protein can’t change in structure. But it can change in sequence to try and disrupt all the ways the virus is trying to antagonize the system.”
The discovery of the structural similarity was surprise enough. But did these bacterial enzymes, like human cGAS, protect against foreign viruses? Yes. That same year, Sorek’s team showed that the bacterial cGAS enzymes did in fact work as an anti-phage defense.
Both teams eventually found STING in bacteria as well. When they confirmed that the bacterial STING functioned in immune defense much like the human STING, “all the dots really started to connect,” Kranzusch said. “Then we had everything.”
The hardest part of this work was neither solving the protein structures nor testing their functions. “The dogma in the field was that immune proteins should not be old. And that was the really hard part to overcome with the cGAS-STING pathway,” Kranzusch said.
“We knew it was conserved, and it was doing the same function, and it was maintained across billions of years of evolution,” he continued. “As the data became overwhelming, that broke down the barriers for all sorts of other findings in the field.”
Cast of Characters
Since the discovery of the uncanny correspondence between bacterial and human cGAS-STING, computational analysis of bacterial defense islands has predicted hundreds of distinct mechanisms of innate immunity. Some of the mechanisms cleave viral DNA or RNA transcripts to kill the viruses; others terminate the reproduction of new viral DNA. “Quite a few of these defense systems turn out to be suicidal,” said Eugene Koonin, an evolutionary biologist at the National Institutes of Health who published a foundational paper on defense islands in 2011. That is, they cause the cell to self-destruct, thereby preventing further spread in the viral population.
But over billions of years, phages have evolved ingenious countermoves to evade such defenses. For example, in response to bacterial cGAS-STING, phages deploy molecules that sponge up the cyclic dinucleotide signals (cGAMP) that connect the sensor (cGAS) to the effector (STING). This effectively short-circuits and overcomes the defense.
Countering that, bacteria evolved a mechanism called Panoptes — first described in 2025 by Whiteley, Morehouse, and their colleagues — which constantly generates cGAMP signals that are similar but not identical to those generated by cGAS. An invading phage then sponges up the cGAMP decoys, allowing the true signal to reach its target (STING) and trigger cellular self-destruction.
This trick works only because the cGAS and Panoptes dinucleotides are different enough for the cell to distinguish them and similar enough that the phage can’t tell the difference. It’s a dangerous balance — one that probably frequently misfires.
“This is the marvel of bacteria,” Whiteley said. “They’re replicating so rapidly that they can try a lot of things that don’t work in order to find the very few that do.”
Another example of fascinating moves and countermoves can be found in a bacterial defense mechanism that depletes NAD, an essential cofactor. NAD is an electron carrier that, every second, greases the wheels of millions of biochemical reactions in the cell. By quickly destroying all cellular NAD, bacteria grind biochemical reactions to a halt, preventing viral replication. But phages, not to be outdone, have evolved ways to reconstitute NAD and evade this bacterial defense, Sorek’s team has found.
Then there’s viperin, a human protein that makes modified nucleotides that quickly terminate viral replication; its mechanism of action was deciphered in 2018. Soon after, Aude Bernheim, who was a postdoctoral fellow in Sorek’s lab and now leads her own research group at the Pasteur Institute in Paris, found homologues of viperin in bacteria. She also showed that they work the same way human viperin does.
Gasdermins are immune proteins present in the cytosol that kill the cell when it senses an infection by piercing a hole in the cell membrane. The mechanism for gasdermins in humans was described in 2015. In 2022, Tanita Wein, who trained as a postdoc in the Sorek lab and now leads her own research group at the Weizmann Institute, discovered that gasdermins work the same way in bacteria as they do in humans.
Researchers initially made headway in the field by mapping existing knowledge of human immunity onto bacterial genomes. Now they are doing the opposite: investigating whether the hundreds of new bacterial immune systems can be used to predict still unknown mechanisms in humans and other eukaryotes.
This predictive framework has already borne fruit. For example, Sorek’s team discovered a bacterial protein that, upon sensing infection, depletes ATP (molecular energy) from the cell, thus preventing the virus from replicating. They later found it in the genomes of animals, including corals and insects (though not humans). When tested in living tissues, the coral and insect proteins worked the exact same way as they do in bacteria.
“This is a very strong revelation, because doing research in bacteria is much easier than doing research in humans,” Sorek said. “One of the most important influences of our research is this ability to use bacteria to study higher organism immunity.”
An Evolutionary Cauldron
Although there are hundreds of bacterial defense systems, most bacteria have only about a dozen. They are inherently costly and always carry the risk of triggering accidental self-destruction, which limits how many any single bacterium can have. Some mechanisms, like restriction-modification enzymes, are common, while others are relatively rare. CRISPR exists in about 40% of prokaryotic genomes; viperins are present in only about 0.5%.
Many of the most common immune mechanisms in prokaryotes have not been inherited by eukaryotes, while some relatively rare ones have been inherited and have “flourished,” Koonin said. The question is: Why? Why don’t our cells have CRISPR? And why did cGAS-STING, a relatively rare immune mechanism in bacteria, become such a central tool in our arsenal?
In some cases, bacterial defense mechanisms could have been acquired by eukaryotes about 2 billion years ago, when an archaeal cell first engulfed a bacterial one — which eventually settled in as a mitochondrion organelle — and seeded the eukaryotic lineage. Other mechanisms may have been acquired later through horizontal gene transfer, a mechanism commonly used by bacteria to swap chunks of DNA, which occurs with less frequency in eukaryotes.
The acquisition “in itself is not such a big problem,” Koonin said. “How and why [rare defenses] replaced the most common prokaryotic defenses — that is more intriguing and, of course, not entirely clear.”
One possibility is that in bacteria, defenses often come with several genetic components organized in small arrays, or operons, that are regulated together. Restriction-modification enzymes, toxin-antitoxin genes, CRISPR and Cas, cGAS and STING — each is a system made of genes that sit next to each other in bacterial genomes. This makes it easy for bacteria to share the entire toolkit.
But in eukaryotes, because of the more complicated way our genomes are organized and regulated, genes’ operon organization is often disrupted. “Once you lose an operon, you are very unlikely to reacquire it by horizontal transfer,” Koonin said. “Once it is disrupted, it is effectively gone.”
When Tera Levin and Edward Culbertson, a postdoctoral fellow in her lab at the University of Pittsburgh, surveyed cGAS and STING proteins across a broad swath of eukaryotes, they found that it’s quite common to find one or the other missing. That begs the question of what one piece is doing when the other isn’t there.
“These components either aren’t there, or aren’t there in the combinations I expected — what are they possibly doing?” Levin said. “That is a question we encounter over and over again in this part of the field.”
It’s possible that the pieces evolved entirely new functions. For example, more than a decade ago, the evolutionary biologist L. Aravind showed that some restriction-modification enzymes became essential enzymes in eukaryotic epigenetics, where they place (or remove) epigenetic marks at specific locations on the chromatin. In 2025, his team showed that Wnt proteins, essential signaling molecules in animal development, also originate in bacterial conflict systems.
In this way, bacteria serve as a kind of “maker space” for accelerated evolutionary experimentation and innovation, generating novelty that then is seeded across life. This seeding is not just a thing of the distant past: Eukaryotes have continued to beg, borrow, and steal from bacteria and phages in more recent evolutionary time.
Several million years ago, the wild fruit fly Drosophila ananassae horizontally acquired a viral toxin gene that cuts DNA, probably from phages of endosymbiotic bacteria. In 2025, Noah Whiteman, a biologist at Berkeley, and his team discovered that the fruit fly uses this phage toxin not to kill viruses but rather to poison parasitic wasps that lay eggs inside fly larvae. Amazingly, the fly manages to wield this powerful new weapon without poisoning its own cells.
Unlike bacteria that can easily swap genes with their neighbors and evolve at warp speed, multicellular organisms are stuck with sexual reproduction — which means that evolution proceeds at a slower pace. By taking advantage of the rapid evolution of bacterial and viral weapons, insects and other eukaryotes can accelerate their own evolutionary process to stay a step ahead of their enemies.
“When you borrow a gene from another organism that’s evolving very rapidly, that’s a great strategy, because you can borrow tools at a faster rate than you could ever make them,” Whiteman said.
It’s now clear that bacterial communities are like a cauldron where new molecular weapons are forged and are always evolving. This microbial arsenal contains hundreds, perhaps thousands, of different defense systems.
Diverse eukaryotic lineages, from unicellular eukaryotes to plants and animals, have repeatedly borrowed, adapted, and lost defenses from this arsenal over evolutionary time. In this cauldron, the rules of all evolutionary warfare — microbial, animal, vegetal — continue to be written.