耶鲁大学教授25年终圆梦：一种“格格不入”的新型疗法成功上市，“撞开”创新药的想象空间 | 对话Craig Crews教授

2026年5月，美国FDA批准了首款蛋白降解靶向嵌合体疗法，标志着一种全新药物类型正式问世。这类药物所基于的理念，曾一度被认为过于激进：药物不再只是抑制致病蛋白的活性，而是可以将其从细胞中选择性清除。耶鲁大学（Yale University）Craig Crews教授，是这一科学故事中的关键人物之一。20多年前，正是他的实验室开创了蛋白降解靶向嵌合体这一技术路线。

在职业生涯中，Crews教授多次跨越学术界与生物技术产业的边界，推动科学突破转化为创新疗法。2003年，他共同创立Proteolix公司，该公司开发的蛋白酶体抑制剂Kyprolis后来获得美国FDA批准，用于治疗多发性骨髓瘤。此后，他又共同创建Arvinas，推动蛋白降解靶向嵌合体疗法从早期科学概念走向临床开发，并最终见证了这一领域迎来首个获批药物。他创立的第三家公司Halda Therapeutics于2025年12月被强生（Johnson & Johnson）收购，该公司专注于开发一种基于诱导接近机制的新疗法——调节诱导接近靶向嵌合体（Regulated Induced Proximity Targeting Chimeras）。

在首款蛋白降解靶向嵌合体疗法迎来获批上市这一历史性里程碑后，Crews教授接受了药明康德的采访。访谈中，他回顾了这一理念的起源，以及该领域早期面对的质疑；也分享了推动靶向蛋白降解成为可行药物模式的关键科学拐点，并解释了他为什么相信诱导接近技术可能从根本上重塑医学的未来。

Crews教授，非常感谢您接受药明康德的采访。也再次祝贺首款蛋白降解靶向嵌合体疗法获批。当您听到这一消息时，脑海中首先浮现的是什么？

Craig Crews教授：对我来说，这一批准代表的意义，远不止一款药物的上市，也不止是一种治疗模式的诞生。我至今仍然记得上世纪90年代，人们第一次得以窥见“生命天书”时的那种兴奋感。几乎是在一夜之间，我们看到人类DNA所编码的众多疾病相关蛋白。对于药物发现而言，那是一个具有变革意义的时刻。

然而，随着时间推移，我们也逐渐意识到，现有治疗手段能够触达的靶点蛋白仍然非常有限。大多数药物依然主要依赖传统小分子抑制方式来调控靶点蛋白。这也正是我们提出蛋白降解靶向嵌合体和诱导接近理念的原因。它让我们有机会跳出经典抑制剂的思维框架，转而利用细胞自身的机制，以全新的方式调控蛋白。

我希望这一批准能够鼓励科学家在思考如何通过调控蛋白相互作用和细胞系统来应对疾病时，变得更富有创造力。因为一旦你意识到“接近”本身也可以成为治疗的基础，可成药靶点的范围就会被大幅拓展。

最初让您想到蛋白可以被清除，而不仅仅是被抑制的灵感来自哪里？

Craig Crews教授：这个故事要追溯到我早年在耶鲁大学担任助理教授的时候。当时，我获得了一笔青年教师研究基金，用来探索异双功能分子的概念，也就是利用一个分子把细胞内的两个蛋白拉到一起。

在一次科学会议上，我遇到了当时还很年轻的Raymond Deshaies教授，他是一名酵母遗传学家，对E3连接酶和泛素系统非常感兴趣。很巧的是，在一次海报展示中，因为作者姓名的字母顺序相邻，我们的海报被安排在了一起。几杯啤酒下肚之后，由于几乎没有人来看我们的海报，我们便开始互相交流。

那次对话后来成为蛋白降解靶向嵌合体理念的起点。我们开始思考，是否有可能通过一个异双功能分子，把靶蛋白拉近到E3连接酶附近，从而“劫持”细胞自身的蛋白降解机制，降解目标蛋白。

Deshaies教授和我在2001年发表了第一篇论文，随后又完成了几项早期概念验证研究。从生物学角度看，这一机制是可行的，但在完成概念验证之后，相关研究一度停滞不前。我也越来越清楚地意识到，如果这些分子继续依赖基于多肽的配体，它们就很难真正发展成为一种治疗模式。

因此，在2008年前后，经过深思熟虑，我决定围绕小分子重新设计整个系统。在接下来的4年里，我的实验室克服了化学、结构生物学以及检测方法方面的挑战，找到了能够结合E3连接酶的小分子配体，从而支持构建完全由小分子部件“组装”而成的蛋白降解靶向嵌合体。对我来说，那才是真正的转折点。当我们迈出了这一步时，我就意识到，它将从根本上改变药物发现的方式。

在该领域早期，蛋白降解靶向嵌合体并不被看好。当时产业界最大的担心是什么？

Craig Crews教授：很大一部分质疑来自当时药物化学领域的主流思维框架，也就是里宾斯基“五规则”（Lipinski’s Rule of Five）。里宾斯基的研究通过回顾性分析，总结了许多具有口服生物利用度的药物所具备的共同特征，对药物化学界产生了深远影响。然而，随着时间推移，人们几乎把这些规则当成了自然法则。因此，任何超出“五规则”的分子都会被“另眼相看”，而蛋白降解靶向嵌合体显然属于这类分子。它们比传统小分子大得多，也复杂得多，这让很多人感到不安。

有意思的是，质疑同时来自两个相反的方向。有些人认为这些分子根本不会起作用，因为他们不相信这么大的分子能够有效进入细胞。另一些人则认为它们的作用可能过强，一旦降解剂进入细胞，其催化特性可能导致广泛毒性。此外，还有所有药物开发都会面对的常规问题：代谢、药代动力学、给药途径和稳定性等。

不过，在技术层面的疑虑之外，一个更根本的问题也被反复提出：我们为什么需要蛋白降解靶向嵌合体？它与传统抑制剂相比有什么优势？对我来说，这是一个非常重要的科学问题。当我们开始选择需要降解的靶点蛋白时，我们有意聚焦于能够产生“差异性生物学”的场景，也就是说，降解能够实现抑制无法实现的功能。

雄激素受体就是一个很好的例子。在前列腺癌中，一种简单的耐药机制就是产生更多受体。当受体水平足够高时，你就无法使用足够剂量的抑制剂来完全阻断不断增加的蛋白。蛋白降解提供了另一种解决方案，因为它清除的是蛋白本身。BRD4是另一个重要案例。如果你抑制BRD4活性，细胞会通过反馈机制产生更多BRD4，这自然会促进耐药性的产生。而降解则从根本上改变了这一生物学过程。

与此同时，业界也意识到另一个更大的问题。到21世纪初，我们基本上已经了解了完整的人类DNA序列，也发现了大量潜在的疾病相关蛋白。然而，从药理学角度看，我们只能通过传统方法调控大约25%的蛋白。许多驱动疾病的蛋白根本不是酶，它们发挥的是支架或结构组织者的作用。只要这些蛋白存在，就可能产生病理作用。如果疾病机制依赖于某个蛋白的存在，那么抑制可能并不够，这个蛋白必须消失。

也正是在那个时候，我意识到，蛋白降解靶向嵌合体正在回应一个现有治疗模式无法解决的需求。我从来没有把蛋白降解靶向嵌合体视为抑制剂的竞争者。我认为，它是对抑制剂的重要补充。

您认为产业界是在什么时候真正开始接受蛋白降解靶向嵌合体这一理念的？

Craig Crews教授：如果我们在PubMed上搜索“PROTAC”或“targeted protein degradation”，那条代表文献数量的曲线已经说明了一切。在我们2015年的论文发表之后，这一领域几乎开始垂直起飞。2015年，我们终于拿出了产业界多年来一直期待看到的东西：完全由小分子构建、并且在体内具有活性的降解剂。

我们的团队接连发表了两篇重要论文。一篇实现了对名为cereblon的E3连接酶的募集，另一篇则成功募集了名为VHL的E3连接酶。两篇论文相隔数周发表。对医药产业来说，一个曾被视为有趣化学生物学概念的想法，突然变成了真实可及的化合物。

我记得当时自己在想：它们或许不是药物化学家见过的最漂亮的分子，但至少我们可以在这个基础上继续优化和改进。人们终于能够看到，蛋白降解靶向嵌合体不再只是基于多肽的学术工具，而是真正能够在动物体内发挥作用、并有潜力成为药物的小分子。

真正推动产业界采用这一技术的原因，是医药公司开始认识到，降解可以解决其现有项目无法解决的问题。这是蛋白降解靶向嵌合体发展历程中的重要拐点。自那以后，这一领域的研发速度迅速提升。

现在首款蛋白降解靶向嵌合体疗法已经获批，您认为这个领域接下来将走向何方？

Craig Crews教授：现在，越来越让我感兴趣的是诱导蛋白-蛋白相互作用更广泛的应用——它不仅仅可以实现降解，还能够在细胞内创造新的分子关系。

分子胶就是一个非常有意思的例子。在某些情况下，我们可以通过对双功能分子内部连接子结构进行精准优化（linkerology），来设计类似胶水的蛋白-蛋白相互作用。我们也看到一些全新方法正在出现，包括DNA编码化合物库（DEL）技术。这些技术能让我们更系统地发现可诱导或稳定蛋白-蛋白相互作用的化合物。

我对配体发现本身也仍然非常感兴趣。人类DNA编码了数百种E3连接酶，而目前研究探索过的只是其中很小一部分。每一种新的连接酶都可能开启新的生物学、新的组织选择性和新的治疗机会。

此外，还有像调节诱导接近靶向嵌合体这样的全新模式。我们正在借鉴从蛋白降解靶向嵌合体中学到的原则，并以不同方式加以应用，从而创造新的药理机制。坦率地说，我认为20年后，传统抑制剂可能不再是药物开发的主流。未来在很大程度上将由诱导接近机制驱动。

从药物化学角度看，这是一个非常有吸引力的理念。如果你可以只在真正关键的组织中抑制一个酶，为什么还要在全身范围内抑制它？这正是组织选择性接近疗法的魅力所在。理论上，你可以围绕一种特异性表达于骨骼肌的蛋白来设计调节诱导接近靶向嵌合体，利用它在骨骼肌中选择性抑制某个靶点，同时避开心肌组织。这种水平的精准性，用传统抑制剂很难实现。

这是一个雄心勃勃的愿景。为了让这样的未来成为现实，您认为目前仍需解决的最大挑战是什么？

Craig Crews教授：我认为核心挑战仍然是配体发现。要让这些组织选择性接近系统真正发挥作用，就需要针对组织特异性蛋白的高质量配体。如果你想在骨骼肌中抑制一个酶，但不影响心脏，那么首先需要找到一个配体，它所结合的蛋白只存在于骨骼肌，而不存在于心肌。因此，真正的瓶颈在于，我们还需要对蛋白质组本身有更深入的理解。

持续驱使我不断向前的，是我称之为“梦想中的小分子”的理念。也就是说，有一天，我们能够拥有一个覆盖几乎整个蛋白质组的配体库。如果拥有这样的化合物库，药物发现就可以变得更加模块化。研发人员可以根据治疗理念，使用可互换的“零部件”来组装疗法：选择一个针对疾病相关蛋白的配体，再选择一个针对组织选择性蛋白的配体，并加上一个用于实现降解或功能调控机制的分子。你可以像从“货架”上取下分子零件一样，构建出全新的药物类型。这仍然是激励我不断探索的愿景。

非常感谢您分享这些洞见。最后想请您从更个人的角度做一个总结。回顾这段非凡的旅程，对您而言，最令人激动或最有意义的一个时刻是什么？

Craig Crews教授：有意思的是，对我来说，那个决定性时刻并不是看到临床数据。那时，我已经确信蛋白降解靶向嵌合体这种生物学机制是可行的。

真正震撼我的时刻，是我第一次看到实际生产出来的原料药。我记得那大约是5公斤蛋白降解靶向嵌合体物料，将被用于人体试验。我至今仍记得自己凝视着那个装满化合物的塑料袋。对大多数人来说，它可能看起来再普通不过。但对我而言，那一刻却让我百感交集。因为突然之间，那个许多年前由两位年轻教授站在几乎无人问津的海报旁、在头脑风暴中开启的想法，已经真正跨入现实。它不再只是停留在化学、生物学或学术层面的概念，而是即将真正进入人体。回望这一路走来的历程，我一时间几乎说不出话来。也正是在那一刻，我才真正感受到，这段旅程的分量远超我的想象。

Reflection on the First Approved PROTAC and the Future of Small Molecule Drugs: A Conversation with Prof. Craig Crews, Yale University

In May 2026, the FDA approved the first PROTAC therapy, marking the arrival of an entirely new class of medicines built on a once-radical idea: instead of merely inhibiting disease-causing proteins, drugs could be designed to eliminate them altogether.At the center of that story is Prof. Craig Crews, whose laboratory at Yale University helped pioneer the PROTAC (PROteolysis TArgeting Chimera) approach more than two decades ago.

Over the course of his career, Prof. Crews has bridged academia and biotechnology in unusual ways. In 2003, he co-founded Proteolix, whose proteasome inhibitor Kyprolis later received FDA approval for multiple myeloma. He subsequently co-founded Arvinas, the company that helped advance PROTAC therapeutics into the clinic and ultimately led to this year’s approval. His third company, Halda Therapeutics, which was recently acquired by Johnson & Johnson, focuses on developing a new induced-proximity-based modality called RIPTAC (Regulated Induced Proximity Targeting Chimeras).

Following the historic approval of the first PROTAC therapy, we spoke with Prof. Crews about the origins of the idea, the skepticism that initially surrounded the field, the scientific turning points that transformed targeted protein degradation into a viable drug modality, and why he believes induced proximity technologies may fundamentally reshape the future of medicine.

Professor Crews, congratulations again on the approval of the first PROTAC therapy. When you heard the news, what went through your mind?

Craig Crews:To me, the approval represents something much larger than a single drug or even a single modality. I still remember the excitement surrounding the Human Genome Project in the 1990s. It felt transformative from a drug-discovery perspective. Suddenly we could see the enormous landscape of potential disease-related proteins across biology.

But over the years, it also became clear how limited our therapeutic reach actually was. Most of our drugs could only target proteins that were controllable through traditional small-molecule inhibition. That’s why we came up with PROTAC and induced proximity.It gives us a way to think beyond classical inhibition and begin using the cell’s own machinery to manipulate proteins in entirely new ways.

I hope it encourages scientists to think more creatively about how we control biology, how we manipulate proteins, interactions, and cellular systems to address disease. Becauseonce you realize proximity itself can become a therapeutic principle, the boundaries of what may be druggable start to expand dramatically.

What originally led you to the idea that proteins could be eliminated rather than simply inhibited?

Craig Crews:The story really goes back to my early years as an assistant professor at Yale. At the time, I had received a junior faculty award to explore this heterobifunctional concept, the idea that you could use a molecule to bring two proteins together inside a cell.

At a conference, I met fellow faculty member named Raymond Deshaies, a yeast geneticist who was deeply interested in E3 ligases and the ubiquitin system. By coincidence, our posters ended up next to each other alphabetically at a poster session. And after a few beers, mainly because nobody was coming by to talk to us about our posters, we started talking to each other instead.

That conversation became the genesis of the PROTAC idea.We began wondering whether it might be possible to hijack the cell’s own protein degradation machinery by using a heterobifunctional molecule to bring a target protein into proximity with an E3 ligase.

Ray and I published the first paper in 2001, followed by several early proof-of-concept studies. The biology worked, but for a while, that was where things stayed.It became increasingly clear to me that this was never going to become a true therapeutic modality if we continued relying on peptide-based ligands for those molecules.

So around 2008, I made a very deliberate decision to redesign the entire system around small molecules. For the next four years, my lab worked through the chemistry, the structural biology, and all the necessary assays to identify small-molecule ligands capable of binding E3 ligases and enabling an all-small-molecule PROTAC.That was really the turning point for me. Once we achieved that, I realized this could fundamentally change the way drug discovery is done.

From the early days of the field, there was still a great deal of skepticism around PROTACs. Why was the industry so doubtful at the time? What were the biggest concerns?

Craig Crews:A lot of the skepticism came from how dominant Lipinski’s “Rule of Five” thinking was in medicinal chemistry at the time. Chris Lipinski’s work was incredibly influential because it retrospectively identified the common properties shared by many orally bioavailable drugs. But over time, people began treating those rules almost as laws of nature. So there was a very strong prejudice against anything that fell beyond the Rule of Five, and PROTACs clearly did. These molecules were much larger and more complex than traditional small molecules, which immediately made many people uncomfortable.

What was interesting is that the skepticism came from opposite directions at the same time. Some people thought the molecules would never work at all. People simply didn’t believe molecules of that size could efficiently enter cells. Others thought they might work too well — that once a degrader entered the cell, their catalytic nature would lead to widespread toxicities. Then there were all the standard pharmaceutical concerns: metabolism, pharmacokinetics, routes of administration, stability, and so on.

But beyond the technical objections, there was also a more fundamental question people kept asking: why do we even need this? What’s the advantage over inhibitors? That became very important for me scientifically.When we started selecting proteins to degrade, we deliberately focused on what I called “differential biology” — situations where degradation could achieve something inhibition fundamentally could not.

The androgen receptor was a great example. In prostate cancer, one resistance mechanism is simply making more receptors. At some point, you can’t realistically give enough inhibitor to fully block the increasing amount of protein. Degradation offers a different solution because you remove the protein itself. BRD4 was another important case. If you inhibit BRD4, the cell responds through feedback mechanisms by producing more of it, which naturally contributes to resistance. Again, degradation changes the biology in a fundamentally different way.

And there was also a larger realization happening at the same time. By the early 2000s, we essentially knew the entire human genome. We had identified enormous numbers of potential disease-related proteins. Yet pharmacologically, we could only control maybe twenty-five percent of the proteome using conventional approaches. Many proteins driving disease don’t function as enzymes at all. They act as scaffolds or structural organizers. Their pathological role is simply that they exist. And if the disease mechanism depends on the existence of the protein, then inhibition may not be enough — the protein has to disappear.

That was the moment I realized PROTACs were addressing a need that existing modalities simply could not solve. I never saw PROTACs as competing with inhibitors. I saw them as complementing them.

When do you think the industry truly began to embrace the idea of PROTACs?

Craig Crews:If we look at a PubMed search for “PROTAC” or “targeted protein degradation,” the curve really tells the story. After our 2015 papers, the field essentially takes off almost vertically. What changed in 2015 was that we finally demonstrated something the industry had been waiting to see for years:, in vivo active degraders built entirely from small molecules.

At roughly the same time, two important papers emerged from our group. One used Cereblon as the recruited E3 ligase, and another used VHL. They came out within weeks of each other. Suddenly, what had previously been viewed as an interesting chemical biology concept became something much more tangible to the pharmaceutical industry.

I remember thinking at the time: they may not be the prettiest molecules medicinal chemists had ever seen, but at least now we had something real to work with. People could now see that these were no longer peptide-based academic tools. They were actual small molecules that could function in animals and potentially become drugs.

From that point on, things accelerated very quickly.What really drove adoption was that companies began recognizing that degradation could solve problems their existing programs could not. That was the inflection point.

Now that we have the first approved PROTAC therapy, where do you think the field is heading next?

Craig Crews:What increasingly interests me is the broader concept of induced protein-protein interactions — not just degradation, but the ability to deliberately create new molecular relationships inside cells.

Molecular glues are a fascinating example of that. In some cases, we can engineer glue-like protein-protein interactions through what people now call “linkerology,” meaning the precise optimization of the linker architecture inside bifunctional molecules. But we’re also seeing entirely new approaches emerge, including DNA-encoded library technologies that may allow us to discover compounds capable of inducing or stabilizing protein-protein interactions much more systematically.

I’m also still very interested in ligand discovery itself. The field has only explored a tiny fraction of the hundreds of E3 ligases encoded in the human genome. Each new ligase potentially opens new biology, new tissue selectivity, and new therapeutic opportunities.

And then there are entirely new modalities like RIPTACs, where we’re taking the principles we learned from PROTACs but applying them in different ways to create new pharmacological mechanisms. To be honest, I don’t think twenty years from now we’ll still mainly be developing traditional inhibitors. I think much of the future will be proximity-driven.

From a medicinal chemistry perspective, it’s a very appealing idea. Why inhibit an enzyme systemically throughout the body if you could inhibit it only in the tissue where you actually want the effect? That’s where tissue-selective proximity therapeutics become very powerful. In principle, you could design a RIPTAC around a protein expressed specifically in skeletal muscle, for example, and use that to selectively inhibit a target in skeletal muscle while sparing cardiac tissue. That level of precision is extremely difficult to achieve with conventional inhibitors.

That’s a remarkably bold vision. To make that future possible, what do you think is the biggest challenge the field still needs to solve?

Craig Crews:I think the central challenge is still ligand discovery. For these tissue-selective proximity systems to work, you need high-quality ligands for tissue-specific proteins. If you want to inhibit an enzyme in skeletal muscle but not in the heart, you first need a ligand for a protein that exists specifically in skeletal muscle and not cardiac muscle. So the real bottleneck becomes building a much deeper molecular understanding of the proteome itself.

What still drives me is what I sometimes call “the small molecules of my dreams” — the idea that one day we could have a comprehensive catalog of ligands for essentially the entire proteome.If we had that, then drug discovery could become much more modular.Researchers could almost assemble therapeutics conceptually from interchangeable components: one ligand for a disease-related protein, another for a tissue-selective protein, another for a degradation or functional mechanism. You could essentially pull molecular parts off the shelf and build entirely new classes of medicines. That’s still very much the vision that motivates me.

Thank you very much for sharing these insights. To conclude on a more personal note, looking back on this remarkable journey, what was the single most exciting or meaningful moment for you? Could you share that moment with our audience?

Craig Crews:Interestingly, for me, the defining moment wasn’t actually the clinical data. By that point, I already believed deeply in the science. I was already convinced the biology worked.

The moment when it truly became real for me was the first time I saw the actual manufactured drug substance — I think it was about five kilograms of PROTAC material that was going to be used in humans. I still remember staring at this plastic bag filled with the compound. To most people, it probably looked completely ordinary. But to me, it was overwhelming. Because suddenly this tiny idea, something that had started fifteen years earlier as a speculative conversation between two young faculty members standing beside largely ignored posters, had crossed into reality. It was no longer just chemistry, or biology, or an academic concept. It was now something that would actually enter a human being. And I remember feeling almost stunned by the arc of that journey. That was the moment when it truly hit me.

免责声明：本文仅作信息交流之目的，文中观点不代表药明康德立场，亦不代表药明康德支持或反对文中观点。本文也不是治疗方案推荐。如需获得治疗方案指导，请前往正规医院就诊。

版权说明：欢迎个人转发至朋友圈，谢绝媒体或机构未经授权以任何形式转载至其他平台。转载授权请在「药明康德」微信公众号回复“转载”，获取转载须知。