内容来源：https://www.quantamagazine.org/why-aging-may-be-a-program-not-a-breakdown-20260814/

内容总结：

衰老不是磨损，而是一场“程序化”的重构？

长期以来，科学界普遍认为衰老是身体器官和分子不断损耗、修复机制逐渐失灵的结果，如同一台年久失修的机器最终走向报废。但洛克菲勒大学细胞生物学家曹俊越提出了一种截然不同的观点：衰老并非随机磨损，而是一个被“程序设计”好的、分阶段进行的秩序化过程，其背后存在明确的分子信号和细胞群体的协同变化。

曹俊越的研究团队利用高通量单细胞基因组学技术，对小鼠衰老过程进行了系统观察。他们分析了来自14个器官或组织、约50只不同年龄小鼠的2100万个细胞，构建了详细的基因表达数据集。研究发现，衰老并非所有细胞同步退化，而是在不同年龄段呈现出明显的阶段性特征：约四分之一的细胞亚型会经历剧烈变化，其余则保持稳定。

具体来看，小鼠在3至6个月龄（相当于人类20至30岁）时，部分脂肪细胞、肌肉细胞及脑内两类具有再生潜力的未成熟细胞开始减少；6至12个月龄（相当于人类30至40多岁），维持组织稳态的肌腱细胞、血管周细胞、结肠平滑肌细胞、肾上皮细胞以及部分组织保护性免疫细胞出现显著耗竭；而到了12个月龄左右（约人类40至50岁），从“细胞减少”转向“细胞扩张”，以免疫细胞为主的细胞群体开始大量增殖；16个月龄后（约人类55岁以后），与衰老相关的特殊免疫细胞进一步扩张，这些“失控”的细胞会逐渐破坏系统，并推高炎症相关疾病的风险。

曹俊越指出，这些阶段性变化与人类“断崖式衰老”的观察相吻合——许多人在40多岁至50多岁之间会感到身体机能骤然下降，血液蛋白质谱也在这一阶段发生显著改变，慢性病风险同步上升。

更重要的是，研究团队发现，每个衰老阶段中，基因组中特定区域的开启或关闭状态高度可重复，而非随机变化。这意味着衰老背后存在一套“程序化指令”。曹俊越比喻说，哺乳动物的衰老就像秋天的树——叶子并非逐片凋零，而是在夏秋之交的两周内集中变色脱落，触发这一变化的是光照时间的变化所引发的信号。同样，哺乳动物衰老过程中的突然变化，表明存在某种信号驱动了整个系统的重构。

曹俊越还发现，驱动衰老的信号既包括内部基因调控蛋白，也包括细胞分泌的细胞因子等外部信号分子。团队已经识别出一些罕见的脆弱细胞类型，可作为抗衰老干预的潜在靶点，并破译了部分驱动衰老的分子代码。

曹俊越强调，一个关键发现是：机体的再生能力和系统韧性在中年之前就已开始下降。因此，若想延缓衰老、实现“逆转”，干预时机应大大提前，而非等到衰老迹象明显后再行动。

中文翻译：

为何衰老可能是一个程序，而非一次崩坏
引言
在某些方面，我们对衰老并不陌生——从头发变白、皮肤起皱，到运动、感官和认知能力的下降，皆可见其踪迹。然而，衰老背后的生物学机制至今仍充满不确定性与争论。许多研究路线支持这样一种理论：衰老是衰败的直接结果——分子（包括蛋白质或DNA）、细胞器、细胞乃至整个器官，因外部侵袭或不可逆的磨损而不可避免地退化。当身体的修复机制无法跟上这些变化时，就像一个设备日渐老化、维修工却寥寥无几的工厂，衰老的种种迹象便随之显现，最终走向死亡。
这一观点听起来很有道理，但在细胞生物学家曹俊越看来，它并不准确。他认为，衰老远非一个随机而线性的损耗过程，而是一个分阶段、有程序、有条不紊的事件。“系统的毁灭在非常早期的阶段就被设定好了。”曹俊越说道。他是洛克菲勒大学单细胞基因组学与群体动力学实验室的负责人。利用能够从全系统视角观察小鼠衰老过程的技术，曹俊越勾勒出衰老的各个独立阶段——类似于胚胎发育的阶段——这些阶段由分子信号和特定细胞群体的变化所界定。在人类身上，这一过程可能在30岁之前就已开始。
曹俊越对衰老的兴趣始于中国河北的高中时期，那时他深切意识到，祖父母和父母不会永生。当许多青少年面对死亡这一现实时，或许会转向诗歌或做出自我毁灭的行为，而曹俊越则转向了科学。找到延缓衰老的方法成为他毕生的目标，也正是因为这个原因，他选择了北京大学生物学作为自己的专业。
最初，曹俊越认为特定蛋白质和蛋白质网络的退化是衰老的罪魁祸首，这与主流模型一致。但大学毕业后，当他在实验室试图识别那些蛋白质时，他意识到与衰老相关的蛋白质数量惊人，且它们的效果取决于其所作用的细胞类型。这幅图景比他原先想象的既更复杂，也更有组织性。
曹俊越决定，他需要数据——大量的数据——涵盖数百种细胞类型中数千种分子变化的数据。作为研究生，他开发了一种高通量技术，能够量化胚胎发育中的这些动态变化。2020年，当他在洛克菲勒建立自己的实验室时，他将这项技术应用于衰老研究。
在一系列实验中，曹俊越及其团队处理了2100万个细胞，这些细胞来自约50只雄性和雌性小鼠在五个生命阶段的14个组织或器官，并为每个细胞构建了基因表达数据集。“这是极其大规模的数据，”曹俊越说，“你知道它来自哪个器官、哪个年龄，还能获得大量的分子信息。”每个阶段都以特定细胞类型的急剧减少或扩增为标志。
他在2025年和2026年发表于《科学》杂志的两篇里程碑式论文指出，随着哺乳动物衰老，构成身体的细胞发生了根本性的重新分布，并描述了指导这一过程的部分表观基因组指令。“衰老过程中存在分子变化，可能还有其他变化，”曹俊越说，“但它们都汇聚于细胞社会的重塑。”
《量子》杂志与曹俊越就衰老程序化理论的证据、每个阶段发生的变化，以及衰老的哺乳动物为何像秋天落叶的树木等问题进行了对话。以下访谈经过删减和编辑，以保证清晰流畅。
你从大学时代就开始研究衰老。你对这个课题的思路是如何演变的？
大学时期，我以为衰老问题需要解决的是一颗药。我以为我们会开发出某种神奇的药物或化合物，用来延长动物的寿命。所以我加入了一个做计算药物设计的实验室，试图设计针对与衰老相关分子的肽。但最大的挑战是，我们不知道该靶向什么分子。
毕业后，我去了美国，在杰克逊实验室（一家总部位于缅因州的生物医学研究非营利机构）做研究助理，研究小鼠体内与衰老相关的分子通路。我以为通过研究通路，我们就能找到一个（药物）靶点，然后在此基础上开发药物来延长寿命。
但几年后，我认识到衰老无法用单一通路或靶点来解释。它涉及许多不同的通路，而这些通路对体内不同细胞类型有着截然不同的影响。我想了解衰老如何在分子层面，跨越不同器官中成百上千种细胞类型而推进。
你是如何做到这一点的？
一个主要挑战是，衰老涉及多个层面的变化——分子、细胞器、细胞、器官以及整个身体。要测量所有这些层面的变化，我们需要既高通量又高分辨率的技术。高通量意味着你可以用它扫描的不仅仅是一个细胞或一种细胞类型，而是整个生物体内数百万个细胞；高分辨率则让你能够看到细胞和分子层面极为细致的变化。
我申请了研究生院，在那里我开发了能够扫描整个哺乳动物机体——从大脑到肾脏再到肺部——的工具，同时还能详细记录数百种不同细胞类型中单个细胞内数万个基因表达的变化。当我到洛克菲勒之后，我们利用这项技术来理解衰老过程中整个系统的动态变化。
你能描述一下那些实验以及你从中得到的发现吗？
在一组研究中，我们从不同年龄的小鼠各器官中提取了超过2000万个细胞：3个月、6个月、12个月、16个月和23个月——大致相当于人类的20岁、30岁、50岁、60岁和75岁。我们分析了每个细胞中2万个基因的表达情况，并利用这些信息来定义细胞类型。然后我们追踪了它们的群体动态变化。
我们发现，并非每种细胞类型都会在衰老中发生变化。我们识别出536种主要细胞类型和1828种亚型。其中只有约四分之一的亚型在衰老过程中表现出显著的变化。其余的在整个生命周期中保持稳定。
令人惊讶的是，衰老的变化并非在所有细胞中普遍发生，而是存在一些更容易受影响的特定细胞群体。
这些群体变化发生在什么时候？
我们发现，衰老可以被划分为若干个不同的时间窗口。在每一个窗口中，特定的细胞群体会表现出协调一致的动态变化。例如，在早期阶段，我们看到某些细胞类型被迅速消耗殆尽。随后是另一个阶段，在这个阶段中，其他细胞大幅扩增。
每个阶段受影响的是哪些类型的细胞？
在第一个阶段，即小鼠3至6个月时（相当于人类约20至30岁），某些脂肪细胞和肌肉细胞会减少，同时大脑中两种具有再生不同类型脑组织能力的未成熟细胞也会减少。
在小鼠6至12个月之间（相当于人类30多岁到40多岁），我们看到维持身体组织所需的细胞急剧减少。这些包括肌腱细胞（肌腱的主要组成部分）；包裹血管并稳定循环系统的细胞；结肠的平滑肌细胞；以及肾脏上皮细胞——它们负责过滤血液中的毒素。同样在减少的还有某些保护特定组织（如肠道）的免疫细胞。
在小鼠约12个月时（大约相当于人类40至50岁），从细胞减少转向细胞扩增。第一波扩增以免疫细胞为主，但也包括肺、肾等器官中的一些选择性细胞，这些细胞的特性因应激或炎症而发生了改变。在小鼠约16个月时（相当于人类接近60岁及以后），与衰老相关的特殊免疫细胞开始扩增。当这些自私的、不受控制的细胞出现后，它们最终会增殖并摧毁整个系统。与此同时，它们可能增加了随年龄增长而罹患炎症性疾病的风险，如心脏病、关节炎、癌症和慢性呼吸系统疾病。
是否有证据表明这些阶段在人类身上也会发生？
有。研究人员在中年人类身上观察到一种被称为“突发性衰老”的现象，这与我们看到的细胞群体动态变化是一致的。当研究人员分析人类血液中的蛋白质特征时，他们发现40多岁中段到50多岁后期之间发生了显著变化。人们也倾向于在中年时报告身体功能的突然下降，而这正是对癌症和一些神经退行性疾病等与年龄相关疾病的易感性上升的时期。
你的发现告诉我们关于衰老本质的什么信息？
以前，人们将衰老视为蛋白质和DNA等分子损伤的线性积累。但我们发现，衰老并不是一个线性过程。它更像一个发育过程，其中有不同的阶段，涉及不同器官中特定细胞类型的协调变化。我们的观点是，衰老与其说是分子损伤，不如说是整个细胞社会的重塑。
是什么控制了衰老中这种细胞重塑？如果有控制者的话？
我们的技术使我们能够检查支配每个细胞功能的基因组程序——基因组中哪些区域是活跃的，哪些是沉默的。如果是随机的分子损伤，我们可能会看到基因组中随机的变化。但我们在每个阶段总是看到相同的区域处于开放（活跃）或封闭（沉默）状态。我们识别出28万个基因组区域，它们在衰老过程中于特定细胞类型中可重复地开放或关闭。
所以身体有一个衰老的程序？
是的。哺乳动物衰老中的突发性变化，加上每个阶段协调一致的细胞动态变化，表明存在控制衰老的上游信号。这就像秋天的树：它的叶子不是线性地凋落，而是在夏秋交替的两周内集中落下。日照的变化触发了信号的释放，然后整个系统随之改变。在哺乳动物衰老中，突发性的变化表明有某种信号在驱动这个系统。
是什么样的信号？
我们发现了内部的分子程序——控制基因表达的蛋白质——以及外部的、被称为细胞因子的分泌型分子，它们驱动着细胞变化。
你的工作对于如何延缓或阻止衰老有什么启示？
我们识别出了一些罕见的脆弱细胞类型，它们可以作为抗衰老干预的靶点，而且我们找到了驱动衰老过程的分子密码，因此我们可以开发一些方法来重新编程这个密码。
至关重要的是，我们已经证明，系统的再生能力和稳健性在中年之前就开始下降。所以如果我们想要挽救衰老，我们应该尽早开始。

英文来源：

Why Aging May Be a Program, Not a Breakdown
Introduction
In some ways, we know aging when we see it, from the graying of hair to the wrinkling of skin to declines in motor, sensory, and cognitive capacities. Yet the underlying biology of aging remains a matter of uncertainty and debate. Many lines of research align with the theory that aging is a direct result of decay — the inevitable degradation of molecules (including proteins or DNA), organelles, cells, or whole organs — from external assault or inexorable breakdown. When the body’s repair mechanisms fail to keep pace with these changes, like a factory with deteriorating equipment and too few mechanics, it manifests as the known signs of aging and, eventually, death.
The idea makes a lot of sense, but according to the cell biologist Junyue Cao, it’s inaccurate. Far from a random but linear process of wear and tear, he argues, aging is a stepwise, programmed, orderly affair. “The destruction of the system is programmed at a very early stage,” said Cao, who heads the Laboratory of Single-Cell Genomics and Population Dynamics at Rockefeller University. Using technology that offers a systemwide view of the aging process in mice, Cao has outlined discrete stages of aging, akin to those of embryonic development, that are defined by changes in molecular signals and specific cell populations. In humans, the process likely begins before age 30.
Cao’s interest in aging began in high school in Hebei, China, when he became acutely aware that his grandparents and parents were not going to live forever. While many teenagers awakening to mortality might turn to poetry or self-destructive behavior, Cao turned to science. Finding a way to slow aging became his lifelong goal, and it’s why he chose biology as his major at Peking University in Beijing.
Initially, Cao assumed that the deterioration of particular proteins and protein networks was responsible for aging, in line with the prevailing model. But then after college, when he was working in a lab trying to identify those proteins, he realized that a daunting number of them were associated with aging, and that their effects depended on the type of cell in which they were operating. It was a picture both more complex and more organized than he had thought.
Cao decided he needed data — lots of it — on thousands of molecular changes across hundreds of cell types. As a graduate student, he developed a high-throughput technology that could quantify these dynamics in embryonic development. When he started his own lab at Rockefeller in 2020, he put this technology to work on aging.
In one series of experiments, Cao and his team processed 21 million cells, sampled from 14 tissues or organs in about 50 male and female mice at five life stages, and built a data set of gene expression for each cell. “It’s extremely large-scale data,” Cao said. “You know which organ it’s from and which age it’s from, and you also know extensive molecular information.” Each stage was marked by a dramatic decline in or expansion of specific cell types.
Two of his landmark papers, published in 2025 and 2026 in Science, point to a radical redistribution of the cells that make up the body as mammals age, and describe some of the epigenomic instructions that guide this process. “There are molecular changes and maybe some other changes in aging,” Cao said, “but they all converge in the remodeling of the cell society.”
Quanta spoke with Cao about evidence for the programmed theory of aging, what happens at each stage, and why aging mammals are like trees shedding leaves. The interview has been condensed and edited for clarity.
You began studying aging as a college student. How has your approach to the topic evolved?
Back in college, I thought the problem of aging that needed solving was the lack of a drug. I thought that we were going to develop some magic drug or chemical that we could use to increase the lifespan of animals. So I joined a lab working on computational drug design, trying to design peptides that target molecules related to aging. But the major challenge was that we didn’t know what molecules to target.
After I graduated, I moved to the U.S. and worked at the Jackson Laboratory [a Maine-based biomedical research nonprofit] as a research assistant studying molecular pathways associated with aging in mice. I thought that by studying pathways, we could identify a [drug] target and, from there, develop drugs to increase lifespan.
But after a few years, I learned that aging cannot be explained by a single pathway or target. It involves many different pathways, and these pathways have very different effects on different cell types in the body. I wanted to understand how aging progresses on a molecular level across hundreds to thousands of cell types in different organs.
How did you do that?
A major challenge is that aging involves changes across many different levels — molecules, organelles, cells, organs, and the whole body. To measure changes at all these levels, we needed technology that was both high throughput, which means you can use it to scan not just one cell or cell type but millions of cells across an entire organism, and high resolution, so you can see very detailed changes at the cellular and molecular levels.
I applied to graduate school, where I developed tools that could scan the entire mammalian organism, from the brain to the kidneys to the lungs, while detailing tens of thousands of gene expression changes within single cells across hundreds of different cell types. When I got to Rockefeller, we used this technology to understand the dynamics of the whole system in aging.
Can you describe those experiments and what you learned from them?
In one set of studies, we extracted more than 20 million cells from various organs from mice of different ages: 3, 6, 12, 16, and 23 months — roughly equivalent to 20, 30, 50, 60, and 75 years in humans. We analyzed the expression of 20,000 genes per cell and used this information to define the cell types. Then we tracked their population dynamics.
We found that not every cell type gets changed in aging. We identified 536 main cell types and 1,828 subtypes. Only about one-quarter of these subtypes show a strong shift in aging. Others remain stable across the lifespan.
It is surprising to find that changes in aging are not universal across all the cells, that there are specific cell populations that are more vulnerable.
When do these population changes occur?
We found that aging can be separated into distinct time windows. In each window, specific groups of cells show coordinated dynamics. In the early phase, for example, we see that some cell types are rapidly depleted. This is followed by another phase, in which other cells are greatly expanded.
What types of cells are affected at each stage?
In the first stage, 3 to 6 months in a mouse [about 20 to 30 years in a human], there is a loss of certain fat and muscle cells, and of two immature cell types in the brain that have the capacity to regenerate different types of brain tissue.
Between 6 and 12 months in a mouse [equivalent to a person in their 30s and 40s], we see dramatic depletion of cells needed to maintain the body’s tissues. These include tenocytes [the primary component of tendons]; the cells that wrap around blood vessels and stabilize the circulatory system; the smooth muscle cells of the colon; and kidney epithelial cells, which filter toxins from the blood. Also in decline are some immune cells that protect specific tissues such as the intestine.
At around 12 months in mice [roughly 40 to 50 years in humans], there is a shift from cell depletion to cell expansion. The first expansion wave is dominated by immune cells, but also includes select cells in the lungs, kidneys, and other organs whose properties have changed as a result of stress or inflammation. At around 16 months in mice [late 50s and beyond in humans], specialized aging-associated immune cells expand. When these selfish, uncontrolled cells emerge, they will eventually proliferate and destroy the system. In the meantime, they may contribute to the increased risk with age of inflammatory conditions such as heart disease, arthritis, cancer, and chronic respiratory illnesses.
Is there evidence that these stages also occur in humans?
Yes, researchers have observed a phenomenon called “abrupt aging” in middle-aged humans that is consistent with the cell population dynamics we saw. When researchers analyze protein signatures in human blood, they see a significant change between the mid-40s and late 50s. People also tend to report an abrupt decline in function in middle age, which is also when there is a rise in susceptibility to age-related diseases such as cancer and some neurodegenerative diseases.
What do your findings tell us about the nature of aging?
Previously, people saw aging as a linear accumulation of damage to molecules such as proteins and DNA. But we found that aging is not a linear process. It’s more like a developmental process, in which there are distinct stages that involve coordinated changes in specific cell types across different organs. Our claim is that aging is not so much molecular damage as a remodeling of the entire cell society.
What controls this cellular remodeling in aging, if anything?
Our technology allows us to examine the genomic program that governs the function of each cell — which regions of the genome are active and which are silent. If it’s random molecular damage, we might see random changes across the genome. But we always see the same regions that are open [active] or closed [silent] at each stage. We identified 280,000 genomic regions that are reproducibly open or closed during the aging process in specific cell types.
So the body has a program for aging?
Yes, the abrupt changes in mammalian aging along with the coordinated cellular dynamics at each stage suggest there are upstream signals that control aging. It’s like a tree in autumn: Its leaves fall not in a linear way but in just two weeks during the transition between summer and autumn. The changes in daylight trigger the release of a signal and then the system gets changed. In mammalian aging, the abrupt changes show there is some kind of signaling that drives the system.
What type of signaling?
We have found both internal molecular programs — proteins that control gene expression — and external, secreted molecules called cytokines that drive the cellular changes.
What does your work tell us about how to slow or stop aging?
We have identified some rare vulnerable cell types that could be targets for anti-aging interventions, and we have found the molecular code that drives the aging process, so we can develop some approaches to reprogram that code.
Crucially, we have shown that the regenerative capacity and robustness of the system decrease before middle age. So if we want to rescue aging, we should start early.