洛克菲勒大學細胞生物學家曹君越透過分析小鼠不同生命階段約兩千一百萬個細胞的基因表達數據，提出衰老並非隨機的分子磨損，而是一個有序的、階段性的程序化過程。他的研究團隊從約五十隻雌雄小鼠的十四種組織或器官中取樣，建立了龐大的單細胞基因表達數據集，發現衰老過程中只有約四分之一的細胞亞型發生顯著變化，其餘則在整個生命週期中保持穩定，這表明衰老並非普遍影響所有細胞，而是針對特定易感細胞群體。（關鍵數字：21, 14, 50, 1,828）
曹君越的研究揭示了哺乳動物衰老的離散階段：在小鼠三至六個月（相當於人類二十至三十歲）時，特定脂肪細胞、肌肉細胞及腦中未成熟細胞開始減少；六至十二個月（人類三四十歲）時，維持身體組織的關鍵細胞如腱細胞、血管周圍細胞和腎上皮細胞大量耗竭；約十二個月後（人類四五十歲），衰老從細胞耗竭轉向細胞擴增，免疫細胞大量增生；十六個月後（人類五十多歲以後），與衰老相關的特化免疫細胞失控擴張，可能導致心臟病、關節炎和癌症等炎症性疾病風險升高。（關鍵數字：20, 30, 40）
這項研究對抗衰老策略具有重要啟示。曹君越的團隊識別出二十八萬個在特定細胞類型衰老過程中可重複開啟或關閉的基因組區域，證明衰老受到上游分子信號——包括控制基因表達的蛋白質和細胞因子——的調控，而非隨機損傷所致。他將此過程比喻為秋天的落葉：並非線性衰退，而是由信號觸發的協調性變化。研究還表明，人體的再生能力在中年之前就已開始下降，因此抗衰老干預應盡早開始。（關鍵數字：280,000, 30）
Cell biologist Junyue Cao at Rockefeller University has challenged the prevailing view that aging results from random molecular wear and tear. By profiling gene expression in 21 million cells sampled from 14 tissues across roughly 50 mice at five life stages, his team discovered that aging proceeds through discrete, coordinated stages resembling embryonic development. Only about one-quarter of the 1,828 cell subtypes identified showed significant shifts during aging, while the rest remained stable, indicating that aging selectively targets specific vulnerable cell populations rather than universally degrading all cells.
Cao's data revealed a structured timeline of cellular remodeling in mice that likely parallels human aging. Early stages (equivalent to the human 20s and 30s) involve the loss of certain fat, muscle, and brain progenitor cells, followed by depletion of tissue-maintaining cells such as tenocytes, vascular support cells, and kidney epithelial cells during the equivalent of the human 30s and 40s. Around midlife, the pattern shifts from cell loss to cell expansion, dominated by immune cells and stress-altered cells in various organs. In later stages, specialized aging-associated immune cells proliferate uncontrollably, contributing to increased risks of inflammatory diseases including heart disease, arthritis, and cancer.
These findings carry profound implications for anti-aging interventions. Cao's team identified 280,000 genomic regions that are reproducibly activated or silenced during aging in specific cell types, pointing to an upstream molecular program rather than stochastic damage. Internal regulators—proteins controlling gene expression—and external signaling molecules called cytokines together orchestrate these cellular changes. Cao compares the process to autumn leaf fall: not gradual but triggered by signals that rapidly transform the system. Critically, his research shows that the body's regenerative capacity declines before middle age, suggesting that effective anti-aging strategies must begin early, potentially before age 30 in humans.