共变不等于共享:XBP1s 的 12 小时振荡器与月级成分,这一刀卡在 30 天这个数字上

Written by

in

Language

Evidence level: B

用细胞系做”ER 应激诱导 → 同时看 12 小时成分与 29.5 天成分”是最便宜的直接检验,但在动手之前有三个坑必须拆:共变不等于共享、12 小时谱峰自己就有三种来源、以及 30 天培养根本测不出 29.5 天——第三个坑是数字层面的硬伤。

正文

《素问·六微旨大论》讲”天气始于甲,地气始于子,子甲相合,命曰岁立”。中国历法有个很少被提起的特点:它从来不试图造一只”总的钟”来记时间。它摆出两套周期——十干、十二支——让它们各自转,靠两套周期互相错开产生的组合来标记唯一时刻。六十年一轮的甲子,不是某只大钟走了一圈,是两只小钟的最小公倍数。

这一篇要检验的东西,正好撞在这种”两套周期”的结构上。

一、要被检验的命题

此前的文章已经理清两条线索。

一条是 XBP1s 的 12 小时振荡器。哺乳动物肝脏里有一套细胞自主的、独立于昼夜钟的 12 小时节律,核心是未折叠蛋白反应(UPR)的转录因子 XBP1s 与上游感受器 IRE1α(PMID 31935211、PMID 41845938);它驱动的是整条”中心法则信息流”——转录起始、mRNA 加工、核糖体生物合成、蛋白质折叠(PMID 31935211),通过核斑点的相分离动力学实现 12 小时节律(PMID 34985945),并与脂质代谢和膜流动性稳态直接挂钩(PMID 33277471)。到 2026 年,组蛋白甲基转移酶复合体亚基 RBBP5 被证明是这套振荡器的上游表观调控节点(PMID 42848836、PMID 39314427)。最关键的一条旁证是:把昼夜钟核心去掉之后,约 12 小时的 mRNA 与蛋白质代谢节律仍在(PMID 37205336)——这套振荡器是真的自主,不是昼夜钟的影子。

另一条线索是人类近月钟的分子实体至今是空的。海洋里的近月钟已被解剖得很细:环节虫的月亮钟由夜间光照授时,在昼夜钟被 CK1δ/ε 抑制打乱后照常运转(PMID 24075994、PMID 39506296);海蚊 Clunio 的自由运行半潮月周期由”数昼夜钟周期数”决定(PMID 38764210)。人类侧的论证这几年确实在增加——2026 年有文章系统给出”人类存在近月钟”的演化与生理论据(PMID 42380750)——但分子实体仍然没找到。

于是 backlog 上这条课题的逻辑很清楚:既然 XBP1s 的 12 小时振荡器被论证为远古潮汐钟的陆地遗留(PMID 33449146、PMID 41845938),而潮汐钟与近月钟在海洋生物里常在同一体内共存(PMID 24075994),那么”两者是否共享上游开关”就是定位人类近月钟最便宜的一刀。候选上游有三个:ER 应激负荷本身(UPR 在生殖生理中本就随周期剧烈波动,PMID 28397763)、公共激酶节点(CK1 家族)、以及频率拍频关系。

二、这一刀砍下去会先撞上三个坑

坑一:共变不等于共享。 这是最要紧、也最容易被跳过的一步。两个完全独立的振荡器,只要频率接近整数比,在有限长度的数据上就会表现出共变——拍频与锁相在观察层面无法与”共享上游”区分。12 小时与 29.5 天恰好接近 1:59 的整数关系。所以”12 小时成分与月级成分共变”这个结果本身不构成共享上游的证据;必须先用主动扰动(敲低 XBP1s、抑制 CK1δ/ε)去看两类成分的阻尼率与相移是否同步改变,才谈得上共享。这道因果门槛不设,后面所有数据都白做。

坑二:12 小时谱峰自己就有三种来源。 2026 年 CHORD 这篇方法学文章把这件事钉死了:转录组里一个 12 小时谱峰可能来自三种机制完全不同的过程——非正弦 24 小时波形的二次谐波、带独立相位的自主 12 小时振荡器(如 IRE1α–XBP1s 轴)、或两个反相 24 小时过程”合成减降解”造成的表观 12 小时(PMID 42703951)。三者在生物学含义上彼此相反,但对 JTK_CYCLE、RAIN、Cosinor 这类标准检测器完全不可区分(PMID 20876817、PMID 25326247)。不做三元分解就开始谈”XBP1s 的 12 小时节律”,本身就是方法错误。

坑三:30 天测不出 29.5 天。 这条最直接:30 天培养光谱只有大约一个完整周期。一个周期给得出”升起来又落下去”,给不出周期长度,也给不出存在性检验的统计量。要估计 29.5 天周期并做显著性检验,最低限度需要 3-4 个完整周期,即 90-120 天以上的连续采样;而 100 天以上的连续培养本身就要面对细胞状态漂移(传代累积、汇合度、营养耗竭、衰老),长周期的”趋势项”与”周期项”在有限数据上很难分离——哺乳动物细胞在培养中的钟基因表达特性与组织内本就不一致(PMID 16448219)。也就是说,原课题写法里的”≥30 天长程表达谱”在数字上是不成立的。

三、还有一个隐藏的授时源

培养箱里没有月亮,但有一个”实验者的作息”。 换液、传代、培养基批次更换、CO₂ 罐更换、周末与工作日的人力差异——这些都是周期性的人为输入。培养箱里的一批细胞,实际上一直跑在一个由实验者排班构成的外部周期里。这些人工周期的频率往往与 29.5 天毫无关系(换液若每 3-4 天一次,引入的是 3-3.5 天成分),但”与目标频率不同”不等于“不会污染目标频率的估计”——非整数比的人为周期在有限采样下通过窗函数泄漏进长周期估计是常事。所以这个实验要成立,必须同时记录并报告所有人工周期的时序,做混杂回归;否则看到 29.5 天成分时,第一嫌疑人是排班表,不是月亮钟。

四、回到”两只钟”

现在可以说回干支了。古人处理多尺度时间从来不靠”一只总钟”,而是摆几套互质的周期让它们自己错开——甲子六十,是十与十二的最小公倍数,没有任何一只”六十年钟”在走。这套思路的好处是:它天然不要求两套周期共享上游。 而现代生物学检验两条节律是否独立时,默认思路恰好相反:相关则必有共因。易学这边给出的提醒是第三种可能——它们可以不共因,只共频,于是看起来相关。

这正好是这条课题最该先解决的事:在动手之前,先把”共变”与”共享”之间那道因果门槛说清楚,再把这道门槛实验化。干支在这里不提供证据,只提供一个提问的角度。

证据与局限

已验证: ①哺乳动物存在细胞自主、独立于昼夜钟的 XBP1s/IRE1α 12 小时振荡器(PMID 31935211、PMID 37205336、PMID 41845938、PMID 33277471);②该振荡器由核斑点相分离动力学实现,并受 RBBP5 表观调控(PMID 34985945、PMID 42848836、PMID 39314427);③12 小时谱峰存在”谐波 / 自主振荡 / 合成-降解交叉”三种机制来源,标准检测器(JTK_CYCLE、RAIN、Cosinor)无法区分(PMID 42703951、PMID 20876817、PMID 25326247);④海洋环节虫近月钟独立于昼夜钟,由夜间光照授时(PMID 24075994、PMID 39506296);⑤海蚊半潮月周期由数昼夜钟周期数决定(PMID 38764210);⑥UPR/ER 应激参与生殖生理的周期调节(PMID 28397763)。

部分验证: ①”XBP1s 12 小时振荡器是远古潮汐钟的陆地遗留”有演化论据支撑(PMID 33449146、PMID 41845938),但”潮汐钟与近月钟共享上游”在同一生物体内尚无直接分子证据;②人类存在近月钟的论证已系统化(PMID 42380750),但分子实体空缺;③培养细胞中确有独立相位、可自主运行的钟基因节律(PMID 15620658、PMID 24498336),说明细胞层面做长程节律检测在方法上可行——但可行距离”测得出 29.5 天”还差一个数量级的时间。

待验证: ①30 天级培养能否检出 29.5 天成分(需先做”周期数 × 采样密度 × 噪声”的功效标定);②培养箱内人工周期(换液/传代/批次/实验者作息)向长周期估计泄漏的量级;③主动扰动(XBP1s 敲低 / CK1δ/ε 抑制)下,12 小时与月级两类成分的阻尼率-相移是否同步改变(共享上游的因果判据);④ER 应激诱导剂(衣霉素、毒胡萝卜素)梯度本身是否足以在细胞系里重建 12 小时振荡的振幅-相位结构(PMID 22038282、PMID 22852048、PMID 31630283)。

相关阅读

证据等级: B(XBP1s 12 小时振荡器的分子身份、CHORD 的谱峰三元分解、环节虫近月钟的独立性均有 PubMed 直接支撑;但”与近月钟共享上游”本身仍是待检验假说,且本文指出原实验设计在时间尺度上存在硬伤)
参考文献: PMID 31935211, PMID 37205336, PMID 41845938, PMID 33277471, PMID 34985945, PMID 42848836, PMID 39314427, PMID 42703951, PMID 20876817, PMID 25326247, PMID 24075994, PMID 39506296, PMID 38764210, PMID 42380750, PMID 33449146, PMID 28397763, PMID 16448219, PMID 15620658, PMID 24498336, PMID 22038282, PMID 22852048, PMID 31630283


免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

Covariation Does Not Equal Sharing: The 12-Hour Oscillator and Monthly Component of XBP1s, This Cut Hits the 30-Day Number

Evidence level: B

Using cell lines to perform “ER stress induction → simultaneous observation of 12-hour and 29.5-day components” is the cheapest direct test, but before proceeding, three pitfalls must be addressed: covariation does not imply sharing, the 12-hour spectral peak itself has three distinct sources, and a 30-day culture cannot measure a 29.5-day cycle—the third is a hard flaw at the numerical level.

Main Text

The Su Wen · Liu Wei Zhi Da Lun states, “Heavenly Qi begins with Jia; Earthly Qi begins with Zi. When Zi and Jia combine, it is called the establishment of the year.” Chinese calendrical systems possess a rarely discussed characteristic: they never attempt to construct a single “master clock” to record time. Instead, they establish two cycles—the Ten Stems and Twelve Branches—allowing them to rotate independently. Unique moments are marked by the combinations generated when these two cycles offset each other. The sixty-year Jiazi cycle is not the result of one large clock completing a revolution; it is the least common multiple of two smaller clocks.

The subject of this paper collides precisely with this “two-cycle” structure.

I. The Proposition to be Tested

Previous articles have clarified two lines of inquiry.

The first is the 12-hour oscillator of XBP1s. In mammalian livers, there is a cell-autonomous rhythm independent of the circadian clock, centered on the transcription factor XBP1s and its upstream sensor IRE1α (PMID 31935211, PMID 41845938). It drives the entire “central dogma information flow”—transcription initiation, mRNA processing, ribosome biogenesis, and protein folding (PMID 31935211)—through phase separation dynamics of nuclear speckles to achieve a 12-hour rhythm (PMID 34985945), and is directly linked to lipid metabolism and membrane fluidity homeostasis (PMID 33277471). By 2026, the subunit RBBP5 of histone methyltransferase complexes has been proven as an upstream epigenetic regulatory node for this oscillator (PMID 42848836, PMID 39314427). The most critical corroborating evidence is that after removing the core circadian clock, ~12-hour mRNA and protein metabolic rhythms persist (PMID 37205336)—this oscillator is truly autonomous, not a shadow of the circadian clock.

The second line of inquiry is that the molecular entity of the human near-lunar clock remains empty to date. Near-lunar clocks in marine environments have been dissected in detail: the lunar clock in annelids is entrained by nocturnal light, continuing to operate even when disrupted by CK1δ/ε inhibition of the circadian clock (PMID 24075994, PMID 39506296); the free-running semilunar cycle in the sea mosquito Clunio is determined by “counting circadian cycles” (PMID 38764210). Arguments regarding the human side have indeed increased in recent years—2026 saw an article systematically providing evolutionary and physiological arguments for “the existence of a near-lunar clock in humans” (PMID 42380750)—but the molecular entity remains undiscovered.

Thus, the logic on this backlog is clear: since the XBP1s 12-hour oscillator has been argued as a terrestrial remnant of ancient tidal clocks (PMID 33449146, PMID 41845938), and tidal clocks often coexist with near-lunar clocks within the same organism in marine species (PMID 24075994), determining whether “they share an upstream switch” is the cheapest cut to locate the human near-lunar clock. There are three candidate upstream nodes: ER stress load itself (UPR fluctuates剧烈 in reproductive physiology across cycles, PMID 28397763), a shared kinase node (CK1 family), or frequency beat relationships.

II. Three Pitfalls This Cut Will Hit First

Pitfall 1: Covariation does not imply sharing. This is the most critical yet often skipped step. Two completely independent oscillators, if their frequencies are close to integer ratios, will exhibit covariation over finite data lengths—beat frequency and phase locking are indistinguishable from “shared upstream” at the observational level. The 12-hour and 29.5-day cycles are close to a 1:59 integer ratio. Therefore, the result of “covariation between the 12-hour component and lunar component” does not constitute evidence for a shared upstream; one must first observe whether the damping rates and phase shifts of both component types change synchronously via active perturbation (XBP1s knockdown, CK1δ/ε inhibition) before claiming sharing. Without setting this causal threshold, all subsequent data is wasted.

Pitfall 2: The 12-hour spectral peak itself has three sources. A 2026 CHORD methodological article cemented this: a 12-hour spectral peak in transcriptomics may arise from three fundamentally different mechanisms—secondary harmonics of non-sinusoidal 24-hour waveforms, an autonomous 12-hour oscillator with independent phase (such as the IRE1α–XBP1s axis), or an apparent 12-hour cycle caused by “synthetic subtraction and degradation” of two anti-phase 24-hour processes (PMID 42703951). These three differ biologically but are completely indistinguishable to standard detectors like JTK_CYCLE, RAIN, and Cosinor (PMID 20876817, PMID 25326247). Discussing “XBP1s’s 12-hour rhythm” without ternary decomposition is itself a methodological error.

Pitfall 3: A 30-day culture cannot measure a 29.5-day cycle. This is most direct: a 30-day culture spectrum yields only approximately one complete cycle. One cycle provides “rising and falling,” but not the period length, nor statistical power for existence testing. To estimate a 29.5-day cycle and perform significance testing, a minimum of 3-4 complete cycles is required, i.e., continuous sampling over 90-120+ days; continuous culture beyond 100 days faces cell state drift (passage accumulation, confluence, nutrient depletion, senescence), making it difficult to separate “trend terms” from “cycle terms” in finite data—mammalian cell clock gene expression characteristics in culture are inherently inconsistent with those in tissues (PMID 16448219). In short, the “≥30-day long-term expression profile” in the original proposal is numerically invalid.

III. A Hidden Entrainment Source

There is no moon in the incubator, but there is an “experimenter’s schedule.” Media changes, passaging, media batch replacements, CO₂ cylinder changes, and differences in human resources between weekends and weekdays are all periodic artificial inputs. A batch of cells in an incubator actually runs continuously within an external cycle constructed by the experimenter’s roster. The frequencies of these artificial cycles often have no relation to 29.5 days (e.g., if media is changed every 3-4 days, a 3-3.5 day component is introduced), but being “different from the target frequency” does not equal “not polluting the estimation of the target frequency”—non-integer ratio artificial cycles frequently leak into long-period estimates via window functions under finite sampling. Therefore, for this experiment to hold, all artificial cycle timings must be recorded and reported, and confounding regression performed; otherwise, when a 29.5-day component is observed, the first suspect is the roster, not the lunar clock.

IV. Returning to “Two Clocks”

We can now return to the Stems and Branches. Ancient people handled multi-scale time not with a “single master clock,” but by setting several coprime cycles to offset each other independently—the sixty-year Jiazi is the least common multiple of ten and twelve, with no “sixty-year clock” moving. The advantage of this approach is: it naturally does not require the two cycles to share an upstream source. Conversely, modern biological testing of whether two rhythms are independent defaults to the opposite logic: correlation implies a common cause. The reminder from Yijing Studies offers a third possibility—they may not share a common cause, only share frequency, thus appearing correlated.

This is exactly what this project should resolve first: before acting, clarify the causal threshold between “covariation” and “sharing,” then experimentalize it. The Stems and Branches provide no evidence here, only a perspective for questioning.

Evidence and Limitations

Verified: ① Mammals possess a cell-autonomous XBP1s/IRE1α 12-hour oscillator independent of the circadian clock (PMID 31935211, PMID 37205336, PMID 41845938, PMID 33277471); ② This oscillator is realized via nuclear speckle phase separation dynamics and regulated epigenetically by RBBP5 (PMID 34985945, PMID 42848836, PMID 39314427); ③ The 12-hour spectral peak has three mechanistic sources (“harmonic / autonomous oscillator / synthetic-degradation crossover”), which standard detectors (JTK_CYCLE, RAIN, Cosinor) cannot distinguish (PMID 42703951, PMID 20876817, PMID 25326247); ④ Annelid near-lunar clocks are independent of circadian clocks and entrained by nocturnal light (PMID 24075994, PMID 39506296); ⑤ The semilunar cycle in sea mosquitoes is determined by counting circadian cycles (PMID 38764210); ⑥ UPR/ER stress participates in periodic regulation of reproductive physiology (PMID 28397763).

Partially Verified: ① “The XBP1s 12-hour oscillator is a terrestrial remnant of ancient tidal clocks” has evolutionary support (PMID 33449146, PMID 41845938), but “shared upstream between tidal and near-lunar clocks” lacks direct molecular evidence within the same organism; ② Arguments for human near-lunar clocks are systematic (PMID 42380750), but the molecular entity is missing; ③ Independent-phase, autonomously running clock gene rhythms exist in cultured cells (PMID 15620658, PMID 24498336), indicating methodological feasibility for long-term rhythm detection at the cellular level—but there is still an order-of-magnitude time gap before measuring a 29.5-day cycle.

To be Verified: ① Whether 30-day-level cultures can detect 29.5-day components (requires first conducting “cycle count × sampling density × noise” power calibration); ② The magnitude of leakage from artificial cycles within incubators (media changes/passaging/batches/experimenter schedule) into long-period estimates; ③ Whether damping rates and phase shifts for the 12-hour and lunar components change synchronously under active perturbation (XBP1s knockdown / CK1δ/ε inhibition)—the causal criterion for shared upstream; ④ Whether gradients of ER stress inducers (calyculin A, bryostatin) alone are sufficient to reconstruct the amplitude-phase structure of the 12-hour oscillator in cell lines (PMID 22038282, PMID 22852048, PMID 31630283).

Related Reading

Evidence Level: B (The molecular identity of the XBP1s 12-hour oscillator, CHORD’s ternary spectral decomposition, and the independence of annelid near-lunar clocks have direct PubMed support; however, “sharing upstream with the near-lunar clock” remains a hypothesis to be tested, and this paper points out a hard flaw in the original experimental design regarding the timescale)
References: PMID 31935211, PMID 37205336, PMID 41845938, PMID 33277471, PMID 34985945, PMID 42848836, PMID 39314427, PMID 42703951, PMID 20876817, PMID 25326247, PMID 24075994, PMID 39506296, PMID 38764210, PMID 42380750, PMID 33449146, PMID 28397763, PMID 16448219, PMID 15620658, PMID 24498336, PMID 22038282, PMID 22852048, PMID 31630283


Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *