Sage — Sulindac (The Kit)
Hypothesis: Msr/sulfide and COX-independent preconditioning on one scaffold — residual after Fuel · Circadian · Strength.
The Kit · Coach · Surfer · Aspirin
Pharmacologic preconditioning, three clocks of mitochondrial set-point, and the residual value of a mid-century NSAID after fuel, circadian, and strength hormesis are already in place
**Val Zudans, MD** · working synthesis, September 2026
*Drawing on the laboratory work of Herbert Weissbach and colleagues, the SDM / reset literature, and the EnSurf metabolic model. A map of a molecule onto an ontology of ATP-generating capacity.*
*This document is a working scientific hypothesis essay written for intellectual exploration. It is not medical advice, not a prescription, not a clinical protocol, and not a recommendation to start, stop, or change any medication --- including sulindac (Clinoril) or any other NSAID.*
*Sulindac is a prescription nonsteroidal anti-inflammatory drug with established gastrointestinal, renal, and cardiovascular risks. The mitochondrial, preconditioning, retinal, anti-aging, and cancer-related mechanisms discussed here are drawn largely from preclinical work and from analogy to other stimuli (exercise, ischemic preconditioning, subthreshold laser). They have not been validated as approved indications. Low-dose or intermittent schedules described in the text are sketches for how a hypothesis might be tested, not doses to take.*
*Nothing in this essay should be used for self-treatment of aging, cancer, retinal disease, or metabolic performance. Decisions about prescription drugs, laser treatment, or any off-label use belong with a licensed clinician who knows the individual patient, the indication, and the contraindications. The author accepts no liability for actions taken from this text.*
> *We are born with perfect biology for an ancestral energy environment. > What we call disease is usually chronic compensatory mode after the > engine that makes ATP has been compromised. The interesting drugs are > the ones that touch the engine --- not another coat of paint on the > dashboard.*
I. Thesis
Sulindac is usually filed under "old NSAID, colon-polyp curiosity, GI risk." That filing is not wrong. It is incomplete. At exposures well below the anti-inflammatory label dose, the same molecule behaves as a chemical ischemic-preconditioning agent: a brief, sub-lethal mitochondrial redox signal that normal cells interpret through PKCε, mitoKATP channels, and a late transcriptional program (iNOS, heat-shock proteins), and that many cancer cells interpret through PKCδ, membrane-potential collapse, and apoptosis.
The metabolic hypothesis is this: sulindac does not create new mitochondria, does not replace fuel quality, does not entrain clocks, and does not substitute for mechanical hormesis. What it may do --- if the trigger is real at human low dose --- is temporarily raise the defensive set-point of the existing mitochondrial population, and, in cells already trapped in Warburg inflexibility, convert that same set-point change into a lethal oxidative load. Those are engine-level actions. They are also mostly redundant with a well-run EnSurf stack. The residual case, if there is one, is narrow: a pulsed chemical echo of late-phase preconditioning on days the organism is not already generating that signal; a possible selective pressure against metabolically incompetent clones; and a historically documented, COX-independent chemopreventive signal that is not fully explained by lifestyle.
Daily NSAID occupancy is the wrong shape for that hypothesis. A pulse that respects Clock 2 (48--72 hours of quiet afterward) is the shape Weissbach documented. A pulse every one to three months is the shape SDM documented for a different stimulus in a different delivery geometry. Those are different experiments. Monthly or quarterly sulindac is not a translation of the 48-hour rat feed. Whether 50 mg is enough of either pulse in a human is unknown.
II. The engine ontology
Aging is the progressive loss of ATP-generating capacity --- falling mitochondrial density, rising heteroplasmy and damaged-organelle burden, shrinking tissue mitochondrial volume. Chronic disease is what that loss looks like when the organism spends years in compensatory mode: hyperinsulinemia, inflammatory tone, apoB elevation, nocturnal non-dipping, sarcopenia, fibrotic remodeling. Cancer is a further distortion of the same engine. The Warburg phenotype is not a curiosity. It is what a clone does when it trades respiratory flexibility for biomass under stress.
Establishment care reads the dashboard. The EnSurf model reads the engine. Fuel quality and timing, circadian entrainment, and strength-plus-aerobic hormesis are the three non-negotiable actuators because they change mitochondrial number, quality, and use simultaneously. Pharmacologic tools are judged by a harsh standard: after those three are in place, what residual degree of freedom is left?
Sulindac is interesting only if it occupies a degree of freedom those three do not fully occupy. To decide that, the molecule has to be taken apart into the two stories that have been collapsed under one trade name, and preconditioning has to be taken apart into the three clocks that have been collapsed under one word.
III. Two stories on one scaffold
Story A --- Msr substrate and the sulfide
Methionine is the most oxidation-prone of the proteinogenic amino acids. Oxidation produces methionine sulfoxide and a new chiral center: Met-S-O and Met-R-O. Unrepaired, oxidized methionines inactivate proteins and register chronic oxidative load. The repair system is the methionine sulfoxide reductase family. MsrA reduces the S-epimer (free and protein-bound, and some small-molecule methyl sulfoxides). MsrB reduces the R-epimer, mostly on protein. A catalytic cysteine attacks the sulfoxide; methionine is released; the enzyme is left as a sulfenic acid; resolving cysteines and thioredoxin/NADPH regenerate the reduced enzyme. Msr is therefore a catalytic protein-repair and ROS-buffering system, not a stoichiometric antioxidant.
Sulindac is a methyl sulfoxide prodrug. Weissbach's group showed it is a direct Msr substrate. MsrA reduces the S-epimer to sulindac sulfide --- the COX-active NSAID. A distinct MsrB-like activity reduces the R-epimer. Liver, kidney, brain, mitochondria, and microsomes all perform the reduction. That fact generated a second idea, which they patented: a recyclable small-molecule sulfoxide that is oxidized by ROS and reduced back by Msr, turning the Msr--thioredoxin axis into a catalytic ROS sink.
This story explains why sulindac is a prodrug, why the sulfide is the anti-inflammatory species, and why the molecule sits inside a protein-repair network that already declines with age. It does not explain the preconditioning phenotype.
Story B --- COX-independent pharmacologic preconditioning
The heart, RPE, and cancer-cell work that makes sulindac metabolically interesting is largely independent of both Msr and COX. Sulindac sulfone --- neither an Msr substrate nor a COX inhibitor --- still protects normal cells and still sensitizes many cancer cells to oxidative stress. Other NSAIDs at matched concentration (aspirin, ibuprofen) do not reproduce the sensitization. Chelerythrine, a PKC blocker, abolishes cardiac protection. ROS scavengers blunt it. The operational identity is ischemic preconditioning without the ischemia.
Keep the two stories separate. Confusing them produces bad dosing. Story A wants some continuous presence of a sulfoxide substrate. Story B wants a brief redox trigger, then a program, then quiet. Most of what follows is Story B. Story A is a bonus that may or may not operate at the doses under discussion. Story B itself splits across three clocks.
--------------------- -------------------------- --------------------------- **Function** **Needs Msr → sulfide?** **Needs COX inhibition?**
**NSAID Yes Yes anti-inflammatory**
**Catalytic ROS Yes, as substrate No cycling via Msr**
**Preconditioning / No No mito protection**
**Cancer-cell No (sulfone works) No sensitization to oxidants** --------------------- -------------------------- ---------------------------
**III.A. Three clocks of preconditioning**
The word preconditioning has been asked to do too much work. A sub-lethal insult today can make tissue harder to kill in an hour, in a day, and in a season, by different carriers. Treating those as one phenomenon produces one schedule, and that schedule will be wrong for two of the three clocks.
**Clock 1 --- early, post-translational**
Classic or first-window ischemic preconditioning opens within minutes of a brief ischemic or pharmacologic trigger and fades within one to three hours. The carriers are already-present proteins and channels: adenosine and opioid receptors, PKC (especially ε in rodents), mitoKATP and sarcKATP, a pulse of signaling ROS, and acute inhibition of mPTP opening. No new transcription is required. This is not the design of the Weissbach feeding experiments, which used 48 hours of exposure and looked for induced proteins. It is the first minutes after a sprint, a remote-limb cycle, or an adenosine bolus.
**Clock 2 --- late-phase protein expression**
The second window opens at about 12--24 hours, is carried by newly made proteins, and lasts on the order of 48--72 hours, sometimes cited out to 72--90. Documented mediators include iNOS, COX-2, aldose reductase, HSP27/70, and MnSOD. Opening of KATP channels remains necessary for the infarct-limiting phenotype to become manifest. This is the clock Weissbach mapped for sulindac. After 48 hours of low-dose feeding, rat myocardium showed PKC-dependent induction of iNOS and HSP27, and isolated hearts (no drug in the perfusate) released less LDH and had smaller infarcts during ischemia--reperfusion. Chelerythrine abolished both the proteins and the protection. ROS scavengers blunted it. In RPE, the same laboratory showed PKCε dependence, mitoKATP dependence, PKG, iNOS, HSP70, and PPARα.
Clock 2 is real. It is also short. iNOS and HSP proteins do not live for a quarter. A dosing cadence built only on Clock 2 cannot be monthly.
**Clock 3 --- reset, memory, and salvage**
Clock 3 is a family of slower traces, not a single enzyme. It is why "is preconditioning only protein expression?" has to be answered no.
***HSP activation versus HSP induction***
Making more HSP molecules is Clock 2. Changing the conformation and repair kinetics of existing HSPs in already-stressed cells is different. SDM literature treats the thermal spike as a kinetic activator of HSP-mediated protein repair, especially in sick RPE whose proteome is already misfolded and short-lived. Catalytic acceleration of repair can outlast the inducing transcript. The repair of the proteome --- including mitochondrial proteins --- is a set-point change, not a pulse of one Western blot band.
SDM is framed as physiologic "reset to default": after HSP and unfolded-protein-response engagement, RPE cytokine output normalizes (VEGF down, PEDF up, and a broader mediator panel), Müller and microglial tone quiet, and retinal autoregulation improves. That is a new neighborhood, not a two-day protein. Clinical function can move within 24 hours to a month, hold near maximum for about three to four months, and drift toward baseline by roughly six to nine months. That is why Vision Protection Therapy is repeated every three to four months. Some patients notice fade at six to eight weeks; retreatment restores the gain. This is the existence proof, in human RPE, that a single sublethal preconditioning-like stimulus can leave a working phenotype measured in months.
Ischemic preconditioning and repetitive hypoxic conditioning change DNA methylation, histone marks, and miRNA pools. The tolerant brain does not merely make more of the same protective proteins; it reprograms the transcriptional response to a later insult --- often a net downregulation of excitability and metabolic demand. Repetitive hypoxic conditioning can stretch neuroprotection from days to months. In some mouse models, resilience is even visible in untreated first-generation offspring. Mitohormesis work now links a brief mitochondrial superoxide signal to citrate export, cytosolic acetyl-CoA, histone acetylation, and durable adult cardioprotection. That is an epigenetic bridge from a short ROS sentence to a months-long phenotype. It is mechanistic permission for Clock 3. It is not proof that oral sulindac writes that bridge.
Transient redox stress can leave an altered mitochondrial network architecture --- mitoengrams --- that progeny cells inherit until the architecture is erased. Cellular memory is not only chromatin. It is also the geometry of the organelle that feeds chromatin.
Myeloid cells remember sterile or infectious stimuli for weeks to months through metabolic and epigenetic rewiring. A systemic ROS or nitric-oxide pulse can in principle train that compartment. This is a Clock-3 candidate for remote effects, not a measured sulindac endpoint.
If a pulse prevents a wave of photoreceptor or myocyte death, the denominator of surviving cells is permanently different. Those cells are still there in December. That looks like long preconditioning and is not. It is a one-time change in stock. Weissbach light-damage retinas that kept more rhodopsin five days after an insult, and rd10 notes of slower early degeneration after intravitreal drug, are compatible with salvage. They are not a 90-day standing-protection study after one systemic pulse.
**What Weissbach measured, and what he did not**
The cardiac experiments fed rats for 48 hours, then isolated the heart and ran ischemia--reperfusion with no drug in the circuit. Protection was present at that moment. They did not take animals off drug and challenge them at 30 or 90 days. The retinal light-damage work was pretreatment, then an acute photic insult, then histology a few days later. The rd10 note was slowing of an ongoing degeneration after intravitreal drug, assessed on the short postnatal clock of that model (rods are largely gone by P60). None of those designs is a one-to-three-month standing-protection study after a single systemic pulse. A months-long protective phenotype after oral sulindac has not been shown; the longer window documented in human RPE belongs to SDM, and some apparent duration after an acute retinal insult is structural salvage rather than a standing preconditioning program.
SDM deposits energy in RPE. A 50 mg tablet does not. Clock 3 in human RPE after laser is not transferable to Clock 3 after oral sulindac by analogy alone.
IV. The pathway in normal cells
Late-phase ischemic preconditioning is an old observation: a sub-lethal insult today makes the tissue harder to kill tomorrow. The early window lasts minutes to a few hours and is post-translational. The late window opens at 12--24 hours, is carried by new proteins, and lasts on the order of 48--72 hours, sometimes cited out to 72--90. Sulindac, at low exposure, appears to enter that late program.
Trigger
A sub-lethal mitochondrial ROS pulse. In the heart-feeding experiments this was inferred from scavenger blockade. The pulse is the point. Too little and nothing is transcribed. Too much and you have injury rather than a signal. This is why concentration and schedule dominate the entire hypothesis.
Sensor and first kinase
PKCε translocates to mitochondria. A PKCε-selective peptide blocker reverses RPE protection. A broad PKC inhibitor (chelerythrine) reverses cardiac protection and blocks induction of the late proteins. PKCδ, the sibling isoform, is the injury-leaning kinase; the ε/δ ratio is a recurring motif in preconditioning literature. Normal cells, under this stimulus, lean ε.
Mitochondrial effectors
Downstream of PKCε: opening of mitochondrial ATP-sensitive potassium channels (mitoKATP), delayed opening of the mitochondrial permeability transition pore, preservation of membrane potential under a later insult. In some tissues PKG and PPARα sit on the same circuit (RPE protection required PPARα). The organelle is not being "stimulated" in the methylene-blue sense. It is being told to refuse the death pore.
Late transcriptional program
iNOS and HSP27 (heart) or HSP70 (RPE) rise in a PKC-dependent manner after 48 hours of feeding or pretreatment. iNOS is not optional in classical late PC; knockout or 1400W abolishes exercise- and ischemia-induced delayed protection. The program is polygenic. COX-2 itself is a documented mediator of late PC in the rabbit heart. That last fact will matter when an NSAID is proposed as the trigger.
Later transcriptional overlay
RNA-seq on sulindac and the non-COX analog MCI-100 implicated Wnt/β-catenin (pro-survival / proliferative) and suppression of TGF-β (less apoptotic drive). MCI-100 protected RPE at lower concentration than sulindac, protected rat hearts after oral dosing, and shared a large induced/repressed gene set with sulindac --- with the reported twist that it did not sensitize cancer cells the same way. The scaffold can be split. Protection and cancer killing are related but separable.
Net phenotype in a normal cell: the next oxidative hit produces less death, less LDH, a smaller infarct, a thicker outer nuclear layer, a higher chance the existing mitochondrial population is still making ATP on the other side of the insult.
V. The pathway in cancer cells
The same trigger is read differently. Cancer cells often already sit near a ROS threshold, run Warburg glycolysis even when oxygen is available, have less flexible residual respiration, and frequently carry an altered PKC isoform balance. In that context sulindac pretreatment plus an oxidant (TBHP, H2O2) or a metabolic forcible (dichloroacetate, which inhibits PDK and drives pyruvate into mitochondria) produces a large ROS burst, loss of ΔΨm on JC-1, cytochrome c release, JNK phosphorylation, PARP cleavage, and TUNEL-positive death.
The sensitization is COX-independent. Normal lung and colon fibroblasts under identical conditions are not killed and are often protected. The 2011 Weissbach-lab thesis stated the fork cleanly: normal cells execute a PKCε survival program; cancer cells execute a PKCδ apoptotic program. DCA synergy is the Warburg-exploitation move --- force the defective respirator to respire, then add a drug that makes that respiration chemically expensive.
This is not conventional cytotoxic chemotherapy. It is a metabolic selectivity argument: cells that cannot cleanly handle an extra mitochondrial ROS load die; cells that can mount PKCε/mitoKATP/iNOS do not. Selectivity will fail in cancers that have intact, flexible OxPhos, and it will fail if the oxidant partner is absent and the residual ROS increment is too small. The dish phenotype is a Clock-2-plus-oxidant phenomenon. It is not evidence of a months-long anti-cancer memory after one tablet.
A small clinical footnote from the same group: topical sulindac plus hydrogen peroxide produced responses in actinic keratoses. That is the oxidant-plus-sulindac idea applied to a precancer the clinician can see. It is not a systemic cancer protocol.
VI. Exercise already runs all three clocks
A single mild treadmill bout in mice is sufficient to put the heart into delayed protection 24 hours later. The cascade is eNOS-derived NO, PKCε translocation, KATP involvement, and iNOS-dependent protection the next day. Chelerythrine blocks it. eNOS or iNOS deletion blocks it. That is Clocks 1 and 2.
Exercise then does what a tablet does not. It induces PGC-1α and biogenesis. It engages mitophagy. It remakes vasculature. It secretes a myokine and EV language. Repeated bouts write Clock 3: epigenetic marks, a larger and cleaner mitochondrial population, trained recovery kinetics. A well-trained organism is therefore already living inside overlapping late-PC windows most days of the week, plus a chronic remodeling layer on top.
The implication for EnSurf is blunt. Strength and aerobic hormesis are not "lifestyle." They are the endogenous, multi-organ, multi-clock version of the pathway under discussion. A chemical IPC agent is a partial mimic of one slice of a program the person is already running.
VII. Hormesis, allostasis, and why schedule is the drug
Homeostasis returns a variable to a set-point. Allostasis changes the set-point so the system can meet a predicted load. Hormesis is a pulse that, after recovery, leaves the set-point higher. Daily occupancy of a receptor is usually homeostasis. A spaced trigger with a decaying program is usually hormesis. Which spacing is correct depends on which clock one is trying to write.
Late-phase preconditioning has a clock. The useful proteins are not permanent. They rise over a day and decay over two to four. That clock is why the concern about daily sulindac is not pedantic.
When pharmacologic preconditioning fails
Pack forty to sixty-five brief ischemic episodes into three or four days in a conscious rabbit and the protection of a single cycle is mostly lost. Infuse an adenosine A1 agonist continuously and tachyphylaxis appears. Give nitroglycerin daily and the endothelial preconditioning of a single exposure is gone within a week. The failure mode is real. It is not theoretical.
When it holds
Give the A1 agonist intermittently for ten days and rabbits stay protected without hemodynamic tachyphylaxis. Repeat remote IPC daily for a week in humans and endothelial protection equals a single day. Exercise every other day does not "use up" preconditioning; each bout re-induces Clock 2 after a trough and slowly writes Clock 3. SDM every three to four months is the clinical version of re-inducing Clock 3 after fade.
The COX-2 collision
COX-2 is a mediator of late PC. In conscious rabbits, antiplatelet-range aspirin left late protection intact; high anti-inflammatory-range aspirin abolished COX-2 activity and the protection with it. An NSAID proposed as an IPC trigger can therefore light the fuse and damp the effector. That collision is dose- and continuity-dependent. It is an independent argument against daily sulindac even if Story B is true. It is also a reason a drug given into another preconditioning window can, in principle, interfere rather than add.
Weissbach's own designs are easy to misread. The heart work was 48 hours of feeding before an acute insult --- a pulse. The NIA longevity proposal was continuous low-level chow --- a different bet, never publicly vindicated as a headline ITP win. Intermittent sulindac (two weeks on, two off) still cut tumors in an azoxymethane mouse model at half the total dose. None of that is human mitochondrial scheduling. All of it says unbroken occupancy is not required for every effect of the molecule.
Boundaries
No public prevention claims. No ophthalmic treatment claims from systemic mito notes. Hypothesis language.
A working hypothesis, open to correction and revised as new evidence arrives. Last revised October 17, 2026.