What carries over, and what does not
APPLICATION NOTEWhat carries over from the ice clock to DNA repair is the structure of the question and the logic of the controls. Time-Gel's clock prediction does not: inside a cell it would be far too small to measure. This page treats repair as a biophysics problem that Time-Gel's question sharpens, not as evidence for Time-Gel.
The starting point is the decision recorded in the ice clock protocol: the clock looks at the moment of change, not at the resulting state (ρI = σ/kB). In repair this means the act of repair matters, not the repaired DNA, and the cost of that act is energy.
| From the ice clock | In DNA repair | Carries over? |
|---|---|---|
| Five roles: source, carrier, processor, state builder, meaning-giver | Damage, DNA strand, recognition proteins, excision, cell response | Yes |
| Search time and the first irreversible step | Finding the damage and cutting on both sides | Yes |
| Control B2: same entropy, no record | Repair proteins burning ATP without repairing | Yes, the core of the design |
| Clock slowing | Phase shift inside a repairing cell | No: roughly 10−11 s, unmeasurable |
| Circadian clock | Repair capacity changing over the day | Shared word only; a chemical clock, not Time-Gel's local time |
Aziz Sancar's field in brief
Aziz Sancar shared the 2015 Nobel Prize in Chemistry with Tomas Lindahl and Paul Modrich for mechanistic studies of DNA repair. Four parts of that work matter here.
- Nucleotide excision repair (NER). UV damage such as a cyclobutane pyrimidine dimer is found, cut on both sides and removed as a short piece: about 12–13 nucleotides in bacteria (UvrABC), about 24–32 in humans. The gap is resynthesised.
- Two search routes. In global genome repair, XPC has to find the damage. In transcription-coupled repair, a stalled RNA polymerase is already there: the owner is waiting before the information arrives.
- Photolyase. Reverses the damage directly with light, without cutting. The record is erased and leaves no trace.
- XR-seq and the circadian clock. Sequencing the excised pieces maps where and when repair happens across the genome; repair capacity in mouse tissues rises and falls over the day with XPA.
These facts are summarised from memory and must be checked against primary sources before any publication. This page does not represent the views of Aziz Sancar or his laboratory.
Five roles in DNA damage
In repair the five roles are more separate than in ice: the source comes from outside (a UV photon) and the owner is a machine independent of the information.
| Role | In NER | Reversible? | Measurable counterpart |
|---|---|---|---|
| Source | UV photon and the CPD it makes | Not on its own; photolyase reverses it | Damage maps, lesion counts |
| Carrier | DNA strand, carrying the damage to the next read | — | Position of the damage |
| Processor | XPC binding and unbinding, TFIIH verification | Yes: a protein bound in the wrong place leaves | Single-molecule dwell times, ATPase rate |
| State builder | Dual incision and resynthesis | No: the first irreversible step | Excised fragment (XR-seq, excision assay) |
| Meaning-giver | Cell response: checkpoints, transcription restart | — | Partly: response markers, not meaning |
Experiment: the cost of finding the owner
The question: how much energy does it cost for information to find its correct owner, and does the rate of wrong owners fall as more energy is spent? The ice clock's B2 control becomes the core: the same proteins spending energy on undamaged DNA build no record.
System. Reconstituted human NER (XPC-RAD23B, TFIIH, XPA, RPA, XPF-ERCC1, XPG) and bacterial UvrABC, on DNA carrying one CPD and on the same sequence undamaged. Measured: correct excision rate, false excision rate, ATP hydrolysis rate, and heat where possible. Entropy production follows from ATP: σ ≈ (hydrolysis rate × |ΔGATP|) / T.
| Arm | What happens | What it controls |
|---|---|---|
| A · Correct owner | Full system, damaged substrate | Test arm |
| B · Ownerless spending (B2) | Full system, undamaged substrate | Energy spent and false cuts with no record |
| C · Energy dial | ATP/ADP ratio varied at fixed total nucleotide | The curve of spending against accuracy |
| D · Verification off | ATPase-dead XPB or XPD | Which step the energy goes to |
| E · Returning repair | Photolyase, direct reversal by light | Cost of repair without cutting or trace |
| F · Search in cells | XR-seq time course after a UV pulse: normal, XPC-deficient, CSB-deficient cells | Search time with the owner waiting or absent |
In the cell arm the birth moment is the UV pulse, and unlike the ice clock the experimenter chooses it. Pulse times are randomised and circadian phase is recorded, because repair capacity changes over the day.
Predictions, rivals and falsification
Here Time-Gel and conventional biophysics predict the same thing; no discrimination is claimed. Time-Gel contributes the control design, not a prediction.
Conventional model: kinetic proofreading (Hopfield 1974, Ninio 1975). An enzyme can make fewer errors than equilibrium allows by adding an irreversible, energy-consuming verification step. Each step lowers the error rate by at most one equilibrium discrimination factor and costs ATP. Its application to NER is open.
| Arm | Kinetic proofreading | Time-Gel (rate reading) | Discriminates? |
|---|---|---|---|
| C · less energy | More false cuts | Same | No |
| B · undamaged DNA | Futile ATP use | Same; entropy without record | No |
| D · ATPase dead | Verification and cutting stop | Same | No |
| Clock shift in the cell | None | About 10−11 s | In principle yes, in practice no |
Falsification of this experiment's own hypothesis (not of Time-Gel): a flat curve in arm C means NER damage selection is not energy-limited proofreading; no measurable ATP use in arm B means the B2 analogy fails in this system; accuracy beyond the proofreading limit means the model is incomplete or the measurement is wrong, and is not counted for Time-Gel.
A first look at real data
The paper adds one thing this page did not have: a reanalysis of a public CPD Damage-seq time course in human skin fibroblasts (Hu, Adebali, Adar & Sancar, PNAS 2017). Damage was counted in 100 kb bins at 0 h, 1 h, 8 h, 24 h, 36 h and 48 h after UV, two replicates each.
- Replicates agree. Rank correlation 0.94–0.99 between replicates, except one low-depth 48 h sample.
- Some damage has not found its owner after 36 hours. About 17% of the genome sits in bins whose damage has not gone down at all.
- That damage sits in blocks, not at random. Mean block length about 560 kb, longest 5.6 Mb.
- Chromosomes differ. 5–7% on chr22, 17, 19 and 1; 23–29% on chr18, 13, X and 21.
Conventional biology explains this as slower repair in closed chromatin and in regions without transcription. The analysis describes the data; it does not test Time-Gel.
Feasibility and limits
The biochemistry arms (A–E) can be done today in a laboratory that already reconstitutes NER; the cell arm (F) needs XR-seq. No arm tests Time-Gel.
Rough clock estimate (all inputs assumed): about 105 lesions per cell after UV, 10–100 ATP per lesion, about 20 kBT per ATP. At most about 2 × 108 kB in about 2 × 10−15 m³, a phase step of about 5 × 10−11 s at ρcrit = 1033 s−1m−3. No clock works at that level inside a cell.
References (to be verified): J. J. Hopfield, PNAS 71, 4135 (1974) · J. Ninio, Biochimie 57, 587 (1975) · A. Sancar, Nobel lecture (2015) · Hu et al., Genes & Development (2015).