Aging & Senescence · §9

Are human aging genes special?

No. Across ten mammals spanning 3.8 to 122 years, every human aging-gene promoter γ sits inside the mammalian distribution (|z|<1), and there is no discontinuous γ longevity switch. The core gate TP53 is nearly flat (CV 3.5%; elephant ≈ human ≈ mouse). The real long-lived switch is off the γ axis: TP53 copy number.

Promoter γ for TP53, CDKN2A, FOXO3 and TERT was measured across ten mammals. Human is within one standard deviation on every gene, no clean short/long threshold appears, and TP53 varies by only CV 3.5% across a 4–211 year span. The elephant's cancer resistance instead comes from ~20 TP53 copies at essentially unchanged per-copy γ 1.4269.

This chapter is the package's direct answer to the question that motivated the study: are human aging genes different from other animals, or is there a special switch? It is the headline result.

The cross-species table

The same promoter-γ pipeline was run for the four masters across ten mammals from the brown rat (3.8 yr) to the human (122 yr), with the bowhead whale (211 yr) listed for span. Coverage is honest: bowhead has no annotated target promoters in NCBI Gene, and CDKN2A is unannotated in several species (the free-text hits were a different gene, CDKN2AIP).

Human is not special

On every gene the human value sits inside the mammalian spread, within one standard deviation, and no single-gene γ–lifespan trend survives a permutation test.

Three of four genes show overlapping short- and long-lived values (no clean threshold). TP53 in particular is nearly flat across the whole span (CV 3.5%): elephant, human and mouse share essentially the same core gate. All four masters do, however, lean the same way — every per-gene rank correlation with lifespan is positive — so the honest question is not whether one gene switches but whether that shared lean is real or an artifact.

Robustness audit: the shared lean is confound, not signal

A dedicated audit re-derives γ bit-for-bit and then corrects the confounds the per-gene tests ignored. First, on this panel γ is essentially a GC-content proxy (ρ(γ, GC) = 0.96–1.00 across genes), so the “γ axis” is a GC axis. Second, the four-gene combined lean — a test the original analysis never ran — reaches only a borderline raw ρ = 0.64 (permutation p = 0.054), and it does not survive correction: body-mass adjustment (the intrinsic-longevity residual) drops it to p = 0.16, and Felsenstein independent contrasts on a dated mammal tree collapse it to corr = 0.17 (p = 0.55) — the short-lived rodents are a single clade, so the raw trend was phylogenetic pseudoreplication. Out of sample, the four γ values classify long- vs short-lived species at only 80% leave-one-out accuracy (label-permutation p = 0.11), not above chance. The human-not-special / no-switch conclusion is therefore robust to every confound this small panel lets us test, and the original “no trend” wording is sharpened to “a weak shared lean that is fully confound-attributable.”

One honest caveat survives the audit. Of the four masters, TERT (telomere maintenance) is the most lifespan-leaning: it is the only gene whose lean is not removed by body-mass correction (size-corrected ρ = 0.54) and it carries the largest phylogenetic-contrast correlation (0.58). It still never reaches significance (p ≈ 0.12), so it is a lead for a larger panel, not a result — graded [O]. It is consistent with known biology that the telomere lever on lifespan is real but acts off the promoter-γ axis: large mammals suppress somatic telomerase to resist cancer (Gomes 2011), a regulatory and dosage effect rather than a promoter-sequence one — the same off-axis pattern as the TP53 copy-number switch below.

The real switch is off the γ axis

Where, then, does extreme longevity come from? Not from rewriting the promoter. The elephant carries ~20 functional TP53 copies at an essentially unchanged per-copy γ of 1.4269 (human 1.4298), giving roughly twenty times the effective gate dosage at a per-copy barrier of 0.5111. The switch is copy number, not promoter γ.

The framework's thesis holds cross-species: a conserved identity γ substrate plus divergent dynamics (dosage, loop gain, reservoir size, senescence rate) produces the lifespan spectrum. Human-not-special, TP53 flatness, and the audited finding that the weak shared lean is fully confound-attributable are graded [V]; the residual TERT lean is [O] (n = 10, p ≈ 0.12, awaiting a larger panel); the off-γ-axis longevity levers are [L] — TP53 copy number (Abegglen 2015 JAMA; Sulak 2016 eLife) and somatic telomerase suppression in large mammals (Gomes 2011).