Krishnamurthy et al. 2004 — Ink4a/Arf expression is a biomarker of aging
Open-access primary study (bronze OA via JCI). doi:10.1172/JCI22475 · PMID 15520862 · PMC524230. Sharpless laboratory, Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill. Published Journal of Clinical Investigation 2004; 114(9):1299–1307. Cited 1,436+ times (top citation percentile).
This paper established that expression of the Ink4a/Arf locus — encoding p16INK4a and p19Arf — rises markedly and broadly across virtually all rodent tissues with advancing age, making it one of the most extensively validated molecular biomarkers of organismal aging. It also demonstrated that this age-associated rise is attenuated by caloric restriction in several tissues and correlates with the in-vivo senescence marker SA-β-gal in the kidney, positioning p16INK4a as a candidate effector — not merely a correlate — of mammalian tissue aging.
TL;DR
Krishnamurthy et al. used quantitative RT-PCR to survey cyclin-dependent kinase inhibitor (CDKI) expression across 15 murine tissues in young (2.5 months) versus old (26 months) C57BL/6 mice, and 12 rat tissues in young (3 months) versus old (28 months) Fischer 344 rats on ad libitum (AL) or caloric restriction (CR) diets. They found:
- p16INK4a rose markedly in 26 of 27 organs examined; the geometric mean fold-increase across 15 murine tissues was ~9.7-fold (described as “approximately 10-fold” in the paper)
- p19Arf showed a more modest ~3.5-fold geometric mean increase
- p21CIP showed only ~1.4-fold average increase — far below p16’s rise
- Caloric restriction attenuated the age-associated rise in kidney, ovary, heart, adrenal, and testis (2- to 16-fold attenuation), but not in lung, lymph node, spleen, or liver
- In the kidney, the p16 attenuation by CR correlated with reduced SA-β-galactosidase activity and decreased nephritis — confirming functional relevance
- Lineage-negative (stem/progenitor-enriched) bone marrow cells showed the principal increase; in lymphoid organs, stromal fractions dominated the rise
- The age-associated rise in p16 strongly correlates with Ets-1 expression (r = 0.62, P < 0.001 across 27 tissues), implicating Ets-1 as a principal transcriptional activator
- p16INK4a and Arf expression are themselves highly correlated across tissues (r = 0.75, P < 0.0001), implying an additional shared coregulatory factor independent of Ets-1
Study Design
Murine age-comparison (primary dataset)
| Parameter | Detail |
|---|---|
| Species / strain | Mus musculus, C57BL/6 |
| Young age | 2.5–3.5 months |
| Old age | 25–26 months |
| n per group per tissue | 4–6 mice |
| Tissues surveyed | 15 murine tissues (adrenal, bone marrow, cecum, cortex, duodenum, heart, kidney, liver, lung, lymph node, ovary, pancreas, spleen, testis, uterus) |
| Analytic method | Quantitative RT-PCR; results expressed as log2(old/young) ratios, normalised to 18S, GAPDH, or TATA-binding protein |
| Validation | IHC and flow-sorted compartment analysis in selected tissues; SA-β-gal staining in kidney and liver slices |
Rat caloric restriction dataset
| Parameter | Detail |
|---|---|
| Species / strain | Rattus norvegicus, Fischer 344 |
| Ages | Young 3 months vs old 28 months |
| Diet arms | Ad libitum (AL) vs caloric restricted (CR) |
| n per group | AL: 2 old males, 2 old females, 2 young males, 2 young females (n=8 total); CR: 2 old males + 2 old females (n=4); 8–16 RT-PCR reactions per tissue per group |
| Tissues | 12 tissues: heart, kidney, lung, lymph node, ovary, uterus, testis + others |
Comparative CDKI panel
In addition to p16INK4a and Arf, the authors surveyed p15INK4b, p18INK4c, p19INK4d, p21CIP, and p27KIP across tissues. The selectivity of p16INK4a as the major age-responsive CDKI was a key finding: only p16 and Arf showed consistent tissue-wide rises; other CDKIs including p21 and p27 did not.
Key Quantitative Results
Tissue-wide p16INK4a increase with age (murine)
The geometric mean old/young ratio across the 15 murine tissues was ~9.7-fold for p16INK4a and ~3.5-fold for Arf. p21CIP averaged only ~1.4-fold (not significantly different from no change in most tissues). A marked increase (3-fold or greater) in p16INK4a was seen in 26 of 27 organs analyzed (combining murine and rat datasets).
The paper text describes cecum, kidney, ovary, and uterus as showing “particularly large (>30-fold) increases.” Specific per-tissue ratios below are read from Figure 1A bar heights (log2 scale) and are approximations; only the >30-fold group and the geometric mean are stated in the prose:
| Tissue | Fold increase (p16, old/young) | Notes |
|---|---|---|
| Uterus (murine) | >96-fold (figure) | Largest fold increase recorded; stromal + epithelium by IHC |
| Cecum | >70-fold (figure) | Below detection in young mice; minimum estimate only |
| Kidney | >34-fold (figure; paper: “>30-fold” group) | Cortical tubules by IHC; CR nearly abolished the rise |
| Ovary | ~30-fold (figure; paper: “>30-fold” group) | CR substantially attenuated |
| Liver | ~29-fold (figure) | IHC not detected in either species; high absolute expression in aged animals |
| Lymph node | ~15-fold (figure) | Stromal: 42-fold; T cell-rich cellular: 13-fold (these compartment values stated in text) |
| Duodenum | >17-fold (figure) | Below detection in young mice |
| Bone marrow | markedly elevated (figure) | Lin− (stem/progenitor-enriched) cells show the principal increase; no specific fold-change given in text for Lin− alone |
| Cortex (brain) | ~9-fold (figure) | gap/needs-replication in specific neuronal vs glial compartments |
| Adrenal | ~8-fold (figure) | CR attenuated |
| Heart | ~7-fold (figure) | CR attenuated; p15INK4b showed ~5-fold rise here (text) |
| Pancreas | above detection threshold in old (figure) | Below detection in young; islets by IHC (consistent with prior human pancreas data cited in Table 1) |
| Spleen | ~3-fold (figure) | Stromal: 15-fold; cellular (T cell-rich): 5-fold (both stated in text) |
| Lung | ~3-fold (figure) | High absolute expression in aged animals; CR did NOT attenuate |
| Testis | ~1-fold (figure) | No meaningful increase in mice; >135-fold in aged AL rats, 58-fold in CR rats, due to Leydig cell tumours |
Table 1 in the paper also compiles published human immunohistochemistry data from prior literature alongside the murine RT-PCR findings, confirming that p16INK4a has been detected in equivalent tissues across species (human: myocardium, cortical tubule rare glomeruli, pancreatic islets, epididymis, uterine epithelium and stroma — drawn from cited references 13–19 in the paper, not from original data generated here).
Selectivity of p16INK4a among CDKIs
The paper explicitly compared 7 CDKI family members:
| CDKI | Geometric mean old/young fold increase |
|---|---|
| p16INK4a | ~9.7-fold |
| p19Arf | ~3.5-fold |
| p18INK4c | ~1.5-fold (non-significant) |
| p15INK4b | ~2-fold (varies by tissue) |
| p19INK4d | ~1.3-fold (minimal) |
| p21CIP | ~1.4-fold (non-significant) |
| p27KIP | ~1.0-fold (no change) |
Only p16 and Arf showed consistent tissue-wide rises; the paper argues p16’s uniqueness among CDKIs reflects its specific role as a senescence effector rather than a general cell-cycle regulator.
Caloric restriction results (Fischer 344 rat)
CR attenuated the age-associated rise in Ink4a/Arf expression in selected tissues:
| Tissue | CR attenuation of p16 | Correlate |
|---|---|---|
| Kidney cortex | Near-complete (2–16-fold attenuation) | SA-β-gal activity also abolished by CR; nephritis reduced |
| Ovary | 2–16-fold attenuation | — |
| Heart | 2–16-fold attenuation | — |
| Adrenal | 2–16-fold attenuation | — |
| Testis | Partial | Leydig cell tumour burden reduced (135-fold AL vs 58-fold CR) |
| Lung | No attenuation by CR | — |
| Lymph node | No attenuation by CR | — |
| Spleen | No attenuation by CR | — |
| Liver | No attenuation by CR | — |
| Uterus | Unexpected increase with CR (paper: “CR appeared to induce a significant increase”) | Mechanism unclear |
SA-β-gal correlation in kidney: In situ SA-β-gal activity (a validated in-vivo senescence marker) was detectable in the renal cortex of aged AL animals and was abolished in CR animals — paralleling p16 expression and confirming that p16 is tracking functionally senescent cells, not merely expressing as an epiphenomenon.
Growth hormone receptor (GHR) deficiency analysis
GHR-deficient (GHR−/−) mice, which have extended longevity, showed no effect on p16INK4a or Arf expression in the kidney, contrasting sharply with CR, which significantly reduced p16 in kidney. However, modest reductions in p16INK4a and Arf were noted in the lung from GHR-deficient animals. The authors conclude that CR and GHR deficiency enhance longevity by molecularly distinct mechanisms — at least in the kidney — since their effects on the Ink4a/Arf locus diverge tissue-specifically.
Stem and progenitor cell compartment
In bone marrow, the principal increase in p16INK4a with aging occurred in lineage-negative cells (enriched for hematopoietic stem and progenitor cells, Lin−: ~2%), not in Lin+ committed progenitor fractions. This finding is consistent with the proposed role of Ink4a/Arf in limiting stem cell self-renewal during aging and provides a mechanistic link between p16 accumulation and age-related stem cell exhaustion in hematopoietic compartments.
Ets-1 / transcriptional regulation
Across 27 tissue types from two species, p16INK4a expression with aging showed a strong positive correlation with Ets-1 (r = 0.62, P < 0.001) — an Ets family transcription factor known to directly activate p16INK4a transcription in cell culture. The paper proposes that stress-induced Ets-1 activation via MAPK/ERK and p38MAPK signalling is a principal transcriptional driver of age-associated p16 accumulation. p16INK4a and Arf expression are themselves highly correlated across all 27 tissues (r = 0.75, P < 0.0001, r² = 0.49 — stated in text). Because Ets-1 does not regulate Arf transcription, this high p16/Arf co-correlation implies an additional shared unknown coregulatory factor(s) that exerts a powerful and independent effect on the Ink4a/Arf locus with aging.
Key Conclusions
- p16INK4a is a near-universal biomarker of tissue aging in rodents. A >3-fold increase was seen in 26 of 27 organs, with the average being ~10-fold across 15 tissues — far greater than any other CDKI family member.
- The rise tracks biological, not merely chronological, age. CR attenuates p16 in parallel with attenuating the senescence-associated phenotype and pathology in the kidney and testis.
- Stem/progenitor cells preferentially accumulate p16. The lineage-negative bone marrow finding directly connects p16 accumulation to stem cell exhaustion — positioning p16 as both marker and candidate effector of impaired tissue regeneration with aging.
- Ets-1 is a principal transcriptional activator, but p16 and Arf co-express at r = 0.75 (P < 0.0001) — since Ets-1 does not regulate Arf, an unknown shared coregulatory factor(s) is implicated.
- p16 expression may be clinically useful for determining donor suitability (organ transplants, bone marrow allografts), predicting disease progression in premorbid syndromes, and as a surrogate for anti-aging therapeutic efficacy.
Human-Tissue Evidence Status
The paper does not generate original human tissue data. Table 1 compiles prior published IHC data from the literature (refs 13–19), showing that p16INK4a has been detected in human tissues (myocardium, renal cortical tubules, pancreatic islets, epididymis, uterine epithelium and stroma) — in parallel with the murine data. The authors describe this as “concordance” with human evidence; the primary human evidence is located in the cited prior publications.
The most rigorous human-tissue follow-up for skin was provided by ressler-2006-p16-human-skin-biomarker (Aging Cell 2006), and for a multi-organ human cross-sectional survey by idda-2020-senescent-markers-human-tissues (Aging 2020).
Model-Organism Extrapolation
| Dimension | Status |
|---|---|
| Pathway conserved in humans? | Yes — CDKN2A/p16INK4a–CDK4/6–RB axis is identical in humans |
| Phenotype (p16 rise with age) conserved in humans? | Yes — replicated in human skin (Ressler 2006), human T lymphocytes, human pancreatic islets, and human adipose/skin biopsies (Hickson 2019, Justice 2019) |
| Replicated in humans as aging biomarker? | Yes — p16 mRNA in human T cells correlates with chronological age (Liu 2009; see p16-rb-pathway); direct tissue-level human data now substantial |
| CR effect on human p16? | Not directly tested as of this paper; indirect evidence suggests CR-mimetics (rapamycin) lower p16+ burden in mice |
See model-organisms/_extrapolation-guide.md for extrapolation tier definitions. Primary rodent species here is mouse (C57BL/6 and Fischer 344 rat); both are well-validated models for human aging biology.
Significance in the Wiki
This paper anchors the p16-rb-pathway page’s “in-vivo aging biology” section — the ~10-fold rodent tissue statistic cited there derives from Table 1 and Figure 1A of this paper. It is co-cited with idda-2020-senescent-markers-human-tissues as the foundational framework for the claim that p16INK4a expression rises with age across many tissues.
The CR-p16 attenuation result is the molecular bridge between p16-as-biomarker and p16-as-effector — establishing that p16 rise is not a bystander to aging but tracks with functional outcomes (nephritis, Leydig tumour burden). This has been widely cited in the context of nutrient-sensing interventions that modulate p16 expression.
Limitations
- Inbred rodent strains only. C57BL/6 and Fischer 344 are susceptible to specific age-related pathologies (lymphoma and kidney disease, respectively) that may amplify tissue-specific p16 signals. The Testis finding (1.1-fold in mice vs >135-fold in rats) is dominated by Leydig cell tumour burden in the Fischer rat — not a universal mammalian aging phenomenon. gap/needs-replication in outbred or heterogeneous-stock animals
- mRNA-based measurement. Most results are RT-PCR from total RNA from grossly dissected organs (fat-trimmed, then mixed cell-type samples). Compartment-level data (IHC, flow sorting) was done in only a subset of tissues. Organ-level mRNA averages may understate cell-type-specific rises.
- No lifespan data. This paper does not show that p16 rise causes aging pathology — it shows correlation with age and CR-sensitive regulation. Functional-consequence evidence came later (Baker 2011 INK-ATTAC) 1.
- Human data is compiled, not original. The Table 1 human IHC column is from prior publications; no original human biopsy data is generated here. See gap/needs-human-replication for the multi-tissue cross-species extension.
- p19Arf complicates interpretation. Because the Cdkn2a locus encodes both p16INK4a and p19Arf (p14ARF in humans) from overlapping sequences, many early assays (and some of the IHC in this paper) could not cleanly separate the two products. The paper acknowledges this and uses gene-product-specific primers for RT-PCR; IHC uses a p16-specific antibody (F-12; Santa Cruz).
Cross-references
- p16-rb-pathway — the signalling pathway this study established as a tissue-wide aging biomarker; quantitative claims on that page (geometric mean ~10-fold, CR attenuation in kidney/ovary/heart) originate here
- cellular-senescence — hallmark page; this study is one of the two foundational rodent-tissue anchors for the molecular evidence base
- cellular-senescence — hallmark MOC
- deregulated-nutrient-sensing — CR attenuates p16 rise; cross-talk with mTOR/AMPK axis
- stem-cell-exhaustion — lineage-negative bone marrow finding links p16 accumulation to the hematopoietic stem cell aging hallmark
- idda-2020-senescent-markers-human-tissues — primary human-tissue cross-sectional confirmation across 10 organs; cites this paper as a foundational reference
- ressler-2006-p16-human-skin-biomarker — direct human skin biopsy validation; the primary human-tissue sequel to this work for skin specifically
Footnotes
Footnotes
-
doi:10.1038/nature10600 · Baker DJ et al. · Nature 2011 · in-vivo · model: BubR1 progeroid mice · INK-ATTAC p16-driven caspase-8 construct; clearance of p16+ cells delayed sarcopenia, fat loss, cataracts · functional consequence evidence for p16-positive senescent cells in aging pathology ↩