Kress et al. 2014 — paravascular clearance in the aging mouse brain
TL;DR
Kress et al. compared young, middle-aged, and old C57BL/6 mice using several complementary assays of cerebrospinal-fluid (CSF) entry, interstitial-solute clearance, vascular pulsatility, extracellular-space properties, and astrocytic aquaporin-4 (AQP4) localization. Relative to young mice, old mice had a reported 40% relative impairment of intraparenchymally injected amyloid-β clearance and a 27% relative reduction in penetrating-arteriole wall pulsatility; paravascular CSF-tracer penetration was also lower, while perivascular AQP4 polarization was broadly lost despite no regional change in total AQP4 immunofluorescence. These are foundational mouse data linking natural aging to impaired glymphatic/paravascular transport, but the study did not measure cognition, protein aggregation, or any human endpoint, and it did not causally show that AQP4 relocalization or weaker pulsatility produced the clearance deficit.1
Design
Animals and age groups
| Group | Age reported for the main aging comparisons |
|---|---|
| Young | 2–3 months |
| Middle-aged | 10–12 months |
| Old | 18 months in the Results and figure captions; 18–20 months in the abstract |
The main CSF-influx and clearance Results identify male C57BL/6 mice. The general animal Methods state that male and female C57BL/6 mice from Charles River and the National Institute on Aging were used unless otherwise noted and give slightly different default middle-age ranges (8–10 months); sex-specific counts are not reported. The paper therefore does not support a sex-stratified conclusion, and its abstract/Methods/Results are not fully consistent about exact ages.1
All anesthetized procedures used ketamine (0.12 mg/g intraperitoneally) plus xylazine (0.01 mg/g intraperitoneally). Intracisternal fluorescent tracers were infused into the cisterna magna at 2 µL/min for 5 minutes (10 µL total). The authors acknowledge that this infusion mildly increased intracranial pressure during infusion, although they report that pressure normalized rapidly afterward.1
Assays
| Question | Assay | Reported sample size |
|---|---|---|
| Does CSF tracer enter aged brain parenchyma less efficiently? | Intracisternal 3-kDa Texas Red dextran (Dex-3) plus 45-kDa fluorescent ovalbumin (OA-45); whole-slice ex-vivo fluorescence 30 minutes after infusion | n=5–8 mice/group |
| Are in-vivo paravascular influx dynamics slower by middle age? | Intracisternal 40-kDa FITC-dextran and two-photon imaging 100 µm below cortex | n=4 mice/group |
| Is interstitial-solute clearance impaired? | 0.5 µL artificial CSF containing 10 nM ^125I-amyloid-β1–40 plus 0.05 µCi ^14C-inulin microinfused into caudate over 5 minutes; brain radioactivity remaining after 60 minutes | n=6–11 mice/group |
| Does vascular pulsatility change with age? | Two-photon diameter line scans of penetrating arteries and ascending veins over 3 seconds | 8–20 vessels from 4 mice/group |
| Do extracellular volume or tortuosity explain the age effect? | Paired awake/anesthetized tetramethylammonium micro-iontophoresis | n=9–20/group as reported |
| Does astrocytic AQP4 localization change? | Regional AQP4/GFAP immunofluorescence and perivascular-polarization analysis | n=4 mice/group for regional analyses; 11–12 vessels from 4 mice/group for perivascular AQP4 intensity profiles; 11–18 vessels from 4 mice/group for endfoot/GFAP analyses; 6–8 regions of interest from 4 mice/group for pial analyses |
The study reports experiment-specific sample sizes but not a total count of unique mice or whether animals were reused across compatible assays; a defensible study-wide n therefore cannot be reconstructed.1
Key results
CSF influx and interstitial-solute clearance declined with age
- Whole-slice penetration by both the 3-kDa and 45-kDa intracisternal tracers was lower with age (one-way ANOVA with Tukey post-hoc comparisons; reported significance thresholds p<0.05 and p<0.001; n=5–8/group). The Results text says both middle-aged and old brains were significantly reduced versus young, whereas the Figure 1 caption explicitly calls old-versus-young significant and describes middle-aged values only as intermediate. Exact group means and confidence intervals are not tabulated in the text.
- In-vivo two-photon imaging showed delayed 40-kDa tracer accumulation in middle-aged versus young cortex (p<0.05, repeated-measures two-way ANOVA with Sidak correction; n=4/group). In young cortex, accumulation began within approximately 5 minutes and peaked at approximately 50 minutes.
- Sixty minutes after caudate co-injection, the Results text reports significantly impaired clearance of metabolically inert ^14C-inulin in both middle-aged and old mice (reported p<0.05 and p<0.01 across comparisons; n=6–11/group), and impaired ^125I-amyloid-β1–40 clearance in both older groups (p<0.001; n=6–11/group). The Figure 1 caption instead explicitly calls old-versus-young clearance significant and describes the middle-aged values as intermediate, without identifying which middle-aged pairwise comparisons reached significance.
- The abstract summarizes old-versus-young amyloid-β clearance as 40% impaired. This is a relative reduction in the measured clearance fraction—not a 40-percentage-point difference—and exact group percentages and confidence intervals are not printed in the Results text.1
The inulin assay was intended to index bulk-flow-associated clearance because inulin is neither retained by brain nor transported across the blood–brain barrier. Amyloid-β clearance is less specific: the authors note that it includes both receptor-mediated blood–brain-barrier efflux and AQP4-associated bulk transport. Thus, the 40% result is not an isolated measurement of one clearance route.1
Penetrating-arteriole pulsatility was selectively lower
Old mice had a reported 27% relative reduction in deep penetrating-arteriole wall pulsatility versus young mice (p<0.05; 8–20 vessels from 4 mice/group). Ascending-vein pulsatility showed no age-related difference. Because pulsatility and clearance were measured as parallel age-associated phenotypes rather than experimentally manipulated in this study, the result supports a candidate driving mechanism but does not establish mediation or causality.1
AQP4 polarization, rather than total regional expression, tracked impaired influx
- In old cortex, AQP4 immunofluorescence was redistributed into tissue surrounding penetrating arterioles, while AQP4 signal in the immediately apposed perivascular endfoot domain was lower (p<0.05; 11–18 vessels from 4 mice/group). The analogous perivascular-endfoot change around capillaries was null.
- Regional total AQP4 immunofluorescence did not differ between young and old brains. In contrast, the relative perivascular polarization measure was reduced across aged brain regions, most consistently in lateral and ventral cortex and also in striatum and hippocampus (region-dependent p<0.05 to p<0.001; n=4/group).
- GFAP immunoreactivity increased with age, but cortical GFAP expression was not associated with CSF-tracer penetration in either young (p=0.3992, r²=0.0377) or old (p=0.6061, r²=0.0143) brain. Total cortical AQP4 expression was also not significantly associated with tracer penetration in young (p=0.0501, r²=0.1634) or old (p=0.2301, r²=0.0749) brain.
- Within each age stratum, lower cortical AQP4 polarization correlated with lower tracer penetration (young: p=0.0225, r²=0.2149; old: p=0.0215, r²=0.2485). The relationship disappeared when young and old data were pooled. The authors explicitly describe this as correlative evidence, not a direct test that AQP4 depolarization causes impaired transport.1
Regional and trans-pial effects
At 30 minutes, OA-45 tracer penetration was most consistently reduced in lateral and ventral cortex, especially in posterior sections (n=4/group). The Results text reports region-dependent p<0.05 to p<0.0001, whereas the Figure 7 caption gives p<0.05 to p<0.001. Near the pial surface, old mice had greater AQP4 expression and lower trans-pial tracer penetration (both p<0.01; 6–8 regions of interest from 4 mice/group), while the local GFAP difference was null.1
Informative null results
- Extracellular volume fraction did not differ by age in awake or anesthetized cortex (age effect p=0.4957; n=9–20/group).
- Extracellular tortuosity did not differ by age (p=0.3375).
- Both young and old mice retained the anesthesia-associated expansion of extracellular volume (p<0.001 for state), arguing against loss of this response as the explanation for lower tracer transport in old mice.
- Ascending-vein pulsatility did not differ by age.
- Total regional AQP4 immunofluorescence did not differ by age; localization, not total signal, was the principal AQP4 phenotype.
- GFAP and total AQP4 expression did not significantly predict cortical tracer penetration within age groups, and the AQP4-polarization association did not persist in pooled-age data.1
Interpretation and relevance to aging
This study provides direct mouse evidence that several measured components of paravascular transport decline by middle and late adulthood: CSF-tracer influx, inulin clearance, and amyloid-β clearance. It also identifies two accompanying candidates—reduced penetrating-arteriole pulsatility and loss of endfoot-localized AQP4—that could connect vascular and astrocytic aging to diminished waste transport in the brain. The findings therefore intersect with loss-of-proteostasis and hypotheses about age-related vulnerability to Alzheimer’s disease, but amyloid deposition, neurodegeneration, cognition, and disease incidence were not measured.1
The proposed chain—aging → weaker arterial pulsatility/AQP4 depolarization → lower paravascular transport → protein aggregation or cognitive decline—is only partly tested here. Age was the exposure, transport phenotypes were measured, and candidate mechanisms were correlated; downstream aggregation and cognition were author hypotheses. gap/no-mechanism
Extrapolation to humans
| Dimension | Status | Notes |
|---|---|---|
| Core anatomy and AQP4 expression conserved? | not assessed here | This mouse study did not test human anatomy, AQP4 localization, or transport. |
| Age-related transport phenotype shown in humans here? | no | All experiments used C57BL/6 mice. |
| Amyloid-β clearance result replicated in aging humans here? | no | The tracer was injected into mouse caudate; no human clearance measurement was performed. |
| Causal mechanism established? | partial/no | Natural aging altered transport, but AQP4 localization and arterial pulsatility were not independently perturbed. |
| Intervention tested? | no | The paper proposes therapeutic targets but did not restore AQP4 polarization, pulsatility, clearance, or cognition. |
The paper’s interpretation that impaired glymphatic clearance contributes to cognitive decline “among the elderly” exceeds its direct evidence. The defensible conclusion is that natural aging was associated with impaired paravascular transport in these mouse assays. Human aging, clinical neurodegeneration, and therapeutic restoration all require independent evidence. gap/needs-human-replication
Limitations and evidence-quality flags
- No human arm and no behavioral endpoint: the study cannot connect the measured transport changes to human aging or even to cognitive performance in the mice. gap/needs-human-replication
- Age is not a mechanistic perturbation: reduced pulsatility, AQP4 depolarization, reactive astrogliosis, and impaired transport co-occurred. The design did not isolate which change was causal.
- Small, assay-specific cohorts: several mechanistic assays used four mice/group, while many vessels, regions, or time points from those same animals contributed observations. The animal, not each vessel or image, is the relevant biological replicate.
- No reported study-wide n: the paper does not state the total number of unique mice or clearly map overlap across assays.
- Sex reporting is insufficient: the general Methods allow both sexes, while the principal aging Results specify males; sex-stratified counts and analyses are absent.
- Age labels are internally inconsistent: old is 18 months in Results/captions but 18–20 months in the abstract, and the default middle-aged range differs between general Methods and the principal Results.
- Anesthesia and tracer delivery matter: most influx imaging used ketamine/xylazine and artificial intracisternal infusion. The authors report a transient infusion-associated intracranial-pressure rise, which may not reproduce spontaneous physiology.
- Tracer influx is not synonymous with net waste clearance: intracisternal fluorescence mainly measured entry/penetration; the separate caudate radiotracer assay measured solute remaining after 60 minutes. These endpoints should not be collapsed into one generic “flow” value.
- Amyloid-β has multiple exit routes: the assay cannot assign the 40% relative clearance impairment solely to glymphatic transport because blood–brain-barrier efflux also contributes.
- AQP4 evidence is associative in this experiment: the pooled-age polarization association disappeared, and no localization-specific rescue was performed.
- Single research program: the assays and interpretation came from the group that introduced the glymphatic model; independent replication and method-sensitive disagreement should be considered on the process page. gap/needs-replication
Funding and conflicts
The work was supported by NIH/NINDS grants NS078167, NS078304, and NS073373 and American Heart Association award 12SDG11820014. The authors declared no competing conflict of interest.1
Downstream propagation candidates
- glymphatic-system: use this as the canonical natural-aging mouse study, preserving the separation between intracisternal tracer influx and interstitial radiotracer clearance.
- brain: replace an unsourced generic age-related glymphatic-decline statement with the mouse-specific 40% amyloid-β-clearance and 27% penetrating-arteriole-pulsatility findings, plus the human-replication caveat.
- astrocytes: add the distinction between unchanged total regional AQP4 signal and reduced perivascular AQP4 polarization; label the tracer association as non-causal.
- loss-of-proteostasis and Alzheimer’s disease: cite only as a candidate age-related clearance mechanism, not evidence that the observed transport deficit caused aggregation or disease.
- arterial-stiffening: if included, describe lower penetrating-arteriole wall pulsatility as a mouse imaging phenotype; this study did not measure arterial stiffness.
- Sleep: propagate only the preserved awake/anesthetized extracellular-volume response in old mice; this study did not test sleep behavior or a sleep intervention.