Tier 3 — preclinical

Neuroprotective functions for the histone deacetylase SIRT6

Kaluski S, Portillo M, Besnard A, Stein D, Einav M, Zhong L, Ueberham U, Arendt T, Mostoslavsky R, Sahay A, Toiber D

Cell Reports 2017 Volume 18, Issue 13, pages 3052–3062

Bibliography

DOI: 10.1016/j.celrep.2017.03.008
PubMed: PMID 28355558
PubMed Central: PMC5389893
Funding: This work was supported by the Israeli Ministry of Science and Space. A.B was supported by 2014 NARSAD Award, Bettencourt-Schueller Foundation, and Philippe Foundation. A.S is supported by Biobehavioral Research Awards for Innovative New Scientists (BRAINS)1-R01MH104175 and NIH-NIA 1R01AG048908-01A1.
Competing interests: The authors don't have any conflict of interest.

Study snapshot

Design	Brain-specific SIRT6 conditional knockout mouse (Nestin-Cre × Sirt6-floxed) with behavioural, molecular, and histological characterisation. Cellular loss-of-function models via CRISPR/Cas9 SIRT6 knockout in SH-SY5Y neuroblastoma cells and shRNA knockdown in N2a cells. Western blot analysis of temporal cortex from sporadic Alzheimer's disease patients and controls. Bioinformatic analysis of public transcriptomic datasets (Allen Brain Atlas, GEO GDS4758, R2 Genomics) for SIRT6 expression by Braak stage and gene co-expression networks.
Model	Brain-specific SIRT6 knockout mice (brS6KO, Nestin-Cre-mediated, 97% C57BL/6J background), 3-5 months old, male and female. SH-SY5Y human neuroblastoma cells with CRISPR/Cas9 SIRT6 knockout using two independent guide RNAs. N2a mouse neuroblastoma cells with shRNA SIRT6 knockdown. Human temporal cortex from sporadic Alzheimer's disease patients and age-matched controls.
Sample	Behavioural experiments: brS6KO n=7, wild-type controls n=12. Immunofluorescence and Western blot of mouse brain: n=4-5 per genotype. TUNEL apoptosis: n=5 per genotype. Human tissue: Alzheimer's disease n=4, controls n=4. Bioinformatic datasets: variable n depending on source.
Intervention	Genetic loss-of-function (Nestin-Cre-driven Sirt6 deletion in brain). Cellular perturbations included ionising radiation (400 rad), proteasome inhibitor MG132, GSK3 inhibitor lithium (1 mM × 50 hours), ATM inhibitor caffeine (100 μM), rescue by SIRT6 wild-type re-expression, and dominant-negative expression of catalytically inactive SIRT6-HY. Tau S199 phospho-mutants (S199A and S199E) tested Tau stabilisation dependence on GSK3 phosphorylation.
Duration	Mice assessed at 3-4 months of age (behavioural testing) and 4 months (molecular and histological analysis). Cellular experiments were short-term (hours to days).
Endpoints	Behavioural: open field locomotor activity and rearing across two days, contextual fear conditioning freezing behaviour, light-dark test time and entries; DNA damage markers: phospho-ATM, γH2AX, H3K56 acetylation, SNF2H recruitment to chromatin; Apoptosis: TUNEL staining in cortex, caspase-3 activation reporter (ApoAlert); Tau pathology: total Tau and phospho-Tau by AT8 and Ser199 antibodies in brain and cells; GSK3 activity: inhibitory phosphorylation of GSK3α and GSK3β; Tau stability: half-life under proteasome inhibition; Tau S199A and S199E phospho-mutant stability; Rescue: apoptosis and Tau phosphorylation after lithium, caffeine, or SIRT6 re-expression; Human tissue: SIRT6 protein and mRNA levels in AD vs control temporal cortex; SIRT6 expression by Braak stage; SIRT6 co-expression with PPP2R1A (Tau phosphatase regulatory subunit); Bioinformatics: SIRT6 co-expression network analysis via Allen Brain Atlas

What the study showed, in plain terms

The 2006 Mostoslavsky paper established that mice completely lacking SIRT6 die at four weeks of age with a rapid aging-like syndrome — but this whole-body catastrophe made it impossible to ask what SIRT6 does specifically in the adult brain. This 2017 paper solved that problem by generating mice in which SIRT6 is deleted only in brain cells, allowing the animals to survive to adulthood so their brain-specific phenotypes could be studied.

The brain-specific SIRT6 knockout mice survived, but their brains showed all the hallmarks of accelerated neurodegeneration by four months of age. DNA damage markers were elevated, apoptotic cell death in the cortex was roughly doubled, and — most striking for the Alzheimer's disease connection — phosphorylated Tau protein, one of the two defining pathological features of Alzheimer's, was significantly increased. The mice showed clear learning and memory deficits in behavioural tests. The mechanism ran through the kinase GSK3: loss of SIRT6 allowed GSK3 to become chronically active, which phosphorylated Tau at position Ser199, which in turn stabilised the Tau protein from normal degradation and let it accumulate.

The translational punchline came at the end of the paper. When the authors examined temporal cortex tissue from sporadic Alzheimer's disease patients, they found SIRT6 protein and mRNA levels dramatically reduced compared with age-matched controls — and the more severe the disease stage (Braak stage), the lower the SIRT6. This paper reframed SIRT6 from a general aging enzyme into a specific neuroprotective factor whose decline may causally contribute to Alzheimer's disease pathology, making it a therapeutic target for neurodegeneration.

Key findings

Brain-specific SIRT6 knockout mice (Nestin-Cre × Sirt6-floxed) survived to adulthood, unlike the whole-body knockouts that die at four weeks — enabling adult neurological phenotyping for the first time.
By 4 months of age, brS6KO brains showed elevated DNA damage markers: increased phospho-ATM, increased γH2AX, elevated H3K56 acetylation, and reduced SNF2H recruitment to chromatin.
Apoptotic (TUNEL-positive) cell numbers in the cortex were roughly doubled in brS6KO vs wild-type at 3-4 months.
brS6KO mice showed impaired non-associative learning (failure to habituate to the open field over two consecutive days) and impaired associative learning (markedly reduced freezing in contextual fear conditioning).
Phosphorylated Tau (AT8 antibody, Ser199) was significantly elevated in brS6KO cortex — a hallmark of Alzheimer's disease and related tauopathies.
Mechanism: SIRT6 loss reduced inhibitory phosphorylation of GSK3α/β, producing chronic GSK3 activation. Active GSK3 phosphorylated Tau at Ser199, which stabilised Tau protein against proteasomal degradation.
Ionising radiation phenocopied SIRT6 loss in wild-type cells: DNA damage alone was sufficient to activate GSK3, phosphorylate Tau, and stabilise it — placing DNA damage signalling upstream of Tau pathology.
Tau S199A phospho-mutant (cannot be phosphorylated at Ser199) failed to accumulate even under proteasome inhibition, confirming Ser199 phosphorylation drives Tau stabilisation. The S199E phospho-mimetic behaved similarly to wild-type Tau.
Pharmacological GSK3 inhibition (lithium, 1 mM), ATM inhibition (caffeine, 100 μM), and SIRT6 re-expression each rescued apoptosis in SIRT6-null cells and reduced Tau phosphorylation.
Sporadic Alzheimer's disease temporal cortex showed dramatically reduced SIRT6 protein (~4-fold lower) and mRNA vs age-matched controls; reduction correlated with Braak stage severity (stages iii-v and iii-vi).
Normal positive co-expression between SIRT6 and PPP2R1A (regulatory subunit of PP2A, the main Tau phosphatase) was lost in Alzheimer's disease brains, suggesting broader disruption of Tau regulation.

What this study can and cannot tell us

Nestin-Cre drives SIRT6 deletion from embryonic day 14-17 onwards, so some behavioural and structural phenotypes could reflect impaired brain development rather than pure post-development degeneration. The authors acknowledge this and use cellular models where development is not a confounder to support their conclusions.
Human tissue analysis used a small sample (n=4 Alzheimer's disease, n=4 control) for Western blot. Broader mRNA analyses used larger public datasets, but the protein-level confirmation is underpowered.
Human data are observational and cross-sectional. They show SIRT6 is reduced in Alzheimer's disease brains but cannot establish whether SIRT6 loss causes AD or is a consequence of AD pathology.
This is a loss-of-function paper: it demonstrates that removing SIRT6 causes neurodegenerative pathology, but does not test whether pharmacologically activating SIRT6 in wild-type aged or AD-model mice rescues these phenotypes. The translational case for SIRT6 activation in neurodegeneration is inferred, not tested.
Behavioural phenotypes (particularly hyperlocomotion) can confound interpretation of anxiety and learning tests. The authors control for this by measuring latency to first enter the lit compartment in the light-dark test.
The paper predates most published SIRT6 activator chemistry. Direct evidence that any specific compound reverses these phenotypes in vivo remains for later work.