Since HDACs have emerged as potential therapeutic targets to treat neurodegenerative diseases, it is of utmost importance to study the role of individual HDACs in brain function. Here we have investigated the role of HDAC6 in memory function and the pathogenesis of AD. We show that Hdac6 is expressed in regions of the adult brain that are implicated in memory function, namely hippocampus and cortex. Within the hippocampal formation HDAC6 was predominantly localized to the cytoplasm and absent from the nucleus, which is in line with previous reports on Hdac6 expression in other cell types (Hubbert et al, 2002). To further study the role of HDAC6, we generated mice deficient in HDAC6 and found that these animals are viable and display no detectable phenotype, which is consistent with previous observations (Zhang et al, 2008). Moreover, brain morphology was normal in Hdac6−/− mice. This supports previous findings from Drosophila models where knockdown of Hdac6 did not cause any detectable phenotype in eye morphology, a commonly used system to assess neurodegeneration in invertebrates (Pandey et al, 2007). Notably, we did not detect a compensatory regulation of other HDACs. Specifically, the levels of SIRT2, another known regulator of tubulin acetylation (Tang & Chua, 2008), were similar in Hdac6−/− mice and wild type littermates. These data suggest that HDAC6 is dispensable for normal development in mice. Moreover, we could not detect changes in the levels of hippocampal bulk histone acetylation or the expression of learning-regulated genes in Hdac6−/− mice compared to wild type littermates. These data are in line with the view that unlike other HDACs, HDAC6 does not directly affect chromatin plasticity (Fischer et al, 2010; Valenzuela-Fernández et al, 2008). Concordant with the finding that tubulin is a major substrate of HDAC6 (Haggarty et al, 2003; Hubbert et al, 2002; Zhang et al, 2008), α-tubulin K40 acetylation was elevated in the cortex and hippocampus of Hdac6−/− mice. Moreover, in agreement with the finding that brain morphology was normal in Hdac6−/− mice, we observed that exploratory behaviour, basal anxiety, motor coordination and long-term associative memory consolidation were indistinguishable between Hdac6−/− mice and wild type littermates. Spatial memory formation in the water maze paradigm was even slightly improved in Hdac6−/− mice. The fact that loss of Hdac6 has almost no effect on cognition is interesting. So far, only class I HDACs have been analysed for their role in memory function using knockout mice and previous studies have shown that loss of Hdac2 (Guan et al, 2009) and Hdac3 (McQuown et al, 2010) could enhance memory function and synaptic plasticity in mice. These findings support the view that targeting class I HDACs enhances memory function (Kilgore et al, 2010). However, the role of other classes of HDACs in memory function still remains to be elucidated.
While loss of Hdac6 did not severely affect cognitive function under basal conditions, we found that spatial and associative memory functions were restored in a mouse model for AD when Hdac6 was knocked out. To this end, APPPS1-21 mice that lacked Hdac6 showed enhanced associative and spatial memory functions when compared to APPPS1-21 that express Hdac6. Aβ plaque load was not affected in APPPS1-21 and APPPS1-21 Hdac6−/− mice suggesting that the improvement of cognition was not due to reduced Aβ load. This view is supported by previous studies where therapeutic effects on cognition have been observed in models for AD despite unaltered Aβ plaque load (Govindarajan et al, 2011; Ricobaraza et al, 2009). Especially, since global loss of Hdac6 does not cause any detectable detrimental phenotype, our data clearly suggest that targeting HDAC6 could be a beneficial therapeutic strategy to treat AD. While our data provides genetic evidence that reducing endogenous HDAC6 levels protects against memory impairment in an AD mouse model, it remains to be tested whether pharmacological inhibition of HDAC6 would have similar effects. In line with our data, recent studies have shown that inhibition of HDAC6 in cortical neurons could rescue neurotoxicity linked to oxidative stress (Rivieccio et al, 2009). Another recent study found that inhibition of HDAC6 rescues axonal degeneration in a model of Charcot-Marie-Tooth disease (d'Ydewalle et al, 2011). However, the therapeutic effect of targeting HDAC6 may not readily translate to all conditions of cognitive impairment and other neurodegenerative diseases. For example, we found that loss of Hdac6 did not rescue age-related memory disturbances (Supporting Information Fig S2). Moreover, in a Drosophila model for the poly-Q disease Spinal Bulbar Muscle Atrophy, in which the ubiquitin proteasome system was inhibited, HDAC6 was found to mediate a compensatory increase in autophagy that was linked to neuroprotection (Pandey et al, 2007). Inhibition of HDAC6 has been shown to be protective in Huntington's disease (HD). In cortical neurons, mutated huntingtin protein impairs intracellular transport of neurotrophic factors such as BDNF. Interestingly, reducing HDAC6 activity was able to rescue this phenotype, which was linked to increased tubulin acetylation (Dompierre et al, 2007). However, while loss of Hdac6 in a mouse model for HD resulted in increased microtubule acetylation, it failed to rescue neurodegenerative phenotypes and deficits in motor coordination (Bobrowska et al, 2011). While the precise role of HDAC6 in poly-Q diseases needs to be studied in greater detail, a direct role for HDAC6 in the pathogenesis of AD is suggested by the finding that HDAC6 protein levels are increased in postmortem tissues samples from AD patients (Ding et al, 2008). Consistently, postmortem analysis revealed reduced α-tubulin K40 acetylation in NFT-containing neurons from AD patients (Hempen & Brion, 1996). Similar effects were detected in neurons upon treatment with Aβ peptides (Henriques et al, 2010). In line with these findings, we observed increased HDAC6 levels in APPPS1-21 mice. Consequently, the levels of tubulin K40ac were decreased in APPPS1-21 mice, a phenotype that was rescued in these mice upon loss of Hdac6. This finding indicates that the therapeutic effect of HDAC6 inhibition in APPPS1-21 mice could be linked to altered tubulin acetylation, which is known to regulate microtubule dynamics and intracellular transport (Creppe et al, 2009; Reed et al, 2006). This view is supported by data indicating that Aβ peptides lead to cytoskeletal abnormalities, specifically the impairment of intracellular transport (Henriques et al, 2010; Stokin et al, 2005). Mitochondrial trafficking, which is essential for normal neuronal function and integrity, was found to be impaired in neurons treated with Aβ peptides (Rui et al, 2006). This corroborates findings from postmortem studies showing abnormal intraneuronal mitochondrial distribution in the brains of AD patients (Wang et al, 2009). It is therefore interesting to note that HDAC6 activity has previously been implicated in intracellular transport (Deribe et al, 2009), including mitochondrial trafficking (Chen et al, 2010). In line with these data, we found that administration of ADDLs to hippocampal neurons isolated from wild type mice caused an impairment of mitochondrial trafficking. In contrast, neurons derived from Hdac6−/− mice did not show any significant impairment in mitochondrial trafficking upon ADDL treatment. Furthermore, we observed a greater accumulation of mitochondria in neuronal somata compared to the stratum radiatum in the hippocampi of APPPS1-21 mice, but not APPPS1-21-Hdac6−/− mice, compared to wild type controls. These results suggest that loss of Hdac6 is protective against Aβ-induced impairment of mitochondrial trafficking in hippocampal neurons in vitro as well as in vivo. Therefore, increased tubulin acetylation linked to mitochondrial trafficking might provide one, however most likely not the only, cellular substrate for improved cognitive function observed in APPPS1-21 mice that lack Hdac6.