Despite these promising results, many problems remain. Existing HDAC inhibitors (including sodium butyrate and relatives, trichostatin A and the FDA-approved suberoylanilide hydroxamic acid) are both promiscuous and pleiotropic. They hit multiple HDACs (11 have been identified so far) and display paradoxical effects of neuroprotection and cytotoxicity, through pathways that remain obscure.

The yin and yang of HDAC inhibitors have just become a bit clearer with a study from Rajiv Ratan and colleagues at the Burke Medical Research Institute in White Plains, New York, and the Weill Medical College of Cornell University in New York. In a report published in the January 2 Journal of Neuroscience, the investigators separate the neuroprotective benefits of HDAC inhibitors from their neurotoxic actions. Transient treatment with a number of HDAC inhibitors in a cell model of oxidative stress-induced death elicits full neuroprotection with no cytotoxicity, they show. Neuroprotection parallels the induction of the cyclin-dependent kinase inhibitor p21waf1/cip1, a result that on the surface suggests that stopping cell cycle entry might be the key to the protective effects of HDAC inhibitors. But unexpectedly, cell cycle reactivation was not observed in this model. Rather, the data suggest that p21 functions in the cytosol, to prevent the activation of apoptosis pathways. The results will help efforts to harness the beneficial effects of HDACs to address neurodegenerative disease, while minimizing or eliminating their toxicities.

Two years ago, Ratan and colleagues demonstrated that HDAC inhibitors protect against oxidative stress-induced cell death in neurons both in vivo and in vitro (see ARF related news story). The goal of the new work, first author Brett Langley explained to ARF, was to identify the genes that are involved in the neuroprotective actions of HDAC inhibitors. To do this, the researchers used cultured rat embryonic cortical neurons treated with homocysteate (HCA), which induces oxidative stress by preventing the cellular uptake of cysteine, and thereby reducing glutathione synthesis. In this system, multiple structurally distinct HDAC inhibitors protected the cells from HCA toxicity. However, the compounds themselves all showed some level of toxicity, which limited the analysis of biochemical pathways that could be done. The investigators noticed that longer exposure to the inhibitors increased toxicity, so they asked whether shorter exposure would decrease it. That was the case: the scientists found that by giving cells a 2-8-hour pulse of HDAC inhibitor, they achieved full protection against a 24-hour challenge with HCA, without toxicity.

Using this paradigm, Langley and colleagues looked at the role of the cyclin-dependent kinase inhibitor p21waf1/cip1 in neurons. The p21 gene is a target for HDAC regulation and has been implicated in the response of tumor cells to inhibitors. In addition, p21 inhibits proapoptotic kinases in the cytosol of cells. In cortical neurons, p21 was strongly induced by HDAC inhibitors, and a 2-hour transient treatment was sufficient to elevate p21 for 24 hours, consistent with the protein having a role in neuroprotection. The p21 message was also induced by HDAC inhibitors in vivo in a rat model of stroke. In that model, the inhibitors decreased the extent of ischemic damage, consistent with previous work. Another bit of physiological data supporting the role of p21 in neuroprotection came from p21 knockout mice, who suffered greater damage after ischemic injury than their wild-type relatives.

The expression of p21 was sufficient for neuroprotection in vitro, as indicated by overexpression experiments in the cortical neurons. Surprisingly, p21 was not necessary for protection in vitro, as neurons from p21 knockout mice were just as protected by HDAC inhibitors as neurons from wild-type mice. Langley and colleagues looked for compensatory expression of related cyclin-dependent kinase inhibitors to explain this phenomenon, but found none.

This result presented the researchers with a double puzzle. One model for neuroprotection by p21 posits that cells under stress attempt to re-enter the cell cycle, but fail and end up dying by apoptosis. By inhibiting cell cycle entry, p21 would be neuroprotective. In the neurons lacking p21, however, there was no evidence that this pathway was operating. That made Langley go back to look for any evidence that the neurons were trying to activate the cell cycle. What he found was that oxidative stress caused no changes in the levels of the cyclins D or E, their kinases Cdk4 or Cdk2, or the retinoblastoma protein, a key marker of cell cycle re-entry in postmitotic neurons. The results indicate that in this model of glutathione depletion, aberrant cell cycle entry is not the cause of cell death. Consistent with this idea, the cell cycle inhibitors roscovitine and olomoucine did not protect the cells against HCA toxicity.

How, then, to explain the protection afforded cells by p21 overexpression? An alternative explanation could lie in p21’s cytosolic actions, where it inhibits the stress kinase (JNK and p38) pathways, as well as their upstream regulator, the pro-apoptotic kinase ASK-1. Supporting this alternative, the investigators found that HDAC treatment induced cytosolic p21 expression, which was absent in untreated cells. Where oxidative stress activated the SapK/JNK pathway, HDAC inhibitors blocked the activation. In neuroblastoma cells overexpressing the apoptosis-inducing kinase ASK-1, HDAC inhibitor treatment resulted in an increased association of p21 with ASK-1. “These results are consistent with a model in which ectopic or HDAC inhibitor-mediated increases in cytosolic p21 result in inhibition of oxidative stress-induced kinase signaling via interaction and inhibition of ASK-1,” the authors conclude.

The second part of the puzzle is how HDAC inhibitors protect cells, with p21 induction appearing dispensable, at least in the in vitro cell model. The answer could be that the compounds inhibit multiple HDACs and/or activate overlapping pathways that culminate in neuroprotection. “The important point is that the pro-apoptotic/pro-death, deleterious effects of HDAC inhibition can be separated from the protective effects,” Langley told ARF. “We are working hard to identify specific HDACs whose inhibition is neuroprotective.” To that end, Langley worked with medicinal chemist Alan Kozikowski at the University of Illinois in Chicago to develop new selective inhibitors based on the HDAC8 crystal structure. “Alan came up with a whole range of HDAC inhibitors, and we have looked at their specificity, neuroprotective actions, and toxicity. We found that some HDAC6-selective inhibitors maintain therapeutic efficacy with no toxicity.” That work was published last summer (Kozikowski et al., 2007). Interestingly, Langley said, the HDAC6 inhibitor does not cause increased acetylation of chromatin nor does it induce p21. Possibly, inhibiting cytosolic deacetylases that work on non-histone substrates could be the key in protecting cells against oxidative stress.

Langley is bullish on HDACs inhibitors for neurodegenerative disease because, he says, they show some signs of promoting repair and regeneration in addition to neuroprotection. “This paper is all about neuroprotection,” he said, “But in chronic neurodegenerative states like Alzheimer disease, one half of the problem is protecting neurons. The other is repair and regeneration. We need to be careful that our therapeutic strategies don’t foul the landscape, where neuroprotection might inhibit neuronal function down the road.” In this regard, he finds it encouraging that the HATs play a role in increasing synaptic plasticity and synaptogenesis. Li-Huei Tsai at the Picower Institute in Cambridge, Massachusetts, and colleagues last year hinted at the possibilities when they showed that HDAC inhibitors could restore memory in one mouse model of Alzheimer disease (see ARF related news story).

The work also contributes to the active debate about the role of cell cycle entry in neurodegeneration. Langley said the lab has been very interested in the role of cell cycle re-entry particularly in response to oxidative stress. “We actually started this study with the hypothesis in mind that oxidative stress conditions induce cell cycle re-entry, which leads to the degeneration of neurons. But we found no evidence of cell cycle re-entry,” he explained. However, the question remains whether re-entry, which does occur in response to some stressors, is a cause or effect of cell death, or perhaps even a protective response. “Whether the re-entry is a part of the neurodegenerative process or a tombstone of the process is something that really needs clarification in the field,” Langley said. For the latest on the cell cycle in neurons, see the ARF coverage of a recent SfN symposium (see ARF related news story) devoted to this topic.—Pat McCaffrey.

Reference:
Langley B, D'Annibale MA, Suh K, Ayou I, Tolhurst A, Bastan B, Yang L, Ko B, Fisher M, Cho S, Beal MF, Ratan RR. Pulse inhibition of histone deacetylases induces complete resistance to oxidative death in cortical neurons without toxicity and reveals a role for cytoplasmic p21waf/cip1 in cell cycle-independent neuroprotection. J. Neurosci. 2008 January 2; 28(1):163-176. Abstract