MicroRNA-22 (miR-22) Overexpression Is Neuroprotective via General Anti-Apoptotic Effects and May also Target Specific Huntington’s Disease-Related Mechanisms

Background Whereas many causes and mechanisms of neurodegenerative diseases have been identified, very few therapeutic strategies have emerged in parallel. One possible explanation is that successful treatment strategy may require simultaneous targeting of more than one molecule of pathway. A new therapeutic approach to have emerged recently is the engagement of microRNAs (miRNAs), which affords the opportunity to target multiple cellular pathways simultaneously using a single sequence. Methodology/Principal Findings We identified miR-22 as a potentially neuroprotective miRNA based on its predicted regulation of several targets implicated in Huntington’s disease (histone deacetylase 4 (HDAC4), REST corepresor 1 (Rcor1) and regulator of G-protein signaling 2 (Rgs2)) and its diminished expression in Huntington’s and Alzheimer’s disease brains. We then tested the hypothesis that increasing cellular levels of miRNA-22 would achieve neuroprotection in in vitro models of neurodegeneration. As predicted, overexpression of miR-22 inhibited neurodegeneration in primary striatal and cortical cultures exposed to a mutated human huntingtin fragment (Htt171-82Q). Overexpression of miR-22 also decreased neurodegeneration in primary neuronal cultures exposed to 3-nitropropionic acid (3-NP), a mitochondrial complex II/III inhibitor. In addition, miR-22 improved neuronal viability in an in vitro model of brain aging. The mechanisms underlying the effects of miR-22 included a reduction in caspase activation, consistent with miR-22′s targeting the pro-apoptotic activities of mitogen-activated protein kinase 14/p38 (MAPK14/p38) and tumor protein p53-inducible nuclear protein 1 (Tp53inp1). Moreover, HD-specific effects comprised not only targeting HDAC4, Rcor1 and Rgs2 mRNAs, but also decreasing focal accumulation of mutant Htt-positive foci, which occurred via an unknown mechanism. Conclusions These data show that miR-22 has multipartite anti-neurodegenerative activities including the inhibition of apoptosis and the targeting of mRNAs implicated in the etiology of HD. These results motivate additional studies to evaluate the feasibility and therapeutic efficacy of manipulating miR-22 in vivo.


Introduction
HD is a fatal hereditary neurodegenerative disease caused by a CAG repeat expansion mutation in exon 1 of the Huntingtin (Htt) gene, which results in a toxic polyglutamine (polyQ) stretch in the encoded Huntingtin (Htt) protein. The clinical features of disease include abnormal involuntary choreiform movements as well as mood and personality changes. Despite the monogenic nature of the HD, the disease demonstrates a complex etiology, likely due to the fact that mutant Htt resides in multiple cellular compartments and complexes, where it may disrupt organellar function and form protein aggregates [1]. Although the roles of various soluble and aggregated Htt species in the disease process remain to be elucidated, HD is widely considered to be a misfolded protein accumulation disorder and intracellular redistribution of Htt generally foretells its neurotoxicity. Specific cellular mechanisms implicated in HD pathogenesis include the dysregulation of transcription, cell signaling, and metabolism. microRNAs (miRNAs) are 19-24 nucleotide noncoding RNAs that act as negative regulators of post-transcriptional gene expression [2,3]. A single miRNA can target many mRNAs in parallel, comprising a potential way to simultaneously modify the activities of multiple pathways. If suitable endogenous or exogenous sequences could be found, miRNAs could comprise a rational new approach to the treatment of mechanistically complex diseases, including neurodegenerative diseases. In addition to their a priori potential for multipartite neuroprotective activities, the dysregulation of miRNA expression has been specifically implicated in the early pathogenesis of neurodegenerative diseases [4], including Huntington's disease (HD) [5][6][7][8]. Therefore, restoring endogenous levels of miRNA expression might also be used to slow disease onset or progression. This has, for example, been shown in amyotrophic lateral sclerosis (ALS), where upregulation of miR-206 has been shown to be part of a cellular autocompensatory mechanism by promoting regeneration at neuromuscular synapses and slowing ALS progression through the downregulation of HDAC4 expression [9].
With the goal of identifying endogenous miRNA sequences that might simultaneously engage a range of HD therapeutic targets, we set out to identify brain miRNAs with a suitable target profile. We also took into account evidence for miRNA dysregulation in HD, and found one miRNA, miR-22, that fulfilled both criteria.

miR-22 Targets Multiple Genes Involved in Huntington's Disease
miR-22 is an miRNA that has previously been shown to be downregulated in both HD and Alzheimer's disease brain [7,10]. Using the miRNA target algorithm TargetScan, we discovered that miR-22 is predicted to target multiple mRNAs that have been implicated in HD pathogenesis. First, miR-22 is predicted to target Rcor1, a crucial regulator of neuronal gene expression; the Restrictive Element 1 Silencing Transcription Factor pathway has been shown to be hyperactive in HD, leading to a large-scale repression of neuronal genes in affected neurons [11]. Second, miR-22 is predicted to target Rgs2; we have recently shown that decreased Rgs2 expression in striatal neurons is protective in in vitro models of HD by decreasing extracellular-signal-regulated kinase (ERK) activation [12]. miR-22 was also predicted to target HDAC4; reversal of acetylation defects have been shown to ameliorate neurodegeneration in cellular and animal models of HD [13], and recent evidence has indicated that the administration of HDAC inhibitor suberoylanilide hydroxamic acid (SAHA) may work via diminishing HDAC4 expression post-transcriptionally [14].
We therefore hypothesized that miR-22 might be a potential therapeutic approach to treating HD via tandem regulation of the above pathways. In order to first confirm the pertinent target predictions, we used 39 untranslated region (UTR) luciferase assays to show that miR-22 specifically interacts with the 39 UTRs of the Rcor1, Rgs2 and HDAC4 mRNAs (Figure 1), compared to control miRNAs that are not predicted to target these mRNAs (miR-146a served as the control for Rcor1 and HDAC4 whereas miR-153 served as a control for Rgs2 because Rgs2 contains a predicted miR-146a binding site).
We next tested the potential for miR-22 to achieve neuroprotection. In these experiments we assessed miR-22 overexpression in the test case to no miRNA expression as a control; we reasoned that this condition provided a high-stringency comparator, since either miRNA overexpression or exposure to additional lentiviral vector exposure could itself potentially compromise neuronal survival and confound the interpretation of our results. In other words, comparing the effect of miR-22 to the effect of another miRNA is problematic because it would be impossible to distinguish whether miR-22 were neuroprotective, or merely less toxic, than the other lentivirally-delivered miRNA.
We first assessed the effect of miR-22 overexpression in in vitro models of HD comprising lentiviral-mediated expression of mutant Htt fragments (Htt171-82Q versus Htt171-18Q) in primary rat striatal or cortical neurons [15,16]; these models have previously been shown to exhibit HD-like neuropathologies and gene expression changes [17]. As expected, Htt171-82Q decreased the viability of both striatal and cortical neurons; moreover, as predicted viability of the neuronal cells was restored by miR-22 overexpression ( Figure 2).
Although miR-22 is not predicted to target Htt, we nonetheless considered whether our observed disease-modifying effects might involve changes in Htt accumulation. This was assessed in parallel cultures immunostained with anti-Htt antibodies (see Methods), in which quantification of Htt-enriched foci of $2 mm (which might represent either compartmentalized of Htt accumulation or Httpositive inclusions) was undertaken. To our surprise, the number of Htt-enriched foci was diminished by miR-22 ( Figure 3). We were able to confirm this activity in multiple experiments, but have not yet been able to identify a predicted target to account for the mechanism of this effect. Nonetheless, the reduction of mutant Htt accumulation (or correction of abnormal Htt distribution) by miR-22 might contribute to its disease-mitigating activity in HD.

miR-22 is Protective in Multiple in vitro Models of Neurodegeneration
In order to test whether non-Htt-targeting effects also contributed to the anti-neurodegenerative activity of miR-22, we employed a different HD model that would not involve Htt expression. Previous studies have shown that administration or accidental ingestion of 3-NP, an irreversible inhibitor of succinate dehydrogenase [18] (a component of the Krebs cycle and complex II/III of the mitochondrial electron transport system), produces neuropathologic and motor phenotypes similar to that of HD [19,20]. We employed an acute in vitro version of this model, which involved treating striatal neurons with 3-NP for 48 hours [21]. We again chose to test the beneficial effects of miR-22 Luciferase constructs expressing the Renilla luciferase gene under the control of the corresponding 39 UTRs were co-transfected in HEK293T cells with either a plasmid expressing miR-22 or an empty plasmid (pcDNA3.1). The intensity of Renilla luciferase luminescence was normalized to firefly luciferase luminescence intensity, which was not under the control of a 39 UTR. Graphs represent mean values 6 SD, n = 6. a indicates p-value ,0.01, by unpaired t test compared to a control sample transfected with the empty vector pcDNA3.1. b indicates p,0.01, by unpaired t test compared to the control sample transfected with a miRNA not predicted to bind to the 39 UTR (miR-146a or miR-153). c indicates p,0.05, by unpaired t test compared to the control sample transfected with a miRNA which is not predicted to bind to the 39 UTR (miR-146a). We then asked whether miR-22 could prevent neurodegeneration in a non-HD-like condition. We thus assessed the effects of miR-22 against neurodegeneration due to long-term culture stress. This model relies on long-term culture of cortical neurons for .5 weeks, after which time a significant decrease in the number of   Figure 4, we observed that initially the exposure to miR-22-expressing lentiviral vector decreased neuronal number (reflecting cell loss due to nonspecific toxicity of the lentiviral vector application). However, as hypothesized, expression of miR-22 increased neuronal long-term neuronal survival, as measured by NeuN-positive cell counts at 5 weeks in vitro ( Figure 5).

miR-22 also Targets Other Cell Death Regulators
Given that none of the HD-related targets explored above have been implicated in neuronal aging, we considered what other targets of miR-22 might be relevant for its broader neuroprotective effects. In so doing, we noted that miR-22 is also predicted by TargetScan to target the 39 UTRs of the Tp53inp1 and MAPK14/p38 mRNAs, whose encoded proteins have known pro-apoptotic activities [23,24]. We subsequently confirmed the targeting of these mRNAs by miR-22 using 39 UTR luciferase assays, and also showed by Western blot that miR-22 overexpression in primary neurons decreases the cellular accumulation of MAPK14/p38 protein ( Figure 6A, B). (We were unable to reliably detect Tp53inp1 protein in our rat primary cultures by Western blot, preventing us from assessing a possible effect of miR-22 on its accumulation.) We also assessed the effects of miR-22 on levels of MAPK14/p38 protein and its activation during neurodegeneration. These analyses showed that both the increased cellular levels of total MAPK14/p38 ( Figure 6C) and activated phosphorylated MAPK14/p38 associated with neurotoxicity were abrogated by the overexpression of miR-22 ( Figure 6D). These data support the conclusion that the inhibition of MAPK14/p38 activity comprises one of the neuroprotective capabilities of miR-22.

miR-22 Inhibits Neuronal Apoptosis
Having determined that miR-22 inhibits the expression of proapoptotic proteins, we next sought to evaluate whether miR-22 inhibited downstream manifestations of apoptosis. We thus tested this hypothesis by measuring the effect of miR-22 overexpression on the activity of effector caspases (3 and 7). As expected, we observed significant activation of effector caspases in all models of neurodegeneration (Htt171-82Q, 3-NP, and aging), which in each case was inhibited overexpression of miR-22 ( Figure 7A, B, C).
Taken together, the above results show a therapeutic potential of miR-22 mediated by both general anti-apoptotic effects (via targeting MAPK14 and Trp53inp1), and specific HD-related effects (via Rcor1, HDAC4, Rgs2 and Htt).

Discussion
The dysregulation of miRNA expression has been implicated in the pathogenesis of a number of neurodegenerative diseases. In addition to being an integral part neuropathogenesis, miRNAs can serve as disease modulators. For example, upregulation of miR-206 has been shown to slow the progression of neurodegeneration in a mouse model of amyotrophic lateral sclerosis by promoting regeneration of neuromuscular synapses [9]. Since a single miRNA can have a broad effect by targeting tens or hundreds of genes, miRNAs may represent a new approach to polytherapy for complex diseases. Here, we have shown that increasing cellular levels of miR-22 might be one such approach for combating neurodegeneration.
We demonstrated that overexpression of miR-22 is neuroprotective in in vitro models of HD, which may be explained by miR-229s targeting multiple genes previously implicated in HD pathogenesis. These include Rcor1 and HDAC4, which have been implicated in HD-related abnormalities in gene expression [11,25]. Rcor1 is a transcriptional co-repressor in the REST/ NRSF pathway, which regulates the expression of neuronal genes including brain-derived neurotrophic factor, an important neuronal plasticity modulator and pro-survival factor for striatal and cortical neurons. Histone hypoacetylation has also been observed in HD and histone deacetylase inhibitors can ameliorate disease phenotype in animal models of HD [26][27][28]. HDAC4 is of particular interest, given that it may be an important mediator of the effects of the HDAC inhibitor SAHA [14]. Not least of all, downregulation of Rgs2 expression is a consistent feature of human HD and multiple HD models, and our own recent findings have demonstrated that inhibition of Rgs2 is protective in HD though enhancement of ERK activity [12]. Moreover, direct or indirect inhibition of mutant Htt accumulation (albeit through a presently unknown mechanism) would be of clear benefit in HD. In addition to HD-specific pathways, miR-22 also demonstrated the ability of to inhibit apoptosis, as shown by its ability to decrease effector caspase activation. We propose that this effect may be  mediated, at least in part, by its targeting pro-apoptotic genes MAPK14/p38 and Trp53inp1.
The targeting of general anti-apoptotic and HD-specific pathways, together with evidence that it is downregulated in HD brain, make miR-22 a particularly interesting strategy for combating HD. In addition, based on its anti-apoptotic properties and abilities to prevent neurodegeneration in non-Htt-based paradigms, it may also be interesting to consider its utility in other neurodegenerative conditions. Our encouraging in vitro results provide an excellent rationale for investigating whether miR-22 can also achieve neuroprotection in vivo. However, it is somewhat unclear whether direct manipulation of miR-22 overexpression would be the preferred strategy to achieve increased miR-22 levels. Such manipulation might be plausible if localized delivery to a small set of affected neurons is sufficient. Alternatively, exploring other ways by which miR-22 levels could be regulated by small-molecule drugs might be another therapeutic option.

Conclusion
Our study demonstrates that enhancing miR-22 expression in neurons has a pronounced neuroprotective against a variety of medically relevant neuronal insults. These results support the perspective that enhancing miR-22 expression might comprise a rational therapeutic strategy for the treatment of HD and other neurodegenerative conditions.  . Protein levels were normalized to b-3 tubulin, which is not a predicted target of miR-22 and whose expression did not change with miR-22 treatment. Graphs represent quantification of the Western blot, error bars represent SEM, ** represents p-value ,0.01, by unpaired t-test. CTRL -non-infected cells. C) and D) Primary striatal neurons were infected on DIV1 with lentiviral vectors encoding WT (Htt18Q) or mutant (Htt82Q) Htt171 fragments under the control of the TRE promoter and co-infected on DIV4 with vector encoding miR-22 and 2.5 weeks post-infection proteins were harvested. Protein levels were normalized to b-3 tubulin, which is not a predicted target of miR-22. Graphs represent quantification of the signals from the Western blot, error bars represent SEM. * represents p-value ,0.05, ** represents p-value ,0.01, by unpaired t-test. doi:10.1371/journal.pone.0054222.g006 PGK-miR-22-WPRE) were produced in human embryonic kidney 293T (HEK293T) cells with a four-plasmid system as described previously [29]. Rno-miR-22 gene was cloned together with the surrounding genomic region of total length 295 bp using the following primers: F-miR-22 cacccagagctaggttcagatggccagtgggtc and R-miR-22 ctgtaaagccgtcaccccagctgtccagac. SIN-PGK-miR-22-WPRE plasmid was also used to overexpress the miRNA by transfection in human embryonic kidney (HEK) T293 cells in 39 UTR-luciferase assays.

Primary Cultures and Lentiviral Infections
Primary cortical and striatal neuron cultures used for neuroprotection assays were prepared from striata (ganglionic eminences) or cerebral cortices of E16 rat embryos as described in [29]. This procedure has been demonstrated to yield a high majority of neuronal cells with some residual glial cells [17]. Cells were plated in 96-well plates, not using the wells in the most outer rows (rows A and H) and columns (1 and 8).
In vitro lentiviral models of HD were prepared as described in [17]. Primary striatal or cortical neuronal cultures were infected at one day in vitro (DIV) with lentiviruses encoding the first 171 amino acids of wild-type or mutant human Htt (SIN-W-TRE-Htt171-18/82Q, respectively). Htt-expressing lentiviruses were applied at a concentration of 25 ng p24 antigen/ml culture medium together with a vector encoding the tetracyclinecontrolled transactivator tTA1 under the control of PGK promoter (at a concentration of 40 ng p24/ml). At DIV4 neurons were infected with lentiviral expression constructs for rno-miR-22 (at a p24 concentration of 25 ng/ml). Three weeks after the infection of Htt constructs, cells were fixed and their survival assessed by NeuN counting. The transduction efficiency of primary neurons under these conditions has been shown to be $ 95% [17,29].
Forty-eight hours later cells were fixed and their survival assessed by NeuN-positive cell counting.

Cellular Model of General Neurodegeneration
Primary cultures of cortical neurons were prepared as in [16] and infected at DIV10 with a lentiviral vector expressing rno-miR-22 (at a p24 concentration of 25 ng/ml). Four weeks after lentiviral infection cells were fixed and their survival assessed by NeuN-positive cell counting.
Images for neuroprotection assays were acquired with a BD Pathway 855 instrument and cell bodies or Htt-positive foci were counted with ImageJ software [30]. Images for quantitative analyses were acquired under non-saturating exposure conditions. Htt accumulation was quantified as numbers of Htt-positive foci of a mean fluorescence intensity .10 times the overall mean pixel intensity and $2 mm in diameter per well. These foci were observed in N171-82Q-expressing, but not N171-18Q-expressing cells [29].

Immunoblotting
For analysis of protein expression cells were harvested in RIPA buffer (Sigma), containing protease inhibitor cocktail (Sigma). Proteins were separated on 10% SDS-polyacrylamide gels and transferred to nitrocellulose membranes. Membranes were blocked for one hour in 30% OdysseyH Blocking Buffer (LI-COR Biosciences), 0.1% Tween-20 in PBS followed by incubation  (Htt82Q), B) treated with 150 mM 3-NP or C) long-term culture (to model aging). The miR-22-overexpressing condition is compared to the control with no miRNA overexpression (see text for rationale). Graphs represent mean 6 SEM, ** represents p-value ,0.01, by unpaired t-test, CTRL -non-infected cells. doi:10.1371/journal.pone.0054222.g007