Connectivity mapping uncovers small molecules that modulate neurodegeneration in Huntington’s disease models
- 1.3k Downloads
Huntington’s disease (HD) is a genetic disease caused by a CAG trinucleotide repeat expansion encoding a polyglutamine tract in the huntingtin (HTT) protein, ultimately leading to neuronal loss and consequent cognitive decline and death. As no treatments for HD currently exist, several chemical screens have been performed using cell-based models of mutant HTT toxicity. These screens measured single disease-related endpoints, such as cell death, but had low ‘hit rates’ and limited dimensionality for therapeutic detection. Here, we have employed gene expression microarray analysis of HD samples—a snapshot of the expression of 25,000 genes—to define a gene expression signature for HD from publically available data. We used this information to mine a database for chemicals positively and negatively correlated to the HD gene expression signature using the Connectivity Map, a tool for comparing large sets of gene expression patterns. Chemicals with negatively correlated expression profiles were highly enriched for protective characteristics against mutant HTT fragment toxicity in in vitro and in vivo models. This study demonstrates the potential of using gene expression to mine chemical activity, guide chemical screening, and detect potential novel therapeutic compounds.
Single-endpoint chemical screens have low therapeutic discovery hit-rates.
In the context of HD, we guided a chemical screen using gene expression data.
The resulting chemicals were highly enriched for suppressors of mutant HTT fragment toxicity.
This study provides a proof of concept for wider usage in all chemical screening.
KeywordsHuntington’s Huntingtin Connectivity Map cMap Chemical screening Therapeutic detection
Huntington’s disease (HD) is a neurodegenerative disease caused by a polyglutamine (polyQ) expansion in the huntingtin (HTT) protein . HD is characterized by mood and cognitive impairments, motor dysfunction (including chorea) and, ultimately, death . The molecular mechanisms underlying the cellular toxicity of mutant HTT are not entirely clear. However, both the loss of normal HTT function and the toxic gain of function of the mutant HTT protein are thought to be important in HD progression [3, 4]. A number of pathological mechanisms have been implicated in the onset and progression of HD, including mitochondrial dysfunction , neurotrophic factor deprivation , and excitotoxicity .
One of the most studied characteristics of mutant HTT is its propensity to form stable inclusions or aggregates . These inclusions have been implicated in activating cellular stress mechanisms such as ER stress  and may play a role in protecting cells from mutant HTT-induced cell death . Current evidence suggests that soluble mutant HTT oligomers produced early in the aggregation process are the major cause of HD neuropathology , and these have been shown to affect mitochondrial function . However, mutant HTT aggregates can sequester a number of transcription factors which contain polyQ-rich regions—such as CBP , TBP  and SP1 —both in the nucleus and in the cytoplasm. Sequestration of these proteins reduces the pool of transcription factors available to activate/repress their target genes , manifesting in transcriptional dysregulation that has been extensively characterized in human samples , in vivo [18, 19] and in vitro models [20, 21].
No drugs are currently available for delaying disease onset or progression in HD. In order to identify chemicals that might interfere with the molecular mechanisms of mutant HTT toxicity and prevent neurodegeneration, various high-throughput chemical screens have been performed in yeast , mammalian cells [23, 24] and cell-free assays . In general, these screens have tested chemical libraries containing hundreds to thousands of chemicals for their effects on either aggregation of amino-terminal fragments of mutant HTT—which contain the polyQ stretch—or cellular apoptosis resulting from expression of these mutant HTT fragments. While such screens have identified benzothiazoles  and mTOR inhibitors  as potential therapeutic agents, these approaches have not yet been successfully translated to human therapies. One underlying issue may be the oversimplified nature of the screens, which focus on single endpoints.
We sought to identify chemicals that ameliorate mutant HTT fragment toxicity using gene expression signature-based chemical screening. Such approaches use a transcription pattern to describe a biological condition. By finding similarities and differences between the transcription patterns of various biological conditions, one can develop hypotheses regarding the similarities and differences of the biological conditions themselves. This approach has been used extensively in the cancer field to distinguish between cancer subtypes . Linking chemical activity with gene expression has been demonstrated in yeast [27, 28], where compendiums of gene expression data for around 100 chemicals were created and hierarchical clustering used to group similar chemicals together with some success. More recently, chemical treatments in human cell lines were used to create a database for the Connectivity Map (cMap) . The cMap is a technique which employs a transcriptional profile of interest to mine a database of gene expression data from 1600 chemical treatments and returns a measure of positive or negative similarity to each chemical as an output. By identifying similarities between the transcription patterns produced by different chemicals, functional similarities can be identified between those compounds. The major advantage of this approach is that this requires no prior understanding of the underlying biological processes at play, and is therefore unbiased. The initial aim of the cMap was to connect chemicals and diseases via similar transcriptional patterns. However, there are as yet few examples where a transcriptional pattern for a disease has been used as a ‘target’ for predicting potential therapeutic agents. The most notable employed a transcription pattern from fasting muscle tissue to identify urosolic acid as an inhibitor of skeletal muscle atrophy .
Transcriptional dysregulation in HD is well documented, making this disorder a particularly good paradigm for testing the disease relevance of the cMap technique. Using gene expression profiling data from human HD brains , we used the cMap to identify chemicals that induced positive (similar) or negative (inverse) transcriptional patterns to that of HD in an unbiased manner. The predicted chemicals were subsequently screened for neuroprotection in well-established acute mammalian cell and Drosophila models of HD that have previously been employed to detect chemicals that protect against mutant HTT toxicity [23, 31]. These HTT fragment-based models are well suited to chemical screening due to the rapid manifestation of a broad range of disease-relevant phenotypes with mechanistic relevance to HD , though they may not capture some disease phenotypes dependent upon full-length HTT protein. Using this approach, we identified an enriched list of compounds that ameliorate disease phenotypes in mammalian cell and Drosophila models of mutant HTT fragment toxicity, all of which produce an opposite transcription pattern to that observed in the HD patient samples. Positively correlated chemicals had no effect and served as an integrated control. These data provide a proof of concept and highlight the promise of the cMap for uncovering novel therapeutic strategies.
PC12 cells stably transfected with a ponasterone A-inducible mutant HTT fragment (103Q) (first 17 amino acids of HTT plus polyQ repeat fused with green fluorescent protein (GFP); referred to as HTT103Q in the text)—known as the Htt14A2.5 cell line (Apostol et al. 2003)—were maintained and passaged in DMEM supplemented with 10 % horse serum (v/v), 5 % FCS (v/v) with 2 mM GlutaMax and 1 mM sodium pyruvate in T75 flasks (Greiner, UK) and incubated at 37 °C in a 5 % CO2 atmosphere.
Drosophila husbandry, compound feeding and assays
Flies were maintained in standard maize food at 25 °C in light/dark cycle of 12:12. The elavGAL4 [c155] fly stocks was obtained from the Bloomington Stock Center (Bloomington, IN). The HTT93Q line was a gift from Larry Marsh and Leslie Thompson . Crosses were set up between male flies carrying elavGAL4 driver and virgin females carrying the HTT93Q transgene. In the F1 generation only females expressed the HTT93Q, while males were used as controls.
Fly treatment, cell viability, caspase induction, aggregate counting and statistical analyses are described in the Supplementary Material.
Connectivity mapping identifies chemicals predicted to modulate HD phenotypes
Inversely connected chemicals reduce mutant HTT-induced caspase activation
Deferoxamine and oligomycin reduce mutant HTT inclusion body formation
Of the seven compounds tested, only deferoxamine and oligomycin significantly altered the formation of HTT103Q-containing inclusion bodies (Fig. 3c, d). Deferoxamine treatment yielded a dose-dependent reduction in the number and the intensity of GFP-tagged HTT103Q aggregates in PC12 cells. While no effect was observed at 5 μM, a significant reduction in aggregation was observed at 10 and 50 μM (P < 0.05 and P < 0.01, respectively), with a maximal reduction of ~50 % (Fig. 3d). Oligomycin also reduced the number of HTT103Q aggregates by ~50 %, with the most robust effect observed at the concentration of 1 μM (P < 0.05) (Fig. 3d). These data highlight that the chemicals identified by the cMap likely have varied mechanisms of protection, as all of the chemicals tested for effects on HTT103Q aggregation prevented HTT103Q-induced toxicity, yet only oligomycin and deferoxamine had a notable effect on HTT103Q aggregation.
Chlorzoxazone and deferoxamine ameliorate neurodegeneration in mutant HTT expressing fruit flies
The modulation of HD gene expression divides negatively correlated chemicals into two distinct groups
Our primary aim in this study was to employ the cMap approach to predict chemicals that may protect against mutant HTT-induced toxicity. In order to achieve this, we employed a comprehensive microarray dataset for HD and control human brain samples  to generate the HD “gene signature.” The largest numbers and magnitudes of transcriptional changes observed were present in the caudate nucleus  (Fig. 1a), corresponding with evidence indicating that the caudate nucleus is one of the earliest and most affected brain regions in HD [33, 38]. However, it should be noted that as many as 50 % of the caudate nucleus neurons may be lost by grade 2 —which undoubtedly contributed to the transcriptional changes observed here. Nonetheless, the gene expression changes showing the most extreme magnitudes of down-regulation in HD—which have been validated by laser-capture microdissection—indicate these likely reflect alterations in expression at the mRNA/cell level .
The cMap database was queried with the gene signature derived from the human HD data. There is an obvious disconnect between the characteristics of neuronal cells in vivo and the cultured immortalized cells used to construct the cMap database. However, in both cases, the disease/chemical induced gene expression was compared with a precise control, therefore minimizing the contribution of cell-specific gene expression. Differing cell machinery would still influence the transcriptional response to a disease or chemical stimulus. Despite this, it was recently demonstrated using the cMap that a gene expression profile for skeletal muscle atrophy generated from mouse muscle tissue was able to identify ursolic acid as a compound whose transcriptomic effect was inversely correlated to that of disease, and which subsequently was found to reduce muscle atrophy in this mouse model . This not only demonstrates the potential of the cMap to identify connections between diseases and chemicals across the in vivo/in vitro boundary, but also with differences in time course and species.
We hypothesized that the HD gene signature reflected a mixture of gene expression changes due to mutant HTT toxicity as well as protective cellular response(s). Therefore, both positively and negatively correlated chemicals were selected in an unbiased manner and screened for effects on mutant HTT fragment toxicity. Of the 12 positive chemical connections—whose expression signatures correlated with those of HD—none significantly reduced HTT103Q toxicity. In contrast, 7 of the 12 chemicals whose expression signatures negatively correlated with those of HD significantly reduced HTT103Q-induced caspase activation (Fig. 2). Here, the positively correlated group of chemicals also serves as a control to emphasize that the negatively correlated chemicals are truly enriched for protective characteristics. These data suggest that several of the corresponding genes in the HD signature reflect the toxicity of mutant HTT, corroborating the idea that significant transcriptional changes observed in the human HD gene signature are caused by the dysfunction of functionally important transcription factors [40, 41].
The role of HTT aggregates in HD pathogenesis is controversial. There is evidence that HTT aggregates cause cellular stress and cell death by apoptosis , as well as conferring a protective mechanism . Indeed, various chemicals that either increase or decrease HTT aggregation have been shown to be cytoprotective . Nonetheless, aggregation of mutant HTT is an important hallmark of HD pathology that provides insight into pathogenesis. With this in mind, we tested whether seven chemicals that reduced HTT103Q toxicity in PC12 cells could modulate HTT103Q aggregation, and found that both oligomycin and deferoxamine significantly reduced HTT103Q aggregation (Fig. 3). Oligomycin is an inhibitor of ATP synthase (complex V) that utilizes the mitochondrial membrane proton gradient to convert ADP to ATP . Oligomycin, as well as other metabolic inhibitors, have previously been shown to rescue cell death in HD models . It is thought that such inhibitors reduce cell death by activating the pro-survival ERK and AKT pathways . Although the ability of oligomycin to rescue mutant HTT-induced cell death has previously been documented, its effect on mutant HTT aggregation has not. Unfortunately, oligomycin is extremely toxic in vivo in mammals, causing metabolic acidosis and nephrotoxicity . Thus, while the identification of oligomycin via the cMap validates this approach, it would not represent a viable therapeutic compound.
Deferoxamine (or desferal) is a bacterial siderophore with a high affinity for iron and has antioxidant properties , which is used therapeutically as an iron-chelating agent. Short exposure to deferoxamine is an effective treatment for acute iron overdose  and is used as a longer-term treatment for patients with blood transfusion-related iron overload . Deferoxmine has also been used to protect against both lipopolysaccharide- and doxorubicin-induced toxicity [48, 49], where reactive oxygen species (ROS) are known to play an important role. Oxidative stress has been implicated in mutant HTT-induced toxicity, with lipid peroxidation products such as malondialdehyde observed in HD-affected regions of the brain . Here, we found that deferoxamine reduced mutant HTT toxicity and aggregation in the PC12 cell model, and as well as ameliorating neurodegeneration in mutant HTT-expressing flies. This complements previous work showing HTT inclusion bodies contain iron, and chelators such as deferoxamine alter aggregate dynamics  and are neuroprotective. This suggests that deferoxamine is effective upstream of HTT-aggregation but may also be protective due to its antioxidant properties, similar to our positive control, ebselen.
Chlorzoxazone significantly reduced HTT-induced caspase activation in PC12 cells and rhabdomere loss in flies expressing HTT93Q exon 1, while having no effect on HTT aggregation. Chlorzoxazone is a clinically used skeletal muscle relaxant which relieves discomfort due to muscle spasm . Previous work has suggested that chlorzoxazone reduces excitatory neurotransmission by modulating calcium-activated SK potassium channels . Interestingly, chlorzoxazone has been proposed to counteract the neurologic effects of CACNA1A mutations . Chlorzoxazone also has anti-inflammatory properties , which may also be relevant for HD, where neuroinflammation has been heavily implicated .
In this study, we have thus validated the effectiveness of the cMap approach to establish therapeutic relationships between chemicals and HD based upon gene expression profiles. We have demonstrated that by using a gene signature for HD, the cMap can identify potential therapeutic agents with a hit rate that far surpasses typical phenotypic screens. Excitingly, we were able to identify protective chemicals with multiple modes of action, a feature that would have been missed in a phenotypic screen of mutant HTT fragment aggregation alone. It is also noteworthy that the multidimensionality of a whole genome gene expression assay allows the identification of chemicals for multiple characteristics—meaning the same data could be used for multiple disease screens. Thus, this work serves as a proof of concept for such screens to be extended to other diseases where transcriptional dysregulation is pivotal, and also for further HD-related screens to be carried out with larger gene-expression datasets (Fig. 5), for instance in mutant-HTT expressing cell lines. Indeed, a greatly expanded version of the cMap database is soon to be released to the scientific community which utilizes the L1000 gene expression measurement method (http://www.broadinstitute.org/LINCS/). The database will be dramatically enlarged from ~6000 profiles for 1300 chemicals in four cell lines, to ~576,000 profiles for 4000 chemicals, 9000 gene knockdowns and 3000 gene overexpressions in ten cell lines. Thus, it is clear that the power of the cMap to identify chemicals/genes that modulate disease phenotypes will be greatly increased. The work presented here also forms a strong rationale for the inclusion of a database with mutant HTT-expressing cell treatments in future cMap iterations. This new resource—combined with the interrogation of additional gene expression data sets at multiple pathological stages for HD and other diseases—will provide a powerful tool for identification of candidate therapeutic chemicals for these disorders.
JLS was supported by a PhD studentship from the Medical Research Council (MRC) Toxicology Unit, whom we also thank for core support. FG was supported by a MRC New Investigator award (FG - G0700090).
Conflict of interest
None of the above authors have a conflict of interest.
- 6.Strand AD, Baquet ZC, Aragaki AK, Holmans P, Yang L, Cleren C, Beal MF, Jones L, Kooperberg C, Olson JM et al (2007) Expression profiling of Huntington's disease models suggests that brain-derived neurotrophic factor depletion plays a major role in striatal degeneration. J Neurosci 27(43):11758–11768CrossRefPubMedGoogle Scholar
- 11.Nucifora LG, Burke KA, Feng X, Arbez N, Zhu S, Miller J, Yang G, Ratovitski T, Delannoy M, Muchowski PJ et al (2012) Identification of novel potentially toxic oligomers formed in vitro from mammalian-derived expanded huntingtin exon-1 protein. J Biol Chem 287(19):16017–16028PubMedCentralCrossRefPubMedGoogle Scholar
- 12.Shirendeb U, Reddy AP, Manczak M, Calkins MJ, Mao P, Tagle DA, Reddy PH (2011) Abnormal mitochondrial dynamics, mitochondrial loss and mutant huntingtin oligomers in Huntington's disease: implications for selective neuronal damage. Hum Mol Genet 20(7):1438–1455PubMedCentralCrossRefPubMedGoogle Scholar
- 16.Steffan JS, Kazantsev A, Spasic-Boskovic O, Greenwald M, Zhu YZ, Gohler H, Wanker EE, Bates GP, Housman DE, Thompson LM (2000) The Huntington's disease protein interacts with p53 and CREB-binding protein and represses transcription. Proc Natl Acad Sci U S A 97(12):6763–6768PubMedCentralCrossRefPubMedGoogle Scholar
- 18.Kuhn A, Goldstein DR, Hodges A, Strand AD, Sengstag T, Kooperberg C, Becanovic K, Pouladi MA, Sathasivam K, Cha JH et al (2007) Mutant huntingtin's effects on striatal gene expression in mice recapitulate changes observed in human Huntington's disease brain and do not differ with mutant huntingtin length or wild-type huntingtin dosage. Hum Mol Genet 16(15):1845–1861CrossRefPubMedGoogle Scholar
- 20.Apostol BL, Illes K, Pallos J, Bodai L, Wu J, Strand A, Schweitzer ES, Olson JM, Kazantsev A, Marsh JL et al (2006) Mutant huntingtin alters MAPK signaling pathways in PC12 and striatal cells: ERK1/2 protects against mutant huntingtin-associated toxicity. Hum Mol Genet 15(2):273–285CrossRefPubMedGoogle Scholar
- 22.Floto RA, Sarkar S, Perlstein EO, Kampmann B, Schreiber SL, Rubinsztein DC (2007) Small molecule enhancers of rapamycin-induced TOR inhibition promote autophagy, reduce toxicity in Huntington's disease models and enhance killing of mycobacteria by macrophages. Autophagy 3(6):620–622CrossRefPubMedGoogle Scholar
- 24.Bodner RA, Outeiro TF, Altmann S, Maxwell MM, Cho SH, Hyman BT, McLean PJ, Young AB, Housman DE, Kazantsev AG (2006) Pharmacological promotion of inclusion formation: a therapeutic approach for Huntington's and Parkinson's diseases. Proc Natl Acad Sci U S A 103(11):4246–4251PubMedCentralCrossRefPubMedGoogle Scholar
- 25.Heiser V, Engemann S, Brocker W, Dunkel I, Boeddrich A, Waelter S, Nordhoff E, Lurz R, Schugardt N, Rautenberg S et al (2002) Identification of benzothiazoles as potential polyglutamine aggregation inhibitors of Huntington's disease by using an automated filter retardation assay. Proc Natl Acad Sci U S A 99(Suppl 4):16400–16406PubMedCentralCrossRefPubMedGoogle Scholar
- 41.Tsunemi T, Ashe TD, Morrison BE, Soriano KR, Au J, Roque RAV, Lazarowski ER, Damian VA, Masliah E, La Spada AR (2012) PGC-1 alpha Rescues Huntington's Disease Proteotoxicity by Preventing Oxidative Stress and Promoting TFEB Function. Sci Transl Med 4(142), 142ra97. doi: 10.1126/scitranslmed.3003799 PubMedCentralCrossRefPubMedGoogle Scholar
- 55.Abdel-Azeem AZ, Abdel-Hafez AA, El-Karamany GS, Farag HH (2009) Chlorzoxazone esters of some non-steroidal anti-inflammatory (NSAI) carboxylic acids as mutual prodrugs: design, synthesis, pharmacological investigations and docking studies. Bioorg Med Chem 17(10):3665–3670CrossRefPubMedGoogle Scholar
- 56.Young D, Mayer F, Vidotto N, Schweizer T, Berth R, Abramowski D, Shimshek DR, van der Putten PH, Schmid P (2013) Mutant huntingtin gene-dose impacts on aggregate deposition, DARPP32 expression and neuroinflammation in HdhQ150 mice. PLoS One 8(9), e75108. doi: 10.1371/journal.pone.0075108 PubMedCentralCrossRefPubMedGoogle Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.