Identification of Drug Modulators Targeting Gene-Dosage Disease CMT1A

Abstract

The structural integrity of myelin formed by Schwann cells in the peripheral nervous system (PNS) is required for proper nerve conduction and is dependent on adequate expression of myelin genes including peripheral myelin protein 22 (PMP22). Consequently, excess PMP22 resulting from its genetic duplication and overexpression has been directly associated with the peripheral neuropathy called Charcot-Marie-Tooth disease type 1A (CMT1A), the most prevalent type of CMT. Here, in an attempt to identify transcriptional inhibitors with therapeutic value towards CMT1A, we developed a cross-validating pair of orthogonal reporter - firefly luciferase (FLuc) and β-lactamase (βLac) - assays capable of recapitulating PMP22 expression, utilizing the intronic regulatory element of the human PMP22 gene. Each compound from a collection of approximately 3,000 approved drugs was tested at multiple titration points to achieve a pharmacological endpoint in a 1536-well plate quantitative high-throughput screen (qHTS) format. In conjunction with an independent counter-screen for cytotoxicity, the design of our orthogonal screen platform effectively contributed to selection and prioritization of active compounds, among which three drugs (fenretinide, olvanil, and bortezomib) exhibited marked reduction of endogenous Pmp22 mRNA and protein. Overall, the findings of this study provide a strategic approach to assay development for gene-dosage diseases such as CMT1A.

The formation of myelin sheath around axons by Schwann cells, a process known as myelination, is a critical component of peripheral nervous system (PNS) postnatal development in vertebrates. The lipid-rich myelin sheath supports axonal stability and enables rapid, saltatory propagation of action potentials (1). Disruption of myelin structure by genetic factors has been recognized as a main cause of human peripheral neuropathies such as Charcot-Marie-Tooth (CMT) disease. Over the past two decades, molecular genetics research has identified over 30 genes whose alterations are associated with the CMT phenotypes including peripheral myelin protein 22 (PMP22) (2).

The PMP22 gene encodes a small tetraspan 22-kDa membrane glycoprotein which plays an essential role in myelin synthesis and assembly, representing 2–5% of PNS myelin proteins (1). The elevated gene-dosage derived from the duplication of the PMP22 gene has been shown to be culpable for pathological hallmarks of CMT type 1A (CMT1A), a predominant type of CMT (3). Typical symptoms of CMT1A include distal muscle atrophy, sensory loss, hyporeflexia, and skeletal deformity. On the other hand, the haploinsufficiency of PMP22 by its reciprocal deletion is associated with a distinct neurological disorder called hereditary neuropathy with liability to pressure palsies (HNPP) (4). Despite the concomitant genetic occurrence, HNPP is less prevalent than CMT1A due to its mild, transient symptoms which render the disease unnoticed or misdiagnosed. The fact that a moderate change in expression by its altered copy number leads to clinical phenotypes indicates that PMP22 levels must be tightly controlled to ensure proper myelination by Schwann cells. Previous transgenic studies have revealed regulatory elements which direct spatial and temporal expression of PMP22 (5, 6). Whereas an upstream region (−10 to −6.5 kb) is sufficient to activate reporter expression at only later stages of PNS development, and thus named the late myelination Schwann cell-specific element (LMSE), a recently discovered intronic element at +11 kb was found to mediate at least part of the induction of PMP22 during myelination.

Studies in rodent models of CMT1A have identified two transcription-based strategies that ameliorate the disease by reducing levels of Pmp22. The first is high dose ascorbic acid which has formed the basis of the only current clinical trial of CMT1A (7). The effects of ascorbic acid may be diverse, but one aspect of its function appears to be reduction in cAMP signaling (8). Unfortunately, the early results of the human clinical trial have so far not revealed a significant effect in CMT1A patients, although the final results of these trials are yet to emerge (9). In addition, proof-of-principle studies using progesterone antagonists to reduce Pmp22 expression in a rat model of CMT1A have shown beneficial effects (10). These studies demonstrate that a relatively subtle (<2-fold) change in PMP22 transcription could effectively treat the most common form of inherited peripheral neuropathy.

There is great potential for pharmacological intervention to downregulate PMP22 levels for a treatment of CMT1A especially since clinical tolerance for excess reduction of PMP22 in CMT1A patients is reflected in mild symptoms of HNPP relative to those of CMT1A. However, one of the major limitations in therapeutic research for CMT1A has been the lack of viable assay systems that can broadly and effectively screen small molecules for their biological relevance to CMT1A. Here, we describe the development of transcription-based cellular assays amenable to quantitative high-throughput screen (qHTS) to identify PMP22-modulating chemical agents, representing the first application of HTS to the translational therapeutic efforts aimed at CMT1A.

Results and Discussion

Development of an Orthogonal Pair of Transcription-Based Assays

A recent study has indicated that tissue-specific expression of PMP22 is governed in part by the regulatory element residing in the largest intron of the PMP22 gene which is activated by transcription factors Egr2 and Sox10, the master regulators of PNS myelination (6). In an attempt to identify chemical agents capable of downregulating PMP22, we utilized the intronic element to develop transcription-based reporter assays to be used as a cell-sensor for PMP22 expression (Figure 1A). A pair of primary cross-validating assays was established with the intronic element conjugated to drive either firefly luciferase (FLuc) or β-lactamase (βLac) expression which was stably transfected into S16 rat Schwann cells previously shown to sustain high levels of Pmp22 expression comparable to those in myelinating Schwann cells (11). Establishment of an assay platform in human cellular context was hindered by unavailability of human-derived Schwann cell lines expressing physiological levels of PMP22. The use of the orthogonal assays allowed discrimination against non-specific activity intrinsic to either FLuc or βLac reporters themselves. The bioluminescent FLuc signal is complemented by the dual βLac output composed of fluorescence at 460nm for βLac expression and FRET at 535nm for cell viability (12). In addition, given that cell-based inhibition assays are inherently obscured by non-specific toxicity, we decided to independently perform the CellTiter-Glo assay for high stringency of selection against cytotoxic compounds (13).

Figure 1. Assay design and validation.

(A) A schematic diagram of reporter constructs derived from the intronic element of the human PMP22 locus. Dimeric Sox10 binding site and Egr2 binding sites are represented by ovals and boxes, respectively. (B) siRNA-induced depletion of Sox10 leads to downregulation of both the Fluc and βLac reporters without affecting cell viability. The reporter (FLuc or βLac) expressing S16 cells were treated with Sox10-siRNA or NT (non-targeting)-siRNA as a control for 48hr. Whereas the FLuc cells were subjected to luciferase assay (i) for FLuc expression or CellTiter-Glo assay (ii) for viability, the βLac cells were processed for βLac expression (iii) and viability (iv) measured at 460nm and 535nm, respectively. Error bars indicate the SD of replicates (n=144 for the FLuc cells, n=192 for the βLac cells).

To validate the biological relevance of the orthogonal reporter assays, we employed siRNA-induced depletion of the transcription factor Sox10 involved in Pmp22 activation in S16 cells (6). Specific downregulation of the reporters, compared to the control siRNA and also warranted by unaffected cell viability in parallel, indicates that the orthogonal assays properly serve to recapitulate Pmp22 expression (Figure 1B).

Assay Miniaturization for qHTS

qHTS provides concentration response profiles of a chemical library directly from an initial primary screen and thus has advantages over traditional schemes that test each library member at a single-concentration. Therefore, we optimized our assays for qHTS in a 1536-well plate format employing 8 titration points (Supplementary Table 1–3) to reduce false positives and enhance the detection of compounds with lower intrinsic or apparent potency (14). For the screen against the NPC library of approximately 3,000 approved drugs (15), the orthogonal FLuc and βLac assays were conducted with reporter enzyme inhibitors, i.e. PTC124 and potassium clavulanate, respectively, as positive controls (16, 17). Direct reporter inhibition was intended to biochemically model the relevant range of biological modulation of Pmp22 expression examined in the assays. The CellTiter-Glo viability assay was normalized to a media only (no cells) control. The assay performances were adequately sensitive and reproducible as indicated by statistical measures for qHTS (Supplementary Table 4). We note that Z’ values in some of these assays were lower than 0.5, a typical threshold for traditional HTS (18). However, Z’ values are insufficient to fully reflect the reliability of assay designs that incorporate an additional pharmacological variable as a test of output as opposed to single-point variability. On the other hand, MSR values, the minimum potency ratio between two measurements having statistical significance, are more relevant to titration-based screening (qHTS) (19) and indicated acceptable variance to ensure reliable estimates of concentration response relationships obtained from our cell-based qHTS.

Selection and Prioritization of Actives

The activity of compounds was plotted over the concentration-range represented by the 8 titration points, generating a single concentration-response curve (CRC) per compound in each assay (Figure 2A). The curves were categorized into three conceptual types (inhibitory, stimulatory, and inactive) and further classified based on efficacy (response magnitude), the number of asymptotes, and quality of fit of the curve to the data (r²) (14) (Supplementary Figure 1).

Figure 2. qHTS profiles and active compound selection.

3D plots from qHTS in the FLuc assay (A), the βLac assay (fluorescence at 460nm), and CellTiter-Glo assay. CRCs are colored according to curve classification defined in Supplementary Figure 1; light blue (class 1), green (class 2), dark blue (class 3), gray (class 4), and magenta (stimulatory). (B) Potency distribution of compounds identified in the FLuc assay. Gray bars indicate the total number of actives identified, blue: concordant actives in the FLuc and βLac assays, red: concordant actives with no significant cytotoxicity. Bottom panel: potency distribution of compounds identified in the βLac assay. Gray bars indicate the total number of actives identified, blue: concordant actives in the FLuc and βLac assays, red: concordant actives with no significant cytotoxicity. (C) Activity plots of the 9 active compounds in FLuc (open circle), βLac (solid circle), and CellTiter-Glo (solid triangle) assays. (D) Potency correlation of the 9 actives (solid circles) from the qHTS assays was indicated by the experimental R² (a solid line) versus the ideal R² of 1 (a dotted line). Open circles represent compounds that are active in both FLuc and βLac (460nm) assays but cytotoxic in the CellTiter-Glo assay. The data from these screens have been deposited into PubChem Summary AID # xxxx.

To identify compounds mediating inhibition of Pmp22, we first selected the high quality inhibitory curve classes (CCs) representing ~10 % and ~5% of activity from the FLuc and βLac (fluorescence at 460nm) assays, respectively (Table 1 and Figure 2B). 175 compounds of inhibitory response greater than 70% were prioritized from the initial 291 actives in the FLuc assay and subsequently cross-referenced with 119 actives from the βLac assay refined by the internal viability FRET (535nm) which contributed to elimination of inhibitory, thus presumably cytotoxic, compounds. This pairwise juxtaposition of the orthogonal data effectively yielded 34 compounds that exhibited concordant activity in both FLuc and βLac (460nm) assays. The utility of the CellTiter-Glo assay, introduced as an additional viability screen, was substantiated by the invalidation of additional toxic compounds that evaded the built-in counter-FRET measurement of the βLac assay. Nine compounds were identified to inhibit expression of both FLuc and βLac derived from the PMP22 intronic element, apparently devoid of non-specific activity attributed to cytotoxicity and reporter interference (Figure 2C and Supplementary Figure 2). The activity profiles represented by potencies (IC₅₀) of the 9 compounds were tightly correlated between the two orthogonal assays as indicated by the Pearson coefficient of R²=0.93 (Figure 2D and Supplementary Table 5). The less sensitive nature of the βLac assay relative to the FLuc assay was also revealed by the displacement of the linear regression by ~8-fold relative to a theoretical fit to assays of equal sensitivity (Figure 2D, dotted line).

Table 1.

Orthogonal Cross-validation Selection Strategy

Stage	Assay	Compounds			Selection Criteria
		Tested	Pass criteria
qHTS	S16 FLuc	2,816	291	10.3%	CC = 1a, 1b, and 2a
			175	6.2%	Inhibitory efficacy > 70%
	S16 βLac	2,816	144	5.1%	CC = 1a, 1b, 2a, and 2b at 460 nm
			119	4.2%	CC ≠ inhibitory CC at 535 nm
	Orthog. cross-validation		34	1.2%	Intersection of FLuc and βLac
	S16 viability	2,816	9	0.3%	CC ≠ inhibitory (counter screen for acute cytotoxicity
Secondary	RT-qPCR analysis	9	4	44%	> 60% ↓ Pmp22 mRNA expression
	Western analysis	4	3†	75%	Downregulation of Pmp22 protein; include additional chemotype/class members‡

^*CC = curve class

^†Hycanthone not available for the analysis

^‡Additional proteasome inhibitors added to further explore drug class

Noticeably, in contrast to the βLac assay, there was a preponderance of stimulatory compounds in the FLuc assay (compare left and middle panel in Figure 2A), and 44% (28 out of 63) of them were identified as FLuc enzyme inhibitors in the biochemical FLuc enzyme assay (Supplementary Figure 3). Although seemingly paradoxical, this can be explained by the phenomenon related to the ability of a compound that binds to the FLuc reporter enzyme and stabilizes its cellular half-life (20). Convincingly, those 28 compounds were also characterized chemically by structures related to known FLuc inhibitors (Supplementary Figure 4) (21).

Conversely, we observed 20 compounds that induced a decrease in the FLuc read-out from the S16 cells but also appeared to be inhibitory in the biochemical FLuc enzyme assay (Supplementary Figure 5). Examination of these compounds revealed that many were complexes with bivalent metal ions (e.g., Hg²⁺) known to decrease quantum yields and shift spectral output from yellow-green to red of FLuc (22), while the remainder were highly colored and therefore capable of attenuating bioluminescence through an inner filter effect (23). It is noteworthy that none of our 9 actives were found in this group of 20 artifact compounds, thus verifying the effectiveness of the orthogonal screen platform in identifying compounds of specific activity.

Inhibition of Endogenous Pmp22 by the Actives Identified

To determine if the 9 actives identified from the reporter-dependent qHTS exhibit physiological effects on Pmp22 expression, quantitative reverse transcriptase (RT)-PCR (qPCR) was utilized to measure the abundance of Pmp22 mRNA transcript in the S16 cells treated with each compound at 10uM, the concentration point that elicits significant response based on CRCs of the 9 actives profiled in the qHTS. Unexpectedly, podophyllotoxin, one of the 9 actives, led to noticeable cytotoxicity upon visual microscopic inspection and therefore was excluded in subsequent analyses. The specific downregulation of Pmp22 was markedly induced by four drugs (hycanthone, fenretinide, olvanil, and bortezomib), compared to the constitutively active Gapdh (glyceraldehyde 3-phosphate dehydrogenase) gene whose expression remained largely unaltered (within the range of 1.3 PCR cycles) across the samples (Figure 3A and Supplementary Figure 6). Similar results were reproduced in primary rat Schwann cells (Figure 3B). Hycanthone, a schistosomicide, was unavailable in a quantity required for subsequent studies. We also introduced FLuc-specific primers to examine how the reporter expression is reflective to that of Pmp22 in the FLuc-expressing S16 cells treated with each compound (Figure 3A). The FLuc expression was consistently repressed, albeit to a lesser degree, which may be explained by different mRNA stability and/or the FLuc reporter regulated solely by the intronic element as opposed to Pmp22 situated in its entire endogenous locus.

Figure 3. Transcription of endogenous Pmp22 is negatively modulated by actives from qHTS.

The FLuc-expressing S16 cells were treated with either DMSO as a control or each compound indicated at 10µM for 24hr and then subjected to quantitative reverse transcriptase-PCR (RT-qPCR) with primers specific to endogenous Pmp22 or FLuc (A). Similar results were observed with Pmp22 and βLac in the βLac-expressing S16 cells (data not shown). Primary rat Schwann cells were cultured and treated with each drug as described in the Supporting Information before subjected to RT-qPCR for Pmp22 expression analysis (B). Data were first normalized to beta-actin (ActB) used as an endogenous loading control across samples and then plotted relative to the untreated sample set to 1 for each target gene. Error bars indicate the SD of three replicates.

There are two variants of the Pmp22 mRNA transcript (Pmp22-1a and -1b) derived from alternative promoters P1 and P2, respectively (24). They differ in the first exon but give rise to the identical Pmp22 protein due to the translation initiation site nestled in the second exon shared between the two variants. To dissect the inhibitory regulation of Pmp22 imposed by the three biologically active drugs, variant-specific primer sets were designed for qPCR analysis in the S16 cells treated with an increasing concentration of each drug. All three of the drugs incurred discernibly preferential inhibition towards Pmp22-1a which is the predominant form of Pmp22 mRNA in myelinating Schwann cells of PNS whereas Pmp22-1b is ubiquitously expressed in other tissues (Figure 4A) (24).

Figure 4. Expression analysis of Pmp22 mRNA transcript variants and protein.

The S16 cells were treated with increasing concentrations of each drug for 24hr and then subjected to either RT-qPCR with primers specific to Pmp22-1a and -b. (A) or Western blot (B). For mRNA analysis, data were first normalized to ActB used as a loading control across samples and then plotted relative to the untreated sample set to 1 for each target. Error bars indicate the SD of three replicates. For protein analysis, fold change was calculated as a ratio of Pmp22 to β-actin (a loading control) normalized to that of the untreated sample set to 1. Higher concentrations of bortezomib were also tested along with other proteasome inhibitors (Supplementary Figure 7). Vehicles used were ethanol for fenretinide and DMSO for olvanil and bortezomib.

A surplus of PMP22 protein derived from its gene duplication provides an etiological basis for the phenotypes of CMT1A (25). Therefore, we further investigated whether the three drugs identified as transcriptional inhibitors for Pmp22 are also capable of modulating the abundance of Pmp22 protein in parallel. The S16 cells treated with a varying concentration of each drug were subjected to Western analysis which revealed downregulation of Pmp22 protein, albeit less pronounced, consistent with the preceding decrease in Pmp22 mRNA (Figure 4B) and therefore validated the transcription-based assay design for such a dose-sensitive, disease-associated protein as Pmp22. In addition, given that a high turnover of Pmp22 is subject to ubiquitin-dependent degradation (26), the counterintuitive decrease of Pmp22 by bortezomib, a proteasome inhibitor, points out an interesting phenomenon of disproportional inhibition of Pmp22 transcription by bortezomib relative to its direct inhibition on the proteasome, which was also observed with other inhibitors in the class (Supplementary Figure 7).

Regulators for Myelination Are Modulated by the Biological Actives

Pmp22 has been known to be dynamically induced along with other myelin genes during postnatal PNS myelination. Therefore, we decided to evaluate the pharmacological specificity of the three drugs towards Pmp22 by qPCR analysis against three major myelin-associated genes: Mpz (P₀-Myelin protein zero), Mag (Myelin-associated glycoprotein), and Cx32 (Connexin 32). Overall, all of the three drugs exhibited transcriptional repression across the myelin genes to some extent at the concentration tested (Figure 5A). In particular, Pmp22 appeared to be most severely inhibited relative to the other myelin genes especially by olvanil, suggesting a titration approach for its enhanced specificity to Pmp22. In an attempt to elucidate how the three drugs exert their inhibitory effects on Pmp22, we also sought to examine known regulators involved in PNS myelination. The qPCR analysis unveiled noticeable downregulation of Egr2 and Sox10 as well as Erbb2 and Erbb3 (27–29), which are required for induction of myelination by axonally expressed neuregulin (Figure 5B). Interestingly, Erbb3 has been identified as a target of Sox10 activity (30). On the other hand, c-jun, a negative regulator for myelination, was significantly upregulated, although Sox2 was unaffected (data not shown) (31, 32). With its broad involvement in cellular processes, this upregulation of c-jun could lead to an adverse effect especially in the central nervous system (CNS) where it has been associated with seizures, pain, memory, and stroke (33). Overall, it is suggested that a reversal of the PNS myelination program is part of the regulatory mechanisms exploited by the three drugs. Additionally, co-titration of combinatorial pairs from the three drugs showed no significant interaction of activity, which suggests they are independently engaged in Pmp22 inhibition (Supplementary Figure 8).

Figure 5. Dynamic changes in expression of myelination regulators.

The S16 cells were treated with each drug at 10uM for 24hr and then applied to RT-qPCR targeting major myelin genes (A) and their regulators (B). Data were first normalized to ActB used as a loading control across samples and then plotted relative to the untreated sample set to 1 for each target gene. Lines are drawn to compare a fold-decrease of Pmp22 with that of the other myelin genes by each drug (A). Error bars indicate the SD of three replicates.

Although previous studies have discovered the 5′-flanking region of Pmp22 involved in Schwann cell-specific expression of Pmp22 (5, 34) as well as post-transcriptional regulation of Pmp22 (35, 36), it still remains elusive how those regulatory pathways specifically converge on PMP22. Due to this limited understanding of druggable targets for PMP22 expression, the phenotypic cellular qHTS assays were devised, functionally hinging upon the PMP22 intronic enhancer which has been well-characterized to interact with crucial myelin regulators (6). In addition, the strategic coupling of the phenotypic assays with the NPC library of annotated drugs was intended to provide not only the prospect of drug re-purposing for a treatment of CMT1A but also critical insight into the spectrum of PMP22 regulatory circuitry.

The design of the primary orthogonal assays supplemented by the viability screen effectively contributed to the optimal selection and prioritization of actives that were shielded from compound-dependent assay interference (20). Subsequently, pharmacological activity for endogenous Pmp22 was successfully confirmed in 4 out of the 9 actives identified from the primary qHTS. The cytotoxicity of podophyllotoxin and the confirmation rate of ~44 % (4 out of 9 actives), however, underscore the importance of carefully-designed follow-up assays that ascertain biological plausibility of primary qHTS. The inter-assay (qHTS and follow-up assays) discrepancy observed with the five non-specific actives cannot be adequately addressed by random integration of the reporters into local chromatin environments that are independent of Pmp22 regulation, since FLuc reporter expression consistently reflected the expression of Pmp22 in the mRNA analysis by qPCR. Rather, while remotely possible, post-translational degradation of the reporters may be facilitated by the drugs, a process which would only be discernible in the primary qHTS assays, not in the follow-up mRNA-based assay. Overall, this discrepancy highlights the challenges in interpreting outcomes of phenotypic assays in primary screening.

One of the Pmp22-modulating drugs, bortezomib, is an anti-cancer 20S proteasome inhibitor widely used for several neoplasm types including multiple myeloma but accompanied by a dose-limiting adverse effect known as bortezomib-induced peripheral neuropathy (BIPN). Consistent with our findings, the underlying cause of BIPN has been recently illuminated by a perturbation of gene expression in Schwann cells including a decrease in Pmp22 (37), which raises the intriguing possibility of using bortezomib as a Paracelsian small-molecule therapy for treatment of CMT1A. Fenretinide is a synthetic analog of retinoic acid (RA), and RA has also been implicated in peripheral nerve development, as it dysregulates PNS myelination via its differential interaction with retinoic acid receptor (RAR) and retinoid X receptor (RXR) (38). However, due to the complexity of the RA signaling pathway, exemplified by the receptor-independent chemopreventive effect of fenretinide (39), it remains unclear how fenretinide regulates Pmp22 expression. Olvanil mainly acts as an agonist for the transient receptor potential vanilloid type 1 (TRPV1), a cation channel with high permeability to Ca²⁺, that mediates thermal nociception and inflammatory pain. Counterintuitively, stimulation of TRPV1 can lead to desensitization of the channel, which is a pharmacological basis for olvanil as an analgesic (40). The previous report that an abnormal influx of Ca²⁺ facilitated by the purinergic receptor P2X7 underpins the pathogenesis of CMT1A suggests that the activity of Ca²⁺ channels such as TRPV1 modulated by olvanil may be associated with PMP22-sensitive neuropathy (41). It is revealing that target-centric re-assessment of the qHTS led to a group of drugs associated with the TRP superfamily among which capsaicin and camphor displayed inhibitory activity to some extent as well (Supplementary Figure 9). Their disqualification was largely attributed to the stringency of the cross-validating system of the orthogonal assays, which can be technically susceptible to false negatives when one of the assays is of lower sensitivity (e.g. βLac assay) and compound potency is low or solubility limited. It is also interesting that enhanced activity of TRPV4 caused by genetic mutations has been causatively linked with another type of CMT (CMT type 2C), suggesting a common physiological denominator for CMT pathology (42). It is notable that our screen did not identify ascorbic acid or progesterone antagonists. However, rodent studies suggest that the ascorbic acid must be used at very high levels to achieve an effect in vivo. In addition, it is possible that other regulatory elements in the Pmp22 gene may mediate the effect of progesterone.

Conclusion

Genomic copy number variation (CNV) provides a basis for genetic predisposition to a range of human diseases (43). In particular, pathogenetic gene duplications involve dosage-sensitive genes such as PMP22 for CMT1A, SNCA for Parkinson’s disease, and APP for Alzheimer’s disease (44). Oftentimes, the scarcity of knowledge of molecular targets involved in expression of affected genes hampers drug discovery efforts for associated diseases. The successful use of the phenotypic transcription-based assays described here lays the foundation for qHTS as an enabling resource not only to identify active compounds but also to help unravel a nexus of regulatory pathways so that a rational therapeutic approach can be further developed to pave a new avenue for treatments of genetic disorders. Moreover, translatability of qHTS to human therapeutics can be subsequently assessed by relevant models such as human (patient)-derived Schwann cells including iPSCs (induced pluripotent stem cells) for CMT1A as we deepen our understanding for sophistication of Schwann cell biology.

Methods

Reporter Constructs and Stable Cell Lines

The PMP22 intronic element (the coordinates 15,091,959-15,092,201 from human chromosome 17 - hg18Mar. 2006 build in UCSC Genome Browser) was conjugated upstream of either firefly luciferase (pGL3 luciferase, Promega) or β-lactamase along with the minimal E1B TATA promoter. An IRES Neomycin cassette from the pEFIRES-N plasmid was inserted downstream of the reporter genes. Stable cell line development with the reporter constructs and cell culture conditions are described in the Supporting Information.

Approved Drug Collection

The NCGC pharmaceutical collection (NPC) contained 2816 pharmacologically active drugs at the time of screening (15). Details on this collection and preparation for testing are described in the Supporting Information.

Reporter Gene and CellTiter-Glo Assays

Assay protocols for qHTS (firefly luciferase, beta-lactamase, and CellTiter-Glo assays) are outlined in the Supporting Information.

Data Analysis for qHTS

Data from each assay were normalized plate-wise to corresponding intra-plate controls as described previously (14), see the Supporting Information for additional details.

siRNA Transfection

1.5µg of either Sox10-siRNA (CAGCGATACCTTAATAAAGTA, Qiagen) or a non-targeting (NT) siRNA control (Cat# 1027280, Qiagen) was transiently transfected into 10⁶ FLuc- or βLac-expressing S16 cells using T-020 program of the Amaxa system (Lonza) in the rat neuron nucleofection solution (Cat# VPG-1003) according to a manufacturer’s protocol. The cells were dispensed in 1536 well plates (500 cells per well) using a BioRAPTR FRD and incubated for 48h before subjected to assays.

Quantitative Reverse Transcription-PCR Analysis

Quantitative PCR was performed in either SYBR green-based reactions or Taqman-based customized 384-well micro fluidic arrays using a ViiA7 system (Applied biosystems). More details including primer sequences are available in the Supporting Information.

Western Blot Bnalysis

Cells were plated in 6 well plates and treated with each compound at concentrations indicated. After 24h, lysates were harvested and analyzed by immunoblotting for PMP22 (1:1000, Cat# ab61220, Abcam) and beta-actin (1:1000, Cat#ab8226, Abcam) as described previously (6). The membranes were probed with DyLight 800 and 680 conjugated secondary antibodies (1:10,000, Cat #35571 and 35518, Pierce) and imaged using a LiCor Odyssey scanner.