Lysosomal Re-acidification Prevents Lysosphingolipid-Induced Lysosomal Impairment and Cellular Toxicity

Neurodegenerative lysosomal storage disorders (LSDs) are severe and untreatable, and mechanisms underlying cellular dysfunction are poorly understood. We found that toxic lipids relevant to three different LSDs disrupt multiple lysosomal and other cellular functions. Unbiased drug discovery revealed several structurally distinct protective compounds, approved for other uses, that prevent lysosomal and cellular toxicities of these lipids. Toxic lipids and protective agents show unexpected convergence on control of lysosomal pH and re-acidification as a critical component of toxicity and protection. In twitcher mice (a model of Krabbe disease [KD]), a central nervous system (CNS)-penetrant protective agent rescued myelin and oligodendrocyte (OL) progenitors, improved motor behavior, and extended lifespan. Our studies reveal shared principles relevant to several LSDs, in which diverse cellular and biochemical disruptions appear to be secondary to disruption of lysosomal pH regulation by specific lipids. These studies also provide novel protective strategies that confer therapeutic benefits in a mouse model of a severe LSD.


Introduction
Lysosomal storage disorders (LSDs) represent some of the most difficult of medical challenges, with poorly understood pathologies and only rare treatment options. Despite having the common property of being caused by mutations in lysosomal enzymes, leading to accumulation of substances that would normally be degraded and to more generally compromised lysosomal function, the more than 40 different LSDs differ greatly in their primary tissue pathology, their severity, and in the specific substances that accumulate within compromised cells. The individuality of these diseases is mirrored by the dominant therapeutic strategies, which are generally focused on replacement of missing enzyme activity (by protein administration or gene expression) or on substrate reduction therapies that have the goal of decreasing availability of a precursor for the substance whose degradation is compromised by enzyme mutation . Such therapies have proven useful in rare cases [40][41][42][43], but progress on therapeutic advances is infrequent and essentially nonexistent for LSDs exhibiting damage to the central nervous system (CNS) [44][45][46]. In addition, progress has tended to be disease specific rather than providing principles that may apply more broadly.
Despite extensive study of LSDs, many critical questions remain unanswered about these diseases. For example, little is known about the biochemical linkage between any particular mutation and lysosomal dysfunction, or even whether there is a direct correlation between accumulation of particular substances and lysosomal dysfunction. In addition, although both lysosomal dysfunction and cellular dysfunctions occur in these diseases, it remains unclear how-or even if-these changes are functionally connected. Moreover, it is unclear whether principles that might be relevant to an individual disease are relevant to the pathology of diseases caused by different mutations.
To attempt to discover principles that might be relevant to LSDs caused by different mutations, we have focused on diseases associated with accumulation of lipids that are able to cause a variety of cellular dysfunctions, up to and including cell death, when applied to cells in vitro. Such diseases include Krabbe disease (KD), metachromatic leukodystrophy (MLD), and Gaucher disease [22,31,[47][48][49][50][51][52][53][54][55]. Although each of these diseases is associated with accumulation of a different lipid (or lipids) and with different disease pathologies, the effects of these lipids on cellular function are severe enough to suggest that such toxicities may contribute to disease pathogenesis.
We now show that a structurally related subset of lipids that accumulate in KD, MLD, or Gaucher disease all induce multiple lysosomal dysfunctions (along with other cellular dysfunctions), thus providing a direct link between enzymatic mutations and lysosomal abnormalities. We further show that it is possible to use drug-repurposing assays to discover single compounds that block a wide range of lipid-induced toxicities. Analysis of the properties of toxic lipids and of protective compounds reveals a previously unsuspected role of lysosomal pH and re-acidification as a potentially valuable therapeutic target. We further provide proof of principle that selecting potential therapies based on their ability to improve lysosomal function without correcting a genetic defect can reveal compounds that offer clinically relevant benefits in a mouse model of a severe LSD.
p < 0.05; Fig 1D, S1E Fig). Psy exposure also increased the activity of resident lysosomal proteases cathepsin D and B, which can cause cellular damage or death upon export to the cytoplasm and the activities of which are known to be elevated in a number of LSDs ( Fig 1E) [86-88]. Psy's ability to disrupt lysosomal function was as great as that seen with bafilomycin A (BafA), which disrupts lysosomal function by antagonizing the lysosomal vacuolar-type H + -ATPase [89].
Exposure of O-2A/OPCs to 1 μM Psy significantly increased intralysosomal pH from 4.88 ± 0.04 to 5.62 ± 0.08 after 24 h, an increase maintained for at least 48 h after a single exposure (p < 0.001; Fig 1F). This elevation in lysosomal pH was observed in both fixed ( Fig 1F) and live (S1F Fig) O-2A/OPCs. Psy exposure was as potent at increasing lysosomal pH as multiple compounds well known to exert such effects, including BafA, chloroquine, or the weak base NH 4 Cl (Fig 1F) [90]. This increase was evident within minutes of exposure to Psy and was comparable to treatment with BafA (S1G Fig, S1-S3 Movies), and the effects on lysosomal pH were sustained over 24-48 h after Psy exposure.

Unbiased Screening Identifies Chemically Diverse Candidate Protective Agents That Prevent Psy-Induced Cellular and Lysosomal Dysfunctions
To identify potential means of preventing Psy-induced toxicities that might be suitable for eventual clinical utilization, we conducted unbiased analysis of multiple concentrations of 1,040 mostly United States Food and Drug Administration (FDA)-approved small molecules Lysosomal Re-acidification and Lysosphingolipid-Induced Lysosomal Impairment and Cellular Toxicity [91] and a custom panel of 12 growth factors with known neuroprotective activity. We examined prevention of Psy-induced suppression of O-2A/OPC division in these analyses (Fig 2A).
We found 16 structurally and functionally diverse compounds (S2A Fig, S1 and S2 Tables), in addition to 4 growth factors, that had the unexpected properties of rescuing cell division (Fig 2B and 2C, S2B Fig). Eight of the 9 most protective agents were effective at rescuing cell division even when their administration was delayed 48 h after Psy exposure (Fig 2D). Importantly, all small molecules were optimally protective at physiologically achievable concentrations (i.e., nanomolar to low micromolar), and most are approved for use in humans (94%) and are blood-brain barrier permeable (>80%) (Fig 2E).
Five agents (chlorotrianisene [1G05], NKH-477 [(9C06), also known as colforsin], clofoctol [8D08], tulobuterol [9E07], and insulin-like growth factor [IGF-1]) revealed by our screens not only significantly rescued cell division but were also able to reduce cell death caused by exposure to higher concentrations of Psy (Fig 2F). With the exception of IGF-I, none of our compounds of interest were previously identified as being able to protect against toxicity of Psy (or of other lipids accumulating in LSDs). Even in the case of IGF-I, previous studies reported that supraphysiological (>10 μg/mL) concentrations decreased Psy-induced apoptosis in OLs [92] and in an O-2A/OPC cell line [82]. In our studies, by contrast, 100 ng/mL IGF-1 shifted Psy's cytotoxicity curve by an order of magnitude and significantly deceased Psydependent suppression of self-renewal (Fig 2G and 2H).
Further examination of three of the most protective agents-clofoctol, NKH-477, and IGF-I-demonstrated that these compounds also prevented Psy-induced alterations in lysosomal function. All three suppressed Psy-induced increases in lipid accumulation and cathepsin activity and restored normal endocytosis (Fig 2I-2K). Moreover, all three compounds significantly decreased Psy-dependent increases in pH ( Fig 2L). In contrast, five randomly selected molecules that did not significantly reduce any Psy toxicities in our screens (caffeine [1E04], acetarsol [5F05], mepartricin [9A06], avobenzone [9H10], and neurotrophin-3; S1 Table) had no effect on Psy-induced increases in lysosomal pH.

Protective Agents Converge on a Limited Number of Common Necessary Pathways for Their Activity
As the protective compounds we discovered are structurally and functionally diverse, we next attempted to define regulatory pathways on which these agents might converge to confer their protective activity. To do this, we focused on multiple signaling pathways previously described to be antagonized by exposure of cells to Psy (e.g., mitogen-activated protein kinase [ 101]. We next generated a secondary screen consisting of pharmacological inhibitors to components of these various signaling pathways. In these experiments, O-2A/OPCs were exposed to Psy; a combination of Psy and a protective agent; or a combination of Psy, a protective agent, and one of 15 pharmacological inhibitors targeting important signaling pathways and proteins (S3 Table). This allowed for the generation of a compound-specific "fingerprint of protection" that revealed putative signaling pathways used by the candidate compound to overcome Psy-induced suppression of division (e.g., Fig 3A and 3B).
We generated fingerprints for 12 of the most efficacious compounds in reducing Psyinduced suppression of division across 5 d using this approach. The results were then hierarchically clustered to identify similarities and dissimilarities between individual compound fingerprints and between the signaling pathways implicated in protection ( Fig 3C). Despite structural and functional diversity among candidate protective agents, there was striking similarity in the signaling pathways needed for protection.
The activity of the diverse protective agents was antagonized by pharmacological inhibition of the Ras/rapidly accelerated fibrosarcoma gene (Raf)/MAPK pathway, Akt, estrogen receptor, protein kinase A (PKA), and geranylgeranyl transferase ([GGT], which is needed for activation of small GTPases that are involved in cell division and migration). Despite their structural diversity, there was a surprisingly high degree of correlation between groups of small molecules; the cluster of structurally and functionally unrelated drugs 2G08 (ethopropazine, an antiparkinsonian drug), 2F11 (estradiol valerate, a synthetic estrogen), and 8D08 (clofoctol, an antibiotic), for example, showed the highest degree of similarity (correl. = 0.97) (S3 Fig). Protection Against Psy-Induced Lysosomal and Cellular Dysfunctions Can Be Provided by Promotion of Lysosomal Re-acidification Lysosomal ion homeostasis, maintained through the activity of several channels and transporters, is critical to the normal function of lysosomes. For example, H + import is necessary for the maintenance of an acidic pH [102] and is achieved through the activity of the V-ATPase, Ca 2+ is important for vesicle trafficking [103] and fusion [104], Na + and K + are required for the regulation of membrane potential [105,106], and Cl − serves as a counterion to regulate lysosomal membrane potential and to facilitate the acidification of the lysosome lumen [107][108][109]. Although any of these may be potential therapeutic targets, we focused our attention on identifying those channels or transporters regulated by signaling pathways uncovered through our fingerprinting analysis.
Of the several pathways that are required for activity of our protective agents, the one for which there is a clearly established linkage to at least one aspect of lysosomal function is the requirement for PKA activity. Previously, it has been shown that cAMP can promote lysosomal re-acidification [110], as can PKA, which is activated by cAMP [111]. In addition, we found that increases in cAMP not only normalized lysosomal pH but also prevented Psyinduced decreases in O-2A/OPC division (Fig 4A and 4B), raising the theoretical possibility that intervention at this point would provide additional benefits beyond that of pH restoration. We therefore focused attention on the role of lysosomal re-acidification as a potential therapeutic target.
The most attractive explanation for how cAMP/PKA activity could restore lysosomal pH would be through activation of the cystic fibrosis transmembrane conductance regulator (CFTR), a PKA-activated transmembrane chloride channel that promotes lysosomal re-acidification [112]. Unlike the CLC-7 Cl − /H + antiporter, another chloride channel that is localized to the lysosomal membrane and thought to play a role in the basal maintenance of lysosomal pH [113], the CFTR channel appears only to be critical for re-acidification. Moreover, although the CFTR can be activated by PKA and cAMP, there is no evidence for such activation of CLC-7. In addition, specific agonists and inhibitors exist for the CFTR, enabling a direct test of whether promoting re-acidification can prevent Psy-induced toxicity.
As we predicted, treatment of cells with the cAMP-independent CFTR agonist RP-107 [114] restored lysosomal pH in cells exposed to Psy. Although control of lysosomal pH and/or lysosomal re-acidification has not been thought to have any upstream role in the multiple cellular dysfunctions caused by Psy exposure, we nonetheless found that RP-107 protected against Psy-induced suppression of division, as well as elevated storage of both neutral lipids and phospholipids (Fig 4C-4E). To test the hypothesis that these benefits were not due to offtarget effects of RP-107, we also co-exposed cells to CFTR-inhibitor 172 (CFTRi-172) [115], which attenuated the protective effects of RP-107 treatment (Fig 4C-4E).
The effects of RP-107 were CFTR dependent, as determined by knockdown of CFTR in O-2A/OPCs using small interfering RNA (siRNA) pools targeting rat CFTR, as well as a pool of nontargeting (NT) siRNAs as a control for transfection; the reduction in CFTR protein levels was confirmed by western blot analysis. Knockdown of CFTR did not significantly affect lysosomal pH when compared to cells exposed to NT controls (4.96 ± 0.13 versus 4.81 ± 0.11, respectively). Moreover, in the presence of Psy, in both NT and CFTR siRNA pools, there was a significant increase in lysosomal pH (5.47 ± 0.08 versus 5.61 ± 0.09, respectively), with no significant difference between these two treatment groups. However, when we tested the effect of RP-107, a specific CFTR agonist, we found that lysosomal pH was significantly reduced in cells exposed to NT siRNA but that CFTR knockdown attenuated RP-107's protective effect (5.18 ± 0.09 versus 5.62 ± 0.03, p < 0.01; S4A Fig). Thus, as with our pharmacological experiments, genetic loss of CFTR does not appear to significantly affect basal lysosomal pH in untreated cells. However, the protective capacity of RP-107 is CFTR dependent. These results are consistent with the original studies demonstrating the role of the CFTR in control of lysosomal re-acidification [112].
Moreover, we found that the most effective protective agents identified in our studies did not themselves reduce the basal acidic pH of lysosomes in the absence of Psy (S4B Fig) but instead seemed to work to promote re-acidification. Indeed, their ability to normalize lysosomal pH, as well as rescue cell division, in cells exposed to Psy was blocked by co-exposure to inhibitors of PKA (Fig 4F and 4G). As these protective agents are able to rescue cells even when applied 48 h after Psy exposure (Fig 2D), it appears that their protective activity is not mediated simply by blocking lysosomal alkalization.

Psy's Free Amine Group Is Critical for Its Toxicity
The observations that multiple Psy-induced lysosomal and cellular dysfunctions can be prevented by lysosomal re-acidification with RP-107 (Fig 4C-4E) and that Psy exposure causes rapid increases in lysosomal pH ( Fig 1F, S1D Fig), raise complementary questions about how Psy causes such changes. One possibility is that Psy disrupts the function of particular proteins involved in lysosomal re-acidification, but another possibility is that structural features of Psy itself are directly relevant to understanding effects on lysosomal pH.
Although multiple studies have attempted to understand the molecular mechanisms underlying Psy's toxicity [61-75], we noted that Psy has unusual physicochemical features that might be of relevance to understanding its effects on lysosomes. Psy is unusual as a cationic, weakly basic lipid, carrying a net positive charge at physiological pH. With a pKa value of 7.18 [116], Psy is predicted to be 99.9% protonated in the acidic pH of the lysosome. If this aspect of Psy's structure is important in altering lysosomal and cellular function, then the protonatable free amine group on Psy should be critical in mediating the changes in lysosomal pH that . Data for all graphs displayed as mean ± SEM; *p < 0.05, **p < 0.01, † p < 0.001 versus untreated; a p < 0.05, b p < 0.01, c p < 0.001 versus Psy-only treatment. See also S1 and S2 Tables for drug names and concentrations. Data presented in this figure can be found in S1 Data.
we observed. We therefore tested whether removing this free amine group altered effects on lysosomal pH and on other outcomes of Psy exposure.
We found that the free amine group on Psy is critical in its ability not only to disrupt lysosomal pH but also to cause other toxic effects. We compared Psy toxicity to that of N-acetyl-Psy (N-AcPsy), a structural derivative containing an amide-linked acetyl group, rendering it no longer protonatable ( Fig 5A). Unlike Psy, N-AcPsy did not induce cell death or alter O-2A/ OPC self-renewal at similar concentrations ( Fig 5B and 5C). Moreover, N-AcPsy did not elevate neutral lipid and phospholipid storage, increase endocytic transport time, increase cathepsin activity, or elevate lysosomal pH (Fig 5D-5G). Thus, the positively charged free amine group present on Psy was critical to increasing lysosomal pH and also to the subsequent lysosomal and cellular impairments observed after exposure in O-2A/OPCs.

Structurally Related Lysosphingolipids from Multiple LSDs Alter Lysosomal Function
To further test the hypothesis that the presence of free amine group on a cationic lipid is critical to lipid-induced toxicities, and that such lipids provide a direct link between enzymatic mutation and lysosomal disruption, we examined a series of lipids known to accumulate in other LSDs. Other lipids of potential interest include lyso-sulfatide (lyso-SF) (which accumulates in MLD [49]), glucosylsphingosine (GlcSph) and glucosylceramide (GlcCer) (which accumulate in Gaucher disease [47]), and lactosylsphingosine (LacSph) and lactosylceramide (LacCer), which accumulate in several LSDs ( Fig 6A) [78, 117,118]. Some of these lipids appear to have been only rarely studied for their effects on cell function in vitro (lyso-SF, GluSph, LacSph, LacCer) [119]. In the case of Gaucher disease, the majority of previous in vitro studies appears to have focused on GlcCer, and studies on both GlcCer and GlcSph often have required lipid concentrations severalfold greater than those at which Psy's effects were observed (e.g., [22,50,51,53,55,[120][121][122]).
In order to eliminate differences in cell types as potential contributors to different outcomes, we examined the survival and self-renewal of O-2A/OPCs exposed to lyso-SF, GlcSph, GlcCer, LacSph, and LacCer. Use of these cells also provided a test of the hypothesis that the structure of a lipid is of primary importance in determining toxicity. We also examined the effects of N-acetyl-sulfatide (N-AcSF) as a direct comparison with N-AcPsy. We found that sphingosinederived lipids that accumulate in different LSDs and that contain a free amine group (and thus are structurally similar to Psy) caused significant cell death and suppression of self-renewal ( Fig  6B and 6C) at similarly low lipid concentrations as we observed with Psy. In contrast, exposure to their ceramide-based counterparts GlcCer and LacCer, or to N-AcSF, did not cause cellular toxicities at comparable or 10-fold higher concentrations (Fig 6B and 6C).
If the hypotheses are correct that other toxic lysosphingolipids that accumulate in LSDs work through similar mechanisms as Psy, and that such mechanisms are relevant to understanding the efficacy of our protective agents, then our protective agents also should rescue cells from the toxic effects of lipids from other LSDs. If correct, such findings would provide both the first structural predictors of toxicity and the first example of protective agents of potential relevance in different LSDs.
We found that our candidate protective agents also reduced the toxic effects of GlcSph, lyso-SF, and LacSph ( Fig 7A). Three of our most effective agents-IGF-1, clofoctol (8D08), and NKH-477 (9C06)-prevented lipid-induced suppression of division and also attenuated increases in neutral lipid and phospholipid accumulation, endocytic import time, cathepsin activities, and lysosomal pH in rat O-2A/OPCs exposed to sublethal concentrations of GlcSph, lyso-SF, or LacSph (

Lysosphingolipids Disrupt Human OL Progenitor Cell Behaviors, and Protective Compounds Rescue Human Cells
We next examined the question of whether the principles revealed in our studies on cells derived from the CNS were applicable to human cells. In these experiments, we used an anti-CD140a (PDGFRα) antibody to enrich for a population of human O-2A/OPCs from the corpus callosal field of mid-gestation fetal tissue (S6 Fig) [128] and exposed cells to Psy and potential protective agents as for rat-derived cells.
Exposure to lysosphingolipids caused death in human cells at concentrations comparable to those used in rat progenitor cells (S4 Table), as well as suppression of cell division and elevation of lysosomal pH at sublethal concentrations, whereas their non-lyso counterparts did not cause similar toxicities (Fig 8A-8C). Notably, cell division and normalization of lysosomal pH were restored in cells exposed to Psy with clofoctol, NKH-477, and IGF-I, as we observed for rat O-2A/OPCs (Fig 8D and 8E).

Alterations in O-2A/OPC Biology in Twitcher Mice Are Like Those Induced By Psy Exposure In Vitro
In the final section of our studies, we asked whether discoveries made on WT cells exposed exogenously to Psy in vitro revealed principles applicable to cells with an LSD-relevant mutation, both in respect to cellular pathologies and to rescue of lysosomal function. These studies were carried out using twitcher mice, a naturally occurring model of KD that recapitulates most human symptoms. Multiple studies have demonstrated that this mouse is a reliable model of KD in respect to enzymatic dysfunction and tissue pathology [77-80, 129,130] and is also one of the most useful models for studying LSDs in general. In particular, twitcher mice progress from a lack of apparent pathology to severe disease over a relatively rapid time course, with function appearing to be normal at birth, followed by disease symptoms manifesting about 20 d after birth and with death ensuing at about 42 d after birth. This time course allows pathology and the effects of treatment to be studied at different stages of disease progression.
In our studies on twitcher mice, we first determined that changes in O-2A/OPC function were like those induced by exposure to low doses of Psy in vitro. We found significant reductions in both myelin content and OL cell number (OLs; Olig2 + /GST + ) in the corpus callosum-the major myelinated tract of the CNS-at P40 when compared to age-matched WT littermates ( Fig 9A and 9B), consistent with previous analyses of human and twitcher tissue [76,131]. We also observed a significant reduction in the percentage of dividing (Ki67 + ) O-2A/OPCs (54.0% ± 1.9% of WT, p < 0.01; Fig 9C) at this late time point, during which time O-2A/OPCs should be undergoing rapid expansion through cell division to replace damaged OLs and myelin.
We additionally found that O-2A/OPC function was compromised in presymptomatic twitcher mice in ways similar to those induced by Psy exposure. We isolated O-2A/OPCs from presymptomatic P15 twitcher mice and examined their self-renewal capacity in vitro. These cells showed impaired self-renewal in comparison with cells of age-matched WT cells when maintained in vitro for 5 d (Fig 9D). Such findings were mirrored by significant reductions in the pool of dividing O-2A/OPCs in vivo at P15 (Fig 9E).
To determine whether cells harboring a mutant GALC gene exhibit changes in lysosomal pH, we examined the endolysosomal pH of corpus callosal O-2A/OPCs acutely isolated from presymptomatic twitcher mice (P17). We found that the lysosomal pH was significantly less acidic than that of cells isolated from age-matched WT littermates (Fig 9F), similar to what was observed in vitro with exogenous Psy treatment ( Fig 1F). Thus, O-2A/OPCs isolated at developmental time points in which symptoms are not obvious (prior to P18-20) show altered lysosomal pH and alterations in critical cellular behaviors like those induced by exposing WT cells to Psy in vitro.

NKH-477 Protects against Multiple Toxicities in Twitcher Mice
We next investigated whether the analytical approach employed in our in vitro studies could identify compounds able to provide clinically relevant benefits in vivo. We focused our studies on NKH-477 (9CO6), a water-soluble derivative of forskolin that is approved for treatment of acute heart failure in Japan [132], as this agent is known to be CNS penetrant and elevates cAMP levels (through direct activation of adenylyl cyclase) in brains of rats after systemic administration [133]. Moreover, unlike the other identified protective agents, the linkage of NKH-477 to PKA regulation (and thus to lysosomal re-acidification) is both defined and mediated through widely expressed proteins, consequentially not requiring cells to express specialized drug-targeted receptors in order to be responsive.   [15]. (N-Q) Quantification of travel speed, stance time, beam traverse time, and relative weights for P25 saline-treated WT and twi, as well as NKH-477-treated twi, mice. (R) Quantification of brain Psy levels at P35. Data for all graphs displayed as mean ± SEM; ns = not significant; *p < 0.05, † p < 0.001 versus WT; b p < 0.01 versus vehicle-treated twi. Data presented in this figure can be found in S1 Data. We initiated treatment at P10, a time when CNS concentrations of Psy are already approaching the range at which we see effects on O-2A/OPCs [78-80], using once-daily intraperitoneal (IP) injections (1 mg/kg; Fig 9G). This is a point in time when disruptions in neuronal function can already be observed in twitcher mice [134], raising the possibility of initiating treatment only after subtle clinical changes are first observable. This delayed initiation of treatment is in marked contrast with the well-studied need to initiate the application of bone marrow transplantation and/or gene therapy in the first few days after birth in order to obtain benefit [1,10,16,40,130].
The primary endpoints of interest in our in vivo studies were whether we could rescue lysosomal and cellular function in O-2A/OPCs and whether once-daily treatment with NKH-477 is sufficient to provide benefit on both parameters. O-2A/OPCs were isolated at P35 to examine the effects of NKH-477 treatment on lysosomal pH, and we found a normalization of pH in cells isolated from treated twitcher mice when compared to vehicle-treated mice (Fig 9H). NKH-477-treated twitcher mice also showed an increase in the numbers of dividing O-2A/ OPCs at P35 to near-normal levels, as well as increases in myelin content and increased OL cell numbers at moribund ages, when compared to vehicle-treated twitcher mice, again to levels not significantly different from WT mice (Fig 9I-9L).
Remarkably, we also found that NKH-477 treatment provided significant lifespan extension that was comparable to published single-therapy treatments aimed at restoring GALC activity, including bone marrow transplantation (the current standard of care in patients) or viralmediated gene therapy ( Fig 9M) [1,2,10,13,15,16]. Moreover, twitcher mice that received daily injections of NKH-477 also showed significantly improved locomotor and gait function (Fig 9N-9P, S7 Fig) and significantly improved weight gain throughout their lifespan when compared to vehicle-treated twitcher littermates (Fig 9Q). These benefits were observed despite the fact that we were not correcting the genetic defect; indeed, we did not find that NKH-477 treatment reduced the overall tissue burden of Psy in the CNS (Fig 9R).

Discussion
Our studies provide multiple novel findings related to the biology of LSDs. We found in both rodent and human cells that structurally related sphingolipids that accumulate in these disorders appear to directly cause multiple lysosomal dysfunctions. We also discovered multiple pharmacological agents, previously approved for other clinical purposes, that prevent all of the sphingolipid-induced lysosomal and cellular toxicities we analyzed, apparently by promoting lysosomal re-acidification. In vivo studies in the twitcher mouse model of KD demonstrated the ability of one of the agents we identified, which is known to be CNS penetrant, to correct lysosomal pH in O-2A/OPCs, as well as to provide multiple therapeutically relevant benefits in the absence of correcting the underlying genetic mutations implicated in the disease.
The finding that pathophysiologically relevant low levels of four different sphingolipids known to accumulate in different LSDs are each sufficient to compromise multiple lysosomal functions appears to provide the first evidence that substances created due to mutations of lysosomal enzymes may be directly responsible for initiating the metabolic dysfunctions that characterize such diseases. Previous studies have speculated that lysosomal dysfunction is caused by such events as intralysosomal accumulation of substances that are not properly degraded (e.g., [19,20,24,26,28,30,32,[135][136][137]), but we could find no prior demonstration-or suggestion-that a specific substance known to accumulate in LSDs is able to simultaneously alter lysosomal pH, endolysosomal trafficking, lipid degradation, and cathepsin activation.
Based on the comparative structures of toxic and nontoxic lipids, we hypothesize that toxicity is caused by disruption of lysosomal pH. All of the four toxic lipids we studied share the presence of a theoretically protonatable free amine group, raising the possibility that their accumulation increases the net positive charge in the lysosomal lumen, altering ion homeostasis and decreasing acidification by suppressing proton influx through the V-type H + ATPase. In contrast, such effects were not caused by other lipids that accumulate in these disorders and that lack this free amine group (i.e., GlcCer, LacCer), or by lysosphingolipids with an acetylated amine group (N-AcPsy and N-AcSF) attenuated toxicity.
The only remotely comparable studies we could find to our own were those of Sillence and colleagues [120,121], who reported that exposure of the virally transformed tumorigenic RAW murine macrophage cell line to 40 μM GlcCer for 48 h altered trafficking of borondipyrromethene (BODIPY)-labeled LacCer to the lysosome, and that an unspecified concentration of GlcCer caused modest increases in lysosomal pH in these cells. However, such studies also demonstrated that exposure to 20 μM GlcCer or GlcSph decreased lysosomal pH in RAW cells exposed to a GlcCer synthase inhibitor, that such effects were not caused by exposure to Psy, and that these GlcCer and GlcSph concentrations caused negligible cell death [120,121]. Thus, these previous results differ markedly from those obtained in our studies examining effects of 10-to 20-fold lower concentrations of Psy, GlcSph, lyso-SF, and LacSph and also do not indicate that lipids with similar structures cause similar lysosomal or cellular pathologies. In addition, although studies on sphingosine (applied at 10-μM concentrations) suggested the free amine group on this lipid is important for its toxicity, these studies considered the role of the amine group was to confer detergent-like properties on sphingosine and did not consider potential relevance to control of lysosomal pH [138,139].
The possibility that changes in lysosomal pH may be of particular importance in understanding the effects of exposure to toxic sphingolipids, and that lysosomal neutralization may be upstream of multiple lysosomal and cellular dysfunctions and may provide a novel therapeutic target, was strongly supported by our findings that we rescued O-2A/OPCs from adverse effects of lipid exposure by activation of the CFTR (which promotes lysosomal re-acidification [112]). Exposure to RP-107, a chemical activator of CFTR [114], prevented alterations in lysosomal pH and also rescued cells from adverse effects on division in a CFTR-dependent manner. In addition to CFTR, there are likely several other lysosomal targets that may be relevant for treatment. Accumulation of undegraded sphingomyelin, for example, has been shown to alter membrane trafficking and lysosomal calcium homeostasis through the impairment of the TRPML1 channel [140].
We think it is also important not to interpret our findings as indicating that activation of chloride flux via the CFTR will be the sole mechanism available for promoting lysosomal reacidification, or that regulation of chloride flux is the only possible way to promote restoration of a normally acidic pH. The CFTR provides a well-studied protein for which there is strong data indicating a role in re-acidification [112], for which useful experimental drugs are available, and for which a role of PKA in activation has been identified. But it seems likely there will be other proteins that offer potential entry points for promoting re-acidification. In respect to the much more studied problem of the control of basal lysosomal pH, there is strong disagreement on whether chloride ion flux through the CFTR or through CLC chloride channels or Cl − /H + exchangers is central to controlling basal lysosomal acidification. Expression of mutant CFTR in alveolar macrophages was reported to be associated with a lack of proper acidification of their degradative compartments [141,142]. In contrast, other authors found that the CFTR was not required for phagolysosomal acidification in macrophages [143] or respiratory epithelial cells [144], with other investigators also questioning the importance of CFTR in promoting lysosomal acidification [145]. These disagreements also extend to the CLC chloride channels or Cl − /H + exchangers, and some investigators have reported that loss of CLC-7 is not associated with alterations in lysosomal acidification [107,108,146]. Moreover, there are also intriguing observations that cation transport may also be important in the regulation of basal lysosomal pH [102,143]. Although it may be that some of these disagreements arise due to use of different techniques [113], it may also be the case that there are nuances of lysosomal regulation that differ in different cell types, and also that lysosomal reacidification may be regulated by flux of cations or anions other than chloride. The extent to which controversies regarding control of basal lysosomal pH are pertinent to studies on control of lysosomal re-acidification is not yet known, however. We hope that the results of our present studies will further increase interest in this important problem and will lead to identification of other regulatory pathways of potential therapeutic relevance.
Although some of the effects of individual toxic sphingolipids that we studied have been observed previously with other cell types (although usually at higher lipid concentrations than we utilized, e.g., [51, 61-75, 93, 95, [119][120][121][122][147][148][149][150][151][152][153][154]), there is no previous indication that all of these forms of damage may ultimately be controlled by a single metabolic parameter or that such a parameter might control lysosomal pH. It is also worth noting that, although multiple mechanisms have been observed to contribute to particular effects of Psy or of other toxic lipids [51, 61-75, 82, 120,123,[147][148][149][150], none has demonstrated the ability to correct the multiple dysfunctions prevented by promotion of lysosomal re-acidification.
Additional support for the hypothesis that control of lysosomal pH is of central importance in understanding the pathology of toxic lipid exposure was provided by the identification, by unbiased drug screening, of novel protective agents that show no known prior overlap in function but that all converged on promoting lysosomal re-acidification. We found that clofoctol, NKH-477, and IGF-1 all restored lysosomal pH in lipid-exposed cells, despite having no known common properties. Restoration of lysosomal pH appears to be due to promotion of re-acidification, as none of these protective agents acidified lysosomes in the absence of Psy. As the only known convergence of the protective substances we identified (including RP-107) is a common ability to promote lysosomal re-acidification, it currently is most likely that it is this aspect of their effects that is most important.
The possibility that regulation of lysosomal pH and re-acidification could represent a convergence point of disease pathology and therapeutic intervention for LSDs appears to be novel. Interest is emerging in the possibility that promoting lysosomal re-acidification may offer therapeutic benefit in situations of lysosomal dysfunction, but studies thus far have been focused only on the possibility that restoring normal lysosomal pH will enhance normal protein degradation [32,110,111,[188][189][190][191][192]. Nonetheless, the possibility that regulation of lysosomal pH could represent a central mechanism in disease pathogenesis and treatment is consistent with the dependence of normal lysosomal function on an acidic pH (as summarized in Fig 10). For example, neutralization can cause release of lysosomal Ca 2+ and cathepsins [32,192]: Ca 2+ release could compromise cytoskeletal function [193] and hence cell division, whereas cathepsin release and activation could initiate cell death [194]. In addition, increasing lysosomal pH would be predicted to decrease function of any lysosomal enzymes evolutionarily optimized for function in an acidic environment.
The question of whether it was possible to modify lysosomal and cellular function in an animal model of a severe LSD was studied by administering NKH-477 to twitcher mice, a severe murine model of an LSD that exhibits a pattern of disease progression similar to that seen in KD patients [195] and that is the most widely used model for studying possible disease interventions [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17]. NKH-477 was the logical choice for such studies, as it was the one small molecule protective agent identified thus far with known CNS penetrance [133] and with a welldefined drug target. In contrast with studies on gene therapy and/or bone marrow transplantation, both of which would be constant in their effects and generally must be initiated shortly after birth when animals are presymptomatic to obtain benefit in experimental models [1,2,10,13,15,16], we only administered NKH-477 once daily beginning 10 d after birth. This starting point was chosen both because this is a time when CNS concentrations of Psy first begin to approach those utilized in our in vitro studies and to test the hypothesis that our approaches could identify interventions able to provide benefit even when initiated after it might be possible to detect early changes in neuronal function [65-67].
The primary goal of our in vivo studies was to determine whether NKH-477 administration could be used to normalize lysosomal pH and improve O-2A/OPC function. After first confirming that O-2A/OPCs isolated from twitcher mice showed similar abnormalities as WT progenitor cells treated with Psy in vitro, we found that daily treatment with NKH-477 normalized lysosomal pH (as analyzed ex vivo in O-2A/OPCs isolated from treated and control twitcher mice) and rescued O-2A/OPC division in vivo.
Benefits of daily NKH-477 treatment extended far beyond rescue of O-2A/OPC division and lysosomal pH and offered several clinically relevant improvements. At the cellular level, daily NKH-477 administration rescued OL numbers and myelin content even at a time when vehicle-treated littermates were moribund. Moreover, mice treated with NKH-477 showed improved motor behavior and weight gain (suggesting that cell types other than O-2A/OPCs also benefitted from this treatment). In addition, survival was significantly extended, even to the same degree previously reported with gene therapy alone (and exceeding that obtained with bone marrow transplants alone) [1,2,10,13,15,16]. These multiple benefits were obtained even though we did not correct the underlying genetic defect nor decrease the overall Psy tissue burden, and there was no prior information on using this compound (or any related compounds) in the context of LSDs. Although NKH-477 increases cAMP levels in the CNS [133] and could thus have other effects beyond promotion of lysosomal re-acidification, the fact that NKH-477 shares the property of promoting re-acidification with the other compounds we identified, and was indeed able to rescue lysosomal pH and O-2A/OPC division in vivo, makes it seem likely that this aspect of drug action is at least partially relevant to the benefits observed.
Even if some of the in vivo benefits we observed were due to other activities of NKH-477 than promotion of lysosomal re-acidification, this would not decrease interest in this agent as a potential candidate for further analysis. Recent reviews of the outcomes of implementing newborn screening (NBS) for detection of early infantile KD (EIKD) in New York state [196] have led to the conclusion that, "in addition to the potential harm to families receiving falsepositive test results, NBS for EIKD appears to have resulted in a reduction in survival in individuals who have the disease. The data from the New York program suggest that NBS for EIKD should be abandoned, pending the development of improved screening or therapies shown to confer both survival and quality-of-life benefits over supportive care. The results of this experience suggest that research efforts should be focused on improving presymptomatic treatment outcomes in children identified by NBS prior to the redeployment of mandatory presymptomatic screening" [197].
As treatment with NKH-477 confers both survival and quality-of-life benefits in the established animal model for KD and already has been approved (in Japan) for use in humans [132], this may provide an attractive starting point for thinking about new approaches to some of the devastating LSDs with severe neuropathology. Moreover, the discovery of mechanisms and protective strategies that apply to distinct lipids accumulating in three different LSDs provides hope that these same general principles will apply to other LSDs characterized by lysosphingolipid accumulation, and perhaps also in other LSDs (such as the neuronal ceroid lipofuscinoses/Batten disease) in which lysosomal pH is abnormally more alkaline [198]. In addition, the ability of re-acidification to rescue a diverse range of lysosomal and cellular dysfunctions raises the question of whether similar strategies might provide broadly useful effects in important diseases in which lysosomal dysfunction has also been implicated, such as Alzheimer's and Parkinson's disease (e.g., [199][200][201][202][203][204][205][206][207][208][209][210][211][212]).

Ethics Statement
The University of Rochester RSRB has reviewed this study and determined that based on federal (45 CFR 46.102) and University criteria the study does not qualify as human subjects research and has waived the need for consent (RSRB#00024759). All animal procedures were performed under guidelines of the National Institutes of Health and approved by the Institutional Animal Care and Utilization Committee (IACUC) of the University of Rochester Medical Center, Rochester, NY (UCAR#2001-140).

Lipids
Lipids used in this study were purchased from Matreya and were of highest purity. All lipids were resuspended in anhydrous dimethyl sulfoxide (DMSO) to 10 mM, stored at -20˚C or -80˚C, and resuspended in media before use. Comparable results for OPC division and survival were obtained with Psy purchased from Sigma-Aldrich and Santa Cruz Biotechnology.

Clonal Analysis
Purified (A2B5 + ) O-2A/OPCs were isolated from P7 rat corpus callosa as above and plated at a density of 25 cells/cm 2 in poly-L-lysine-coated 24-well plates in DMEM:F12 complete media supplemented with 10 ng/mL PDGF-AA immediately after purification. Experiments were initiated after 24 h. Cells were fixed after 5 d in 4% paraformaldehyde, stained with antibodies against A2B5 (1:4; in-house hybridoma, ATCC) and GalC (1:4; in-house hybridoma, ATCC), and counterstained with DAPI (1 μg/mL; Invitrogen #D1306), and the size and composition of each clone was then scored. Spontaneous generation of GalC + OLs (i.e., in the absence of differentiation conditions) or Type 2 astrocytes was not detected in this paradigm, and so only the number of progenitor cells (A2B5 + GalC − ) per clone is reported.

Cell Migration
Cell migration with agarose drops was performed as previously reported [214,215]. Briefly, purified (A2B5 + ) O-2A/OPCs were isolated from P7 rat corpus callosa as above, resuspended in 0.3% low-melt agarose (at 37˚C; Sigma #A0701), and diluted in DMEM:F12 complete media supplemented with 10 ng/mL PDGF-AA at a density of 4 x 10 4 cells/μL, and 1.5 μL of the cell-agarose mixture was plated in the center of a poly-L-lysine-coated 24-well plate. The agarose was allowed to gel at 4˚C for 10 min before DMEM:F12 complete media supplemented with 10 ng/mL PDGF-AA, with and without Psy. (PDGF-AA was omitted in some wells as controls for migration, as O-2A/OPC motility is stimulated in vitro by PDGF). Half of the media, with and without Psy, was replaced daily for 3 d, after which point the cells were loaded with calcein-AM (200 nM) and imaged. The distance migrated from the agarose drop to the leading edge was quantified as reported [215].

Immunocytochemistry
Cells were fixed with 4% paraformaldehyde for 20 min before permeabilization for 10 min with 0.5% Triton X-100 (

Analysis of Cell Survival
Cells were incubated with 200 nM calcein-AM and 1 μg/mL prodium iodide for 30 min at 37˚C to determine the proportion of live and dead cells, respectively, in an experimental condition. Single-cell analysis was performed using a Celigo cytometer (Nexcelom) and the %Live corrected values are reported (calcein + PI − ).

Analysis of Cell Proliferation Rate
The total number of cells per well was determined using Brightfield analysis with a Celigo cytometer (Nexcelom) daily across 5 d, and cell numbers were normalized to the number of cells at the beginning of the experiment (Day 0). The proliferation rate was calculated as the fold change in cell number per unit time (days) using linear regression (0.95 < R 2 < 1.0 over 5 d) for each well.

Small-Molecule Screen
For analyses of cell division, purified (A2B5 + ) O-2A/OPCs were exposed to 1 μM Psy with each of the 1,040 compounds in the NINDS II Custom Collection library (Microsource) diluted to a final concentration of 0.2, 1, and 5 μM, each in duplicate, 15 compounds per plate. Each plate had control wells (in triplicate) of cells exposed to either vehicle (0.01% DMSO) or Psy (1 μM) alone. The proliferation rate was determined by daily cell counting using a Celigo cytometer (Nexcelom) across 5 d, as outlined above, and the increases in cell number within each well were internally normalized to the number of cells in that well at Day 0 (before the addition of Psy/compounds). The calculated proliferation rates were then normalized to the mean proliferation rate of in-plate vehicle-treated controls.
Selected "hits" in both screens (15 for survival and 36 for proliferation) were rescreened at 0.2, 1, and 5 μM for their ability to significantly reduce Psy-induced suppression of division across 5 d. Those compounds that significantly reduced Psy toxicities with at least one of the three selected concentrations (22 in total) were rescreened using nine-point dose-response curves, ranging from 1 nM to 10 μM, to identify optimally protective concentrations to significantly reduce Psy-induced cell death at 5 d and/or suppression of division across 5 d. The list of 22 compounds that significantly reduced Psy-induced suppression of division was shortened to 15 by elimination of those minimally protective compounds for which commercial sources were cost prohibitive or unavailable.

Growth Factor Screen
For analyses of cell division, cells were exposed to 1 μM Psy with each growth factor (including BSA, with which all growth factors were diluted) diluted to a final concentration of 10, 33, and 100 ng/mL, each in triplicate, six growth factors per plate. Each plate had control wells (in triplicate) of cells exposed to either vehicle (0.01% DMSO) or Psy (1 μM) alone. The proliferation rate was determined by daily cell counting using a Celigo cytometer (Nexcelom) across 5 d, as outlined above, and the increases in cell number within each well were internally normalized to the number of cells in that well at Day 0 (before the addition of Psy/compounds). The calculated proliferation rates were then normalized to the mean proliferation rate of in-plate vehicle-treated controls.

Fingerprint Analysis
Concentrations of small-molecule inhibitors of signaling proteins were selected based on their ability to reduce phosphorylation of target proteins when analyzed by immunoblot when possible; the selected concentrations minimally enhanced-or did not alter-Psy's effects on cell division when examined in the absence of protective agents. Purified (A2B5 + ) O-2A/OPCs were pretreated with the small-molecule inhibitors for 1 h before the addition of protective agent and Psy (each in triplicate). All plates had vehicle-and Psy-treated wells (each in triplicate), as well as Psy combined with the protective agent of interest, wells as controls. The proliferation rates were analyzed and normalized against vehicle-treated controls, as outlined above. Data were hierarchically clustered with an unweighted Euclidean distance similarity metric (complete linkage clustering) using Cluster 3.0 and visualized using TreeView.

Delayed Administration of Protective Agents
For experiments in which the administration of candidate protective agents was delayed, purified (A2B5 + ) O-2A/OPCs were exposed to the indicated concentrations of Psy for 2 d before each small molecule or IGF-1, prepared at a 10X concentration in DMEM:F12 complete media supplemented with 10 ng/mL PDGF-AA, was diluted to 1X so as to minimize dilution of Psy and perturbation of cells. An equal volume of DMEM:F12 complete media supplemented with 10 ng/mL PDGF-AA was added to untreated and Psy-only control wells. The proliferation rate was calculated from the daily change in cell number from time of administration (Day 2) for three days (Day 5), as outlined above.

Lysosomal pH Measurements
Purified (A2B5 + ) O-2A/OPCs were plated on poly-L-lysine-coated glass-bottom microwell dishes (MatTek Co., Ashland, MA; #P35G-1. [5][6][7][8][9][10][11][12][13][14] in DMEM:F12 complete media supplemented with 10 ng/mL PDGF-AA after passaging. Cells were loaded with 500 μg/mL LysoSensor Yellow/Blue Dextran (Invitrogen) in complete media with PDGF for 24 h prior to treatment. After 24 h, the cells were fixed in 4% paraformaldehyde and were imaged using a Leica TCS SP5 laser confocal microscope with a 63X oil immersion lens. Using an excitation wavelength of 335 nm (405 diode), emission spectra at 450 nm (acidic) and 521 nm (alkaline) were quantified, and the ratio of these emissions was calculated using the Leica Advanced Fluorescence software. Live-cell imaging was performed as above, with the exception that cells were not fixed in PFA prior to imaging and analysis. To generate the lysosomal pH calibration curve, the pH of pre-loaded O2A/OPCs was measured as previously described [112]. Briefly, the cells were incubated in calibration buffers (20 mM MES, 110 mM KCl, and 20 mM NaCl containing 10 μM monensin and 20 μM nigericin; Sigma) adjusted to known pH values between 4.0 and 6.0 at 0.5 increments using HCl/NaOH for 1 h prior to imaging, and ratiometric quantification, as above. Calibration curves were generated using both fixed and live cells.

Endocytosis Measurements
Purified (A2B5 + ) O-2A/OPCs were exposed to indicated conditions for 24 h. Cells were trypsinized and resuspended at a density of 10 6 /mL in conditioned (treated) medium and 1:1000 FluoSpheres polystyrene beads (Invitrogen). The cell:bead suspension was incubated at room temperature and gently inverted every 5 min. At indicated time points, 10 μL (10,000 cells) were transferred to 1 mL of ice-cold PBS and pelleted at 13,000 rpm at 4˚C for 5 min. Cell pellets were resuspended in ice-cold 2% paraformaldehyde/PBS and transferred to PLL-coated 96-well dishes to adhere during fixation. The integrated fluorescence intensity per cell was measured using a Celigo cytometer (Nexcelom) and plotted over time; the time-to-half-maximal intensity was calculated using curve-fitting software (Prism).

In Situ Cathepsin Activity Measurements
Cathepsin B activity was measured as per manufacturer's instructions (MagicRed Cathepsin B substrate, ICT). Briefly, purified (A2B5 + ) O-2A/OPCs were treated as indicated for 24 h before exposure to cell-permeant CathB substrate (1 μM); substrate cleavage occurred at 37˚C for 1 h before the integrated fluorescent intensity per cell was quantified with a Celigo cytometer (Nexcelom). For cathepsin D measurements, fluorescently labeled, cell-permeant CathD active-site inhibitor (BODIPY-FL Pepstatin A; 10 μM) was added to O-2A/OPCs that had been treated as indicated for 24 h; active CathD labeling occurred at 37˚C for 1 h before the integrated fluorescent intensity per cell was quantified with a Celigo cytometer (Nexcelom).

In Situ Lipid Accumulation
Neutral lipid and phospholipid accumulation were quantified with the HCS LipidTox Phospholipidosis and Steatosis Detection Kit (Invitrogen) as per the manufacturer's directions. Positive controls cyclosporin A (10 μM) and propranolol (10 μM) were used, respectively.

CFTR Knockdown
Rat O-2A/OPCs were exposed to either 50-nM pools of four siRNA constructs targeting rat CFTR or 50-nM pools of four control siRNA constructs that do not target the rat genome with DharmaFECT-1 transfection reagent (1:1000), as per manufacturer's directions (Dharmacon), for 24 h. Four d post transfection, cells were passaged and either lysed for western blot analysis or plated on glass-bottom dishes for analysis of lysosomal pH, as outlined above. Western blot analysis was performed as previously reported [216] using an anti-rabbit CFTR antibody (Cell Signaling) and HRP-conjugated beta-actin (Santa Cruz).

Psy Quantification
Mice were killed at P35 and transcardially perfused with ice-cold PBS. Isolated tissue was flash-frozen in liquid nitrogen and stored at -80˚C until analysis. Analysis of sphingolipids was performed by Dr. Jacek Bielawski from the Lipidomics Core at the Medical University of South Carolina (MUSC) using liquid chromatography-mass spectrometry (LC-MS/MS) and supercritical fluid chromatography-mass spectrometry (SFC-MS/MS) methodologies, as described previously [217].

Animal Treatment
Adult heterozygote (Galc twi/+ ) C57Bl/6J (B6.CE-Galctwi/J) mice were originally obtained from Dr. Ernesto Bongarzone (University of Illinois at Chicago, Chicago, IL) and used as breeder pairs to generate homozygous (twi; Galc twi/twi ) twitcher mice and WT (Galc +/+ ) C57Bl/6J mice. All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Rochester School of Medicine and Dentistry and conformed to the requirements of the Animal Welfare Act. In total, three cohorts of aged-matched mice from different litters received daily IP injections beginning at P10: WT mice receiving saline (n = 6); twi receiving saline (n = 6); twi receiving 1 mg/kg NKH-477 (Tocris) in saline (n = 8).
Mice were euthanized at P35 and tissue was isolated after transcardial perfusion with 4% paraformaldehyde. Immunohistochemical analyses were completed as outlined above. For survival analysis, animals were provided moistened chow and hydragel water packs and monitored daily for weight gain. Animals were euthanized when moribund, as assessed by when the animals could no longer ambulate to maintain food and water intake or exhibited clinical signs of pain such as hunched posture and ruffled fur, as determined on a daily basis. Animals were euthanized using CO 2 exposure and cervical dislocation.

Motor Behavior Testing
Aged-matched mice from different litters received daily IP injections beginning at P10: WT mice receiving saline (n = 4); twi receiving saline (n = 3); twi receiving 1 mg/kg NKH-477 (Tocris) in saline (n = 4) were analyzed for locomotive ability and gait using the Phenoscan suite and Runwayscan software (CleverSys, Inc) at P25. Multiple parameters, including stance, stride, swing, brake, and propulsion time (milliseconds), stride length (millimeters), and average speed (millimeters/s) were collected for each animal over three compliant trials and averaged for both front and rear paws.

Statistical Analyses
Bar graphs are plotted as mean ± SEM and represent at minimum three independent biological replicates performed in triplicate, except where noted. Two-group comparisons were analyzed using a Student's t test, and multiple-group comparisons were analyzed using an ANOVA with Bonferroni post-hoc test. Prism (v5.0; GraphPad) was used for data analysis and presentation. ng/mL PDGF + 10 ng/mL bFGF ("PDGF+FGF"); 100 pg/mL PDGF ("-PDGF"); 1 ng/mL PDGF + 40 ng/mL T3/T4 ("+T3/T4"); and 1 ng/mL PDGF + 10 ng/mL BMP4 ("+BMP4") for 5 d in mass culture. A2B5 + : glial progenitor cells; GalC + : OLs; GFAP: astrocytes; A2B5 + /GFAP + : Type-2 astrocytes; Ki67 + : mitotically active cells; Olig2 + : oligodendroglial-lineage cells. (B) Quantification of (A). (C) Quantification of human fetal O-2A/OPCs maintained as in (A) for 5 d, except at clonal density. The mean number of cells immunopositive for the indicated stain per clone, as well as the percentage of clones containing at least one immunopositive cell, are reported. Note that NeuN + or Tuj1 + neurons were never detected in mass or clonal culture. Data for all graphs displayed as mean ± SD for one GW20 human sample. All experiments were repeated in four human GW19-21 samples with comparable results. Data presented in this figure can be found in S1 Data. (TIFF) S7 Fig. NKH-477 treatment reduces twitcher mouse gait abnormalities. Quantification of gait for P25 vehicle-treated WT (n = 3-4), vehicle-treated twitcher mice (n = 3), and NKHtreated twitcher mice (n = 4), including measurements of stance, break, propel, swing, and stride time, as well stride length, for front and rear paws. Data for all graphs displayed as mean ± SEM; Ã p < 0.05, ÃÃ p < 0.01, † p < 0.001 versus WT; a p < 0.05, c p < 0.001 versus vehicle-treated twitcher. Data presented in this figure can be found in S1 Data. (TIFF) S1 Movie. Time-lapse movie of rat O-2A/OPCs loaded with a ratiometric lysosomal pH dye and exposed to vehicle (0.01% DMSO) for 5 min.