Induced pluripotent stem cell technology for disease modeling and drug screening with emphasis on lysosomal storage diseases

The recent derivation of disease-specific induced pluripotent stem cells (iPSCs) from somatic cells of patients with familial and sporadic forms of diseases and the demonstration of their ability to give rise to disease-relevant cell types provide an excellent opportunity to gain further insights into the mechanisms responsible for the pathophysiology of these diseases and develop novel therapeutic drugs. Here, we review the recent advances in iPSC technology for modeling of various lysosomal storage diseases (LSDs) and discuss possible strategies through which LSD-iPSCs can be exploited to identify novel drugs and improve future clinical treatment of LSDs.


Introduction
By using a combination of transcription factors that had previously yielded success in reprogramming mouse somatic cells [1], Yamanaka and colleagues demonstrated that human somatic cells can be directly reprogrammed to a pluripotent state similar to that of their embryonic stem cell (ESC) counterparts in vitro [2]. Th e human induced pluripotent stem cells (iPSCs) created by the reprogramming process exhibit the typical characteristics of human ESCs, showing morphology and growth requirements and surface and pluripotent-related marker expression similar to those of their inner cell massderived counterparts [2][3][4]. Furthermore, both in vitro diff erentiation and in vivo teratoma formation analyses indicate that human iPSCs are truly PSCs as they are able to give rise to cell types representing all three embryonic germ layers [2,5].
Lysosomal storage diseases (LSDs) are individually rare but collectively common; their estimated total prevalence is 1 out of 8,000 live births [6]. As current therapies have limited eff ect and most LSDs progress relentlessly, therapies that are more effi cient are urgently needed. To develop effi cient therapies, a more thorough understanding of the pathophysiological development of LSDs at the cellular level is essential. Several LSD disease models have been established in knockout mice for disease modeling and drug tests [7][8][9][10][11][12]. However, at present, these models are not able to mimic the whole spectrum of LSD conditions. Th erefore, lack of appropriate human cells aff ected by LSDs for drug screening and toxicity testing may be a major obstacle in the development of new therapies for LSDs. Since pluripotency reprogramming technology off ers an easy and effi cient means to generate patient-specifi c iPSCs, the iPSCs derived from patients with familiar or sporatic disease off er a valuable methodology through which to study the mechanisms involved in the initiation and progression of LSDs in vitro and further screen remedies for clinical treatment.
Th is review focuses on the current status of the application of iPSCs in LSD research by summarizing the LSD-iPSC cellular/ultrastructural fi ndings that have been reported in the literature and by presenting useful strategies for drug designing/screening using LSD-iPSCs. We hope this review will encourage additional translational research in LSD drug development using this novel stem cell technology.
organelle dysfunctions relating to the accumulation of specifi c substrates (Table 1) [14]. LSDs also encompass two other types of rare disorders caused by either transport defects through the lysosomal membrane [14,15] or defective vesicular traffi cking (Table 1) [14,16,17]. Th e majority of LSDs are clinically progressive and currently have no defi nite cures. Recent development of enzyme replacement therapy (ERT) has successfully alleviated the symptoms of patients with certain types of LSDs, such as Gaucher's disease, Pompe disease, Fabry disease, and type I, II, and VI mucopolysaccharidosis (MPS) [18], but the benefi t of ERT on the neurological manifestations of LSDs is less obvious [19]. Hematopoietic stem cell transplantation [20], substrate reduction therapy [21], and pharmaceutical chaperones [22] have also been developed to treat selected patients with LSDs and showed benefi cial eff ects [23]. Nevertheless, for patients with advanced-stage or late-onset LSDs, the current treatment results are still not satisfactory.

History and progress of induced pluripotent stem cell technology
In 2006, Yamanaka and colleagues [1,2] demonstrated that forced expression of only four transcription factors (Oct3/4, Sox2, Klf4, and c-Myc) is suffi cient to convert both mouse and human fi broblasts into ESC-like cells. Other researchers reported similar results at almost the same time [24][25][26]. Th e cells generated were named iPSCs. Such a breakthrough circumvented the thorny ethical issues surrounding research that employs human embryos and also off ered the hope of providing replaceable human cells with less immune rejection for use in clinical applications. A major disadvantage of iPSC technology is its low effi ciency. However, an increasing number of modifi ed protocols employing chemicals and small molecules have been forwarded to improve the effi ciency of iPSC derivation; these methods have been reviewed extensively elsewhere [27][28][29]. In addition, alter native strategies have been developed to avoid the problem of integration of viral genes, including piggyBac vectors [30], recombinant proteins [31], modifi ed mRNAs [32], microRNAs [33], and Sendai virus [34]. In general, however, it is accepted that the most effi cient method to generate human iPSCs is still through lentiviral and retroviral transduction [35]. While iPSC research in vitro has progressed considerably, the large-scale application of iPSCs in clinical practice in the near future still hangs in the balance, mainly because of the concern of tumorigenicity that is comparable to that of ESCs. In addition, a recent report [36] demonstrated that the teratomas from inbred C57BL/6 fi broblast-derived iPSCs still cannot prevent the occurrence of immune rejection when transplanted back to the syngeneic mice, although Okita and colleagues [37] argued against this conclusion and reasoned that teratoma formation might not be a good approach through which to evaluate immune rejection. Consistent with this viewpoint, transplantations of mutation-corrected iPSC-derived cells into aff ected mice have been shown to result in the rescue of disease phenotypes in mice with sickle cell anemia [38]. Nevertheless, it should be borne in mind that de novo immuno genicity might be produced during iPSC derivation and maintenance, which could be caused by, for example, the viral antigens generated by viral vectors or the animal antigens contained in the serum or supplements used for cell culture. Despite all of these uncertainties regarding clinical application, it is well recognized that human iPSCs are an unprecedented and powerful tool that is highly promising for modeling numerous human genetic diseases in vitro.

Induced pluripotent stem cells for disease modeling and drug testing
So far, more than 40 iPSC disease models have been successfully generated from patients with genetic diseases [5], and the length of the list keeps increasing. Notable examples of models developed so far are Duchenne and Becker muscular dystrophy [7], Huntington disease [7], Shwachman-Diamond syndrome [7], Lesch-Nyhan syndrome [7], amyotrophic lateral sclerosis [39], spinal muscular atrophy [40], familial dysautonomia [41], dyskera tosis congenital [42], Friedreich's ataxia [43], fragile X syndrome [44], LEOPARD (lentigines, electrocardiographic conduction abnormalities, ocular hypertelorism, pulmonary stenosis, abnormal genitalia, retarded growth, deafness) syndrome [45], long-QT syndrome [46,47], Rett syndrome [48], and Hutchinson-Gilford progeria [49], although not all of these iPSCs exhibit disease-specifi c phenotypes. Th e derivatives of some of these diseasespecifi c iPSCs have been employed as in vitro disease models to test the phenotype-correcting eff ects of small numbers of promising drugs, such as neurons diff erentiated from spinal muscular atrophy-specifi c iPSCs [40] and Rett syndrome-specifi c iPSCs [48] and cardiomyocytes diff erentiated from iPSCs for long-QT syndrome [47], suggesting the probability of their use as platforms for performing high-throughput screenings of large chemical libraries to identify novel drug candidates for these diseases.

Lysosomal storage disease-specifi c iPSCs and their cellular pathology
Gaucher disease: the initial proof-of-principle lysosomal storage disease-iPSCs Since there is no cure for most LSDs in current medicine, LSD-specifi c iPSCs may provide a unique opportunity for dissecting unexplored disease pathogeneses and identifying new drugs. Several disease-specifi c iPSCs have been successfully generated from either mouse models for LSDs or patients with LSDs (Table 2). Th e iPSCs derived from a patient with Gaucher disease type III accounted for perhaps the fi rst reported human LSD-iPSC line established [7], although few Gaucher diseasespecifi c phenotypes have been described in detail.

Mouse lysosomal storage disease-specifi c iPSCs
Using tail-tip fi broblasts from mouse models of Fabry disease [8], Krabbe disease [8], MPS VII [8], and Pompe disease [9], Eto and colleagues have generated corresponding disease-specifi c iPSCs and characterized them. In addition to reporting defi cient enzyme activities and substrate accumulation in these cells, the authors reported impaired embryonic body formation in MPS VII-specifi c iPSCs [8], a novel phenotype that, as the authors suggested, is possibly attributed to an elevated level of hyaluronic acid and may not be easily identifi ed without using disease-specifi c iPSCs. According to the authors, two in vivo phenotypes are possibly related to impaired embryoid body formation in MPS VII iPSCs: (1) the lower-than-expected 25% of MPS VII (GUSB −/− ) mice born from heterozygous mating based on Mendelian inheritance and (2) hydrops fetalis, which is known to be relatively common in patients with MPS VII [8]. Moreover, the same group has shown that mouse Pompe disease iPSCs can be diff erentiated into skeletal muscles that are positively stained for myosin heavy chain, accumulate glycogen in lysosomes, and have typical ultrastructural features, including Z-, I-, A-, and H-bands [9]. Such an achievement is remarkable because skeletal muscles are among the most diffi cult mammalian cell types to obtain from in vitro diff erentiation of pluripotent stem cells [50].

Human Pompe disease
Using a unique acid alpha-glucosidase (GAA) rescuebased strategy and fi broblasts from two patients with Pompe disease, we also successfully generated four Pompe disease-specifi c iPSC lines [10]. All of these lines exhibit Pompe disease-specifi c phenotypes, such as very low GAA activity and high glycogen content, and can be diff erentiated into cardiomyocytes that have disarrayed myofi brils and abundant glycogen-containing vacuoles. Unexpectedly, we also found that Pompe disease iPSCs have defective cellular respiration (whereas ESCs and normal iPSCs do not) and this is supported by our fi nding that the mitochondria of cardiomyocytes derived from Pompe disease iPSCs exhibit abnormal morphology. We have tested several drugs/chemicals by using Pompe disease iPSC-derived cardiomyocytes and found that glycogen accumulation can be reduced by recombinant GAA and that the deteriorated mitochondrial functions can be partially rescued by L-carnitine. In addition, using comparative transcriptome analysis, we have identifi ed six marker genes whose expression robustly correlates with the therapeutic eff ect of recombinant GAA or Lcarnitine [10]. We are currently using these Pompe disease iPSCs to test the therapeutic eff ects of other compounds (for example, autophagy inhibitors) because excessive autophagic build-ups have been documented as an important feature of cells of patients with Pompe disease [51,52].

Human mucopolysaccharidosis 1H
Presuming that iPSC-derived hematopoietic cells may have fewer immunological complications than regular donor cells in hematopoietic cell transplantation (HCT), Tolar and colleagues [11] generated iPSCs from a patient with MPS type IH, a severe form of alpha-L-iduronidase defi ciency that can be treated by HCT but not by ERT, and successfully corrected the gene defect by using iPSCs transduced with lentivirus harbor ing the wild-type gene (called IDUA) encoding this enzyme. Th e authors reported that alpha-L-iduronidase activity is not required for stem cell renewal and that MPS IH-iPSCs already have lysosomal storage of GAG. Moreover, MPS IH-iPSCs can be diff erentiated into hemato poietic progeny with a colony-forming capacity comparable to those of IDUA-corrected and wild-type iPSC-derived hematopoietic progeny. Th e authors empha sized the advantages of iPSCs for possible application in HCT; for example, iPSCs are more feasible for long-term culture, and gene correction of iPSCs is easier than that of hematopoietic stem cells, which are more sensitive to ex vivo mani pulation. Clarifi cation of whether such IDUA-corrected iPSCs can be used for therapeutic purposes without immunological complications awaits further studies.

Human mucopolysaccharidosis IIIB
A prominent advantage of iPSC technology is that it off ers access to patient neuronal cells, because neuronal cells are not diffi cult to derive from iPSCs [40]. Th is is critical for studying diseases -such as MPS type IIIB, a fatal LSD caused by the defi ciency of α-N-acetylglucosaminidase -that involve primarily the central nervous system. Lemonnier and colleagues [12] successfully generated MPS IIIB-iPSCs and demonstrated that severe cellular pathology, including storage vesicles and disorganized Golgi complex, exists in undiff erentiated iPSCs and diff erentiated neurons but not in neuronal progenitors. It is worth noting that the authors used exogenous recombinant enzyme to complement the enzymatic defect in order to clone MPS IIIB-iPSCs effi ciently, an approach similar to our rescue strategy mentioned above [10]. Considering quantitative reverse transcription-polymerase chain reaction and Western blot results, the authors suggested that accumulation of heparin sulfate modifi es the extracellular matrix constituents and related signaling pathways, which cause disorganization of Golgi architecture. Mild phenotypes in fl oating neuronal progenitors were explained by their non-adherent nature and less dependence on extracellular matrix-bound signals.

Pharmacological chaperones
Th e introduction above reveals that researchers around the world have generated proof-of-principle patientspecifi c iPSCs for several LSDs. Most of the published results of LSD-iPSC research have not advanced to the stage of drug design or medium-scale drug testing, not to mention attempts at high-throughput screening of chemical libraries for novel drug targeting of LSDs using iPSCs. However, compared with other disease iPSCs, LSD-iPSCs are more suitable for the purpose of drug design and high-throughput chemical screening because they are caused by defects of lysosomal enzymes, whose activities are measurable in vitro on a large-scale basis and the accumulated lysosomal substrates in LSDs can be assayed by either biochemical or immunocytochemical methods. Moreover, novel drugs can be designed or identifi ed to enhance the activity, stability, or traffi cking of mutant enzymes by assisting their folding or to target the pathways that synthesize the accumulated substrates.
One class of small molecules termed pharmacological chaperones, which are reversible and competitive inhibitors of their target enzymes, may be suitable drug candidates.
In the past decade, many pharmacological chaperones have been developed to target the aff ected enzymes in various LSDs; this topic has been extensively reviewed elsewhere [22,53]. Th e chaperones can be taken orally, cross the blood-brain barrier, and have biodistributions that are better than those of ERT. Some of these chaperones are already prescribed formally in clinical practice [53]. Th e common strategies for identi fying candidate pharmacological chaperones include both looking for molecules that have structural homology with the target natural substrates [54] and direct highthrough put screenings of compound libraries [55]. Th e methodology comprises initial in vitro assays such as enzyme inhibition assays in diff erent pH environments [22,56], physical stability assays [22,57] using recombinant wild-type enzymes, and subsequent cell-based assays to estimate the eff ect of chaperones on enzyme activity and enzyme traffi cking [22]. Chaperones identifi ed to work for wild-type enzymes are not necessarily helpful for mutant enzymes and need be tested in diff erent patients' fi broblasts or cell lines to evaluate the therapeutic eff ect [22]. For this purpose, LSD-iPSCs can off er a limitless source of human diseased cells containing various mutations for the second-round drug screening that is based on various cell-based assays, especially for those biochemical or cellular phenotypes seen in distinct diff erentiated cells (for example, neurons) that can be obtained only through directed diff erentiation of iPSCs. It is also theoretically possible that LSD-specifi c iPSCs can be used directly in fi rst-round or even large-scale cell-based screening using similar strategies because a number of these cell-based assays have been successfully developed to meet the requirements of high-throughput screening formats [22], including the high-content imaging platforms [58] to evaluate enzyme traffi cking.

Proteostasis regulators and other compounds
In addition to pharmacological chaperones, other ways to improve the protein folding of mutant enzymes are available. Two alternative methods have been developed [59][60][61]. First, it has been demonstrated [59,60] that two common L-type calcium channel blockers [59], either diltiazem or verapamil, and ryanodine receptor blockers, such as lacidipine [60], can partially restore the activity of two glucocerebrosidase mutants in fi broblasts derived from patients with Gaucher disease; the authors suggest ed that these drugs exert their eff ects by upregulating a subset of molecular chaperones, such as BiP and Hsp40, which in turn ameliorate the capacity of the endoplasmic reticulum to rescue misfolded mutant enzymes. Second, Mu and colleagues [61] showed that two proteostasis regulators, celastrol and MG-132, can increase the concen trations and functions of mutant enzymes associated with two LSDs: Gaucher disease and Tay-Sachs disease. Moreover, the authors demonstrated that the combined use of pharmacological chaperones and such proteostasis regulators can generate a synergistic rescue eff ect on mutant enzymes in cells derived from patients with either LSD [61]. A more comprehensive review on the multiple aspects of protein folding or degradation that are related to pharmacological intervention can be found elsewhere [62]. Obviously, future eff orts should be aimed at identifying more compounds of these two classes by using LSD-iPSCs and proving that the chemicals identifi ed by using these strategies can be successfully applied next in animal studies and fi nally in clinical trials.
On the other hand, novel adjunct therapies also deserve to be developed to better preserve various cellular functions after correcting the disease-specifi c cellular pathology in diff erent organelles (for example, the autophagic build-up and mitochondrial dysfunction in Pompe disease [52] and Golgi abnormalities in MPS type IIIB [12]). In this respect, candidate drugs may include some well-known autophagy inhibitors used in clinical trials [63], drugs or nutrient supplements [64] for treating mito chondrial dysfunction, and chemicals known to reverse endoplasmic reticulum-to-Golgi traffi cking defects [65]. High-throughput screening of chemical libraries for this purpose is also a possibility because similar approaches in other fi elds using ESCs/iPSCs have been reported [66,67] and because proper screening-based formats of some amenable biochemical and cellular assays for such organelle dysfunctions have been reported [65,68,69] and may be exploited in the iPSC system as well. However, this is still a challenging task because successful purifi cation of diff erentiated cells is a prerequisite and iPSCs may need to be passaged as single cells that can survive. A Rho-associated kinase inhibitor [70] or Accutase (Millipore Corporation, Billerica, MA, USA) [71] may be helpful to achieve this goal.

Caveats in drug screening for lysosomal storage diseases based on iPSCs
Several caveats are associated with using patient-specifi c iPSCs for drug screening [72,73]. First, mutational hetero geneity exists in most LSDs, but establishment of patient-specifi c iPSCs is time-consuming and laborintensive. Th us, it may not be practical for a single laboratory to generate patient-specifi c iPSCs for all of the mutations found in a single disease, and determining whether the drugs identifi ed to work for certain mutants are also helpful for other mutants would be a problem. Second, as seen in other classic cell-based platforms for drug screening, there is no guarantee that drug candidates identifi ed from in vitro assays can be used successfully in vivo. Th ird, to test drugs that target the phenotypes existing only in diff erentiated cells (for example, electrophysiological anomalies in cardiomyocytes), a highly homogenous cell population diff erentiated from iPSCs may be needed to obtain consistent readouts [22]. Th erefore, extensive collaborations among diff erent laboratories on the basis of consensus and standardized protocols for generating iPSC lines and comparing drug eff ects will be essential. Moreover, other assays and systems -such as using LSD animal models to determine the pharmacokinetics, pharmacodynamics, and toxicities of candidate drugs -should always be used to complement iPSC-based drug screening. Finally, a more effi cient and economic, and less time-consuming, set of protocols for obtaining homogenous diff erentiated cell types of interest in LSDs should be established in advance.

Conclusions
iPSC technology off ers a revolutionary method for model ing LSDs and other diseases and the hope of future cell-based therapy. Since most LSDs are characterized by defects in enzymes whose activities are readily measurable in vitro, disease-specifi c iPSCs off er an ideal in vitro cellular system for designing LSD-specifi c pharmacological chaperones and for possible high-throughput compound screening. However, the application of iPSC technology for drug discovery is still at an early stage, and several major challenges -such as the diffi culty of generating highly enriched disease-relevant desired cell types in large quantities from diseased iPSCs, the infl uence of culture and reprogramming artifacts on cell behavior, and the inability to recapitulate disease features in the diseased iPSC derivatives -must be resolved before it can be rendered an effi cient and robust system for developing drugs targeting LSDs. With the advances in generating transgenic human pluripotent stem cells [74], the diffi culty of enriching desired cell types from diff erentiating diseased iPSCs is likely to be resolved soon as specifi c cell types can be purifi ed from cell mixtures by lineage-specifi c genetic markers or directly diff erentiated by ectopically expressed lineage determinant(s) in diseased iPSCs or both. In addition to iPSCs, induced somatic cells, which directly convert from fi broblasts by defi ned transcription factors [75,76], can potentially provide another source of disease-relevant cell types for the purpose of disease modeling. Th e advantage of the direct cell fate conversion technology is that a desired cell type can be directly generated from a patient's fi broblasts by using a fast and simple protocol without the need of further cell purifi cation. However, it has been suggested that the conventional pathological phenotypes of certain forms of diseases depend on cell interaction and may require a longer time to arise in a disease iPSC model [77,78]. Nevertheless, the disease process might be initiated much earlier than the emergence of clinical symptoms. Th erefore, iPSC disease modeling can potentially provide an opportunity for earlier identifi cation of phenotypic changes in diseases of interest. Despite these concerns, evidence supporting iPSC disease modeling of genetic diseases as a valuable in vitro cellular system through which to understand the mechanisms underlining the patho logies of diseases and future drug discovery has rapidly accumulated in the past few years.