X-ALD is a complex peroxisomal disorder with variable expressivity. Although its primary genetic basis has been known for some time [1–4], the exact nature of X-ALD pathogenesis and its genetic and environmental modifiers have not been elucidated. Here, we generated iPSC resources for the longer-term purpose of developing novel tissue culture models for elucidating the pathogenesis of X-ALD and screening for more effective drug therapies. In keeping with prior reports , skin fibroblasts from ABCD1 mutation carriers can be reprogrammed to form iPSCs with the hallmark molecular properties of pluripotency, including the expression of appropriate gene and protein biomarkers and changes in DNA methylation levels, as summarized in Additional file 1. Patient iPSCs could be differentiated into embryoid bodies and differentiated in vitro into representative cell types of all three germ layers. Most importantly, patient iPSCs formed teratomas with evidence of cell types from all three germ layers.
Consistent with prior reports [64–66], we identified de novo CNCs over 10-kb in length in approximately half of our iPSCs. These CNCs were only found in healthy donor iPSCs and not those generated from patient cells (Additional file 4). The observed genomic deletions always provided evidence of affecting only one allele and genomic amplifications always involved a limited increase in copy number (three to four copies maximum). Due to the fact that we conducted global expression and DNA methylation analyses on these samples, we could investigate the effects that these CNCs have on the expression of genes located within affected genomic segments. In almost all circumstances, their expression levels were within the range of diploid samples. Although multiple factors likely contribute to these observations, we favor the explanation that this primarily reflects the effects of selection whereby CNCs are only tolerated in iPSCs if they involve genomic regions that do not influence the initiation of reprogramming or maintenance of pluripotency.
As a result of our genomic characterization of these cell resources, we acquired global gene expression data from patient and control fibroblasts. Many DEGs were previously reported to be associated with the site of biopsy . This is reasonable given that the patient and control fibroblasts were acquired from different institutions even though all biopsies involved the upper limbs of donors. We sought to determine if there was enrichment for functional categories or biological processes in the DEGs, keeping in mind the limitations of using cultured cells to study complex diseases involving interactions between multiple organ systems. Only very broad functional categories or KEGG pathways were highlighted in these analyses, with none of them showing a direct relation to disease.
Since there are likely to be gaps in public databases of processes relevant to peroxisome biology and X-ALD pathogenesis, we conducted a manual inspection of gene annotations provided by the DAVID bioinformatics resource and found multiple DEGs involved in immune related processes, but only two (CBLB and RAB27A) of these genes were not associated with the site of biopsy. CBLB plays a critical role in antigen-induced immune tolerance and Cblb-deficient mice immunized with myelin basic protein are more susceptible to experimental autoimmune encephalomyelitis (EAE), a model for multiple sclerosis [67, 68]. RAB27A mutations can lead to an uncontrolled T lymphocyte and macrophage activation syndrome in humans, with some individuals showing possible leukocyte brain infiltration . In one Saudi Arabian kindred, RAB27 mutations were associated with immunodeficiency and progressive demyelination of brain white matter .
The DEGs found in patient and control iPSCs did not overlap with those found in fibroblasts and instead were consistent with several leading hypotheses regarding X-ALD pathogenesis. This suggests that the reprogramming process can minimize the confounding influence the site of skin biopsy has on the gene expression profiles of cultured fibroblasts. In particular, we highlight the reduced expression of PEX11B, a major controller of peroxisome proliferation and neuroinflammatory genes, in patient relative to control iPSCs. Pex11B null mice show several pathologic features, including neuronal migration defects, enhanced neuronal apoptosis, developmental delay, hypotonia and neonatal lethality . Despite these severe phenotypes, Pex11B null mice displays only mild defects in peroxisomal fatty acid beta-oxidation and ether lipid biosynthesis . Intriguingly, the deletion of a single Pex11B allele leads to a slightly increased number of peroxisomes, increased levels of oxidative stress in brain tissue, and neuronal cell death in mice . In addition, the ULK1, whose yeast homolog plays a critical role in the autophagy-mediated peroxisome turnover , showed higher expression in CCALD patient relative to control iPSCs. It is tempting to speculate that if nervous and immune system cells of ABCD1 mutation carriers indeed have aberrant PEX11B and ULK1 activity, this could lead to differences in peroxisome abundance and/or activity that increase reactive oxygen species (ROS) levels and promote X-ALD pathogenesis. It remains to be determined if ABCD1 mutation carriers have abnormal peroxisome abundance in their pertinent nervous and immune system cells and tissues.
In a similar vein, the increased NAAA, THBS1, BSG (aka CD147 aka EMMPRIN) and NOTCH1 gene expression in patients relative to control iPSCs is supportive of hypotheses regarding a predisposition to neuroinflammation that is a prelude to devastating autoimmune responses. NAAA hydrolyzes palmitoylethanolamide (PEA), a naturally occurring lipid amide that, when administered as a drug, inhibits inflammatory responses . In principle, increasing leukocyte NAAA levels could reduce PEA levels and promote inflammation. In fact, a chemical inhibitor of NAAA function attenuates inflammation and tissue damage and improves recovery of motor function in mice with spinal cord trauma . Intriguingly, CD200 has been proposed to play a role in the immune privileged status of the CNS when CD200-mediated immune suppression occurs via neuron-microglial as well as glial-glial interactions in inflammatory conditions . THBS1 is linked to neuroinflammatory processes involving astrocyte and microglia through its role in processing and activating the TGF-β ligand  and is also implicated in responses to oxidative stress . Likewise, Notch1 is involved in microglial associated inflammation . Also of relevance are emerging reports that BSG acts a master regulator of matrix metalloproteinases (MMP) implicated in most diseases involving neuroinflammation and thus has been proposed to play a role in the immune-privileged status of the CNS .
Although we highlight the possible implications of the gene expression profiles observed in patient iPSCs, we note alternative hypotheses regarding their origins and biological significance. While the iPSCs described in this study have the hallmark properties of pluripotency, their gene expression profiles could reflect subtle ABCD1 mutation status-dependent differences in their predisposition to differentiate into specific cell types and lineages. Comparisons of the gene expression profiles of mature cell types derived from patient- and healthy donor-derived iPSCs will be especially informative. The persistence or elimination of groups of DEGs reflective of biological processes and pathways could provide a means of assessing the tissue specificity of disease and enhance the ability to discern biologically informative gene expression signatures from noise resulting from confounding variables, such as tissue culture conditions.
Although ABCD1 mutation carriers show elevated sVLCFA levels in their blood and urine and reduced sVLCFA catabolic activity in their cultured fibroblasts, the role of sVLCFA in disease pathogenesis is still under discussion. The significance of decreased plasmalogen levels in the patients' brain white matter also is unclear [44, 78]. As expected, CCALD patient fibroblasts had elevated VLCFA levels, but similar PE plasmalogen levels, relative to those from healthy donors (Figure 3A, C). Likewise, iPSCs from CCALD patient and healthy control donors also showed similar PE plasmalogen levels (Figure 3D). The fact that all patient and control iPSCs tested had low VLCFA levels, based on C26:0-lysophosphorylcholine measurements, (Figure 3B) is puzzling, yet consistent with prior reports . VLCFA levels are determined by their rate of synthesis (that is, catalyzed by fatty acid elongating enzyme ELOVL1 ), degradation (affected by ABCD1 and ABCD2 function ) and uptake of these fatty acids from the culture medium. As such, one hypothesis is that the rate of VLCFA synthesis is lower in iPSCs relative to fibroblasts under the culture conditions evaluated. Consistent with this hypothesis and prior reports , the expression of the ELOVL1 gene was significantly lower (2.6-fold, FDR < 2.2 × 10-9) in iPSCs relative to fibroblasts. Nevertheless, we note that ELOVL1 was not differentially expressed in patient relative to control fibroblasts or iPSCs. An alternate hypothesis that the ABCD2 gene is compensating for the impaired ABCD1 function in patient iPSCs; however, ABCD2 was not differentially expressed in patient relative to control fibroblasts or iPSCs. This does not preclude the possibilities that ABCD2 activity is being increased on the protein level or that another gene is playing a major role in significantly lowering VLCFA levels in CCALD iPSCs. We also note a prior hypothesis that the rapid growth rate of iPSCs could reduce their VLCFA levels, independent of their ABCD1 mutation status . Fibroblasts have altered morphology and slowed growth in iPSC media relative to fibroblast media, which according to the growth rate hypothesis could contribute to their reduced VLCFA levels. Given that iPSCs can rapidly differentiate in fibroblast media, iPSC growth media provides an imperfect, but necessary, compromise in direct comparisons between cultured fibroblasts and iPSCs. We note the potential contribution of MEF feeder cells to iPSC lipid profiles and the advantages of using feeder-free media in future experiments.
Future applications and directions
The impending implementation of newborn screening for X-ALD based on blood lipid profiles will increase the demand for model systems to screen for more effective therapeutic interventions [36, 81]. Early detection would provide physicians with a window of opportunity to treat presymptomatic patients prior to the development of CCALD, and may also prevent or delay AMN onset. Therapeutic interventions, such as Lorenzo's Oil, help prevent the onset of cerebral disease in some individuals, but are not effective for the majority of CCALD patients and, likewise, there are no effective options for AMN [75–77]. A compelling attribute of iPSC model systems is that they represent the exact ABCD1 mutations found in the patient population and thus provide an opportunity to test therapeutic agents tailored to a patient's genotype in cell populations most affected by disease. Examples of genotype-dependent therapeutic strategies include nonsense suppressor drugs  and molecular chaperones  for individuals with nonsense and missense mutations, respectively.
The fact that CCALD iPSCs show gene expression profiles similar to those derived from healthy controls may reflect the fact that X-ALD clinical symptoms do not manifest at birth but, instead, occur in early childhood or later in life. Given that ABCD1 mutant mice show clinical aspects of X-ALD with increasing age , it is possible that later passage CCALD iPSCs and their derivatives may manifest gene expression profiles and/or functional properties more consistent with disease pathogenesis and progression. In this regard, a comparison of the properties of iPSCs and their derivatives previously obtained from other CCALD and AMN patients  as a function of in vitro passage number could be informative. Despite the promise of iPSC approaches, it will remain a significant challenge to generate and optimize in vitro model systems for X-ALD and other complex disorders that involve multiple organ systems as well as unknown gene-environment interactions and genetic modifiers.