Altered WNT signaling in hiPSC NPCs derived from four schizophrenia patients

Dear Editor

Schizophrenia (SZ) is a devastating psychiatric disorder hypothesized to be a neurodevelopmental condition (1, 2); arising as a consequence of dysregulation of brain development (3, 4). WNT signaling is important for neural patterning, proliferation and migration, and synapse formation (reviewed by (5)); moreover, converging post-mortem (6, 7), rodent (8, 9) and pharmacological (10) evidence suggests that WNT signaling may contribute to SZ (reviewed by (11, 12)). We utilized human induced pluripotent stem cell (hiPSC) derived forebrain patterned neural progenitor cells (NPCs) (13, 14) to investigate canonical WNT activity in a pilot cohort of four SZ patients.

Because all research described herein was performed on deidentified human samples obtained for broadly consented scientific research by either American Type Culture Collection (ATCC) or the Coriell Cell Repository, is was found to be exempt by the Internal Review Committee of the Icahn School of Medicine at Mount Sinai. This work was also reviewed by the Embryonic Stem Cell Research Oversight Committee at the Icahn School of Medicine at Mount Sinai.

We compared global transcription of forebrain hiPSC NPCs from six control and four SZ patients by RNAseq (Table 1; GSE63738), cultured as described (13, 14). As previously reported, hiPSC forebrain NPCs differentiate to a mixed neuronal population of glutamatergic and GABAergic neurons; there was no difference in the ability of control or SZ hiPSC NPCs to generate βIII-TUBULIN-positive neurons (14) and neither transcriptional nor immunohistochemical characterization revealed any diagnosis-dependent differences in the regional patterning of forebrain NPCs (13). Multi-dimensional scaling (MDS) resolved most SZ and control hiPSC NPC samples (Fig. 1A); 848 genes were significantly differentially expressed (FDR<0.05) (SI Table 1), as illustrated by a heat map (Fig. 1C) and a volcano plot (Fig. 1D). The differentially expressed genes in SZ hiPSC NPCs were significantly 3.6-fold enriched when compared to WNT target genes (p < 10e-20) predicted by standard Classification and Regression Tree (CART) methods (16). The differentially expressed genes (FDR < 0.05) were submitted to DAVID (http://david.abcc.ncifcrf.gov), which identified several significantly enriched pathways (Fig. 1B; Tables 2–3), including the WNT signaling pathway: 17.3-fold enrichment (p < 10e-13; FDR < 10e-11). The perturbed WNT genes are marked by red stars in the WNT signaling pathway diagram (Fig. 1E; Table 2). 6/6 differentially expressed WNT genes identified by RNAseq were confirmed when tested by qPCR (Fig. 1F; Table 2).

Table 1.

Description of known clinical information for the four Coriell SZ patients.

Coriell ID	Sex	Ethnicity	Age	Age of Onset	Phenotype	Hospitalizations?	Family History
GM02038	M	Causasian	22y	6 years	suicide	?	unknown
GM01792	M	Causasian Jewish/Scandanavian	26y	unknown	episodes of agitation, delusions of persecution, and fear of assassination; at age four mild features of pervasive developmental disorder	?	father and sister affected; brother autistic at age four
GM01835	F	Causasian Jewish	27y	unknown	drug abuse; schizo-affective disorder	Yes	father and brother affected
GM02497	M	Causasian Jewish	23y	15 years	paralogical thinking, affective shielding, splitting of affect from content, and suspiciousness	Yes	affected father, anorexic/schizoid sister

Fig. 1. RNAseq comparisons of control and SZ hiPSC NPCs.

A. Multidimensional scaling (MDS) of RNAseq gene expression of hiPSC NPCs from each of six control and four SZ patients segregates samples along the two leading fold change dimensions. Gene expression analysis was performed on passage-matched hiPSC forebrain NPCs cultured on matrigel. Cells were lysed in RNA BEE (Tel-test, Inc). RNA was chloroform extracted and treated with RQ1 RNAse-free DNAse (Promega). RNAseq samples were prepared using the Illumina HiSeq 2500 RNA kit for 100nt/single end reads, four samples were run per lane. Raw cDNA reads were aligned to the hg19 reference with the spliced gap aligner Spliced Transcripts Alignment to a Reference (STAR) software, with count-based quantitation carried out via the Subread package featureCounts (http://bioconductor.org/packages/release/bioc/html/Rsubread.html) at both the geneic and exonic levels for UCSC and ensemble annotation builds. B. Pathway enrichment analysis based on DAVID. X-axis represents fold enrichment; Y-axis denotes pathways. The FDR are labeled on the right of the bar plot. C. Heat map of control and SZ hiPSC NPCs of 848 unique genes (FDR<0.05). The count data were normalized and modeled as over-dispersed Poisson data using a negative binomial model in the Bioconductor package edgeR (15). Fold changes, p-values and false discovery rates (FDRs) are obtained from the same package for integrative analysis. D. Volcano plots of −log10 p-value versus log2 fold-change mRNA levels for control and SZ hiPSC NPCs. Key canonical WNT signaling genes, including Lymphoid Enhancer-Binding Factor 1 (LEF1), Dickkopf-1 (DKK1), DKK2, Secreted frizzled-related protein-2 (SFRP2), SFRP4, growth differentiation factor 5 (GDF5) and GDF10, are indicated. E. Wnt signaling pathway. The differentially expressed genes by RNAseq are marked by red stars. F. qPCR validation of perturbed WNT gene expression, normalized to the expression of the housekeeping genes GAPDH and ACTIN: LEF1, DKK1, DKK2, SFRP2, SFRP4, and GDF10. Error bars are s.e. *P < 0.05, **P < 0.01, ***P < 0.001.

Table 2.

Selected WNT signalling genes differentially expressed in SZ hiPSC NPCs.

		RNAseq			qPCR
Symbol	Refseq ID	Fold-Change	P-value	FDR	Fold-Change	P-value
DKK1	NM_012242	2.32	2.23E-06	1.43E-04	2.43	7.78E-04
DKK2	NM_014421	3.35	1.45E-09	2.38E-07	7.11	1.08E-03
SFRP2	NM_003013	1.33	1.21E-06	8.61E-05	2.08	3.14E-04
SFRP4	NM_003014	2.05	1.22E-04	3.87E-03	4.31	3.37E-03
GDF5	NM_000557	3.42	1.75E-06	1.18E-04	-	-
GDF10	NM_004962	1.82	4.16E-03	6.58E-02	2.24	3.77E-02
LEF1	NM_016269	1.68	1.19E-03	2.51E-02	4.09	1.73E-03

Table 3.

Enrichment analysis for BMP signaling, hedgehog signaling or GPCR signaling pathways.

Category	Term	Fold Enrichment	FDR
KEGG_PATHWAY	Hedgehog signaling pathway	18.7	0.02
REACTOME_PATHWAY	Signaling by BMP	0.7	1.00
REACTOME_PATHWAY	Signaling by GPCR	2.1	1.00

We investigated canonical WNT activity using the well-established T-cell factor (TCF) / Lymphoid enhancer-binding factor (LEF) (TOPFlash) assay, in which transcriptional activation of TCF/LEF binding sites drives expression of a luciferase reporter (17, 18). NPCs were infected 3–7 days prior to analysis with a Lentiviral (LV)-TOPFLASH luciferase reporter, generously provided by Karl Willert (UCSD), as well as a constitutive LV-renilla reporter for normalization. SZ hiPSC NPCs showed increased canonical WNT signaling relative to controls (p<10e-5) (Fig. 2A, Fig. 3), though increased WNT signaling was not necessarily present in every patient and significant outliers often skewed results (Fig. 3). Across six independent experimental replicates, the following fold-changes in canonical WNT signaling we observed: 2.8, 4.2, 3.1, 2.7, 4.7 and 3.3 (Fig. 3). The ultimate effector of canonical WNT signaling is β-CATENIN; Western blot analysis for β-CATENIN protein (1:10,000; Millipore), normalized to β-ACTIN (1:10,000; Ambion), revealed increased β-CATENIN protein levels in SZ hiPSC NPCs (Fig. 2B).

Fig. 2. Perturbed WNT signaling in SZ hiPSC forebrain NPCs.

A. Comparison of canonical WNT activity (assayed as LV-TOPFLASH reporter levels relative to LV-renilla florescence) between control and SZ hiPSC forebrain NPCs, averaged by diagnosis. Luciferase levels were determined using the Dual-Glo Luciferase Assay System (Promega), measured on a FlexStation 3 (Molecular Devices) and then normalized to LV-renilla florescence. B. Increased β-CATENIN protein levels in SZ hiPSC forebrain NPCs. Western blot comparison of β-CATENIN and β-ACTIN levels in control and SZ hiPSC NPCs. Western blots were repeated twice using independent protein lysates; Student’s T tests were used to test statistical differences between control and SZ western blot β-CATENIN levels. β-ACTIN was used as a loading control because we have found no evidence, by microarray or Nanostring nCounter gene expression assays or SILAC quantitative protein mass spectrometry, that it is differentially expressed in SZ hiPSC NPCs or neurons (13, 14). C. No effect of WNT on aberrant migration in SZ hiPSC forebrain NPCs. Control and SZ neurosphere outgrowth when cultured with PBS, canonical WNT3A (20 ng/ml) and noncanonical WNT7B (5 ng/ml). Error bars are s.e. *P < 0.05, **P < 0.01.

Fig. 3. Experimental variability in assaying WNT signaling in SZ hiPSC forebrain NPCs.

A. Six experimental replicates comparing canonical WNT activity (assayed as LV-TOPFLASH reporter levels relative to LV-renilla florescence) between control and SZ hiPSC forebrain NPCs, averaged by diagnosis. With increasing passage, NPC lines can show reduced ability to differentiate to neurons or undergo spontaneous transformation to a highly proliferative cell with rounded morphology that cannot undergo neural differentiation at all; when either event occurred, that NPC line was dropped from subsequent experiments, for this reason, not all NPC lines were analysed in independent experiments. B. Six experimental replicates comparing canonical WNT activity (assayed as LV-TOPFLASH reporter levels relative to LV-renilla florescence) between control and SZ hiPSC forebrain NPCs, averaged by individual. (Top row: Mean +/− s.e. Bottom row: Variability chart showing individual data points). For phenotypic analysis, statistical analysis was performed using JMP (Carey, NC). Box-Cox transformation of raw data was performed to correct non-normal distribution of the data and means were compared within diagnosis by Oneway analysis using both Student’s T test and Tukey Kramer HSD. A nested analysis of values for individual patients was performed using standard least squares analysis comparing means for all pairs using Student’s T test for specific pairs and Tukey Kramer HSD for multiple comparisons. Error bars are s.e. *P < 0.05, **P < 0.01, ***P < 0.001.

WNT signaling has been implicated in neural migration (19). Following 48 hours of culture with either canonical (20ng/ml WNT3A) or non-canonical (5ng/ml WNT7B) WNT signals, neither control nor SZ hiPSC forebrain NPCs showed significant changes in radial migration (98 total SZ neurospheres were analyzed relative to 56 total control neurospheres) (Fig. 2C); increased canonical WNT signaling was not sufficient to recapitulate SZ aberrant migration in control hiPSC derived neurospheres (14).

Consistent with evidence suggesting that the WNT pathway could be aberrant in SZ (20), we demonstrate that SZ hiPSC forebrain NPCs derived from four patients have perturbations in WNT signaling, but caution that i) due to our small sample size, these phenotypes may not generalize across all SZ patients and ii) there was substantial variation in the specific SZ hiPSC NPC lines with increased WNT signaling between experimental replicates. SZ hiPSC NPCs with elevated canonical WNT signaling showed significantly increased experimental variation, suggesting that this phenotype might be more accurately reflect an increased variability in WNT signaling, perhaps due to increased susceptibility to an extrinsic factor, rather than implying a cell-autonomous difference in canonical WNT signaling.

A question of immediate interest is whether WNT signaling is also perturbed in SZ hiPSC neurons, and if so, in which neuronal cell types this is most evident. WNT signaling has been implicated in neural patterning, proliferation, differentiation, migration and activity-dependent synaptic modulation (12, 19, 21–25). Given that WNT signaling is typically believed to increase neurogenesis (26), and that we and others have reported reduced neuronal connectivity in SZ hiPSC neurons (14, 27), we note that the (increased) direction of change in WNT signaling observed in our hiPSC NPCs is potentially surprising, though it may reflect an attempt at compensation for neural defects in other pathway(s). Perturbations in canonical WNT signaling in SZ hiPSC NPCs foretells a practical confound for future hiPSC-based studies of SZ because aberrant canonical WNT signaling might affect the specification of SZ hiPSCs to certain neural fates. During neuronal differentiation, active WNT signaling is required for the specification of hippocampal (28) and midbrain dopaminergic fate (29, 30), while repression of WNT signaling is required for cortical interneuron (31, 32) and striatum (33, 34) neuronal patterning; two recent publications have reported differing abilities of SZ hiPSCs to differentiate into dopaminergic neurons (27, 35).

Recent rodent-(36), hiPSC-(13, 27, 37, 38) and olfactory neural stem cell-(39) based studies of SZ have reported increased oxidative stress and reactive oxygen species. There is a well-documented cross-talk between redox and WNT/β-catenin signaling (40–44); for example, treatment of cells with a low dose of H₂O_2, induces a rapid stabilization of β-catenin (43), while down-regulation of canonical WNT signaling can decrease oxidative stress (45). If increased oxidative stress does indeed contribute to perturbed canonical WNT signaling in SZ hiPSC NPCs, small variations in tissue culture induced oxidative stress between experimental replicates may be one source of the large experimental variation observed between SZ patients. Future studies, across larger patient cohorts, will be necessary to determine whether aberrant canonical WNT signaling is a causal molecular factor contributing to aberrant neural patterning and neuronal maturation in SZ, or simply a non-cell autonomous consequence of increased oxidative stress (46).

ABBREVIATIONS

SZ

schizophrenia

hiPSC

human induced pluripotent stem cell

NPC

neural progenitor cell

FDR

false discovery rate