Familial t(1;11) translocation is associated with disruption of white matter structural integrity and oligodendrocyte–myelin dysfunction

Although the underlying neurobiology of major mental illness (MMI) remains unknown, emerging evidence implicates a role for oligodendrocyte–myelin abnormalities. Here, we took advantage of a large family carrying a balanced t(1;11) translocation, which substantially increases risk of MMI, to undertake both diffusion tensor imaging and cellular studies to evaluate the consequences of the t(1;11) translocation on white matter structural integrity and oligodendrocyte–myelin biology. This translocation disrupts among others the DISC1 gene which plays a crucial role in brain development. We show that translocation-carrying patients display significant disruption of white matter integrity compared with familial controls. At a cellular level, we observe dysregulation of key pathways controlling oligodendrocyte development and morphogenesis in induced pluripotent stem cell (iPSC) derived case oligodendrocytes. This is associated with reduced proliferation and a stunted morphology in vitro. Further, myelin internodes in a humanized mouse model that recapitulates the human translocation as well as after transplantation of t(1;11) oligodendrocyte progenitors were significantly reduced when compared with controls. Thus we provide evidence that the t(1;11) translocation has biological effects at both the systems and cellular level that together suggest oligodendrocyte–myelin dysfunction.


INTRODUCTION
Schizophrenia (SZ) and other major mental illnesses (MMI) such as bipolar and major depression show high heritability. Accumulating evidence from GWAS studies points to a multifactorial polygenic inheritance, with individual genes conferring a modest increased susceptibility, as well as pleiotropy 1 . In contrast, rare genetic variants, such as a balanced chromosomal translocation in a large Scottish family, that co-segregate with MMI show highly penetrance [2][3][4][5][6] . The range of psychiatric phenotypes observed in people carrying the balanced t(1:11) translocation suggests its study will be of considerable value for improved understanding of biological processes underlying MMI. This translocation disrupts the DISC1 gene and segregates with schizophrenia and affective disorders in this large family Despite multiple lines of evidence from pathological, gene expression and radiological studies, the role of glia is understudied 2,7-15 . As oligodendrocytes enable rapid impulse propagation and provide trophic and metabolic support to axons [16][17][18][19] , their dysfunction is likely to result in altered neuronal homeostasis. Further, the sole protein-coding gene disrupted by the t(1;11) translocation; Disc1, is known to affect specification and differentiation of oligodendrocytes [20][21][22] in animal models. Hence a direct study of the impact of the t(1;11) translocation on oligodendrocytes is important. Furthermore, given the limitations of animal models of neuropsychiatric disorders, complementary models of human glia in the context of MMI are particularly necessary. Indeed, recent studies using patient derived iPS cells have shown impaired glial maturation suggesting a causal link with schizophrenia 23,24 . 4 A powerful approach to interrogate the structural and cellular white matter consequences of the t(1;11) translocation is to undertake combined water diffusion MRI (dMRI) and biological studies of iPSC-derived oligodendrocytes from patients carrying the t(1;11) translocation. Previous studies using iPSCs from individuals with DISC1 mutations and MMI have predominantly studied neural precursor and/or neuronal processes 2,25,26 . Here, we demonstrate abnormalities of white matter integrity using brain dMRI in t(1;11) translocation carrying cases compared to familial controls. In addition, case iPSC-derived oligodendrocytes display cellular and structural abnormalities in vitro as well as upon transplantation into hypomyelinated mice. This study elucidates a cell biological basis of oligodendrocyte-myelin deficits in MMI. In addition, we establish a human platform for future mechanistic studies.

Global changes in white matter structure and connectivity due to the t(1;11) translocation
To begin to understand the effect of t(1;11) translocation at a whole brain as well as cellular level, we used a multi-tiered approach integrating patient and control whole brain imaging, and in vitro and in vivo stem cells along with transgenic studies (Figure 1a).
To study the impact of t(1;11) translocation on global white matter structure and structural connectivity we undertook whole-brain probabilistic tractography using dMRI on 21 individuals from the previously reported Scottish family known to carry the t(1:11) translocation of whom 8 were carriers of the t(1;11) translocation 27 . All 8 carriers had a psychiatric diagnosis; 1 with schizophrenia, 4 with major depression (MDD), 3 with cyclothymia while only 1 of the 13 family members who did not have the t(1;11) translocation, had a psychiatric diagnosis 27 . The affected non-carrier however is described to carry modifier loci on chr11q2 and chr5q that might contribute to the development of MMI (described in 28 ). High resolution T1-weighted structural and dMRI data were combined to create structural connectivity matrices of each participant's brain. In these matrices (Figure 1b-e), the 85 grey matter regions, parcellated from the structural MRI data, are the nodes of the brain structural network while the connecting white matter pathways, identified from whole brain dMRI tractography, are its edges. The edge connection strength between nodes is obtained by recording the mean fractional anisotropy (FA), a measure of white matter microstructure, along tractography streamlines connecting all ROI (network 6 node) pairs (Figure 1b-e). Global graph theory measures of brain structural connectivity, such as network degree (number of connections one node has to other nodes), network strength (the average sum of edge connections per node) and global efficiency (the average of the inverse shortest path length between nodes), can then be calculated for each subject and compared across populations to assess differences in brain structure. In our case, we hypothesised that carriers have reduced connectivity (e.g. lower degree, strength and global efficiency) than noncarriers. We estimated the effect of translocation status (carrier versus non-carrier) and age on global connectivity measures using a Markov chain Monte Carlo (MCMC) approach 29

Altered differentiation and gene expression in t(1;11) derived oligodendroglia
In order to begin to determine the cellular and molecular basis underlying the observed abnormalities in white matter integrity, we next studied oligodendrocytes from affected and unaffected individuals -see Suppl. Table 1. iPSC lines were   established, using an episomal non-integrating method, from 4 individuals carrying   7   the mutation (Case 1 -cyclothymia, Case 2 -MDD, Case 3 -MDD, Case 4 -SZ; see   Supplementary table 1) and 3 unaffected family controls (Supplementary figure 1). H3K27 trimethylation staining indicated typical X-chromosome inactivation in the female iPSC lines as seen by distinct foci in the nucleus while the male lines showed diffuse staining (Supplementary figure2). We generated enriched oligodendrocyte lineage cells using a previously described protocol 30 (Figure 2a-c). Quantitative analysis showed no difference in progenitor specification as assessed by OLIG2 + staining on day 1 (Figure 2d)  were also found to be dysregulated in t(1;11) carrying lines (NKX2.  Figure   3d).

In vitro morphology as well as in vivo internodal length is severely affected in t(1;11) carrying oligodendrocytes
In support of the dysregulation of actin related genes in case lines, DISC1 has previously been shown to affect microtubule organization and neurite outgrowth 32,33 .
In oligodendrocytes, this is likely to affect morphological development and myelin formation. Case derived O4+ oligodendrocytes were found to be severely stunted and dysmorphic in comparison to familial controls (Figure 4a translocation.

DISCUSSION
In this study we provide converging lines of evidence from human imaging, molecular and cellular analyzes including of chimeric mice that suggests that the t(1;11) translocation causes oligodendrocyte-myelin deficits.
The results from structural connectivity, measured using graph theory analysis are consistent with widespread t(1;11) translocation dependent widespread changes to white matter connectivity. It has previously been shown that family members who carry the balanced translocation have a pattern of cortical thinning similar to that observed in patients with schizophrenia 36 . Here, we provide evidence disruption of white matter topology and organisation building on earlier voxel-based findings from ourselves showing a link between FA values and psychotic symptoms in those with the t(1;11) translocation 37 and between white matter organization, genetic risk markers and cognition in the patients with 'idiopathic' schizophrenia scanned as part of the same project 38 . The recent large ENIGMA schizophrenia diffusion tensor imaging study of over 4000 individuals has recently shown white matter microstructural differences in schizophrenia 39 and it is likely that the t(1;11) translocation is causing an altered regulation of schizophrenia candidate genes and the DISC1 core interactome as alternative pathways for risk in major mental illness 40 .
There has been much debate about the importance of DISC1 as a genetic risk factor for schizophrenia 41,42 and it is more appropriate to instead classify DISC1 as a genetic risk factor for major mental illness with a broad clinical phenotype. As such DISC1 interacts with many protein partners and mechanistically affects several neurodevelopmental pathways despite not being identified as having a clear 13 association with schizophrenia from genome-wide studies 1 . This is exemplified from the clinical phenotypes of the 4 affected carriers in this study, one with schizophrenia and three with affective disorders. Interestingly, OPCs and oligodendrocytes derived from a juvenile-onset, treatment resistant individual showed severe developmental, structural and functional changes across tests (Case 4; described in Supplementary   Table 1). In contrast, cells derived from individuals presenting a less severe symptomatology (as in Case 1) also showed comparatively milder changes in vitro.
This both highlights the need for multiple patient lines as well as the promise of iPSbased systems to model aspects of disease variation. The use of several iPS lines allows stratification based on cellular or gene expression changes as well as identification of common phenotypes to aid drug-screening approaches. In line with this, we uncovered shared gene expression changes across patient lines.
Robust and comparable differentiation of case lines compared to familial controls showed that the translocation does not interfere with early specification and patterning of cells to the oligodendroglial lineage. Case lines however, showed premature cell cycle exit of OPCs, a finding consistent with reports for cortical neurons derived from DISC1 exon 8 interrupted iPSC lines 26 and supported also by the observed increased expression of NKX2.2, ZNF488 and SLC8A3 genes. Of these, NKX2.2 and ZNF488 are described to be specific to the oligodendroglial lineage as well as having a pro-differentiation effect on OPCs [43][44][45] . Thus, their increased expression in our dataset is consistent with the observed phenotype of precocious differentiation. Our analysis also highlights genes identified via GWAS studies and implicated in psychosis (ZNF804A) 46 and suicidal behaviour (SKA2) 47 .
Additionally, several genes (UGT8, GAL3ST1, ZNF488, CTTNBP2, SLC8A3, GRIA4, 14 and KCND2) identified in our analysis agree with a recent study of glial progenitors from childhood onset schizophrenia patient derived iPSC lines 23 . This study of humanised glial chimeric mice suggests a potential causal role for impaired glial progenitor cell differentiation and found premature migration of GPCs, reduced white matter expansion and hypomyelination relative to controls as well as deficits in astrocyte differentiation and morphology 23 . This points towards convergent mechanisms for oligodendrocyte dysfunction in MMI despite differences in genetic basis. Future studies employing functional genetics approaches such as CRISPR and siRNA mediated knockdown could clarify the role of these genes in the pathophysiology.
Our finding of dysregulation of actin associated genes, is of considerable interest noting that DISC1 has been shown to play a role in microtubule reorganization and neurite outgrowth 33 . In keeping with these findings, we observed case derived oligodendrocytes to have a morphological phenotype of shorter and less complex processes. Critically, we also observed in the Der1 mouse model, that recapitulates the effects of the human translocation upon DISC1 expression, shorter internodes that was further supported by a reduction in myelin segment length found upon transplantation of case and control derived OPCs into hypomyelinated mice.
Transgenic rodent models of Disc1 have shown a range of anatomical changes including cerebral cortical thinning, reduced neurite outgrowth as well as behavioural changes (reviewed in 48 ) but finer structure deficits in myelination have thus far not been shown. A crucial role for the actin remodeling pathway in myelination has been described [49][50][51] and could serve as an important pharmacological target in future studies. Finally, our study also points towards a galactolipid imbalance in case 15 derived oligodendrocytes leading to myelin defects. In keeping with this, mice genetically altered for galactolipid pathway genes present neurological manifestations due to altered axo-glia interactions and unstable myelination [52][53][54] .
The future evaluation of in vivo functional consequences of shorter internodes would be of great interest since shorter internodes are associated with reduced conduction velocities and even degenerative changes. This is particularly relevant when considering that gamma oscillations between different brain regions are known to be affected in MMI. Parvalbumin expressing interneurons, the major cell-type thought to regulate gamma oscillations are extensively myelinated 55

MRI Imaging and Analysis
All imaging data were collected on a Siemens Magnetom Verio 3T MRI scanner running Syngo MR B17 software (Siemens Healthcare, Erlangen, Germany). For each subject, whole brain diffusion MRI (dMRI) data were acquired using a singleshot spin-echo echo-planar (EP) imaging sequence with diffusion-encoding gradients 19 applied in 56 directions (b=1000 s/mm2) and six T2-weighted (b=0 s/mm2) baseline scans. Fifty-five 2.5 mm thick axial slices were acquired with a field-of-view of 240 mm and matrix 96 × 96 giving 2.5 mm isotropic voxels. In the same session, a 3D T1-weighted inversion recovery-prepared fast spoiled gradient-echo (FSPGR) volume was acquired in the coronal plane with 160 contiguous slices and 1 mm isotropic voxel resolution. Image processing and tractography is described in Supplementary Methods.

Animal Ethics
All animal experiments were conducted in accordance with the UK Animals images through the whole section were taken using a Leica SP8 confocal microscope. Images were analyzed blind to genotype by assigning random numbers to the animals and using the ImageJ simple neurite tracer plugin.

Neonatal transplantation of OPCs into MBP Shi/Shi ;Rag2 -/mice
Homozygous MBP Shi/Shi ;Rag2 -/pups between P1-P2 were anaesthetized with isoflurane and maintained on a heat-mat for the duration of the transplantation.

Induced pluripotent stem cell line maintenance
All iPSC were derived from human donor dermal skin fibroblasts using integrationfree episomal methods 58 . iPS lines were maintained in Matrigel (BD Biosciences) coated plastic dishes in E8 medium (Life Technologies) at 37 o C and 5% CO 2 .

Karyotyping, pluripotency markers and X-chromosome inactivation staining
Standard G-banding chromosome analysis was performed to confirm chromosome number and gross genetic abnormalities over the course of this study. Pluripotency of case and control iPS lines was confirmed by staining for SOX2, OCT3/4 and TRA1-60. X-chromosome inactivation in female iPSC lines was confirmed by staining for H3K27 trimethylation marks 59 .

OPC and Oligodendrocyte generation
Derivation of neural precursors (NPCs) was performed as described 30

RNA Sequencing
Library prepraration, sequencing and analysis is described in detail in Supplementary methods. Sequencing reads will be made available via ArrayExpress or European Genome-Phenome Archive (EGA) prior to publication.

Quantitative Real-Time PCR (qRT-PCR)
RNA was extracted from cell pellets using the RNeasy Mini kit (QIAGEN). 500ng of RNA was used to prepare cDNA by random hexamer extension using MMuLV reverse transcriptase (ThermoFisher) according to the manufacturer's instructions.
Gene specific primers were used in a 96 well plate format to quantify differences in cDNA levels using a BioRad CFX96 qPCR machine.

Δ Δ
Ct method was used to normalise and quantify relative fold changes in gene expression.

Image acquisition and analysis
Images were acquired either using a Zeiss ObserverZ1 wide field or Zeiss LSM710 confocal microscope (Carl Zeiss Microimaging).
Images were converted to TIFF format, corrected for brightness and contrast and analyzed using ImageJ (NIH) or Adobe Photoshop (Adobe Inc.). Figures for preparation were assembled using Adobe Illustrator (Adobe Inc.)

Cell morphometry
Cellular area was measured in ImageJ using O4 stained images 61 . Images were thresholded, segmented and individual cells outlined using the wand selection tool.
Measurement of internodal lengths was performed using Simple Neurite Tracer plugin in ImageJ. Electron microscopy measurements were made using the freehand line tool in ImageJ. Areas was calculated by the in-built area measurement function.

Statistical Analysis
Statistical analysis of the data was performed with Prism (GraphPad software, USA).
All tests were performed as two-tailed. Tractography data was analyzed by Markov Chain Monte Carlo modelling 29 .
For experiments involving iPSCs, each derivation was considered as a n=1 and data was collected from a minimum of 3 derivations for each line. Normality of data distribution was analyzed by D'Agostino and Shapiro-Wilk's tests. Normally distributed data was analyzed using Student's t-test or by 1-way ANOVA test with Holm-Sidak's multiple comparison correction (for more than 2 groups). Welch's correction was applied when standard deviations of the two groups differed from each other. Quantitative PCR data was analyzed using non-parametric tests such as Mann-Whitney or Kruskal-Wallis tests (for 2 and >2 groups respectively). Network nodes were identified from high-resolution T 1 -weighted volume scans using Median with upper and lower quartiles is shown. The whiskers depict the range.
Cases studied using iPS cells in this particular study have been highlighted   (g-i) Electron microscopy analysis of the corpus callosum of wild-type and Der1/+ mice (g) shows that thickness of myelin (h) and the relative frequency of myelinated fibers (i) are not affected due to the translocation. >250 axons measured in each group from 4 animals per genotype.
(j-l) Single-oligodendrocyte analysis using by CNPase staining (j) at P42 supports findings in (e,f) and shows an increase in number of processes per oligodendrocytes (k) but a decrease in internodal length (l). Each data points represents one cell and the graph contains pooled data from n=28 cells each from 4 animals per genotype.