Journal of Affective Disorders 190 (2016) 429–438
Contents lists available at ScienceDirect
Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad
The microtubule-associated molecular pathways may be genetically disrupted in patients with Bipolar Disorder. Insights from the molecular cascades Antonio Drago a,1, Concetta Crisafulli b,n,1, Antonina Sidoti b, Marco Calabrò b, Alessandro Serretti a a b
Department of Biomedical and Neuromotor Sciences – DIBINEM – University of Bologna, Bologna, Italy Department of Biomedical Science and Morphological and Functional Images, University of Messina, Via Consolare Valeria, 98125 Messina, Italy
art ic l e i nf o
a b s t r a c t
Article history: Received 24 April 2015 Received in revised form 24 September 2015 Accepted 10 October 2015 Available online 23 October 2015
Bipolar Disorder is a severe disease characterized by pathological mood swings from major depressive episodes to manic ones and vice versa. The biological underpinnings of Bipolar Disorder have yet to be deﬁned. As a consequence, pharmacological treatments are suboptimal. In the present paper we test the hypothesis that the molecular pathways involved with the direct targets of lithium, hold signiﬁcantly more genetic variations associated with BD. A molecular pathway approach ﬁnds its rationale in the polygenic nature of the disease. The pathways were tested in a sample of ∼7000 patients and controls. Data are available from the public NIMH database. The deﬁnition of the pathways was conducted according to the National Cancer Institute (http://pid.nci.nih.gov/). As a result, 3 out of the 18 tested pathways related to lithium action resisted the permutation analysis and were found to be associated with BD. These pathways were related to Reelin, Integrins and Aurora. A pool of genes selected from the ones linked with the above pathways was further investigated in order to identify the ﬁne molecular mechanics shared by our signiﬁcant pathways and also their link with lithium mechanism of action. The data obtained point out to a possible involvement of microtubule-related mechanics. & 2015 Elsevier B.V. All rights reserved.
Keywords: Pathway analysis Aurora Reelin Integrin Microtubule-related process
1. Introduction Bipolar Disorder (BD) is a mental disease characterized by pathological mood swings, both depressive and manic. It affects ∼1% of the population world-wide and results in high economic and societal costs for the communities in which patients live (American Psychiatric Association, 2000). Pharmacological treatment is considered necessary for BD (American Psychiatric Association, 2002), even if the present treatments present some (major) ﬂaws. First the efﬁcacy of drugs is limited: up to one third of subjects, and more than half of patients diagnosed with BD may experience relapses during their lifetime despite treatment. Further, the drugs that are in use today (mood stabilizers, second-generation antipsychotics and antidepressants), though generally effective in the short term, do not dramatically change the natural history of the disorder (Angst and Sellaro, 2000). Finally, the appearance of severe side effects is a somewhat common experience. The absence n
Corresponding author. E-mail address: [email protected]
(C. Crisafulli). 1 Authors that contributed equally to the work.
http://dx.doi.org/10.1016/j.jad.2015.10.016 0165-0327/& 2015 Elsevier B.V. All rights reserved.
of a detailed knowledge about the biological background of BD is one of the reasons for the lack of a highly efﬁcient and/or speciﬁc drug treatment. Consistently, we are witnessing a shift of the scientiﬁc paradigm of the biological basis of BD from the monoamines to a more comprehensive picture of biological pathways involved with neurodevelopment, neurodegeneration and in general with the normal functioning of the brain (Berk et al., 2011). For decades the investigation of monoamines did not yield a solid breakthrough for the understanding of BD, and now that the perspective is changing, pressed by the ﬁndings form genomewide investigations among others, we are still waiting for a consistent biological understanding of BD. Thus, it is still impossible to engineer a drug based on consistent biological evidence about the pathophysiology of BD. Genetics holds the potential to unravel the genes that harbor the variations that signiﬁcantly segregate in patients versus healthy controls. This ﬁeld has been extensively and completely reviewed recently (Sullivan et al., 2012). Things are really far to be elucidated though, and this ﬁeld of research appears to be still in its infancy. Moreover, the brain proved to be complex to investigate and the polygenic nature of BD further hinders any potential biological understanding of the disorder. Finally, both rare and common variations may be involved in the
A. Drago et al. / Journal of Affective Disorders 190 (2016) 429–438
Table 1 Sample under analysis. Sample
Trinity College Dublin University of Edinburgh Pritzker Neuropsychiatric Disorders Research Consortium
Irish Scottish European– American Systematic Treatment Enhancement Program for Bipolar Disorder European– (STEP1) American Systematic Treatment Enhancement Program for Bipolar Disorder European– (STEP2) American Thematically Organized Psychosis (TOP) Study Norwegian University College London (UCL) British Grand total /
pathophysiology of BD (Sullivan et al., 2012), therefore the investigation of very large samples is needed to tell apart the noise from the true signal, and to extrapolate the impact of rare variations. For such a reason, during the recent years some international consortiums joined forces to create the larger genetic samples ever seen for psychiatric diseases, BD included. Some of these datasets are available for international researchers to be investigated. In the present paper we studied ∼7000 BD patients and controls from the public NIMH dataset (Table 1). We used a molecular pathway analysis to investigate the sample, a technique which proved to be more powerful compared to single SNPs analysis and which takes into account the polygenetic nature of the disorder in a more speciﬁc way compared to common genome-wide investigations (Holmans, 2010). For the identiﬁcation of biological pathways possibly involved with BD we choose to use, as starting point, the molecular pathways involved with lithium, given its relevance for the treatment of BD. We examined several molecular networks related to these elements under the hypothesis that the pathways involved in the pharmacodynamics of lithium could explain, at least in part, the pathophysiology of the disorder. A case-control mega-sample was used to test the hypothesis that the molecular networks derived from lithium's molecular targets are also involved in the genetic risk of BD. This process entails a cross-relation between the pharmacodynamics of a benchmark drug for the treatment of a disease, and its genetic background. This cross-relation was not thoroughly previously demonstrated in literature and the analysis process that is undertaken in the present work is therefore to be considered as exploratory.
2. Materials and methods 2.1. Datasets Genetic and phenotypic data were available from the NIMH (https://www.nimhgenetics.org). Table 1 reports the characteristics of the samples under investigation. 3803 cases and 3470 controls were analyzed. 2.2. Quality control of single datasets Quality control was performed on genotypes generated by various GWAS platforms, with quality control conducted separately using a common approach. The SNPs successfully genotyped in each study and common to all platforms were pruned to remove high LD and lower frequency SNPs and were then used for relatedness testing in each sample and in the combined total sample.
Case (n) Case ♂ (n) Case ♀ (n) Control (n) Control ♂ (n) Control ♀ (n) 150 282 1130
72 121 426
78 161 704
797 275 718
236 141 366
561 134 352
203 457 3803
83 182 1583
120 275 2220
349 495 3470
176 212 1546
173 283 1924
Common quality control parameters were applied: (i) missing rate per SNP o0.05 (before sample removal below), (ii) missing rate per individual o0.02, (iii) missing rate per SNP o0.02 (after sample removal above), (iv) missing rate per SNP difference in cases and controls o0.02, (v) SNP frequency difference to HapMap o0.15, and (vi) Hardy–Weinberg equilibrium (controls) P4 1 10 6. Samples size varied between ∼150 and ∼1200 individuals. The number of SNPs per study after quality control varied between 250,000 and 680,000. On average, the quality control processes excluded 15 individuals per study (with a range of 0–100 individuals) and 38,000 SNPs per study (with a range of 5000–160,000 SNPs). The pool of eligible SNPs relatedness testing and population structure analysis, were further pruned to remove LD (leaving no pairs with r2 40.05) and lower frequency SNPs (minor allele frequency o0.05). 2.3. Genetic quality control Genetic quality control included relatedness testing and principal components analyses. Identical subjects (pi-hat 4 0.9) were identiﬁed and one was included after random choice. Relatives (pihat40.2) were excluded as well, by choosing randomly one representative. Principal component estimation was performed with the non-related subset of individuals. The population structure and the study of origin were chosen as covariates along with gender. Cases were grouped in a single dataset and a relatedness test was conducted. Pairs (pi-hat 40.9) and relatives (pi-hat 4 0.2) were excluded and one randomly chosen representative was reintegrated in the sample groups. The genetic heterogeneity across samples was controlled at the level of the genetic quality control in the analysis' ﬂow. The rationale under this choice was that the SNPs found to speciﬁcally segregate in one or another sample, were excluded from the analysis, was lower in the analysis' ﬂow. Nevertheless, this cannot rule out the possibility that pooling heterogeneous samples may lead to spurious differences between patients and controls. 2.4. Flowchart of the analyses As a ﬁrst step of the work, the pathways under analysis were selected from the National Cancer Institute (http://pid.nci.nih.gov/) database according to their involvement with lithium-based therapy. Below are reported the pathways under investigation.
Androgen receptors glucose metabolism LKB1 Reelin
Aurora Hedgehog LPA Stem
Calcineurin Insulin p53 Trans
cdc42 Integrin Presenilin Trk
A. Drago et al. / Journal of Affective Disorders 190 (2016) 429–438
The pathways were then tested (each pathway considered as a single and independent entity), to check if they harbored a signiﬁcantly larger number of variations associated with the phenotype than expected by chance. After the identiﬁcation of the signiﬁcant pathways among the ones under investigation, the genes within them that were more affected by the variation, were selected. In this phase a custom molecular network was also built to evaluate the quality and number of the interactions across the selected genes (nodes). Finally, we extracted information from Drugbank database to ﬁnd the known molecules that act as direct interactors with lithium. We added these information to the Cytoscape software, then proceeded to enrich and re-evaluate of the custom network created in the previous step. This process was executed in the attempt to trace the chain of molecular events and targets possibly involved with the action of lithium in BD subjects. 2.5. Statistical analyses 2.5.1. Test for association: method and procedure The selection of the random pathways was undertaken following two stringent criteria: (1) having the same number of variations of the original set of genes and (2) all the SNPs randomly selected had to belong to genes. 106 random SNPs lists (permutations) were created to limit false positives. The same number of Fisher Tests was then conducted to test whether or not there were more variations associated with the phenotype in the original pathways compared to the random ones. Finally, the permutated p value of the 106 associations was calculated as previously reported (Phipson and Smyth, 2010). We used this latter approach in previous works (Drago et al., 2011) but it may be biased to the limit that “one pathway eventually always wins”. We then designed a method that allows for a “there is no winner” output. Moreover, we set the permutation analysis to test 106 random pathways against the real phenotype and chose not to permute the phenotype itself against the index pathway, as other available programs do. This perspective was chosen to test the hypothesis that “there are random pathways that better associate with the real phenotype” rather than “there are random phenotypes that better associate with the real pathway”. In all facts, the pathway analysis can be conducted through other available programs. Inrich (Lee et al., 2012) is one of the most interesting and valuable programs but when permutating within the genome to extract the p values of enrichment, it selects random intervals without strictly selecting SNPs that belong to genes. This was considered to be a limitation because the starting point for the original set of SNPs and the random ones is different (all genetic variations vs completely random variations). Magenta (Segre et al., 2010) is another program that conducts a pathway analysis but the permutation is based on phenotypes while we were interested in simulating other possible genetic backgrounds based on a selection of genes. For these reasons, we deﬁned an algorithm that takes into account these issues. 2.5.2. Imputation Imputation was run for the genes that belong to the pathway under analysis in order to decrease the computational effort. The CEU HapMap 1000 genomes served for the analysis. Pruning was undertaken after imputation.
2.5.3. Pathway enrichment analysis The genomic sections that contain the genes selected for the analysis were identiﬁed through PLINK annotation. These sections were imputed, checked for quality (info 40.9) and pruned (r2 40.2), please refer to previous paragraphs, and their association with the phenotype (Bipolar patients vs controls) tested. From the remaining part of the genome a random selection of variations was taken, their number summing up to the number of analyzed SNPs within the pathway of interest, and all of them belonging to genes randomly selected throughout the genome. The association of this random list of variations was tested against the phenotype under analysis. Then, the prevalence of variations signiﬁcantly (P o0.01) associated with the outcome in the pathway under analysis and the random pathway were tested for a signiﬁcant different distribution (Fisher exact test). The 0.01 level was chosen in order to limit false positive ﬁndings while starting with the analysis, while keeping the higher probabilities to catch any true positive association. This operation was repeated 106 times (permutation) selecting different groups of random SNPs as a control group, under the same parameters described above. 106 Fisher Tests where then conducted and the permuted p was extracted according to Phipson and Smyth (2010). 2.5.4. Toolset for the analysis Executables (BASH language for LINUX based OS) were written for operationalize the analysis ﬂow. They are available upon request. R (Core Team, 2014), PLINK (Purcell et al., 2007), SNPTESTv2.3.0 (Marchini and Howie, 2010) and IMPUTE2 (Marchini and Howie, 2010) were part of the executables and served for the analyses.
3. Results In the following table the description of the sample under investigation is reported (Table 1). The target of the ﬁrst batch of analyses was to highlight which of the pathways, involved in lithium action (see method), were signiﬁcantly associated with BD. Out of the selected pathways, Reelin, Integrin and Aurora held more variations signiﬁcantly associated with BD within their related genes, compared to what expected by chance. In particular the Reelin related SNPs with a signiﬁcant association with the phenotype were twice the number expected by chance (32 vs 14). The number of SNPs within the Integrin were two times more than expected (22 vs 12) while the Aurora pathway had four times the number of signiﬁcantly associated SNPs than expected by chance (21 vs 8). In Table 2 the details of association for each pathway are reported. Table 2 Pathways associated with BD after 106 permutations at an exploratory p level of 0.01. Pathway Number of available genes
Perm p SNPs associated with phenotype at P o0.01
28 Index_path Random_path Integrin 45 Index_path Random_path Aurora 31 Index_path Random_path
True 32 (2%) 14 (1%)
False 1111 (97%) 1129 (98%)
22 (3%) 12 (1%)
582 (96%) 592 (98%)
21 (6%) 8 (2%)
294 (93%) 307 (97%)
The random pathway is one out of the 106 permutated random pathways used for the permutation analysis and the prevalence of the signiﬁcant SNPs associated with the outcome is a chance ﬁnding.
Table 3 Characteristics of the genes with the major number of SNPs associated with BD within investigated pathways. Gene long name
Cytoplasmic Polyadenylation Element Binding Protein 1 NCK adapter protein 2
Snail Family Zinc Finger 1
Reelin, Integrin Integrin
Transforming, acidic coiledcoil containing protein 3
Phosphoinositide-3-Kinase, Regulatory Subunit 1 (Alpha)
RuvB-like AAA ATPase 2
Tudor domain containing 7
TPX2, microtubule-associated Glutamate receptor, ionotropic, N-methyl D-aspartate 2B Diaphanous-related formin 1
HSP90AA1 Heat Shock Protein 90 kDa Alpha (Cytosolic), Class A Member 1 PARVB Parvin, beta
LIM and senescent cell antigen-like domains 2
Protein phosphatase 2, regulatory subunit B', delta Aurora kinase B
totSNP Sig_0.01 Freq Characteristics
totSNP: Total number of SNPs within the gene Sig_0.01: Number of SNPs signiﬁcantly associated with the pathology Freq.: Signiﬁcant SNPs/total SNPs%.
Encodes for a sequence-speciﬁc RNA-binding protein that regulates mRNA cytoplasmic polyadenylation and translation initiation during oocyte maturation, early development and at postsynapse sites of neurons. Encodes for an adapter protein of NCK family which plays a role in ELK1-dependent transcriptional activation in response to activated Ras signaling Encodes for a zinc ﬁnger transcriptional repressor which is involved in induction of the epithelial to mesenchymal transition (EMT), formation and maintenance of embryonic mesoderm, growth arrest, survival and cell migration. Encodes for a member of the transforming acidic colied-coil protein family. It is involved in the microtubule-dependent coupling of the nucleus and the centrosome and regulates centrosome-mediated interkinetic nuclear migration (INM) of neural progenitors. Encodes for a subunit of Phosphatidylinositol 3-kinase enzyme which is necessary for the insulin-stimulated increase in glucose uptake and glycogen synthesis in insulin-sensitive tissues. Plays an important role in signaling in response to FGFR1, FGFR2, FGFR3, FGFR4, KITLG/SCF, KIT, PDGFRA and PDGFRB. Likewise, plays a role in ITGB2 signaling This gene encodes the second human homolog of the bacterial RuvB gene, a DNA helicase. Plays an essential role in oncogenic transformation by MYC and also modulates transcriptional activation by the LEF1/TCF1CTNNB1 complex. May also inhibit the transcriptional activity of ATF2 Encodes a component of cytoplasmic RNA granules which acts by binding to speciﬁc mRNAs and regulating their translation. Encodes form TPX2 protein that is required for normal assembly of microtubules during apoptosis. Encodes for a subunit of the ionotropic receptor of Glutamate. It is essential for the correct functioning of the brain.
Previous ﬁndings in literature Thiess (1968), Huang et al. (2003) Suzuki et al. (2002), Poul and Fait (1975) –
Yang et al. (2012), Wurdak et al. (2010) Chen et al. (2014)
Satoh et al. (2013)
Satoh et al. (2013) Martucci et al. (2006), Zhao et al. (2011), Szczepankiewicz et al. (2009) This gene encodes for a poteind required for the assembly of F-actin structures. May as a scaffold protein for Kuo et al. (2006) MAPRE1 and APC to stabilize microtubules and promote cell migration and it seem to have a neurite outgrowth promoting activity. This genes encodes for an actin-binding protein associated with focal contacts. Probably plays a role in the – regulation of cell adhesion and cytoskeleton organization Martins-de-Souza et al. (2009) Encodes for a is stress induced molecular chaperonen that functions as homodimer. Promotes the maturation, structural maintenance and proper regulation of speciﬁc target proteins involved for instance in cell cycle control and signal transduction. – Encodes for a member of the parvin family of actin-binding proteins, which play a role in cytoskeleton organization and cell adhesion. Adapter protein that plays a role in integrin signaling via ILK and in activation of the GTPases CDC42 and RAC1 by guanine exchange factors, such as ARHGEF6. Is involved in the reorganization of the actin cytoskeleton and formation of lamellipodia. Encodes for member of a small family of focal adhesion proteins and is involved in the modulation of cell – spreading and migration. This adapter protein links beta-integrins to the actin cytoskeleton, and bridges this cytoplasmic complex to cell surface receptor tyrosine kinases and growth factor receptors. This gene encodes for one of the four major Ser/Thr phosphatases, that is implicated in the negative control – of cell growth and division. Encodes for a member of the Aurora kinase subfamily of serine/threonine kinases. It is the catalytic com- – ponent of the chromosomal passenger complex (CPC) that is critical for the correct progression through and completion of mitosis.
A. Drago et al. / Journal of Affective Disorders 190 (2016) 429–438
A. Drago et al. / Journal of Affective Disorders 190 (2016) 429–438
Following the identiﬁcation of the relevant pathways, we proceeded to select the genes involved in the above processes, that harbor more SNPs associated with the phenotype than expected by chance. The results of this selection are following genes: NCK2, GRIN2B, PIK3R1, SNAI1, TACC3, RUVBL2, DIAPH1, PARVB, PARVG, HSP90AA1, LIMS2, PPP2R5D, TPX2, CPEB1, AURKB and TDRD7. The main characteristics of the above results are summarized in Table 3. In this phase, we also collected information on the genes coding for the molecules directly interacting with lithium (from Drugbank database). It resulted in a list of 5 genes: IMPA1, IMPA2, INPP1, GSK3β and GRIA3. Using the two selections of genes as prompt for further pathway analyses, we built an enriched Interaction Network. The result is displayed in Fig. 1. The analysis of the various interactions and the roles of the elements within this interaction network seems to indicate that these genes are mainly involved in the regulation of microtubulebased processes and in the phosphorylation of phosphatidylinositol molecules. In Fig. 2 the processes and sub-processes linked with the Interaction Network are summarized.
4. Discussion The present paper investigates the possible liability genes of BD starting from a different perspective compared to standard genetic analyses. We investigated pathways linked to lithium mechanism of action. In fact lithium has been used in the treatment of acute bipolar mania and in maintenance treatment for patients with BD for over 60 years (Baldessarini and Tondo, 2000) and it is generally considered the most effective treatment for BD, but without a complete knowledge of its molecular action in the brain. Using this incomplete knowledge as starting point for our analyses may help to unravel possible pathways that are associated with BD. 4.1. Main results and implications Our ﬁrst set of analyses underlined the signiﬁcant association of three biological pathways with BD. The Reelin, Integrin and Aurora kinases signaling pathways (Table 2). These three networks take part in various physiological events primarily related in cell survival, differentiation and tissue organization, contributing on several aspects of cell maintenance. Triggering these cascade result in a variety of short-term and longterm responses that may inﬂuence mood stability. Data in literature already demonstrated how Reelin signaling pathway is closely connected with the physiology of the brain. Several neuronal cells subtypes express the Reelin protein (Frotscher, 1998; Curran and D'Arcangelo, 1998; Del Rio et al., 1997), which is also present in glial and astrocytic processes, even if at lower levels than in neurons (Roberts et al., 2005). The induction of the Reelin cascade leads to clustering of the receptors resulting in recruitment and activation of various factors which seems to ultimately lead to a control of neuronal migration (Assadi et al., 2003; Bock and Herz, 2003; Beffert et al., 2002; Hiesberger et al., 1999; D'Arcangelo et al., 1999). Also, this induction may be important for synapse formation and for the modulation of synaptic transmission and synaptic plasticity (Ventruti et al., 2011; Groc et al., 2007; Qiu and Weeber, 2007; Herz and Chen, 2006; Chen et al., 2005; Beffert et al., 2005). Alterations of this process may cause severe consequences in the physiology of the brain, and may be involved in the onset and development of psychiatric disorders (Folsom and Fatemi, 2013; Ovadia and Shifman, 2011). Integrins seem to be also deeply involved in the physiology of
the brain. These elements have a role in brain development, maturation of neural circuits, and adult neuroplasticity (Lubbers et al., 2014). In particular, the integrin signaling pathway, seems to be able to modulate changes in cytoskeleton function. These changes, regulate the blockade or induction of neuronal activity (Dunaevsky et al., 1999) acting through the regulation of spine motility. This spine motility may turn out to be important in regulating brain function as an adjunct element that changes the value of excitability independently from afferent stimulation. Of interest, a decrement of dendritic spine density exists in the neocortex of psychiatric subjects (Glantz and Lewis, 2000; Garey et al., 1998). Some elements belonging to the integrin family were already studied for association with psychiatric disorders, in particular the integrinβ3 (ITGβ3) (Probst-Schendzielorz et al., 2015; Napolioni et al., 2011; Weiss et al., 2006). The role of the ITGβ3 gene in the brain is to control platelet function, cell-adhesion, and cell signaling (Weiss et al., 2006). Our third signiﬁcant pathway is related to Aurora Kinases enzymes. These elements represent a family of serine/threonine kinases whose main role appear to be the modulation of cell cycle. In particular, two of the three main molecules that form this family, Aurora A and Aurora B have multiple roles in centrosome function, chromatid separation, and cytokinesis (Marumoto et al., 2003; Hauf et al., 2003; Glover et al., 1995). Little is known about the third member, Aurora-C. Recent reports show that this enzyme is also a chromosomal passenger protein, and that it binds directly to INCENP and survivin in vitro (Yan et al., 2005). Aurora kinases are mainly studies for their involvement in tumorigenesis; they can be found overexpressed in a variety of tumor cell lines (Tatsuka et al., 1998). Although it is known, as brieﬂy summarized before, that our signiﬁcant pathways are involved in the mechanisms of the brain, the exact molecular correlation of the above processes with BD and lithium treatment is somewhat undeﬁned. In front of these facts, we decided to design further analyses in order to better elucidate why these pathways resulted signiﬁcantly associated with BD and how they are linked with lithium. For doing so, we decided to focus only on a selection of genes linked with the pathways. The choice of the genes was done according to the number of variation associated with BD harbored with each candidate, and was also inﬂuenced by the quality and the number of their interactions. Cytoscape software and the related tools were used to perform this selection. After the selection of the core members that are the main reason for the positive association we obtained, we performed a second selection of genes whose products directly interact with lithium. A simple research through the Drugbank database was performed to complete this task. The two selections were then used to build an enriched custom network. This network contains the ﬁne molecular mechanics shared by our signiﬁcant pathways (resulted from the ﬁrst selection of genes) and it also presents several possible connections that link the core members with the direct interactors of lithium (resulted from the second selection of genes and the enrichment). Once obtained this enriched interaction network, we analyzed it to highlight the principal processes linked to this network. The results we obtained may be interesting: the core genes from the three pathways seems to converge on a speciﬁc biological process that is involved in the organization of the microtubular structures and microtubule-associated mechanics2. In particular, they regulate cytokinesis. This is a typical process that follows 2 Our data also evidenced an involvement in the phosphorylation of phosphatidylinositol, however we did not deepen this aspect, as this process is too common in too many cell types, while microtubule-related transport may be more speciﬁc for the correct functioning of the brain.
A. Drago et al. / Journal of Affective Disorders 190 (2016) 429–438
Fig. 1. Interaction Network of the candidate genes under analysis. In ﬁgure above is represented the interaction network built from the signiﬁcant data resulted from our analyses. As indicated in the legend under the network, shape and colors of both Nodes and Edges indicate a precise property of the genes (Nodes) and of their interactions (Edges). The form of the nodes indicates the origin of the nodes. Rounded Nodes are the genes resulted from our previous analyses. Octagonal Nodes represent the direct interactors of lithium (from drugbank Database). Triangular Nodes indicate the enrichment derived interactors of the initial elements (rounded and octagonal nodes). The color of the nodes indicates the biological mechanism or mechanisms in which each element is involved. Light Green: Regulation of microtubule-based process. Dark Green: phosphatidylinositol phosphorylation. For a better distinction, genes belonging to the initial cluster have a red name while enrichment-derived names are indicated with a black name. The edges represent indicate the type of interaction existing between the various elements of the network. The shapes at the ends of each edge indicate the direction of interaction and the type of action each elements has to its interactor (e.g. Node A - Node B ¼Node A activates Node B). The color of the edges indicates the type of interaction Green: Activation; Blue: Binding; Yellow: Expression; Purple: post-translational modiﬁcations (ptmod). The above data is based on Cluepedia_String-Actions database (v9). (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.)
mitosis/meiosis. It was demonstrated that this kind of mechanisms plays an additional role in neurons where it controls polarity and migration of these cells (Falnikar et al., 2013). Defects on transport can severely inﬂuence the functions of these cells and thus may have a crucial role in the development of neurological disorders. The stabilization of the microtubular trafﬁc may have a positive effect on psychotic phenotypes like BD. Some experimental studies evidenced in fact an association between BD and an impaired microtubule transport (Kinoshita, 2012; Ogawa et al., 2014). We tested the enriched custom network to evaluate its association with BD. In Table 3 we reported a summary of the more signiﬁcantly associated components. At last, in order to better clarify the ﬁne molecular mechanics involved, we tried to analyze the elements included in this
network. 4.2. Genes and molecular events possibly involved in lithium action According to the interaction data on Drugbank database (Kutmon et al., 2013; Cytoscape-Consortium, 2001), lithium directly associates with 5 proteins. Three of them, IMPA1, IMPA2 and INPP1 are involved in the phosphatidylinositol signaling pathway. Lithium acts as an inhibitor of these targets and causes a reduction of myo-inositol levels, thus inﬂuencing DAG and IP3 levels, two important effectors that control neuronal growth, hippocampal LTP and stress induced cognitive impairment related mechanics. An additional target of lithium is represented by GSK3β. GSK3β is a negative regulator of glucose homeostasis and it seems to play an important role in ERBB2-dependent stabilization of microtubules
A. Drago et al. / Journal of Affective Disorders 190 (2016) 429–438
Fig. 2. Principal and sub- processes linked with the Interaction Network. In the above ﬁgure are reported the biological processes involved with our Interaction Network. On the left are indicated the sub-processes with the % of the genes/Term ratio. (Genes ¼ genes involved with the speciﬁc sub-process, Term ¼the total number of genes within the Interaction Network). Colors indicate the principal processes linked with the sub-processes. Light green¼ “Regulation of microtubule-based process”; Dark green¼ “Phosphatydilinositol phosphorylation”. Overall, ∼55% of the term within the interaction network indicate a correlation of this network with the processes regulating the functioning of microtubules, while ∼45% are related to the phosphatydilinositol phosphorylation. These results are inﬂuenced by the presence of terms connected to both the principal processes. The data was obtained by the use of ClueGO, an addon of Cytoscape software tool. Ontology used: GO_BiologicalProcess; GO_MolecularFunction; KEGG; REACTOME Statistical Test Used ¼Enrichment (Right-sided hypergeometric test) Correction Method Used¼ Bonferroni step down. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.)
at the cell cortex. The last known direct target of lithium is Gria3. Gria3 is a glutamate receptor ionotropic, AMPA3. Glutamate receptors are the predominant excitatory neurotransmitters receptors in mammalian brain. In particular, it seems to produce a large potentiation of the GluR3 ﬂop splice variant, that inhibits rapid desensitization (Karkanias and Papke, 1999). As summarized in Fig. 1, GSK3β seems to have a relevant role in the network which was built on the basis of our ﬁndings, while it is also evidenced a modest involvement of GRIA3 (and then glutamate related neurotransmission) [bottom-left corner of ﬁgure] and only a marginal involvement on INPP1 (and the related phosphatydil inositol pathway) [in the upper part of image]. Most of the nodes seems to be involved in the regulation of microtubules dynamics. And GSK3β may mediate the effects of lithium on the various biological processes involved. Some of our signiﬁcant genes seems to interact with GSK3β. One of the most relevant interactions is the one between GSK3β and NCK2. This molecule is an adapter protein which associates with tyrosine-phosphorylated growth factor receptors or their cellular substrates. This protein plays a role in cytoskeletal reorganization and takes part in the processes involved in both in Reelin and Integrin–related pathways. Through interaction with DOCK5 and AKT1, NCK2 is able to phosphorylate and inactivate GSK3β, thus regulating the microtubule dynamics (Ogawa et al., 2014). Subjects with an altered NCK2 present an up-regulated GSK3β that may be normalized through lithium. Another hypothesis is that the inhibiting effects of lithium on GSK3β may be dependent from NCK2 action, so an altered NCK2 may vary the dynamics of lithium. According to our pathway data NCK2 also interact with ILK, which seems to be able to inhibit GSK3β (posttranslational modiﬁcation). In general, the interaction of these molecules seems to play a role in the regulation of cell shape, polarity, cell survival and differentiation (Hehlgans et al., 2007). SNAI1, a zinc ﬁnger transcriptional repressor whose primary function is to down-regulate the expression of ectodermal genes within the mesoderm, may also play a relevant role in lithium's pharmacodynamics. It has a role in the neural crest differentiation, but also for mechanisms related to cell survival (Nieto, 2002). It is
interesting to note that this protein is under the control of GSK3β which can control its exportation to the cytoplasm and its degradation. PPP2R5D gene belongs to the phosphatase 2A regulatory subunit B family and it is implicated in the negative control of cell growth and division. It was shown that PPP2R5D interacts with the kinase in charge of phosphorylating the tau protein, GSK3β. DIAPH1 is a protein required for the assembly of F-actin structures and is important for driving tangential migration of cortical interneurons in rodents (Ercan-Sencicek et al., 2015). It interacts with MEMO1 and RHOA, and together they play an important role in ERBB2-dependent stabilization of microtubules at the cell cortex. While still being related to microtubule mechanics, some of our signiﬁcant genes seems to have only a low probability of being inﬂuenced by GSK3β (too many nodes of distance from this protein, meaning that they can be inﬂuenced/controlled by other factors). At a more deep analysis, it seems that some of them may suffer a greater inﬂuence from inositol associated pathway, even if not directly depicted in our resulting network. This is the case of genes like: PIK3R1, PARVB, PARVG and LIMS2. PIK3R1, the regulatory subunit of Phosphatidylinositol 3-kinase, may represent the link between inositol pathway and GSK3β. According to Reelin pathway, PIK3R1 may interact with akt1 and then inactivates GSK3Β causing microtubule depolymerization. The probable correlation with BD, however, goes through its action on PTEN. PIK3R1 mutations (activating) destabilizes PTEN, a negative regulator of the PIK3 pathway (Cheung et al., 2011). Its destabilization may cause abnormalities in oligodendrocytes, exerting severe myelination defects included thickening and unraveling of the myelin sheath (Fraser et al., 2008). Thus, the destabilization of PTEN should be the key point that triggers psychiatric abnormalities. Some studies already associated these two genes with suicidal behavior (Le-Niculescu et al., 2013; Karege et al., 2011) in psychiatric subjects. PARVB, PARVG and LIMS2 seem to be associated with PIK3R1. All of them need to interact with ILK to work properly. ILK (Integrin-linked kinase) is a cytoplasmic protein serine/threonine kinase which is an important target of PI3K-mediated integrin signal transduction (Dedhar et al., 1999;
A. Drago et al. / Journal of Affective Disorders 190 (2016) 429–438
Hannigan et al., 1996). Various experiments have indicated that ILK activation is dependent on PI3K activity (Delcommenne et al., 1998). Moreover, an increase of PI3K activity stimulates ILKmediated signaling (Persad et al., 2000; Morimoto et al., 2000).
NARSAD, and the Netherlands Genetic Computing Cluster for their sponsorship of the PGC.
4.3. Limitations There are limitations that must be considered while evaluating our data. As said before in the text, our approach has to be considered speculative in that we wanted to test the association of molecular networks targeted by lithium with BD without the presence of strong evidences in literature. Thus, our results have to be considered exploratory that need further studies to conﬁrm their role in BD. Nevertheless, this data may be used as base for further analyses aimed to draw a clearer picture of the biological mechanics involved with BD. Secondly, there are limits that concern pathway analyses in general, in particular the fact that not all the genes and their functions are known. The deﬁnition of a molecular pathway is based on the current knowledge of the biological mechanisms available in literature. Thus, it may not include potential elements whose role is yet to be deﬁned (currently not included in any pathway).
5. Conclusions In the present paper we used lithium-related targets as tools to individuate biological mechanics possibly involved with the pathophysiology of BD. According to our ﬁndings, the most interesting pathways are related to the mechanics involved in migration and, in particular, polarization of neurons. In particular, the control of the microtubule system seems to be essential for the correct function of neurons. Variations in the physiological functions of these mechanics may partly explain the development of BD. Further, these mechanics may also play a role in therapy: when stimulated, they may trigger a repositioning of neurons which counteract speciﬁc defects in neural circuits. This mechanism could be associated with the therapeutic effects of lithium. Further and more speciﬁc tests between subjects responsive and resistant to lithium treatment are needed to evaluate the precise molecular chain involved with both BD and the action of lithium. Additional investigations may also investigate if Reelin-, Integrin- and Aurora kinases- associated pathways may play an independent role in the pathophysiology of BD or, as we indicated in the paper, the positive association is dependent from the microtubule related mechanics.
Ethical standards The paper is based on public accessible data from NIHM database. The data was collected and published according to the current laws of the countries of origin and were approved by the ethical committee. Role of funding source The present study was not funded.
Acknowledgments Thanks to the NIMH Center for Collaborative Genetic Studies on Mental Disorders and the Psychiatric GWAS Consortium (PGC) for providing their full sample GWAS results, individual quality controlled genotypes, and individual level dosage data for distribution to qualiﬁed investigators. We are deeply grateful to the NIMH,
American Psychiatric Association, 2000. Diagnostic and statistical manual of mental disorders: DSM-IV-TR, Revised 4th ed. American Psychiatric Association, 2002. Practice guideline for the treatment of patients with bipolar disorder. Am. J. Psychiatry 159 (Suppl. 4), 1–50, Revision. Angst, J., Sellaro, R., 2000. Historical perspectives and natural history of bipolar disorder. Biol. Psychiatry 48 (6), 445–457. Assadi, A.H., Zhang, G., Beffert, U., McNeil, R.S., Renfro, A.L., Niu, S., Quattrocchi, C.C., Antalffy, B.A., Sheldon, M., Armstrong, D.D., Wynshaw-Boris, A., Herz, J., D'Arcangelo, G., Clark, G.D., 2003. Interaction of reelin signaling and Lis1 in brain development. Nat. Genet. 35 (3), 270–276. http://dx.doi.org/10.1038/ng1257. Berk, M., Kapczinski, F., Andreazza, A.C., Dean, O.M., Giorlando, F., Maes, M., Yucel, M., Gama, C.S., Dodd, S., Dean, B., Magalhaes, P.V., Amminger, P., McGorry, P., Malhi, G.S., 2011. Pathways underlying neuroprogression in bipolar disorder: focus on inﬂammation, oxidative stress and neurotrophic factors. Neurosci. Biobehav. Rev. 35 (3), 804–817. http://dx.doi.org/10.1016/j. neubiorev.2010.10.001. Baldessarini, R.J., Tondo, L., 2000. Does lithium treatment still work? Evidence of stable responses over three decades. Arch. Gen. Psychiatry 57 (2), 187–190. Bock, H.H., Herz, J., 2003. Reelin activates SRC family tyrosine kinases in neurons. Curr. Biol. 13 (1), 18–26. Beffert, U., Morﬁni, G., Bock, H.H., Reyna, H., Brady, S.T., Herz, J., 2002. Reelinmediated signaling locally regulates protein kinase B/Akt and glycogen synthase kinase 3beta. J. Biol. Chem. 277 (51), 49958–49964. http://dx.doi.org/ 10.1074/jbc.M209205200. Beffert, U., Weeber, E.J., Durudas, A., Qiu, S., Masiulis, I., Sweatt, J.D., Li, W.P., Adelmann, G., Frotscher, M., Hammer, R.E., Herz, J., 2005. Modulation of synaptic plasticity and memory by Reelin involves differential splicing of the lipoprotein receptor Apoer2. Neuron 47 (4), 567–579. http://dx.doi.org/10.1016/j. neuron.2005.07.007. Core Team, R., 2014. R: A Language and Environment for Statistical Computing. Chen, C., Zhang, C., Cheng, L., Reilly, J.L., Bishop, J.R., Sweeney, J.A., Chen, H.Y., Gershon, E.S., Liu, C., 2014. Correlation between DNA methylation and gene expression in the brains of patients with bipolar disorder and schizophrenia. Bipolar Disord. 16 (8), 790–799. http://dx.doi.org/10.1111/bdi.12255. Curran, T., D'Arcangelo, G., 1998. Role of reelin in the control of brain development. Brain Res. Rev. 26 (2–3), 285–294. Chen, Y., Beffert, U., Ertunc, M., Tang, T.S., Kavalali, E.T., Bezprozvanny, I., Herz, J., 2005. Reelin modulates NMDA receptor activity in cortical neurons. J. Neurosci.: Off. J. Soc. Neurosci. 25 (36), 8209–8216. http://dx.doi.org/10.1523/ JNEUROSCI.1951-05.2005. Cytoscape-Consortium, 2001. Cytoscape. 〈http://cytoscape.org〉. (accessed 07.04.11). Cheung, L.W., Hennessy, B.T., Li, J., Yu, S., Myers, A.P., Djordjevic, B., Lu, Y., StemkeHale, K., Dyer, M.D., Zhang, F., Ju, Z., Cantley, L.C., Scherer, S.E., Liang, H., Lu, K.H., Broaddus, R.R., Mills, G.B., 2011. High frequency of PIK3R1 and PIK3R2 mutations in endometrial cancer elucidates a novel mechanism for regulation of PTEN protein stability. Cancer Discov. 1 (2), 170–185. http://dx.doi.org/10.1158/ 2159-8290.CD-11-0039. Drago, A., Crisafulli, C., Serretti, A., 2011. The genetics of antipsychotic induced tremors: a genome-wide pathway analysis on the STEP-BD SCP sample. Am. J. Med. Genet. B: Neuropsychiatr. Genet.: Off. Publ. Int. Soc. Psychiatr. Genet. 156B (8), 975–986. http://dx.doi.org/10.1002/ajmg.b.31245. Del Rio, J.A., Heimrich, B., Borrell, V., Forster, E., Drakew, A., Alcantara, S., Nakajima, K., Miyata, T., Ogawa, M., Mikoshiba, K., Derer, P., Frotscher, M., Soriano, E., 1997. A role for Cajal–Retzius cells and reelin in the development of hippocampal connections. Nature 385 (6611), 70–74. http://dx.doi.org/10.1038/385070a0. D'Arcangelo, G., Homayouni, R., Keshvara, L., Rice, D.S., Sheldon, M., Curran, T., 1999. Reelin is a ligand for lipoprotein receptors. Neuron 24 (2), 471–479. Dunaevsky, A., Tashiro, A., Majewska, A., Mason, C., Yuste, R., 1999. Developmental regulation of spine motility in the mammalian central nervous system. Proc. Natl. Acad. Sci. U.S.A. 96 (23), 13438–13443. Dedhar, S., Williams, B., Hannigan, G., 1999. Integrin-linked kinase (ILK): a regulator of integrin and growth-factor signalling. Trends Cell. Biol. 9 (8), 319–323. Delcommenne, M., Tan, C., Gray, V., Rue, L., Woodgett, J., Dedhar, S., 1998. Phosphoinositide-3-OH kinase-dependent regulation of glycogen synthase kinase 3 and protein kinase B/AKT by the integrin-linked kinase. Proc. Natl. Acad. Sci. U.S.A. 95 (19), 11211–11216. Ercan-Sencicek, A.G., Jambi, S., Franjic, D., Nishimura, S., Li, M., El-Fishawy, P., Morgan, T.M., Sanders, S.J., Bilguvar, K., Suri, M., Johnson, M.H., Gupta, A.R., Yuksel, Z., Mane, S., Grigorenko, E., Picciotto, M., Alberts, A.S., Gunel, M., Sestan, N., State, M.W., 2015. Homozygous loss of DIAPH1 is a novel cause of microcephaly in humans. Eur. J. Hum. Genet. 23 (2), 165–172. http://dx.doi.org/ 10.1038/ejhg.2014.82. Frotscher, M., 1998. Cajal-Retzius cells, Reelin, and the formation of layers. Curr. Opin. Neurobiol. 8 (5), 570–575. Folsom, T.D., Fatemi, S.H., 2013. The involvement of Reelin in neurodevelopmental disorders. Neuropharmacology 68, 122–135. http://dx.doi.org/10.1016/j. neuropharm.2012.08.015. Falnikar, A., Tole, S., Liu, M., Liu, J.S., Baas, P.W., 2013. Polarity in migrating neurons
A. Drago et al. / Journal of Affective Disorders 190 (2016) 429–438
is related to a mechanism analogous to cytokinesis. Curr. Biol. 23 (13), 1215–1220. http://dx.doi.org/10.1016/j.cub.2013.05.027. Fraser, M.M., Bayazitov, I.T., Zakharenko, S.S., Baker, S.J., 2008. Phosphatase and tensin homolog, deleted on chromosome 10 deﬁciency in brain causes defects in synaptic structure, transmission and plasticity, and myelination abnormalities. Neuroscience 151 (2), 476–488. http://dx.doi.org/10.1016/j. neuroscience.2007.10.048. Groc, L., Choquet, D., Stephenson, F.A., Verrier, D., Manzoni, O.J., Chavis, P., 2007. NMDA receptor surface trafﬁcking and synaptic subunit composition are developmentally regulated by the extracellular matrix protein Reelin. J. Neurosci.: Off. J. Soc. Neurosci. 27 (38), 10165–10175. http://dx.doi.org/10.1523/ JNEUROSCI.1772-07.2007. Glantz, L.A., Lewis, D.A., 2000. Decreased dendritic spine density on prefrontal cortical pyramidal neurons in schizophrenia. Arch. Gen. Psychiatry 57 (1), 65–73. Garey, L.J., Ong, W.Y., Patel, T.S., Kanani, M., Davis, A., Mortimer, A.M., Barnes, T.R., Hirsch, S.R., 1998. Reduced dendritic spine density on cerebral cortical pyramidal neurons in schizophrenia. J. Neurol. Neurosurg. Psychiatry 65 (4), 446–453. Glover, D.M., Leibowitz, M.H., McLean, D.A., Parry, H., 1995. Mutations in aurora prevent centrosome separation leading to the formation of monopolar spindles. Cell 81 (1), 95–105. Holmans, P., 2010. Statistical methods for pathway analysis of genome-wide data for association with complex genetic traits. Adv. Genet. 72, 141–179. http://dx. doi.org/10.1016/B978-0-12-380862-2.00007-2. Huang, Y.S., Carson, J.H., Barbarese, E., Richter, J.D., 2003. Facilitation of dendritic mRNA transport by CPEB. Genes. Dev. 17 (5), 638–653. http://dx.doi.org/ 10.1101/gad.1053003. Hiesberger, T., Trommsdorff, M., Howell, B.W., Gofﬁnet, A., Mumby, M.C., Cooper, J. A., Herz, J., 1999. Direct binding of Reelin to VLDL receptor and ApoE receptor 2 induces tyrosine phosphorylation of disabled-1 and modulates tau phosphorylation. Neuron 24 (2), 481–489. Herz, J., Chen, Y., 2006. Reelin, lipoprotein receptors and synaptic plasticity. Nat. Rev. Neurosci. 7 (11), 850–859. http://dx.doi.org/10.1038/nrn2009. Hauf, S., Cole, R.W., LaTerra, S., Zimmer, C., Schnapp, G., Walter, R., Heckel, A., van Meel, J., Rieder, C.L., Peters, J.M., 2003. The small molecule Hesperadin reveals a role for Aurora B in correcting kinetochore-microtubule attachment and in maintaining the spindle assembly checkpoint. J. Cell. Biol. 161 (2), 281–294. http://dx.doi.org/10.1083/jcb.200208092. Hehlgans, S., Haase, M., Cordes, N., 2007. Signalling via integrins: implications for cell survival and anticancer strategies. Biochim. Biophys. Acta 1775 (1), 163–180. http://dx.doi.org/10.1016/j.bbcan.2006.09.001. Hannigan, G.E., Leung-Hagesteijn, C., Fitz-Gibbon, L., Coppolino, M.G., Radeva, G., Filmus, J., Bell, J.C., Dedhar, S., 1996. Regulation of cell adhesion and anchoragedependent growth by a new beta 1-integrin-linked protein kinase. Nature 379 (6560), 91–96. http://dx.doi.org/10.1038/379091a0. Kuo, C.T., Zhu, S., Younger, S., Jan, L.Y., Jan, Y.N., 2006. Identiﬁcation of E2/E3 ubiquitinating enzymes and caspase activity regulating Drosophila sensory neuron dendrite pruning. Neuron 51 (3), 283–290. http://dx.doi.org/10.1016/j. neuron.2006.07.014. Kinoshita, M., 2012. Role of septin cytoskeleton in dopaminergic neurotransmission and neurodegeneration. Nihon shinkei seishin yakurigaku zasshi Jpn. J. Psychopharmacol. 32 (1), 25–29. Kutmon, M., Kelder, T., Mandaviya, P., Evelo, C.T., Coort, S.L., 2013. CyTargetLinker: a cytoscape app to integrate regulatory interactions in network analysis. PLoS ONE 8 (12), e82160. http://dx.doi.org/10.1371/journal.pone.0082160. Karkanias, N.B., Papke, R.L., 1999. Lithium modulates desensitization of the glutamate receptor subtype gluR3 in Xenopus oocytes. Neurosci. Lett. 277 (3), 153–156. Karege, F., Perroud, N., Burkhardt, S., Fernandez, R., Ballmann, E., La Harpe, R., Malafosse, A., 2011. Alterations in phosphatidylinositol 3-kinase activity and PTEN phosphatase in the prefrontal cortex of depressed suicide victims. Neuropsychobiology 63 (4), 224–231. http://dx.doi.org/10.1159/000322145. Lee, P.H., O'Dushlaine, C., Thomas, B., Purcell, S.M., 2012. INRICH: interval-based enrichment analysis for genome-wide association studies. Bioinformatics 28 (13), 1797–1799. http://dx.doi.org/10.1093/bioinformatics/bts191. Lubbers, B.R., Smit, A.B., Spijker, S., van den Oever, M.C., 2014. Neural ECM in addiction, schizophrenia, and mood disorder. Prog. Brain Res. 214, 263–284. http: //dx.doi.org/10.1016/B978-0-444-63486-3.00012-8. Le-Niculescu, H., Levey, D.F., Ayalew, M., Palmer, L., Gavrin, L.M., Jain, N., Winiger, E., Bhosrekar, S., Shankar, G., Radel, M., Bellanger, E., Duckworth, H., Olesek, K., Vergo, J., Schweitzer, R., Yard, M., Ballew, A., Shekhar, A., Sandusky, G.E., Schork, N.J., Kurian, S.M., Salomon, D.R., Niculescu 3rd, A.B., 2013. Discovery and validation of blood biomarkers for suicidality. Mol. Psychiatry 18 (12), 1249–1264. http://dx.doi.org/10.1038/mp.2013.95. Marchini, J., Howie, B., 2010. Genotype imputation for genome-wide association studies. Nat. Rev. Genet. 11 (7), 499–511. http://dx.doi.org/10.1038/nrg2796. Martucci, L., Wong, A.H., De Luca, V., Likhodi, O., Wong, G.W., King, N., Kennedy, J.L., 2006. N-methyl-d-aspartate receptor NR2B subunit gene GRIN2B in schizophrenia and bipolar disorder: Polymorphisms and mRNA levels. Schizophr. Res. 84 (2–3), 214–221. http://dx.doi.org/10.1016/j.schres.2006.02.001. Martins-de-Souza, D., Gattaz, W.F., Schmitt, A., Maccarrone, G., Hunyadi-Gulyas, E., Eberlin, M.N., Souza, G.H., Marangoni, S., Novello, J.C., Turck, C.W., Dias-Neto, E., 2009. Proteomic analysis of dorsolateral prefrontal cortex indicates the involvement of cytoskeleton, oligodendrocyte, energy metabolism and new potential markers in schizophrenia. J. Psychiatr. Res. 43 (11), 978–986. http://dx.
doi.org/10.1016/j.jpsychires.2008.11.006. Marumoto, T., Honda, S., Hara, T., Nitta, M., Hirota, T., Kohmura, E., Saya, H., 2003. Aurora-A kinase maintains the ﬁdelity of early and late mitotic events in HeLa cells. J. Biol. Chem. 278 (51), 51786–51795. http://dx.doi.org/10.1074/jbc. M306275200. Morimoto, A.M., Tomlinson, M.G., Nakatani, K., Bolen, J.B., Roth, R.A., Herbst, R., 2000. The MMAC1 tumor suppressor phosphatase inhibits phospholipase C and integrin-linked kinase activity. Oncogene 19 (2), 200–209. http://dx.doi.org/ 10.1038/sj.onc.1203288. Napolioni, V., Lombardi, F., Sacco, R., Curatolo, P., Manzi, B., Alessandrelli, R., Militerni, R., Bravaccio, C., Lenti, C., Saccani, M., Schneider, C., Melmed, R., Pascucci, T., Puglisi-Allegra, S., Reichelt, K.L., Rousseau, F., Lewin, P., Persico, A.M., 2011. Family-based association study of ITGB3 in autism spectrum disorder and its endophenotypes. Eur. J. Hum. Genet. 19 (3), 353–359. http://dx.doi.org/10.1038/ ejhg.2010.180. Nieto, M.A., 2002. The snail superfamily of zinc-ﬁnger transcription factors. Nat. Rev. Mol. Cell. Biol. 3 (3), 155–166. http://dx.doi.org/10.1038/nrm757. Ovadia, G., Shifman, S., 2011. The genetic variation of RELN expression in schizophrenia and bipolar disorder. PLoS ONE 6 (5), e19955. http://dx.doi.org/ 10.1371/journal.pone.0019955. Ogawa, F., Malavasi, E.L., Crummie, D.K., Eykelenboom, J.E., Soares, D.C., Mackie, S., Porteous, D.J., Millar, J.K., 2014. DISC1 complexes with TRAK1 and Miro1 to modulate anterograde axonal mitochondrial trafﬁcking. Hum. Mol. Genet. 23 (4), 906–919. http://dx.doi.org/10.1093/hmg/ddt485. Ogawa, K., Tanaka, Y., Uruno, T., Duan, X., Harada, Y., Sanematsu, F., Yamamura, K., Terasawa, M., Nishikimi, A., Cote, J.F., Fukui, Y., 2014. DOCK5 functions as a key signaling adaptor that links FcepsilonRI signals to microtubule dynamics during mast cell degranulation. J. Exp. Med. 211 (7), 1407–1419. http://dx.doi.org/ 10.1084/jem.20131926. Phipson, B., Smyth, G.K., 2010. Permutation P-values should never be zero: calculating exact P-values when permutations are randomly drawn. Stat. Appl. Genet. Mol. Biol. 9. http://dx.doi.org/10.2202/1544-6115.1585, Article39 (1-12). Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M.A., Bender, D., Maller, J., Sklar, P., de Bakker, P.I., Daly, M.J., Sham, P.C., 2007. PLINK: a tool set for wholegenome association and population-based linkage analyses. Am. J. Hum. Genet. 81 (3), 559–575. http://dx.doi.org/10.1086/519795. Poul, J., Fait, M., 1975. Veriﬁcation of suitability of the child shoes with open heel part (author's transl)]. Acta Chir. Orthop. Traumatol. Cechoslov. 42 (5), 479–481. Probst-Schendzielorz, K., Scholl, C., Eﬁmkina, O., Ersfeld, E., Viviani, R., Serretti, A., Fabbri, C., Gurwitz, D., Lucae, S., Ising, M., Paul, A.M., Lehmann, M.L., Steffens, M., Crisafulli, C., Calabro, M., Holsboer, F., Stingl, J., 2015. CHL1, ITGB3 and SLC6A4 gene expression and antidepressant drug response: results from the Munich Antidepressant Response Signature (MARS) study. Pharmacogenomics 16 (7), 689–701. http://dx.doi.org/10.2217/pgs.15.31. Persad, S., Attwell, S., Gray, V., Delcommenne, M., Troussard, A., Sanghera, J., Dedhar, S., 2000. Inhibition of integrin-linked kinase (ILK) suppresses activation of protein kinase B/Akt and induces cell cycle arrest and apoptosis of PTEN-mutant prostate cancer cells. Proc. Natl. Acad. Sci. U.S.A. 97 (7), 3207–3212. http: //dx.doi.org/10.1073/pnas.060579697. Qiu, S., Weeber, E.J., 2007. Reelin signaling facilitates maturation of CA1 glutamatergic synapses. J. Neurophysiol. 97 (3), 2312–2321. http://dx.doi.org/10.1152/ jn.00869.2006. Roberts, R.C., Xu, L., Roche, J.K., Kirkpatrick, B., 2005. Ultrastructural localization of reelin in the cortex in post-mortem human brain. J. Comp. Neurol. 482 (3), 294–308. http://dx.doi.org/10.1002/cne.20408. Sullivan, P.F., Daly, M.J., O'Donovan, M., 2012. Genetic architectures of psychiatric disorders: the emerging picture and its implications. Nat. Rev. Genet. 13 (8), 537–551. http://dx.doi.org/10.1038/nrg3240. Segre, A.V., Consortium, D., investigators, M., Groop, L., Mootha, V.K., Daly, M.J., Altshuler, D., 2010. Common inherited variation in mitochondrial genes is not enriched for associations with type 2 diabetes or related glycemic traits. PLoS Genet. 6 (8). http://dx.doi.org/10.1371/journal.pgen.1001058. Suzuki, S., Mizutani, M., Suzuki, K., Yamada, M., Kojima, M., Hatanaka, H., Koizumi, S., 2002. Brain-derived neurotrophic factor promotes interaction of the Nck2 adaptor protein with the TrkB tyrosine kinase receptor. Biochem. Biophys. Res. Commun. 294 (5), 1087–1092. http://dx.doi.org/10.1016/S0006-291X(02) 00606-X. Satoh, J., Kawana, N., Yamamoto, Y., 2013. Pathway Analysis of ChIP-Seq-Based NRF1 Target Genes Suggests a Logical Hypothesis of their Involvement in the Pathogenesis of Neurodegenerative Diseases. Gene Regul. Syst. Biol. 7, 139–152. http://dx.doi.org/10.4137/GRSB.S13204. Szczepankiewicz, A., Skibinska, M., Suwalska, A., Hauser, J., Rybakowski, J.K., 2009. No association of three GRIN2B polymorphisms with lithium response in bipolar patients. Pharmacol. Rep. 61 (3), 448–452. Thiess, A.M., 1968. Observation on health hazards and poisonings by o-phthalodinitril. zen- fur Arbeitsmedizin und Arbeitsschutz 18 (10), 303–312. Tatsuka, M., Katayama, H., Ota, T., Tanaka, T., Odashima, S., Suzuki, F., Terada, Y., 1998. Multinuclearity and increased ploidy caused by overexpression of the aurora- and Ipl1-like midbody-associated protein mitotic kinase in human cancer cells. Cancer Res. 58 (21), 4811–4816. Ventruti, A., Kazdoba, T.M., Niu, S., D'Arcangelo, G., 2011. Reelin deﬁciency causes speciﬁc defects in the molecular composition of the synapses in the adult brain. Neuroscience 189, 32–42. http://dx.doi.org/10.1016/j.neuroscience.2011.05.050. Wurdak, H., Zhu, S., Min, K.H., Aimone, L., Lairson, L.L., Watson, J., Chopiuk, G., Demas, J., Charette, B., Halder, R., Weerapana, E., Cravatt, B.F., Cline, H.T., Peters, E.C., Zhang, J., Walker, J.R., Wu, C., Chang, J., Tuntland, T., Cho, C.Y., Schultz, P.G.,
A. Drago et al. / Journal of Affective Disorders 190 (2016) 429–438
2010. A small molecule accelerates neuronal differentiation in the adult rat. Proc. Natl. Acad. Sci. U.S.A. 107 (38), 16542–16547. http://dx.doi.org/10.1073/ pnas.1010300107. Weiss, L.A., Kosova, G., Delahanty, R.J., Jiang, L., Cook, E.H., Ober, C., Sutcliffe, J.S., 2006. Variation in ITGB3 is associated with whole-blood serotonin level and autism susceptibility. Eur. J. Hum. Genet. 14 (8), 923–931. http://dx.doi.org/ 10.1038/sj.ejhg.5201644. Yang, Y.T., Wang, C.L., Van Aelst, L., 2012. DOCK7 interacts with TACC3 to regulate interkinetic nuclear migration and cortical neurogenesis. Nat. Neurosci. 15 (9),
1201–1210. http://dx.doi.org/10.1038/nn.3171. Yan, X., Cao, L., Li, Q., Wu, Y., Zhang, H., Saiyin, H., Liu, X., Zhang, X., Shi, Q., Yu, L., 2005. Aurora C is directly associated with Survivin and required for cytokinesis. Genes Cells: Devoted Mol. Cell Mech. 10 (6), 617–626. http://dx.doi.org/10.1111/ j.1365-2443.2005.00863.x. Zhao, Q., Che, R., Zhang, Z., Wang, P., Li, J., Li, Y., Huang, K., Tang, W., Feng, G., Lindpaintner, K., He, L., Shi, Y., 2011. Positive association between GRIN2B gene and bipolar disorder in the Chinese Han Population. Psychiatry Res. 185 (1–2), 290–292. http://dx.doi.org/10.1016/j.psychres.2009.11.026.