Ablation of Mrds1/Ofcc1 Induces Hyper-γ-Glutamyl Transpeptidasemia without Abnormal Head Development and Schizophrenia-Relevant Behaviors in Mice

Mutations in the Opo gene result in eye malformation in medaka fish. The human ortholog of this gene, MRDS1/OFCC1, is a potentially causal gene for orofacial cleft, as well as a susceptibility gene for schizophrenia, a devastating mental illness. Based on this evidence, we hypothesized that this gene could perform crucial functions in the development of head and brain structures in vertebrates. To test this hypothesis, we created Mrds1/Ofcc1-null mice. Mice were examined thoroughly using an abnormality screening system referred to as “the Japan Mouse Clinic”. No malformations of the head structure, eye or other parts of the body were apparent in these knockout mice. However, the mutant mice showed a marked increase in serum γ-glutamyl transpeptidase (GGT), a marker for liver damage, but no abnormalities in other liver-related measurements. We also performed a family-based association study on the gene in schizophrenia samples of Japanese origin. We found five single nucleotide polymorphisms (SNPs) located across the gene that showed significant transmission distortion, supporting a prior report of association in a Caucasian cohort. However, the knockout mice showed no behavioral phenotypes relevant to schizophrenia. In conclusion, disruption of the Mrds1/Ofcc1 gene elicits asymptomatic hyper-γ-glutamyl-transpeptidasemia in mice. However, there were no phenotypes to support a role for the gene in the development of eye and craniofacial structures in vertebrates. These results prompt further examination of the gene, including its putative contribution to hyper-γ-glutamyl transpeptidasemia and schizophrenia.


Introduction
Genetic studies have implicated the Mrds1/Ofcc1/Opo gene in multiple developmental processes although its biological function remains largely unknown. Martinez-Morales et al. identified a medaka (a small fish similar to zebrafish) mutant, ojoplano (opo), which is defective in optic cup morphogenesis, and identified the gene responsible for this phenotype. In addition to the defective optic cup, opo mutants also show aberrant craniofacial development and deficits in other structures, including the brain. The gene product of medaka Opo is strikingly similar to that of the mammalian MRDS1/OFCC1/Mrds1/Ofcc1 gene [1], suggesting a function in vertebrate eye and craniofacial development.
This gene has been implicated in two human pathological conditions: orofacial cleft and schizophrenia. Davies et al. [2] reported OFCC1 as a potential orofacial cleft susceptibility gene in humans, based on the observation of three unrelated patients with orofacial cleft, all harboring a chromosomal break within or close to the uncharacterized OFCC1 gene. Schizophrenia is a severe and debilitating mental illness that afflicts approximately 1% of the world population. While the exact genetic mechanisms of disease pathology remain elusive, recent progress in large-scale analyses has isolated a number of potentially predisposing loci or genes. Human chromosome 6p21-24 has repeatedly been identified as a linkage locus for schizophrenia susceptibility [3,4,5].
Straub et al. analyzed an Irish family cohort, and identified the DTNBP1 gene [coding for the dystrobrevin binding protein 1 (dysbindin)] from 6p22.3 as a schizophrenia susceptibility gene [6]. Forty nine families in their sample (18% of total) gave a positive linkage signal at 6p24, when the twenty-seven 6p22.3-linked families (10% of total) were omitted from the analysis. This raises the possibility that the locus harbors more than one schizophrenia susceptibility gene [7]. Follow-up analysis by Straub et al. on 262 Irish schizophrenic pedigrees identified a second gene, MRDS1/ OFCC1 (encoding orofacial cleft 1 candidate 1) gene, located on 6p24.3, as a potentially novel candidate gene for schizophrenia [8,9]. The positive association of schizophrenia with DTNBP1 has been replicated in independent studies [10,11,12,13,14,15,16,17,18,19,20], but association with the MRDS1/OFCC1 gene has not been tested in other ethnic populations.
In this study, we first investigated the relationship between MRDS1/OFCC1 and normal craniofacial and eye development, as well as schizophrenia, by generating and examining mice deficient in the gene corresponding to human MRDS1/OFCC1 and medaka opo. In addition, we tested the hypothesis of genetic association between the MRDS1 and schizophrenia using family-based association analysis, in a Japanese cohort.

Generation of knockout mouse
All animal experiment protocols were approved by the Animal Experiment Committee of RIKEN Brain Science Institute and BioResouce Center. Mice were housed in groups of four or five in standard cages in a temperature and humidity controlled room, with a 12-h light/dark cycle (lights on at 08:00), and had access to standard lab chow and water ad libitum. The animals were used for behavioral tests from 3 to 4 months of age.
Major experimental tasks (design and construction of targeting vector, establishment of targeted ES cells, generation of the chimera and F1 mice, and removal of the neo cassette) to generate the Mrds1/Ofcc1 knockout mice were performed by Unitech, Co. (Kashiwa, Japan). The knockout mice were produced on a background of C57BL/6. The founder mouse (+/Flox [neo+]) was crossed with the FLPe transgenic mouse, producing conditional knockout mice without the neo cassette (+/Flox [neo-]). Deletion of this cassette was confirmed by PCR. In this study, we created mice with a constitutionally deleted (KO) allele by crossing the +/Flox [neo-] mice with the CAG-Cre-Tg mouse, that systemically expresses Cre recombinase, under the control of a CAG promoter [21]. After backcrosses for at least four generations to C57BL/6N Slc (Japan SLC, Shizuoka, Japan), heterozygotes were intercrossed to obtain wild-type (WT), heterozygote and KO mice. Genotyping was performed by PCR with primer sets: (forward 1: 59-CATGTGTACAGAACA-TGTGTGTAGC-39, forward 2: 59-AGCTCCTCTATTCTC-TGGTTGAGG-39, and reverse: 59-AGGTAGAGGCTCCTC TGAGCTTAG-39). Template DNA was extracted from tails or other tissues in some cases, producing 190 bp (forward 1 and reverse primers) and 270 bp (forward 2 and reverse primers) products from the WT and KO alleles, respectively.

Behavioral Analyses
Behavioral analyses were performed as described elsewhere [22,23,24]. All values in the text and figures represent mean 6 SEM (n = 10-15/group). We analyzed the data using Student's ttest, and showed P values for comparisons between knockout animals and their control littermates. P,0.05 was considered significant. ANOVA (analysis of variance) followed by post hoc Tukey multiple comparison test was also used if required.

The Japan Mouse Clinic
The knockout mice were screened for abnormalities using systematic mouse abnormality screening called the 'Japan Mouse Clinic', a research system established and performed at the RIKEN BioResource Center (http://www.brc.riken.jp/lab/jmc/ mouse_clinic/m-strain_jp.html). In this study, we analyzed the heterozygotes obtained by the intercross between heterozygous animals (backcrossed animals) as well as KO animals and WT littermates.

Human DNA sample
All subjects were recruited from a geographic area located in central Japan. The family-based study samples consisted of 124 families with 376 individuals, of whom 163 subjects were schizophrenics and the remainder, unaffected. The types and numbers of families included 80 independent and complete trios (schizophrenic offspring and their parents), 15 probands with one parent, 13 probands with affected siblings, and 30 probands with discordant siblings. Best-estimate lifetime diagnosis was made by direct interview with at least two experienced psychiatrists and based on DSM-IV criteria, using all available information from medical records, hospital staff and family informants. Written informed consent was obtained from all participants and the study protocol was approved by the Committee on Ethics at RIKEN.

Genotyping of human subjects and statistical analyses
The genomic structure of human MRDS1/OFCC1 is based on Human Feb. 2009 (GRCh37/hg19) of the UCSC database (http://genome.ucsc.edu/) (Fig. S1). The single nucleotide polymorphism (SNP) search was performed using Entrez SNP from the NCBI database (http://www.ncbi.nlm.nih.gov/SNP/). We selected 32 SNPs within the MRDS1/OFCC1 gene, including SNPs that showed strong evidence of association with schizophrenia in a prior study [8]. Twenty-five SNPs that could be unequivocally genotyped and showed sufficient heterozygosity (minor allele frequency .0.1) in our preliminary examinations were used for further analyses (Table S1 and Figure S1). MRDS1/OFCC1 SNPs span a genomic stretch of 483,168 bp and the average marker-tomarker interval was ,20 kb. These SNPs were selected to cover the MRDS1/OFCC1 region as evenly as possible.
DNA was extracted from whole blood according to a standard protocol. We used Assays-by-Design SNP genotyping products to score SNPs (Applied Biosystems, Carlsbad, CA, USA) (http:// www.appliedbiosystems.com/), based on the TaqMan assay methods. Genotypes were determined using an ABI7900 sequence detection system instrument (Applied Biosystems) and the SDS v2.0 software package (Applied Biosystems).
The normalized LD coefficient D9 [26] and squared correlation coefficient r 2 [27], parameters indicating a degree of LD between markers, were calculated, and haplotype structures were constructed, using HAPLOVIEW version 4.2 (http://www.broad.mit. edu/mpg/haploview/) [28]. Transmission distortions in the family panel were evaluated using the pedigree disequilibrium test (PDT) program, v5.1 (http://www.chg.duke.edu/research/software.html) [29][30]. The PDT program computes two statistical measures, PDT-sum and PDT-ave. Briefly, PDT-sum gives more weight to larger families, whereas PDT-ave places equal weight on all families. The suitability of both statistical methods depends on family structure and genetic models [29].

Isolation of mouse Mrds1/Ofcc1 transcripts
The N-terminal region of the Ofcc1/Mrds1 protein is evolutionarily conserved among vertebrates including human, mouse, turkey and medaka fish ( Figure 1A). However, the genomic structure of mouse Mrds1/Ofcc1 is still unclear, particularly the 39 region. We thus performed 39-rapid amplification of cDNA ends (RACE) analysis with PCR primers placed in the conserved 59 region. Three Mrds1/Ofcc1 isoforms were isolated: variants 1, 2 and 3 ( Figure 1B) with open reading frames (ORF) of 498 bp, 651 bp and 825 bp, respectively. The deduced amino-acid sequences of the 3 variants were not homologous to any known mouse proteins. Importantly, we found that while exons 2 through 8 of Mrds1/Ofcc1 are shared with the mouse Opo gene [31], exons 1 (non-coding) and 9 were not ( Figure 1B), suggesting that Mrds1/Ofcc1 variants 1, 2 and 3 are shorter isoforms of the mouse Opo transcript.
Next, we examined cellular localization of Mrds1/Ofcc1 protein in mouse embryonic fibroblast (MEF) cells transfected with an expression vector for either variant 1 or 2 tagged with enhanced green fluorescent protein (EGFP) at the C-terminus. The variant 1-EGFP construct showed diffuse expression in the cytoplasm, in a partial net-like structure within MEF cells. In contrast, the majority of the variant 2-EGFP protein was localized as dot or vesicle like structures in the cytoplasm. Nuclear localization was also evident in cells with high expression of the variant 2-EGFP construct ( Figure 1C).

Generation of Mrds1/Ofcc1 knockout (KO) mice
We created two different types of KO mouse, systemic and conditional. In the conditional knockout alleles of the gene, exons 3 and 4 of the Mrds1/Ofcc1 gene are flanked by two loxP sequences ( Figure 1B). In this study, we analyzed the (systemic) KO mice, which were created by mating mice with conditional KO alleles with transgenic (Tg) mice expressing Cre recombinase ( Figures 1B  and 2A). The homozygous KO mice were born according to the Mendelian ratio and grew normally to adulthood (data not shown).
RT-PCR analysis using primer pairs designed between exons 3 and 4, and exons 3 and 6 showed a lack of transcripts for variants 1, 2 and 3 in the whole brain and trunk from embryonic day 14.5 (E14.5), KO embryos ( Figure 2B, left and middle panels). As expected, mRNA expression in KO mice was detected with primer sets between exons 18 and 21 ( Figure 2B, right panel). Deletion of exons 3 and 4, in all likelihood created out-of-frame (frame-shifted) transcripts ( Figure 1B), producing biologically inactive proteins in the mutant mice.

No craniofacial abnormalities in the KO mice
No orofacial clefts (cleft lip and/or cleft palate) were found in the greater than 50 examined mutant mice. There were no discernible differences between the typical appearance of WT and KO faces ( Figure 3A). Nor did the head skull structures of the E18.5 KO fetuses differ from those of the control mice ( Figure 3B). The medaka opo mutant, where an ethylnitrosyrea (ENU) -induced mutation disrupts normal splicing of the opo gene, which is a potential ortholog for mouse and human Mrds1/Ofcc1/ MRDS1/OFCC1, develops the ojoplano ('flat eye') phenotype due to defective eye cup morphogenesis [1]. No apparent eye abnormalities were observed in our KO mice (Figure 3 A and C), although we did not perform histological or immunohistochemical examinations. These data indicate that penetrance of the craniofacial abnormality generated by the deficiency of Mrds1/Ofcc1 is, if existent, extremely low in mice.

Genetic association of MRDS1/OFCC1 with schizophrenia, but with no disease related behavioral abnormalities in KO mice
We examined whether the KO mice exhibited behavioral changes related to mental illnesses, including schizophrenia. We used a battery of tests consisting of open field, tail suspension, elevated plus maze, home cage activity, prepulse inhibition (PPI), and fear conditioning tests. Results are summarized in Table 1 and Figure 4. Importantly, we detected no differences between WT and KO in PPI [two-way repeated ANOVA (prepulse level x genotype), genotype effect, male, F (1,42) = 0.761, P = 0.388; female, F (1,44) = 0.102, P = 0.751], a test of sensorimotor gating function ( Figure 4A), although deficits in PPI have been reported in patients with schizophrenia. In addition, tests on our KO mice related to affection and cognition axes detected no behavioral changes which correlate with phenotypes seen in human mental disorders ( Figure 4B and Table 1).
In contrast to results from animal studies, we found an association between schizophrenia and MRDS1/OFCC1 in a Japanese cohort. Five SNPs that span a 197 kb region of the gene (rs7741209-rs855394), showed significant transmission distortion in a family-based association study (lowest P = 0.0082 for rs7741209) ( Figure S1 and Table S1). Pairwise marker-tomarker linkage disequilibrium (LD) and haplotype block structures based on the standard Lewontin D9 statistic and squared correlation coefficient r 2 statistic are shown ( Figure S2, A and B). The SNPs that showed significant association with schizophrenia were not in tight LD to each other ( Figure S2, A and B). This may reflect an allelic heterogeneity of the gene that confers susceptibility to schizophrenia in the Japanese population.

Significant increase in serum c-glutamyl transpeptidase (GGT) levels in KO mice
To obtain insight into the biological function of Mrds1/Ofcc1, we analyzed the KO mice using the 'Japan Mouse Clinic', a comprehensive and systematic abnormality screening system conducted by the RIKEN BioResource Center [32] (http:// www.brc.riken.jp/lab/jmc/mouse_clinic/m-strain_jp.html). This system analyzes mice in a two-step manner, via pipeline 1 and The in-depth screen is performed using 11 methods: rotarod test, auditory brain stem response (ABR), insulin tolerance test (ITT), oral glucose tolerance test (OGTT), lactate measurements in blood, adipokines and clinical biochemical tests (glucose and adipose related), fundoscopy, ERG (electroretinography), blood pressure, body fat percentage, bone mineral density (by DEXA; dualenergy X-ray absorptiometry) and echocardiography. The extensive pipeline 1 screen revealed that two measures were significantly different among the genotypes: c-glutamyl transpeptidase (GGT) levels ( Figure 5A Figure 5A).
GGT is a reliable and established marker for liver function in human clinical practice. However, we did not find any other markers suggesting liver dysfunction in serum or ( Figure 5, B-D) in liver tissue (data not shown). In humans, elevated levels of GGT are seen in the early phase of alcoholic hepatopathy. Additionally, human kidney and pancreas are GGT enriched tissues, but we found no evidence suggesting abnormalities of these two organs in the KO mice (data not shown). Therefore, it remains unclear why the KO mice have notably increased serum levels of GGT. To our knowledge, this is the first report of asymptomatic hyper-cglutamyl transpeptidasemia in mice.
Pipeline 2 is a behavior-oriented screen, consisting of light/dark transition, open-field, home-cage activity, passive avoidance, tail suspension, hot plate and tail flick tests. We detected no significant differences among the three genotypes, WT, HET and KO in these tests. The optional three tests, PPI, object exploration and fear conditioning tests were not performed.

Discussion
The MRDS1/OFCC1/Mrds1/Ofcc1/Opo gene is structurally conserved among vertebrates, suggesting a phylogenically important role for this gene. In this study, we created and analyzed Mrds1/Ofcc1 KO mice, where the gene construct lacks exons 4 and 5, generating a frame-shift mutation in the mRNA and thereby producing a premature termination codon that potentially abolishes the biological activity of the protein. Mice were analyzed thoroughly using amongst other tests, the Japan Mouse Clinic, a systematic abnormality screening system for mutant mice. Despite its genetic involvement in craniofacial and eye development and the pathogenesis of schizophrenia, no discernible abnormalities in mutant mice were detected, except for surprisingly high levels of GGT in the sera. A possible explanation for these results is that an as yet unidentified gene with structural homology to Mrds1/Ofcc1 compensates for the loss of the Mrds1/Ofcc1 at a functional level. This is however unlikely, as an extensive homology search of the database by researchers including ourselves, has failed to identify any gene homologous to Mrds1/Ofcc1 in the mouse genome.
In medaka fish, a hypomorphic mutation in Opo, results in multiple malformations of craniofacial structures, including the eyes. It is possible that the function of mammalian Mrds1/Ofcc1/ MRDS1/OFCC1 differs from that of medaka Opo. It is also possible that the genetic background of mice may influence the penetrance of the predicted eye phenotypes. Alternatively, one or more biologically active, uncharacterized transcripts that do not include exons 4 and 5 may be present. In this study, we detected a transcript containing exons 18 through 21, in the KO mice ( Figure 2B, right panel). This scenario seems unlikely as medaka opo proteins are expressed as two isoforms of 42 and 130 kDa [1], and a hypomorphic mutation of the gene results in a reduction of the two proteins.
As stated earlier, OFCC1 is also a positional candidate gene for orofacial cleft (OFC1, OMIM119530). However, we detected no abnormalities in craniofacial development in the KO mice. Although we cannot rule out the possibility that abrogated MRDS1/OFCC1/Mrds1/Ofcc1 is insufficient to cause craniofacial abnormalities, the simplest explanation for the current observation is that MRDS1/OFCC1 is not a causal gene for orofacial cleft seen in subjects with a chromosomal break at 6p24. Consistent with this idea, is the identification of the other candidate gene for OFC1 locus, TFAP2A [2] which lies proximal to MRDS1/OFCC1. While this gene does not appear to be directly disrupted by chromosomal breaks in the three affected subjects, Tfap2a knockout mice exhibit pleiotropic morphological abnormalities, including orofacial cleft [33]. Interestingly, our KO mice exhibited strikingly high levels of GGT in sera. This phenotype of high but asymptomatic levels of GGT has not been reported previously in mammals. None of the examined schizophrenic offspring showed extremely high levels of GGT in the absence of other markers for tissue damage. Discovery of this asymptomatic, high c-glutamyl transpeptidasemia phenotype would have been difficult, without the availability of the Japan Mouse Clinic, a research tool established by our institute. This discovery could assist in identifying corresponding human subjects in a clinical setting. Examining the relationship between schizophrenia and liver function-related parameters, some patients with schizophrenia manifest asymptomatic hyperbilirubinemia [34]. However, our KO mice showed normal levels of bilirubin. Recent studies have found the relationship GGT levels and cardiovascular diseases independent on the liver function, although we found no sign for cardiovascular defect in the KO mice [35,36].
In this study of KO mice, we were unable to identify any schizophrenia-relevant behavioral abnormalities. It may be interesting to examine how the behavior of the KO mice is modulated when they are placed under stress-loaded conditions such as separation of pups from dams or administrated with psychostimulants like methamphetamine. In the human study, we have found evidence of a genetic association between the MRDS1/ OFCC1 gene and schizophrenia in a Japanese sample. To corroborate this association, further studies using larger sample sizes will be needed.
Finally, future analysis of our Mrds1/Ofcc1-KO mice could be valuable in deciphering the mechanism of high c-glutamyl transpeptidasemia and its possible relationship to schizophrenia. Figure S1 Genomic structure of the MRDS1/OFCC1 gene and gene-centric association analysis. Genomic structure of the MRDS1/OFCC1 gene is shown with chromosomal positions according to the human genome database (http:// genome.ucsc.edu/). The negative logarithm of P value for association is plotted as a function of the order of SNP markers. Statistical analyses were performed using the pedigree disequilibrium test. Red diamond-shaped marks denote SNPs with P,0.05, while blue marks denote SNPs with P.0.05.