Address correspondence and reprint requests to J. de Belleroche, Division of Neuroscience and Mental Health, Faculty of Medicine, Imperial College London, Charing Cross Hospital, Fulham Palace Road, London W6 8RF, UK. E-mail: email@example.com
The aetiology of schizophrenia is complex and the pathological mechanisms involved are still not fully understood. The aim of this project was to gain insight into the underlying molecular changes occurring in schizophrenia through the analysis of gene expression. Using suppression subtractive hybridization to isolate differentially expressed genes in superior temporal cortex (BA22), we detected one prominent sequence with reduced expression in schizophrenia and represented in at least nine clones. This was then selected for further validation. This 190-bp partial transcript showed identity to part of the Dickkopf-3 (Dkk3) gene sequence. Differential expression was initially confirmed in BA22 by slot blot hybridization where expression was decreased by 35% (p < 0.026). These results were further authenticated in a larger panel (12 control and 11 schizophrenia cases) using SYBR Green I real-time quantitative RT–PCR, in which a 41% decrease in expression of Dkk3 mRNA in schizophrenia was obtained (p < 0.012). Furthermore, using in situ hybridization, Dkk3 mRNA was shown to be abundantly expressed in cortical neurones, with prominent expression in layers II/III and V/VI of BA22. Dkk3 belongs to a novel family of Dkk proteins, which have been shown to be potent inhibitors of the neurodevelopmental wingless (Wnt) signalling pathway, and is therefore a putative candidate for further investigation into the aetiology of schizophrenia.
Schizophrenia is a major psychotic illness, which generally occurs in late adolescence or early adulthood with a lifetime risk of 1% in the general population. It is defined by a group of heterogeneous symptoms of varying severity. These include positive symptoms such as delusions, hallucinations (particularly auditory) and thought disorder, and negative symptoms such as lack of emotional response, social withdrawal and speech disorders. Family, twin and adoption studies have identified a genetic component to schizophrenia but there are no genes or biological markers that are consistently and uniquely associated with the illness. In addition, several plausible hypotheses have been proposed to explain the cause of schizophrenia, including neurodevelopmental defects and abnormalities in dopamine, 5-hydroxytryptamine and glutamate neurotransmission (for reviews see, McKenna 1997; Wong and Van Tol 2003).
Brain imaging technologies have been fundamental in identifying some of the most replicable findings in schizophrenia pathology, such as third and lateral ventricular enlargements (Johnstone et al. 1976; Haug 1982) and decreased cortical volumes (Nelson et al. 1998). However, these features are not always disease specific (Roy et al. 1998; Lim et al. 1999) and cannot provide the detailed molecular information that is necessary to understand the underlying pathology of the disease. Indeed, molecular studies have already started to identify the genes that contribute to the aetiology of schizophrenia (see review by Elkin et al. 2004). Recently there has been progress in identifying genes within schizophrenia vulnerability loci such as neuregulin 1 on chromosome 8p (Stefansson et al. 2002), dysbindin on chromosome 6 (Straub et al. 2002) and catechol-O-methyl transferase on chromosome 22 (Herken and Erdal 2001). Furthermore, analysis of gene expression in post-mortem tissue, particularly using microarray techniques, has yielded a number of genes involved in synaptic transmission, neurodevelopment, lipid and myelin function, mitochondrial function and oxidative metabolism that are changed in schizophrenia (Mirnics et al. 2000; Hakak et al. 2001; Vawter et al. 2001; Mimmack et al. 2002; Prabakaran et al. 2004). Association studies of a gene originally identified from microarray studies, RGS4, have provided support for this potential candidate gene (Mirnics et al. 2001; Chowdari et al. 2002). In addition, transgenic studies of positional candidates have substantiated their functional relevance; for example, targeted mutations disrupting the expression of the neuregulin gene display a similar hyperactivity to mouse models of schizophrenia based on impaired glutamate neurotransmission (Stefansson et al. 2002; Elkin et al. 2004).
In order to identify transcripts that are differentially expressed in schizophrenia, we have used the technique of suppression subtractive hybridization (SSH) as our method of choice. This method is based on subtractive hybridization combined with suppression PCR, which identifies differentially expressed genes and also enriches low-abundance genes, equalizing them with higher-abundance genes (Diatchenko et al. 1996). Not only does SSH provide a comprehensive profile of differentially expressed genes but it can also isolate genes that are of low abundance in the mRNA populations, which are not always detected with other methods such as microarrays (Cao et al. 2004).
We have applied SSH to analyse differential gene expression in superior temporal cortex (BA22). Our aim in this study was to focus initially on superior temporal cortex and subsequently extend the findings to other regions implicated in schizophrenia. The temporal lobe is a region that has figured prominently in studies of schizophrenia owing to its importance in auditory and language processing, two processes that are notably abnormal in schizophrenia. Structural neuroimaging and neuropathological studies in this region have generally shown abnormalities in patients with schizophrenia compared with controls (for reviews see, Harrison 1999; Shenton et al. 2001). Furthermore, in one of our previous studies of BA22, we showed a region-specific change in NR1 NMDA subunit mRNA expression which correlated with decline in cognitive function. This effect did not occur in frontal cortex (BA10) (Humphries et al. 1996).
We have detected a differentially expressed transcript in superior temporal cortex (BA22), Dickkopf 3 (Dkk3), which is highly represented among the down-regulated transcripts in schizophrenia and serves an important role in neurodevelopment. In view of the potential importance of Dkk3 and its relevance to schizophrenia, we have completed a preliminary validation of this candidate and confirmed the marked down-regulation in the expression of Dkk3 in schizophrenia. We have also determined the pattern of expression of the Dkk3 gene in human brain.
Materials and methods
Patients and controls
Patients were recruited from a psychiatric nursing home and a mental hospital as part of a prospective study and represent a group of chronic schizophrenic patients. They were all assessed at the time of recruitment (before 1994) and fulfilled Diagnostic and Statistical Manual III-Revised (DSM-IIIR) criteria for chronic residual schizophrenia, which was in use at the time. Most had a DSM IIIR score of 295.62. After consenting to brain donation, symptoms and cognition were assessed and a full history was taken. The causes of death in both the control and schizophrenia group were similar, being mainly coronary artery occlusion, ischaemic heart disease or bronchopneumonia. Case notes were available after death for all patients. Control brain samples were obtained from mentally normal individuals, matched to the patient group for approximate age, duration of agonal status and post-mortem interval (PMI). At autopsy, gross examination of brains revealed no major atrophy. Cases of widespread damage due to stroke were excluded. Histological analysis was also carried out to exclude cases of Alzheimer's disease, Parkinson's disease or multiple sclerosis. Measurement of pH was carried out in tissue homogenates and varied between 5.7 and 6.5 with a mean ± SD of 6.17 ± 0.23 in all cases. Age at onset of schizophrenia ranged between 22 and 44 years, with a mean ± SD of 30.67 ± 8.51 years. Medication ranged between 9 and 400 (mean ± SD 110.4 ± 126.73) mg chlorpromazine equivalents per day. For slot blot analysis, the control group comprised six subjects aged 75.7 ± 16.3 (range 58–94) years and a mean ± SD PMI of 13.8 ± 14.1 (range 9.5–20) h. The schizophrenia group consisted of six subjects aged 80 ± 10.7 (range 61–89) years and a PMI of 8.5 ± 4.8 (range 3.5–16.5) h. For real-time PCR analysis, the control group comprised 12 subjects aged 76.5 ± 16.95 (range 41–94) years and a PMI of 13 ± 4.3 (range 7.5–16.5) h. The schizophrenia group consisted of 11 subjects aged 78.36 ± 10.33 (range 61–91) years and a PMI of 9.45 ± 7.83 (range 3.5–30) h. Patient details are summarized in Table 1. No significant differences in any of the variables were noted between control and schizophrenia samples. Studies were approved by the Riverside Research Ethics Committee and carried out according to approved protocols.
Table 1. Summary of details of patients whose brains were used in this study
Schz, schizophrenic. All patients were included in the real-time SYBR Green I PCR analysis. *Tissue from these patients was also used in the slot blot analysis.
Rapid dissection of tissue blocks (< 1 cm3) was carried out at post-mortem and samples were rapidly frozen using either liquid nitrogen (tissue extracts) or cardice (in situ hybridization). Brodmann area 22 (BA22) was dissected from the lateral surface of the whole brain extending from the temporal pole (BA38) along the superior surface of the temporal lobe. The samples were then stored at − 80°C until required. For in situ hybridization studies, sections were cut (15 µm) on a cryostat (Bright Instruments Co. Ltd., Huntington, UK), thaw mounted on to Superfrost slides (VWR International, Poole, UK) and stored at −80°C until use.
Total RNA was extracted from schizophrenia and normal control samples using RNAzol B™ (Biogenesis Ltd., Poole, UK), a preparation based on the guanidinium thiocyanate method (Chomczynski and Sacchi 1987). RNA concentrations were determined using a GeneQuant™ spectrophotometer (Biochrom Co., Cambridge, UK). Absorbance values were recorded at 260 nm and 280 nm. Readings at 260 nm were used to calculate nucleic acid concentrations assuming that an optical density (OD) of 1 is equivalent to 40 µg/mL RNA. The ratio of OD260/OD280 was used to evaluate the purity of the nucleic acid samples and the quality of the extracted total RNA was determined using agarose gel electrophoresis. Messenger RNA extraction was carried out using a Dynabeads mRNA purification kit (Dynal Biotech, Oslo, Norway), in which oligo dT residues (dT)25 coupled to the surface of the Dynabeads bind mRNA.
SSH involves two rounds of hybridization between the two mRNA populations obtained from control and schizophrenia tissue samples. The mRNA population containing the specific differentially expressed genes is referred to as the tester (schizophrenia) and the reference mRNA is referred to as the driver (control). Initially, the tester and driver mRNAs are converted into cDNA. Adaptors containing long inverted terminal repeats are then ligated on to tester cDNA ends (Lukyanov et al. 1995), and tester and driver cDNA are hybridized removing the common hybridized sequences. The remaining cDNAs represent genes that are expressed in the tester but not the driver (Sagerström et al. 1997) and these represent sequences up-regulated in schizophrenia. These differentially expressed cDNAs are selectively amplified using suppression PCR, in which primers specific to each adaptor sequence are used. The suppression effect is a negative selection for fragments that do not have the same adaptor at both ends. In order to identify sequences that are down-regulated in schizophrenia, the control mRNA was used as the tester sample and schizophrenia mRNA as the driver (reverse subtracted cDNA). SSH was carried out using a PCR-Select™ cDNA Subtraction Kit (BD Biosciences Clontech, Oxford, UK) according to the protocol recommended by the manufacturer. The procedure is described below in brief.
Synthesis of cDNA was carried out using 2 µg Dnase-treated mRNA from each population, 1 µL cDNA synthesis primer (10 µm) and 20 units of avian myeloblastosis virus (AMV) reverse transcriptase. One microlitre of [α32P]dCTP (1 mCi/mL) was added to monitor the progress of cDNA synthesis. Second-strand synthesis was carried out by using a second-strand enzyme cocktail (DNA polymerase I, Rnase H, Escherichia coli DNA ligase) followed by incubation with T4 DNA polymerase. The double-stranded cDNA was then extracted using phenol : chloroform and ethanol precipitation. Short blunt-ended double-stranded cDNA fragments were generated by restriction enzyme digestion using RsaI. Specific adaptors were ligated to two separate aliquots of RsaI-digested tester cDNA; one aliquot was ligated to adaptor-1 and the other was ligated to adaptor-2R using T4 DNA ligase, The first hybridization was carried out by adding 1.5 µL of driver to 1.5 µL of either adaptor-1- or adaptor-2R-ligated tester in 1 µL hybridization buffer. The samples were overlaid with mineral oil and denatured at 98°C for 1.5 min and left to anneal for 8 h at 68°C. After the first hybridization, the two samples were combined and fresh denatured driver was added to further enrich for differentially expressed sequences. The samples were hybridized overnight at 68°C
PCR amplification of differentially expressed transcripts
For each subtraction, two successive PCR reactions were carried out to amplify differentially expressed cDNAs. In the primary PCR, 1 µL diluted subtracted cDNA was mixed with 1 µL PCR primer 1 (10 µm), 0.5 µL dNTP mix (10 mm), 2.5 µL 10 × PCR reaction buffer and 0.5 µL 50 × Advantage cDNA polymerase mix (Advantage cDNA PCR kit; Clontech) in a total volume of 25 µL. Initially, the samples were incubated at 75°C for 5 min to fill in the missing strands of the adaptors. The PCR reactions were then carried out in a Perkin-Elmer Thermal Cycler 480 (Boston, MA, USA) as follows: 94°C for 30 s, 64°C for 30 s and 71°C for 1.5 min for 27 cycles. The amplified products were then diluted 10-fold and 1 µL of each dilution was used for secondary nested PCR amplification. This reaction contained 1 µL template DNA, 2.5 µL 10 × PCR reaction buffer, 1 µL nested PCR primer 1 (10 µm), 1 µL nested PCR primer 2R (10 µm), 0.5 µL dNTP mix (10 mm) and 0. 5 µL 50 × Advantage cDNA polymerase mix in an end volume of 25 µL. The PCR parameters were as follows: 94°C for 30 s, 68°C for 30 s and 72°C for 1.5 min for 12 cycles. The success of the subtraction experiment was confirmed by PCR analysis, which assessed the presence of the glyceraldehyde-3-phosphate dehydrogenase (G3PDH) gene in the subtracted and unsubtracted controls. The analysis of subtraction efficiency was performed on diluted subtracted and unsubtracted secondary PCR products. The PCR reaction mixes were then subjected to 33 cycles of PCR (94°C for 30 s, 60°C for 30 s and 68°C for 2 min). Aliquots were removed after 18, 23, 28 and 33 cycles, and were then analysed on a 2% agarose gel.
Cloning of subtracted cDNA
The cDNA fragments obtained from the forward and reverse subtractions were cloned into a pT-Adv vector. Briefly, 1 µL secondary PCR (∼30 ng) product was ligated into 50 ng vector in a final volume of 10 µL containing 4 U T4 DNA ligase. The ligation mixtures were incubated overnight at 14°C. TOP10F′ competent cells (50 µL/reaction) were thawed on ice and 2 µL 0.5 mβ-mercaptoethanol were added to each tube and gently mixed by stirring with a pipette tip. To each batch of competent cells, 2 µL of the appropriate ligation reaction mixture was added and samples were incubated for 30 min on ice. The cells were then heat-shocked for 30 s in a water bath at 42°C with no mixing or stirring. Some 250 µL SOC medium (2% tryptone, 0.5% yeast extract, 10 mm NaCl, 2.5 mm KCl, 10 mm MgCl2, 10 mm MgSO4 and 20 mm glucose) was added and samples were incubated at 37°C for 1 h at 225 rpm in a rotary shaking incubator. Aliquots (50 and 200 µL) from each transformation were spread on Laura Bertani medium (LB)/5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (xGal)/isopropyl-β-D-thiogalactopyranoside (IPTG) agar plates containing 50 µg/mL kanamycin. The plates were then incubated overnight at 37°C, followed by incubation for 3 h at 4°C to allow further colour development.
DNA extraction of clones of interest
The white colonies, which contain the insert, were grown overnight in 5 mL LB broth medium (1% tryptone, 0.5% yeast extract and 1% NaCl at pH 7.0) with 50 µg/mL kanamycin. The plasmid DNA of interest was extracted using Wizard plus SV minipreps DNA purification system (Promega, Madison, WI, USA). Plasmid DNA was further purified by centrifugation column chromatography before elution with sterile distilled water. The purified plasmid DNA was stored at − 20°C until required
Automated fluorescent DNA sequencing
Sequencing reaction mixtures were composed of 50 ng PCR product or up to 1 µg plasmid DNA, 4 µL Big Dye enzyme mix (Perkin Elmer) and 6.4 pmol primer made up to a final volume of 10 µL with sterile distilled water. Cycling conditions were as follows: 96°C for 10 s, 50°C for 5 s, and 60°C for 4 min for 25–35 cycles. The sequences obtained were analysed using EditView (ABI Automated Sequence Viewer, version 1.0.1, Applied Biosystems, Warrington, UK). Sequence alignments were analysed using the AssemblyLign™ program (Oxford Molecular Genetics, Oxford, UK). All sequences obtained were identified by searching for known homologous sequences and expressed sequence tags present in the National Centre for Biotechnology Information (NCBI) database.
Quantification of mRNA by slot blot hybridization
Total RNA samples were resuspended in diethyl pyrocarbonate distilled water (DEPC-dH2O) to give a concentration of 40 µg per 10 µL, which was then serially diluted and blotted at 30–10 µg for each sample as described previously (Virgo et al. 1995; Humphries et al. 1996). To each sample, 200 µL slot blot buffer [7.5 mL 20 × saline sodium citrate buffer (SSC), 7.5 mL DEPC-dH2O and 5 mL formaldehyde] was added and RNA was denatured by incubation at 65°C for 10 min. The RNA samples were then loaded on to a slot blot apparatus and drawn through wells under vacuum on to a nylon transfer membrane (Hybond-NX; Amersham Biosciences UK Ltd., Little Chalfont, UK) that had been prewetted with 10 × SSC solution. The RNA was then cross-linked on to the membrane by exposure to UV light for 2 min. The RNA slot blot filters were preincubated for 3–4 h in Rapid Hyb hybridization buffer (Amersham Pharmacia Biotech, Piscataway, NJ, USA) at 65°C. Radioactively labelled probes were added to the prehybridized filters and further incubated overnight at 65°C. The filters were then washed twice in 3 × SSC and 0.1% sodium dodecyl sulphate (SDS), once in 1 × SSC/0.1% SDS, once in 0.5 × SSC/0.1% SDS and finally with 0.1 × SSC/0.1% SDS, according to the required stringency. All washes were carried out for 30 min at 65°C for cDNA probes and 55°C for oligonucleotide probes. The membranes were then exposed to Hyperfilm MP (Amersham Biosciences UK Ltd., Little Chalfont, UK) at − 70°C. The results were quantified using a scanning densitometer (Chromoscan III; Joyce Loebel) and expressed as a ratio of the candidate gene to the internal reference.
DNA probes were labelled using the Mega prime™ DNA labelling kit (Amersham Pharmacia Biotech). Template DNA was used at a starting concentration of 5 ng/µL and 25 ng was used for each labelling reaction. After addition of 5 µL random nanomer primer mix (provided in kit) to the template DNA, the solution was denatured by boiling for 5 min, spun briefly and placed on ice. Some 10 µL labelling buffer (dATP, dGTP and dTTP in Tris-HCl pH 7.5, β-mercaptoethanol and MgCl2), 2 µL Klenow enzyme [1 U/µL DNA polymerase I Klenow fragment (cloned in 100 mm potassium phosphate pH 6.5, 10 mmβ-mercaptoethanol and 50% glycerol] and 2 µL [32P]dCTP (10 mCi/mL) were added to the DNA and primer solution, and the volume adjusted to 50 µL. The mixtures were then incubated at 37°C for a minimum of 30 min. The radioactively labelled DNA probe was purified using ProbeQuant G50 columns (Amersham Biosciences UK Ltd., Little Chalfont, UK).
Quantification of mRNA by real-time PCR analysis
After total RNA extraction, 1 µg total RNA was incubated with 1 µL 1 × Dnase I buffer (Sigma, Poole, UK) and 2 U amplification-grade Dnase I enzyme in a final volume of 10 µL. After incubation for 30 min at 37°C, 1 µL stop solution (50 mm EDTA) was added and samples were then incubated at 70°C for 10 min to inactivate the enzyme. High-temperature RT was carried out using RETROscript first-strand synthesis kit (Ambion, Austin, TX, USA). Some 2 µL random decamers (50 µm) were added to each Dnase-treated sample and incubated at 80°C for 3 min then chilled on ice. This was followed by the addition of 2 µL RT buffer, 4 µL dNTPs (2.5 mm each), 1 µL Rnase inhibitor (10 U/µL) and 1 µL (100 U) moloney murine leukaemia virus (MMLV) reverse transcriptase. All samples were then incubated at 42°C for 1 h, after which the reaction was stopped by incubation at 92°C for 10 min.
Real-time PCR was performed using the Applied Biosystems Prism 7700 sequence detection system (Applied Biosystems, Warrinton, UK). All reactions were performed in a volume of 25 µL containing 1 µL cDNA template (25 ng), 0.5 µm primers, 2 mm MgCl2, nucleotides, TaqDNA polymerase and buffer, all of which were included in the SYBR Green I Master Mix (Applied Biosystems). Each sample was analysed in triplicate and a no-template control was performed for each primer set used. The PCR cycling parameters were: 50°C for 2 min, 95°C for 10 min, and 40 cycles of 94°C for 15 s, 60°C for 1 min, with a final recording step of 78°C for 30 s to further prevent interference from any primer dimer formation. This temperature was chosen above the melting temperature of the primer dimers to ensure that the fluorescence obtained derives from the target amplification product itself. The primers used for RT–PCR were: Dkk3/regulated in glioma (RIG), 5′-CGAACACTGAACTCTACGCC-3′ (forward) and 5′-CAGTTGAAGTGATTTATGCTTGAT-3′ (reverse); Dkk3 (NM_013253), 5′-CAGGCTTCACAGTCTGGTGCTTG-3′ (forward) and 5′-ACATTGTTTCCATCTCCTCCCCTC-3′ (reverse); RIG (NM_006394), 5′-GATGGTGGAGAACAGGATTGGTTTC-3′ (forward) and 5′-GTGCTCAGGAAGCGACATTTGG-3′ (reverse); and β-actin (NM_001101) 5′-CTGGAACGGTGAAGGTGACA-3′ (forward) and 5′-AAGGGACTTCCTGTAACAATGCA-3′ (reverse).
PCR product specificity was confirmed by 2.5% agarose gel electrophoresis to ensure the correct product size was obtained, DNA sequencing to verify the sequence and melting point analysis of PCR products using the Applied Biosystems 7700 real-time instrumentation. Changes in fluorescence were recorded as the temperature was increased from 65°C to 95°C at a rate of 0.2°C/s to obtain a DNA melting curve. The characteristic peak at the melting temperature of the target product distinguishes it from amplification artefacts that melt at lower temperatures in broader peaks or products of the same length with different GC/AT ratios.
DNA sequencing of PCR products was carried out using QIAquick™ PCR purification kits (Qiagen, Valencia, CA, USA) with the forward and reverse primers independently. Sequencing reaction mixtures were composed of 50 ng PCR product, 4 µL Big Dye enzyme mix and 6.4 pmol primer made up to a final volume of 10 µL with sterile dH2O. Cycling conditions were as follows: 96°C for 10 s, 50°C for 5 s and 60°C for 4 min for 25–35 cycles. The sequences obtained were analysed using EditView (ABI Automated Sequence Viewer, version 1.0.1) and identified by searching for known homologous sequences and expressed sequence tags featured in the NCBI database.
The data were analysed using the Sequence Detector software (version 1.7; Applied Biosystems). The raw data obtained were then directly exported into an Excel workbook (Microsoft, Redmond, WA, USA) for analysis of amplification plots and predicted threshold cycle (Ct) values from the exponential phase. Data analysis was simplified by automating all calculations in an Excel workbook entitled Data Analysis for Real-Time PCR (DART) (developed by Peirson et al. 2003). The workbook enables the rapid calculation of Cts, amplification efficiencies and resulting initial fluorescence values (R0) from raw data exported from Sequence Detection System version 1.7. Differences in amplification efficiency were assessed using one-way anova, based upon the null hypotheses: (1) that amplification efficiency was comparable within sample groups (outlier detection) and (2) amplification efficiency was comparable between sample groups (amplification equivalence).
In situ hybridization
Specific antisense oligonucleotides were synthesized (Sigma-Genosys Ltd., Cambridge, UK) for use in the in situ hybridization studies. For detection of human Dkk3 mRNA, four probes (A, B, C and D) were used which were based on the human Dkk3 sequence (Krupnik et al. 1999). Probe A is a 36mer oligonucleotide complementary to nucleotides 384–419, probe B is a 36mer oligonucleotide complementary to nucleotides 486–521, probe C is a 36mer oligonucleotide complementary to nucleotides 1128–1163 and probe D is 33mer oligonucleotide complementary to nucleotides 1503–1535. All four Dkk3 probes yielded identical patterns of expression. The data presented in this paper were obtained using probe A. Oligonucleotide probes were 3′ end-labelled using terminal deoxynucleotidyl transferase (Promega) and 35S-labelled dATP (Amersham Pharmacia Biotech).
Cryostat sections were processed and in situ hybridization carried out as described previously (Akbar et al. 2001). In brief, slides were fixed in 4% paraformaldehyde in phosphate-buffered saline (4°C) for 10 min, rinsed in phosphate-buffered saline twice each for 5 min and then treated with 0.25% acetic anhydride in 0.1 m triethanolamine/0.9% NaCl for 10 min. Following dehydration in increasing concentrations of ethanol, the sections were delipidated in chloroform for 5 min, rinsed in ethanol and allowed to air dry. Sections were then hybridized overnight at 42°C with 2 × 106 cpm labelled probe in 100 µL hybridization buffer [5 × Denhardt's (0.02% bovine serum albumin, 0.02% Ficoll and 0.02% polyvinylpyrrolidone), 4 × SSC, 50% formamide, 10% dextran sulphate, 200 µg/mL polyadenylic acid, 200 µg/mL sheared single-stranded salmon sperm DNA, 50 mm phosphate buffer, 100 µg/mL tRNA and 100 mm dithiothreitol]. The following day, sections were washed stringently (1 × SSC at room temperature (20–22°C) for 30 min, 1 × SSC at 55°C for 30 min, 0.5 × SSC at 55°C for 30 min, 0.1 × SSC at 55°C for 30 min and 0.1 × SSC at room temperature), dehydrated in an ascending series of ethanols and air-dried. When dry, sections were exposed to X-ray film (BioMax MR, Eastman Kodak Co., Rochester, NY, USA) for 21 days and developed. Sections were then dipped in photographic emulsion (Ilford K5, Ilford Imaging UK Ltd., Knutsford, UK) and exposed for 3 weeks at 4°C, developed and counterstained with toluidine blue for light microscopic analyses.
Values obtained were analysed using the Mann–Whitney U-test. Spearman rank correlation tests were performed to determine the effects of patient variables such as age and PMI on mRNA levels. All statistical analyses were performed using Instat 2.01 computer program (Graphpad Software, San Diego, USA).
SSH (Diatchenko et al. 1996) was used to identify differentially expressed gene transcripts between schizophrenia and normal control cases. Messenger RNA was prepared from superior temporal cortex (BA22) tissue samples from age-matched control and schizophrenia cases with a short post-mortem delay (Humphries et al. 1996), purified using oligo dT beads and used for cDNA synthesis. The cDNA was digested using RsaI to generate fragments in the range of 100–1200 bp. Following hybridization between control and schizophrenia cDNA samples, differentially expressed mRNA populations were obtained. A shift in the banding patterns of the subtracted samples and their respective controls indicates a differential expression between samples. This shift is shown in the agarose gel electrophoresis of SSH secondary PCR products (Fig. 1). Common sequences were eliminated during hybridization whereas others were amplified during PCR and thus showed a strong signal.
Differentially regulated PCR products were ligated into a vector and cloned in TOP10F′ cells. Over 150 positive clones representing both up- and down-regulated genes were isolated. Plasmid DNA was extracted from 39 clones representing down-regulated cDNAs. Sequence analysis showed that one down-regulated transcript was highly represented, being present in nine of the 39 cDNAs sequenced. This 190-bp sequence has identity to both Dkk3 [also known as REIC (reduced expression in immortalized cells) and RIG], which is transcribed in the reverse orientation. RIG was first identified as being down-regulated in tumour cells where it is thought to function in tumour suppression at the RNA level (Ligon et al. 1997). Dkk3 is a member of a family of secreted glycoproteins (Dkk1–4), which suppress posterior wingless (Wnt) signalling and are critical for embryonic head development (Fedi et al. 1999; Krupnik et al. 1999). It has been shown in mouse that Dkk1 deletion leads to reduced head development anterior to the midbrain with abnormalities in limb morphogenesis (Mukhopadhyay et al. 2001). Therefore, considering its functional relevance to schizophrenia, we selected this Dkk3 sequence as our first candidate for validation.
Slot blot hybridization
Initially, slot blot hybridization of immobilized mRNA was carried out using a radioactively labelled cDNA probe representing the full 190-bp Dkk3/RIG sequence and the signal obtained was related to the expression of the internal reference gene G3PDH. The levels of expression of Dkk3/RIG mRNA were found to be significantly decreased by 35% in schizophrenia cases compared with controls (p < 0.026) (Fig. 2a). Spearman rank correlation tests did not show any significant effects of age (r = 0.39, p = 0.2), PMI (r = 0.024, p = 0.93) or pH (r = 0.07, p = 0.8) on the levels of expression of Dkk3/RIG mRNA. In addition, there were no significant differences in the levels of Dkk3/RIG mRNA in females compared with males (p > 0.05).
SYBR Green I real-time quantitative RT–PCR (Peirson et al. 2003) was used to further authenticate the decrease in Dkk3 expression in a larger panel of subjects. Three separate quantification experiments were devised using specific primer sets, one for the Dkk3/RIG common sequence, one for a sequence unique to Dkk3 (Fig. 2b) and the third for a sequence unique to RIG. Using the Dkk3/RIG primers, a significant (p < 0.026) decrease of 48% in Dkk3/RIG related to β-actin (internal reference gene) was obtained in the schizophrenia cases relative to controls (Fig. 2c). Similarly, using the Dkk3-specific primers a significant (p < 0.012) decrease of 41% was also obtained (Fig. 2d). Similar significant results were obtained when the signal was related to G3PDH. No significant change was detected using the RIG primers. Quantification of Dkk3 mRNA levels in the frontal cortex (BA10) did not show a significant change in Dkk3 levels, indicating the regional specificity of the effect on Dkk3 (data not shown). Spearman rank correlation tests showed that there were no effects of PMI (Dkk3/RIG: r = 0.1, p = 0.4; Dkk3: r = − 0.37, p = 0.08), age (Dkk3/RIG: r = 0.3, p = 0.1; Dkk3: r = − 0.326, p = 0.13), pH (Dkk3/RIG: r = − 0.02, p = 0.92; Dkk3: r = − 0.23, p = 0.26) and, where data were available, age at onset (Dkk3/RIG: r = 0.2, p = 0.23; Dkk3: r = − 0.326, p = 0.13) and chlorpromazine equivalents (Dkk3/RIG: r = 0.36, p = 0.44; Dkk3: r = − 0.37, p = 0.24).
In situ hybridization studies
Regional localization of Dkk3 mRNA in BA22 was investigated using in situ hybridization (Akbar et al. 2001). Four Dkk3 oligonucleotide probes (probes A, B, C and D) directed against the human sequence (Krupnik et al. 1999) were tested. All four probes produced identical Rnase A-sensitive patterns of expression in brain sections from control patients (data not shown). Probe A was used for further characterization of CNS expression of Dkk3 mRNA in post-mortem brain tissue from both control patients and those with schizophrenia.
Film autoradiograms showed abundant expression of Dkk3 mRNA in cerebral cortex with two broad bands of intense labelling extending across layers II–III and layers V–VI (Fig. 3). In BA22 from control cases, intense labelling for Dkk3 mRNA was seen in layers II and VI, with high to moderate expression levels in layer III and V. A much reduced signal was seen in layers I and IV (Fig. 3). No signal was detected in sections pretreated with Rnase A (Fig. 3). In tissue from schizophrenia cases, there appeared to be an overall reduction in signal intensity compared with controls and this reduction was most marked in layers II and III. Further work is needed to quantify these regional differences.
In entorhinal cortex (from a control case) there appeared to be an overall similar distribution profile for Dkk3 mRNA, with strong labelling of cell clusters in layers II and III, and a more uniform signal across layers V and V1 (Fig. 4a). A similar laminar distribution was shown in frontal cortex (BA10), with the most intensively labelled cell populations seen in layers V and VI, and a less intense diffuse band of labelling in layers II and III (schizophrenia case; Fig. 4b). In hippocampal sections from a schizophrenia case, intense labelling for Dkk3 mRNA was seen in CA3 and dentate gyrus with moderate expression in the CA1 subfield (Fig. 4c).
Emulsion-dipped sections revealed a predominantly neuronal labelling pattern, with grain clusters seen over several cell populations including pyramidal cells (Fig. 5). Although Dkk3 mRNA expression was widespread there were clearly subgroups of cells that did not express this gene (Fig. 5). White matter labelling was not seen.
Schizophrenia is a severe and disabling neuropsychiatric disorder with a complex aetiology including both genetic and environmental components (Lewis and Lieberman 2000). The pathological mechanisms involved are intricate and still not fully understood. The aim of our study was to obtain insight into the underlying molecular changes occurring in schizophrenia through the analysis of differential gene expression in cerebral cortex of schizophrenics compared with normal controls. Using SSH, we carried out both forward and reverse subtraction of mRNA populations from schizophrenia and control cases to obtain genes that were either up- or down-regulated in schizophrenia respectively. We have now isolated over 150 clones and have found one sequence to be most highly represented. It was found in nine of the 39 sequences generated from the reverse direction representing genes that are down-regulated in schizophrenia. In view of its functional relevance in schizophrenia, we have selected this sequence as our first candidate for validation. The 190-bp sequence has identity to both Dkk3 and RIG.
Dkk3 is a member of a family of secreted glycoproteins (Dkk1–4) that suppress posterior Wnt signalling and are critical for embryonic head development (Fedi et al. 1999; Krupnik et al. 1999). It has been shown in mouse that Dkk1 deletion leads to reduced head development anterior to the midbrain with abnormalities in limb morphogenesis (Mukhopadhyay et al. 2001). In mature brain, Dkks are involved in synaptic function and apoptosis (Wodarz and Nusse 1998). Dkk3 is decreased in human immortalized and tumour-derived cell lines, e.g. non-small cell lung carcinomas (Tsuji et al. 2001) where it is thought to function as a tumour suppressor. RIG contains no known protein domains, no RIG protein has been detected and its full-length mRNA lacks a polyadenylation site. It is thought that it may function in tumour suppression at the RNA level as known to occur for H19 RNA (Brannan et al. 1990). The expressed sequence tags (ESTs) previously described as being identical to RIG have now been shown to be derived from the Dkk3 gene (ESTIB666, EST56233, EST48217, EST59418) (Ligon et al. 1997). Therefore, the down-regulated sequence we have isolated is also most likely to be Dkk3.
This study enabled us to obtain a good preliminary validation of the decrease in expression of Dkk3 obtained from the suppression subtractive hybridization study using two separate methodological approaches, slot blot hybridization and real-time PCR. Differential expression was initially confirmed in BA22 by slot blot hybridization where expression was decreased by 35% (p < 0.03). These results were further authenticated in a larger panel (12 control and 11 schizophrenia cases) using SYBR Green 1 real-time quantitative RT–PCR where a 41% decrease in expression of Dkk3 mRNA in schizophrenia was obtained (p < 0.012).
We further characterized the expression of Dkk3 by in situ hybridization and showed for the first time the regional expression profile for Dkk3 mRNA in human superior temporal cortex (BA22), frontal cortex (BA10), entorhinal cortex and hippocampus. A distinct laminar distribution pattern was seen for Dkk3 mRNA, whereby labelling intensity varied according to cortical layer. In BA22, little labelling for Dkk3 was seen in layers I and IV, with moderate labelling in layers III and V, and strong labelling in layers II and VI. This expression profile for DKK3 mRNA was also seen in BA10, whereas that observed in the entorhinal cortex was broadly similar with strong labelling of cell clusters in layers II/III and a more consistent signal across layers V and VI. Emulsion-dipped sections revealed a predominantly neuronal expression profile for Dkk3 mRNA, with labelling of several cell types including pyramidal cells. Light microscopic analyses of emulsion autoradiograms showed that the intense signal for Dkk3 mRNA seen in layers II and VI appears to reflect the packing density of cells in these regions. White matter labelling was not detected. In adult mouse CNS, Dkk3 mRNA appears to be restricted to the cortex and hippocampus (Krupnik et al. 1999). The present study supports the cortical localization of Dkk3 mRNA seen in mouse cortex but provides more detailed information concerning cellular distribution.
Dkk3 mRNA has a distinctive pattern of distribution, being expressed in the major cortical neuronal populations including both pyramidal cells and interneurones. In addition, layer VI neurones, which play a key role in neuronal migration during brain development, were labelled. The high density of Dkk3 expression in superior temporal cortex, together with the reductions seen in patients with schizophrenia, indicate that Dkk3, through its potent actions on the Wnt signalling pathway, may play an important role in the aetiology of schizophrenia.
The Wnt pathway plays a fundamental role in neuronal development through regulation of cell migration and differentiation controlling the initial formation of the neural plate and subsequent patterning decisions, and in later stages regulating the neuronal cytoskeleton and differentiation of synapses in the cerebellum. The effects of extracellular Wnt ligands associating with Frizzled (FDZ) and low-density lipoprotein-related receptors (LRP5/6) in the membrane are transduced to the cytoplasmic protein dishevelled, leading to the stabilization of cytosolic β-catenin, which translocates to the nucleus, binds to T cell factor transcription factors and regulates the expression of target oncogenic and developmental genes. In the absence of Wnt, β-catenin is phosphorylated by glycogen synthase kinase (GSK)-3, which leads to its degradation (Patapoutian and Reichardt 2000).
Dkk proteins bind to the LRP5/6 receptor and prevent the Wnt–Frizzled complex from interacting with LRP5/6 but Dkk3 has not been shown to bind directly to the LRP5/6 receptor (Mao et al. 2001). However, transfection of either Dkk3 or dominant negative LRP5 into osteosarcoma cells significantly reduced β-catenin nuclear translocation in these cells (Hoang et al. 2004). In addition, in PC12 cells co-transfected with LRP5 or LRP6 and Wnt7a, Dkk3 negatively modulated Wnt7a signalling through the canonical Wnt/β-catenin pathway, as reflected by increased transcription of a T cell factor-responsive luciferase reporter gene (Caricasole et al. 2003). Taken together these studies indicate that Dkk3 is able to modulate the Wnt pathway.
There is currently considerable interest in the involvement of the Wnt pathway in schizophrenia. For example, Wnt-1-positive cells (Miyaoka et al. 1999) are significantly increased and β- and γ-catenin levels significantly reduced (Cotter et al. 1998) in hippocampus in schizophrenia cases compared with controls. Furthermore, reduced levels of GSK-3β mRNA have been reported in dorsolateral prefrontal cortex of patients with schizophrenia (Kozlovsky et al. 2004), together with reduced levels of protein and GSK activity (Kozlovsky et al. 2000, 2001). In addition, deletion of the Dishevelled-1 gene in mice leads to reduced social interaction and sensory motor gating abnormalities which model schizophrenia (Lijam et al. 1997). Moreover an association between single nucleotide polymorphisms (SNPs) in the FZD3 locus on 8p21 and schizophrenia has been reported (Yang et al. 2003).
In summary, we have demonstrated the differential expression of Dkk3 in schizophrenia in superior temporal cortex (BA22), which has important implications in the understanding of the aetiology of the disease. However, this is a preliminary study using a small patient sample; further investigations are needed to extend this finding at the mRNA level in a larger cohort and to confirm that levels of Dkk3 are not influenced by variables such as antipsychotic treatment. In addition, it is important to determine whether the mRNA changes are reflected by changes in protein concentration. Nevertheless, our current study has provided strong support for the expression of Dkk3 playing a putative role in schizophrenia pathology. It further implicates the Wnt signalling pathway in schizophrenia pathology and supports the neurodevelopmental hypothesis of schizophrenia.
We are grateful to the Stanley Foundation for funding this project.