The Pharmacogenomics Journal (2009) 9, 103–110 & 2009 Nature Publishing Group All rights reserved 1470-269X/09 $32.00 www.nature.com/tpj ORIGINAL ARTICLE

Further evidence for association of the RGS2 with antipsychotic-induced parkinsonism: protective role of a functional polymorphism in the 30-untranslated region

L Greenbaum1, RC Smith2,3, RGS2 (regulator of G- signaling 2) modulates dopamine receptor 1 4 1 signal transduction. Functional variants in the gene may influence suscepti- A Rigbi , R Strous , O Teltsh , bility to extrapyramidal symptoms (EPS) induced by antipsychotic drugs. To 1 5 K Kanyas , M Korner , further investigate our previous report of association of the RGS2 gene with D Lancet6, E Ben-Asher6 susceptibility to antipsychotic-induced EPS, we performed a replication and B Lerer1 study. EPS were rated in 184 US patients with schizophrenia (115 African Americans, 69 Caucasian) treated for at least a month with typical 1Biological Psychiatry Laboratory, Department of antipsychotic drugs (n ¼ 45), risperidone (n ¼ 46), olanzapine (n ¼ 50) or Psychiatry, Hadassah–Hebrew University Medical clozapine (n ¼ 43). Six single nucleotide polymorphisms (SNPs) within or 2 Center, Jerusalem, Israel; Department of flanking RGS2 were genotyped (rs1933695, rs2179652, rs2746073, rs4606, Psychiatry, New York University Medical School, New York, NY, USA; 3Manhattan Psychiatric rs1819741 and rs1152746). Odds ratios (ORs) and 95% confidence intervals Center, New York, NY, USA; 4Beer Yaakov Mental (CIs) were calculated by logistic regression. Our results indicate association of Health Center, Beer Yaakov, Israel; 5Center for SNP rs4606 with antipsychotic-induced parkinsonism (AIP), as measured by Genomic Technologies, Hebrew University, the Simpson Angus scale, in the overall sample and in the African-American Jerusalem, Israel and 6Department of Molecular Genetics, Weizmann Institute of Science, subsample, the G (minor) allele having a protective effect. ORs for AIP among Rehovot, Israel rs4606 G-allele carriers were 0.23 (95% CI 0.10–0.54, P ¼ 0.001) in the overall sample, and 0.20 (0.07–0.57, P ¼ 0.003) in the African-American Correspondence: subsample. In the previously studied Israeli sample the OR was 0.31 (0.11– Professor B Lerer, Biological Psychiatry 0.84, P ¼ 0.02). We completely sequenced the RGS2 gene in nine patients Laboratory, Department of Psychiatry, Hadassah–Hebrew University Medical Center, with AIP and nine patients without, from the Israeli sample. No common Ein Karem, Jerusalem 91120, Israel. coding polymorphisms or additional regulatory variants were revealed, E-mail: [email protected] suggesting that association of the rs4606 C/G polymorphism with AIP is biologically meaningful and not a consequence of linkage disequilibrium with another functional variant. Taken together, the findings of the current study support the association of RGS2 with AIP and focus on a possible protective effect of the minor G allele of SNP rs4606. This SNP is located in the 30-regulatory region of the gene, and is known to influence RGS2 mRNA levels and protein expression. The Pharmacogenomics Journal (2009) 9, 103–110; doi:10.1038/tpj.2008.6; published online 18 March 2008

Keywords: typical antipsychotics; extrapyramidal symptoms; antipsychotic-induced parkinson- ism; akathisia; regulator of G-protein signaling 2

Received 13 December 2007; revised 22 January 2008; accepted 13 February 2008; published online 18 March 2008 RGS2 and antipsychotic-induced parkinsonism L Greenbaum et al 104

Introduction systems in the central nervous system including dopami- nergic,11 serotonergic,12 cholinergic13 and opioid14 recep- Antipsychotic (neuroleptic) drugs, the mainstay of pharma- tors. Involvement of RGS2 in conditions, such as anxiety15 cological treatment for schizophrenia, are associated with and hypertension16 has been suggested and an association the development of extrapyramidal symptoms (EPS), some of the gene with panic disorder has been reported.17 reversible (acute dystonia, parkinsonism and akathisia) and Since replication studies are essential in the field of some long-lasting (tardive dyskinesia and dystonia).1,2 EPS pharmacogenetics, we report a further evaluation of our are a major problem in the treatment of schizophrenia initial findings concerning the association of RGS2 with AIP because of their negative effect on adherence.3 The majority in a sample of schizophrenia patients from the United States of studies show that atypical, second-generation antipsy- treated with typical and atypical antipsychotics drugs. We chotics are less likely to cause EPS than typical, first- genotyped all six SNPs used in the original study, and generation drugs.4 Antipsychotic-induced parkinsonism performed a cross-sectional analysis of RGS2 as a potential (AIP) is the most frequent manifestation of acute EPS. The candidate gene for AIP. In addition to the 30-UTR SNP incidence of AIP varies from 15 to 40%.5 AIP mimics rs4606, the rs2746073 SNP is intronic, two SNPs are idiopathic Parkinson’s disease symptoms such as bradykine- upstream the RGS2 locus (rs1933695 and rs2179652) and sia and rigidity, but tremor is reported to be less frequent.5 two downstream (rs1819741 and rs1152746), all within About two-thirds of the patients recover within 2 months 20 000 bp flanking region. Furthermore, to determine but some may require over 6 months for their parkinsonism whether functional polymorphisms in RGS2 other than to resolve.1 AIP is also observed as a late onset (tardive) rs4606 might be associated with AIP and whether the manifestation.6 association of rs4606 might be a consequence of linkage Risk factors for AIP include older age, female gender and disequilibrium (LD) with a different functional variant, we the use of antipsychotic drugs that block the dopamine D2 sequenced the entire RGS2 gene in 18 patients from our receptor with a high level of potency.5 Genetic factors may original Jewish-Israeli sample, selected on the basis of their contribute to inter-individual differences in susceptibility.7 AIP phenotype and rs4606 genotype status. Recently, we reported association of the regulator of G-protein signaling 2 gene (RGS2) with susceptibility to AIP Results in patients with schizophrenia treated with antipsychotic 8 drugs. Five of six single nucleotide polymorphisms (SNPs) Clinical variables within or flanking RGS2 were nominally associated with There were no significant differences between PARK þ development or worsening of parkinsonian symptoms in a (n ¼ 141) and PARKÀ patients (n ¼ 43) as regards demo- sample of 115 Jewish, Israeli, acutely psychotic inpatients graphic and clinical data, such as age, sex, ethnic origin with schizophrenia. In the context of a prospective study, (African-American or Caucasian) and type of antipsychotic these patients were treated for at least 2 weeks with typical treatment (Table 1). However, mean Positive and Negative antipsychotic medication, alone or in combination with the Symptoms scale (PANSS) score was significantly higher atypical antipsychotic, risperidone. Of particular interest is among PARK þ than PARKÀ patients (P ¼ 0.00002) and the association of rs4606, a C1114G polymorphism located PARK75% compared to PARKÀ patients (P ¼ 0.001). A similar 0 in the 3 -untranslated region (UTR) of the gene, which difference was observed in the African-American subsample remained significant after correction for multiple testing. (PARK þ vs PARKÀ, P ¼ 0.004; PARK75% vs PARKÀ, The minor G allele of rs4606 was associated with a lesser P ¼ 0.015). propensity toward development or worsening of AIP. Functionality of this polymorphism was demonstrated by SNP genotyping Semplicini et al.,9 who showed that RGS2 expression was Two SNPs (rs1933659 and rs2746073) showed a significant significantly reduced in fibroblasts carrying the rs4606 G (Po0.05) allele frequency difference between African Amer- allele in comparison with those with the CC genotype and G icans and Caucasians and were therefore analyzed in the protein receptor-mediated signaling was significantly in- African-American subsample only and not in the sample as a creased. In our study8 two haplotypes made up of tagging whole. Of the four SNPs that were suitable for association SNPs (including rs4606) within and flanking the gene were testing in the overall sample, one (rs4606) was nominally associated with AIP after correction for multiple testing. The associated with AIP (P ¼ 0.033), the minor (G) allele being haplotype that included the rs4606 G allele was associated more frequent in the PARKÀ compared to the PARK þ group with a lesser susceptibility to development or worsening of (Table 2), as previously reported by Greenbaum et al.8 When AIP. Association of RGS2 with akathisia was also observed comparing the PARKÀ and the PARK75% groups, the but at a lesser level of significance. significance level was stronger for rs4606 (P ¼ 0.016) and RGS2 encodes the RGS2 protein, a member of a large rs1819741 also merged as significant (P ¼ 0.026). Separate protein family defined by a common RGS domain, respon- analyses were performed on the patients of African- sible for regulation of G protein-coupled receptors by American origin, taking into account all six SNPs genotyped. binding to the G-a subunit, stimulating GTPase activity As shown in Table 2, rs4606 was associated with AIP in the and terminating downstream signals.10 RGS2, a 211 amino- African-American subsample when allele frequencies were acid protein, influences several major receptor signaling compared in the PARK þ :PARKÀ and PARK75%:PARKÀ

The Pharmacogenomics Journal RGS2 and antipsychotic-induced parkinsonism L Greenbaum et al 105

Table 1 Demographic and clinical features of patients with (PARK+) or without (PARKÀ) antipsychotic-induced parkinsonism, assessed by the SAS

(A) Whole sample (B) African-American subsample

PARK+ PARKÀ PARK+ PARKÀ

Mean, n (%) s.d. Mean, n (%) s.d. Mean, n (%) s.d. Mean, n (%) s.d.

Number 141 (76.6) 43 (23.4) 86 (74.8) 29 (25.2) Age 40.3 9.8 39.9 9.1 40.29 9.7 41.1 8.8 Male 129 (93.5) 40 (95.2) 82 (95.3) 27 (93.1)

Antipsychotic treatment Typical 37 (26.2) 8 (18.6) 24 (27.9) 7 (24.1) Risperidone 32 (22.7) 14 (32.6) 23 (26.7) 10 (34.5) Olanzapine 38 (27) 12 (27.9) 21 (24.4) 7 (24.1) Clozapine 34 (24.1) 9 (20.9) 18 (21) 5 (17.3)

Rating scales PANSS total 70.7** 17.3 59.9 12.0 68.7* 15.2 59.4 12.4 SAS total 2.70** 2.83 0 0 2.59** 1.81 0 0 BAS total 1.028* 1.649 0.314 1.05 0.687 1.175 0.397 1.256

Ethnicity African-American 86 (61) 29 (67.4) Hispanic 30 (21.3) 11 (25.6) White 25 (17.7) 3 (7) Caucasiana 55 (39) 14 (32.6)

Abbreviations: BAS, Barnes Akathisia scale; PANSS, Positive and Negative Symptoms scale; SAS, Simpson Angus scale.*Po0.01, **Po0.0001. aHispanic+white. groups (P ¼ 0.033, P ¼ 0.031). rs1819741 was significant haplotype analysis across the overall US sample, the African- when comparing the PARKÀ and PARK75% groups American subsample and the Israeli sample,8 it is note- (P ¼ 0.046). None of these comparisons survive Bonferroni worthy that the TGCA haplotype, which has a frequency correction for multiple testing. range of 0.22–0.25, was overrepresented in the PARKÀ group Haplotype analysis was performed encompassing the four across all samples by 10–17%. Results for the CCTA and SNPs that could be analyzed in the overall sample. Two of CCTG haplotypes were less consistent among the three the SNPs (rs4606 and rs1819741) are in an LD block, samples. When haplotypes derived from the nominally according to the confidence interval (CI) algorithm of significant rs4606 SNP and the near-significant rs1819742 Gabriel et al.18 (Figure 1a). The results of the haplotype SNP were examined, the CT haplotype, which has a analysis are shown in Table 3 for the overall sample and the frequency range of 0.69–0.72, was overrepresented in the African-American sample separately; the results from the PARK þ group across all samples by 12–16%. Jewish-Israeli sample8 are added for comparison. As shown Logistic regression analysis, controlling for PANSS scores in Table 3, the haplotype CCTA (made up of rs2179652-C, (Table 4), emphasizes the significant protective effect of the rs4606-C, rs1819741-T and rs1152746-A) was overrepre- rs4606 G allele against AIP in the overall US sample, the sented among PARK þ (PARK þ :PARKÀ ratio, 0.28:0.13, African-American subsample and the Israeli sample P ¼ 0.006, survives Bonferroni correction for multiple test- (although age and gender are well-known risk factors for ing) and PARK75% patients (PARK75%:PARKÀ ratio AIP, they we not included in the regression models because 0.24:0.10, P ¼ 0.016). The TGCA haplotype (made up of there was no significant difference between PARK þ and rs2179652-T, rs4606-G, rs1819741-C and rs1152746-A) was PARKÀ patients regarding these two variables (Table 2)). overrepresented among PARKÀ patients (PARK þ :PARKÀ The analysis revealed that carriers of the rs4606 G allele ratio, 0.20:0.30, P ¼ 0.047; PARK75%:PARKÀ ratio, (as heterozygotes or homozygotes) were 3.2–5 times less 0.16:0.31, P ¼ 0.013). The CCTG haplotype (made up of likely to be in the PARK þ group than CC homozygotes. The rs2179652-C, rs4606-C, rs1819741-T and rs1152746-G) was effect of the TGCA haplotype was similar but weaker, also overrepresented among PARKÀ patients particularly in the Israeli sample. In contrast to AIP, no (PARK þ :PARKÀ ratio, 0.15:0.25, P ¼ 0.024) but the differ- association of SNPs or haplotypes in the RGS2 gene with ence did not emerge as significant for the PARK75%:PARKÀ antipsychotic-induced akathisia, as measured by the Barnes ratio (0.19:0.27, P ¼ 0.194). Comparing the results of the Akathisia scale (BAS), was observed.

The Pharmacogenomics Journal RGS2 and antipsychotic-induced parkinsonism L Greenbaum et al 106

À Sequencing Two rare heterozygous variants were found in gene-coding regions and UTRs in the 18 DNAs sequenced. One was an exon 5-synonymous SNP, G559A (arginine), which had a

0.064 0.051 frequency of 1/18 in PARKÀ patients and 0 in PARK þ 75% vs. PARK patients; the second was a 50-UTR G29A SNP with a À

P genotype (d.f. 2) frequency of 1/18 in PARK þ patients and 0 in PARKÀ

PARK patients. Because of the rarity of these two variants, further genotyping in our sample was not warranted. Three intronic À RGS2 SNPs with minor allele frequency 40.05 were detected in the current screen: rs3053226 (intron 3), rs10598428 (intron 4) and rs17647363 (intron 4). These SNPs are already documented in dbSNP and located in an LD block with 0.026 0.046 rs4606. Since these are intronic SNPs with no putative

P allele (d.f. 1) biological function and to avoid spuriously positive results

PARK 75% vs PARK due to high LD with rs4606, we did not further study their association with AIP. Two rare (1/36 frequency), previously

À unknown, heterozygous intronic SNPs were found as well. À

Discussion PARK+ vs PARK

P genotype (d.f. 2) In the context of a cross-sectional replication study, we 75) vs PARK examined association of the RGS2 gene with EPS (AIP and À À akathisia) in a US sample of mixed ethnicity that included schizophrenic and schizoaffective patients treated for at least 1 month with a typical or atypical antipsychotic drug.

0.033 0.0430.033 0.016 0.037We found 0.021 0.031 that SNP rs4606 0.041 was associated with AIP in the

P allele (d.f. 1) overall sample and in the African-American subsample, the PARK+ vs PARK G allele being protective. This SNP has been shown by another research group to influence RGS2 .9 Another SNP, rs1819741 was associated with susceptibility to the extreme phenotype. Haplotypes made up of SNPs within 75%) and flanking the gene were associated with AIP in the

MinAF (PARK overall sample and in the African-American subsample. Considering the current observations in conjunction with our original report,8 the most consistent finding was for SNP

MinAF rs4606. The G (minor) allele of rs4606 was overrepresented (PARK+) in PARKÀ patients in the Jewish Israeli sample of

) Greenbaum et al.8 and also in the currently examined overall À US sample and African-American subsample, indicating a

MinAF protective effect. Logistic regression analysis, controlling for (PARK illness severity (PANSS total score), showed that carriers of gene, and comparison of allele and genotype frequency in antipsychotic-treated patients with (PARK+) or without the rs4606 G allele (as heterozygotes or homozygotes) were 3.22 (Israeli sample), 4.34 (overall US sample) and 5.0 RGS2 (African-American subsample) times less likely to manifest minor AIP than noncarriers of the G allele. A similar pattern was observed for the TGCA haplotype but the odds ratios (ORs) were slightly weaker; the effect is most likely driven by the rs4606 G allele and the haplotype does not appear to ), and of patients whose SAS score was above the 75th percentile (PARK

À manifest a predictive value greater than the single allele. Subject to further replication, our findings suggest that a priori analysis of genetic variation in RGS2 could contribute to prediction of risk for AIP among schizophrenia Details of SNPs in the patients who are candidates for treatment with typical antipsychotic drugs, in conjunction with demographic and

0.05 in bold font. clinical variables and additional variants in RGS2 and other rs2179652rs4606 191036449rs1819741rs1152746 191051461 191047795 T, 191063528 C T,rs1933659 C C, G A,rs2179652 G 191031443 0.42rs2746073 191036449rs4606 191045850 0.33 G, 0.39rs1819741 0.44 A 0.34 T,rs1152746 191051461 C 191047795 T, 0.24 191063528 A 0.27 0.25 0.45 0.07 T, C C, G 0.37 A, G 0.20 0.24 0.07 0.04 0.31 0.42 0.36 0.43 0.697 0.08 0.30 0.05 0.091 0.22 0.37 0.27 0.116 0.27 0.05 0.633 0.18 0.23 0.351 0.26 0.489 0.169 0.953 0.241 0.059 0.588 0.461 0.643 0.429 0.805 0.613 0.064 0.575 0.596 0.847 0.994 0.562 0.817 0.579 0.698 1.000 0.694 0.793 o parkinsonism (PARK dbSNP no.Overall Position sample Alleles major, African Americans Table 2 Abbreviation: MinAF, minor alleleP frequency. that have still to be identified.

The Pharmacogenomics Journal RGS2 and antipsychotic-induced parkinsonism L Greenbaum et al 107

Figure 1 Linkage disequilibrium (LD) between single nucleotide polymorphisms (SNPs) within and flanking RGS2 (regulator of G-protein signaling 2), in the overall US sample (a) and the African-American subsample (b). LD blocks, defined by confidence intervals (Gabriel et al.18) are marked.

Table 3 Frequency of four SNP haplotypes (derived from rating. The replication study employed a cross-sectional rs2179652, rs4606, rs1819741 and rs1152746) and two SNP design with patients rated for AIP after at least a month of haplotypes (derived from rs4606 and rs1819741) in the RGS2 treatment with a single typical or atypical antipsychotic gene in PARK+ and PARKÀ patients drug. The PARK þ phenotype was defined as any score exceeding 0 on the Simpson Angus scale (SAS). SAS scores Sample Frequency Case:control w2 P-value were substantially higher in the Greenbaum et al.8 study frequency ± ± (PARK+:PARKÀ) (PARK þ¼12.2 3.1, PARKÀ¼13.9 4.7 at baseline) than in the current study (PARK þ¼2.7 þ 2.8, PARKÀ¼0). The CCTA lower severity of parkinsonian symptoms most likely Overall US sample 0.24 0.28:0.13 7.68 0.006* reduced the sensitivity of the current study to detect African Americans 0.25 0.28:0.15 4.23 0.039 significant effects. In this context, the replicated association Israeli sample 0.35 0.36:0.3 0.06 0.805 of SNP rs4606 is noteworthy. The very stringent 0 threshold set for defining the PARKÀ phenotype, necessitated by the TGCA distribution of SAS scores in the sample, allowed the Overall US sample 0.22 0.20:0.30 3.95 0.047 protective effect of the G allele of rs4606 to be demons- African Americans 0.22 0.19:0.32 3.89 0.048 trated. A second limitation of our study is that patients Israeli sample 0.25 0.13:0.30 6.83 0.009 receiving one of four different antipsychotic drugs were studied, only one of the drugs being typical. Since atypical CCTG Overall US sample 0.17 0.15:0.25 5.13 0.024 antipsychotics drugs are associated with a lower rate of EPS African Americans 0.14 0.12:0.18 1.17 0.278 than typical drugs, this explains the low SAS scores that were Israeli sample 0.12 0.23:0.08 8.71 0.003 observed. Nevertheless, the SAS score of the group receiving typical antipsychotics was not significantly higher than that CT of the other treatment groups. Third, the small sample size Overall US sample 0.7 0.73:0.61 4.63 0.031 should be taken into account; this would limit power to African Americans 0.69 0.73:0.57 4.7 0.03 detect differences and reduce the level of significance, Israeli sample 0.72 0.84:0.68 5.81 0.016 contributing to the fact that only one of the nominally Po0.05 in bold font. *Survived Bonferroni correction for multiple testing. significant univariate associations survived stringent Bonferroni correction for multiple testing. At the same time, multivariate analysis of the rs4606 SNP that controlled Several limitations need to be taken into account when for PANSS score showed strong ORs for a protective effect considering the results of this replication study. First, the against AIP and stronger significance levels than the design differs from that of our original study8 that was univariate analyses. The higher PANSS score in the PARK þ prospective. In our original study, patients were treated with group reflects greater severity of illness, which would most typical antipsychotics (with or without risperidone) and AIP likely have resulted in these patients receiving higher doses was rated at baseline and after 2 weeks of treatment. The of antipsychotic drugs. It was important to take this PARK þ phenotype was defined as development or worsen- potentially confounding variable into account, which could ing of parkinsonism from the pre- to the post-treatment not be done in the univariate analyses.

The Pharmacogenomics Journal RGS2 and antipsychotic-induced parkinsonism L Greenbaum et al 108

Table 4 Protective effect of RGS2-rs4606 G allele and RGS2-TGCA haplotype against antipsychotic-induced parkinsonism

PARK+ PARKÀ Odds ratio 1/Odds ratio 95% CI P-value rs4606 G allele carriers (Nagelkerke R2; P-value) Israeli sample (0.11; 0.01) 0.25 0.51 0.31 3.22 0.11–0.84 0.02 Overall US sample (0.20; 0.00003) 0.45 0.67 0.23 4.34 0.10–0.54 0.001 African-American subsample (0.21; 0.0002) 0.43 0.71 0.20 5 0.07–0.57 0.003

TGCA haplotype carriers (Nagelkerke R2; P-value) Israeli sample (0.08; 0.04) 0.27 0.49 0.40 2.5 0.15–1.02 0.057 Overall US sample (0.18; 0.00001) 0.38 0.58 0.32 3.12 0.15–0.70 0.004 African-American subsample (0.20; 0.0002) 0.32 0.59 0.25 4 0.09–0.66 0.005

Analysis controls for PANSS total score.

The SNP most consistently associated with AIP, rs4606, is The pathophysiology of AIP is unclear. Nigrostriatal located in the 30-UTR of the RGS2 gene. Polymorphisms in pathway dopamine D2 receptor occupancy by antipsycho- UTRs may play an important role in posttranscriptional tics is directly related to parkinsonism20 and all clinically regulation, including translation, stability and degradation, effective antipsychotics are D2 receptor blockers.21 D2 subcellular localization and other mechanisms.19 The receptor occupancy of more than 80%, as produced by influence of rs4606 on RGS2 mRNA and protein levels in typical antipsychotics, significantly increases the risk of cultured fibroblasts was described by Semplicini et al.9 RGS2 AIP.5 D2 receptor occupancy produced by atypical antipsy- gene expression was significantly reduced in fibroblasts chotics is less than 80%, depending on specific drug type carrying the G allele in comparison with those with the CC (for example, clozapine exhibits D2 occupancy less than genotype. Furthermore, there was higher angiotensin 70%).21 Other hypotheses of AIP pathophysiology focus on II-stimulated intracellular calcium increase and ERK1/2 the differences between typical and atypical drugs. Faster phosphorylation, which are mediated via a G protein- dissociation from D2 receptors and/or blockade of serotonin coupled receptors, in fibroblasts that carry the G allele. receptors (HT2A) by atypical antipsychotics have also been Both processes were negatively correlated with RGS2 suggested as possible mechanisms.20 RGS2 (a member of the expression.9 Using computer modeling, Semplicini et al.9 R4 subfamily of RGS ) may influence susceptibility found that the predicted secondary structure of the RGS2 to AIP in several ways. RGS2 is reported to modulate D1 and mRNA is influenced by the C/G substitution, by the creation D2 dopamine receptor signal transduction pathways,11,22,23 of an internal loop in the transcript structure that may cause and RGS2 induction by D1 receptor agonist was enhanced in a destabilization of the mRNA molecule. These findings hemiparkinsonian rats.24 Functional variants in the gene indicate that rs4606 is a functional 30-UTR SNP, which may contribute to susceptibility to develop AIP by influen- influences RGS2 mRNA expression and probably protein cing intracellular dopaminergic signaling. Other GPCR levels. systems whose signaling is reported to be influenced by On sequencing gene-coding regions and UTRs two RGS2 may be influenced by functional variation in the gene, previously unknown heterozygous variants were found—a such as M1 and M3 muscarinic receptors.13,25 It is note- 50-UTR SNP in a PARK þ patient and an exon 5-synonymous worthy in this context that administration of muscarinic SNP in a PARKÀ patient. Given their rarity, it is unlikely that antagonists ameliorates AIP.5 RGS2 also plays a role in these variants influence susceptibility to AIP in our sample. serotonergic signaling via 5’HT2A receptors.12 This is Taken together with the functional role rs4606 plays in relevant because blockade of 5’-HT2A receptors is one of regulating RGS2 expression,9 these findings support the the mechanisms thought to underlie the lesser propensity of hypothesis that association of the s4606 C/G polymorphism atypical antipsychotics to induce AIP. Several studies have with AIP is biologically meaningful. On the other hand, as shown significant change in RGS2 mRNA level after typical yet unidentified functional variants in protein-coding or atypical antipsychotic treatment.26,27 Further studies of regions or UTRs may be responsible, by LD, for the observed the interaction of RGS2 with other relevant signaling system trend toward association of rs1819741, which is located genes in the context of AIP are indicated. outside the gene. All intronic RGS2 SNPs with MAF40.05 Notwithstanding the limitations, this replication study detected in the current screen are already documented in modestly supports the findings of our original report8 with dbSNP and located in an LD block with rs4606. However, the regard to two RGS2 SNPs associated with AIP and identifies possibility that they may play a role independent of one of them (rs4606) as protective in all the samples studied. rs4606 in regulating gene expression cannot be entirely These findings render RGS2 a promising candidate gene for excluded. Association of additional variants in RGS2 with AIP. Further studies in larger samples are warranted. The role susceptibility to AIP was observed in our original study8 but of the gene in idiopathic Parkinson’s disease and in other not supported in the current study, possible due to lack of pathologies related to the dopaminergic system, such as power. schizophrenia and ADHD, has yet to be studied.

The Pharmacogenomics Journal RGS2 and antipsychotic-induced parkinsonism L Greenbaum et al 109

Patients and methods containing 0.22 ml20Â assay reagent and 2.5 ml2Â TaqMan Universal PCR Master Mix (Applied Biosystems) in a total Clinical methods volume of 5 ml in 384-well plate. PCR conditions were 2 min Details of the sample and study design are described in detail at 50 1C, 10 min at 90 1C and 45 cycles of 15 s at 95 1C and in a previous publication.28 In brief, this was a cross- 1 min at 60 1C. Real-time PCR was performed and analyzed sectional study of patients with schizophrenia or schizoaf- in an ABI PRISM 7900HT Sequence Detection System fective disorder diagnosed according to Diagnostic and (Applied Biosystems) with the SDS 2.3 software. For the Statistical Manual of Mental Disorders, 4th edn. criteria who purpose of quality control, B10% of the samples were were hospitalized at one of three tertiary care public genotyped twice; the match rate was 99%. hospitals in the United States and had been treated with a single antipsychotic agent (clozapine, olanzapine, risperi- Sequencing done or a first-generation antipsychotic) for at least a To identify additional protein-coding or regulatory variants month. Patients gave written informed consent for partici- that might influence susceptibility to AIP, we completely pation in the study after the purposes and procedures were sequenced the RGS2 gene in nine patients from the original explained. The protocol and consent forms were approved sample of Greenbaum et al.8 who fulfilled criteria for by the internal review board of each institution. Recruit- parkinsonism (PARK þ ) and nine patients who did not ment was consecutive and sampling procedures were (PARKÀ). Eight of the patients were rs4606 G-allele carriers continued until there were approximately 50 patients in (CG/GG genotypes, seven PARKÀ, one PARK þ ) and ten each of the four groups. Clinical state was evaluated by the were rs4606 CC genotype carriers (eight PARK þ , two PANSS,29 parkinsonian symptoms by the SAS30 and akathisia PARKÀ). This design, taking into account both rs4606 allele by the BAS.31 The scales were administered on two separate status and AIP phenotype of the patients, allowed us to occasions, separated by at least a week, by the same specifically search for variants that might be in LD with clinician. The mean score of the two assessments was used rs4606 and account for the protective effect of the G allele in for data analysis. For DNA extraction, 30 cc of blood was PARKÀ G-allele carriers and for susceptibility variants in collected in EDTA tubes. In addition, the patients were PARK þ CC genotype carriers. Primers were constructed evaluated for fasting blood glucose and lipids; the results of using the Primer3 program.32 The RGS2 sequence fragments these assays and their relationship to antipsychotic treat- were amplified by PCR (primers sequences available on ment are reported in Smith et al.28 request), and the amplicons were purified using ExoSAP-IT The overall sample for the current study (clinical ratings (USB Corporation, Cleveland, OH, USA). Then the products and DNA available) consisted of 184 patients of whom 115 were sequenced by an automated sequencer at the Hebrew were African-American, 41 Hispanic and 28 white. Hispanic University Center for Genomic Technologies. The sequence and white patients were combined as a single Caucasian analysis was performed with Sequencher software (Gene group (n ¼ 69). Distribution among the antipsychotic treat- Codes Corporation, USA). ment groups was as follows: typical antipsychotic drugs, n ¼ 45; risperidone, n ¼ 46; olanzapine, n ¼ 50 and clozapine, Data analysis n ¼ 43. No significant difference was seen in mean SAS scores Patients with an SAS score of 0 on two evaluations were among patients in the different antipsychotic groups. grouped as null (PARKÀ) for the AIP phenotype and patients whose average SAS score was above 0 were grouped as Genotyping positive (PARK þ ). To further explore association of the Genomic DNA was extracted from whole blood using the RGS2 gene with AIP we defined the upper quartile of Puregene DNA Purification System (Gentra Systems, MA, patients according to SAS scores as PARK75% and compared USA). Six SNPs within or flanking RGS2 (upstream and them to PARKÀ patients. The same approach was used to downstream) were genotyped: rs1933695, rs2179652, analyze the akathisia phenotype according to the BAS. Allele rs2746073, rs4606, rs1819741 and rs1152746. These are and genotype frequencies were compared in PARKÀ vs identical to the six SNPs that were genotyped and analyzed PARK þ and PARK75% patients and in akathisiaÀ and in our original report of association of the RGS2 and EPS.8 akathisia þ patients by w2-tests. Bonferroni correction for Two SNPs (rs1933695 and rs2746073) that showed a multiple testing was applied. Haploview (version 3.12; Broad statistically significant (Po0.05) allele frequency difference Institute, Cambridge, MA, USA) was used to examine LD between African Americans and Caucasians were excluded between SNPs, to define LD blocks (according to the CI from the analysis of the overall sample but were analyzed in algorithm of Gabriel et al.18), to detect significant departure the African-American subsample. No SNPs showed signifi- from HWE and to perform haplotype population frequency cant deviation from Hardy–Weinberg equilibrium (HWE). estimation. Individual haplotypes were extracted from the SNP genotyping was performed using the TaqMan Assay- population genotype data with the program, PHASE v.2.33,34 On-Demand purchased from Applied Biosystems (Foster ORs and 95% CIs were calculated by logistic regression, City, CA, USA). The assay contains two primers and two controlling for PANSS scores since these were significantly MGB-TaqMan probes. The PCR reaction was performed higher among PARK þ compared to PARKÀ patients. Age, according to the manufacturer’s instructions. In short, gender, ethnicity and drug treatment group did not 10–30 ng of gDNA were added to a reaction mixture significantly influence the model and were not included.

The Pharmacogenomics Journal RGS2 and antipsychotic-induced parkinsonism L Greenbaum et al 110

Acknowledgments 16 Heximer SP, Knutsen RH, Sun X, Kaltenbronn KM, Rhee MH, Peng N et al. Hypertension and prolonged vasoconstrictor signaling in RGS2- This study was supported in part by internal funds from Hadassah deficient mice. J Clin Invest 2003; 111: 445–452. Medical Organisation (to BL) and by an Independent Investigator 17 Leygraf A, Hohoff C, Freitag C, Willis-Owen SA, Krakowitzky P, Fritze J et al. Rgs 2 gene polymorphisms as modulators of anxiety in humans? Grant from Eli Lily (to RCS). J Neural Transm 2006; 113: 1921–1925. 18 Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B Duality of interest et al. The structure of haplotype blocks in the . Science 2002; 296: 2225–2229. 19 Chen JM, Fe´rec C, Cooper DN. A systematic analysis of disease- The authors have no competing interests. associated variants in the 30 regulatory regions of human protein-coding genes I: general principles and overview. Hum Genet 2006; 120: References 1–21. 20 Casey DE. Pathophysiology of antipsychotic drug-induced movement 1 Blanchet PJ. Antipsychotic drug-induced movement disorders. Can disorders. J Clin Psychiatry 2004; 65(Suppl): 25–28. J Neurol Sci 2003; 30(Suppl 1): S101–S107. 21 Miyamoto S, Duncan GE, Marx CE, Lieberman JA. Treatments for 2 Reynolds GP. Receptor mechanisms in the treatment of schizophrenia. schizophrenia: a critical review of pharmacology and mechanisms of J Psychopharmacol 2004; 18: 340–345. action of antipsychotic drugs. Mol Psychiatry 2005; 10: 79–104. 3 Lindenmayer JP, Eerdekens E, Berry SA, Eerdekens M. Safety and efficacy 22 Ho G, Wang Y, Jones PG, Young KH. Activation of serum response of long-acting risperidone in schizophrenia: a 12-week, multicenter, element by D2 dopamine receptor is governed by Gbetagamma- open-label study in stable patients switched from typical and atypical mediated MAPK and Rho pathways and regulated by RGS proteins. oral antipsychotics. J Clin Psychiatry 2004; 65: 1084–1089. Pharmacology 2007; 79: 114–121. 4 Park S, Ross-Degnan D, Adams AS, Sabin J, Kanavos P, Soumerai SB. Effect 23 Stanwood GD, Parlaman JP, Levitt P. Genetic or pharmacological of switching antipsychotics on antiparkinsonian medication use in inactivation of the dopamine D1 receptor differentially alters the schizophrenia: population-based study. Br J Psychiatry 2005; 187: 137–142. expression of regulator of G-protein signalling (RGS) transcripts. Eur J 5 Hirose G. Drug induced parkinsonism: a review. J Neurol 2006; Neurosci 2006; 24: 806–818. 253(Suppl 3): iii22–iii24. 24 Taymans JM, Kia HK, Groenewegen HJ, Leysen JE, Langlois X. Bilateral 6 Lerner V, Libov I, Kapstan A, Miodownik C, Dwolatzky T, Levine J. The control of brain activity by dopamine D1 receptors: evidence from prevalence of neuroleptic drug-induced tardive movement subsyn- induction patterns of regulator of G protein signaling 2 and c-fos mRNA dromes among schizophrenic and schizoaffective patients residing in in D1-challenged hemiparkinsonian rats. Neuroscience 2005; 134: the southern region of Israel. Isr J Psychiatry Relat Sci 2007; 44: 20–28. 643–656. 7 Arranz MJ, de Leon J. Pharmacogenetics and pharmacogenomics of 25 Tovey SC, Willars GB. Single-cell imaging of intracellular Ca2+ and schizophrenia: a review of last decade of research. Mol Psychiatry 2007; activity reveals that RGS 2, 3, and 4 differentially 12: 707–747.Jun 5; e-pub ahead of print. regulate signaling via the Galphaq/11-linked muscarinic M3 receptor. 8 Greenbaum L, Strous RD, Kanyas K, Merbl Y, Horowitz A, Karni O et al. Mol Pharmacol 2004; 66: 1453–1464. Association of the RGS2 gene with extrapyramidal symptoms (EPS) 26 Robinet EA, Geurts M, Maloteaux JM, Pauwels PJ. Chronic treatment induced by treatment with antipsychotic medication. Pharmacogenet with certain antipsychotic drugs preserves upregulation of regulator of Genomics 2007; 17: 519–528. G-protein signalling 2 mRNA in rat striatum as opposed to c-fos mRNA. 9 Semplicini A, Lenzini L, Sartori M, Papparella I, Calo` LA, Pagnin E et al. Neurosci Lett 2001; 307: 45–48. Reduced expression of regulator of G-protein signaling 2 (RGS2) in 27 Fatemi SH, Reutiman TJ, Folsom TD, Bell C, Nos L, Fried P et al. Chronic hypertensive patients increases calcium mobilization and ERK1/2 olanzapine treatment causes differential expression of genes in frontal phosphorylation induced by angiotensin II. J Hypertens 2006; 24: cortex of rats as revealed by DNA microarray technique. Neuropsycho- 1115–1124. pharmacology 2006; 31: 1888–1899. 10 Hollinger S, Hepler JR. Cellular regulation of RGS proteins: modulators 28 Smith R, Lindenmayer J-P, Bark N, Warner-Cohen J, Vaidhyanathaswamy S, and integrators of G protein signaling. Pharmacol Rev 2002; 54:527–559. Khandat A. Clozapine, risperidone, olanzapine, and conventional 11 Taymans JM, Leysen JE, Langlois X. Striatal gene expression of RGS2 and antipsychotic drug effects on glucose, lipids, and leptin in schizophrenic RGS4 is specifically mediated by dopamine D1 and D2 receptors: clues patients. Int J Neuropsychopharmacol 2005; 8: 183–194. for RGS2 and RGS4 functions. J Neurochem 2003; 84: 1118–1127. 29 Kay S, Fiszbein A, Opler LA. The Positive And Negative Syndrome scale 12 Ghavami A, Hunt RA, Olsen MA, Zhang J, Smith DL, Kalgaonkar S et al. for schizophrenia. Schizophr Bull 1987; 13: 261–276. Differential effects of regulator of G protein signaling (RGS) proteins on 30 Simpson G, Angus MP. Scale for assessment extrapyramidal side effects. serotonin 5-HT1A, 5-HT2A, and dopamine D2 receptor-mediated Acta Psychiatr Scand 1970; 212: 11–19. signaling and activity. Cell Signal 2004; 16: 711–721. 31 Barnes TR. A rating scale for drug induced akathisia. Br J Psychiatry 1989; 13 Bernstein LS, Ramineni S, Hague C, Cladman W, Chidiac P, Levey AI 154: 672–676. et al. RGS2 binds directly and selectively to the M1 muscarinic 32 Rozen S, Skaletsky H. Primer3 on the WWW for general users and for acetylcholine receptor third intracellular loop to modulate Gq/11alpha biologist programmers.In: Krawetz S, Misener S (eds). Bioinformatics signaling. J Biol Chem 2004; 279: 21248–21256. Methods and Protocols: Methods in Molecular Biology 2000. Humana 14 Potenza MN, Gold SJ, Roby-Shemkowitz A, Lerner MR, Nestler EJ. Effects Press: Totowa, NJ.pp. 365–386. of regulators of G protein-signaling proteins on the functional response 33 Stephens M, Smith NJ, Donnelly P. A new statistical method for of the mu-opioid receptor in a melanophore-based assay. J Pharmacol haplotype reconstruction from population data. Am J Hum Genet 2001; Exp Ther 1999; 291: 482–491. 68: 978–989. 15 Yalcin B, Willis-Owen SA, Fullerton J, Meesaq A, Deacon RM, Rawlins JN 34 Stephens M, Donnelly P. A comparison of Bayesian methods for et al. Genetic dissection of a behavioral quantitative trait locus shows haplotype reconstruction from population genotype data. Am J Hum that Rgs2 modulates anxiety in mice. Nat Genet 2004; 36: 1197–1202. Genet 2003; 73: 1162–1169.

The Pharmacogenomics Journal