.Nicotinamide-N-methyltransferase (NNMT) in schizophrenia: Genetic association and decreased frontal cortex mRNA levels
Emerging evidence suggests impaired one-carbon metabolism in schizophrenia. Homocysteine is one of the key components of one-carbon metabolism. Elevated plasma homocysteine levels were reported in schizophrenia. A linkage study found that nicotinamide-N-methyltransferase (NNMT), an enzyme involved in one-carbon metabolism, is a determinant of plasma homocysteine levels. In an association study the rs694539 NNMT single nucleotide polymorphism (SNP) was found significantly associated with hyperhomocysteinaemia. Aiming to assess the possible involvement of NNMT in the aetiology of schizophrenia we (1) performed an association study of eight NNMT tagged SNPs in 202 families sharing the same ethnic origin including healthy parents and a schizophrenia proband; (2) assessed NNMT mRNA levels in post-mortem frontal cortex of schizophrenia patients. Genotyping was performed using the ABI SNaPshot and the HRM methods. Individual SNPs and haplotypes were analysed for association using the family-based association test (UNPHASED software). NNMT mRNA levels were measured using RT real-time PCR. In the single SNP analysis, rs694539, previously reported to be associated with hyperhomocysteinaemia, and rs1941404 were significantly associated with schizophrenia (p<0.004 and p=0.033, respectively, following permutation test adjustment). Several haplotypes were also significantly associated with schizophrenia (global p values <0.05 following permutation test adjustment). This is the first study demonstrating an association of NNMT with schizophrenia. Post-mortem frontal cortex NNMT mRNA levels were ~35% lower in schizophrenia patients vs. control subjects. Our study favours the notion that NNMT is involved in the aetiology of schizophrenia.
in schizophrenia: genetic association and
decreased frontal cortex mRNA levels
*, Elad Lerer
*, Madhara Udawela
, Elizabeth Scarr
, Brian Dean
R. H. Belmaker
, Richard Ebstein
and Galila Agam
Department of Clinical Biochemistry, Faculty of Health Sciences, Ben-Gurion University of the Negev, Beer-Sheva, Israel
Psychiatry Research Unit, Faculty of Health Sciences, Ben-Gurion University of the Negev and Mental Health Center,
Department of Human Genetics, Hebrew University, Jerusalem, Israel
Rebecca L. Cooper Research Laboratories, Victorian Brain Bank Network, Mental Health Research Institute, Parkville,
Emerging evidence suggests impaired one-carbon metabolism in schizophrenia. Homocysteine is one of
the key components of one-carbon metabolism. Elevated plasma homocysteine levels were reported in
schizophrenia. A linkage study found that nicotinamide-N-methyltransferase (NNMT), an enzyme
involved in one-carbon metabolism, is a determinant of plasma homocysteine levels. In an association
study the rs694539 NNMT single nucleotide polymorphism (SNP) was found signiﬁcantly associated with
hyperhomocysteinaemia. Aiming to assess the possible involvement of NNMT in the aetiology of
schizophrenia we (1) performed an association study of eight NNMT tagged SNPs in 202 families sharing
the same ethnic origin including healthy parents and a schizophrenia proband; (2) assessed NNMT
mRNA levels in post-mortem frontal cortex of schizophrenia patients. Genotyping was performed using
the ABI SNaPshot and the HRM methods. Individual SNPs and haplotypes were analysed for association
using the family-based association test (UNPHASED software). NNMT mRNA levels were measured
using RT real-time PCR. In the single SNP analysis, rs694539, previously reported to be associated with
hyperhomocysteinaemia, and rs1941404 were signiﬁcantly associated with schizophrenia (p<0.004 and
p=0.033, respectively, following permutation test adjustment). Several haplotypes were also signiﬁcantly
associated with schizophrenia (global p values <0.05 following permutation test adjustment). This is the
ﬁrst study demonstrating an association of NNMT with schizophrenia. Post-mortem frontal cortex NNMT
mRNA levels were y35 % lower in schizophrenia patients vs. control subjects. Our study favours the
notion that NNMT is involved in the aetiology of schizophrenia.
Received 14 September 2010; Reviewed 14 November 2010 ; Revised 22 June 2011; Accepted 22 June 2011 ;
First published online 27 July 2011
Key words: Frontal cortex, genetic association, mRNA levels, nicotinamide-N-methyltransferase
An emerging body of evidence suggests impaired one-
carbon metabolism in schizophrenia (Frankenburg,
2007; Muntjewerﬀ et al. 2006). Homocysteine is one of
the key components of one-carbon metabolism.
Elevated plasma homocysteine levels were reported in
schizophrenia patients, particularly in young males
(Akanji et al. 2007; Applebaum et al. 2004; Levine et al.
2002; Nevo et al. 2006; Regland et al. 1995). Moreover, a
recent study by Kale et al. (2009) found elevated plas-
ma homocysteine levels accompanied by reduced
plasma folate and B
in ﬁrst-episode never-medicated
schizophrenia patients. Brown et al. (2007) found that
elevated maternal third-trimester serum homocysteine
levels were associated with >2-fold increase in
schizophrenia risk in the oﬀspring (Weaver et al. 2007).
A meta-analysis of eight cross-sectional case-control
studies suggested that a 5 m
M increase in homocysteine
Address for correspondence : G. Agam, Ph.D., Department of Clinical
Biochemistry and Psychiatry Research Unit, Faculty of Health
Sciences, Ben-Gurion University of the Negev, PO Box 4600,
Beersheva 84170, Israel.
Tel. : +972-8-6401737 Fax : +972-8-6401740
Email : firstname.lastname@example.org
* These authors contr ibuted equally to this work.
International Journal of Neuropsychopharmacology (2012), 15, 727–737. f CINP 2011
A R T I C L E
levels is associated with a 70% higher risk for schizo-
phrenia (Muntjewerﬀ et al. 2006).
Moderately elevated plasma homocysteine levels in
the range of 7.3–24.4 m
M were found to be associated
with human brain atrophy (Sachdev et al. 2002) and
with neurotoxicity via NMDA receptors dependent on
glycine levels (Lipton et al. 1997). Homocysteine was
reported to accelerate oxidative damage of endothelial
cells (Lentz, 2005) and to induce mitochondrial dys-
function (Zieminska et al. 2006). Since NMDA recep-
tor-involved neurotoxicity (Lau & Zukin, 2007),
elevated brain cell apoptosis (Glantz et al. 2006),
mitochondrial dysfunction (Ben-Shachar, 2002) and
hypoxia (Murray, 1994) have been reported to contrib-
ute to the pathophysiology of schizophrenia, each of
the eﬀects of homocysteine described above, or any
combination of them, may be involved in the aetiology
of the disorder.
Hyperhomocysteinaemia is a complex trait deter-
mined by multiple genetic and environmental factors
(Carr et al. 2009; Malinowska & Chmurzynska, 2009).
Methylenetetrahydrofolate reductase (MTHFR) is one
of the key enzymes of one-carbon metabolism. The
677T allele of MTHFR, associated with elevated plasma
homocysteine (Frosst et al. 1995), was found to be
associated with neuropsychiatric disorders including
schizophrenia (Joober et al. 2000; Regland et al. 1997 ;
Sazci et al. 2003). Another enzyme involved in one-
carbon metabolism is nicotinamide-N-methyltrans-
ferase (NNMT). NNMT methylates nicotinamide
and other pyridine compounds in a reaction that
transfers the methyl group released from S-adenosyl-
methionine (SAM) while the latter is converted into
S-adenosylhomocysteine (SAH) which is further con-
verted into homocysteine (Aksoy et al. 1994). In ad-
dition to being the predominant route of the
metabolism of nicotinamide methylation by NNMT
is also an important conjugation reaction in the bio-
transformation of many drugs and xenobiotics
The NNMT gene is approximately 16.5 kb in length
and consists of three exons and two introns. Fourteen
tagged single nucleotide polymorphisms (SNPs) were
identiﬁed across the NNMT gene. All the SNPs are
located in the non-coding regions of NNMT, but they
may be involved in the regulation of the gene’s ex-
pression. In a genome-wide study in 398 individuals
from 21 extended families, performed by Souto and
colleagues to allocate chromosomal loci linked with
plasma homocysteine levels, the strongest linkage
signal was found on chromosome 11q23 where the
NNMT gene is mapped (Souto et al. 2005). One of ten
SNPs studied (rs694539) and one haplotype were
signiﬁcantly associated with homocysteine levels
(Souto et al. 2005). A further study in healthy Japanese
men found the same SNP to be associated with higher
plasma homocysteine levels among the hyperhomo-
cysteinaemic subgroup (Zhang et al. 2007). Recently,
two studies found an association of polymorphisms
across the NNMT gene with abdominal aortic aneur-
ism (Giusti et al. 2008) and with paediatric acute lym-
phoblastic leukaemia (de Jonge et al. 2009), disorders
known to be associated with hyperhomocysteinaemia.
Parkinson’s disease is also associated with hyper-
homocysteinaemia (Rodriguez-Oroz et al. 2009) and
interestingly, NNMT, abundant in multiple tissues in-
cluding brain (Sano et al. 1992), was found to be over-
expressed in post-mortem cerebella of Parkinson’s
patients (Parsons et al. 2003).
To assess the possible involvement of the NNMT
gene in the aetiology of schizophrenia we (1) per-
formed a comprehensive association study using eight
tagging SNPs spanning across the complete NNMT
gene, including the SNP found to be associated with
hyperhomocysteinaemia by Souto et al. (2005), in
Israeli families including a schizophrenia proband
and healthy parents ; (2) measured NNMT mRNA
levels in post-mortem frontal and parietal cortices of
schizophrenia patients and healthy control subjects;
(3) studied whether NNMT’s expression in post-
mortem frontal cortex of schizophrenia patients and
control subjects is aﬀected by the genotype.
Subjects and methods
Clinical samples for genotyping
For the association study DNA samples from 202 un-
related families including a schizophrenia patient and
her/his parents, all of Israeli origin of the same eth-
nicity, were collected in two independent locations
(Beer-Sheva Mental Health Center, Beer-Sheva and
Emek Medical Center, Afula) after informed consent.
The original protocol was approved by the Helsinki
Committee (Beer-Sheva and Emek). All patients were
interviewed by an experienced psychiatrist using the
structural clinical interview for DSM-IV criteria
(SCID). Diagnosis of schizophrenia was assigned on
the basis of interview and medical records according
to DSM-IV criteria (APA, 1994).
To study whether NNMT’s expression (see below) is
aﬀected by the genotype, post-mortem frontal cortex
[Brodmann’s Area (BA) 9] samples from 13 schizo-
phrenia patients and 13 matched normal controls
obtained from the Victoria Brain Tissue Repository at
the Mental Health Research Institute of Victoria,
728 A. Bromberg et al.
Melbourne, Australia were used. The demographic
data of these samples is summarized in Table 1. Study
of these samples was approved by the IRB committee
of the Mental Health Research Institute of Victoria.
Clinical samples for NNMT mRNA levels
NNMT mRNA levels were measured in the same post-
mortem frontal cortex samples (13 schizophrenia
patients, 13 matched normal controls) used for geno-
DNA from peripheral lymphocytes was extracted
using the phenol procedure. DNA from brain was ex-
tracted using the MasterPure
DNA Puriﬁcation kit
(Epicentre, USA). SNPs were identiﬁed by searching
through the dbSNP public database (http://www.
ncbi.nlm.nih.gov/SNP/). Altogether 14 tagged SNPs
were identiﬁed across the NNMT gene by Haploview
and HapMap data (http://www.hapmap.org) using
the pairwise tagging option with r
=0.8 as a cut-oﬀ
threshold. We chose the threshold of 5 % rare allele
frequency for a SNP to be included in the association
study. Hence, seven SNPs were included in the study.
An additional NNMT SNP (rs694539) reported by
Souto et al. (2005) to be signiﬁcantly associated with
plasma homocysteine levels was also included in the
study. The locations of the eight SNPs studied across
the NNMT gene are summarized in Table 2.
Six of the NNMT SNPs (rs4646335, rs3819100,
rs2301128, rs10891645, rs2155806, rs949374) were
genotyped using the SNaPshot method (Applied
Biosystems, USA). This method relies upon the exten-
sion of a primer immediately adjacent to the SNP
using ﬂuorescently labelled ddNTPs. The ﬂuores-
cently labelled extension primers were visualized
by electrophoresis on a capillary ABI PRISM 310
automated sequencer. Two additional SNPs, rs1941404
and rs694539 were assessed by high-resolution melt
(HRM) analysis, based on ﬂuorescence tracking of
DNA melting across the deﬁned temperature range,
generating melting proﬁles that can be used to identify
the presence of single base variation within the am-
plicon. The pairs of primers used for the SNaPshot
method and the pairs of primers used to amplify the
rs1941404 and rs694539 SNPs using the HRM method
are listed in Table 3.
PCR cycling conditions for the SNaPshot method
were as follows : samples were initially heated at
94 xC for 5 min followed by 35 cycles of 94 xC (30 s),
55 xC (30 s), 72 xC (90 s) and a ﬁnal extension step of
72 xC for 5 min. After the ﬁrst PCR cycle the PCR
product was cleaned with ExoSAP at 37 xC for 30 min
and then at 80 xC for 15 min. The conditions for the
second PCR step were as follows : 96 xC (10 s), 50 xC
(5 s) and 60 xC (30 s) for 25 cycles. The second PCR
product was cleaned using shrimp alkaline phospha-
tase (SAP) initially at 37 xC for 1 h followed by 72 xC
for 15 min.
PCR reactions of the HRM method were performed
using 5 ml Thermo-Start Master Mix (Thermo Fisher
Scientiﬁc Inc., USA), 2 ml primers (2.5 m
M), 1 ml SYTO9
(dye), and 1 ml water in a total volume of 9 ml and 1 ml
genomic DNA. The PCR conditions were as follows:
activating enzyme step at 95 xC for 15 min, 95 xC (5 s),
58 xC (15 s) and 72 xC (10 s) for 45 cycles. The reaction
proceeded to a hold at 40 xC for 2 min, a second hold
at 65 xC for 2 min and then the melt procedure ramped
from 75 xC to 85 xC increasing by 0.05 xC every 3 s.
For samples in which a Mendelian error occurred
genotyping of the entire family was repeated. If the
error persisted the family was not included in the
analysis. The overall Mendelian error rate following
our double-check procedure was <0.5%.
RNA puriﬁcation and cDNA synthesis
RNA was puriﬁed using the RNeasy kit (Qiagen Inc.,
USA). RNA concentration and purity were quantiﬁed
according to absorbance at 260 nm and at 280 nm
(GeneQuant II; Pharmacia Biotech, USA). The samples
were treated with DNase I (RNase free) included in
the kit prior to the synthesis of the cDNA to remove
any genomic DNA. cDNA synthesis was performed
using the Verso cDNA kit (Thermo Fisher Scientiﬁc
Inc.). One mg RNA was reverse-transcribed into cDNA
for 45 min at 42 xC in a ﬁnal volume of 20 ml. To obtain
PCR results within the linear range of detection both
NNMT and b-actin cDNA products were diluted 1 :5.
Measurement of mRNA levels by RT real-time PCR
Real-time quantitative PCR was performed using
Rotor Gene 3000 (Corbett Research, Australia) using
the house-keeping gene b-actin as an internal stan-
dard. b-actin has already been shown to exhibit the
same level of expression in post-mortem brain of
schizophrenia patients and control subjects (Ben-
Shachar & Karry, 2008). The primers for RT–PCR were
designed using primer 3 software (http://frodo.wi.
mit.edu/primer3/) and veriﬁed using BLAST software
to span the intron, to ensure no DNA contamination.
The primers used in the reaction were as follows:
NNMT forward (GTTTGGTTCTAGGCACTCTG) and
NNMT in schizophrenia 729
reverse (GCAGGTTCTGGTCTGAGTAG) ; b-actin for-
ward (TCCCTGGAGAAGAGCTACG) and reverse
Reactions were performed in a ﬁnal volume of 25 ml
consisting of 2 ml puriﬁed cDNA (NNMT or b-actin),
M of each forward and reverse primers and
12.5 ml ABsolute QPCR SYBR Green Mix (ABgene, UK)
providing dNTPs, Taq DNA polymerase and reaction
buﬀer. NNMT and b-actin real-time PCR assays in-
cluded an initial step of 15 min at 95 xC to activate Taq
polymerase, followed by 40 cycles comprised of de-
naturation at 95 xC for 15 s, annealing at 58 xC for 20 s,
and extension at 72 xC for 15 s. The ﬂuorescence of the
accumulating product was acquired at each cycle after
an additional step of 5 s melting at 78 xC.
Quantitation of each sample was obtained by inter-
polation of C
values from the generated standard
curve using Rotor-Gene software. All samples were
run three times. The presented results are the average
of the three runs. Each of the runs included a cali-
bration curve of both b-actin and NNMT using a DNA
pool which included four samples having the highest
concentrations. All assays included a no template
Veriﬁcation of the results of NNMT mRNA levels
using three additional reference genes
Total RNA was isolated from 100 mg frozen tissue
using 1.0 ml TRIzol
reagent (Invitrogen, Australia),
according to the manufacturer’s instructions. The
RNA was treated with DNAse I (Applied Biosystems,
Australia) at 37 xC for 25–30 min, then puriﬁed by
phenol/chloroform extraction and stored at x80 xC.
RNA quantity and quality were determined by
spectrophotometer readings. DNA contamination was
checked by PCR using primers speciﬁc for genomic
First-strand cDNA was synthesized from 2 mg RNA
using 100 U M-MLV-RT (Applied Biosystems) with
M random decamers and 2.5 mM oligo dT primers
(Applied Biosystems), 0.5 m
M of each dNTP and 20 U
RNase inhibitor in 1rRT buﬀer [50 m
(pH 8.3), 75 m
M KCl, 3 mM MgCl
, 5 mM DTT] in a ﬁnal
volume of 20 ml. The reaction was incubated at 44 xC
for 1 h then inactivated at 92 xC and the product stored
at x20 xC.
Real-time reactions were performed in triplicate on
cDNA template with SYBR Green detection using a
Table 1. Post-mortem brain cohort’s demographic data
DoI CoD Neuroleptics
CoDAge Sex PMI Age Sex PMI
1 38 M 36 11 Hanging Flunazepine decanoate (25 mg, 3-weekly) 35 M 35 Coronary artery atheroma
2 35 M 47 17 Perforated gastric ulcer Flunazepine decanoate (50 mg, 3-weekly) 38 M 44 Coronary artery atheroma
3 47 M 41.5 21 Multiple injuries Chlorpromazine (400 mg, daily) 48 M 24 Coronary artery atheroma
4 53 M 43 7 Aspiration/food Haloperidol (150 mg, 2-weekly) 53 M 12 Pulmonary thromboembolism
Chlorpromazine (200 mg, daily)
5 69 M 44.5 47 Ischaemic heart disease Triﬂuoperazine HCl (5 mg, daily) 62 M 66 Acute myocardial infarct
6 61 M 37.5 38 Ischaemic heart disease Flunazepine decanoate (62.5 mg, 2-weekly) 68 M 69 Coronary artery atheroma
7 65 F 50 18 Ruptured abdominal
Flunazepine decanoate (25 mg, 2-weekly),
Haloperidol (5 mg, daily)
66 F 43 Acute myocardial infarct
8 57 M 24 28 Coronary artery atheroma Flunazepine decanoate (18.75 mg, 3-weekly) 57 M 27 Ischaemic heart disease
9 69 M 48 6 Carbon monoxide poisoning Haloperidol (7 mg, daily) 68 M 41 Aortic stenosis
10 23 M 78 5 Multiple injuries Haloperidol (150 mg, 4-weekly) 42 M 51 Exsanguination
11 26 M 52 2 Carbon monoxide poisoning Haloperidol(75 mg, 2-weekly) 25 M 35 Right ventricular hypertrophy
12 41 M 35 15 Hanging Flunazepine decanoate (25 mg, ? weekly) 32 F 56 Coronary artery atheroma
13 47 F 50 20 Pneumonia Risperidone (12 mg, daily) 47 F 24 Pulmonary embolus
PMI, Post-mortem interval (h) ; DoI, duration of illness (yr) ; CoD, cause of death.
730 A. Bromberg et al.
Bio-Rad iQ5 Real-Time PCR detection system (Bio-Rad
Laboratories, USA). The identities of the amplicons
were conﬁrmed by sequencing. Reactions were
performed in 50 ml volume containing cDNA diluted
1: 125, 0.4 n
M primers and 1r IQ SYBR Green super-
mix (Bio-Rad Laboratories), with cycling conditions of
95 xC for 3 min, 40 cycles of 30 s each at 95 xC, 57 xC
and 72 xC, followed by a melt curve. qPCR data
was acquired using IQ5 optical system 2.0 software
values were calculated against a reference
sample. Relative quantity was adjusted for reaction
eﬃciencies calculated from standard curves for each
gene and each plate (85–110% eﬃciency range) using
the Pfaﬄ method (Pfaﬄ, 2001). Ampliﬁcation levels
of the target gene were normalized to the geometric
mean of three reference genes: peptidylprolyl iso-
merase A (cyclophilin A; PPIA), alpha synuclein
Table 2. The position of the studied tagged single nucleotide polymorphisms (SNPs)
number SNP Location Position
distance (bp) p value
rs4646335 A/T utr 113672152 – 0.527
rs3819100 A/G int 1 113672686 53 0.321
rs2301128 G/A int 1 113673209 523 0.583
rs1941404 C/T int 2 113674248 1039 0.033
rs10891645 C/A int 2 113676457 2209 0.199
rs2155806 T/C int 2 113677720 1263 0.867
rs949374 T/G int 2 113678579 859 0.604
rs694539 A/G Upstream to
113638629 49 0.0034
Signiﬁcant p values are in bold.
Table 3. Primers used in the genotyping of NNMT tagged single nucleotide polymorphisms (SNPs)
SNP First PCR primers Second PCR primer extensions
Primers for the SNaPshot method
rs4646335 F: CCTGTCTCTCTGAACTTTGG
rs2301128 F: CTGGATCTGGTGTTCAGTCT
rs10891645 F : GGCTCTGCATATGGTCTATC
rs2155806 F: GGTTCAACTGGGGGAAAT
rs949374 F: GGAGGGAGGGATAAAAACAT
Primers for the HRM method
rs1941404 F: CTCCCTGTGTCTTCCATTAC
Amplicon size 168 bp
rs694539 F: AAGTGCTGACAGGTGATAGG
Amplicon size 187 bp
F, Forward; R, reverse.
NNMT in schizophrenia 731
(SNCA) and glyceraldehyde-3-phosphate dehydro-
genase (GAPDH) (Vandesompele et al. 2002). (See also
pages 40–44 for detailed calculations.)
Reference gene primer sequences:
CycA – F: ATGGTCAACCCCACCGTGTTCTTCG,
CycA – R: CGTGTGAAGTCACCACCCTGACACA,
A-syn – F: CTGCTGCTGAGAAACCAAA,
GAPDH – F: TGCACCACCAACTGCTTAGC,
GAPDH – R: GGCATGGACTGTGGTCATGAG.
We used the logistic-based variant of the transmission
disequilibrium test (TDT) to assess the association
without correcting for a confounding eﬀect of popu-
lation stratiﬁcation. All tests were performed using the
latest version (3.1.3) of the UNPHASED software for
association analysis of multilocus haplotypes derived
from UNPHASED genotype data (http://www.mrc-
To correct for multiple testing we used the permu-
tation test option as provided in UNPHASED.
Student’s t test was used to compare post-mortem
brain NNMT mRNA levels (relative to b-actin) be-
tween patients and controls.
All eight NNMT-tagged SNPs genotyped (Fig. 1) were
in Hardy–Weinberg equilibrium. As presented in
Table 2, single SNP analysis indicated an association
with schizophrenia of two SNPs, rs1941404 in intron 2
(p=0.033) and rs694539, upstream of the NNMT gene,
previously found to be associated with hyperhomo-
cysteinaemia (Souto et al. 2005 ; Zhang et al. 2007)
(p<0.004). We proceeded to multimarker association
analysis using the sliding window approach and
haplotypes comprised of consecutive 2–8 SNPs were
examined. This analysis indicated a signiﬁcant associ-
ation with schizophrenia (adjusted by the permutation
test) for several haplotypes that include one of the
rs1941404 and the rs694539 SNPs or both (Table 4). It
should be noted that since the SNPs analysed were
tagging SNPs selected out of all the SNPs in the NNMT
gene region using Haploview and HapMap data, there
is no linkage disequilibrium (LD) between the SNPs.
Each of the SNPs selected reﬂects all the SNPs be-
longing to its haplotype block. This was further veri-
ﬁed by applying Haploview software and the tagger
algorithim on the results in our population ; no LD
between the genotyped SNPs was revealed. As a
Translation initiation site
5′ intranslated region
Fig. 1. Genomic organization of human NNMT and locations of the studied single nucleotide polymorphisms within its locus.
732 A. Bromberg et al.
signiﬁcant global p value was obtained between
certain haplotypes and schizophrenia we further
examined the association of speciﬁc individual haplo-
types. Individual haplotypes exhibiting signiﬁcant
association with schizophrenia (adjusted by permu-
tation tests) are listed in Table 5.
Brain mRNA levels
Post-mortem frontal cortex NNMT mRNA levels nor-
malized for b-actin were about 35 % lower in schizo-
phrenia patients vs. control subjects [1.3 arbitrary units
S.D.) vs. 2.0¡0.6, p<0.007] (Fig. 2). To verify
these results the samples were measured for NNMT
levels compared to an additional three reference
genes, PPIA, SNCA and GAPDH. The datasets of
NNMT, PPIA, SNCA and GAPDH were shown to be
normally distributed using the D’Agostino & Pearson
omnibus normality test. As shown in Fig. 3 the result
ﬁrst obtained with b-actin as a housekeeping gene was
replicated using the three other normalizing genes.
Twenty-ﬁve percent statistically signiﬁcant decrease
in NNMT mRNA levels was obtained (1.04¡0.59 vs.
NNMT expression and genotype
There was no diﬀerence in frontal cortex NNMT
mRNA levels among genotypes in either of the
diagnosis groups, except for the rs949374 SNP
in schizophrenia patients (ANOVA: F
p=0.047), but this result did not remain signiﬁcant
after correction for multiple testing.
This investigation is the ﬁrst study reporting a genetic
association between the NNMT gene and schizo-
phrenia. Two single SNPs as well as 2–6 window-size
haplotypes across the NNMT gene were signiﬁcantly
associated (following permutation correction for mul-
tiple testing) with schizophrenia. The rs694539 SNP,
located upstream of the NNMT gene, has previously
been found to be associated with elevated plasma
homocysteine levels in 398 Spanish subjects (Souto
et al. 2005) and with higher plasma homocysteine
levels in a hyperhomocysteinaemic subgroup of
healthy Japanese men (Zhang et al. 2007). Earlier
studies carried out in human liver biopsies (Smith et al.
1998; Yan et al. 1999) detected no SNPs or insertion/
deletion events within either the exons or the 5k-
ﬂanking region of NNMT. The authors did ﬁnd eight
SNPs within intron 1, none of which was related to the
level of NNMT activity, but these SNPs were other
than those studied by us. Smith et al. (1998) found
phenotypic diﬀerences in NNMT activity in the general
population but no cDNA transcript diﬀerences among
subjects with low, intermediate and high NNMT
activity. They concluded that the activity diﬀerences
stem from diﬀerences in mRNA steady-state levels.
The authors note that diﬀerences in steady-state
Table 4. Haplotypes of the NNMT gene found to be signiﬁcantly associated with schizophrenia
0.003 (0.009) 2
4 0.0014 (0.0039)
0.0016 (0.002) 5
Both nominal and adjusted signiﬁcance levels are presented. The signiﬁcance levels following the permutation test are shown in
bold within parentheses.
NNMT in schizophrenia 733
mRNA levels may relate to the rate of its synthesis or
its catabolism and suggest that diﬀerences in the
NNMT’s gene transcription rate may be due to dif-
ferences in regulatory regions of the gene resulting in
diﬀerent enzymatic activity.
Plasma homocysteine levels were not available for
the DNA samples of our cohort, but schizophrenia is
strongly associated with hyperhomocysteinaemia
(Akanji et al. 2007; Kale et al. 2009 ; Levine et al. 2005b;
Muntjewerﬀ et al. 2006). Interestingly, two groups
have recently reported association of NNMT with
additional hyperhomocysteinemia-related disorders.
Giusti et al. (2008) found an association between
NNMT and abdominal aortic aneurism and de Jonge
Table 5. Individual NNMT haplotypes signiﬁcantly associated with schizophrenia
Window rs694539 rs4646335 rs3819100 rs2301128 rs1941404 rs10891645 rs2155806 rs949374 x
2 A A 10.7 0.001
G A 12.6 0.0004
T A 4.43 0.035
G C 4.04 0.044
G T 5.76 0.016
C A 10.34 0.0013
3 A A A 11.06 0.0009
G A A 6.03 0.014
G T A 6.0 0.014
G C A T 6.03 0.014
C A 9.51 0.002
4 A A A A 9.38 0.0022
A A A G 6.4 0.011
T G G C 5.84 0.016
G G T C 4.05 0.44
G C A T 4.98 0.026
C A T T 8.59 0.0034
C C C T 5.49 0.019
5 A A A A C 10.62 0.0011
A A A G C 8.42 0.0037
G A G G C 4.29 0.038
G T G G C 4.19 0.040
A G G C C 4.22 0.040
A A C A T 4.08 0.043
A G T A C 4.09 0.043
G C A T T 4.77 0.029
6 A A A A C C 9.86 0.0017
A A A G C C 4.87 0.027
A A A G C A 8.54 0.0035
G A G G C C 4.86 0.027
A A G T A C 4.49 0.03
7 A A A A C C T 10.04 0.0015
A A A G C A T 5.51 0.019
A A A G C C T 5.12 0.024
G A A G T A C 7.29 0.0069
A G G T A T T 4.17 0.03
8 A A A A C C T T 8.86 0.003
A A A G C C T T 4.2 0.04
A A A G T A C G 10.7 0.001
A A G G T C T T 5.05 0.025
G A A G T C T G 3.95 0.047
734 A. Bromberg et al.
et al. (2009) found an association between NNMT and
paediatric acute lymphoblastic leukaemia.
Post-mortem frontal cortex NNMT mRNA levels
were about 30% decreased in schizophrenia patients.
The ﬁnding does not simplistically correlate with hy-
perhomocysteinaemia reported in this disorder. It has
previously been reported that treatment with an
NNMT inhibitor resulted in reduced rather than in-
creased homocysteine release (Riederer et al. 2009).
Since homocysteine cycles through methylation–
demethylation reactions (from homocysteine to meth-
ionine, to SAM, to SAH and back to homocysteine)
it is diﬃcult to predict whether increased or de-
creased NNMT activity would lead to hyperhomo-
The NNMT SNPs included in the haplotypes found
now to be associated with schizophrenia are located
either in introns or upstream of the coding region and
are therefore not translated. However, the possibility
that these SNPs may have a regulatory role in the
transcription of the gene and could account for the
reduced mRNA levels cannot be ruled out. We there-
fore correlated NNMT genotype to its expression in
post-mortem frontal cortex samples. No correlation
was found. This might have resulted from the very
small sample size of the post-mortem samples, poss-
ibly reﬂecting a type II error.
Cerebrospinal ﬂuid (CSF) homocysteine levels are
about 2-fold lower than those in plasma but there are
controversies whether CSF and plasma homocysteine
levels correlate (Levine et al. 2005a ; Obeid et al. 2007 ;
Valentino et al. 2010). Given the neurotoxic eﬀect
of high homocysteine (Bisschops et al. 2004; Lipton
et al. 1997 ; Zieminska et al. 2006) it is of interest that
elevated CSF homocysteine levels have recently
been found in neurodegenerative diseases such as
Alzheimer’s disease (Isobe et al. 2009, 2005),
Parkinson’s disease (Postuma & Lang, 2004) and
amyotrophic lateral sclerosis (ALS) (Valentino et al.
2010). In the ALS study both plasma and CSF homo-
cysteine levels were elevated but no correlation be-
tween the two variables was found. This may suggest
that CSF homocysteine originates directly in the cen-
tral nervous system. If this is the case it is tempting
to speculate that decreased frontal cortex NNMT ex-
pression is related to elevated homocysteine levels in
this brain region resulting in exacerbated neuronal
The present study suﬀers from two central limita-
tions: (a) our association study cohort consisted of a
relatively small sample size ; (b) the SNPs included in
the haplotypes found to be associated with schizo-
phrenia are located either in introns or upstream of the
coding region and are therefore not translated, ques-
tioning their functional implication.
Our study supports the involvement of NNMT in
the aetiology of schizophrenia but awaits replication
in additional populations. Further investigation of the
relationship between NNMT polymorphisms and ex-
pression and between NNMT expression and plasma
and CSF homocysteine levels are warranted.
The authors thank their colleagues for the collection
of samples: blood DNA samples – Wolfgang Maier
(University of Bonn, Germany), M. Rietschel (Central
Institute of Mental Health, University of Heidelberg,
Mannheim, Germany) and T. Schulze (Department of
Genetic Epidemiology in Psychiatry, Central Insti-
tute of Mental Health, University of Heidelberg,
Mannheim and the Department of Psychiatry, Uni-
versity of Go
ttingen, Germany) ; post-mor-
tem brain samples – B. Dean (Rebecca L. Cooper
mRNA levels (AU)
Fig. 2. Scatter plot demonstrating signiﬁcantly lower frontal
cortex NNMT mRNA levels in schizophrenia patients vs.
controls using b-actin as a normalizing gene. n=13 in each
group (t test, * p<0.007).
Fig. 3. Bar graph demonstrating signiﬁcantly lower frontal
cortex NNMT mRNA levels in schizophrenia patients
vs. controls using PPIA, SNCA and GAPDH as normalizing
genes. t test, * p=0.04. &, Controls; %, schizophrenia.
NNMT in schizophrenia 735
Research Laboratories at the Mental Health Research
Institute of Victoria, Australia).
Statement of Interest
Akanji AO, Ohaeri JU, Al-Shammri SA, Fatania HR (2007).
Associations of blood homocysteine concentrations in
Arab schizophrenic patients. Clinical Biochemistry 40,
Aksoy S, Szumlanski CL, Weinshilboum RM (1994).
Human liver nicotinamide N-methyltransferase. cDNA
cloning, expression, and biochemical characterization.
Journal of Biological Chemistry 269, 14835–14840.
APA (1994). Diagnostic and Statistical Manual of Mental
Disorders, 4th edn. Washington, DC : American Psychiatric
Applebaum J, Shimon H, Sela BA, Belmaker RH, et al.
(2004). Homocysteine levels in newly admitted
schizophrenic patients. Journal of Psychiatry Research 38,
Ben-Shachar D (2002). Mitochondrial dysfunction in
schizophrenia: a possible linkage to dopamine. Journal of
Neurochemistry 83, 1241–1251.
Ben-Shachar D, Karry R (2008). Neuroanatomical pattern of
mitochondrial complex I pathology varies between
schizophrenia, bipolar disorder and major depression.
PLoS One 3, e3676.
Bisschops RH, van der Graaf Y, Mali WP, van der Grond J
(2004). Elevated levels of plasma homocysteine are
associated with neurotoxicity. Atherosclerosis 174, 87–92.
Brown AS, Bottiglieri T, Schaefer CA, Quesenberry CP,
et al. (2007). Elevated prenatal homocysteine levels as a
risk factor for schizophrenia. Archives of General Psychiatry
Carr DF, Whiteley G, Alﬁrevic A, Pirmohamed M (2009).
Investigation of inter-individual variability of the
one-carbon folate pathway: a bioinformatic and genetic
review. Pharmacogenomics Journal 9, 291–305.
de Jonge R, Tissing WJ, Hooijberg JH, Jansen G, et al.
(2009). Polymorphisms in folate-related genes and risk
of pediatric acute lymphoblastic leukemia. Blood 113,
Frankenburg FR (2007). The role of one-carbon metabolism
in schizophrenia and depression. Harvard Reviews of
Psychiatry 15, 146–160.
Frosst P, Blom HJ, Milos R, Goyette P, et al. (1995).
A candidate genetic risk factor for vascular disease :
a common mutation in methylenetetrahydrofolate
reductase. Nature Genetics 10, 111–113.
Giusti B, Saracini C, Bolli P, Magi A, et al. (2008). Genetic
analysis of 56 polymorphisms in 17 genes involved in
methionine metabolism in patients with abdominal aortic
aneurysm. Journal of Medical Genetics 45, 721–730.
Glantz LA, Gilmore JH, Lieberman JA, Jarskog LF (2006).
Apoptotic mechanisms and the synaptic pathology of
schizophrenia. Schizophrenia Research 81, 47–63.
Isobe C, Abe T, Terayama Y (2009). Homocysteine May
Contribute to Pathogenesis of RNA Damage in Brains
with Alzheimer’s Disease. Neurodegenerative Diseases 6,
Isobe C, Murata T, Sato C, Terayama Y (2005). Increase of
total homocysteine concentration in cerebrospinal ﬂuid in
patients with Alzheimer’s disease and Parkinson’s disease.
Life Sciences 77, 1836–1843.
Joober R, Benkelfat C, Lal S, Bloom D, et al. (2000).
Association between the methylenetetrahydrofolate
reductase 677C–>T missense mutation and schizophrenia.
Molecular Psychiatry 5, 323–326.
Kale A, Naphade N, Sapkale S, Kamaraju M, et al. (2009).
Reduced folic acid, vitamin B(12) and docosahexaenoic
acid and increased homocysteine and cortisol in
never-medicated schizophrenia patients : implications for
altered one-carbon metabolism. Psychiatry Research 175,
Lau CG, Zukin RS (2007). NMDA receptor traﬃcking in
synaptic plasticity and neuropsychiatric disorders. Nature
Reviews Neuroscience 8, 413–426.
Lentz SR (2005). Mechanisms of homocysteine-induced
atherothrombosis. Journal of Thrombosis and Haemostasis
Levine J, Agam G, Sela BA, Garver DL, et al. (2005a). CSF
homocysteine is not elevated in schizophrenia. Journal of
Neural Transmission 112, 297–302.
Levine J, Sela BA, Osher Y, Belmaker RH (2005b). High
homocysteine serum levels in young male schizophrenia
and bipolar patients and in an animal model. Progress in
Neuropsychopharmacology and Biological Psychiatry 29,
Levine J, Stahl Z, Sela BA, Gavendo S, et al. (2002).
Elevated homocysteine levels in young male patients
with schizophrenia. American Journal of Psychiatry 159,
Lipton SA, Kim WK, Choi YB, Kumar S, et al. (1997).
Neurotoxicity associated with dual actions of
homocysteine at the N-methyl-
Proceedings of the National Academy of Sciences USA 94,
Malinowska A, Chmurzynska A (2009). Polymorphism
of genes encoding homocysteine metabolism-related
enzymes and risk for cardiovascular disease. Nutrition
Research 29, 685–695.
Muntjewerﬀ JW, Kahn RS, Blom HJ, den Heijer M (2006).
Homocysteine, methylenetetrahydrofolate reductase and
risk of schizophrenia: a meta-analysis. Molecular
Psychiatry 11, 143–149.
Murray RM (1994). Neurodevelopmental schizophrenia:
the rediscovery of dementia praecox. British Journal of
Psychiatry 25 (Suppl.), 6–12.
Nevo GA, Meged S, Sela BA, Hanoch-Levi A, et al. (2006).
Homocysteine levels in adolescent schizophrenia patients.
European Neuropsychopharmacology 16, 588–591.
736 A. Bromberg et al.
Obeid R, Kostopoulos P, Knapp JP, Kasoha M, et al. (2007).
Biomarkers of folate and vitamin B
are related in blood
and cerebrospinal ﬂuid. Clinical Chemistry 53, 326–333.
Parsons RB, Smith SW, Waring RH, Williams AC, et al.
(2003). High expression of nicotinamide
N-methyltransferase in patients with idiopathic
Parkinson’s disease. Neuroscience Letters 342, 13–16.
Pfaﬄ MW (2001). A new mathematical model for relative
quantiﬁcation in real-time RT-PCR. Nucleic Acids Research
Postuma RB, Lang AE (2004). Homocysteine and levodopa:
should Parkinson disease patients receive preventative
therapy? Neurology 63, 886–891.
Regland B, Germgard T, Gottfries CG, Grenfeldt B, et al.
(1997). Homozygous thermolabile
methylenetetrahydrofolate reductase in schizophrenia-like
psychosis. Journal of Neural Transmission 104, 931–941.
Regland B, Johansson BV, Grenfeldt B, Hjelmgren LT, et al.
(1995). Homocysteinemia is a common feature of
schizophrenia. Journal of Neural Transmission General Section
Riederer M, Erwa W, Zimmermann R, Frank S, et al. (2009).
Adipose tissue as a source of nicotinamide N-
methyltransferase and homocysteine. Atherosclerosis 204,
Rodriguez-Oroz MC, Lage PM, Sanchez-Mut J, Lamet I,
et al. (2009). Homocysteine and cognitive impairment in
Parkinson’s disease: a biochemical, neuroimaging, and
genetic study. Movement Disorders 24, 1437–1444.
Sachdev PS, Valenzuela M, Wang XL, Looi JC, et al. (2002).
Relationship between plasma homocysteine levels and
brain atrophy in healthy elderly individuals. Neurology 58,
Sano A, Endo N, Takitani S (1992). Fluorometric assay of
rat tissue N-methyltransferases with nicotinamide and
four isomeric methylnicotinamides. Chemical Pharmacology
Bulletin (Tokyo) 40, 153–156.
Sazci A, Ergul E, Guzelhan Y, Kaya G, et al. (2003).
Methylenetetrahydrofolate reductase gene polymorphisms
in patients with schizophrenia. Brain Research. Molecular
Brain Research 117, 104–107.
Smith ML, Burnett D, Bennett P, Waring RH, et al. (1998).
A direct correlation between nicotinamide
N-methyltransferase activity and protein levels in human
liver cytosol. Biochimica Biophysica Acta 1442, 238–244.
Souto JC, Blanco-Vaca F, Soria JM, Buil A, et al. (2005).
A genomewide exploration suggests a new candidate
gene at chromosome 11q23 as the major determinant
of plasma homocysteine levels: results from the
GAIT project. American Journal of Human Genetics 76,
Valentino F, Bivona G, Butera D, Paladino P, et al. (2010).
Elevated cerebrospinal ﬂuid and plasma homocysteine
levels in ALS. European Journal of Neurology 17, 84–89.
Vandesompele J, De Preter K, Pattyn F, Poppe B, et al.
(2002). Accurate normalization of real-time quantitative
RT-PCR data by geometric averaging of multiple internal
control genes. Genome Biology 3, research0034.
Weaver IC, D’Alessio AC, Brown SE, Hellstrom IC, et al.
(2007). The transcription factor nerve growth factor-
inducible protein a mediates epigenetic programming :
altering epigenetic marks by immediate-early genes.
Journal of Neuroscience 27, 1756–1768.
Weinshilboum R (1988). Pharmacogenetics of methylation :
relationship to drug metabolism. Clinical Biochemistry 21,
Yan L, Otterness DM, Weinshilboum RM (1999). Human
nicotinamide N-methyltransferase pharmacogenetics :
gene sequence analysis and promoter characterization.
Pharmacogenetics 9, 307–316.
Zhang L, Miyaki K, Araki J, Nakayama T, et al. (2007). The
relation between nicotinamide N -methyltransferase gene
polymorphism and plasma homocysteine concentration in
healthy Japanese men. Thrombosis Research 121, 55–58.
Zieminska E, Matyja E, Kozlowska H, Staﬁej A, et al. (2006).
Excitotoxic neuronal injury in acute homocysteine
neurotoxicity: role of calcium and mitochondrial
alterations. Neurochemistry International 48, 491–497.
NNMT in schizophrenia 737