MECHANISMS OF DISEASE
Mechanisms of disease
A new NOS2 promoter polymorphism associated with increased
nitric oxide production and protection from severe malaria in
Tanzanian and Kenyan children
Maurine R Hobbs, Venkatachalam Udhayakumar, Marc C Levesque, Jennifer Booth, Jacquelin M Roberts,
Ariana N Tkachuk, Ann Pole, Hilary Coon, Simon Kariuki, Bernard L Nahlen, Esther D Mwaikambo, Altaf L Lal,
Donald L Granger, Nicholas M Anstey, J Brice Weinberg
Summary
Background Nitric oxide (NO) is a mediator of immunity to
malaria, and genetic polymorphisms in the promoter of the
inducible NO synthase gene (NOS2) could modulate
production of NO. We postulated that NOS2 promoter poly-
morphisms would affect resistance to severe malaria.
Methods We assessed genomic DNA from healthy children
and from those diagnosed with malaria from Tanzania (n=47
and n=138, respectively) and Kenya (n=1106) for
polymorphisms by single-stranded conformational poly-
morphism (SSCP) analysis and sequencing. We also
measured in-vivo NO production in Tanzanian children.
Findings We identified a novel single nucleotide
polymorphism, 1173 CT, in the NOS2 promoter that was
significantly associated with protection from symptomatic
malaria (odds ratio 012, 95% CI 003048, p=00006) in
179 Tanzanian children, and significantly associated with
protection from severe malarial anaemia (adjusted relative
risk 025, 95% CI 009066, p=00005) in 1106 Kenyan
children studied over 5 years. The risk of parasitaemia was
not significantly different in wild-type or 1173 CT
individuals. 1173 CT protection in Tanzanians was
independent of the previously recognised NOS2954 GC
polymorphism. The (CCTTT)n NOS2 polymorphism (Tanzania
and Kenya) was not associated with severe malaria
outcomes. 1173 CT was associated with increased
fasting urine and plasma NO metabolite concentrations in
Tanzanian children, suggesting that the polymorphism was
functional in vivo.
Department of Internal Medicine, University of Utah and VA
Medical Centers, Salt Lake City, UT, USA (M R Hobbs PhD,
J Booth BA, A N Tkachuk BA, A Pole BA, Prof D L Granger MD); Division
of Parasitic Diseases, National Center for Infectious Diseases,
Centers for Disease Control and Prevention, Atlanta, GA, USA
(V Udhayakumar PhD, J M Roberts MS, S Kariuki MD, B L Nahlen MD,
Prof A L Lal PhD); Department of Medicine, VA and Duke University
Medical Centers, Durham, NC, USA (M C Levesque MD,
Prof J Brice Weinberg MD); Department of Psychiatry, University of
Utah, Salt Lake City, UT, USA (H Coon PhD); Rollback Malaria
Program, WHO, Geneva, Switzerland (B L Nahlen); Department of
Paediatrics and Child Health, Hubert Kairuki Memorial University,
Dar es Salaam, Tanzania (Prof E D Mwaikambo MD); Division of
Infectious Diseases, Menzies School of Health Research and
Flinders University Northern Territory Clinical School, Darwin, NT
Australia (N M Anstey FRACP)
Correspondence to: Dr J B Weinberg, VA and Duke Medical Centers,
508 Fulton Street, Durham, NC 27705, USA
(e-mail: brice@duke.edu)
1468
For personal use. Only reproduce with permission from The Lancet Publishing Group.
Interpretation The NOS2 promoter 1173 CT single
nucleotide polymorphism is associated with protection
against cerebral malaria and severe malarial anaemia.
Increased NO production in individuals with the 1173 CT
polymorphism lends support to a protective role for NO
against these syndromes. Targeted interventions to increase
NO delivery or production could provide novel preventive and
therapeutic strategies against these major causes of
mortality in African children.
Lancet 2002; 360: 146875
Introduction
The WHO estimates that 1527 million deaths per year
are caused by severe Plasmodium falciparum malaria, with
most deaths occurring in African children younger than
age 5 years.1 The major syndromes of severe malaria in
these children are CEREBRAL MALARIA, SEVERE MALARIAL
ANAEMIA, and respiratory distress or metabolic acidosis.2
Prevalence of cerebral malaria and severe malarial anaemia
varies by geographic region and by intensity of
transmission of malaria.3 Distinct forms of dysregulation of
the immune system seem to give rise to the two
syndromes.4 To better understand the pathophysiology of
severe malaria and the mechanisms of protection from the
disease, we sought to identify genetic traits that confer
protection from, or susceptibility to, malaria and to assess
the function of these genetic traits in acute infection.
NITRIC OXIDE (NO) is toxic to malaria parasites in vitro5
and protects rodents from disease in vivo.6 The induction
and regulation of the human NITRIC OXIDE SYNTHASE TYPE 2
(NOS2) gene is complex and involves sequences up to 16 kb
upstream of the gene.7 Up to five-fold induction of reporter
constructs can be achieved with human NOS2 promoter
sequences, using the proximal 78 kb of sequence (8 kb
to +1 bp)7 in cells stimulated with various cytokines.
Induction and expression of NOS2 varies dependent on
the induction signals used and the cell types studied.8
Two NOS2 promoter polymorphisms, 954 GC and
the (CCTTT)n repeat, have been examined for potential
roles in malarial outcomes (figure 1). The 954 GC
polymorphism has been associated with mild malaria
disease severity in Gabon,9 and a long (CCTTT)n repeat
length (defined as 11 repeats) has been associated with
protection from fatal cerebral malaria in Gambian
children.10 However, in a previous experiment11 we noted
no association between these previously described
polymorphisms and risk of cerebral malaria in Tanzanian
children, or with NO production in vivo.11
In earlier work, in a controlled, prospective study,
Anstey and colleagues12 noted high NO production and
mononuclear cell immunoreactive NOS2 in Tanzanian
malaria-exposed, healthy control children. NO production
and NOS2 expression were significantly lower in children
THE LANCET Vol 360 November 9, 2002 www.thelancet.com

GLOSSARY
Methods
CEREBRAL MALARIA (CM)
Unarousable coma (WHO) with a Blantyre coma score of 2 or less for at
least 30 min after the last convulsion, no other cause of coma evident
from clinical or cerebrospinal fluid analysis, and any level of P falciparum
parasitaemia on thick film examination.
MUTAGENICALLY SEPARATED PCR (MS-PCR)
A method for amplifying DNA fragments with different sequences
(polymorphisms) from the same genomic region with three primers
designed such that one template sequence produces a smaller or larger
amplified product relative to the wild-type product.
NITRIC OXIDE (NO)
NO is a gas that mediates antimicrobial activity, smooth muscle
relaxation, neurotransmission, and modulation of cytokine production
and vascular cell adhesion molecule expression.
NITRIC OXIDE SYNTHASE TYPE 2 (NOS2)
The inducible isoform of nitric oxide synthase gene (NOS type 2) is an
enzyme induced in response to infection and cytokines. NOS2 is
capable of generating large quantities of NO from its substrate
L-arginine.
SEVERE MALARIAL ANAEMIA (SMA)
A haemoglobin concentration of less than 60 g/L and a trophozoite
count greater than 10 000 trophozoites/L of blood, with no other overt
cause.
SINGLE NUCLEOTIDE POLYMORPHISM (SNP)
Single base pairs in genomic DNA at which different sequence
alternatives (alleles) exist in normal individuals in some population(s);
the least frequent allele has an abundance of 1% or greater.
SINGLE-STRANDED CONFORMATIONAL POLYMORPHISM (SSCP)
ANALYSIS
A method for distinguishing between DNA fragments with different
sequences (polymorphisms) amplified from the same genomic region
based on differences in the mobility of the single-stranded DNA during
polyacrylamide gel electrophoresis.
admitted to hospital with uncomplicated malaria or
cerebral malaria, than in controls. This finding suggests
that high concentrations of NO could protect against
cerebral malaria in coastal Tanzania, a region with a high
incidence of the disease. We postulated, therefore, that the
differing degrees of NOS2 expression and NO
concentration in children with different clinical manifes-
tations of malaria might be due to as yet unidentified
genetic differences in the NOS2 gene.
Polymorphism
NFB binding site
DNase 1
hypersensitive site
C/EBP binding site
TATA box
25 kb 1173 CT
16 kb CCTTT repeat 954 GC
Figure 1: Schematic diagram of the first 2500 bp of the NOS2 promoter
The relative positions of the (CCTTT) repeats and the 1173 CT and 954 CG single nucleotide
n
polymorphisms, with respect to the start site of transcription, are shown. Also shown is the TATA box,
and the binding sites for the DNA-binding proteins NF-B and C/EBP, and a DNAse hypersensitive
site in the region.
THE LANCET Vol 360 November 9, 2002 www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
Participants
Between 1992 and 1995, we studied two groups of
patients who had been previously involved in other studies
done by us; one group was from Tanzania12,13 and the
other from Kenya.14
For the Tanzanian participants, we identified inpatients
prospectively recruited from the Muhimbili Medical
Centre, Dar es Salaam, Tanzania, for a study examining
the role of NO in severe and uncomplicated malaria.12 We
recruited children into three groups: healthy controls,
uncomplicated malaria, and cerebral malaria. Controls
were consecutive, asymptomatic, malaria-exposed
children aged 6 months to 9 years, who had no fever or
history of fever within the past 2 weeks, a normal white
blood cell count, and no acute illness. Most were awaiting
elective surgery for non-inflammatory conditions or for
long-bone fractures older than 1 week.12 We defined
uncomplicated clinical malaria as a febrile illness with
P falciparum parasitaemia greater than 10 000 tropho-
zoites/L, no history of convulsions, no other evident
cause of fever, fully alert, normoglycaemia, and no severe
respiratory distress. Cerebral malaria was defined as
unarousable coma (WHO definition) with a Blantyre
coma score of 2 or less for at least 30 min after the last
convulsion, no other cause of coma evident from clinical
or cerebrospinal fluid analyses, and P falciparum
parasitaemia (any amount) on thick film examination. We
did analyses on whole venous blood that had been
previously collected into EDTA-containing tubes; the
blood had been spotted onto blotting paper (3M, St Paul,
MN, USA) and stored at air-conditioned room
temperature (2025C) for 12 months, and then at 80C
for 3 years. We also randomly selected samples of sera
from a collection of stored sera of 2300 pregnant women,
attending the prenatal clinic at Mwananyamala District
Hospital, Dar es Salaam, between 1992 and 1994.13
For the Kenyan cohort, we did analyses on DNA
isolated from the blood samples of Kenyan children who
participated in the Asembo Bay Cohort Project (ABCP).14
Briefly, the study was done in an area of western Kenya
where more than 95% of the residents belong to the Luo
ethnic group. Pregnant women were enrolled mostly in
their last trimester and their newborn babies were enrolled
soon after the birth. The village monitors visited each
mother-child pair every fortnight for
clinical observation. At monthly
intervals, blood samples were obtained
for measurement of parasitaemia and
haemoglobin concentrations. Only
children from singleton births and the
first child of mothers who contributed
more than one birth to the cohort were
included in our study. We defined
severe malarial anaemia as a
haemoglobin concentration of less
Transcription than 60 g/L and a trophozoite count
NFB
start site greater than 10 000 trophozoites/L of
blood.
The College Research and
Publications Committee at Muhimbili
Medical Centre and the Institutional
Review Board of Duke University
Medical Centre approved the study.
Informed consent was obtained in
Kiswahili from all parents or guardians
of Tanzanian children. Consent was
also obtained for use of 368 sera of
pregnant women. At the time of
1469
MECHANISMS OF DISEASE

dividuals, we examined the first 8 kb of the NOS2

Primers used for amplification of the 1173 CT
fragment and for MS-PCR
Frag8.5: 5 -GACAAAGTTAATATATTATGCTG-3 outcomes. In the Tanzanian group, to identify any genetic
Frag8.3: 5 -CCCATACATATGCATTTTTCC-3 mutations, we amplified whole genome15 DNA from blood
MS-P2F: 5 -GACAAGAAGGAAATGAGTGGACACAGGTAGCAAAGT
GTTGAGAC-3 Technologies, Alameda, CA, USA). The whole genome
MS-P3R: 5 -GCATTTTTCCATCATAAAAGTAA-3 amplification products were diluted one in 100 in double
MS-P4F: 5 -GTGGTAGCAAATGTTGGAAT-3 distilled water or tris/EDTA for use as template in
Genotype
CT/TT CC Odds ratio p malaria who had the lowest NOx values, by SINGLE-
(Tanzania) and
relative risk (Kenya)
(95% CI)*
Tanzania (case control)
Clinical malaria 4 (30) 130 (970)
(n=134)
Healthy controls 9 (200) 36 (800) 012 (003048) 00006
(n=45)
Kenya (longitudinal cohort)
SMA (n=144) 4 (28) 140 (972)
No SMA (n=916) 81 (88) 835 (912) 031 (012084) 00120
Data are number (%) unless otherwise indicated. SMA=severe malarial
anaemia. *Data not adjusted for potential confounders. One of nine controls
with the C/T genotype had microscopically detectable asymptomatic
parasitaemia versus six of 36 children with the C/C genotype (p=081).
Individuals had one episode of SMA each. 105 (75%) individuals had one
episode of SMA, 26 (186%) had two, eight (57%) had three, and one had
nine.
Table 1: Effect of NOS2 1173 CT genotype on malarial
outcomes in Tanzanian and Kenyan children
drawing blood for the ABCP, consent was obtained to use
the Kenyan samples in other experiments.
Protocol
NO is rapidly oxidised to the stable metabolites nitrite and
nitrate (NOx) in vivo, and fasting NOx concentrations
have been validated as a measure of nitric-oxide
production in children.12 In the Tanzanian group,12 we
controlled exogenous nitrate ingestion to minimise the
contribution of diet to the urine and plasma nitrate
concentrations. In brief, we gave children a low nitrate
dinner, and they then fasted overnight. The next day, the
first morning-voided urine was discarded and, after a
distilled water challenge, the second fasting spot urine and
venous blood were collected. Concentrations of urine and
plasma NOx were measured as described,12 and expressed
as a ratio of NOx/creatinine to control for variability in
urine concentration and renal function, respectively. We
did not assess concentrations of NOx in the blood or urine
of the Kenyan group.
To investigate the relation between malarial disease
phenotype and NOS2 gene regulation in Tanzanian in-
promoter (8 kb to +1 bp) (Genbank accession number
AF017634) for polymorphisms associated with malaria
spotted on filters with 15-mer random primers (Operon
subsequent PCR amplifications. We did the initial
polymorphism detection screening on 12 Tanzanian
samples from the healthy controls with the highest NOx
values and on 12 samples from the patients with cerebral
STRANDED CONFORMATIONAL POLYMORPHISM (SSCP) analysis.
Of the single nucleotide polymorphisms identified, one
was differentially distributed between controls and
patients with cerebral malaria, and was explored further.
We amplified the 1173 CT fragment with the primers
Frag85 and Frag83 (panel). We separated the denatured
PCR products on precast 20% 125XTBE(tris/borate/
EDTA)-acrylamide gels (Invitrogen, San Diego, CA,
USA). We stained the gels in 1% Cyber Gold (Molecular
Probes, Eugene, OR, USA) in 125XTBE, and visualised
the conformations by illumination on a shortwave ultra-
violet light box (Fotodyne, New Berlin, WI, USA). We
investigated the Kenyan cohort by MUTAGENICALLY
SEPARATED PCR (MS-PCR).16 We used the primers MS-P2F at
01 mol/L, MS-P3R at 05 mol/L, and MS-P4F (panel)
at 045 mol/L, in 15 mmol/L magnesium chloride,
1Biolase PCR buffer, 5U Biolase DNA polymerase
(Bioline, Springfield, NJ, USA), and TaqStart antibody
(CloneTech, Palo Alto, CA, USA) as per the
manufacturers instructions. We separated the PCR
products on a 4% MetaPhor agarose gel (BioWhittaker
Molecular Applications, Rockland, MA, USA), and
visualised them by ethidium bromide staining. CCTTT
repeats were amplified as previously described.17
Statistical analysis
We assessed the significance of the 1173 CT single
nucleotide polymorphism in the Tanzanian study by
Fishers exact test on the 22 cross-tabulation of cases
(clinical malaria) and controls by genotype. We used
multiple linear regression (Stata version 7.0) to model the
relation between NOS2 promoter genotype and NOx
production (log transformed urine Nox/creatinine and
plasma NOx/creatinine), controlling for age, parasitaemia,
and interleukin-12 p40 promoter genotype (the only other
genotype known to affect NO production in this
population18). Data from the Kenyan cohort were
analysed with univariate and multivariate Poisson
regression, to ascertain the association between genotypes
and malaria-associated morbidity. We used generalised
estimating equations to adjust for the correlation between
multiple visits from the same individual. Covariates were

Genotype Number of events/number of total visits (%) Risk ratio (95% CI)* p
Outcome
Severe malarial anaemia C/T 4/1399 (03%) 025 (009066) 0005
C/C 190/15 680 (12%) 100 (reference)
Severe anaemia C/T 36/1399 (26%) 072 (046112) 0146
C/C 536/15 680 (34%) 100 (reference)
Parasitaemia C/T 239/1651 (145%) 086 (072101) 0081
10 000 trophozoites/L of blood C/C 3042/18 309 (166%) 100 (reference)
Parasitaemia >0 and <10 000 trophozoites/L of blood C/T 1116/1651 (676%) 106 (098114) 0117
C/C 11526/18309 (629%) 100 (reference)
*Poisson regression analysis with generalised estimating equations (to adjust for correlation that exists between observations from the same individual) controlling
for sickle-cell trait status.
Table 2: Effect of 1173 CT genotype on development of severe malarial anaemia and parasitaemia in Kenyan children

1470 THE LANCET Vol 360 November 9, 2002 www.thelancet.com

For personal use. Only reproduce with permission from The Lancet Publishing Group.
 MECHANISMS OF DISEASE

p=0038* one (1%) was homozygous T/T.

p=0014*
Analysis of the distribution of the
10 30 NOS2 1173 CT single nucleotide

Plasma NOx/creatinine (uM/uM)

09 polymorphism between the different
Urine NOx/creatinine (uM/uM)

25 disease groups in Tanzania showed

08
07
20
06
05 15
04
10
03
02
05
01
0 0
C/C C/T
(n=34) (n=9)
1173 genotype
Figure 2: Concentrations of NOx in urine and plasma of fasting Tanzanian controls,
according to C-1173T genotype
Boxes=2575 percentiles. Whiskers=1090 percentiles. Horizontal bar=median. Dot=geometric mean.
*After controlling for age, parasitaemia, and interleukin-12 p40 promoter genotype.
identified a priori, and included the mothers survival
(died at some time during follow-up), fetal gestational age
(<37 weeks vs 37 weeks), birthweight (<2500 g vs
2500 g), sex of child, and sickle cell trait (HbAS)
genotype. We judged a p value of less than
005 significant. We computed regression models with
SAS (version 8.0).
Role of the funding source
The sponsors of the study had no role in study design,
data collection, data analysis, data interpretation, or
writing of the report.
Results
We identified the polymorphisms in the NOS2 promoter
in 185 malaria-exposed Tanzanian children with uncom-
plicated and complicated clinical malaria (n=138) and
without clinical malaria (controls; n=47) (table 1). 20% of
the controls had malaria parasites on thick blood film
examination, but all were asymptomatic. The PCR
fragment spanning base pair 1173 (relative to the NOS2
transcription start site8) produced three different
conformation patterns by SSCP gel analysis (T/T, C/C
[wild-type], and C/T), each of which we sequenced. Six
(32%) samples could not be amplified. Of the remaining
179 samples, 927% (n=166) were homozygous wild-type
C/C genotype, 67% (n=12) were heterozygous C/T
genotype, and one sample (06%) was homozygous
variant genotype T/T at base pair 1173. 49 of 50
(965%) children with uncomplicated malaria were
homozygous wildtype C/C genotype, and one (2%) was
heterozygous C/T. 82 of 85 (965%) of the children with
cerebral malaria were wild-type, two (2%) were C/T, and
that the odds of developing clinical
malaria was 88% lower in those with
the C/T or T/T genotype than in
those with the wild-type genotype
(table 1). The protective association
of 1173 CT in the control group
was also highly significant compared
with the groups with uncomplicated
(p=0004) or cerebral (p=0005)
C/C C/T malaria.
(n=33) (n=9) To confirm and extend this
finding, we analysed the 1173 CT
single nucleotide polymorphism by
MS-PCR16 in 1106 samples from a
birth cohort from the holoendemic
area of Asembo Bay in Western
Kenya, where severe malarial
anaemia is more prevalent than
cerebral malaria. In this open cohort study,14 at least
monthly data on the severe outcomes associated with
malaria were available up to 5 years of age. Clinical data
were available for 1060 of those genotyped.
The C/T or T/T genotypes were significantly associated
with protection from severe malarial anaemia (table 1).
When we controlled these data for various confounders,
the risk ratio for developing severe malarial anaemia over
time was 75% lower in children with the C/T genotype
than in those with the C/C genotype (table 2). The
protection afforded by the 1173 CT polymorphism
seemed to be specific for severe malarial anaemia,
although there was a modest, non-significant, reduction in
the risk for overall severe anaemia in those with the 1173
CT polymorphism (table 2). The C/T genotype was not
associated with reduced risk for high-density parasitaemia
or overall parasite rates (table 2). This finding contrasts
with the sickle-cell trait-associated protection against
severe malarial anaemia that was accompanied by a
significant reduction in high density parasitaemia in the
same cohort.19 Furthermore, there was no significant
difference in the overall mortality rate between the two
NOS2 genotypes in the first 5 years of life (data not
shown).
A case-control study in children does not allow
estimates of the frequency of a polymorphism in the
general population. To ascertain the incidence of the
1173 CT polymorphism in the general adult
Tanzanian population, we tested for the polymorphism in
DNA from sera of 368 pregnant women.13 8% of these
individuals were heterozygous (C/T) for the poly-
morphism, and one was homozygous T/T. The C-allele
frequency was 0958 and the T-allele frequency was 0042

Genotype* Number of events/number of total visits Risk ratio (95% CI) p

Outcome
Severe malarial anaemia Long 132/12 228 (11%) 071 (043115) 0163
Short 30/2023 (15%) 100 (reference)
Severe anaemia Long 398/12 288 (32%) 082 (058114) 0232
Short 73/2023 (36%) 100 (reference)
Parasitaemia 10 000 trophozoites/L of blood Long 2379/14 386 (165%) 106 (092122) 0423
Short 371/2356 (158%) 100 (reference)

Parasitaemia >0 and <10 000 trophozoites/L of blood Long 9064/14 386 (630%) 096 (093103) 0334
Short 1539/2356 (653%) 100 (reference)
*Long=CCTTTn polymorphism with n11 (n=760), and short=CCTTTn polymorphism with n<11 (n=135).
Table 3: Effect of CCTTTn long versus short genotype on development of severe malarial anaemia and parasitaemia in Kenyan children

THE LANCET Vol 360 November 9, 2002 www.thelancet.com 1471

For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE

Genotype* Number of outcome events/number of total visits Risk ratio (95% CI) p
Outcome
Severe malarial anaemia CT/12-13 3/1157 (03%) 024 (008074) 0013
CC/12-13 129/11 131 (12%) 100 (reference)
Severe anaemia CT/12-13 24/1157 (21%) 060 (038096) 0033
CC/12-13 374/11 131 (34%) 100 (reference)
Parasitaemia CT/12-13 211/1362 (155%) 090 (076107) 0245
10 000 trophozoites/L of blood CC/12-13 2168/13 024 (166%) 100 (reference)
Parasitaemia >0 and <10 000 trophozoites/L of blood CT/12-13 909/1362 (667%) 106 (098115) 0155
CC/12-13 8155/13 024 (626%) 100 (reference)
*CT/12-13 signifies the 1173 CT polymorphism and a CCTTT1213 in the same individual (n=68), and CC/12-13 signifies the wildtype 1173C and a CCTTT1213 in
the same individual (n=379). Poisson regression analysis with generalised estimating equations (to adjust for correlation that exists between observations from the
same individual) controlling for sickle-cell trait status.
Table 4: Effect of 1173 CT/CCTTT1213 genotype on development of severe malarial anaemia and parasitaemia in Kenyan children

in these women. The population seemed to be in Hardy-
Weinberg equilibrium, with 066 T/T individuals
expected from a sample size of 368. This number
correlated with the observed frequency in this sample set.
In the Kenyan population, 1004 of 1106 (91%) samples
were homozygous wild type (C/C), 89 (8%) heterozygous
(C/T), none homozygous variant (T/T), and 13 (1%)
could not be amplified. This population had a T-allele
frequency of 0041, and thus two homozygous T/T
individuals would be predicted out of 1106 samples. The
absence of individuals with the T/T genotype is within the
sampling error of the assay. Using SSCP analysis, we did
not find the 1173 CT polymorphism in any of 132
unrelated samples taken from white people of northern
European descent (CEPH samples).20
We examined previously measured fasting concen-
trations of NO metabolites12 in Tanzanian children to
investigate the relation between the 1173 CT
polymorphism and NO production. Too few of the
children with uncomplicated and cerebral malaria had the
1173 CT polymorphism (one and two, respectively) to
test for significance between this polymorphism and NOx
concentrations in these samples, so we confined our
analysis to the healthy group. In the controls, those with
the 1173 CT polymorphism had significantly higher
fasting urine NOx excretion and plasma NOx
concentrations than did the children with a wild-type
genotype (figure 2).
We compared the distribution of the 1173 CT
polymorphism with that of the NOS2 promoter
954 GC polymorphism and the (CCTTT)n repeats,
22
20 NOS2 1173 C T
HbAS
NOS2 1173 C T or HbAS (%)
previously examined in the Tanzanian children.11 Of those
Tanzanian children with the 1173 CT polymorphism,
none had a short allele genotype. Further analysis revealed
that 12 of 13 (92%) 1173 CT children had a
CCTTT13 repeat, compared with a frequency of the
CCTTT13 repeat in the CC children of 32% (42 of 130)
with clinical malaria and 22% (8 of 36) of control
children. The children with the 954 GC poly-
morphism most frequently had either the CCTTT8 (41%)
or CCTTT11 (30%) repeat sized alleles. Although the
1173 CT polymorphism was present in two individuals
with the 954 GC polymorphism (one control and one
with uncomplicated malaria), in each instance the
individual had a CCTTT13 repeat and a CCTTT8 or a
CCTTT12 repeat, respectively. This finding indicates that
the 1173 CT polymorphism and the 954 GC
polymorphism were present on separate copies of
chromosome 17. We conclude that the 1173 CT
polymorphism sorts independently from the 954 GC
polymorphism, and that the 1173 CT polymorphism
seems to have arisen on a haplotype background linked to
a CCTTT13 repeat.
Tables 3 and 4 show detailed analyses of the
(CCTTT)n repeats and the 1173 CT polymorphism in
the Kenyan data set. There was no significant relation of
the (CCTTT)n, considering long versus short repeats,
with malaria disease severity or with parasitaemia
(table 3). In this population, 70 of 74 children with both
1173 CT and CCTTT genotype data had either the
sized 12 or 13 repeat CCTTT1213. Moreover, when we
considered those who had the 1173 CT polymorphism
and a CCTTT1213 genotype (CT/1213), they were
significantly less likely to have severe malarial anaemia
than those with the wildtype 1173 C and CCTTT1213
genotype (CC/1213), or those with any other CCTTT
repeat size (CC/non1213, table 4) when we controlled

18 for sickle-cell trait. All individuals without a CCTTT1213

16 were wild-type 1173 C (CC/non-12-13), and showed no
significant protection against severe malarial anaemia (risk
14 ratio=108, 95% CI=069168, p=07381) or high
12 density parasitaemia (092, 083102, p=01) (controlled
for sickle-trait status) compared with 1173 CC/
10 CCTTT1213 individuals (reference, 100). This finding
indicates that the 1173 CT polymorphism (and not
8
CCTTT repeat length) affects the severity of malaria in
6 these children. The presence or degree of parasitaemia did
not differ significantly between those with or without the
4 1173 CT polymorphism or long or short CCTTT
2 polymorphism.
Sickle-cell trait (HbAS) protects against severe
0 outcomes of malaria, and it is highly prevalent in east
Controls Uncomplicated Cerebral African populations who live in malaria endemic areas.19
malaria malaria To compare the frequency and distribution of the
Figure 3: NOS2 1173 CT genotype or sickle-cell trait 1173 CT single nucleotide polymorphism and HbAS,
(HbAS) relative to disease category in Tanzanian children we analysed the Tanzanian samples for HbAS by

1472 THE LANCET Vol 360 November 9, 2002 www.thelancet.com

For personal use. Only reproduce with permission from The Lancet Publishing Group.
 MECHANISMS OF DISEASE

Number of Odds ratio (OR) or relative risk (RR)

Comparison Country patients Reference

1173 C T Control vs clinical malaria Tanzania 179 This paper OR

1173 C T No SMA vs SMA Kenya 1060 This paper RR

954 G C Gabon 200 9 RR

Uncomplicated malaria vs SM
954 G C Tanzania 180 11 OR
Control vs clinical malaria
954 G C Control vs cerebral malaria Tanzania 134 11 OR

OR
CCTTT>11 SMA vs cerebral malaria died The Gambia 160 10
OR
CCTTT>11 Cerebral malaria lived vs died The Gambia 290 10
RR
CCTTT>11* No SMA vs SMA Kenya 895 This paper
OR
CCTTT>11 Control vs cerebral malaria Tanzania 131 11
CCTTT15 Uncomplicated malaria Thailand 283 29 OR
vs cerebral malaria
0 05 10 15 20 25 30

Figure 4: Comparison of reported odds ratios or relative risks for NOS2 promoter polymorphisms in malaria
Relative risks and odds ratios presented as mean (95% CI). SMA=severe malarial anaemia. SM=severe malaria. *Unadjusted for effects of 1173 CT
(table 4). 954 GC was noted in only one of 75 patients.

amplification of a portion of the beta globin gene, and
studied associations of HbAS with malaria severity in this
population. The sickle-cell trait was found in seven of 42
control children (17%) and in one of 45 children with
uncomplicated malaria (2%), but it was absent in 69
children with cerebral malaria (figure 3). As expected, the
protective association of HbAS was highly significant
(odds ratio 004, 95% CI 002037, p=00004).
The frequency of 1173 CT polymorphism was
similar to that of HbAS (figure 3) in controls and in
children with uncomplicated malaria, and the traits sorted
independently (the gene for NOS2 is on chromosome 17,
and the beta globin gene is on chromosome 11), since
only one control child had both the 1173 CT
polymorphism and the sickle-cell trait. Although HbAS is
more protective than the 1173 CT polymorphism
(HbAS odds ratio 004, 95% CI 001037, p=00004;
1173 CT polymorphism 012, 005027, p=00006),
combined, 37% of the Tanzanian control children had
either the 1173 CT polymorphism or the HbAS
protective polymorphism, compared with 3% of the group
with clinical malaria. In the Kenyan cohort, the
prevalence of HbAS was 174% and was associated with
significant protection against severe malarial anaemia and
high-density parasitaemia.19 This potential confounder
was controlled for in the multivariate analysis (tables 2, 3,
and 4) for this population.
Discussion
We have established the protective association of the
1173 CT NOS2 promoter polymorphism in two
cohorts of populations living in areas with different
malaria endemicity and distinct patterns of severe disease.
Validation of the protective association in the large
longitudinal Kenyan birth cohort provides convincing
evidence for it being an important genetic marker of
protection from severe malaria. This 1173 CT
polymorphism is as prevalent as the sickle-cell trait in the
two East African populations studied, and it is not found
in white populations that lack a history of exposure to
malaria. The 1173 CT polymorphism is significantly
associated with protection from both cerebral malaria and
severe malarial anaemia, and with increased NO
production in vivo. These associations agree with clinical
studies that lend support to a protective role for NOS2 in
malarial outcomes.12,2124

Wildtype NOS2 1173 CT NOS2

C T
DNA
Cytokines Cytokines
DNA
Microbes Microbes
NOS2
NOS2 mRNA Toxins
Toxins mRNA
L-arginine
NOS2 NOS2
protein protein

Malaria parasite growth Malaria parasite growth

NO
NO NO
Cytoadherence ICAM-1 ICAM-1 Cytoadherence
NO
TNF toxicity TNF TNF TNF toxicity

Figure 5: Proposed mechanism of the protective effect of the 1173 CT NOS2 promoter polymorphism against clinical malaria
TNF=tumour necrosis factor.

THE LANCET Vol 360 November 9, 2002 www.thelancet.com 1473

For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
The 1173 CT polymorphism lies in the region of the
promoter (25 kb to +1 bp) that produces strong
induction of NOS2 transcription in the mouse,25 but only
basal transcription in the human NOS2 gene.7 The
polymorphism maps to the region of the human promoter
that correlates with the enhancer region (R II) of the
mouse NOS2 promoter,25 but lies within a 16 bp sequence
not found in mice. The C to T change predicts the
creation of a new sequence recognition site for the
GATA-1 or GATA-2 transcription factors. These factors,
proteins that bind to specific DNA sequences and affect
transcription of mRNA from DNA, could potentially
account for an increased degree of transcription from the
1173 CT promoter.26 However, neither GATA-1 nor
GATA-2 bound to wild-type or 1173 CT DNA
sequences in gel shift assays (data not shown). Therefore,
GATA-1 or GATA-2 binding is probably not the
mechanism through which increased NO production
arises. An alternate hypothesis would be that the
1173 CT polymorphism disrupts the binding of a
repressor of transcription at this site. Pance and co-
workers27 reported increased transcription of a reporter
gene from NOS2 promoters with deletions in the proximal
portion of the NOS2 promoter. Further studies will be
needed to test this hypothesis.
Figure 4 summarises results from studies1113,28 on the
effect of various NOS2 promoter polymorphisms with
respect to risk of malaria. As noted, some investigators
have reported significant protection by the 954 GC or
CCTTT long repeat polymorphisms, whereas others have
failed to confirm this finding in individuals from other
countries.29 In our previous studies,11 neither the
954 GC polymorphism nor the CCTTT long repeat
polymorphisms were related to disease category or NO
production in Tanzania. The results of this study indicate
that the 1173 CT polymorphism sorts independently
from the 954 GC polymorphism in Tanzanians, and in
both populations the 1173 CT polymorphism was
strongly linked to the mean large repeat lengths of
CCTTT13 or CCTTT1213. A significant effect on malaria
severity of CCTTT1213 was noted only in those individuals
who also had the 1173 CT polymorphism. These data
raise the possibility that the reported association of long
repeat size with protection from fatal cerebral malaria in
Gambian children10 is due to a specific single nucleotide
polymorphism linked to the long repeat sizes in that
population. Our data cannot exclude the possibility that
the 1173 CT is in linkage disequilibrium with another
(unidentified) polymorphism in the NOS2 gene.
Nevertheless, based on the higher in-vivo NO production
in the Tanzanian children with the 1173 CT genotype
than in those with the wild-type genotype, we postulate
that the 1173 CT polymorphism causes a functional
difference in NOS2 expression or function, or both,
resulting in increased NO production (figure 5).
Our results concur with those of previous studies12,2124
that indicate that increased NOS2 expression and NO
production in individuals is associated with less severe
disease. Some have postulated that local tissue production
of NO could be deleterious in severe malaria,30 but our
data firmly support a protective role. The NOS2
1173 CT polymorphism could be protective in at least
three ways (figure 5). First, NO could have direct
antiparasitic effects, either on liver or blood stages. An
important role for NO in attenuating the growth of
malaria parasites in the liver stage has been well
documented in experimental malaria studies,6 and
immune responses to liver-stage parasites have been
previously linked to protection from severe malaria.28
1474
For personal use. Only reproduce with permission from The Lancet Publishing Group.
Although NO-related activity kills blood stages in vitro,
we observed that the frequencies of both low-level and
high-level parasitaemia were not significantly different in
the C/C and C/T individuals. This finding is consistent
with data from several rodent and human malaria studies
that suggest that NO has more antidisease effects than
direct antiparasitic effects against blood-stage infections.21
Second, NO decreases expression of the cytokine-
inducible endothelial adhesion molecules intercellular
adhesion molecule-1 (ICAM-1), vascular cell adhesion
molecule-1 (VCAM-1), and E-selectin31 involved in
cytoadherence and microvascular sequestration of
parasitised red cells.32 If the 1173 CT polymorphism
results in enhanced NOS2 expression and NO production
in activated endothelium and local macrophages, then
expression of adhesion molecules could be diminished,
decreasing organ microvascular sequestration of
P falciparum infected red blood cells, and end-organ
complications such as cerebral malaria. Third, NO
potently decreases macrophage production of tumour
necrosis factor.33 There is strong evidence that tumour
necrosis factor, lymphotoxin, and other proinflammatory
cytokines are directly related to the pathogenesis of severe
malaria.34,35 Increased NO production, resulting from the
1173 CT polymorphism, could result in lower
proinflammatory cytokine production in response to
infection and might explain the decreased risk of both
uncomplicated clinical illness and severe disease in those
with the 1173 CT polymorphism.
Our results warrant a more detailed genetic study of
NOS2 and its promoter in malaria and other infectious
diseases. Our work shows that NO is associated with
protection against two of the most severe manifestations
of childhood malaria. As such, targeted interventions to
increase NO delivery or production could provide novel
preventive and therapeutic strategies against these major
causes of death in African children.
Contributors
N M Anstey, D L Granger, E D Mwaikambo, and J B Weinberg
designed, coordinated, and did the Tanzanian part of the study. A L Lal,
V Udhayakumar, S Kariuki, and B L Nahlen designed, coordinated, and
did the Kenyan part of the study. M R Hobbs designed, directed, and
undertook genotyping and analysed the data. J Booth, A N Tkachuk, and
A Pole did the genotyping. M R Hobbs, M C Levesque, H Coon, and
N M Anstey did the statistical analysis for the Tanzanian part of the
study, and J M Roberts did the statistical analysis for the Kenyan part of
the study. M R Hobbs, V Udhayakumar, N M Anstey, D L Granger,
M C Levesque, A L Lal, and J B Weinberg interpreted data and wrote the
report.
Conflict of interest statement
None declared.
Acknowledgments
We thank Mushtaq Y Hassanali, Chris Hahn, Stella Stanslaus, and
Denis Manyenga for assistance with specimen collection and processing in
Dar es Salaam; Michael L Greenberg and John D Hamilton for making
the adult sera available; Lesa Hall for assistance with artwork; and
Jean-Marc Lalouel, Barton Haynes, and Mark Leppert for critical review
of the manuscript.
This work was supported by the National Institutes of Health
(AI41764 and 5P30CA42014), the National Center for Research
Resources to the University of Utah School of Medicine General Clinical
Research Center (M01-RR00064), an American Society of Tropical
Medicine and Hygiene Fellowship in Tropical Medicine, and the VA
Research Service. The US Agency for International Development
(HRN6001-A-00401000) supported the Asembo Bay Cohort Study,
and the NCID/CDC Emerging Infectious Disease Fund supported the
genetic component of the study.
References
1 Anon. World malaria situation in 1993. Wkly Epidemiol Rec 1996; 71:
1722.
2 Marsh K, Forster D, Waruiru C, et al. Indicators of life-threatening
malaria in African children. N Engl J Med 1995; 332: 1399404.
THE LANCET Vol 360 November 9, 2002 www.thelancet.com

3 Snow RW, Omumbo JA, Lowe B, et al. Relation between severe
malaria morbidity in children and level of Plasmodium falciparum
transmission in Africa. Lancet 1997; 349: 165054.
4 Akanmori BD, Kurtzhals JA, Goka BQ, et al. Distinct patterns of
cytokine regulation in discrete clinical forms of Plasmodium falciparum
malaria. Eur Cytokine Netw 2000; 11: 11318.
5 Rockett KA, Awburn MM, Cowden WB, Clark IA. Killing of
Plasmodium falciparum in vitro by nitric oxide derivatives. Infect
Immun 1991; 59: 328083.
6 Seguin MC, Klotz FW, Schneider I, et al. Induction of nitric oxide
synthase protects against malaria in mice exposed to irradiated
Plasmodium berghei infected mosquitoes: involvement of interferon
gamma and CD8+ T cells. J Exp Med 1994; 180: 35358.
7 de Vera ME, Shapiro RA, Nussler AK, et al. Transcriptional
regulation of human inducible nitric oxide synthase (NOS2) gene by
cytokines: initial analysis of the human NOS2 promoter. Proc Natl
Acad Sci USA 1996; 93: 105459.
8 Levesque MC, Hobbs MR, Anstey NM, et al. A review of
polymorphisms in the human gene for inducible nitric oxide
synthase (NOS2) in patients with malaria. Sepsis 2001; 4:
21731.
9 Kun JF, Mordmuller B, Lell B, Lehman LG, Luckner D,
Kremsner PG. Polymorphism in promoter region of inducible nitric
oxide synthase gene and protection against malaria. Lancet 1998; 351:
26566.
10 Burgner D, Xu W, Rockett K, et al. Inducible nitric oxide synthase
polymorphism and fatal cerebral malaria. Lancet 1998; 352:
119394.
11 Levesque MC, Hobbs MR, Anstey NM, et al. Nitric oxide synthase
type 2 promoter polymorphisms, nitric oxide production, and disease
severity in Tanzanian children with malaria. J Infect Dis 1999; 180:
19942002.
12 Anstey NM, Weinberg JB, Hassanali MY, et al. Nitric oxide in
Tanzanian children with malaria: inverse relationship between
malaria severity and nitric oxide production/nitric oxide synthase type
2 expression. J Exp Med 1996; 184: 55767.
13 Anstey NM, Tissot Dupont H, Hahn CG, et al. Seroepidemiology of
Rickettsia typhi, spotted fever group rickettsiae, and Coxiella burnetti
infection in pregnant women from urban Tanzania. Am J Trop Med
Hyg 1997; 57: 18789.
14 Bloland PB, Ruebush TK, McCormick JB, et al. Longitudinal cohort
study of the epidemiology of malaria infections in an area of intense
malaria transmission, 1: description of study site, general
methodology, and study population. Am J Trop Med Hyg 1999; 60:
63540.
15 Zhang L, Cui X, Schmitt K, Hubert R, Navidi W, Arnheim N.
Whole genome amplification from a single cell: implications for
genetic analysis. Proc Natl Acad Sci USA 1992; 89: 584751.
16 Rust S, Funke H, Assmann G. Mutagenically separated PCR (MS-
PCR): a highly specific one step procedure for easy mutation
detection. Nucleic Acids Res 1993; 21: 362329.
17 Xu W, Liu L, Emson PC, Harrington CR, Charles IG. Evolution of a
homopurine-homopyrimidine pentanucleotide repeat sequence
upstream of the human inducible nitric oxide synthase gene. Gene
1997; 204: 16570.
18 Morahan G, Boutlis CS, Huang D, et al. A promoter polymorphism
in the gene encoding interleukin-12 p40 (IL12B) is associated with
mortality from cerebral malaria and with reduced nitric oxide
production. Genes Immunity 2002; 3: 41418.
THE LANCET Vol 360 November 9, 2002 www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
19 Aidoo M, Terlouw DJ, Kolczak M, et al. Protective effects of the
sickle cell gene against malaria morbidity and mortality. Lancet
2002; 359: 131112.
20 Dausset J, Cann H, Cohen D, Lathrop M, Lalouel JM, White R.
Centre detude du polymorphisme humain (CEPH): collaborative
genetic mapping of the human genome. Genomics 1990; 6:
57577.
21 Anstey NM, Weinberg JB, Granger DL. Nitric oxide in malaria. In:
Fang F, ed. Nitric oxide and infection. New York: Plenum
Publishing Corporation, 1999: 31141.
22 Perkins DJ, Kremsner PG, Schmid D, Misukonis MA, Kelly MA,
Weinberg JB. Blood mononuclear cell nitric oxide production and
plasma cytokine levels in healthy Gabonese children with prior mild
or severe malaria. Infect Immun 1999; 67: 497781.
23 Kun JF, Mordmuller B, Perkins DJ, et al. Nitric oxide
synthase 2 (Lambarene) (954 GC), increased nitric oxide
production, and protection against malaria. J Infect Dis 2001; 184:
33036.
24 Pombo DJ, Lawrence G, Hirunpetcharat C, et al. Immunity to
malaria after administration of ultra-low doses of red cells infected
with Plasmodium falciparum. Lancet 2002; 360: 61017.
25 Alley EW, Murphy WJ, Russell SW. A classical enhancer element
responsive to both lipopolysaccharide and interferon-gamma
augments induction of the iNOS gene in mouse macrophages. Gene
1995; 158: 24751.
26 Heinemeyer T, Wingender E, Reuter I, et al. Databases on
transcriptional regulation: TRANSFAC, TRRD and COMPEL.
Nucleic Acids Res 1998; 26: 36267.
27 Pance A, Chantome A, Reveneau S, Bentrari F, Jeannin JF. A
repressor in the proximal human inducible nitric oxide synthase
promoter modulates transcriptional activation. Faseb J 2002; 16:
63133.
28 Hill AV, Allsopp CE, Kwiatkowski D, et al. Common west African
HLA antigens are associated with protection from severe malaria.
Nature 1991; 352: 595600.
29 Ohashi J, Naka I, Patarapotikul J, Hananantachai H,
Looareesuwan S, Tokunaga K. Significant association of longer
forms of CCTTT microsatellite repeat in the inducible nitric oxide
synthase promoter with severe malaria in Thailand. J Infect Dis
2002; 186: 57881.
30 Clark I, Awburn M. Migration inhibitory factor in the cerebral and
systemic endothelium in sepsis and malaria. Crit Care Med 2002; 30
(suppl 5): S26367.
31 De Caterina R, Libby P, Peng HB, et al. Nitric oxide decreases
cytokine-induced endothelial activation: nitric oxide selectively
reduces endothelial expression of adhesion molecules and
proinflammatory cytokines. J Clin Invest 1995; 96: 6068.
32 Turner G. Cerebral malaria. Brain Pathol 1997; 7: 56982.
33 Florquin S, Amraoui Z, Dubois C, Decuyper J, Yoldman M. The
protective role of endogenously synthesized nitric oxide in
staphylococeal enterotoxin B-induced shock in mice. J Exp Med
1994; 180: 115358.
34 Grau GE, Taylor TE, Molyneux ME, et al. Tumor necrosis factor
and disease severity in children with falciparum malaria. N Engl J
Med 1989; 320: 158691.
35 Engwerda CR, Mynott TL, Sawhney S, De Souza JB, Bickle QD,
Kaye PM. Locally up-regulated lymphotoxin alpha, not systemic
tumor necrosis factor alpha, is the principle mediator of murine
cerebral malaria. J Exp Med 2002; 195: 137177.
1475