Antibody Responses to VAR1CSA and VAR2CSA
• JID 2007:196 (1 July) • 155
M A J O R A R T I C L E
Effects of Sex, Parity, and Sequence Variation
on Seroreactivity to Candidate Pregnancy Malaria
Andrew V. Oleinikov,1Eddie Rossnagle,1Susan Francis,1Theonest K. Mutabingwa,1,3,4,5Michal Fried,1,2
and Patrick E. Duffy1,2
1Seattle Biomedical Research Institute and
2University of Washington, Seattle;
3London School of Hygiene & Tropical Medicine, London, United
5Muheza Designated District Hospital, Muheza, Tanzania
4National Institute for Medical Research, Dar es Salaam, and
in the human placenta, and pregnancy malaria (PM) is associated with the development of disease in and the
death of both mother and child. A PM vaccine appears to be feasible, because women become protected as they
develop antibodies against placental infected erythrocytes (IEs). TwoIEsurfacemolecules,VAR1CSAandVAR2CSA,
bind CSA in vitro and are potential vaccine candidates.
We expressed all domains of VAR1CSA and VAR2CSA as mammalian cell surface proteins, using a
novel approach that allows rapid purification, immobilization, and quantification of target antigen. For serum
samples from East Africa, we measured reactivity to all domains, and we examined the effects of host sex and
parity, as well as the effects of parasite antigenic variation.
Serum samples obtained from multigravid women had a higher reactivity to all VAR2CSA domains
than did those obtained from primigravid women or from men. Conversely, serum samples obtained from men
had consistently higher reactivity to VAR1CSA domains than did those obtained from gravid women.Seroreactivity
was strongly influenced by antigenic variation of VAR2CSA Duffy binding–like domains.
Women acquire antibodies to VAR2CSA over successive pregnancies, but they lose reactivity to
VAR1CSA. Serum reactivity to VAR2CSA is variant specific, and future studies should examine the degree to which
functional antibodies, such as binding-inhibition antibodies, are variant specific.
Plasmodium falciparum–infected erythrocytes adhere to chondroitin sulfate A (CSA) to sequester
Plasmodium falciparum parasites sequester in the hu-
man placenta , and pregnancy malaria (PM) is as-
sociated with the development of disease in and the
death of both mother and child [2–5]. Previous studies
identified chondroitin sulfate A (CSA) as a major re-
ceptor molecule for sequestration of infected erythro-
cytes (IEs) in the placenta . Malariaparasitesvariably
express antigens on the IE surface that bind a variety
of endothelial receptors [7, 8], including CSA. PfEMP1
is a variant surface antigen family encoded by ∼60 var
Received 22 November 2006; accepted26January2007;electronicallypublished
23 May 2007.
Potential conflicts of interest: none reported.
Financial support: The Bill & Melinda Gates Foundation (grant 29202); National
Institutes of Health (grant R01AI52059 to P.E.D.).
Reprints or correspondence: Dr. Andrew V. Oleinikov, 307 Westlake Ave. N, Ste.
500, Seattle, WA 98109 (firstname.lastname@example.org).
The Journal of Infectious Diseases
? 2007 by the Infectious Diseases Society of America. All rights reserved.
genes per malaria parasite genome , and these pro-
teins have been implicated in a number of binding
interactions. The sequences of var genes vary substan-
tially within and between genomes. PfEMP1 forms are
expressed in a mutually exclusive manner , creating
extensive antigenic variation and the potential for mul-
tiple adhesion profiles. This variation is a major obsta-
cle to the development of a PfEMP1-based antimalarial
Resistance to PM increases over successive pregnan-
cies  as women acquire antibodies against placental
parasites. Serum samples obtained from immune mul-
tigravid women, but not those from men, can inhibit
binding of placental IEs to CSA , even IEs collected
in distant geographic regions. This serum activity is
related to protection from infection and disease during
pregnancy [12, 13]. Two PfEMP1 molecules, VAR1CSA
and VAR2CSA, have been implicated in PM and are
potential vaccine candidates (reviewed in ). Both
are large molecules of 1350 kDa with 7 and 6 distinct
156 • JID 2007:196 (1 July) • Oleinikov et al.
Hybrid protein for expression of Plasmodium falciparum antigens on the surface of mammalian cells. GFP, green fluorescent protein.
Duffy binding–like (DBL) domains, respectively, and each is
extensively cross-linked by disulfide bonds.
To study the role of these molecules in protective immunity,
we expressed all domains of VAR1CSA and VAR2CSA on the
surface of mammalian cells as green fluorescent protein (GFP)
fusion proteins, by use of a novel vector that allowed rapid
purification, immobilization, and quantification of antigen.We
prepared arrays of individual VAR1CSA and VAR2CSA do-
mains from laboratory strains and field isolates, and we tested
their immunoreactivity by use of serum samples obtainedfrom
East African donors, to determine the effects of host sex and
parity, as well as the effects of parasite antigenic variation, on
MATERIALS AND METHODS
Vector for the expression of malaria antigens on the surface
of mammalian cells.
The DNA sequence encoding enhanced
GFP (EGFP) was excised from pEGFP-N1 (Clontech)bymeans
of XhoI/NotI digestion. The sequence encoding the transmem-
brane and cytoplasmic (TMC) domains of the rat surface re-
ceptor megalin  was amplified by use of reverse-transcrip-
tion polymerase chain reaction (PCR)performedusingforward
and reverse primers with EcoRI and XhoI sites at their 5?ends,
respectively (forward primer: 5?-TTTGAATTCCTCCAGGGA-
CGACAATGGCTGTT-3?; reverse primer: 5?-TTTCTCGAGTA-
CGTCGGATCTTCTTTAACGAG-3?). The sequence then was
digested with EcoRI and XhoI. Plasmid vector pSecTag2C (In-
vitrogen) was digested with BamHI and EcoRI and then was
ligated to a double-stranded (ds) oligonucleotide adaptor
(AdEx) with a multicloning site created by annealing 2 single-
stranded (ss) oligonucleotides: 5?-GATCCTTAAGTCCGGAG-
CTAGAGGCGCCTCCGGACTTAAG-3?. The resulting vector
was digested with EcoRI and XhoI and then was ligated to the
megalin TMC fragment described above. This construct, in
turn, was digested with XhoI and Bsp120I and was ligated to
the EGFP fragment. The resultingvectorwasdigestedwithXhoI
and AgeI to remove double-digestion sites, and it then was
ligated to a ds oligonucleotide adaptor (created by annealing
the following 2 ss oligonucleotides:5?-TCGAGCTGAAGCTTC-
TCGACTGCAGGATTCGAAGCTTCAGC-3?) that introduced
point mutations to eliminate unwanted restriction sites. The
resulting vector, known as “pAdEx,” was used to clone and
express the P. falciparum antigens described inthepresentstudy
(figure 1). The integrity of the construct was verified by re-
striction digestion and sequencing.
Cloning malaria antigen genes into the pAdEx vector.
DNA encoding each antigen was amplified by PCR from strain
FCR3 and strain 3D7 P. falciparum genomic DNA or from
cloned placental parasite sample 661 cDNA (see below), by use
of PCR performed using primer pairs with appropriate restric-
tion enzyme sites (table 1). After PCR was performed,amplified
DNA fragments and the pAdEx vector were digested, ligated,
and cloned. The integrity of each construct was verified by
Cloning and sequencing of var2csa from placental parasite
Clinical placental parasite sample 661 was a pla-
cental intervillous blood sample obtained, after delivery, from
a woman at Muheza Designated District Hospital (Muheza,
Tanzania) who was participating in the Mother-Offspring Ma-
laria Studies (MOMS) Project (describedin).Parasitesam-
ples were stored in RNALater (Ambion) at ?20?C. RNA was
isolated using Trizol (Invitrogen) according to the manufac-
turer’s instructions. Purified RNA was treated with DNA-free
reagent (Ambion) to remove genomic DNA. RNA was then
reverse-transcribed using Superscript III and randomhexamers
(Invitrogen) for 2 h at 42?C. DBL6 forward primer 5?-AAGA-
ACATTGTTCTAAATGTC-3?and reverse primer 5?-TGTAA-
ATATTGTTCAATAAAATCC-3?were designed by aligning
PFL0030c sequences from strains 3D7 and ITG (GenBank ac-
cession no. AY372123) to identify conserved sequences that
flank the DBL6 domain. The PCR product was cloned into
pCR2.1 vector by use of the TOPO TA Cloning System (In-
vitrogen) and was sequenced in both directions.
Preparation of quantitative protein arrays with malaria
COS-7 cells (50%–70% confluent) were transfected
with various constructs by use of Fugene transfection reagent
(Roche) according to the manufacturer’s protocol. Cells from
Antibody Responses to VAR1CSA and VAR2CSA • JID 2007:196 (1 July) • 157
Table 1. Polymerase chain reaction primers for the amplification of antigen domains.
DBL1a-CIDR (271–2280) CCCGGATCCAGGATCATAAGGAACATACTAATTTACGG
DBL2b (2440–3402) CCCCCTTAAGTCTAATCGTAATCTTGGTTTTTCAAATG
DBL3g (3802–4698) TTCGGATCCTTAAAGAAAACGATGGAAAGAAAC
DBL4? (4855–5805) CCCGGATCCAGGAAAATGACGACAAATATACTAACATT
DBL5g (5968–7146) CCCGGATCCAGGACGATGAACCAAAAGAAGTTGAAGG
DBL7? (8761–9540) CCCGGATCCAGAAGGAATTACAAACTTTTACCTTCTG
DBL1X (1–1347) CCCCTCCGGAATGGATAAATCAAGTATTGCTAAC
DBL3X (3580–4557)CCCGGATCCAGAAGGAAAATGAAAGTACCAATAATAAAA CCCGAATTCCATCACTCGCAGATTTTCCTACATATTTA
DBL4? (4708–5643) CCCGGATCCAGGAGAAAAAAAATAATAAATCTCTTTG
DBL5? (5944–7008) CCCGGATCCAGTTAGATAGATGTTTTGACGACAAG
MSP-1 19-kDa CTD
aIn the sequences of var1csa (FCR3 strain; GenBank accession no. AJ133811), var2csa (3D7 strain; PlasmoDB accession no. PFL0030c), ama1 (3D7 strain;
PlasmoDB accession no. PF11_0344), and msp1 (3D7 strain; PlasmoDB accession no. PFI1475w).
bRestriction enzyme sites are underlined.
each 150-mm2flask were lysed 48 h after transfection (trans-
fection efficiency, 180%) with 5 mL of CellLytic reagent
(Sigma). Recombinant products were confirmed on Western
blots with anti-GFP monoclonal antibody (MAb) (1:500 di-
lution; Clontech), followed by horseradish peroxidase (HRP)–
conjugated anti–mouse IgG (1:1000 dilution; Sigma). Concen-
trations of fusion proteins (expressed in relative fluorescence
units) were measured using GFP fluorescence with the use of
the Fluoroskan Ascent FL fluorometer/luminometer (Thermo
Labsystems), and they then were equalized by dilution with
lysate of nontransfected cells (lysate K). A total of 100 mL of
diluted lysate was added to each well of 384-well white plates
coated with anti-GFP antibody (Pierce), and plates were in-
cubated at 4?C overnight. Undiluted lysate K was used as a
control for nonspecific background fluorescence and chemi-
luminescence. Lysate prepared from cells transfected with con-
trol construct (pAdEx vector alone without malaria antigen
fusion partner) was used as a negative control in each assay.
After washing with washing buffer (PBS plus 0.05% Tween-
20), plates were ready for immunoprofiling experiments in-
volving serum samples from humans.
Validation of expressed merozoite surface protein–1 (MSP-
1) antigen by structure-sensitive monoclonal antibody.
Recombinant MSP-119or control construct product AdEx was
immobilized in anti-GFP plates as described above and then
was incubated with mouse MAb 12.10 (1:5000 dilution),which
is reactive only to the properly folded structure of MSP-1 
(provided by Dr. J. A. Lyon, Walter Reed Army Institute of
Research), followed by HRP-conjugated anti–mouse IgG (1:
1000 dilution; Sigma). Reactivity signals were obtained (ex-
pressed in relative luminescence units) by use of 100 mL of ECL
chemiluminescence substrate (Amersham Biosciences) per well
and a Fluoroskan luminometer.
The human serum samples used in these
studies were collected from East African donors, under pro-
participants, who provided written, informed consent before
donating samples, included adult men and multigravid women
from Kenya [18, 19], as well as multigravid and primigravid
women from Tanzania . In brief, 18- to 45-year-old mul-
tigravid women and 18- to 50-year-old men (median age, 28
and 29 years, respectively;P p .62
to 45-year-old gravid women from Tanzania, were included in
the study. Serum samples obtained from pregnant womenwere
collected at the time of delivery and were tested individually.
The number of serum samples used in each experiment is
indicated in the corresponding figure legends. Serum samples
obtained from 10 randomly selected nonimmune donors in the
) from Kenya, as well as 18-
158 • JID 2007:196 (1 July) • Oleinikov et al.
fluorescent protein (GFP) fusion proteins in COS-7 cells and were immobilized individually for antigen arrays. As a positive control, apical membrane
antigen–1 (AMA-1)–GFP fusion protein was used. Cyt, cytoplasmic domain; DBL, Duffy binding–like domain; TM, transmembrane domain.
Plasmodium falciparum protein domains expressed and used for seroreactivity studies. Indicated domains were expressed as green
United States were separated from whole blood obtained from
commercial sources (Valley Biomedical) and were used in a
pool as a negative control.
Immunoprofiling study of malaria antigens.
samples were preincubated at 4?C for at least 24 h with an
equal volume of 10 mg/mL goat IgG, to eliminate nonspecific
reactivity against goat anti–GFP IgG bound to the wells of 384-
well plates. The preincubated serum samples were further di-
luted 1:100 with Superblock (Pierce) and were incubated with
the antigen array for 2 h at room temperature. After 3 washes
with washing buffer, plates were incubated with donkey anti–
human IgG (H+L) affinity-purified antibody conjugated to
HRP (Jackson Immunoresearch) diluted 1:1000 in Superblock.
After 1 h at room temperature, the wells were washed; 100 mL
of ECL chemiluminescence substrate (Amersham Biosciences)
were then added per well, and chemiluminescence and fluo-
rescence signals were measured. The use of the chemilumines-
cence substrate does not affect the fluorescence measurement.
Chemiluminescence signal reflects immune reactivity, and
fluorescence signal reflects theamount ofimmobilizedantigen–
GFP fusion proteins. Fluorescence signal was corrected by sub-
traction of background values measured in lysate K wells, and
then the immunoreactivity signal (chemiluminescence) was
normalized to the amount of immobilized antigen (fluores-
cence) in each well. Average reactivity was calculated for du-
plicate wells, and a final specific immunoreactivity (expressed
as arbitrary units [AUs]) was calculated by subtracting thecon-
trol value (defined as either the average reactivity of the same
serum sample to control construct +3 SD or the reactivity of
pooled nonimmune serum samples to the same antigen +3 SD,
man’s rank test. Differences between group reactivities were
tested for significance by use of the Mann-Whitney U test.
was considered to be statistically significant. GraphPad
Prizm software was used for all statistical analyses.
P ! .05
RESULTS AND DISCUSSION
Features and performance of quantitative protein arrays.
Heterologous expression of malaria surface antigens is known
to be difficult, in part because of their high AT content (up to
80%) and their highly conformational cysteine-rich structure.
An expression system that provides a transmembrane protein
trafficking pathway and cell-surface presentation may signifi-
cantly improve the cotranslational folding of PfEMP1 surface
molecules, in which each domain contains 6–9 disulfide bonds.
We engineered a pAdEx vector encoding a hybrid receptorwith
a signal peptide (from the immunoglobulin k chain), an ex-
tracellular domain, and individual transmembrane and cyto-
plasmic domains (both from the single-spanning transmem-
brane receptor megalin) (figure 1). The cytoplasmic domain
has signals that direct this protein to the plasma membrane
surface. In addition, the GFP-reporter protein is fused to the
cytoplasmic domain and reports protein expression levels,
which can be quantified. The multicloning site allows simple
and rapid preparation of different constructs that express Plas-
modium antigen extracellular domains on the surface of mam-
var1csa and var2csa genes, in addition to other P. falciparum
antigens (the apical membrane antigen–1 [AMA-1] and MSP-
1 19-kDa carboxy-terminal fragment), as GFP fusion proteins
(figure 2). All antigens were successfully expressed using the
native malaria coding sequence. Cysteine-rich interdomain re-
gion (CIDR)–a domains always follow DBL-a domains, and
they may act as a single functional domain ; therefore,
Antibody Responses to VAR1CSA and VAR2CSA • JID 2007:196 (1 July) • 159
7 cells. A, Western blot of expressed antigens with monoclonal anti-GFP antibody. Lane 1, Duffy binding–like (DBL) 3g region from VAR1CSA (predicted
molecular weight [MW], 85 kDa); lane 2, control construct minimegalin with extracellular domain containing the first ligand-binding domain of rat
receptor megalin (nt 1–1882)  (predicted MW, 106 kDa); and lane 3, merozoite surface protein–1 (MSP-1) 19-kDa fragment (predicted MW, 70
kDa). B, Interaction of structure-sensitive anti–MSP-1 monoclonal antibody (MAb) 12.10  with MSP-1 fusion protein. The control protein expressed
from vector without insert (AdEx control) or MSP-1 recombinant protein was immobilized in the wells of anti-GFP plates and was tested for reactivity
with monoclonal antibody (MAb) 12.10 followed by secondary anti–mouse horseradish peroxidase (HRP)–conjugated antibody (MAb 12.10) or with
secondary antibody only (2nd only). Signals were measured using chemiluminescent substrate. Bars denote the average of 3 measurements, and error
bars denote SEs. RLU, relative luminescence units.
Characterization of malaria antigens cloned as green fluorescent protein (GFP) fusion proteins in the pAdEx vector and expressed in COS-
var1csa DBL1-a domain was expressed together with CIDR1-
a domain. For negative control wells, we used a GFP fusion
protein (AdEx) containing an irrelevant extracellular domain
of 37 aa that resulted from the translation of the multicloning
site in the pAdEx DNA construct.
The integrity of fusion proteins was tested by Western blot
analysis with anti-GFP antibodies (figure 3A). Recombinant
proteins demonstrated the expected molecular weight and pro-
duced green fluorescence in cells as well as in cell lysates. Fluo-
rescence was preserved after immobilization of fusion proteins
in 384-well plates. GFP fluorescence has been shown to be a
good indicator of properly folded membrane proteins when
GFP is fused to the cytoplasmic tail . We also tested the
reactivity of the disulfide-rich MSP-1 19-kDa fusion protein by
use of conformation-dependent MAb 12.10 , which readily
recognized the antigen (figure 3B), thereby confirming correct
Malaria antigens were organized into arrays by use of a sin-
gle-step procedure performed in 384-well plates. The GFP fu-
sion partner has a number of advantages. First, the tag can be
used for immobilization and purification of antigens in a single
simple step. Second, the GFP allows the amount of antigen in
each lysate to be measured and equalized, thereby reducing
variance. Third, the immunoreactivity of serum samples (mea-
sured by chemiluminescence) can be normalized to theamount
of antigen (measured simultaneously by GFP fluorescence) in
each well, which further reduces variance.
Seroreactivity to irrelevant antigens.
earlier studies [22, 23], we found that serum samples obtained
from immune individuals in malaria-endemic regionsfrequently
react to completely irrelevant proteins (datanotshown),andthis
nonspecific reactivity corresponds to an elevated reactivity to
malaria antigens. In contrast, serum samples obtained from
nonimmune individuals (NISS) living in areas of nonende-
micity have low nonspecific reactivity. For this reason, NISS
control may not be adequate to demonstrate specific reactivity
of serumsamples testedinseroepidemiologicstudiesofmalaria,
because this approach may falsely identify serum samples with
high levels of nonspecific reactivity as having a positive result.
The use of the control construct provides the means toquantify
and, therefore, correct for nonspecific reactivity of each con-
struct in each serum sample.
Seroreactivity to VAR1CSA and VAR2CSA associated with
a dichotomous pattern related to sex.
reactivity of East African and nonimmune individuals to do-
mains of VAR1CSA and VAR2CSA expressed as GFP fusion
proteins. AMA-1 was used as a positive control because it is
known to react strongly to the majority of serum samples ob-
tained from individuals in malaria-endemic regions . As
expected, serum samples obtained from immune individuals
uniformly showed high levels of reactivity to relatively con-
served AMA-1, and seroreactivity did not differ between men
and multigravid women (inset in figure 4A) (median for 44
serum samples obtained from men, 5289 AUs; median for 52
As was observed in
We measured the sero-
160 • JID 2007:196 (1 July) • Oleinikov et al.
from men to VAR1CSA domains. Seroreactivity to VAR2CSA and VAR1CSA domains (after subtraction of the control value [see Materials and Methods])
is indicated according to donor group. White bars denote serum samples obtained from men, and gray bars denote serum samples obtained from
multigravid women. AU, arbitrary units. P values are the results of a 2-tailed Mann-Whitney U test (for 52 serum samples obtained from multigravid
women and 44 serum samples obtained from men [left] and for 32 serum samples obtained from multigravid women and 32 serum samples obtained
from men [right]). The top of the box denotes the 75th percentile, the bottom of the box denotes the 25th percentile, and the line through the middle
of the box denotes the 50th percentile (i.e., the median). The whiskers denote the 10th and 90th percentiles. The inset shows the reactivity of apical
membrane antigen–1 (AMA-1) for both groups.
Preferential reaction of serum samples from multigravid women to VAR2CSA domains and preferential reaction of serum samples obtained
serum samples obtained from multigravid women, 5872 AUs;
).P p .46
Immune responses to PfEMP1 domains were substantially
lower and more variable (figure 4) than wereAMA-1responses.
Two VAR1CSA domains (DBL6b and DBL7?) and 1 VAR2CSA
domain (DBL2X) were found to be nonreactive or minimally
reactive in our screening tests. Nonreactivity of VAR2CSA
DBL2X was likely the result of rapid degradation of this fusion
protein during and after cell lysate preparation, as detected by
Western blot analysis (data not shown). The reason for non-
reactivity of VAR1CSA DBL6 and DBL7 is not clear, because
the proteins were stable. The results suggest that host immu-
noreactivity is weak against these domains, but we cannot ex-
clude the possibility that the proteins were incorrectly folded
in a way that disrupted or masked structural epitopes.
The variable response to VAR1CSA and VAR2CSA was re-
lated to the sex of the serum donors. Consistent with the find-
ings of earlier studies from West Africa [25–27], the reactivity
of all VAR2CSA domains (other thanDBL-2X)wassignificantly
higher with serum samples from multigravid women than with
serum samples from men (figure 4). Of the 54 serum samples
that were obtained from Kenyan multigravid women and were
tested in this experiment, 10 were obtained from women with
PM. Antibody levels were not significantly different (data not
shown) in women with PM versus those without PM, possibly
reflecting that the duration of infection is brief in the multi-
gravid women [3, 28, 29] or that antibody levels may be max-
imal in this parity group by the time of delivery. Previous
studies in West Africa that examined 3 DBL domains (DBL1,
DBL5, and DBL6) of VAR2CSA expressed in baculovirus [25,
26] found that seroreactivity to domains 5 and 6, but not to
domain 1, was significantly higher in multigravid women than
in men. The increased reactivity against all VAR2CSA domains
noted for serum samples obtained from multigravid women in
East Africa supports the idea that this PfEMP1 molecule is
preferentially expressed by PM parasites and that women ac-
quire antibodies against this protein as they become protected.
Reactivity to the DBL1X domain was significantly correlated
with reactivity against 3 other domains (Spearman correlation
for DBL3,[ ]; for DBL5,r p 0.29 P p .04[ ];r p 0.41 P p .003
Antibody Responses to VAR1CSA and VAR2CSA • JID 2007:196 (1 July) • 161
White bars denote serum samples obtained from primigravid women, and gray bars denote serum samples obtained from multigravid women. AU,
arbitrary units. P values are results of a 2-tailed Mann-Whitney U test (np32 for each group). The top of the box denotes the 75th percentile, the
bottom of the box denotes the 25th percentile, and the line through the middle of the box denotes the 50th percentile (i.e., the median). The whiskers
denote the 10th and 90th percentiles.
Increases in serum reactivity to VAR2CSA domains with gravidity. Seroreactivity to individual VAR2CSA domains is stratified by gravidity.
and for DBL6,
ferent VAR2CSA domains is acquired concordantly.
The pattern of reactivity to VAR1CSA versus VAR2CSA do-
mains diverged markedly and was consistent against all tested
domains (figure 4). Serum samples obtained from multigravid
women reacted more strongly to VAR2CSA domains, whereas
serum samples obtained from men reacted more strongly to
VAR1CSA domains. The increased antibodylevelsnotedinmen
versus those noted in multigravid women were statistically sig-
nificant for 2 VAR1CSA domains (DBL1a and DBL5g). Men
and multigravid women had similar reactivity to AMA-1 (see
above) and MSP-1 (data not shown), indicating that multi-
gravid women specifically lose reactivity to VAR1CSA. Our
studies of AMA-1 and MSP-119are similar to numerous earlier
studies, which found that seroreactivitytovariousnon-PfEMP1
malaria antigens did not vary with the pregnancy status or
parity of the sample donors .
The dichotomous pattern of reactivity of men and multi-
[ ]) but not against AMA-1
), suggesting that immunity to dif-
r p 0.48 P p .0003
;r p 0.07 P p .6
gravid women may be explained by mutually exclusive ex-
pression of var genes in P. falciparum . PM is caused by
CSA-binding parasites  that preferentially express var2csa
, and peripheral parasites in pregnant women have features
similar to those of placental parasites [16, 32]. Thus, the up-
regulation of var2csa in placental parasitesmaybeaccompanied
by a down-regulation of other commonly expressed var genes,
such as var1csa. Antibodies against VAR1CSA domains may
therefore diminish in pregnant women, who would receive an-
tigenic stimulation by VAR2CSA but not VAR1CSA during ep-
isodes of PM.
Gravidity-related increases in seroreactivity to VAR2CSA.
We compared immunoreactivity to VAR2CSA domains in se-
rum samples obtained from multigravid women (all 32 of
whom did not have PM) and those obtained from primigravid
women (8 of whom had PM and 24 of whom did not) in
Tanzania. As was observed elsewhere [25, 26] (see below),
VAR2CSA seroreactivity increased with the number of preg-
nancies (figure 5) and, consequently, with protection status.
162 • JID 2007:196 (1 July) • Oleinikov et al.
sample 661 DBL6? domain sequences. Sequence alignment was performed using Clustal W at the GenomeNet Web site (available at: http://
align.genome.jp). Stars denote conserved residues, and colons and dots denote more- and less-conservative substitutions, respectively. a2, a4, a5,
and a7 are helical regions identifiable according to information found in . Blue boxes denote regions of low homology. Yellow boxes denote
cysteines conserved between these 2 variants. B, Reactivity of the 3D7 strain and the placental parasite sample 661 DBL6? domains with serum
samples obtained from 5 multigravid women. The 5 samples were randomly selected from the set of serum samples used in previously described
experiments. Black bars denote 3D7 DBL6?; white bars, 661 DBL6?.
Antigenic variation and seroreactivity of VAR2CSA Duffy binding–like (DBL) 6? domain. A, Comparison of strain 3D7 and placental parasite
Differences in seroreactivity betweengroupsof differinggravid-
ities were statistically significant for 3 domains (DBL1, DBL3,
and DBL6). These differences remained significant in analyses
that included only serum samples from women without PM.
In earlier studies, seroreactivity of the DBL5 domain  and
3 VAR2CSA domains (DBL1, DBL5 and DBL6)  correlated
significantly with gravidity, but the levels of seroreactivity were
not significantly different between groups of differing gravid-
ities. Interestingly, antibody levels to VAR2CSA domains 1, 3,
and 6 were significantly higher (, for all comparisons;P ! .05
Antibody Responses to VAR1CSA and VAR2CSA • JID 2007:196 (1 July) • 163
data not shown) among first-time Tanzanian mothers with PM
versus those without PM in our study, suggesting specific re-
sponses to the antigen during PM. Separate studies will need
to examine whether the antibodies produced by first-time
mothers during malaria episodes have functional activity.
Similar studies were previously undertaken in West Africa
with 2 VAR1CSA DBL domains (DBL1 and DBL2) expressed
in Escherichia coli  and with varying numbers of VAR2CSA
domains (2 , 3 , or 6 ) expressed in the baculovirus
system. In those studies, for serum samples obtained from men
and women, the reactivity against VAR1CSA domains did not
differ significantly. Of note, E. coli–expressed DBLantigensmay
not recreate the extensive disulfidebondsandfoldsofthenative
protein , and expression of the DBL1 domainseparatefrom
the CIDR domain may disrupt a single functional domain and
alter its conformation. In our studies, VAR1CSA domains were
the native structure of these complex antigens,whichmayallow
better discrimination of differences in seroreactivity.Theearlier
studies of VAR2CSA, which we discussed in detail above, gen-
erally observed a sex-specific and parity-specificpatternofreac-
tivity, supporting the idea that VAR2CSA is preferentially ex-
pressed by placental parasites and is targeted by antibodies that
correlate with immunity.
In our work, we expanded on these previous studies to in-
corporate all domains from each PfEMP1 protein, expressed
each domain in a mammalian system to increasetheprobability
of correct folding, and studied them together by use of serum
samples from a distinct geographic region, East Africa. To our
knowledge, these are the first studies to show a higher level of
samples obtained from men versus those obtained from mul-
tigravid women, and reactivity is highest against the first
VAR1CSA domain (DBL1a plus CIDR1a). We also demon-
strated that all immunoreactive VAR2CSA domains react most
strongly to serum samples obtained from multigravid women,
and we confirmed that this reactivity is parity specific.
Variant-specific reactivity to VAR2CSA.
reactivity of immune serumsamplestovariantformsofdomain
DBL6? representing laboratory isolate 3D7 and fresh placental
parasite sample 661. These variant forms have a high level of
homology throughout most of their sequence (figure 6A). In-
dividual serum samples obtained from multigravidwomenvar-
ied substantially in their reactivity to variant forms of DBL6?
(figure 6B). Antigenic variation in this domain is limited to
areas comprising ∼30% of the domain sequence, primarily in
the loops between helices a2 and a4, as well as those between
helices a5 and a7 . Because the remainder of the domain
is largely conserved, and because the immune response against
these 2 homologous domains is significantly different, we spec-
ulate that the immune response is predominantly directed to-
We compared the
ward regions of sequence variability, including the loops. This
may also indicate that the most conserved parts of the domain
are poorly immunogenic. We also saw a similar pattern of
differential reactivity with VAR2CSA DBL1X domains (identity
was ∼80% between variants [data not shown]).
A previous study by Tuikue Ndam et al.  demonstrated
no association between serum levels of anti-3D7 VAR2CSA
antibodies and anti–CSA-binding antibodies in 4 of 6 placental
isolates. This may have resulted from VAR2CSA sequence var-
iation between placental samples, as the authors suggested, or
it may indicate that functional antibodies are a minor subset
of total antibodies. Our results regarding differential reactivity
of laboratory isolate (3D7) versus placental parasite (sample
661) DBL domains do not confirm one or the other of these
possibilities. If the former possibility is correct, then protective
immunity in multigravid women may reflect the acquisition of
antibodies against the VAR2CSA variants present in a com-
munity. A globally related pool of polymorphisms accounts for
sequence variation in VAR2CSA , and, therefore, a limited
number of variants may be adequate to elicit broadly reactive
antibodies. Such a vaccine may be able to target only the loop
regions, which could significantly simplify the task of devel-
oping a vaccine
Future studies will need to identify the malaria antigen, do-
main, or domain variant(s) and fragment(s) thatarespecifically
targeted by protective antibodies, as well as those that elicit
broadly reactive antibodies. This information could providethe
basis for a PM vaccine.
We thank Vlad Malkov for fruitful discussion. J. Lyon providedantibody
1. Walter PR, Garin Y, Blot P. Placental pathologic changes in malaria: a
histologic and ultrastructural study. Am J Pathol 1982;109:330–42.
2. Brabin BJ. An analysis of malaria in pregnancy in Africa. Bull World
Health Organ 1983;61:1005–16.
3. McGregor IA. Thoughts on malaria in pregnancy with consideration
4. Murphy SC, Breman JG. Gaps in the childhood malaria burden in
Africa: cerebral malaria, neurological sequelae, anemia, respiratorydis-
tress, hypoglycemia, and complications of pregnancy. Am J Trop Med
5. Guyatt HL, Snow RW. Impact of malaria during pregnancy on low
birth weight in sub-Saharan Africa. Clin Microbiol Rev2004;17:760–9.
6. Fried M, Duffy PE. Adherence of Plasmodium falciparum to chondroi-
tin sulfate A in the human placenta. Science 1996;272:1502–4.
7. Baruch DI, Ma XC, Singh HB, Bi X, Pasloske BL, Howard RJ. Iden-
tification of a region of PfEMP1 that mediates adherence of Plasmo-
dium falciparum infected erythrocytes to CD36: conserved function
with variant sequence. Blood 1997;90:3766–75.
8. Rowe JA, Moulds JM, Newbold CI, Miller LH. P. falciparum rosetting
164 • JID 2007:196 (1 July) • Oleinikov et al. Download full-text
mediated by a parasite-variant erythrocyte membrane protein and
complement-receptor 1. Nature 1997;388:292–5.
9. Gardner MJ, Hall N, Fung E, et al. Genome sequence of the human
malaria parasite Plasmodium falciparum. Nature 2002;419:498–511.
10. Scherf A, Hernandez-Rivas R, Buffet P, et al. Antigenic variation in
malaria: in situ switching, relaxed and mutually exclusive transcription
of var genes during intra-erythrocytic development in Plasmodium
falciparum. EMBO J 1998;17:5418–26.
11. Fried M, Nosten F, Brockman A, Brabin BJ, Duffy PE. Maternal an-
tibodies block malaria. Nature 1998;395:851–2.
12. Duffy PE, Fried M. Antibodies that inhibit Plasmodium falciparum
adhesion to chondroitin sulfate A are associated with increased birth
weight and the gestational age of newborns. Infect Immun 2003;71:
13. Staalsoe T, Shulman CE, Bulmer JN, Kawuondo K, Marsh K, Hviid L.
Variant surface antigen-specific IgG and protectionagainstclinicalcon-
sequences of pregnancy-associated Plasmodium falciparum malaria.
14. Rowe JA, Kyes SA. The role of Plasmodium falciparum var genes in
malaria in pregnancy. Mol Microbiol 2004;53:1011–9.
15. Saito A, Pietromonaco S, Loo AK, Farquhar MG. Complete cloning
and sequencing of rat gp330/“megalin,” a distinctive member of the
low density lipoprotein receptor gene family. Proc Natl Acad Sci USA
16. Fried M, Domingo GJ, Gowda CD, Mutabingwa TK, Duffy PE. Plas-
modium falciparum: chondroitin sulfate A is the major receptor for
adhesion of parasitized erythrocytes intheplacenta.ExpParasitol2006;
17. Angov E, AufieroBM, TurgeonAM,etal.Developmentandpre-clinical
analysis of a Plasmodium falciparum merozoite surface protein-1(42)
malaria vaccine. Mol Biochem Parasitol 2003;128:195–204.
18. Kurtis JD, Lanar DE, Opollo M, Duffy PE. Interleukin-10 responses
to liver-stage antigen 1 predict human resistance to Plasmodium fal-
ciparum. Infect Immun 1999;67:3424–9.
19. Fried M, Muga RO, Misore AO, Duffy PE. Malaria elicits type 1 cy-
tokines in the human placenta: IFN-g and TNF-a associated with
pregnancy outcomes. J Immunol 1998;160:2523–30.
20. Mutabingwa TK, Bolla MC, Li JL, et al. Maternal malaria and gravidity
interact to modify infant susceptibility to malaria. PLoS Med 2005;2:
21. Drew DE, von Heijne G, Nordlund P, de Gier JW. Green fluorescent
protein as an indicator to monitor membrane protein overexpression
in Escherichia coli. FEBS Lett 2001;507:220–4.
22. Knobloch J, Schreiber M, Grokhovsky S, Scherf A. Specific and non-
specific immunodiagnostic properties of recombinant and synthetic
Plasmodium falciparum antigens. Eur J Clin Microbiol 1987;6:547–51.
23. Fesel C, Goulart LF, Silva Neto A, et al. Increased polyclonal immu-
noglobulin reactivity toward human and bacterial proteinsisassociated
with clinical protection in human Plasmodium infection. Malar J 2005;
24. Cortes A, Mellombo M, Masciantonio R, Murphy VJ, Reeder JC, An-
ders RF. Allele specificity of naturally acquired antibody responses
against Plasmodium falciparum apical membrane antigen 1. Infect Im-
25. Salanti A, Dahlback M, Turner L, et al. Evidence for the involvement
of VAR2CSA in pregnancy-associated malaria. J Exp Med 2004;200:
26. Tuikue Ndam NG, Salanti A, Le-Hesran JY, et al. Dynamics of anti-
VAR2CSA immunoglobulin G response in a cohort of Senegalesepreg-
nant women. J Infect Dis 2006;193:713–20.
27. Barfod L, Nielsen MA, Turner L, et al.Baculovirus-expressedconstructs
induce immunoglobulin G that recognizes VAR2CSA on Plasmodium
falciparum-infected erythrocytes. Infect Immun 2006;74:4357–60.
28. McGregor I. Malaria—recollectionsandobservations.TransRSocTrop
Med Hyg 1984;78:1–8.
29. Brabin B, Rogerson S. The epidemiology and outcomes of maternal
malaria. In: Duffy PE, Fried M, eds. Malaria in pregnancy. New York,
London: Taylor and Francis, 2001:27–52.
30. Fievet N, Cot M, Chougnet C, et al. Malaria and pregnancy in Ca-
meroonian primigravidae: humoral and cellular immune responses to
Plasmodium falciparum blood-stage antigens. Am J Trop Med Hyg
31. Salanti A, Staalsoe T, Lavstsen T, et al. Selective upregulation of asingle
distinctly structured var gene in chondroitin sulphateA–adheringPlas-
modium falciparum involved in pregnancy-associated malaria.MolMi-
32. Ofori MF, Staalsoe T, Bam V, et al. Expression of variant surface an-
tigens by Plasmodium falciparum parasites in the peripheral blood of
clinically immune pregnant women indicates ongoing placental infec-
tion. Infect Immun 2003;71:1584–6.
33. Jensen AT, Zornig HD, Buhmann C, et al. Lack of gender-specific
antibody recognition of products from domains of a var gene impli-
cated in pregnancy-associated Plasmodium falciparum malaria. Infect
34. Singh SK, Hora R, Belrhali H, Chitnis CE, Sharma A. Structural basis
for Duffy recognition by the malaria parasite Duffy-binding-like do-
main. Nature 2006;439:741–4.
35. Trimnell AR, Kraemer SM, Mukherjee S, et al. Global genetic diversity
and evolution of var genes associated with placental and severe child-
hood malaria. Mol Biochem Parasitol 2006;148:169–80.