Immune signaling pathways regulating bacterial and malaria parasite infection of the

mosquito Anopheles gambiae

Stephan Meister, Stefan M. Kanzok, Xue-li Zheng, Coralia Luna, Tong-Ruei Li, Ngo T. Hoa,
John Randall Clayton, Kevin P. White, Fotis C. Kafatos, George K. Christophides, and Liangbiao
Zheng
PNAS 2005;102;11420-11425; originally published online Aug 2, 2005;
doi:10.1073/pnas.0504950102
This information is current as of May 2007.

Online Information High-resolution figures, a citation map, links to PubMed and Google Scholar,
& Services etc., can be found at:
www.pnas.org/cgi/content/full/102/32/11420
Supplementary Material Supplementary material can be found at:
www.pnas.org/cgi/content/full/0504950102/DC1
References This article cites 36 articles, 16 of which you can access for free at:
www.pnas.org/cgi/content/full/102/32/11420#BIBL
This article has been cited by other articles:
www.pnas.org/cgi/content/full/102/32/11420#otherarticles
E-mail Alerts Receive free email alerts when new articles cite this article - sign up in the box
at the top right corner of the article or click here.
Rights & Permissions To reproduce this article in part (figures, tables) or in entirety, see:
www.pnas.org/misc/rightperm.shtml
Reprints To order reprints, see:
www.pnas.org/misc/reprints.shtml

Notes:
Immune signaling pathways regulating bacterial
and malaria parasite infection of the mosquito
Anopheles gambiae
Stephan Meister*†‡, Stefan M. Kanzok‡§¶, Xue-li Zheng§储, Coralia Luna§, Tong-Ruei Li**, Ngo T. Hoa§,
John Randall Clayton*¶, Kevin P. White**, Fotis C. Kafatos*†, George K. Christophides*†,††‡‡, and Liangbiao Zheng§††‡‡
*European Molecular Biology Laboratory, Meyerhofstrasse 1, 69117 Heidelberg, Germany; and Departments of §Epidemiology and Public Health and
**Genetics, Yale University School of Medicine, 60 College Street, New Haven, CT 06520
Contributed by Fotis C. Kafatos, June 13, 2005
We show that, in the malaria vector Anopheles gambiae, expres-
sion of Cecropin 1 is regulated by REL2, an NF-␬B-like transcription
factor orthologous to Drosophila Relish. Through alternative splic-
ing, REL2 produces a full-length (REL2-F) and a shorter (REL2-S)
protein isoform lacking the inhibitory ankyrin repeats and death
domain. RNA interference experiments show that, in contrast to
Drosophila Relish, which responds solely to Gram-negative bacte-
ria, the Anopheles REL2-F and REL2-S isoforms are involved in
defense against the Gram-positive Staphylococcus aureus and the
Gram-negative Escherichia coli bacteria, respectively. REL2-F also
regulates the intensity of mosquito infection with the malaria
parasite, Plasmodium berghei. The adaptor IMD shares the same
activities as REL2-F. Microarray analysis identified 10 additional
genes regulated by REL2, including CEC3, GAM1, and LRIM1.
innate immunity 兩 NF-␬B 兩 Relish 兩 Cecropin 兩 Imd
rom insects to mammals, NF-␬B-like transcription factors play
F central roles in induction and regulation of innate immune
responses. In Drosophila, three NF-␬B-like proteins have been
identified. Dorsal is largely engaged in early developmental pro-
cesses, whereas Relish and Dif are involved in immune signaling
leading to transcriptional activation of genes, including those en-
coding antimicrobial peptides (AMPs) (1). The Toll pathway, which
is used, principally, to counteract infections by fungi (2) and
Gram-positive bacteria (3), is activated through nuclear transloca-
tion of Dif; one of the transcriptionally induced genes encodes the
AMP Drosomycin (4, 5). In contrast, the immunodeficiency (Imd)
pathway deals primarily with Gram-negative bacterial infections (6)
through nuclear translocation of Relish, leading to transcriptional
induction of an array of AMPs, including Cecropins A and B,
Attacin, Drosocin, and Diptericin (7).
Relish is a modular protein consisting of two main domains. The
Rel-homology domain (RHD) binds to DNA and activates gene
expression, whereas the inhibitory I␬B domain, harboring several
ankyrin (ANK) repeats, blocks the nuclear localization signal and,
thus, keeps Relish in the cytosol. After bacterial infection, the RHD
domain is separated from ANK by targeted proteolysis and is
translocated into the nucleus. Recent studies have shown that the
caspase Dredd is involved, directly or indirectly, in the proteolytic
activation of Relish (8). In the yellow fever mosquito Aedes aegypti,
three products of the Relish gene are produced via alternative
splicing and encode the full-length protein, the RHD, and the ANK
domain (9). Activation of the Aedes full-length protein presumably
requires proteolytic processing, as in Drosophila.
The Anopheles gambiae genome sequence (10) has allowed us to
compare putative immunity genes between Anopheles and Dro-
sophila (11). The majority of intracellular components of the Toll
and Imd pathways are conserved, but several differences exist. Most
striking is the absence of Dif in Anopheles, suggesting that a
functional Toll pathway would require REL1 (the orthologue of
Dorsal, previously designated as Gambif1) (12) or REL2 (the
orthologue of Relish) to substitute for Dif. Here, we characterize
11420 –11425 兩 PNAS 兩 August 9, 2005 兩 vol. 102 兩 no. 32
the Anopheles gambiae REL2 and show that it regulates expression
of the AMP genes CEC1, CEC3, and GAM1 and a key parasite
antagonist, LRIM1 (13). Two REL2 gene products are produced via
alternative splicing: A full-length form (REL2-F) includes the
carboxyl-terminal ANK and death domains, and a shorter form
(REL2-S) lacks these domains. RNAi-dependent gene silencing in
adult Anopheles mosquitoes revealed that REL2-F and IMD are
implicated in defense against the Gram-positive bacterium Staph-
ylococcus aureus, whereas REL2-S is involved in defense against the
Gram-negative bacterium Escherichia coli. Thus, through alterna-
tive splicing, Anopheles uses a single NF-␬B gene (REL2) to mediate
immune reactions, for which Drosophila employs two distinct genes,
Relish and Dif. We show that the Anopheles Imd pathway leading to
REL2-F activation is also involved in limiting Plasmodium parasite
infections. Thus, a classical immune signaling pathway is implicated
in parasite survival in the Anopheles mosquito vector. Finally,
REL2-F is involved in regulation of the melanization reaction;
silencing of REL2-F causes limited but consistent melanization of
Plasmodium ookinetes in the Anopheles midgut.
Materials and Methods
cDNA Production. Total RNA was extracted from 25 larvae, pupae,
or adult Anopheles gambiae and ⬇5,000 embryos with TRIzol
reagent (Invitrogen) and treated with DNaseI, and 0.2 ␮g were used
for reverse transcription using the Omniscript kit (Qiagen, Santa
Clarita, CA).
dsRNA Construction. PCR amplicons tailed with T7 promoter se-
quences (see Table 2, which is published as supporting information
on the PNAS web site) were used to synthesize dsRNAs with the
MEGAshortscript kit (Ambion) that were DNaseI-treated, cleaned
up by phenol:chloroform extraction, precipitated with isopropanol,
and resuspended in nuclease-free water.
Luciferase Assays. Cells were plated in 24-well plates and, upon
reaching ⬇70% confluence, were incubated for 30 min at room
temperature with 7 ␮g of dsRNA in FBS-free medium. Transfec-
tions with CEC1 or KIN1 reporter (300 ng per well) and reference
Abbreviations: AMP, antimicrobial peptide; ANK, ankyrin; Imd, immunodeficiency; KD,
knockdown; RHD, Rel-homology domain; RNAi, interfering RNA.
†Presentaddress: Department of Biological Sciences, Imperial College London, SW7 2AZ
London, United Kingdom.
‡S.M. and S.M.K. contributed equally to this work.
¶Present address: Department of Molecular Microbiology and Immunology, Johns Hopkins
Bloomberg School of Public Health, Baltimore, MD 21205.
储Present
address: Department of Parasitology, Southern Medical University, Guangzhou,
Guangdong Province, China.
††G.K.C. and L.Z. contributed equally to this work.
‡‡To whom correspondence may be addressed. E-mail: g.christophides@imperial.ac.uk or
liangbiao.zheng@yale.edu.
© 2005 by The National Academy of Sciences of the USA
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0504950102
Fig. 1. Genomic organization and alternative splicing of the Anopheles REL2 gene. Introns (lines) and exons (open boxes) are shown, based on Ensembl gene
prediction (A) and actual data (B). The length (bp) of introns and exons is shown below and above the lines and boxes, respectively. E, putative ␬B elements.
(C) REL2 has two alternative 5⬘ exons. The 5⬘ end of REL2-S has not been confirmed and is indicated by dashed intron lines. Alternative splicing occurs at the 3⬘
end with a REL2-S-specific exon. Protein domains are depicted by shades of gray inside the exons. Q兾H, glutamine and histidine-rich region; RHD, Relish homology

IMMUNOLOGY
domain; IPT兾TIG, DNA-binding domain; NLS, nuclear localization signal; ANK, ANK-repeat domain; DD, death domain. Putative start- and stop-codons and target
regions of dsRNA constructs (RHD dsRNA and ANK dsRNA) are indicated.

pAct5C-Renilla (3 ng per well) DNA constructs were performed
with the Qiagen Effectene kit (14). Cells were harvested 16–18 h
later and subjected to Dual Luciferase (Promega) assays according
to the manufacturer’s instructions. Relative light units per second
were measured in a luminometer (Berthold, Bundoora, Australia).
Gene Silencing in Adult Anopheles gambiae. One- to two-day-old
adult female mosquitoes were injected with 207 ng of dsRNA, as
described in ref. 15, and, after 4 days, was subjected to infection
assays.
Immune Challenge in Mosquito Cell Lines. Lipopolysaccharides from
E. coli O26:B6 (Sigma Aldrich) were dissolved in water and added
to cells at 10 ␮g兾ml. Micrococcus luteus and E. coli (DH5␣) bacteria
and Beauveria bassiana spores (American Type Culture Collection)
were heat-inactivated by boiling for 5 min in PBS and added to cells
at 100 microbes per cell.
Bacterial and Malaria Infection of Adult Anopheles gambiae. Bacteria
were grown to OD600 ⫽ 0.7, precipitated, washed, and resuspended
in PBS. Bacterial suspensions of 69 nl (E. coli at OD600 ⫽ 0.01 and
S. aureus at OD600 ⫽ 0.4) were injected into the mosquito thorax
with a nanoinjector (Nanoject, Drummond Scientific, Broomall,
PA). For malaria infections, mosquitoes were fed for 30 min at 21°C
on CD1 mice infected with transgenic GFP-fluorescing Plasmo-
dium berghei (16, 17).
Quantitative RT-PCR. cDNA was produced as described above.
Transcript abundance was measured with an Applied Biosystems
7500 Real-Time PCR system according to the manufacturer’s
instructions. PCRs of 25 ␮l consisted of 1⫻ SYBR green mix, 900
nM forward and reverse primers, and cDNA, corresponding to 0.02
␮g of total RNA. Primer sequences are shown in Table 2.
Microarray Analysis. Slides were printed as described in ref. 18;
prehybridized in 5⫻ standard saline citrate (SSC) (1⫻ SSC ⫽ 0.15
M sodium chloride兾0.015 M sodium citrate, pH 7), 0.1% SDS, and
1% BSA at 42°C for 45 min; washed at room temperature with
deionized water; dipped in 100% isopropanol; and dried. mRNA
samples were prepared by using the PolyTract isolation system
(Promega) and labeled as described in ref. 19. Equal volumes of
Cy-3- and Cy-5-labeled samples were combined in hybridization
buffer supplemented with poly(A)-DNA and hybridized to mi-
croarray slides at 42°C overnight. Hybridized arrays were washed,
air-dried, and scanned by using the ScanArray 3000 (GSI Lumonics,
Billerica, MA). Two independent experiments with dye swaps were
performed. Expression ratios were calculated and analyzed by using
the TM4 microarray software package (www.tigr.org兾software兾
tm4兾). After spot quality filtering, expression data were normalized
by using the Lowess (Locfit) algorithm, and clustered with k-mean
and hierarchical clustering with the program TMEV (The Institute
for Genomic Research) (19). Elements with at least 500 signal
intensity values were considered.
Results
Genomic Organization and Expression of REL2. RT-PCR and RACE
using RNA from adult female Anopheles gambiae were used to
determine the structural organization of REL2. This gene consists
of 11 exons, encompassing 10.9 kb (Fig. 1). It encodes putative
protein(s) with a glutamine- and histidine-rich region, potentially
implicated in protein–protein interactions, followed by RHD, IPT兾
TIG (DNA binding), ANK and death domains, and a nuclear
localization signal.
We detected two REL2 transcripts: full-length REL2-F, and a
shorter form, REL2-S, encompassing a unique exon at its 3⬘ end.
For mRNA splicing, this exon uses a splice acceptor that is 4 bp 3⬘
upstream to that used in REL2-F and results in a downstream early
termination codon. Thus, REL2-F encodes a protein with all of the
above features, whereas REL2-S is missing the ANK and death
domains. Additional alternative splicing at the 5⬘ end of REL2 may
also produce two transcript variants for each form (Fig. 1C).
REL2-F shows significant similarity to the Aedes (55%) and
Drosophila (29%) Relish (see Fig. 7, which is published as support-
ing information on the PNAS web site).
REL2-F and -S transcripts are expressed constitutively through-
out Anopheles gambiae development, albeit at different levels:
REL2-F is expressed more strongly (Fig. 2A). Both transcripts are
also expressed in cultured cell lines, Sua1B (Fig. 2B) and 4a3A (data
not shown). Cells challenged with heat-killed E. coli (Gram-
negative) or purified E. coli lipopolysaccharides (LPS) up-regulate

Meister et al. PNAS 兩 August 9, 2005 兩 vol. 102 兩 no. 32 兩 11421

 Fig. 2. Expression profile of REL2 in mosquito cultured cells
and developmental stages. (A) Relative abundance of REL2-F
and REL2-S transcripts at different stages of mosquito devel-
opment. (B) Transcriptional regulation of REL2 transcripts in
immune-challenged cultured Sua1B cells. REL2-F was detected
by using primer pairs targeting the ANK domain. REL2-S was
detected with one primer specific to the REL2-S transcript (see
Materials and Methods). PCR conditions: 30 cycles, 60°C an-
nealing temperature, 1.5 min extension time, and 2 ␮g of total
RNA used per reaction. *, a band which includes the intron
sequence of REL2-S. It is unclear whether this transcript is part
of an additional splice variant of REL2. The identity of REL2-S
bands was confirmed by sequencing. Control RT-PCR with rpS7
showed no genomic DNA contamination. CTRL, control; B.b.,
B. bassiana; M.l., M. luteus; E.c., E. coli; 么, males; 乆, females.

REL2-F significantly and REL2-S to a lower level. We cannot (⬇60% and 68%, respectively), although not as efficiently as REL2
exclude the possibility that induction after LPS challenge was due KD. This is consistent with evidence from Drosophila imd mutants,
to the common peptidoglycan contaminants of LPS preparations in which Cecropin A expression is reduced but not abolished (23).
(20). No significant up-regulation was detected after cell challenge Surprisingly but consistently, silencing of CASPL1 and IKK1
with the fungus Beauveria bassiana or the Gram-positive bacterium or IKK2 had opposite effects in the two cell lines. In Sua1B cells,
M. luteus. KD of CASPL1, but not of IKK1 or IKK2, substantially reduced
CEC1 promoter expression, whereas, in 4a3A cells, KD of IKK1
Expression of CEC1 Is Regulated by REL2-F. A putative AMP target or IKK2, but not of CASPL1, reduced CEC1 promoter expres-
gene of Anopheles immune signaling pathways is CEC1 (previously sion. We used semiquantitative RT-PCR to estimate the level of
designated as Cecropin A) (14, 21). Compared with Drosophila gene silencing (Fig. 4A). Importantly, KD of IKK1 and IKK2 in
Cecropins, which are mostly active against Gram-negative bacteria, Sua1B was highly effective, suggesting that these genes are not
CEC1 has a wide spectrum of antibacterial activities and is induced implicated in the regulation of CEC1 in Sua1B cells. KD of
by Plasmodium and both Gram types of bacteria (21). Sua1B and CASPL1 in the same cells was less effective (⬇40%) and, yet,
4a3A cells have high levels of constitutive CEC1 expression that are
associated with substantial reduction in CEC1 activity (Fig. 3B).
further enhanced by treatment with heat-killed E. coli or M. luteus
A plausible explanation that requires further investigation is that
(data not shown). We examined the effect of REL proteins on
IKK1兾IKK2 and CASPL1 are implicated in alternative branches
CEC1 gene expression in Sua1B cells after transfection with a
CEC1-promoter–luciferase-reporter construct (14), silencing of
REL1 and REL2 genes by RNAi, and subsequent monitoring of
luciferase activity. Targeting either the RHD or ANK domains
of REL2 resulted in substantial reduction of luciferase activity,
typically 5- to 10-fold. In contrast, REL1 knockdown (KD) had no
significant effect (Fig. 3A).
Because Anopheles gambiae shows no transitive RNAi effects,
i.e., dsRNAs are effective only against mRNAs bearing the corre-
sponding sequences (22), targeting the RHD should silence both
REL2 forms, whereas ANK-domain KD should affect only
REL2-F. We observed that KDs using ANK dsRNA result in a
major (⬇91%) reduction in CEC1-promoter expression, essentially
indistinguishable from that observed with RHD dsRNA. Thus, it
appears that REL2-F largely controls constitutive CEC1-promoter
expression in Sua1B cells, whereas REL2-S has no major effect
(Fig. 3A). Similar results were obtained for the 4a3A cells (data not
shown). Proteolytic cleavage is necessary to remove the inhibitory
ANK domain of Drosophila Relish, and this processing would also
be expected for REL2-F activation. By extension, these results
suggest that REL2-F is constitutively activated in these mosquito
cells.
Other Anopheles orthologues of Drosophila Imd-pathway com-
ponents, the Imd adaptor (IMD) and the caspase Dredd
(CASPL1), were identified in ref. 11; orthologues of IKK␤ (IKK1)
and IKK␥ (IKK2) are reported here (see Table 3, which is published
Fig. 3. CEC1 promoter activity in cultured Anopheles cells. (A) REL2, but not
as supporting information on the PNAS web site). To examine the
REL1, is required for expression of CEC1 promoter in Sua1B cells, which is
degree of functional Imd-pathway conservation between Anopheles equally affected by dsRNAs encompassing the RHD and the ANK-repeats
and Drosophila, we silenced these genes in Sua1B and 4a3A cells by domain. (B) REL2, IMD, and CASPL1, but not IKK1 or IKK2, are required for
using RNAi and observed significant down-regulation (Fig. 3 B and CEC1 expression in Sua1B cells. (C) REL2, IMD, IKK1, and IKK2, but not CASPL1,
C). In both cell lines, IMD KD reduced CEC1-promoter expression are required for CEC1 expression in 4a3A cells. CTRL, control.

11422 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0504950102 Meister et al.

 genesis, a protein disulfide isomerase, KIN1 (see below), and
LRIM1, a leucine-rich repeat immune protein with strong anti-
parasitic activity (13). The CEC3 gene is tightly clustered with
CEC1 and CEC2, and its promoter region includes two NF-␬B-
binding sites (14), suggesting regulatory similarities to CEC1. At
least 1, and as many as 10, ␬B elements were detected in adjacent
genomic regions of all regulated genes (see Table 4, which is
published as supporting information on the PNAS web site).
KIN1 encodes a short mature cationic polypeptide with a kinino-
gen domain, which is enriched in histidine (26 residues, 36%), plus
an amino-terminal signal peptide. KIN1 is strongly up-regulated by
immune challenges (18) and Plasmodium infection (25). Examina-
tion of a KIN1-promoter luciferase construct in Sua1B cells showed
that REL2 is, indeed, required for KIN1 expression (see Fig. 8,
which is published as supporting information on the PNAS web
site).

REL2 Is Required for Antibacterial Defense of Adult Anopheles gam-

Fig. 4. Efficacy of RNAi-mediated gene silencing. (A) KD of IKK1, IKK2, and
biae. We investigated, by RNAi analysis, the possible role of REL2
and other components of the Imd pathway in defense against
bacterial infection in adult mosquitoes. DsRNA was injected into
newly emerged adult females that were infected, 4 days later, with
bacteria representative of the Gram-positive (S. aureus) or the
Gram-negative (E. coli) types, as described in ref. 15. Real-time

CASPL1 genes in Sua1B cultured cells. RNAi effectiveness was determined by
semiquantitative RT-PCR (Left). Quantification of PCR products (Right) was
performed by using rpS7 as an internal reference and untreated cells as a
calibrator. (B) REL2 gene silencing in adult mosquitoes after injections with
either RHD or ANK dsRNA. Relative REL2-S and REL2-F transcript levels in KD
compared with control GFP dsRNA-treated mosquitoes were determined in
the carcass, midgut, and whole mosquitoes 4 days after injections by using
quantitative RT-PCR.
of the CEC1-activation pathway that are differentially active in
these two cell types.
REL2 Regulates the Expression of AMPs and Other Immunity Genes.
We performed a DNA microarray analysis (18) to identify genes
regulated by REL2 in Anopheles 4a3A cells by comparing the
expression profiles of cells treated with either REL2 (RHD domain)
or control lacZ dsRNAs. Cells were challenged, 24 h after dsRNA
addition, by a 12-h exposure to heat-killed E. coli, and expression
profiles were assessed at 0, 2, 6, and 12 h. From 3,840 EST clones
present on the microarray, nine were consistently down-regulated
in REL2 KD cells by at least 1.7-fold and one by 1.2-fold (Table 1).
Six of these genes have been categorized previously as defense兾
immunity-related (24). They encode the AMPs CEC3 and GAM1
(previously CecB and gambicin, respectively), the serine protease
CLIPB14, another two CLIP-domain serine proteases, a fibrinogen-
domain lectin, a Brix-domain protein implicated in ribosome bio-
IMMUNOLOGY
quantitative RT-PCR confirmed that RHD dsRNA injection si-
lences both REL2-S and REL2-F transcripts equally (⬇50%),
whereas ANK dsRNA decreases the levels of only REL2-F (Fig.
4B). Interestingly, REL2-S transcript levels were up-regulated by
1.5-fold in dsANK-injected mosquitoes.
Mosquito survival was assessed daily for 7 days after bacterial
infection (Fig. 5). Simultaneous KD of REL2-F and REL2-S
transcripts (RHD dsRNA) significantly compromised mosquito
defense against both S. aureus and E. coli. In contrast, KD of
REL2-F alone (ANK dsRNA) reduced the survival of S. aureus- but
not of E. coli-infected mosquitoes. Similar results were obtained
with IMD KD, suggesting that IMD acts in concert with REL2-F in
defense against S. aureus. Survival of CASPL1-KD mosquitoes did
not detectably decrease after infection (data not shown). Prelimi-
nary results indicated that KD of REL1 had no measurable effect
on mosquito survival, suggesting that REL2 is sufficient to confer
resistance to bacterial infections.
The Anopheles Imd Pathway Regulates Malaria Parasite Infection. It
has been reported previously that CEC1 and other AMP genes are
up-regulated when Plasmodium crosses the Anopheles midgut (11,
21, 26). Furthermore, our microarray results indicated that REL2
regulates expression of the major Plasmodium antagonist LRIM1.
We tested the effect of REL2 and IMD KD on P. berghei survival
to the oocyst stage in the mosquito midgut. At 4 days after dsRNA
treatment, mosquitoes were infected with GFP-marked parasites

Table 1. Effect of REL2 pathway gene silencing on parasite development

Gene KD Midguts Parasites per midgut Prevalence, % Mel, % P (KS) P (t)

Control (5) 79 86 87.34 0.07

IMD (5) 79 153 96.20 0.13 0.044 0.017
Control (4) 49 51 87.76 0.63
REL2-RHD (4) 49 104 91.84 4.37 0.147 0.035
Control (3) 42 82 97.62 0.00
REL2-ANK (3) 42 166 95.24 4.87 0.093 0.008

Numbers of oocysts and melanized parasites are reported as three experimental datasets for knocked-down
genes or REL2 domains and control dsRNA-treated mosquitoes (GFP). The number of replicates is indicated in
parenthesis in the first column. Each replicate used different batches of mosquitoes, which were fed on the same
infected mouse. Identical numbers of midguts from control and KD mosquitoes were assessed for each replicate
(see also Table 5). P values indicate whether the distributions of parasites in KD and control midguts in each
dataset are similar and were determined by KS and Student’s t test. Mel, melanized parasites (ookinetes); KS,
Kolmogorov–Smirnov test. Parasites per midgut are oocysts and melanized ookinetes.

Meister et al. PNAS 兩 August 9, 2005 兩 vol. 102 兩 no. 32 兩 11423


Fig. 5. Mosquito survival to bacterial infection. Survival of adult mosquitoes
to Gram-positive S. aureus (A) and Gram-negative E. coli (B) infections after KD
of IMD, REL2-RHD, and REL2-ANK. GFP dsRNA-treated mosquitoes were used
as a control. The survival assay was repeated at least 3 and up to 10 times per
gene. Error bars represent the standard error.
(16, 17), and oocysts were counted 6–8 days later. Silencing of
REL2 (RHD dsRNA) or IMD resulted in a statistically significant
twofold increase of oocyst numbers, suggesting that the pathway is,
indeed, implicated in parasite killing before or during midgut
invasion (Fig. 6, Table 1; and see Table 5, which is published as
supporting information on the PNAS web site). Quantitatively
similar results were obtained when silencing REL2-F alone (ANK
dsRNA), indicating that REL2-F, rather than REL2-S, is impli-
cated in this reaction. Both REL2-RHD and REL2-ANK, but not
IMD KDs, led to limited occurrence of parasite melanization (⬇5%
of the ookinetes). This finding would suggest that REL2-F is
involved in the control of the melanization reaction, in a manner
independent of IMD.
Discussion
Previously, our comparative genomic analysis of immunity-related
genes in the fruit fly Drosophila melanogaster and the malaria
mosquito Anopheles gambiae (11) revealed that, in general, intra-
cellular components of immune signaling pathways are well con-
served between the two insects. However, differences were noted,
particularly in potential components of the Toll pathway, which may
indicate important variations in immune signaling. Importantly, the
NF-␬B-like transcription factor Dif, which activates gene expression
upon recognition of Gram-positive bacteria and fungi, has no
orthologue in Anopheles. This may imply that in a putative Toll
pathway, one of the other two Anopheles NF-␬B-like proteins,
REL1 (the Dorsal orthologue) or REL2 (the Relish orthologue),
may substitute for the role of Dif. Furthermore, Toll genes have
evolved through independent gene duplication and sequence di-
vergence, and, thus, the Drosophila Toll receptor has no clear
orthologue in Anopheles. In contrast, intracellular components of
11424 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0504950102
Fig. 6. REL2- and IMD-mediated Plasmodium killing and melanization in the
mosquito vector. (A) Averaged P. berghei oocyst load per midgut after KD of
IMD, REL2-ANK, and REL2-RHD. Oocyst load after individual gene KD was
normalized relative to oocyst load of parallel dsGFP treatment and presented
as a percentage. Every gene KD was repeated three times or more. Black
sections of columns represent melanized ookinete fraction. Error bars repre-
sent the standard error. (B) Representative sections of infected Anopheles
gambiae midguts infected with GFP-tagged P. berghei. Arrows point to
melanized ookinetes.
the Imd pathway, which is activated mainly in response to Gram-
negative bacteria, are highly conserved between the two insects.
Diversification of proteins implicated in immune recognition
upstream of both pathways is very pronounced. This category
includes peptidoglycan-recognition proteins (PGRPs) and Gram-
negative binding proteins. For example, the Drosophila PGRP-LC,
the main receptor of the Imd pathway (27, 28), and its Anopheles
orthologue have similar architectures, but their PGRP domains
have apparently evolved independently from a single domain-
containing ancestral protein (11). Thus, recognition properties of
PGRP-LC could differ substantially between the two species.
Interestingly, the other Imd-pathway-associated receptor, PGRP-
LE, has no Anopheles orthologue.
Here, we provide evidence from detailed gene-structure deter-
mination and reverse genetic analysis of function, illuminating the
structural and functional evolution of immune signaling in Dro-
sophila and Anopheles gambiae. The first major difference we
detected in Anopheles is alternative splicing of the REL2 gene. This
gives rise to two isoforms, REL2-F and REL2-S, which are differ-
entially involved in defense against Gram-positive and Gram-
negative bacteria, respectively. In this way, the single mosquito gene
REL2 mediates alternative immune responses, which, in the fly,
require two genes: REL2’s orthologue Relish and Dif. Thus, through
posttranscriptional processing of REL2, Anopheles appears to com-
pensate for the absence of Dif.
In addition to this gene substitution, we discovered a switch in the
roles of the NF-␬B transcriptional factors during the ⬇250 million
years of evolution (29, 30) since the last common ancestor of
Meister et al.
Drosophila and Anopheles. The Anopheles REL2-F is used for
defense against the Gram-positive bacterium S. aureus, whereas the
orthologous and structurally similar Relish in the fruit fly is
required for defense against Gram-negative bacteria. Likewise,
REL2-S deals with Anopheles defense against the Gram-negative
bacterium E. coli, whereas Dif, which, like REL2-S, does not have
an inhibitory domain, deals with fruit fly responses against Gram-
positive bacteria. However, both REL2-F and Relish mediate their
functions in concert with the death-domain adaptor IMD, indicat-
ing that the pathway structure has remained conserved.
REL2-F, unlike its fly counterpart Relish but similar to ortholo-
gous genes in other higher eukaryotes, encodes a death domain
located at the carboxyl terminus of the deduced protein. This
domain might be used for oligomerization with the respective
domain of IMD or other death-domain-encoding proteins. Perti-
nent information is available from studies on the Relish gene of
Aedes aegypti, the vector of yellow and dengue fever, which belongs
to a different mosquito subfamily (Culicinae rather than Anopheli-
nae). The Relish gene of Aedes is structurally similar to Anopheles
REL2, in that it encodes a death domain. Aedes REL2 produces
three alternatively spliced transcripts: two are similar to Anopheles
REL2-F and REL2-S, and the third encodes a protein encompass-
ing only the ANK domain (9).
In Aedes, transgenic overexpression of a truncated version of the
REL2-S homologue (C8), which lacks the putative glutamine- and
histidine-rich transactivator domain, results in susceptibility to
Gram-negative but not to Gram-positive bacteria (31). Additional
studies are required to determine whether C8 acts as a dominant-
negative allele of REL2-S. In this case, our postulated switch in
pathway function would probably have occurred in the common
ancestor of anopheline and culicine mosquitoes.
Whether Aedes, like Anophele, also lacks Dif is not yet resolved.
However, recent studies have shown that the Aedes orthologue of
REL1 is not involved in defense against bacteria but only against
fungal infections (32). Likewise, our preliminary results in Anoph-
eles indicate that REL2 is the only NF-␬B factor implicated in the
antibacterial defense against E. coli and S. aureus. Thus, it is likely
that the two mosquitoes use similar strategies to deal with bacterial
infections.
Control of NF-␬B activation is achieved through inhibitory I␬B
domains by the use of two alternative strategies. In the fly, Relish
contains its own inhibitory I␬B domain (ANK) located at the
carboxyl terminus of the protein, whereas the inhibitory domain for
Dif is provided by another protein, Cactus. Accordingly, Anopheles
REL2-S, in noninduced condition, might be constitutively acti-
1. Hoffmann, J. A. & Reichhart, J. M. (2002) Nat. Immunol. 3, 121–126.
2. Lemaitre, B., Nicolas, E., Michaut, L., Reichhart, J. M. & Hoffmann, J. A. (1996) Cell 86, 973–983.
3. Michel, T., Reichhart, J. M., Hoffmann, J. A. & Royet, J. (2001) Nature 414, 756–759.
4. Meng, X., Khanuja, B. S. & Ip, Y. T. (1999) Genes Dev. 13, 792–797.
5. Rutschmann, S., Jung, A. C., Hetru, C., Reichhart, J. M., Hoffmann, J. A. & Ferrandon, D.
(2000) Immunity 12, 569–580.
6. Lemaitre, B., Kromer-Metzger, E., Michaut, L., Nicolas, E., Meister, M., Georgel, P.,
Reichhart, J. M. & Hoffmann, J. A. (1995) Proc. Natl. Acad. Sci. USA 92, 9465–9469.
7. Silverman, N., Zhou, R., Stoven, S., Pandey, N., Hultmark, D. & Maniatis, T. (2000) Genes
Dev. 14, 2461–2471.
8. Stoven, S., Silverman, N., Junell, A., Hedengren-Olcott, M., Erturk, D., Engstrom, Y.,
Maniatis, T. & Hultmark, D. (2003) Proc. Natl. Acad. Sci. USA 100, 5991–5996.
9. Shin, S. W., Kokoza, V., Ahmed, A. & Raikhel, A. S. (2002) Proc. Natl. Acad. Sci. USA 99, 9978–9983.
10. Holt, R. A., Subramanian, G. M., Halpern, A., Sutton, G. G., Charlab, R., Nusskern, D. R.,
Wincker, P., Clark, A. G., Ribeiro, J. M., Wides, R., et al. (2002) Science 298, 129–149.
11. Christophides, G. K., Zdobnov, E., Barillas-Mury, C., Birney, E., Blandin, S., Blass, C., Brey,
P. T., Collins, F. H., Danielli, A., Dimopoulos, G., et al. (2002) Science 298, 159–165.
12. Barillas-Mury, C., Charlesworth, A., Gross, I., Richman, A., Hoffmann, J. A. & Kafatos,
F. C. (1996) EMBO J. 15, 4691–4701.
13. Osta, M. A., Christophides, G. K. & Kafatos, F. C. (2004) Science 303, 2030–2032.
14. Zheng, X. L. & Zheng, A. L. (2002) Insect Mol. Biol. 11, 517–525.
15. Blandin, S., Moita, L. F., Kocher, T., Wilm, M., Kafatos, F. C. & Levashina, E. A. (2002)
EMBO Rep. 3, 852–856.
16. Vlachou, D., Zimmermann, T., Cantera, R., Janse, C. J., Waters, A. P. & Kafatos, F. C.
(2004) Cell. Microbiol. 6, 671–685.
17. Franke-Fayard, B., Trueman, H., Ramesar, J., Mendoza, J., van der Keur, M., van der Linden,
R., Sinden, R. E., Waters, A. P. & Janse, C. J. (2004) Mol. Biochem. Parasitol. 137, 23–33.
18. Dimopoulos, G., Christophides, G. K., Meister, S., Schultz, J., White, K. P., Barillas-Mury,
C. & Kafatos, F. C. (2002) Proc. Natl. Acad. Sci. USA 99, 8814–8819.
Meister et al.
vated, inhibited by another I␬B-like protein, or inhibited by
REL2-F through heterodimer formation. The detected low con-
stitutive expression of the CEC1 promoter in Anopheles cells is
caused by REL2-F and not REL2-S, suggesting that REL2-S is
normally associated with an inhibitory protein. Further studies are
required to determine the inhibitory partner of REL2-S. Our
preliminary data suggest that the putative partner is not CACT, the
orthologue of Cactus (data not shown).
Anopheles gambiae is the main vector of human malaria in
subSaharan Africa, causing death to more than one million people
every year. An important finding of our work is that the IMD兾
REL2-F signaling cascade is involved in limiting the number of
P. berghei oocysts that develop in the mosquito midgut. This
reaction could be due to the detected low constitutive activation of
REL2-F, to pathway activation by Plasmodium, or both. A recent
study has shown that ectopic overexpression of CEC1 drastically
reduces susceptibility of mosquitoes to Plasmodium (33). Other
REL2-regulated proteins that might explain the observed pheno-
type are Gambicin (GAM1), another AMP with antiparasitic
activity in vitro (34), and LRIM1, a key antagonist of ookinete-to-
oocyst development (13).
The silencing of REL2-F also leads to melanization of some
ookinetes during their passage through the Anopheles midgut
IMMUNOLOGY
epithelium. This reaction is independent of IMD, pointing to either
an alternative signaling cascade or to the detected REL2-F con-
stitutive activation. In Drosophila, the Imd pathway seems to
regulate melanization upstream of Serpin27A (35), which is an
inhibitor of the final steps of prophenoloxidase activation and is
controlled by the Toll pathway (36).
In conclusion, our data suggest significant divergence of immune
signaling between the Anopheles mosquito and the fruit fly D. mela-
nogaster. The documented differences most likely reflect different
lifestyles and, consequently, different infectious agents that the two
insects encounter during their lifetimes. In mosquitoes, one of these
agents is the malaria parasite Plasmodium.
We thank G. Dimopoulos and members of our and the Aksoy labora-
tories for helpful discussions and the use of equipment and Dr. Gareth
Lycett for critical reading of the manuscript. This work was supported,
in part, by National Institutes of Health Grants R01 A43035 (to L.Z.),
P01 AI044220-06A1 (to F.C.K.), and K22 HG00045 (to K.P.W.); by
European Union Grant HPRN-CT-2000-00080 (to F.C.K.); and by the
European Molecular Biology Laboratory. S.M.K. was supported, in part,
by the Deutsche Forschungsgemeinschaft, and N.T.H. was supported by
Wellcome Trust Fellowship 064874.
19. Hegde, P., Qi, R., Abernathy, K., Gay, C., Dharap, S., Gaspard, R., Hughes, J. E., Snesrud,
E., Lee, N. & Quackenbush, J. (2000) BioTechniques 29, 548–550, 552–554, and 556.
20. Leulier, F., Parquet, C., Pili-Floury, S., Ryu, J. H., Caroff, M., Lee, W. J., Mengin-Lecreulx,
D. & Lemaitre, B. (2003) Nat. Immunol. 4, 478–484.
21. Vizioli, J., Bulet, P., Charlet, M., Lowenberger, C., Blass, C., Muller, H. M., Dimopoulos,
G., Hoffmann, J., Kafatos, F. C. & Richman, A. (2000) Insect Mol. Biol. 9, 75–84.
22. Hoa, N. T., Keene, K. M., Olson, K. E. & Zheng, L. (2003) Insect Biochem. Mol. Biol. 33,
949–957.
23. Onfelt Tingvall, T., Roos, E. & Engstrom, Y. (2001) EMBO Rep. 2, 239–243.
24. Dimopoulos, G., Casavant, T. L., Chang, S., Scheetz, T., Roberts, C., Donohue, M., Schultz,
J., Benes, V., Bork, P., Ansorge, W., et al. (2000) Proc. Natl. Acad. Sci. USA 97, 6619–6624.
25. Nachou, D., Schlegelmilch, T., Christophides, G. K. & Kafatos, F. C. (2005) Curr. Biol. 15,
1185–1195.
26. Dimopoulos, G., Richman, A., Muller, H. M. & Kafatos, F. C. (1997) Proc. Natl. Acad. Sci.
USA 94, 11508–11513.
27. Choe, K. M., Werner, T., Stoven, S., Hultmark, D. & Anderson, K. V. (2002) Science 296, 359–362.
28. Gottar, M., Gobert, V., Michel, T., Belvin, M., Duyk, G., Hoffmann, J. A., Ferrandon, D.
& Royet, J. (2002) Nature 416, 640–644.
29. Yeates, D. K. & Wiegmann, B. M. (1999) Annu. Rev. Entomol. 44, 397–428.
30. Gaunt, M. W. & Miles, M. A. (2002) Mol. Biol. Evol. 19, 748–761.
31. Shin, S. W., Kokoza, V., Lobkov, I. & Raikhel, A. S. (2003) Proc. Natl. Acad. Sci. USA 100, 2616–2621.
32. Shin, S. W., Kokoza, V., Bian, G., Cheon, H. M., Kim, Y. J. & Raikhel, A. S. (2005) J. Biol.
Chem. 280, 16499–16507.
33. Kim, W., Koo, H., Richman, A. M., Seeley, D., Vizioli, J., Klocko, A. D. & O’Brochta, D. A.
(2004) J. Med. Entomol. 41, 447–455.
34. Vizioli, J., Bulet, P., Hoffmann, J. A., Kafatos, F. C., Muller, H. M. & Dimopoulos, G. (2001)
Proc. Natl. Acad. Sci. USA 98, 12630–12635.
35. Takehana, A., Yano, T., Mita, S., Kotani, A., Oshima, Y. & Kurata, S. (2004) EMBO J. 23, 4690–4700.
36. Ligoxygakis, P., Pelte, N., Ji, C., Leclerc, V., Duvic, B., Belvin, M., Jiang, H., Hoffmann, J. A.
& Reichhart, J. M. (2002) EMBO J. 21, 6330–6337.
PNAS 兩 August 9, 2005 兩 vol. 102 兩 no. 32 兩 11425