Wolbachia density and virulence attenuation after
transfer into a novel host
E. A. McGraw*†, D. J. Merritt‡, J. N. Droller*, and S. L. O’Neill*†§
*Section of Vector Biology, Department of Epidemiology and Public Health, Yale University School of Medicine, 60 College Street, New Haven, CT 06520;
and‡Department of Zoology and Entomology, University of Queensland, St. Lucia, Qld 4072, Australia
Edited by Margaret G. Kidwell, University of Arizona, Tucson, AZ, and approved January 3, 2002 (received for review September 4, 2001)
The factors that control replication rate of the intracellular bacte-
rium Wolbachia pipientis in its insect hosts are unknown and
difficult to explore, given the complex interaction of symbiont and
host genotypes. Using a strain of Wolbachia that is known to
over-replicate and shorten the lifespan of its Drosophila melano-
gaster host, we have tracked the evolution of replication control in
both somatic and reproductive tissues in a novel host?Wolbachia
association. After transinfection (the transfer of a Wolbachia strain
into a different species) of the over-replicating Wolbachia popcorn
strain from D. melanogaster to Drosophila simulans, we demon-
strated that initial high densities in the ovaries were in excess of
what was required for perfect maternal transmission, and were
fitness costs associated with ovary infection rapidly declined in the
generations after transinfection. The early death effect in D.
simulans attenuated only slightly and was comparable to that
induced in D. melanogaster. This study reveals a strong host
involvement in Wolbachia replication rates, the independence of
density control responses in different tissues, and the strength of
natural selection acting on reproductive fitness.
lularly in the ovaries of all described hosts, as well as in a number
of other tissues depending on the particular Wolbachia?host
combination (1, 2). Infection of the host germ line enables
transovarial transmission as well as induction of a number
of reproductive abnormalities. These include cytoplasmic in-
compatibility (CI), parthenogenesis, feminization, and male
killing, all of which enhance the spread of Wolbachia in host
In Drosophila simulans and Drosophila melanogaster, most
infections induce CI (4), where Wolbachia modify developing
sperm such that only the presence of the same Wolbachia strain
in the egg can rescue the modification, allowing successful
completion of karyogamy (5, 6) and the subsequent normal
development of the embryo. Uninfected females cannot rescue
the sperm modification, and so the development of their off-
spring is blocked. The result is that with each subsequent
generation, infection frequencies rise and fewer and fewer
uninfected females can successfully reproduce (7).
replication rates must be sufficiently high to ensure fidelity of
transovarial transmission, while being low enough to not cause
overt host pathology. This replication control can be considered
a defining feature that separates vertically transmitted symbionts
from horizontally transmitted pathogens. The mechanisms that
this control is influenced by host or bacterial genotype are not
well understood. Studies exploring the link between CI expres-
sion and Wolbachia infection densities have revealed that it is
possible to select for increased Wolbachia densities in both D.
melanogaster (8) and Nasonia. In the latter case, however,
heritability was limited, and densities appeared to autocorrect
around a mean from one generation to the next (9). The results
of transinfection experiments in Drosophila have argued for the
olbachia pipientis are common bacterial endosymbionts of
arthropods and filarial nematodes. They occur intracel-
presence of a host component by demonstrating that the same
Wolbachia strain is maintained at higher densities in D. simulans
than in D. melanogaster (8, 10), and that there may also be
species-specific colonization patterns of different tissues (11).
Variation in the behavior of different Wolbachia strains that
infect the same host species (15) supports the role of bacterial
genomic contributions. As such, it appears that control of
In nearly all reported cases, direct fitness costs suffered by the
host because of Wolbachia infection appear to be negligible (12)
and in some instances Wolbachia infections appear to be bene-
ficial (13, 14). An exception to this rule was recently described
in D. melanogaster (15), in which a Wolbachia strain was iden-
tified that caused a dramatic reduction in the lifespan of its host.
The strain, named popcorn (wMelPop), replicates to abnormally
high densities, especially in adult nervous and muscle tissue,
resulting in host cell pathology. Interactions between the pop-
corn strain and its host provide an ideal system for studying the
factors that regulate Wolbachia replication rate. We used em-
bryonic microinjection to transfer wMelPop from its natural D.
melanogaster host into an uninfected D. simulans genetic back-
ground. We were then able to determine the role of the popcorn
genotype in the over-replication phenotype and temporally track
the attenuation of fitness costs associated with popcorn in the
newly infected host.
Materials and Methods
Strain Nomenclature. Wolbachia?host species combinations are
named with a two-part designation, the first referring to the host
species (Dmel or Dsim for D. melanogaster or D. simulans,
respectively) and the second indicating the bacterial strain (i.e.,
wMelPop and wRi for the popcorn and Riverside infections,
respectively). Any strain name followed by a ‘‘T’’ (e.g., Dsim
wRiT) indicates that the original host has been cured of its
infection by standard methods of tetracycline treatment (16).
Embryonic Microinjection. Egg cytoplasm from Dmel wMelPop
(15) was microinjected into Dsim wRiT (16) by standard tech-
niques (8, 17, 18). Dsim wRiT was chosen as the recipient host
strain, because the infection it harbored before tetracycline
treatment was known to cause only slight reproductive costs to
its host (19) and to be capable of inducing strong CI expression
(16, 19). Embryos were incubated at 27°C for 48 h. Larvae were
transferred to vials containing fruit fly diet (Sigma) and moved
to 21°C for all subsequent rearing except where noted.
This paper was submitted directly (Track II) to the PNAS office.
Abbreviation: CI, cytoplasmic incompatibility.
†Present address: Department of Zoology and Entomology, University of Queensland,
St. Lucia, Qld 4072, Australia.
§To whom reprint requests should be addressed. E-mail: email@example.com.
The publication costs of this article were defrayed in part by page charge payment. This
article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C.
§1734 solely to indicate this fact.
March 5, 2002 ?
vol. 99 ?
Rearing and Screening for Infection Status. Surviving G0females
were crossed with Dsim wRiT males and used to establish
isofemale lines. After we found evidence of viable G1offspring,
females were killed and DNA was extracted by using the
STE?boil method (STE ? 100 mM NaCl?10 mM Tris?HCl, pH
8.0?1 mM EDTA) (20). Presence of Wolbachia in the samples
was detected by PCR using primers for the wsp gene (81F and
691R) (21). Amplification of DNA was carried out in 50 mM
KCl?10 mM Tris?HCl (pH 9.0)?0.1% Triton X-100?2.5 mM
MgCl2?0.25 mM each dNTP?0.5 ?M primers. The thermal
cycling regime was as follows; 3-min denaturation at 94°C, 35
cycles of denaturation at 94°C, annealing at 55°C, and extension
at 72°C each for 1 min, followed by an extra 10-min extension
step at 72°C. Reactions were run in either a PTC-200 Thermal
Cycler (MJ Research) or a Mastercycler (Eppendorf). Roughly
350 embryos were injected. All (9 of 9) surviving G0males and
7 of 12 surviving G0females were PCR positive for Wolbachia.
To ensure the stability of the infection in early generations,
screening were used as parental stock. After 11 generations, we
removed the selection regime and began rearing populations of
flies in bottles. Several lines were lost during selection, such that
descendents of a single original female comprised the entire line.
Tests for CI Rescue and Fitness Measurements. Single mating pairs
of 1- to 2-day-old virgin flies were placed in empty food bottles
with molasses plate lids that were dotted with a yeast suspension.
15 or more eggs was counted and saved to measure egg hatch-
ability. If females did not produce eggs within 3 days, the cross
was discarded from the experiment. The number of unhatched
eggs was recorded 72 h after collection, and the percentage of
egg hatch was calculated. Dsim wMelPop was assayed at gener-
ation 20 after transinfection for CI rescue by mating infected
males to uninfected (tetracycline control) females (n ? 75).
Fecundity was simply a summation of the first 3 days of egg
production. Productivity was the same measure, extended for 10
days. Fecundity was measured from the same crosses used to
measure hatch rate, and is therefore likely to be an overestimate,
given that all nonproducing crosses were removed. All estimates
were compared with parallel measurements from control crosses
(uninfected ? uninfected flies generated by tetracycline treat-
ment, for each line). Statistical significance of hatch rates for
various crosses was determined by using a Kruskal–Wallis test
followed by comparisons using the Simultaneous Test Procedure
method (22). Fecundity measurements were compared with a
general one-way ANOVA and a sequential Bonferroni adjusted
? value for planned comparisons.
Maternal Transmission. We measured the transmission efficiency
15 and 22 by selecting 10 PCR positive females (mated to
uninfected control males) from each line and by assaying, with
PCR, the proportion of their offspring (10 each) that were
infected with Wolbachia.
Real-Time Quantitative PCR. DNA from either head or ovary
dissections was extracted as above. All hybridization probes,
primers, and PCR procedures were as reported (11). In short,
PCR amplification was carried out independently for both the
single-copy wsp and su(fk)C genes of bacterial and host origin,
respectively, in the presence of specific molecular beacon hy-
bridization probes (23). Initial gene copy number was estimated
by comparison to a standard curve using Roche LIGHTCYCLER
data analysis software v3.1.02. The copy number of the host gene
Three replicates were run and averaged for each sample before
construction of relative copy number ratios. For each strain at
each time point, we collected measurements on 10 samples.
Because variances tended to be correlated with means, we log
transformed all ratio data before performing t tests. If variances
remained unequal (F test) we then used the nonparametric
Mann–Whitney U test. A Bonferroni correction was used to
compensate for multiple comparisons.
Fluorescence Microscopy. Embryos less than 45 min old were
collected from 3-day-old females and dechorionated in 50%
bleach (0.352 M NaClO) for 5 min, shaking periodically, fol-
KCl?2.3 mM NaHCO3?1.0 mM CaCl2?0.055 mM NaH2PO4)
until a bleach odor could no longer be detected. Embryos were
then exposed to equal parts fixative (4% formaldehyde in PBS)
and heptane for 20 min, shaking vigorously every 5 min. All
fixative was then removed, and the eggs were washed with fresh
heptane. Devitellinization was achieved by adding an equal
1 min. All heptane was removed, and the eggs were washed with
100% methanol. Eggs were washed three times, 5 min each with
PBS containing 0.25% BSA and 0.4% Triton-X (PBT), and then
exposed to a blocking agent [normal goat serum (NGS) diluted
in PBT] for 1 h at room temperature. After removing PBT–NGS,
the embryos were incubated in primary antibody (Hsp60, LK2;
from Sigma) (2) overnight at 4°C. The primary antibody was
removed with four 10-min rinses in PBT. Eggs were incubated in
mouse FITC-conjugated anti-IgG for 3 h at room temperature.
The secondary antibody was removed by two washes in PBT, 10
min each. Eggs were then stained with a 1 ?g/ml 4?,6-diamidino-
2-phenylindole (DAPI) solution in 80% glycerol for 5 min. After
four 10-min rinses in PBS, the embryos were removed to an 80%
glycerol solution. After slide mounting, embryos were examined
by using a Zeiss Axioskop FS microscope equipped with epi-
fluorescence optics. Images were recorded with a Cohu gray-
scale charge-coupled device camera, digitized with a Scion
framegrabber under control of SCION IMAGE 1.62, and compiled
by using PHOTOSHOP version 5.5.
Lifespan Measurements. For each strain included in the analysis
wMelPop, and Dmel wMelPopT), four vials of 20 flies each of
gender treatment) were reared at 26°C with a relative humidity
of 40% in standard yeast-inoculated food vials. Each day we
recorded the number of new deaths. Flies were moved to fresh
food vials every 3 days. When all flies in all lines were dead,
survival curves of the various treatment groups were compared
with a Cox Proportional Hazard regression model of survival
analysis (STATVIEW 5.01) followed by Z tests for individual
CI Characteristics of wMelPop Infection in D. simulans. The initial
characterization of wMelPop (based on matings to 3-day-old
males) revealed that it was incapable of modifying sperm in a D.
melanogaster background and as such could be considered a
mod?strain (15). After transinfection into D. simulans, however,
the association gained the ability to cause CI, indicating that
sperm modification was host dependent (11). We also wished to
compare the ability of the infection in the two hosts to rescue
sperm defects induced by mod?Wolbachia variants carried by
the same host species (Fig. 1). Dsim wMelPop females at
generation 20 mated to Dsim wRi (strong CI inducer) males
produced embryos with a mean hatch rate of 65.93% (SE ?
2.39%; n ? 50), which was significantly greater than nonrescuing
uninfected controls (0.49% ? 0.29%; n ? 50, P ? 0.01). D.
melanogaster (yw67C23) males infected with wMel induced only
moderate CI expression (83.9% ? 2.2%; n ? 50), which was fully
McGraw et al.
March 5, 2002 ?
vol. 99 ?
no. 5 ?
rescued by wMelPop-infected females (98.9% ? 0.53%, n ? 50).
As such, it appears that the wMelPop infection behaves as a
mod?, resc?strain in a D. melanogaster background and as a
mod?, resc?strain in a D. simulans host background.
wMelPop Replication in D. simulans. Ovaries. All measurements of
taken between G20–22from four ages of adults, and again at G29
for 28-day-old flies only (Fig. 2a). Ovary density estimates from
Dsim wRi and Dmel wMelPop were comparatively low and
increased very little during the lifespan of the fly. In contrast,
Wolbachia densities (as determined by wsp gene relative copy
number) in the ovaries of Dsim wMelPop rose rapidly with age,
and by 28 days were roughly 5-fold greater (24.2 ? 4.5 copies per
cell; n ? 10) on average than either Dsim wRi (7.37 ? 1.6; n ?
10) or Dmel wMelPop (2.7 ? 0.54; n ? 10). Fluorescent
microscopy using both an anti-Hsp60 antibody (2) and the
DNA-specific dye (DAPI) (11, 26, 27) corroborate the data
collected from the real-time quantitative PCR assay. The
transinfected line shows much higher densities of Wolbachia at
G20in comparison to Dsim wRi (Fig. 3). By G29, densities of
wMelPop in the ovaries were greatly reduced (2.4 ? 0.54; n ?
8) to a level that was no longer greater than densities associated
with the wRi infection (P ? 0.05) indicating a rapid attenuation
of replication rate. At G40we again assayed 28-day-old females
for wMelPop densities in ovaries, and found that they had risen
slightly (8.4 ? 0.21; n ? 12) compared with G29(P ? 0.05), but
were still well below mean densities found at G20(P ? 0.01).
hatch (?SE, n ? 50 per cross) from wMelPop and wMelPopT females in both
D. simulans and D. melanogaster hosts mated to Dsim wRi or Dmel wMel,
respectively. White columns indicate hatch rates from crosses involving unin-
fected females that cannot rescue sperm modification, and gray columns
represent crosses involving wMelPop-infected females which show partial
rescue of the wRi infection and complete rescue of the wMel infection.
Ability of the wMelPop infection to rescue CI. Mean percentage egg
of various ages plus a set of measurements for both G29and G40Dsim wMelPop at 28 days of age. Measurements are of ovaries dissected from virgin females.
Mean Wolbachia densities in ovaries and heads (?SE, n ? 10 per each point) as determined by real-time quantitative PCR for three lines of infected flies
Embryos were stained with 4?,6-diamidino-2-phenylindole (DAPI; blue) and
Hsp60 immunostaining (green). (a) D. simulans embryos infected with wMel-
Pop. (b) D. simulans infected with wRi. (c) D. melanogaster infected with
wMelPop. In all cases Wolbachia are distributed throughout the periphery of
the embryo. Densities appear the highest in Dsim wMelPop embryos.
Wolbachia distribution in embryos from 3-day-old infected females.
www.pnas.org?cgi?doi?10.1073?pnas.052466499 McGraw et al.
Heads. Wolbachia densities in heads were similar in young flies
for all three lines compared, Dsim wRi, Dsim wMelPop, and
Dmel wMelPop (Fig. 2 b and c). All infections increased in
density with age of the fly, particularly wMelPop. We focused on
comparisons between heads collected from older flies because
they are more relevant to the early death phenotype. Our
primary interest was to determine whether the density of wMel-
Pop was higher in D. simulans G20(females: 19.7 ? 2.0, n ? 10;
males: 29.5 ? 5.8, n ? 18) than in D. melanogaster (females:
25.6 ? 4.8, n ? 14; males: 40.7 ? 7.0, n ? 10). For females, we
determined that the densities of wMelPop were not different in
the two hosts (P ? 0.05), but that the density of wMelPop in both
hosts was significantly higher (P ? 0.01) than Dsim wRi (5.8 ?
0.60, n ? 10). The pattern was the same for densities in males,
with no differences between the wMelPop infections in D.
simulans and D. melanogaster (P ? 0.05), but wMelPop having
significantly higher densities than wRi (15.5 ? 5.6; n ? 12; P ?
0.01). Variability in our density measurements was, in general,
greatest for the wMelPop infection, most especially for male
Fitness Costs Associated with wMelPop Infection. Fecundity and
hatch rate. Because Wolbachia are maternally transmitted, bac-
terial densities in the ovaries have direct bearing on transmission
efficiencies. Higher densities of Wolbachia in the ovaries that
favor greater transmission may also confer fitness costs. We
estimated the effects of the popcorn infection on the reproduc-
tive fitness of females in both hosts by measuring fecundity and
hatch rates. The same measures were taken for D. simulans
infected with the wRi infection as well as uninfected D. simulans
and D. melanogaster lines for comparison (Fig. 4). Surprisingly,
wMelPop in its native D. melanogaster has no effect on female
reproductive fitness, with mean egg hatch rate (98.2 ? 0.5%, n ?
75) and fecundity (114.1 ? 2.0 eggs in 3 days, n ? 75) indistin-
guishable from uninfected controls (data not shown). The wRi
infection demonstrated mild, but significant, reductions (P ?
0.01) in both egg hatch (92.6%, ? 1.0%, n ? 75) and in fecundity
(96.0 ? 3.3, n ? 75) when compared with uninfected D. simulans
(98.0% ? 0.5, n ? 75; and 110.0 ? 2.0, n ? 75, respectively).
Reduced fecundity in laboratory populations of Dsim wRi has
been reported (28). Egg hatch rates for the transinfected line
measured at G5were much reduced (72.4% ? 2.8%, n ? 27), but
G5(P ? 0.05). Fecundity of Dsim wMelPop was also significantly
(P ? 0.01) lower at G5(41.5 ? 2.3%, n ? 27), less than half that
of uninfected controls (98.3 ? 2.5%, n ? 75). The effect lessened
most significantly between G14and G22(P ? 0.05), but had not
disappeared by G22(71.5 ? 2.2%, n ? 75), with fecundities still
below controls (P ? 0.05).
The tetracycline-treated transinfected control line (Dsim
wMelPopT) showed normal levels of hatch rate and fecundity,
which indicated that the observed reductions in Dsim wMelPop
were not host effects associated with either inbreeding or genetic
drift. The reduced fecundity in the transinfected line could also
have been explained by delayed egg production, and therefore
we compared productivity from G14Dsim wMelPop and Dsim
wRi lines. Mean productivity for the transinfected line (130.0 ?
60.3, n ? 10) was significantly lower (P ? 0.01) than Dsim wRi
(337.6 ? 20.6, n ? 10). The disparity in numbers could be
explained by a decline in egg production of the transinfected line
to near zero after only 5 days of collection.
Maternal transmission. We hypothesized that the higher Wol-
bachia densities in the ovaries of Dsim wMelPop might translate
into greater maternal transmission efficiencies. Modeling by
Turelli (29) has suggested that significant reductions in host
fitness may be stable in populations if they are linked to gains in
Wolbachia transmission and high CI expression. The transin-
fected line demonstrated 100% transmission efficiency as as-
sayed by PCR at generations 15 and 22. Both Dsim wRi and
Dmel wMelPop also displayed perfect transmission. Although
our results are preliminary, given the small sample sizes, it is
clear that a comparison of the three lines that levels of popcorn
Wolbachia in the ovaries are in excess of what is required for
perfect transmission in the laboratory.
Longevity. The effect of the wMelPop infection on lifespan of
D. simulans was tested at G20and G29post transinfection (Fig.
5). For both experiments, survival of Dsim wRi and Dmel
wMelPop populations was tracked simultaneously for compar-
ison, as well as all uninfected counterpart strains. Survival curves
for males and females of each treatment group were measured
independently. The Dsim wMelPop line demonstrated a reduced
lifespan in comparison with both uninfected controls (Fig. 5b)
and to Dsim wRi (Fig. 5a), confirming the dominant role of the
wMelPop genome in the phenotypic expression of shortened
host lifespan. The proportional hazard of death caused by
wMelPop infection for D. simulans at G20versus the nonvirulent
wRi infection (Fig. 5 a vs. b) was significantly greater for both
relative risk ratios of 6.42 (95% confidence interval, 5.76–7.08)
and 11.35 (95% confidence interval, 11.05–11.65), respectively.
The risk ratio demonstrates the degree to which either variable
(in this case infection type) causes a greater hazard of dying.
Were the wMelPop and wRi infections associated with identical
death risk in males and females, their relative risk ratios would
have been 1.0. The effects of early death in the new host
compared with the native host were no more extreme for males
at either G20(Fig 5 b vs. d) or G29(Fig. 5 c vs. d) (Z ? 1.0 and
0.25, P ? 0.05). Females at G20did demonstrate a more extreme
4.95 times as great (95% confidence interval, 4.39–5.51) (Fig. 5
by G29(Z ? 2.3, P ? 0.05) with a relative risk ratio of 1.18 (95%
confidence interval, 0.64–1.7) (Fig. 5 c vs. d). The latter result
indicates that transinfection initially led to heightened virulence
in females but over multiple generations this effect attenuated
back to levels observed in its natural host. Comparisons of death
hazards for Dsim wMelPop G20 versus G29 (Fig. 5 b vs. c)
demonstrated a significant difference for females only (Z ? 8.9,
P ? 0.01). The relative risk ratio was 2.23 (95% confidence
interval, 1.68–2.81). Mean time to death lengthened from 14.8
days at G20to 20.8 days at G29. For males the mean time to death
(?SE) assayed by mating females of each line to uninfected males. Data are
wRi, and Dmel wMelPop.
Mean egg hatch (%) and fecundities (number of eggs in first 3 days)
McGraw et al.
March 5, 2002 ?
vol. 99 ?
no. 5 ?
was not large enough to produce a significant difference (Z ?
0.84, P ? 0.05) in relative risk (1.06, 95% confidence interval,
0.55–1.57). For both genders, survival curves at G29appear less
steep with an increasing variability in survival time, with a subset
of flies living longer.
A comparison of the CI expression of wMelPop in D. simulans
and D. melanogaster indicates the complexity of host and bac-
terial genotypic interactions that determine expression of repro-
ductive phenotypes. Previously we reported that wMelPop could
not modify sperm in D. melanogaster, but gained that ability in
a D. simulans host (11). The cause appeared to be differential
tissue tropism in the two hosts. In D. simulans, the bacterium
localized to both testes sheath cells and developing sperm
bundles, whereas in D. melanogaster, wMelPop was found only in
the sheath cells. The strain’s ability to rescue was more stable,
expressing itself in two host backgrounds and in response to
modification by two different infections. This stability is likely to
be a byproduct of the need for Wolbachia strains to maintain
themselves in the female germ line to be transovarially trans-
mitted. This observation, in turn, places them in the correct
cellular compartment for CI rescue. The same restriction is not
placed on testes infections because the male germ line is not
directly involved in vertical transmission.
The ability of the wMelPop infection to completely rescue the
wMel infection is not unexpected, given that wMelPop is very
closely related (they share identical wsp sequences; ref. 11). The
incomplete rescue of the wRi infection is in keeping with
previous estimates of hatch rates from Dsim wMel ? Dsim wRi
crosses (10). Both results indicate the dominant role of bacterial
genotype in determination of CI crossing type.
wMelPop densities in ovaries of D. melanogaster were surpris-
ingly low and constant throughout the host’s lifespan. We
expected the gonads to be a site of over-replication and to
experience increasing densities with age as observed by Min and
Benzer in other tissues (15). But if Wolbachia density correlates
with reduced fitness, then lower titers explain the complete lack
of fecundity and hatch rate reductions in D. melanogaster. In
contrast, the same infection transinfected into D. simulans
initially infects ovaries at much higher densities and in a manner
consistent with the observed reductions in reproductive fitness.
The attenuation of such fitness costs could be explained by
parallel reductions in Wolbachia replication rates in the ovaries.
Comparisons of maternal transmission rates for the infections
indicate that the low ovary densities in the first few days of egg
production are sufficient to secure perfect transmission in the
Patterns of bacterial densities in heads for the different
strain?host combinations did not parallel those for the ovaries.
The wMelPop infection increased in density with age, regardless
than wRi. The simultaneous increases in variability associated
with wMelPop may be an inherent property of such high titers
or may reflect a recently established association with low se-
lection pressure for optimal bacterial densities in nervous
The host-dependent differences in both wMelPop ovary den-
sities and reproductive fitness costs can be explained by two
hypotheses, which may not be mutually exclusive. The first is that
there may be intrinsic host-species-specific differences in control
of wMelPop replication. The differences in tissue tropism of
wMelPop in the two hosts (11) and the observation that Wol-
bachia densities are generally higher in D. simulans than in D.
melanogaster (8) suggest this is quite likely. The second expla-
to a point where the replication rate in the ovaries has been
reduced to where it does not induce reproductive fitness costs,
but is still high enough to guarantee perfect transmission. Such
endpoints are predictable in a host–parasite system that relies on
vertical transmission (31). The rapid attenuation of fecundity
and hatch rate in association with reductions in Wolbachia
density demonstrate that the establishment of the Wolbachia
symbiosis is an active process. This finding is in contrast to the
image of Wolbachia as a stealthy manipulator of the host implied
by the results of transfer experiments that show little host
involvement in phenotypic outcomes (32) or by what appears to
be the lack of any interaction between Wolbachia and the innate
immune system of the host (33).
The reduction of adult lifespan observed in the novel host
indicates the strong involvement of the wMelPop genome in
expression of the early death phenotype. In contrast to fecundity
and egg hatch, the reduced lifespan phenotype did not attenuate.
This reduced attenuation may be explained by a number of
reasons. First, the level of selection on adult lifespan may be
negligible under most laboratory rearing conditions, in which
flies are typically maintained at low temperatures and passaged
when young. Second, the characteristics of the wMelPop genome
Wolbachia strain?host species combinations. Black lines represent infected
flies and gray lines represent uninfected tetracycline-treated counterparts.
Survival curves of populations of female and male flies of various
www.pnas.org?cgi?doi?10.1073?pnas.052466499McGraw et al.
that confer loss of replication control in nervous tissue may not Download full-text
be easily reversible by compensatory or back mutations. Inter-
estingly, the progressive attenuation of virulence in ovaries
compared with maintenance of virulence in nervous tissue
suggests that there are tissue-specific controls on Wolbachia
replication. Again these results point to the complexity of the
interaction between Wolbachia and the host. It should be
possible in the future to determine the loci involved in replica-
tion control now that complete genome sequence is available for
both the Drosophila host and the Wolbachia strain that infects it.
It has been proposed that Wolbachia-induced longevity re-
ductions might be able to be used as a form of vector-borne
disease control because old insects are responsible for the vast
majority of disease transmission (34). Such a scenario would rely
as reducing the adult lifespan of its host. The behavior of
wMelPop in D. simulans indicates that this strain might be able
to be used in transinfections of mosquitoes or other disease
vectors because the strain is capable of inducing both strong CI
expression and life shortening in a novel host. Moreover, the
reproductive costs associated with the infection quickly attenu-
ated, but the life-shortening phenotype appeared stable under
laboratory conditions. Alternatively, understanding the genetic
basis of longevity reduction by wMelPop might allow for the
so as to produce the same phenotype.
We thank John Brownstein for his assistance with measuring fitness
characteristics, Ros Schumacher for help with the fluorescence micros-
copy, and Oleg Kruglov for aid in Drosophila care and rearing. This work
was supported by National Institutes of Health Grant AI40620 (to
S.L.O.), University of Queensland Grant 01?UQESEG009 (to D.J.M.),
and National Science Foundation Postdoctoral Fellowship 0074396 (to
1. Dobson, S. L., Bourtzis, K., Braig, H. R., Jones, B. F., Zhou, W., Rousset, F.
& O’Neill, S. L. (1999) Insect Biochem. Mol. Biol. 29, 153–160.
2. Taylor, M. J. & Hoerauf, A. (1999) Parasitol. Today 15, 437–442.
3. O’Neill, S. L., Hoffmann, A. A. & Werren, J. H., eds. (1997) Influential
Passengers: Inherited Microorganisms and Invertebrate Reproduction (Oxford
Univ. Press, Oxford).
4. Hoffmann, A. A. & Turelli, M. (1997) in Influential Passengers: inherited
Microorganisms and Invertebrate Reproduction, eds. O’Neill, S. L., Hoffmann,
A. A. & Werren, J. H. (Oxford Univ. Press, Oxford), pp. 42–80.
5. Callaini, G., Dallai, R. & Riparbelli, M. G. (1997) J. Cell Sci. 110, 271–280.
6. Lassy, C. W. & Karr, T. L. (1996) Mech. Dev. 57, 47–58.
7. Turelli, M. & Hoffmann, A. A. (1995) Genetics 140, 1319–1338.
8. Boyle, L., O’Neill, S. L., Robertson, H. M. & Karr, T. L. (1993) Science 260,
9. Perrot-Minnot, M. J. & Werren, J. H. (1999) J. Evol. Biol. 12, 272–282.
10. Poinsot, D., Bourtzis, K., Markakis, G., Savakis, C. & Mercot, H. (1998)
Genetics 150, 227–237.
11. McGraw, E. A. Merritt, D. J., Droller, J. N. & O’Neill, S. L. (2001) Proc. R. Soc.
London Ser. B 268, 2565–2570.
12. Poinsot, D. & Mercot, H. (1997) Evolution (Lawrence, Kans.) 51, 180–186.
13. Hoerauf, A., Nissen-Pahle, K., Schmetz, C., Henkle-Duhrsen, K., Blaxter,
M. L., Buttner, D. W., Gallin, M. Y., Al-Qaoud, K. M., Lucius, R. & Fleischer,
B. (1999) J. Clin. Invest. 103, 11–18.
14. Dedeine, F., Vavre, F., Fleury, F., Loppin, B., Hochberg, M. E. & Bouletreau,
M. (2001) Proc. Natl. Acad. Sci. USA 98, 6247–6252.
15. Min, K. & Benzer, S. (1997) Proc. Natl. Acad. Sci. USA 94, 10792–10796.
16. Hoffmann, A. A., Turelli, M. & Simmons, G. M. (1986) Evolution (Lawrence,
Kans.) 40, 692–701.
17. Ashburner, M. (1989) Drosophila: A Laboratory Manual (Cold Spring Harbor
Lab. Press, Plainview, NY).
18. Sinkins, S. P., Braig, H. R. & O’Neill, S. L. (1995) Proc. R. Soc. London Ser.
B 261, 325–330.
19. Hoffmann, A. A. & Turelli, M. (1988) Genetics 119, 435–444.
20. O’Neill, S. L., Giordano, R., Colbert, A. M. E., Karr, T. L. & Robertson, H. M.
(1992) Proc. Natl. Acad. Sci. USA 89, 2699–2702.
21. Braig, H. R., Zhou, W., Dobson, S. L. & O’Neill, S. L. (1998) J. Bacteriol. 180,
22. Sokal, R. R. & Rohlf, F. J. (1995) Biometry (Freeman, New York).
23. Tyagi, S. & Kramer, F. R. (1996) Nat. Biotechnol. 14, 303–308.
24. Partridge, L. & Andrews, R. (1985) J. Insect Physiol., 393–395.
25. Partridge, L., Fowler, K., Trevitt, S. & Sharp, W. (1986) J. Insect Physiol. 32,
26. Bressac, C. & Rousset, F. (1993) J. Invert. Pathol. 61, 226–230.
27. O’Neill, S. L. & Karr, T. L. (1990) Nature (London) 348, 178–180.
28. Hoffmann, A. A., Turelli, M. & Harshman, L. G. (1990) Genetics 126,
29. Turelli, M. (1994) Evolution (Lawrence, Kans.) 48, 1500–1513.
30. Ebert, D. (1994) Science 265, 1084–1086.
31. Anderson, R. M. & May, R. M. (1982) Parasitology 85, 411–426.
32. Giordano, R., O’Neill, S. L. & Robertson, H. M. (1995) Genetics 140,
33. Bourtzis, K., Pettigrew, M. M. & O’Neill, S. L. (2000) Insect Mol. Biol. 9,
34. Sinkins, S. P. & O’Neill, S. L. (2000) in Insect Transgenesis Methods and
Applications, eds. Handler, A. M. & James, A. A. (CRC Press, New York).
McGraw et al.
March 5, 2002 ?
vol. 99 ?
no. 5 ?