Diversity and recombination in Wolbachia and Cardinium from Bryobia spider mites.

1.6 MB·English

by  Ros, Vera I D

Rosetal.BMCMicrobiology2012,12(Suppl1):S13
http://www.biomedcentral.com/1471-2180/12/S1/S13
RESEARCH Open Access
Diversity and recombination in Wolbachia and
Cardinium from Bryobia spider mites
Vera I D Ros1,4*, Vicki M Fleming2,3, Edward J Feil2, Johannes A J Breeuwer1
Abstract
Background: Wolbachia and Cardinium are endosymbiotic bacteria infecting many arthropods and manipulating
host reproduction. Although these bacteria are maternally transmitted, incongruencies between phylogenies of
host and parasite suggest an additional role for occasional horizontal transmission. Consistent with this view is the
strong evidence for recombination in Wolbachia, although it is less clear to what extent recombination drives
diversification within single host species and genera. Furthermore, little is known concerning the population
structures of other insect endosymbionts which co-infect with Wolbachia, such as Cardinium. Here, we explore
Wolbachia and Cardinium strain diversity within nine spider mite species (Tetranychidae) from 38 populations, and
quantify the contribution of recombination compared to point mutation in generating Wolbachia diversity.
Results: We found a high level of genetic diversity for Wolbachia, with 36 unique strains detected (64 investigated
mite individuals). Sequence data from four Wolbachia genes suggest that new alleles are 7.5 to 11 times more
likely to be generated by recombination than point mutation. Consistent with previous reports on more diverse
host samples, our data did not reveal evidence for co-evolution of Wolbachia with its host. Cardinium was less
frequently found in the mites, but also showed a high level of diversity, with eight unique strains detected in 15
individuals on the basis of only two genes. A lack of congruence among host and Cardinium phylogenies was
observed.
Conclusions: We found a high rate of recombination for Wolbachia strains obtained from host species of the
spider mite family Tetranychidae, comparable to rates found for horizontally transmitted bacteria. This suggests
frequent horizontal transmission of Wolbachia and/or frequent horizontal transfer of single genes. Our findings
strengthens earlier reports of recombination for Wolbachia, and shows that high recombination rates are also
present on strains from a restrictive host range. Cardinium was found co-infecting several spider mite species, and
phylogenetic comparisons suggest also horizontal transmission of Cardinium among hosts.
Background widespread among arthropods and nematodes. It is esti-
Wolbachia and Cardinium are intracellular bacteria mated that around 66% of all insects are infected with
infecting a wide range of arthropod species. They have Wolbachia [7]. This diverse genus has been subdivided
been classified as reproductive parasites, being able to into 11 “supergroups” (A-K) on the basis of molecular
manipulate their host’s reproductive system in order to phylogenetic analysis [8-13]. Cardinium was more
promote their own transmission [1-3]. Recently, benefi- recently discovered and has so far been found in 6-7%
cial effects of Wolbachia have been identified as well, as of all arthropods, though seems to be more common in
Wolbachia can protect hosts against virus infection Chelicerates than in insects [2,14-18]. Wolbachia and
[4,5]. Cardinium may also exert beneficial effects [6] Cardinium have been found co-infecting the same host
and in many other cases the effect of Wolbachia or Car- species [2,15,17-21].
dinium is unknown. Wolbachia is well studied and is Although Wolbachia and Cardinium are generally
considered to be clonally inherited via vertical transmis-
sion, there is now a large body of molecular evidence
*Correspondence:[email protected]
1EvolutionaryBiology,InstituteforBiodiversityandEcosystemDynamics, for discordant phylogenies of host and endosymbiont
UniversityofAmsterdam,Amsterdam,theNetherlands [22-29]. Distantly related Wolbachia or Cardinium
Fulllistofauthorinformationisavailableattheendofthearticle
©2012Rosetal;licenseeBioMedCentralLtd.ThisisanopenaccessarticledistributedunderthetermsoftheCreativeCommons
AttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,andreproductionin
anymedium,providedtheoriginalworkisproperlycited.
Rosetal.BMCMicrobiology2012,12(Suppl1):S13 Page2of15
http://www.biomedcentral.com/1471-2180/12/S1/S13
strains can infect closely related host species, and closely mite family Tetranychidae, by analyzing strains recov-
related strains may infect distantly related host species. ered from seven Bryobia species, Tetranychus urticae,
Such patterns suggest horizontal transmission of bac- and Petrobia harti. We consider strain diversity between
teria (or at least of some bacterial genes) between hosts, tetranychid host species, within single host species
although direct evidence for horizontal transmission is (investigating multiple populations; up to 20 populations
rare [30-32]. for B. kissophila) and within single populations and indi-
Horizontal transfer has been further supported by evi- viduals. Both Wolbachia and Cardinium have been
dence for recombination [33]. For Wolbachia, recombi- reported from this family. Wolbachia has been detected
nation has been found between genes (intergenic) as in at least six asexual and one sexual Bryobia species
well as within genes (intragenic). Intergenic recombina- and strains from both supergroup B and K have been
tion is evident from inconsistencies between gene trees found [12,47,48]. Supergroup K is a new supergroup
[34-36]. Intragenic recombination has been observed that has only been detected in Bryobia so far [12]. We
within the genes wsp, ftsZ, and gltA and within and investigate intra- and intergenic recombination in Wol-
between supergroups A and B [34,37-41]. Recently, a bachia (four genes) and Cardinium (two genes), and
genomic comparison of A-group Wolbachia strains by quantify the rate of recombination relative to mutation
Klasson et al. [42] showed highly recombining genomes, for Wolbachia, by analyzing the variation between pairs
implying frequent horizontal gene transfer. of very closely related strains. We compare this endo-
Cardinium has been less well investigated: although symbiont diversity to the degree of host congruence (co-
there is some evidence for inconsistent phylogenies speciation), host mitochondrial DNA diversity, and geo-
between Cardinium and their hosts, which is compatible graphical distribution.
with horizontal transmission, only a few studies have
been performed, mainly focusing on a single gene (16S Results
rDNA) [2,15,18,21,23], and there is currently no evi- We included Wolbachia strains from seven Bryobia spe-
dence regarding recombination in Cardinium. cies (B. berlesei, B. kissophila, B. praetiosa, B. rubriocu-
Phylogenetic and evolutionary studies on Wolbachia lus, B. sarothamni, B. spec. I, andB. spec. V) and T.
have mainly focused on samples representing a wide urticae, and Cardinium strains from B. rubrioculus, B.
range of host species [26,34,37,38,43,44]. Based on two sarothamni, T. urticae, andP. harti (Figure 1 and Addi-
genes, Jiggins [38] showed that among strains from a tional file 1) [49]. We were unable to reliably determine
wide range of host species, the rate of recombination is the infection status of the other Bryobia host species
similar to that of a horizontally transmitted bacterium (Figure 1) due to the lack of adequate material and/or
(Cowdria ruminantium). It remains however unclear to inconsistent amplification of the bacterial genes, there-
what extent these conclusions will be supported by the fore these species were excluded from further analyses.
analyses of much more tightly defined samples such as The dataset includes strains from sexually (B. sar-
thoserecoveredfromcloselyrelatedhostgenera,oreven othamni, T. urticae, P. harti) and asexually (the remain-
fromasinglehostspeciesfromasinglegeographicaland ing species) reproducing species.
temporalsource.Mostcurrentstudieswhichaddressthis
have used only one or two genes or a restricted number Multiple infections
ofspeciesorpopulations[31,36,41,45].AstudybyBaldo Wolbachia and Cardinium were found co-infecting B.
etal.[22]includedamoredetailedstudyoftheextentof rubrioculus, B. sarothamni, and T. urticae. In B. rubrio-
recombination and horizontal transfer in a single spider culus and B. sarothamni, Wolbachia and Cardinium
genus and revealed that horizontal transfer explains a strains were obtained from doubly infected individuals,
large part of the Wolbachia distributionpatterns within whereas in T. urticae they were obtained from singly
the genus. Exact rates of recombination among Wolba- infected individuals (Additional file 1). Multiple Wolba-
chiastrainshavehowevernotbeeninferredsofar,which chia strains infecting a single host individual were not
makes it difficult to draw direct comparisons with rates detected, and neither were multiple Cardinium strains.
found for other bacteria. Recombination rates can be Sequence chromatograms revealed no double peaks and
obtainedfrommultilocussequencedata.Strainsthatdif- cloning and sequencing of eleven PCR products did not
fer at only a single locus are grouped into clonal com- reveal multiple infections.
plexes. Subsequently, the allele sequences are examined
todeterminewhethersingle allelicvariantswithinaclo- Wolbachia diversity
nal complex result from point mutation or homologous Sequences from the four Wolbachia genes (wsp, ftsZ,
recombination[46]. groEL, and trmD) were recovered for 65 Wolbachia
We present here a detailed study of the diversity of infected individuals, except for wsp from B. spec. V
Wolbachia and Cardinium in the phytophagous spider (ITA11). The Wolbachia strain infecting B. spec. V
Rosetal.BMCMicrobiology2012,12(Suppl1):S13 Page3of15
http://www.biomedcentral.com/1471-2180/12/S1/S13
Figure1PhylogeneticrelationshipbetweenthetetranychidhostspeciesfromwhichWolbachiaandCardiniumstrainswereobtained.
Maximumlikelihoodcladogram(28SrDNA)ofthegenusBryobiaandfouroutgroupspeciesofthegenusPetrobiaisshown[49].Tetranychus
urticaewasdepictedseparatelyastheexactpositionofT.urticaerelativetotheotherhostspecieswasnotstudiedsofar.ThegenusTetranychus
belongstoanothersubfamily(Tetranychinae)thanBryobiaandPetrobia(bothBryobiinae)ofthefamilyTetranychidae.Themodeofreproduction
isgivenforeachhostspecies(A=asexual,S=sexual)inaseparatecolumn,andthesubsequentcolumnsindicatefromwhichhostspecies
WolbachiaandorCardiniumstrainswereincludedinthisstudy.SpeciesnamesarecoloredasinFigure2,4,5,andAdditionalfile3.Hostspecies
ingreywerenotincludedinthisstudy.Numbersabovebranches(bold)indicateMLbootstrapvaluesbasedon1,000replicates,numbersbelow
branches(plain)depictBayesianposteriorprobabilities(onlyvalueslargerthan50areindicted).
Rosetal.BMCMicrobiology2012,12(Suppl1):S13 Page4of15
http://www.biomedcentral.com/1471-2180/12/S1/S13
belongs to the newly described supergroup K [12], changes, and could be either derived by point mutations
which is highly divergent from supergroup B strains or recombination (the chance of two independent muta-
infecting other tetranychid mites. We excluded the tions both occurring in one out of four genes is 0.25).
supergroup K strain from phylogenetic and recombina- One of these variant alleles was found elsewhere in the
tion analyses. No insertions or deletions were found dataset, implicating recombination. All other allelic var-
within ftsZ, groEL, andtrmD . Within wsp small indels iants differed from the founder alleles at four or more
(3-9bp) were found in a few strains but all sequences sites and were considered as putative recombinational
could be unambiguously aligned. imports. Ignoring alleles with one non-unique and two
The sequenced Wolbachia strains reveal a high diver- nucleotide changes, the estimated ratio of recombina-
sity. From the 64 Wolbachia strains (excluding the tional events to mutational events per gene fragment is
supergroup K Wolbachia strain in B. spec. V), 36 strains 11:1. If we include non-unique changes as recombina-
(sequence types; STs) were found unique (Additional tional imports, and unique changes as point mutations,
file 2). Between 11 (groEL) and 18 (trmD) alleles were the ratio is 15:2. We therefore conclude that new alleles
found per locus (Table 1). Nucleotide diversity was 5-11 were 7.5 to 11 times more likely to be generated by
times higher for wsp than for the other loci (Table 1). recombination than by point mutation. This is a conser-
The dN/dS ratio was < 1 for all loci, indicating that the vative estimate because single nucleotide changes were
genes where not subjected to positive selection. The wsp attributed to point mutation and not to recombination,
gene also revealed a high rate of intragenic recombina- although recombination between similar alleles could
tion (see below), with two sites identified within hyper result in a single nucleotide change. Further, a high rate
variable region 1 (HVR1) under positive selection of recombination is consistent with the observed incon-
(HyPhy: codons 20 and 30; unpublished data). Despite gruence between the four gene tree topologies (Addi-
this high nucleotide diversity and recombination rate for tional file 3).
wsp, we found the most alleles for trmD. Intragenic recombination is another process that may
Forty-four out of the 64 strains were grouped into five contribute to the origin of new Wolbachia genotypes.
clonal complexes (I-V; Figure 2 and Table 2). All other We detected intragenic recombination within the trmD
strains differed at more than one locus from the strains and wsp genes (Figure 3). The alignment of wsp genes
in these complexes. A total of 17 alleles deviated from shows that the polymorphic sites are not randomly dis-
the alleles from the founding genotypes within the clo- tributed, but clearly shows a mosaic pattern consistent
nal complexes (Table 2). A significant higher number of with recombination. Intragenic recombination is not
these variant alleles were found for trmD compared to restricted to Wolbachia strains from the same host spe-
the other loci (Table 2; Chi-square test; p=0.003), which cies, but also involves strains infecting different host
is consistent with the observation that this locus con- species. For example, the wsp sequence obtained from
tains the most alleles. Wolbachia in B. sarothamni (all populations) is a
Recombination between Wolbachia strains recombinant between the wsp sequences obtained from
We investigated intergenic recombination by analysis of Wolbachia in B. kissophila (FR13) and T. urticae (T3)
allelic variation within the clonal complexes. This (Figure 3).
approach reveals whether variant alleles arose by point Cospeciation of Wolbachia and host species
mutation or by recombination. Of the 17 variant alleles, Examination of the concatenated Wolbachia phylogeny
four differed from the typical allele in the clonal com- reveals that there is generally a lack of cospeciation
plex by a single nucleotide change (Table 2). Three of between host and parasite (Figure 4). Wolbachia strains
these single nucleotide changes, however, were non- obtained from a single host species do not clearly clus-
unique. Two other alleles differed by two nucleotide ter. For example, strains from B. rubrioculus are found
Table 1Diversityobserved for four Wolbachiagenes andtwo Cardinium genes.
Locus Size(bp) Alleles Variablesites π p-distancemax.(%) dN/dSratio
n %
Wolbachia(n=64) wsp 525 13 155 29.52 0.1030 20.08 0.60
ftsZ 507 14 20 3.94 0.0126 2.37 0.07
groEL 491 11 18 3.67 0.0087 1.83 0.29
trmD 453 18 34 7.51 0.0176 4.42 0.23
Cardinium(n=15) CLO 407 6 15 3.69 0.0151 2.22 -
gyrB 631 8 127 20.13 0.0839 14.9 0.06
π=nucleotidediversity
Rosetal.BMCMicrobiology2012,12(Suppl1):S13 Page5of15
http://www.biomedcentral.com/1471-2180/12/S1/S13
Figure2SchematicoverviewoftheclonalrelatednessoftheWolbachiaSTsaspredictedbyeBURST.EachSTisrepresentedbyablack
dot,thesizeofwhichisproportionaltothenumberofstrainsofthatST.STsthatdifferatasinglelocusarelinkedbylines.Onlyonevariantis
likelyduetoamutationalevent(indicatedby*),theothervariantsaremostlikelyduetorecombinationevents.STsthatarenotlinkedtoother
STsdonotshareatleastfouridenticalalleleswithanyotherST.HostspeciesnameinwhicheachSTwasdetectedisindicated:BB=B.berlesei;
BK=B.kissophila(A-DindicatedifferentCOIclades,seetext);BP=B.praetiosa;BR=B.rubrioculus;BS=B.sarothamni;BspI=B.spec.I;TU=T.urticae.
Figure 3 Examples of recombination within trmDand wsp. Only polymorphic sitesare shown (position in alignmentisgivenon top).
Sequencesarenamedbytheirsamplecode(Additionalfile1)andabbreviatedhostspeciesname(seelegendFigure2).Eachsequencemay
havebeenfoundindifferentpopulationsorhostspecies,seephylogeniesoftrmDandwspinAdditionalfile3.Differentshadingsindicate
possiblerecombinantregions(seeresults).Differencesandidentities(dots)comparedtothemiddlesequenceareshown.*=alsodetectedin
BspI,BK-A,BK-C,andBP.^=alsodetectedinBR.
Rosetal.BMCMicrobiology2012,12(Suppl1):S13 Page6of15
http://www.biomedcentral.com/1471-2180/12/S1/S13
Figure4ConcatenatedMLphylogenyforWolbachia,based on fourgenes (wsp,ftsZ,groEL,andtrmD).Thirty-sixuniquestrainsare
shown.Samplecode(Additionalfile1)andhostspeciesnameinwhicheachstrainwasdetectedareindicated(forabbreviationsseelegend
Figure2).MLbootstrapvalues(topnumber,bold)andBayesianposteriorprobabilities(bottomnumber,plain)aredepicted(onlyvalueslarger
than50areindicated).*=thetopologywithinthiscladeisslightlydifferentfortheMrBayestopology.Thebaratthebottomindicatesabranch
lengthof10%likelihooddistance.IndependentphylogeniesforeachgenearedepictedinAdditionalfile3.
Rosetal.BMCMicrobiology2012,12(Suppl1):S13 Page7of15
http://www.biomedcentral.com/1471-2180/12/S1/S13
at different places in the phylogeny. The same is true for haplotypes and Wolbachia haplotypes may be linked
strains from B. spec. I. On the other hand, the Wolba- [50-53]. As this has serious implications for population
chia phylogeny is not completely random with respect studies based on mtDNA [54], we were motivated to
to host species. Some Wolbachia strains from B. kisso- examine this possibility for B. kissophila. High levels of
phila cluster together, whilst others cluster with strains diversity at the mitochondrial COI locus were observed
from B. spec I (BEL4_2) or B. rubrioculus (FR15). Two withinB.kissophila,whichresolvedintofourclades(A-D)
B. kissophila-derived strains (NL9 and FR13) are very [49]. However, there was little evidence for correlation
divergent from all other B. kissophila strains. In the between the COI haplotypes and the Wolbachia strains
exceptional case of B. sarothamni, the same Wolbachia (Figure 2 and 4). A total of 20 populations were investi-
genotype was found in all five populations (from Bel- gatedforB.kissophila,andahighlydivergentsetofWol-
gium and France; except for a minor difference in trmD bachiastrainswasfoundwithinthisspecies.Twenty-one
for BEL6; Figure 2, 4, and Table 2). This strain was not Wolbachiastrainswerefound,fourofwhichwereshared
found in any of the other species, although it closely between populations. Within several populations (BEL1,
resembles the Wolbachia strain infecting B. berlesei at FR2,NL1,NL3,NL6,SP3, and SP4)morethanoneWol-
three of the four genes (wsp is highly divergent between bachiastrainwasdetected.
the two strains). Bryobia sarothamni and B. berlesei Bryobia kissophila COI clade A was highly divergent
share the same host plant species, Cytisus scoparius, fromallotherCOIclades,andcontainsWolbachiastrains
which potentially facilitates horizontal transmission of thataredivergentfromtheonesfoundintheotherclades.
both Wolbachia strains and genes. However, the two investigated populations belonging to
Finally,asingleB.praetiosaindividualwasinvestigated. clade A (NL9 and FR13) harbor divergent Wolbachia
AlthoughthisspecieswasfoundtoharborauniqueWol- strains.Also,someallelesofthesestrainsaresharedwith
bachia strain, this strain shares each of its alleles with other B. kissophila clades (for groEL and trmD) or with
strains in (multiple) other host species. Although allelic other Bryobia species (for all four genes) (Additional file
identity by descent cannot be ruled out without more 3).WolbachiastrainsfromcladeB,C,andDshowamix-
detailed analysis, this observation is also consistent with tureofdifferentWolbachiastrains.Thereisnocorrelation
frequentinter-allelerecombination.Withintheotherspe- withCOIhaplotype,althoughtherearenostrainsshared
cies, divergent Wolbachia strains were found between amongpopulationsbelongingtodifferentCOIclades.
populationsandalsowithinpopulations(Figure4).Infive There is a similar lack of congruence between Wolba-
B.rubrioculusmitepopulations, sixdivergentWolbachia chia strain diversity and geographic location of the host
strainswerefound:populationPL5containstwodivergent populations. Very distant populations may harbor identi-
Wolbachiastrains.ForB.spec.IthreeWolbachiastrains cal Wolbachia strains (e.g., BEL2 and SA1; B. kisso-
were detected in two populations: two individuals from phila), while nearby populations harbor very divergent
BEL4 harbor highly divergent Wolbachia strains (mainly Wolbachia strains (e.g., NL15 and NL16; B. rubrioculus).
duetodifferencesatwspandftsZ). Also within populations divergent strains are found.
Correlation between Wolbachia and host mitochondrial
diversity or geographical location Cardinium diversity
IthasbeensuggestedthatinfectionbyWolbachiaaffects Cardinium diversity was investigated by sequencing part
host mitochondrial diversity and that mitochondrial of the 16S rDNA and gyrB gene. Sequences were
Table 2Clonalcomplexes found forWolbachia
Complex I II III IV V
STa 4 9 5 2 1 11 6 30 29 28 27 16 15 14 13 36 10 12 24 25 33 34
wsp 1 - - - - - 518 12 - - - 5 38 - - - - 416 12 - 6 -
ftsZ 2 - - - 11 - - 10 - - - 3 - - - - - - 10 - 14 -
groEl 8 - - 44 44 - - 8 - - - 8 - 44 - - - - 12 112 3 -
trmD 1 31* 1015 - - 67 67 1 149 67 21* 1 - - 52* 179 158 158 8 - 9 111*
Speciesb BK-B BK-B BK-B BK-B BK-B BK-D BK-D BK-B BspI BK-D BK-B BK-D BK-D BK-D BR BR BspI BR BK-D BK-D BS BS
Freq.c 2 1 1 12 1 1 1 1 1 1 1 2 1 1 1 1 1 1 3 1 8 1
aAllelicvariantswithineachclonalcomplexaredepicted,listedpersequencetype(ST).Likelyrecombinationalchangesaredepictedinplaintext,andputative
mutationsareshowninbold.Disputablecasesarehighlightedinitalics(possiblerecombinationalchanges).Foreachallelicvarianttheallelenumberisgiven,
withinsuperscriptthenumberofpolymorphicsitesbetweentheallelicvariantandthetypicalalleleoftheclonalgroup.*indicatesnon-uniquemutations(in
caseswhereoneortwomutationswerefound).
bHostspeciesnameinwhicheachSTwasdetectedisindicated(forabbreviationsseelegendFigure2).
cThefrequencyofeachsequencetypeislisted.
Rosetal.BMCMicrobiology2012,12(Suppl1):S13 Page8of15
http://www.biomedcentral.com/1471-2180/12/S1/S13
successfully recovered from all Cardinium infected indi- closely related host species, all belonging to the family
viduals and all sequences could be unambiguously Tetranychidae, andmainly from the genusBryobia.Our
aligned. No insertions or deletions were found within study shows that even within a single host genus there
gyrB. Within 16S rDNA, one insertion and one deletion exists a high level of Wolbachia diversity. Wolbachia
(both 1bp) were found. For 16S rDNA six alleles were strainsbelongingtotwohighlydivergentsupergroups(B
found, with 3.7% variable sites, a maximum p-distance and K) were detected. Even within Wolbachia super-
of 2.2%, and a nucleotide diversity of 0.015 (Table 1). group B, 36 unique STs were obtained from 64 infected
Diversity for gyrB was much higher, with eight alleles, hosts.Althoughtherewaslittlecorrelationbetweenhost
20.1% variable sites, a maximum p-distance of 14.9%, speciesandWolbachiastrains,strainswerenotdistribu-
and a nucleotide diversity of 0.084. ted randomly among different species (Figure 2 and 4),
In total, eight strains were detected within eight popu- sothatacertainlevelofspecificitywasobserved.Strains
lations, belonging to four mite species, and phylogenetic within clonal complex I were restricted to B. kissophila
analysis resolved these eight stains into two major clades and within clonal complex V to B. sarothamni. Other
(Figure 5). The Cardinium strain found in P. harti complexes however contain strains from different host
(CH1) is divergent from two other clades (named I and species.Itisstrikingthatmanyallelesaresharedamong
II), which were detected in B. sarothamni and B. rubrio- the different STs, even from different host species, indi-
culus (both clade I and II), and in T. urticae (clade I). catingthatrecombinationcontributessubstantiallytothe
These two clades are highly supported. Clade I and II geneticdiversityofWolbachia.Recombinationisfurther
differed at 1.7% of nucleotide sites for 16S rDNA and at evidenced by the many phylogenetic conflicts observed
10.6% for gyrB, while differences within clades are small among the individual gene trees and a high recombina-
(<1.2% for both genes). Generally, there is congruence tionratecomparedtomutationrate.Analysisofthevar-
between the phylogenies obtained for 16S rDNA and iant alleles in the clonal complexes reveals that the rate
gyrB which suggests less recombination than in Wolba- of recombination compared to point mutation in the
chia, although the evidence is equivocal. However, there diversificationoflineagesrangesbetween7.5:1and11:1.
is no obvious association between Cardinium genotype The observed recombination rate and diversity is
and host species. Clade I contains strains found in three much higher than what would be expected for clonal
B. rubrioculus populations and in one T. urticae and organisms. Recombination is rare in other clonally
one B. sarothamni population, while clade II contains inherited, obligate intracellular bacteria [55,56]. The
highly related strains found in two B. sarothamni popu- high recombination rate we found is comparable to
lations and one B. rubrioculus population. One strain rates of horizontally transmitted human pathogens. For
was found infecting two host species: B. rubrioculus example, for Streptococcus pneumoniae a recombination
(NL15_1-4) and B. sarothamni (FR21_3). Other strains to mutation ratio of 10:1 was found, for Neisseria
belonging to B. sarothamni population FR21 group meningitidis a ratio of 5:1 [57]. Horizontal transmission
within clade II (FR21_1-2). These patterns imply hori- of Wolbachia has been observed, but examples are rare
zontal transfer of strains (or genes) between and within [30-32]. Although many studies based on molecular data
host species. have suggested extensive horizontal gene transfer of
Wolbachia [22,25,35,36,42,43], it is unclear if bacteria
Discussion are transmitted horizontally, or if the transfer concerns
This detailed study of reproductive parasites in nine tet- single genes, possibly via bacteriophages [58]. The high
ranychid mite species reveals a high genetic diversity. rate of recombination found in this study, the observa-
Wolbachia strains belonging to two highly divergent tion that individual alleles are shared among Wolbachia
supergroups (B and K) were detected (see also [12]). strains from different host species but complete STs are
The diversity within supergroup B was high, with 36 not, and the fact that Wolbachia is mainly clonally
unique strains found in 64 investigated individuals. The inherited, suggest that individual genes rather than com-
level of recombination detected is extremely high, sup- plete bacteria are exchanged. Alternatively, transfer of
porting the mosaic genome structure of Wolbachia [42]. bacteria leading to mixed infections and subsequent
Cardinium was less frequently found in the mites than recombination may explain these observations. Although
Wolbachia, but also showed a high level of diversity, our cloning data suggest that mixed infections are rare,
with eight unique strains detected in 15 individuals on this possibility cannot be excluded (see also [59]). The
the basis of only two genes. observation that the trees are not completely random
with respect to host species suggests that vertical trans-
Wolbachia diversity mission does occur [26,43].
We investigated Wolbachia diversity at a fine scale with Homologous recombination in bacteria can occur by
respecttohostdiversity,bycomparingstrainsfromnine transformation, conjugation, or transduction.
Rosetal.BMCMicrobiology2012,12(Suppl1):S13 Page9of15
http://www.biomedcentral.com/1471-2180/12/S1/S13
Figure516S rDNA,gyrB,andconcatenatedML phylogeniesforCardinium. Samplecode(Additionalfile1)and hostspeciesnamein
whicheachstrainwasdetectedareindicated:BR=B.rubrioculus;BS=B.sarothamni;PH=P.harti.TwocladesarenamedIandII.MLbootstrap
values(topnumber,bold)andBayesianposteriorprobabilities(bottomnumber,plain)aredepicted(onlyvalueslargerthan50areindicated).
Thebaratthebottomindicatesabranchlengthof10%likelihooddistance.
Rosetal.BMCMicrobiology2012,12(Suppl1):S13 Page10of15
http://www.biomedcentral.com/1471-2180/12/S1/S13
Conjugation and transformation require physical con- prevalent for Cardinium. Horizontal transmission
tact, or close proximity, of donor DNA and recipient seemed also to explain diversity patterns found for Car-
bacteria. Ecological circumstances may create opportu- dinium infecting Cybaeus spiders [17]. Patterns of
nities for recombination, e.g., Wolbachia strains from B. recombination and horizontal transfer should however
sarothamni and B. berlesei share three of the four be further studied including more genes. An MLST set
alleles, and their mite hosts feed on the same host plant for Cardinium is desirable, for reliable strain typing and
species (we found both species co-occurring on the for investigating patterns of recombination, horizontal
same individual plant). Other ecological interactions transmission, or host manipulation. This requires the
have been suggested as means for bacteria or gene use of several independent markers, sufficiently distant
exchange, e.g., host-parasite interactions or double Wol- from each other within the genome.
bachia infections [28,36,45]. However, in many other
cases, opportunities for recombination are less obvious. Phenotypic effects of Wolbachia and Cardinium in spider
Transduction involving vectors (e.g., plasmids, phages, mites
or viruses) is a more likely manner of gene exchange. We analyzed Wolbachia and Cardinium strains from
Good vector candidates are bacteriophages, as these both asexual and sexual host species. Weeks and
have been isolated from Wolbachia infected populations Breeuwer [48] showed that Wolbachia is involved in
[60-62] and seem to be common in Wolbachia genomes causing asexuality in at least two species: B. praetiosa
[42,63]. Phylogenetic analyses suggest that the bacter- and an unidentified species. Wolbachia is possibly caus-
iophage WO is horizontally transferred between differ- ing asexuality in the other infected asexual Bryobia spe-
ent Wolbachia strains, and is able to infect new cies as well. The general observation is that all
Wolbachia hosts [60,61,64]. Other, free-living, bacteria individuals within the asexual Bryobia species are
might even be involved in phage-transfer. We also infected with Wolbachia. No males have ever been
noted the presence of a bacteriophage in an individual observed, neither in cultures nor in the field, and addi-
of B. spec. I. The bacteriophage sequence, detected coin- tional lab experiments including at least 20 individuals
cidentally with groEL primers, appeared similar to the per species (except for B. berlesei) show a fixed infection
sequence of the Wolbachia bacteriophage WOcauB1 with Wolbachia (unpublished data). Moreover, Weeks
from Cadra cautella (GenBank: AB161975; 12% p-dis- and Breeuwer [48] analyzed 240 B. kissophila, 144 B.
tance) [65], and to part of the sequenced genome praetiosa, and 24 B. rubrioculus individuals and found
(located within the gene dnaA) of Wolbachia from Dro- all individuals infected with Wolbachia. We detected
sophila melanogaster (GenBank: AE017196; 11% p- Cardinium in one asexual species, B. rubrioculus. This
distance). species is doubly infected with both Wolbachia and Car-
With strict vertical transmission, strong linkage dise- dinium, although Cardinium was not found in all indivi-
quilibrium between host mtDNA and Wolbachia would duals. It is unclear if Cardinium is having an effect on
be expected. However, recombination may uncouple the host species, but it is unlikely that it induces the
such associations, and could be a reason for the asexuality as not all individuals are infected. We
observed lack of congruence between host mtDNA and detected both Wolbachia and Cardinium in the sexually
Wolbachia STs. There are some signs of congruence, reproducing species B. sarothamni and T. urticae. Both
with related host strains (with identical COI sequences) species appear polymorphic for infection with both bac-
sharing identical or closely related Wolbachia strains, teria. Cardinium induces strong CI in B. sarothamni,
but due to the high rate of recombination such associa- while no effect for Wolbachia has been found so far
tions are broken up rather quickly. [47]. Previously, Wolbachia was found inducing CI in T.
urticae [66-69], but no effect of Cardinium on T. urticae
Cardinium diversity was found so far [68]. We detected only Cardinium in
For Cardinium, the two investigated genes showed P. harti, but Weeks et al. [2] also report Wolbachia
highly similar phylogenies, giving no clear evidence for from P. harti. The effects of both Wolbachia and Cardi-
intergenic recombination. Also, no signs of intragenic nium in P. harti, andT. urticae require further
recombination were found. There was however no con- investigation.
gruence between Cardinium strains and associated host
species: similar strains were found in B. rubrioculus, B. Conclusions
sarothamni, and T. urticae. Only the strain infecting P. We found a relatively high rate of recombination for
harti was clearly distinct from all other strains. The Wolbachia strains obtained from host species of the
sharing of strains among different host species, and the family Tetranychidae. Considering the fact that Wolba-
occurrence of divergent strains in one host population chia is widely distributed among arthropods, we investi-
(FR21), suggest that horizontal transmission is also gated strains from a restrictive host range. It remains to