Evolutionary transition in symbiotic syndromes enabled diversification of phytophagous insects on

18 Pages·2015·3.33 MB·English

by  Sailendharan Sudakaran

TheISMEJournal(2015)9,2587–2604
©2015InternationalSocietyforMicrobialEcology Allrightsreserved 1751-7362/15
www.nature.com/ismej
ORIGINAL ARTICLE
Evolutionary transition in symbiotic syndromes
enabled diversification of phytophagous insects on
an imbalanced diet
Sailendharan Sudakaran1, Franziska Retz1, Yoshitomo Kikuchi2, Christian Kost3,4 and
Martin Kaltenpoth1,5
1InsectSymbiosisResearchGroup,MaxPlanckInstituteforChemicalEcology,Jena,Germany;2Bioproduction
Research Institute, National Institute of Advanced Industrial Science and Technology (AIST) Hokkaido,
Sapporo, Japan; 3Experimental Ecology and Evolution Research Group, Max Planck Institute for Chemical
Ecology, Jena, Germany and 4Institute of Microbiology, Friedrich Schiller University, Jena, Germany
Evolutionary adaptations for the exploitation of nutritionally challenging or toxic host plants
represent a major force driving the diversification of phytophagous insects. Although symbiotic
bacteria are known to have essential nutritional roles for insects, examples of radiations into
novel ecological niches following the acquisition of specific symbionts remain scarce. Here we
characterized the microbiota across bugs of the family Pyrrhocoridae and investigated whether the
acquisition of vitamin-supplementing symbionts enabled the hosts to diversify into the nutritionally
imbalanced and chemically well-defended seeds of Malvales plants as a food source. Our results
indicate that vitamin-provisioning Actinobacteria (Coriobacterium and Gordonibacter), as well as
Firmicutes (Clostridium) and Proteobacteria (Klebsiella) are widespread across Pyrrhocoridae, but
absent from the sister family Largidae and other outgroup taxa. Despite the consistent association
with a specific microbiota, the Pyrrhocoridae phylogeny is neither congruent with a dendrogram
based on the hosts’ microbial community profiles nor phylogenies of individual symbiont strains,
indicating frequent horizontal exchange of symbiotic partners. Phylogenetic dating analyses based
on the fossil record reveal an origin of the Pyrrhocoridae core microbiota in the late Cretaceous
(81.2–86.5 million years ago), following the transition from crypt-associated beta-proteobacterial
symbionts to an anaerobic community localized in the M3 region of the midgut. The change in
symbiotic syndromes (that is, symbiont identity and localization) and the acquisition of the
pyrrhocorid core microbiota followed the evolution of their preferred host plants (Malvales),
suggesting that the symbionts facilitated their hosts’ adaptation to this imbalanced nutritional
resource andenabledthe subsequentdiversification inacompetition-poor ecological niche.
TheISME Journal (2015) 9, 2587–2604; doi:10.1038/ismej.2015.75; published online29 May2015
Introduction race, with plants continuously evolving novel che-
mical defenses or imbalanced nutritional composi-
The evolutionary success of herbivorous insects and
tiontoreduceherbivoreattacks,andinsectsadapting
their diversification into a wide range of ecological
by developing strategies to overcome defenses and
niches is closely connected to the diversification of
nutritional challenges. Thus, the diversification of
their host plants (Ehrlich and Raven, 1964). Herbi-
terrestrialplantsopenedupamultitudeofecological
vores and plants engage in an evolutionary arms
niches, permitting the adaptive radiation of herbi-
vorous insects (Farrell and Mitter, 1994). This is
probably best exemplified by the coevolution
Correspondence: M Kaltenpoth, Department for Evolutionary between butterflies and their host plants, with the
Ecology, Institute for Zoology, Johannes Gutenberg University diversification of several lepidopteran lineages fol-
Mainz,Mainz,Germany.
lowing their adaptation to a particular group of
or C Kost, Experimental Ecology and EvolutionResearch Group,
chemically well-defended plants (Ehrlich and
MaxPlanckInstituteforChemicalEcology,Hans-Knoell-Strasse8,
07745Jena,Germany. Raven, 1964), for example, Pieridae butterflies on
E-mail:[email protected]@gmail.com their Brassicales host plants (Wheat et al., 2007).
5Current address: Department for Evolutionary Ecology, Institute However, host plant selection and exploitation as
for Zoology, Johannes Gutenberg University Mainz, Mainz,
anutritionalresourcearenotonlydeterminedbythe
Germany.
metabolic capabilities of the insects themselves, but
Received 24 November 2014; revised 25 March 2015; accepted
3April2015;publishedonline29May2015 also their associated microbiota (Douglas, 2009;
Symbiont-enabledradiationofphytophagousinsects
SSudakaranetal
2588
Hosokawa et al., 2007). Microbial symbionts can for the functionality of the symbioses, the evolu-
confer important ecological traits to their hosts, tionary consequences of changes in pentatomomor-
including contributions to digestion (Breznak and phan symbiotic syndromes remain enigmatic.
Brune, 1994; Warnecke et al., 2007; Lundgren and Among pentatomomorphan bugs, the Pyrrhocor-
Lehman, 2010), detoxification (Dowd 1989; Genta idaeappeartobeexceptionalwithregardtoboththe
et al., 2006) and nutrient provisioning (Borkott and localization of the symbionts and the microbiota
Insam, 1990; van Borm et al., 2002). Consequently, composition. Previous studies on Pyrrhocoris
suchsymbioticinteractionscanhaveacrucialrolein apterus and Dysdercus fasciatus revealed that they
the evolutionary diversification of herbivorous harbor a distinct and stable microbiota consisting of
insects by facilitating expansion into novel ecologi- obligate and facultative anaerobes including Actino-
cal niches (Moran, 2007; Janson et al., 2008). bacteria (Coriobacterium glomerans and Gordoni-
Accordingly, expansion of the host plant range bactersp.),Firmicutes(Clostridiumsp.)andGamma-
and/or an increased diversification have been Proteobacteria (Klebsiella sp.). These bacteria are
observed in gall midges after the acquisition of localizedintheventricoseregion(M3)ofthemidgut
fungal symbionts (Joy, 2013). Furthermore, the (HaasandKönig,1987;Sudakaranetal.,2012;Salem
replacement of an ancestral beta-proteobacterial et al., 2013), which is the main region for the
symbiont in sharpshooters (Cicadellidae) with digestion of the ingested food particles (Silva and
Baumannia—a symbiont with a comparatively large Terra, 1994; Kodrík et al., 2012). Concordantly, the
genome (686kb) encoding for pathways to produce midgut crypts of Pyrrhocoris apterus and Dysdercus
vitamins and cofactors in addition to amino acids— fasciatus are reduced in size and do not contain
likely facilitated the shift from phloem sap as the any symbiotic microbes (Glasgow, 1914; Sudakaran
main nutrient source to the even more nutritionally et al., 2012). Similar crypt morphologies have been
imbalanced xylem sap (Takiya et al., 2006). How- reported for other genera such as Antilochus and
ever, despite the wealth of information that is Probergrothius, suggesting that the M3-associated
available on the benefits microbes can provide to microbial community may be widespread among
theirinsecthosts,theroleofsymbiontsindrivingthe Pyrrhocoridae (Glasgow, 1914; Rastogi, 1964; Bentz
diversification of insects and their expansion into and Kallenborn, 1995; Singh and Singh, 2001; Goel
novel ecological niches remains poorly understood and Chatterjee, 2003). In Pyrrhocoris apterus and
(Janson et al., 2008). Dysdercusfasciatus,thegutmicrobiotawasfoundto
Within the megadiverse insect order Hemiptera, be vital for growth and survival of the host (Salem
the infraorder Pentatomomorpha contains over et al., 2013), through the supplementation of
12500 species (Schaefer, 1993; Schuh and Slater, B vitamins by the dual actinobacterial symbionts
1995;Henry,1997),manyofwhichharborbeneficial C. glomerans and Gordonibacter sp. (Salem et al.,
symbiontsthatcontributesignificantlytohostfitness 2014).AsthepredominantfoodsourceofPyrrhocor-
(Muller, 1956; Huber-Schneider, 1957; Schorr, 1957; idae, that is, seeds of the plant order Malvales
Abe et al., 1995; Fukatsu and Hosokawa, 2002; (Kristenová et al., 2011), is deficient in B vitamins,
Kikuchi et al., 2009; Tada et al., 2011; Salem et al., this symbiont-mediated nutritional upgrading plays
2013). Interestingly, symbiotic syndromes (that is, an important role by allowing the hosts to exploit
identity and localization of the symbionts) vary a nutritionally inadequate diet (Salem et al., 2013).
greatly among Pentatomomorpha, indicating fre- In this study, we aimed at elucidating the
quent transitions during the evolutionary history of ecological and evolutionary implications of a major
this group. The most common symbiont-bearing transition in symbiotic syndromes. Specifically, we
organsacrossthesuperfamiliesLygaeoidea,Coreoidea testedthehypothesisthattheevolutionarytransition
and Pentatomoidea are specialized sacs or tubular to a characteristic midgut core microbiota enabled
outgrowths, called crypts or gastric ceca, in the the diversification of pyrrhocorid bugs on the
posterior region of the midgut that harbor beneficial nutritionally imbalanced diet of Malvales seeds.
symbionts belonging to the gamma- or beta- To this end, we characterized the microbiota across
proteobacteria (Glasgow, 1914; Miyamoto, 1961; 25 species of Pyrrhocoroidea (22 Pyrrhocoridae and
Buchner, 1965; Fukatsu and Hosokawa, 2002; threeLargidaespecies)throughacombinationof454
Prado and Almeida, 2009; Hosokawa et al., 2010; pyrosequencing and quantitative PCR. In addition,
Kikuchi et al., 2011a,b). However, several other we reconstructed a dated phylogeny of the hosts
symbiotic syndromes occur across Pentatomomor- through calibration with the fossil record and
pha, including paired or unpaired bacteriomes with compareditwithadistancedendrogramofmicrobial
intracellular symbionts in some Lygaeoidea community profiles, as well as strain-level phyloge-
(Kuechleretal.,2012;Matsuuraetal.,2012),aswell nies of the two vitamin-provisioning actinobacterial
as more complex microbial communities in midgut symbionts. The results allow us to assess
regions devoid of crypts (in Pyrrhocoroidea; the distribution of the characteristic M3 midgut
Sudakaran et al., 2012). Thus, evolutionary transi- microbiota across bugs of the superfamily Pyrrho-
tions in symbiotic syndromes must have occurred corideaandtoidentifytheevolutionaryoriginofthis
repeatedly in Pentatomomorpha. Although such symbiotic syndrome. Subsequently, a comparison
transitions are expected to have major implications withtheageofMalvalesplantsallowedustotestthe
TheISMEJournal
Symbiont-enabledradiationofphytophagousinsects
SSudakaranetal
2589
hypothesis that the acquisition of a specific micro- 16mM (NH ) SO and 0.01% Tween 20), 2.5mM
4 2 4
biota preceded the bugs’ diversification on this MgCl ,240μMdNTPs,0.8μMofeachprimerand0.5
2
nutrient-deficient food source. Furthermore, host– U of Taq DNA polymerase (VWR International
symbiont co-cladogenetic analyses shed light on the GmbH,Darmstadt,Germany).Cycleparameterswere
evolutionary stability and maintenance of the char- as follows: 3min at 94°C, followed by 35 cycles of
acteristic core microbiota in Pyrrhocoridae. Taken 94°C for 40s, 55°C for 40s and 72°C for 40s, and
together, the results provide novel insights into the a final extension step of 4min at 72°C. PCR
evolutionary transitions and ecological relevance of products were then sequenced bidirectionally on an
symbiotic microbial communities in the diverse ABI 3730xl capillary DNA sequencer (Applied
insect order Hemiptera. Biosystems, Foster City, CA, USA). Protein-coding
sequences (COI and COII) were curated and then
aligned based on their amino-acid translation in
Materials and methods Geneious Pro 5.4 (Biomatters, Auckland, New
Zealand), whereas partial 18S rRNA sequences were
Insect sample collection and DNA extraction alignedusingtheSINAaligner(Pruesseetal.,2012).
For characterizing the microbial community and The individual alignments were concatenated and
reconstructingsymbiontandhostphylogeniesacross usedforphylogeneticreconstructionwithmaximum
the Pyrrhocoroidea superfamily and outgroup taxa, likelihood algorithms (ML) and Bayesian Inference
live adult specimens of Pyrrhocoridae (22), Largidae (BI), respectively. An ML tree was computed with
(3), Lygaeidae (2), Oxycarenidae (1) and Rhopalidae FastTree2.1usingtheGTRmodel,andlocalsupport
(1) were collected from their respective habitats values were estimated with the Shimodaira–Hase-
across four different continents (Supplementary gawa test based on 1000 resamplings without
Table S1). Bugs were killed and preserved in 70% reoptimizing the branch lengths for the resampled
ethanol until further analysis, and at least one alignments (Price et al., 2010). For BI (computed
individual per species was kept in ethanol as a using MrBayes 3.1.2; Huelsenbeck and Ronquist,
voucher specimen. Before DNA extraction, samples 2001), the data set was partitioned into the three
were surface sterilized by rinsing with sterile Milli- genes, with six substitution types for the CO genes
pore water, 1% sodium dodecyl sulfate, and then (GTR model), and one for the ribosomal gene (F81
again sterile Millipore water (Billerica, MA, USA). model). Owing to saturation in substitutions, third
Up to six complete specimens per bug species (or codonpositionswere excluded from the analysis for
fewer, if less than seven individual specimens were the two mitochondrial genes (COI and COII). The
available), were homogenized under liquid nitrogen analysis was performed with four chains and a
withsterilepestles.FortheJapanesebugspecimens, temperature of 0.2 for 10000000 generations, and
however, only the already dissected midgut was we confirmed that the standard deviation of split
available and used instead of whole individuals. frequencies was consistently below 0.01. Trees were
DNA was extracted using the MasterPure DNA sampled every 1000 generations, and a ‘burn-in’ of
Purification Kit (Epicentre Technologies, Madison, 1000 was used (=10%). We computed a 50%
WI, USA) according to the manufacturer’s instruc- majorityruleconsensustreewithposteriorprobabil-
tions. An additional lysozyme incubation step ities for every node.
(30min at 37°C; 4μl of 100mgml–1 lysozyme,
Sigma-Aldrich, St Louis, MO, USA) was included
before proteinase K digestion to break up Gram- Dating of the host phylogeny
positive bacterial cells (see Sudakaran et al., 2012). Divergence time estimations for the Pyrrhocoroidea
Individual extracts were used for quantitative PCR superfamily were inferred using BEAST v1.8.0
(qPCR)analysis,aswellasforPCRandsequencingof (Drummond and Rambaut, 2007), by testing various
host and symbiont genes for phylogenetic analysis. substitution models and parameter settings
PooledDNAextractsfromeachspecieswereusedfor (see Supplementary Table S2 and S3) on a fixed
454 pyrosequencing of the associated bacterial input tree (the BI tree, see above). Two fossil
communities. calibration points were used: (i) two Dysdercus
fossils from Florissant beds in Colorado (37.0–33.1
million years ago (mya)) (Scudder, 1890), and (ii)
Reconstruction of the host phylogeny a Pyrrhocoris tibialis fossil from Rott-am-
The phylogenetic relationships among the Pyrrho- Siebengebirge in Germany (28.5–23.8mya) (Statz
coridae, its sister family Largidae and outgroup taxa and Wagner, 1950). A hard upper boundary for the
were reconstructed using PCR amplification and ageoftherootwassetto160mya,basedontotheage
sequencing of two mitochondrial (cytochrome of the oldest Pentatomomorpha fossil, as well as the
oxidase I and II) and one nuclear gene (18S estimated age of the Pentatomomorpha infraorder
ribosomal RNA (rRNA)) for all host species using (~152.9mya, Upper Jurassic) (Li et al., 2012; Misof
primers listed in Table 1. PCR was performed in a et al., 2014). Evaluation and comparison of model
total reaction volume of 12.5μl, containing 1μl of parameters were performed using Tracer v1.5
template DNA, 1×PCR buffer (20mM Tris-HCl, (Drummond and Rambaut, 2007). The maximum
TheISMEJournal
Symbiont-enabledradiationofphytophagousinsects
SSudakaranetal
2590
11
00
edmicrobiota Reference Lietal.,2005Lietal.,2005Lietal.,2005Lietal.,2005Lietal.,2005Lietal.,2005Lietal.,2005Lietal.,2005Simonetal.,1994Simonetal.,1994Kaltenpothetal.,2009Weisburgetal.,1991utin-Ganacheetal.,20utin-Ganacheetal.,20Sudakaranetal.,2012Sudakaranetal.,2012Sudakaranetal.,2012Sudakaranetal.,2012ThisstudyThisstudyThisstudyKaltenpothetal.,2009Ishaketal.,2011Ishaketal.,2011
ciat BoBo
o
ss e
a Us 111111111122223333333344
r
ei
dth gth 433235008001890000098998
n n 222222221222112222211111
a e
a L
e
d
oi v.
rrhocor Fwd./re Fwd.Fwd.Rev.Rev.Fwd.Rev.Fwd.Rev.Rev.Fwd.Fwd.Rev.Fwd.Rev.Fwd.Rev.Fwd.Rev.Rev.Fwd.Fwd.Rev.Fwd.Rev. encing.
y u
P q
(PCR,cloning/sequencing,454pyrosequencing)andquantification(qPCR)of ′–4′TargetgenePrimernamePrimersequence(53) 18SrRNAPyr18S_2FGGGAGGTAGTGACAAAAAATAACG18SrRNAPyr18S_4FATCCTTTAACGAGGATCTATTGG18SrRNAPyr18S_3RACATACTTGGCAAATGCTTTCGC18SrRNAPyr18S_4RGTTAGAACTAGGGCGGTATCTGCOIC1-J-2183-FCAACATTTATTTTGATTTTTTGGCOITL2-N-3014-RTCCAATGCACTAATCTGCCATATTACOIC1-J-2530-FGGAGTAATTCTAGCCAACTCCOIC1-N-2609-RGAATACTGCTCCTATGGATACOIITK-N3796_revACTATTAGATGGTTTAAGAGCOIITL-J3033_fwdTCTAATATGGCAGATTAGTGCA16SrRNACor-2FGGTAGCCGGGTTGAGAGACC16SrRNArP2ACGGCTACCTTGTTACGACTT16SrRNAM13FTGTAAAACGACGGCCAGT16SrRNAM13RGGAAACAGCTATGACCATG16SrRNAClostridium_1050-fwdCTCGTGTCGTGAGATGTTGG16SrRNAClostridium_1248-revGCTCCTTTGCTTCCCTTTGT16SrRNAKlebsiella_250-fwdCAGCCACACTGGAACTGAGA16SrRNAKlebsiella_453-revGTTAGCCGGTGCTTCTTCTG16SrRNAD.fas_Egg.2R_qPCCCGTATCTCAGTCCCAATGT16SrRNAAct-2FGCGAACGGGTGAGTAACAC16SrRNAGray519FCAGCMGCCGCNGTAANAC16SrRNACor-1RACCCTCCCMTACCGGACCC16SrRNAGray28FGAGTTTGATCNTGGCTCA16SrRNAGray519RGTNTTACNGCGGCKGCTG PCR;Rev.,reverse;rRNA,ribosomalRNA.es,(2)cloning/sequencingofCoriobacteriaceaesymbionts,(3)qPCRand(4)454se
acterization onibacter quantitativegofhostgen
echar /Gord qPCR,uencin
Table1Primersusedforth TargetTargettaxon HostsHeteropteraHeteropteraHeteropteraHeteropteraHeteropteraHeteropteraHeteropteraHeteropteraHeteropteraHeteropteraSymbiontsCoriobacteriumEubacteriaEubacteriaEubacteriaClostridiumClostridiumKlebsiellaKlebsiellaGordonibacterGordonibacterCoriobacteriumCoriobacteriumEubacteriaEubacteria Abbreviations:Fwd.,forward;Use:(1)amplificationandseq
TheISMEJournal
Symbiont-enabledradiationofphytophagousinsects
SSudakaranetal
2591
clade credibility (consensus) tree was inferred with Quantification of core microbes
TreeAnnotator using a burn-in of 5,000 and qPCRswereperformedforthefourdominantbacterial
a posterior probability limit of 0.5 (Drummond and symbionts in Pyrrhocoridae (Coriobacterium glomer-
Rambaut, 2007). The consensus tree was visualized ans,Gordonibactersp.,Clostridiumsp.andKlebsiella
with FigTree v1.3.1 (http://tree.bio.ed.ac.uk/soft sp.), using specific 16S rRNA primers (Table 1) on a
ware/figtree/), including highest posterior density RotorgeneQcycler(Qiagen,Hilden,Germany)infinal
(HPD) intervals. reaction volumes of 25μl, containing 1μl of template
DNA (usually a 1:10 dilution of the original DNA
Characterization of microbial community profiles extract), 2.5μl of each primer (10μm) and 12.5μl of
Bacterial tag-encoded FLX amplicon pyroseqencing SYBR Green Mix (Rotor-Gene SYBR Green kit,
(bTEFAP) was performed to characterize the micro- Qiagen). Standard curves were established using
bial community composition of members belonging 10−8–10−2 ng of specific PCR product as templates
to Pyrrhocoridae, Largidae and outgroup taxa. DNA for the qPCR. A NanoDrop 1000 spectrophotometer
was sent to external service providers (Research & (Peqlab Biotechnology Limited, Erlangen, Germany)
TestingLaboratories,Lubbock,TX,USA,orMRDNA was used to measure DNA concentrations for the
Lab, Shallowater, TX, USA), and amplification was templatesofthestandardcurve.PCR conditionswere
achieved using the 16S rRNA primers Gray28F and as follows: 95°C for 5min, followed by 35 cycles of
Gray519R (Table 1) (Ishak et al., 2011, Sun et al., 60°C for 30s, 72°C for 20s and 95°C for 15s; then a
2011). Sequencing libraries were generated through melting curve analysis was performed by increasing
one-step PCR with 30 cycles, using a mixture of Hot the temperature from 60°C to 95°C within 20min.
Start and HotStar high-fidelity Taq polymerases The efficiencies of all four quantitative PCR assays
(Qiagen, Valencia, CA, USA). Sequencing extended were confirmed to be 499%. Based on the standard
from Gray28F, using a Roche 454 FLX instrument curves, the 16S copy numbers of the four dominant
(Branford, CT, USA) with Titanium reagents and symbiontscouldbecalculatedforeachindividualbug
procedures. All low-quality reads (quality cut-off from the qPCR threshold values (Ct) by the absolute
=25) and sequences o200bp were removed follow- quantification method (Lee et al., 2006, 2008), taking
ing sequencing, which left between 1990 and 30361 the dilution factor and the absolute volume of DNA
sequences per sample for subsequent analysis. extract into account. The absolute 16S copy numbers
Processing of the high-quality reads was performed were log transformed and then used to visualize the
using QIIME (Caporaso et al., 2010b). Sequences were quantitative differences of the bacterial symbionts
clustered into operational taxonomic units (OTUs) across different host genera using box plots.
using multiple OTU picking in combination with
chimera checking using usearch (Edgar, 2010) fol-
lowed by cdhit (Fu et al., 2012) with 97% similarity Symbiont (Coriobacteriaceae) strain-level phylogenies
cut-offs. For each OTU, one representative sequence In order to reconstruct strain-level phylogenies for
was extracted (the most abundant) and aligned to the the two actinobacterial symbionts Coriobacterium
Greengenes core set (available from http://greengenes. and Gordonibacter, we followed two different stra-
lbl.gov/)usingPyNast(Caporasoetal.,2010a),withthe tegies, based on (i) the bTEFAP sequencing data
minimum sequence identity set to 75%. Taxonomy alone, and (ii) a combination of bTEFAP sequences
was assigned using the Ribosomal Database Project and Sanger sequencing of cloned 16S rRNA ampli-
(RDP) classifier (Wang et al., 2007), with a minimum cons. For the first approach, OTUs were picked
confidence to record an assignment set to 0.80. individually for each host species, using the para-
AnOTUtablewasgenerateddescribingtheabundance meters as described above. This was necessary to
of bacterial phylotypes within each sample conserve differences among symbiont strains that
(Supplementary Table S4). The table was then manu- were below 3% sequence divergence, which would
ally curated by removing low-abundance OTUs be lost in the combined OTU picking strategy used
(o0.1%ineachofthesamples)andthroughBLASTn for assessing the general microbial community
of the representative sequences (see Supplementary composition in Pyrrhocoridae. For each OTU, the
Data S1) against the NCBI and RDP databases. To longest representative sequence was extracted, and
visualize the results, OTUs with the same genus-level sequences affiliated with the family Coriobacteria-
assignments were combined based on the BLASTn ceaewereextractedafterRDPandBLASTclassifica-
results. The genus-level table was used to construct tion (see Supplementary Data S2). The resulting
heatmapsusingtheRpackage‘gplots(heatmap.2)’.For representative sequences were aligned to reference
beta-diversityanalysisandUPGMAclustering,theraw sequences of all Coriobacteriaceae type strains
OTU table was subsampled to the depth of 1500 obtained from the RDP (Cole et al., 2014) using the
sequences per sample, and distance matrices were SINAaligner(Pruesseetal.,2012),andphylogenetic
calculated using Bray–Curtis and Jaccard metrics. For relationships were computed using ML as described
visualization, two-dimensional principal coordinate for the reconstruction of the host phylogeny.
analysis plots and dendrograms based on UPGMA For the second approach, the bTEFAP data were
clusteringwereconstructedbasedonthebetadiversity complemented byacloning/sequencingapproachin
distance matrices. order to obtain longer and hence more informative
TheISMEJournal
Symbiont-enabledradiationofphytophagousinsects
SSudakaranetal
2592
reads for phylogenetic analysis. For this purpose, with the resulting distribution of codiversification
PCR amplifications of the 3′-region of the 16S events in the randomized data set. For Parafit
rRNA of both Coriobacteriaceae symbionts (that is, analysis,ahostdistancematrixwascomputedusing
Coriobacterium glomerans and Gordonibacter sp.) theRpackage‘Ape(cophenetic.phylo)’basedonthe
were carried out with the primers Cor-2F and rP2 phylogenetic tree, and symbiont distance matrices
using a Biometra thermocycler (Analytik Jena, Jena, were computed in BioEdit 7.0.5.3 (Hall, 1999) based
Germany) in total reaction volumes of 12.5μl on the concatenated alignments. Permutation tests
containing 1μl of template DNA, 1×PCR buffer (1000replicates)wererunasimplementedinParaFit
(20mM Tris-HCl, 16mM (NH ) SO and 0.01% (Legendreetal.,2002).Inordertoassessthepossible
4 2 4
Tween 20), 2.5mM MgCl , 240μM dNTPs, 0.8μM obscuring effect of interspecific predation on co-
2
of each primer and 0.5U of Taq DNA polymerase phylogeneticpatterns,werepeatedtheanalysesafter
(VWR International GmbH). Cycle parameters were omission of known carnivorous host taxa (Antilo-
as follows: 3min at 94°C, followed by 35 cycles of chusspp.,Dindymuslanius)thatmayhaveacquired
94°C for 40s, 68°C for 40s and 72°C for 40s, and a the Coriobacteriaceae symbionts horizontally via
final extension step of 4min at 72°C. PCR products feeding on heterospeficic pyrrhocorid bugs.
were cloned into Escherichia coli using the Strata-
Clone PCR Cloning Kit (Stratagene, Agilent Tech- Results
nologies, Santa Clara, CA, USA) according to
the manufacturer's instructions. Transformed E. coli Host phylogeny and divergence time estimates
cells were grown on LB agar containing 10mgml–1 To elucidate the evolutionary origin of the Pyrrhocor-
ampicillin and 2% 5-bromo-4-chloro-indolyl- idae–microbiotaassociation,thephylogeneticrelation-
β-d-galactopyranoside (X-gal) (Sigma-Aldrich) for ships across bugs of the Pyrrhocoroidea superfamily
blue/white screening. Colony PCR was performed were reconstructed. The combination of partial 18S
on eight randomly selected transformants for each rRNA, COI and COII gene sequences resolved most of
insect host with vector primers M13F and M13R thetaxonomicrelationshipswithinthePyrrhocoroidea
(Table 1) using the above-mentioned reaction mix (Figure 1b and Supplementary Figure S1), and
and cycling conditions, except that an annealing divergence time estimations based on two fossil
temperature of 55°C was used. PCR products were calibration points and a hard lower boundary for the
checked for the expected size on a 1.5% agarose gel root age yielded consistent age estimates for selected
(130V, 30min) and purified using the peqGOLD nodesofinterestacrossarangeofdifferentsubstitution
MicroSpin Cycle Pure Kit (Peqlab Biotechnologies models(GTR+I+G,HKY+G,HKY+I+G,TN93+G,TN93
GmbH, Erlangen, Germany) before sequencing with +I+G)andparametersettings(SupplementaryTableS2
M13 primers. Nearly full-length Coriobacterium and S3). Omitting the Pyrrhocoris tibialis fossil
glomerans and Gordonibacter sp. 16S rRNA calibration point did not affect age estimates, whereas
sequences from different Pyrrhocoridae were omitting either the Dysdercus fossil calibration or the
obtained by combining the short sequences from root boundary resulted in significantly increased age
OTUs picked for each individual species with estimates (Supplementary Table S3). Based on the
bTEFAP and the sequences obtained by PCR/cloning tracer analysis of effective sample sizes and marginal
for the respective OTUs. In cases with multiple likelihood values, the TN93+I+G model with two
Coriobacterium (for Scanthius aegypticus, Scanthius codon partitions for the protein-coding genes (1+2,
obscurus, Pyrrhocoris apterus, Pyrrhocoris sibiricus and 3), estimated base frequencies, and a relaxed
and Dysdercus fasciatus) or Gordonibacter OTUs uncorrelated lognormal clock model yielded the most
(for Scanthius aegypticus and Dysdercus fasciatus) robust results.
perhostspecies,themostabundantOTUwaschosen, The phylogenetic analyses revealed an estimated
and the identity of bTEFAP and cloned sequences age of 135.7mya for the superfamily Pyrrhocoroidea
wasconfirmedintheoverlappingregiontoreducethe (95% HPD interval: 104.4–159.9mya, Figure 1b).
risk of chimera formation. Sequence alignment and The families Pyrrhocoridae and Largidae formed
phylogenetic tree reconstruction using ML and BI monophyletic sister taxa that split about 125.4mya
were done as described above. (95% HPD interval: 92.9–154.6mya; Figure 1b).
Within the Pyrrhocoridae family, the genus Prober-
Cophylogenetic analysis of host and symbiont grothius diverged from the common ancestor of all
To test for codiversification between hosts and their other taxa about 86.5mya (95% HPD interval: 63.1–
Coriobacteriaceae symbionts, the phylogenies of the 109.3mya). Around 81.2mya (95% HPD interval:
two Coriobacteriaceae symbionts (based on the 58.4–102.2mya), the clade Dindymus+Antilochus
bTEFAP data alone or the combination with clon- split from the group comprising Dysdercus, Derma-
ing/sequencing data) were compared with the host tinus, Scanthius, Pyrrhocoris, the unknown pyrrho-
phylogeny using Treemap 3 (Page RDM, 1995) and corid taxon, and Cenaeus.
Parafit (Legendre et al., 2002). In Treemap, the
positions of taxa were randomized on the host and Microbial community composition of pyrrhocorid bugs
symbiont trees (1000 replicates), and the number of The microbiota of several species of Pyrrhocoridae
observed codiversification events was compared (n=22)andLargidae(n=3),aswellasoutgrouptaxa
TheISMEJournal
Symbiont-enabledradiationofphytophagousinsects
SSudakaranetal
2593
Figure1 Dated host phylogeny andmicrobiota profile of 22 species within the family Pyrrhocoridae as well as outgroups(Largidae,
Lygaeidae, Oxycarenidae and Rhopalidae). (a) Photographs of selected Pyrrhocoridae host species: adult and fifth instar nymph of
Pyrrhocoris apterus, adult Dysdercus cingulatus, adult and nymphs of Dysdercus fasciatus, and a mating pair of Probergrothius
sanguinolens (from left to right). (b) Phylogenetic relationships of the hosts (Pyrrhocoridae n=22, Largidae n=3, Lygaeidae n=2,
Oxycarenidaen=1, Rhopalidaen=1species),reconstructedusingpartial18SrRNA,cytochromeoxidaseIandcytochromeoxidaseII
genesequences.DivergencetimeestimateswerederivedusingBEASTanalyses(TN93+I+Gmodel).Selectednodeagesareshowninmya
with95%HPDintervalbars.Dashedbranchesrepresentpyrrhocoridtaxawithoutthecharacteristiccoremicrobiota.Thegreencolored
barindicatestheestimatedoriginofthehostplantorderMalvales(72–96mya)(Wangetal.,2009).(c)2DPrincipalCoordinateAnalysis
(PCoA)showingtheclusteringofhostspeciesbasedontheirmicrobialcommunityprofilesusingBray–Curtis(left)andJaccard(right)
distancematrices,respectively.Colorsforindividualsamplescorrespondtothecoloringoftaxainb.(d)Relativeabundanceofmicrobial
taxa as obtained from 454 pyrosequencing of 16S rRNA amplicons (305,179 reads in total), represented as a heatmap based on log-
transformedvalues.OTUswerecombinedonthegenuslevelforbettervisualization,andonlygenerathatamountto41%ofthemicrobial
communityinatleastoneofthehostspeciesaredisplayed.Thedendrogramabovetheheatmaprepresentstheclusteringofmicrobialtaxa
accordingtotheirdistributionandabundanceacrosshostspecies.Notethedistinctclusteringofthefourcoremicrobialtaxaassociated
withPyrrhocoridae.
TheISMEJournal
Symbiont-enabledradiationofphytophagousinsects
SSudakaranetal
2594
(n=4) were characterized using 454 amplicon Phylogenetic analysis of the Coriobacteriaceae
pyrosequencing of bacterial 16S rRNA (bTEFAP), symbionts
which yielded a total of 305179 sequences. After The Pyrrhocoridae core microbiota contains two
removing singletons, chimeric sequences and OTUs actinobacterial symbionts that were previously
below 0.1% abundance, the sequences were clus- shown to be essential for growth and survival in
tered into 356 OTUs. Bray–Curtis and Jaccard P. apterus and D. fasciatus through the supplemen-
clusteringofthehostspeciesbasedontheirbacterial tation of B vitamins (Salem et al., 2013, 2014). The
community profiles revealed a well-defined cluster phylogeny of both Coriobacteriaceae symbionts was
containing the genera Dysdercus, Scanthius, Pyrrho- reconstructed based on the set of short read
coris, Dindymus, Antilochus and the unknown sequences from bTEFAP representing the Coriobac-
Pyrrhocoridae species, asecond cluster with Prober- teriaceae symbiont OTUs from each pyrrhocorid
grothius, Dermatinus and Cenaeus, and separate species (Figure 3). In addition, we amplified and
clusters for members of the Largidae family and the sequenced the symbionts’ 3’-region of the 16S rRNA
outgroup composed of other Pentatomomorphan gene sequences from 13 different host species, as
bugs (Oxycarenus hyalinipennis, Leptocoris augur, well as only Coriobacterium glomerans from
Lygaeus equestris and Spilostethus hospes), respec- Dysdercus decussatus, and only Gordonibacter sp.
tively (Figure 1c). UPGMA dendrograms of the from two Probergrothius species (that is, P. nigricor-
bacterial communities associated with the host nis and P. sanguinolens). Subsequently these
species computed using the beta diversity matrices sequences were combined with the bTEFAP repre-
(Bray–Curtisand Jaccard) yielded qualitatively simi- sentative sequences to enhance the resolution of the
lartopologieswiththemajorgroupingsbeinglargely phylogenetic tree (Supplementary Figure S3). The
identical (Supplementary Figure S2). phylogenetic analyses revealed that the symbiotic
Combining OTUs on the genus level revealed that Coriobacterium glomerans and Gordonibacter sp.
themicrobiotaofPyrrhocoridaebugswasdominated strainsformtwodistinctmonophyleticcladeswithin
by four core bacterial taxa: Coriobacterium glomer- thefamilyCoriobacteriacae,whichisconsistentwith
ans, Gordonibacter sp. (Actinobacteria), Clostridium a single acquisition event for each symbiont and
sp. (Firmicutes) and Klebsiella sp. (Proteobacteria) subsequenthost–symbiontcoevolution(Figure3and
(Figure 1d). These taxa were consistently present Supplementary Figure S3). A possible exception are
across Pyrrhocoridae in abundances ranging from the Gordonibacter symbionts of the basal pyrrho-
104 to 108 16S rRNA gene copies per individual corid genus Probergrothius, which group within the
(Figure 2), with the exception of the genera Prober- monophyletic symbiont cluster in the combined
grothius,DermatinusandCenaeus,whichlackedthe phylogeny (Supplementary Figure S3), yet fall out-
Coriobacteriaceae symbionts (Figures 1d and 2). side when only the bTEFAP sequences are consid-
Furthermore, although OTUs associated with the ered (Figure 3). Interestingly, several host species
genera Clostridium and Klebsiella were present in contained two or more dominant OTUs for one or
most species of thesethree hostgenera,qPCR assays both of the actinobacterial symbionts. Specifically,
specific for the Pyrrhocoridae-associated Clostri- more than one Coriobacterium OTU was observed
dium and Klebsiella OTUs were negative for all for Scanthius aegypticus, Scanthius obscurus, Pyr-
samplesexcepttwooftheProbergrothiusspecimens, rhocoris apterus, Pyrrhocoris sibiricus and Dysder-
indicating that the Clostridium and Klebsiella OTUs cus fasciatus, whereas two or more Gordonibacter
associated with these three genera differ from those OTUs were detected for Scanthius aegypticus and
of the other Pyrrhocoridae (Figure 2). Thus, the host Dysdercus fasciatus. For Pyrrhocoris apterus and
genera Probergrothius, Dermatinus and Cenaeus Dysdercus fasciatus, the occurrence of two distinct
lacked the microbiota that is characteristic for other Coriobacterium sequences, respectively, was pre-
Pyrrhocoridae, which is also reflected in their viously confirmed using cloning and sequencing
separate clustering in the principal coordinate (Kaltenpoth et al., 2009). Thus, the multiple
analyses (Figure 1c). Coriobacterium and Gordonibacter OTUs observed
The microbiota of members of the family Largidae here for several host species likely reflect true
(the sister taxon to the Pyrrhocoridae) was dominated symbiont microdiversity rather than 454 sequencing
by Burkholderia and completely lacked the artifacts. Although at present we cannot exclude the
Pyrrhocoridae-associated core microbes (Figures 1d possibility that the multiple Coriobacteriaceae
and2).Similarly,thecoremicrobiotawasabsentfrom sequencesfoundinindividualPyrrhocoridaespecies
otherpentatomomorphanoutgroupspecies:themicro- represent divergent 16S rRNA copies within the
biota of both Oxycarenus hyalinipennis (Oxycareni- same symbiont genome, the presence of multiple
dae) and Leptocoris augur (Rhopalidae) was distinct strains seems much more likely given the
dominated by Wolbachia sp. and Bartonella sp., highdegreeofsimilarityofthetwo16SrRNAcopies
whereas the Lygaeidae species Lygaeus equestris and (99.72%) in the sequenced genome of C. glomerans
Spilostethus hospes contained consortia of Trabul- isolated from P. apterus (Stackebrandt et al., 2013).
siella sp. and Stenotrophomonas sp. (L. equestris), or In addition to the occurrence of multiple OTUs
Wolbachia sp. and Rickettsia sp. (S. hospes), within individual host species, the incongruence of
respectively. thephylogeniesofbothCoriobacteriaceaesymbionts
TheISMEJournal
Symbiont-enabledradiationofphytophagousinsects
SSudakaranetal
2595
Figure 2 Evolutionary transitions in symbiotic syndromes in Pyrrhocoroidea, and abundance of core microbial taxa. (a) Schematic
phylogenyofPyrrhocoridaeandLargidaegenera(adaptedfromFigure1b).Pyrrhocoridaetaxawithcoremicrobiotaaregiveninblack,
thosetaxawithoutthecoremicrobiotaareindarkgray,andtheLargidaewithcrypt-associatedsymbiontsareinlightgray.Reconstructed
evolutionarytransitionsinsymbioticsyndromesareindicatedonthephylogeny.Numbersofvalidlydescribedextantspeciesaregiven
behindeachgenusname(fromHussey,1929).(b)Abundanceofthefourcoresymbionttaxa(Coriobacteriumglomerans,Gordonibacter
sp., Clostridium sp. and Klebsiella sp.) across multiple specimens of the nine different genera of Pyrrhocoridae (Dysdercus (n=37),
Dermatinus(n=2), Scanthius(n=5), Pyrrhocoris(n=7), unknownPyrrhocoridae (n=6), Cenaeus(n=5), Dindymus (n=6), Antilochus
(n=7)andProbergrothius(n=23))andtwogeneraofLargidae(Physopelta(n=4)andLargidae(n=1)).Abundancewasassessedas16S
rRNAgenecopynumbersusingqPCR,basedononetosixreplicateindividualsperhostspecies,whichwerethencombinedonthegenus
level.Linesrepresentmedians,boxescomprisethe25–75percentilesandwhiskersdenotetherange.
withthePyrrhocoridaehostphylogeny(Coriobacter- apparently played an important role during the
ium glomerans: Parafit: P=0.974, TreeMap: evolutionofthissymbiosis.AstheCoriobacteriaceae
P=0.345; Gordonibacter sp.: Parafit: P=0.978, Tree- symbionts are localized in the same region of the
Map: P=0.449) (Supplementary Figure S3) suggest midgutandcanbeco-transmittedbothverticallyand
horizontal exchange of symbionts between host horizontally (Kaltenpoth et al., 2009), we also tested
species.Giventhesymbiontmicrodiversityobserved for co-cladogenesis of the two symbiont lineages.
in the bTEFAP data, we paid special attention to Randomization of phylogenetic trees or distance
avoid the generation of possible chimeric sequences matrices and subsequent statistical evaluation, how-
when combining bTEFAP reads and sequences ever, yielded no evidence for co-cladogenesis
obtained after cloning of PCR amplicons. Owing to betweenCoriobacteriumglomeransandGordonibac-
the high similarity of different symbiont strains, ter sp. strains across host taxa (Parafit: P=0.898,
however,thepossibilityofchimeraformationforthe TreeMap: P=0.251).
symbionts of individual host taxa cannot be com-
pletely ruled out, which may hinder accurate co-
phylogenetic analyses. To exclude the possibility Discussion
that co-phylogenetic patterns were additionally
obscured by interspecific predation among pyrrho- In this study, we characterized the microbiota
corid bugs resulting in transient Coriobacteriaceae associated with bugs of the hemipteran families
being picked up in the bTEFAP sequences, we Pyrrhocoridae and Largidae, and investigated the
repeated the analyses after excluding known pre- origin and evolutionary dynamics of the host–
datory taxa (Antilochus spp., Dindymus lanius). microbiota association on both the community and
Although some symbiont taxa clustered according strain level. The results provide insights into an
to their host genera (particularly Pyrrhocoris and evolutionary transition from individual crypt-
Scanthius for Gordonibacter, and Dysdercus for associated symbionts to a more complex microbiota
Coriobacterium), the tests for co-cladogenesis that is localized in the insect’s midgut. This transi-
remained nonsignificant. Thus, although the Pyr- tion coincided with the evolution of the hosts’
rhocoridae maintain a specific microbiota, horizon- preferred food plants and preceded the major
tal transmission between co-occurring species radiation of pyrrhocorid bugs, highlighting the
TheISMEJournal
Symbiont-enabledradiationofphytophagousinsects
SSudakaranetal
2596
possible importance of the microbial community in Malvales, with a few notable exceptions such as
adapting to novel ecological niches. Probergrothius angolensis, which feeds on seeds of
the ancient gymnosperm Welwitschia mirabilis
(Wetschnig and Depisch, 1999). Despite being
Nutritional contributions of the core microbiota phylogenetically distant, these host plants share
associated with pyrrhocorid host similar phytochemical defenses, particularly cylco-
Members of the Pyrrhocoridae are predominantly propenoic fatty acids (CPFAs) (Allen et al., 1967).
phytophagous, feeding on seeds of the plant order These compounds are known to be toxic to insects
Figure3 Cophylogeneticanalysisof(a)thedualactinobacterialsymbionts(CoriobacteriumglomeransandGordonibactersp.)and(b)
their Pyrrhocoridae hosts. The symbiont phylogeny is based on partial 16S rRNA bTEFAP sequences from OTUs picked for each
individualspecies(seeSupplementaryDataS2).ColorsforindividualsymbiontstrainscorrespondtothecoloringofhosttaxainFigure
1b.ForeachCoriobacteriaceaeOTU,therelativeabundance(inrelationtotherespectivehost’scompletemicrobialcommunity)isgivenin
bracketsbehindthestraindesignation.Host–symbiontassociationsareshownbyconnectinglines.Valuesatthenodesrepresentlocal
supportvaluesfromtheFastTreeanalysis(GTRmodel).
TheISMEJournal