Transcriptome dynamics of Arabidopsis during sequential biotic and abiotic stresses

19 Pages·2016·2.78 MB·English

ThePlantJournal(2016)86,249–267 doi:10.1111/tpj.13167
RESOURCE
Transcriptome dynamics of Arabidopsis during sequential
biotic and abiotic stresses
SilviaCoolen1,†,SilviaProietti1,†,RichardHickman1,†,NelsonH.DavilaOlivas2,†,Ping-PingHuang3,†,MarcelC.VanVerk1,4,
JohanA.VanPelt1,AlexanderH.J.Wittenberg5,MartinDeVos5,MarcelPrins5,JoopJ.A.VanLoon2,MarkG.M.Aarts3,
MarcelDicke2,Corne(cid:1)M.J.Pieterse1andSaskiaC.M.VanWees1,*
1Plant–MicrobeInteractions,DepartmentofBiology,UtrechtUniversity,POBox800.56,3508TB,Utrecht,TheNetherlands,
2LaboratoryofEntomology,WageningenUniversity,POBox16,6700AAWageningen,TheNetherlands,
3LaboratoryofGenetics,WageningenUniversity,POBox16,6700AAWageningen,TheNetherlands,
4Bioinformatics,DepartmentofBiology,UtrechtUniversity,POBox800.56,3508TB,Utrecht,TheNetherlands,and
5KeygeneN.V.,POBox216,6700AEWageningen,TheNetherlands
Received17December2015;revised7March2016;accepted8March2016;publishedonline15March2016.
*Forcorrespondence([email protected]).
†Theseauthorscontributedequallytothiswork.
SUMMARY
Innature,plantshavetocopewithawiderangeofstressconditionsthatoftenoccursimultaneouslyorin
sequence. To investigate how plants cope with multi-stress conditions, we analyzed the dynamics of
whole-transcriptome profilesofArabidopsisthalianaexposedtosixsequentialdoublestressesinflictedby
combinationsof:(i)infectionbythenecrotrophicfungusBotrytiscinerea,(ii)herbivorybychewinglarvaeof
Pierisrapae,and(iii)droughtstress.Eachofthesestressesinducedspecificexpressionprofilesovertime,in
which one-third of all differentially expressed genes was shared by at least two single stresses. Of these,
394genesweredifferentiallyexpressedduringallthreestressconditions,albeitofteninoppositedirections.
Whentwostresseswereappliedinsequence,plantsdisplayedtranscriptomeprofilesthatwereverysimilar
tothesecondstress,irrespectiveofthenatureofthefirststress.Nevertheless,significantfirst-stresssigna-
tures could be identified in the sequential stress profiles. Bioinformatic analysis of the dynamics of co-
expressedgeneclustershighlightedspecificclustersandbiologicalprocessesofwhichthetimingofactiva-
tion or repression was altered by a prior stress. The first-stress signatures in second stress transcriptional
profileswereremarkablyoftenrelatedtoresponsestophytohormones,strengtheningthenotionthathor-
mones are global modulators of interactions between different types of stress. Because prior stresses can
affecttheleveloftoleranceagainstasubsequentstress(e.g.priorherbivorystronglyaffectedresistanceto
B.cinerea),thefirst-stresssignaturescanprovideimportantleadsfortheidentificationofmolecularplayers
thataredecisiveintheinteractionsbetweenstressresponsepathways.
Keywords: combinatorial plant stress, transcript profiling, Botrytis cinerea, Pieris rapae, drought stress,
generegulatorynetwork,planthormones,RNA-Seq,Arabidopsisthaliana.
INTRODUCTION
Plants are continuously threatened by a wide range of phytohormonesjasmonicacid(JA),ethylene(ET),abscisic
harmful microbial pathogens and insect herbivores. acid (ABA), and salicylic acid (SA), and to a lesser extent
Besides these biotic stresses, plants are also exposed to with cytokinin, brassinosteroids and auxin (Robert-
extremeabioticenvironmentalconditionssuchasdrought, Seilaniantz etal., 2011; Pieterse etal., 2012; Giron etal.,
heat,cold,waterlogging,highsalinityortoxicity.Adaptive 2013; O’Brien and Benkova(cid:1), 2013; Kazan and Lyons, 2014;
plant responses to single biotic and abiotic stresses have Broekgaardenetal.,2015).JAandETaregenerallyinvolved
been extensively studied. Both biotic and abiotic stress in defense against pathogens with a necrotrophic life-
responses are associated with the action of the style, whereas defenses against biotrophs are commonly
©2016TheAuthors 249
ThePlantJournal©2016JohnWiley&SonsLtd
250 SilviaCoolenet al.
controlledbySA(Glazebrook,2005).ABAisassociatedwith thesedifferencesmaybemuchsmalleroraretheresultof
plantdevelopment andabioticstresses(Yamaguchi-Shino- ashiftinthephasingoftheexpressionprofiles.
zaki and Shinozaki, 2006), such as drought, but its role in In order to gain detailed insight into how plants cope
modulating JA-dependent defenses against insect herbi- with multiple stresses, we here investigated how a first
vores and SA-dependent defenses against pathogens is stressinfluencesthenatureanddynamicsofthetranscrip-
becomingincreasinglyevident(Yasudaetal.,2008;Verhage tional response that is induced by a second stress. We
etal.,2011;Vosetal.,2013b).Antagonisticandsynergistic chose to study the response of the model plant species
interactions between hormonal signal-transduction path- Arabidopsis to two biotic stresses (infection by the necro-
ways are thought to provide the plant with a regulatory trophic fungus Botrytis cinerea and herbivory by larvae of
potentialtoadapttoitscomplexbioticandabioticenviron- Pieris rapae) and to one abiotic stress (drought stress by
mentwhileutilizingitsresourcesinacost-efficientmanner water withhold). These stresses were chosen because in
(ReymondandFarmer,1998;Robert-Seilaniantzetal.,2011; previous studies it was demonstrated that the plant hor-
Pieterseetal.,2012;Vosetal.,2013a,2015). mones JA, ABA, and/or ET are involved in adaptive plant
In natural and agricultural settings, plants often have to responses to these respective stresses. We hypothesized
copewithmultiplestressconditionssimultaneously.Inthe thatcombiningthesestressesmayleadtohormonalsignal
context of climate change, it is highly likely that the fre- interactions that potentially affect the outcome of the
quency and complexity of these multi-stress conditions response to the second stress. Several previous studies
will increaseandfurther threatencrop yield.Abiotic stres- have identified thousands of Arabidopsis genes that
ses can significantly affect plant responses to biotic stres- change in expression in response to the selected single
ses and vice versa, depending on the timing, nature, and stresses (Reymond etal., 2000, 2004; De Vos etal., 2005;
severity of the stresses (Atkinson and Urwin, 2012; Appel Ferrari etal., 2007; Huang etal., 2008; Rowe etal., 2010;
etal., 2014; Rejeb etal., 2014). How plants regulate and Birkenbihl etal., 2012; Windram etal., 2012; Rehrig et al.,
prioritize their adaptive response when exposed to multi- 2014; Clauw etal., 2015), but their dynamic behavior dur-
ple stresses is largely unknown. Several studies have ingmulti-stressconditionsislargelyunknown.
investigated plant responses to different stress factors Botrytis cinerea is considered to be the second most
occurringsimultaneouslyorsequentially(MohrandCahill, importantplantpathogen(Deanetal.,2012),infectingover
2003;DeVosetal.,2006;VanOostenetal.,2008;Atkinson 200 cultivated plant species and causing significant eco-
etal., 2013; Prasch and Sonnewald, 2013; Rasmussen nomic damage to crops worldwide. Moreover, B. cinerea
etal.,2013;Santinoetal.,2013;Kissoudisetal.,2014;Riv- has become an important model for studying interactions
ero etal., 2014; Sewelam etal., 2014; Stam etal., 2014; between plants and necrotrophic pathogens (Van Kan,
Suzuki etal., 2014; Ramegowda and Senthil-Kumar, 2015; 2006; Laluk and Mengiste, 2010). As a necrotroph, B.
Sham etal., 2015). From these studies, the picture cinereakillsplanttissuepriortofeedingbyusingdifferent
emergedthatdifferentstresssignalingpathwaysareinter- mechanismsthat causeplantdecay,e.g. enzymaticdegra-
connectedinanetworkthatisundercontrolofkeyregula- dationofthecellwalls,generationoftoxicreactiveoxygen
tors such as MAP kinases, transcription factors and the compounds, or secretion of host non-selective toxins. JA
above-mentioned stress-related hormones (Fujita etal., and ET participate in the defense response of Arabidopsis
2006; Pieterse etal., 2009; Robert-Seilaniantz etal., 2011; against B. cinerea (Thomma etal., 1998, 1999; Diaz etal.,
Rejeb etal., 2014; Caarls etal., 2015). In order to gain 2002;Geraatsetal.,2002;Roweetal.,2010;ElOirdietal.,
insightinthecomplexityoftheplantresponsetocombina- 2011), while ABA and SA can have a negative effect on B.
torialstresses,severalrecentstudiesinvestigatedchanges cinerearesistance(ElOirdietal.,2011;Liuetal.,2015;Vos
in the transcriptome of Arabidopsis thaliana (hereafter etal.,2015).
called Arabidopsis) in response to simultaneous exposure Insect herbivores consume over 15% of the plant bio-
to abiotic and bioticstresses (Atkinson etal., 2013; Prasch massproducedannuallyintemperateandtropicalecosys-
and Sonnewald, 2013; Rasmussen etal., 2013; Suzuki tems making insect herbivory a major conduit by which
etal., 2014; Ramegowda and Senthil-Kumar, 2015; Sham energy flows through food webs (Cyr and Pace, 1993;
etal.,2015).Generally,theresponsestothesinglestresses Agrawal, 2011; Johnson, 2011). The Small Cabbage White
weredifferentfromthosetothedoublestresses.However, butterfly P. rapae is one of the most destructive pests of
these studies often focused on a single time point, repre- cruciferous plants because it has adapted to the glycoside
sentingonlyasnapshotofthetranscriptionalchangesthat toxins known as glucosinolates that are produced by cru-
are induced by a single or combinatorial stress. The influ- cifers as chemical defenses (Hopkins etal., 2009). Ara-
ence of one stress on the other may primarily have an bidopsis and other plants activate additional defense
effect on the timing of the response to the second stress, responses that reduce the performance of leaf-chewing P.
causingthedetectionoflargetranscriptionaldifferencesin rapae caterpillars on pre-infested plants (De Vos et al.,
combinatorial stresses at one time point, while over time 2006). It has been shown that this herbivore- or wound-
©2016TheAuthors
ThePlantJournal©2016JohnWiley&SonsLtd,ThePlantJournal,(2016),86,249–267
Transcriptomedynamicsduringsequentialstresses 251
induced resistance also extends systemically to undam-
aged plant parts (Howe and Jander, 2008; Vos etal.,
2013b). JA is an important primary signal in herbivore-
induced local and systemic defenses in various plant–her-
bivoreinteractions,whileABAhasamodulatingroleinthe
JA responsiveness (Bodenhausen and Reymond, 2007;
Howe and Jander, 2008; Soler etal., 2013; Vos etal.,
2013b).SAisreportedtoinhibittheJA-dependentdefense
pathway that is induced by P. rapae feeding (Koornneef
etal.,2008).
Drought is one of the most frequently experienced abi-
otic environmental stresses in plants. Low water availabil-
ity in the rhizosphere leads to a reduction in leaf stomatal
conductance and growth (Schachtman and Goodger,
2008). Adaptive responses to drought also involve meta-
bolic, osmotic, and structural adjustment, as well as the
production of proteins with DNA damage control and
repairfunctions(IngramandBartels,1996).ABAaccumula-
tion is essential for the adaptation to drought, but also
ABA-independent regulatory systems are involved in
drought stress-responsive gene expression. In the latter, Figure1. Experimental schedule of treatments and harvests for RNA-Seq
JA and ET have been implicated as important regulators timeseriesofsingleandsequentialdoublestresses.
Thescheduleshowsthetimingoftreatmentsandtimepointsofharvestfor
(Bray,1997;Shinozakietal.,2003;Rieraetal.,2005;Huang
the three main treatments, B. cinerea (Bc, red), P. rapae (Pr, green) and
etal.,2008). drought (Dr, yellow), and the respective pre-treatments. Each single and
Inthisstudy,weusedRNA-Seqtoanalyzethedynamics sequential double stress sample had a mock/control (not visualized) that
washarvestedatthesametimepointasthestresstreatment.Mock-treated
ofthetranscriptomechangesthatoccurredinArabidopsis
plants were cultivated under the same conditions as their respective B.
overfourtimepointsinresponsetoB.cinereainfection,P. cinerea-treatedplants(sameperiodsof100%RH).Untreatedcontrolplants
rapae feeding, drought stress, and all six combinations of werecultivatedunderthesameconditionsastheirrespectiveP.rapae-and/
ordrought-treatedplants.FirststresseswerestoppedbyeitherloweringRH
sequentialdoublestresses.Ourresultsshowthatirrespec-
from100to70%(after1dayincaseofBcpre-treatment),removingcaterpil-
tive of the first stress, Arabidopsis is capable of swiftly larsfromplants(after1day),orre-wateringaftera7-dayperiodofdrought
adaptingitstranscriptometorespondtothesecondstress. (7+1;blue).Incasethesecondstresswasdrought,thepre-treatmentswith
B. cinerea and P. rapae were performed right after the last moment of
Over time, this second stress-induced transcriptome is
watering.100%;periodof100%RHinsteadofstandard70%RH;timeindica-
highly similar to that of plants that did not receive a first tionsatthebottomindicatetimeofthedayatwhichplantswereharvested.
stress,butcontainsclearfirst-stresssignatures,whichmay
playaroleinthephenotypicinteractionbetweenconsecu-
tivestresses. earliest transcriptional changes can be observed around
6h after application of the inoculum, while at 24h after
RESULTS inoculation massive changes in gene expression can be
detected (Windram etal., 2012; Vos etal., 2015). For the
ExperimentalapproachforRNA-Seqanalysisofsingleand
study of P. rapae stress, we chose a time span between 3
sequentialstresstimeseries
and 24h after infestation by larvae of stage L1 because
Inordertocaptureamaximaldynamicrangeofthestress previous studies demonstrated that this would yield a
responses,theresponsetoeachofthethreemainstresses maximaldynamicrangeof transcriptional responses(Rey-
was monitored in a different time frame of four time mondetal.,2000,2004;DeVosetal.,2005;Verhageet al.,
points, depending on how quickly the stress response 2011).Fortheinductionofdroughtstress,4-week-oldAra-
developed(Figure1). Thetranscriptionalresponseto each bidopsis plants that had previously been watered with
single and sequential stress was compared at each time equal amounts of water were subsequently withheld from
pointtoanon-treatedcontrol(fortreatmentsnotinvolving water for 7days. At day 5 of water withhold, drought-
B. cinerea) or a mock-treated control (same 100% relative stressed plants were clearly smaller and darker colored
humidityconditionsasB.cinereatreatments)thatwashar- than the watered control plants, a phenotype that pro-
vested at the same time as the stress treatment. For the gressedfurtheronday6and7whentheywereatthepoint
studyofB.cinereastress,atimespanbetween6and24h ofwilting.Thetranscriptometimeserieswerechosenat5,
afterinoculationwitha5-lldropletof59105sporesml(cid:1)1 6and7daysafterwaterwithhold,andatday8(7 +1day),
was chosen, because previous studies showed that the which was 1day after re-watering. The recovery response
©2016TheAuthors
ThePlantJournal©2016JohnWiley&SonsLtd,ThePlantJournal,(2016),86,249–267
252 SilviaCoolenet al.
at day 8 was chosen as the fourth time point of the time series was selected according to their significance in
drought time series because this recovery response after fold-change expression (false discovery rate (FDR) <0.05)
droughtstressisinterestingbyitself,andatthistimepoint and an additional threshold level of at least two-fold
the sequential treatment with B. cinerea and P. rapae was change ((cid:1)1> log > 1) in comparison to the respective
2
executedandthuscouldfunctionasareferencetreatment. control (Table S1). The first observation that can be made
Priortoapplyingthesecondstress,furtherdevelopmentof from the RNA-Seq results is that over time there are clear
thefirststresswasstoppedbychangingthe100%relative differences in the number of genes that are significantly
humidity condition to 70% (first stress B. cinerea), by activated or repressed during the different single stress
removing the caterpillar (first stress P. rapae), or by re- conditions (Figure2). For responses to B. cinerea (total
watering the plants (first stress drought). Developmental 2076uniqueDEGs)andP.rapae(total3952uniqueDEGs),
leafnumber8wasusedforapplyingB.cinereaorP.rapae a strong increase in the number of activated genes is
as second stress. For all treatments, leaf number 8 was observed over time, while relatively few genes are
harvested for RNA-Seq analysis. When leaf number 8 was repressed. Upon exposure to drought stress (total 4032
not damaged by P. rapae, the next-closest P. rapae-dam- uniqueDEGs for thefirst three time points, plus 482addi-
aged leaf was harvested. Three biological replicates per tional unique DEGs for the 1day after re-watering time
treatmentandtimepointweresubjectedtoRNA-Seq.Each point), relatively more genes become repressed than acti-
ofthethreebiologicalreplicatesconsistedoffour‘number vated.Apriorstressdidnotdramaticallychangethenum-
8’ leaves that were pooled to form one sample. After har- ber of DEGs relative to the single stresses (Figure2).
vest,leaveswereprocessedandsubjectedtoRNA-Illumina ClusteringtheunionofDEGsofthesinglestresssets(total
sequencing. On average, 14.6million reads (range 8.5– 7355 unique DEGs), and subsequent gene ontology (GO)
29.8million) were generated per sample with >90% of analysis (Boyle etal., 2004) of overrepresented biological
sequences aligning to the Arabidopsis genome after qual- processesineachclusterhighlightsthedifferentiallyregu-
ityfiltering(VanVerketal.,2013). latedbiologicalprocessesduringtheplantresponsetothe
single stresses and uncovers similarities and contrasts
Timeseriestranscriptomeprofilingfollowingsingleand
between the different stress responses (Figure3 and
sequentialstresses
TableS2).
In this study, our aim was to analyze the dynamic tran-
CoreDEGsthataresharedbetweenthesinglestress
scriptomechangesthataretriggeredbythesinglestresses
responses
and investigate how the nature and dynamics of these
transcriptome profiles were affected by pre-exposure to To investigate to what extent genes and biological pro-
each of the other two stresses. First, a set of differentially cesses are shared between the three single stress
expressed genes (DEGs) derived from each single stress responses, we compared their DEGs. Figure4(a) shows
Figure2. NumbersofDEGsatdifferenttimepointsinsingleandsequentialstressresponses.
Graphsshowthenumberofactivated(redbars)andrepressed(bluebars)differentiallyexpressedgenes(DEGs)forallsinglestressesandtheircorresponding
sequentialdoublestressesatdifferenttimepointsaftertreatment(FDR<0.05;>2-fold).Theonedayafterre-wateringtimepointofthedroughttreatmentsis
indicatedas‘7+1day’.Bc,B.cinerea;Pr,P.rapae;Dr,drought;Pr+Bc,Dr+Bc,Bc+Pr,Dr+Pr,Bc+Dr,andPr+Dr,respectivesequentialdoublestresses.
©2016TheAuthors
ThePlantJournal©2016JohnWiley&SonsLtd,ThePlantJournal,(2016),86,249–267
Transcriptomedynamicsduringsequentialstresses 253
Figure3. ClusteringofthesinglestressDEGs.
Huneiaotnmaopf dshifofewreinngtiatlhlyeeexxpprreesssseidongepnaettser(nDsEGofs)thine –3 0 3 Bc Pr days dDaysr days days
the three single stresses at different time points
after induction (total 7173 unique genes). DEGs
Response to chi(cid:2)n
were clustered using mclust yielding nine gene
clusters(coloredbarsontheleft).Ontherightside, Defense response,
themostsignificantGOtermforeachcluster(full incompa(cid:2)ble interac(cid:2)on
datasetinTableS2).Bc,B.cinerea;Pr,P.rapae;Dr,
drought. For drought stress, the time point 1day
Response to s(cid:2)mulus
after re-watering (7+1day) was included in the
cluster analysis. Blue–red color key for change in
geneexpressionlevel:(cid:1)3>log2foldchange>3. Photosynthesis
Cell wall macromolecule
metabolic process
Response to chemical s(cid:2)mulus
Systemic acquired resistance
RNA methyla(cid:2)on
Response to oxygen-containing
compound
that there is a large overlap between the DEGs of the sin- system process’) is activated by B. cinerea, but repressed
gle stress responses, ranging from 1716 genes shared by P. rapae and drought. Conversely, cluster 10 (top GO
between the drought and P. rapae sets, to 788 genes terms ‘multidimensional cell growth’, ‘response to light
between the drought and B. cinerea sets, and 777 genes stimulus’, and ‘cell wall organization’) is activated by P.
betweentheP.rapaeandB.cinereasets.Ofall7173DEGs rapae, but repressed by B. cinerea and drought. The fact
(excluding the 1day after re-watering time point), 2493 that there is an overlap in the expression of genes under
DEGs(35%)aresharedwithoneorbothoftheotherstres- all three single stresses, whether in the same or in oppo-
ses. A core set of 394 DEGs (5%) was differentially site directions, suggests that these genes or their regula-
expressedinresponsetoallthreesinglestresses,andclus- tors may act as a point of convergence if plants were to
tered into 12 co-expressed gene clusters (Figure4b). experiencethesestressesincombination.
AmongthiscoresetofsharedDEGsareseveralwell-char-
Botrytiscinereadataset:effectofherbivoryanddrought
acterized hormone-responsive marker genes, including
stressondynamicsofB.cinerea-inducedgeneexpression
LOX2 (At3g45140), JAZ7 (At2g34600), and JAZ8
(At1g30135) (JA responsive), PDF1.2 (At5g44420) and ToinvestigatetheeffectofP.rapaeinfestationanddrought
ORA59 (At1g06160) (JA/ET responsive), PR4 (At3g04720), stress on the dynamics of the transcriptome changes that
ERF5 (At5g47230), ERF6 (At4g17490), and ACS2 areinducedbyB.cinereainfection,weanalyzedtheexpres-
(At1g01480) (ET responsive), Rap2.6L (At5g13330), and sion patterns over time of all 2076 B. cinerea-responsive
HAI1(At5g59220)(ABAresponsive),andPR-1(At2g14610), DEGs.Clusteringofthisgroupofgenesyielded10clusters
PR-5 (At1g75040), and FRK1 (At2g19190) (SA responsive) of co-expressed genes across the B. cinerea single and
(Table S1 for details on their expression patterns). In Fig- sequentialstressdatasets.Geneclustersthatareactivated
ure4(c)theexpressionpatternsof wellcharacterized mar- in response to B. cinerea infection are enriched for GO
ker genes of the response to B. cinerea (PDF1.2; termssuchas‘responsetochitin’(Figure5(cluster3)and
At5g44420), P. rapae (LOX2; At3g45140), and drought Table S4), reflecting recognition of fungal chitin by the
stress (RAD18; At5g66400) are depicted, confirming that plant immune system (Pel and Pieterse, 2013), and ‘re-
the different stress treatments resulted in the expected sponsetoETstimulus’,reflectingthehighlevelofETemis-
response.Whenlookingattheco-expressedgeneclusters, sion that is related to plant responses to B. cinerea
only the genes of cluster 1 (top GO terms related to ‘re- infection(Broekgaardenetal.,2015).Inaddition,geneclus-
sponsetooxygen-containingcompound’,‘responsetoJA’ ters that are repressed in response to B. cinerea infection
and‘responsetowounding’;TableS3)areregulatedinthe are associated with GO terms such as ‘multidimensional
samedirection(activated)duringallthreeindividualstress cellgrowth’(Figure5(cluster9)andTableS4),highlighting
conditions (Figure4b). All other gene clusters behave the antagonistic relationship between plant growth and
clearly different in response to the three single stresses defense(WangandWang,2014).
and are often regulated in opposite directions (Figure4b). Interestingly, the expression patterns over time in the
For example cluster 8 (top GO terms related to ‘response sequentialdoublestresstreatmentsappearingeneralvery
to other organism’, ‘defense response’ and ‘immune similar to the ones of the B. cinerea treatment alone. This
©2016TheAuthors
ThePlantJournal©2016JohnWiley&SonsLtd,ThePlantJournal,(2016),86,249–267
254 SilviaCoolenet al.
(a) (b)
(c)
Bc Pr+Bc Dr+Bc Pr Bc+Pr Dr+Pr Dr Bc+Dr Pr+Dr
SA: PR-1, PR-5, FRK1
JA: LOX2, JAZ7, JAZ8
JA/ET: PDF1.2, ORA59
ET: PR-4, ERF6, ERF7, ACS2
ABA: Rap2.6L, HAI1
PDF1.2 (At5g44420)
LOX2 (At3g45140)
RAD18 ( At5g66400)
Figure4. SharedDEGsbetweenthesinglestressresponses.
(a)VenndiagramshowingtheoverlapbetweentheDEGsofeachofthesinglestressresponses.ThetotalnumberofuniqueDEGspersinglestressoveralltime
pointsisshowninred(fulldatasetinTableS1).
(b)Hierarchicalclusteringofthe394coreDEGsthataresharedbetweenthethreesinglestresses(Cosinesimilaritymetric;12clustersarecolorcodedinthe
squareboxesontheleft).Ontherightside,themostsignificantGOtermforthelargestclusters(fulldatasetinTableS3).
(c)Comparisonoftheexpressionpatternsofthe394coreDEGsinresponsetothesingleandrespectivesequentialdoublestresses.Differentlanesin(c)reflect
thetranscriptionprofilesatthetimepointsaftertreatmentasindicatedabovethelanesin(b).Genenamesinthelowerrightcornerrepresentmarkergenesof
theSA,JA,JA/ET,ETandABAresponsepathwaysthatareamongthe394coreDEGs.PDF1.2,LOX2,andRAD18representknownmarkergenesforthe
responsetoB.cinerea,P.rapae,anddrought,respectively.Bc,B.cinerea;Pr,P.rapae;Dr,drought;Pr+Bc,Dr+Bc,Bc+Pr,Dr+Pr,Bc+Dr,andPr+Dr,
respectivesequentialdoublestresses.Blue-redcolorkeyforchangeingeneexpressionlevel:(cid:1)3>log foldchange>3.
2
suggests that Arabidopsis swiftly reprogrammes its tran- cinereasingletreatment.Thiseffectisalsovisibleintheleft
scriptome to the response that is induced by B. cinerea panel of Figure4(c) where the expression patterns of the
infection, thereby overruling effects of the prior stresses coresetof394DEGsareplotted.AlsopriorexposuretoP.
herbivory and drought on B. cinerea-responsive gene rapae inflicted clear differences in the expression patterns
expression.For example,at themomentdrought pre-trea- oftheB.cinerea-responsivegenesduringthefirsttwotime
tedplantswereinoculatedwithB.cinerea(onedayafterre- points (compare the first two time points in the Pr+Bc
wateringofdrought-treatedplants),morethan1000genes treatmentswiththoseintheBctreatment),butatlatertime
werestilldifferentiallyexpressedintheanalyzedleaf(num- points the gene expression patterns of the core DEGs
ber 8) due to the prior drought treatment (508 DEGs acti- become very similar to that of the B. cinerea single stress
vated and 610 DEGs repressed at time point 7+1day; treatment. Nevertheless, during the sequential stress
TableS1;Figure3lastlane).At6hafterB.cinereainocula- responsessomeclustersshowafirst-stresssignature,e.g.
tionthisdrought-inducedeffectisstillclearlyvisibleinthe thegenesinclusters2,3and5inFigure5showadelayed
B.cinerea-responsivegeneset(comparethe6hlaneinBc activation when plants experienced herbivory or drought
versusDr+ BctreatmentinFigure5),butatthelatertime stresspriortoB.cinereainfection.
pointsthiseffectquicklydampensoffandtheB.cinerea-re- In order to identify in greater detail co-regulated genes
sponsivegenesstarttofollowthesamepatternasintheB. of which the expression pattern in response to B. cinerea
©2016TheAuthors
ThePlantJournal©2016JohnWiley&SonsLtd,ThePlantJournal,(2016),86,249–267
Transcriptomedynamicsduringsequentialstresses 255
Figure5. Dynamics of the expression of the B.
cinerea set of DEGs during single and sequential
doublestresses.
Heatmap showing the expression patterns of the
2076B.cinerea-responsiveDEGsduringB.cinerea
infection on mock pre-treated (Bc), P. rapae pre-
infested (Pr+Bc) or drought pre-treated (Dr+Bc)
Arabidopsisplants.TheB.cinerea-responsiveDEGs
were clustered using mclust yielding 10 clusters
(colored bars on the left). On the right side, the
mostsignificantGOtermforeachcluster(fulldata
setinTableS4).Blue–redcolorkeyforchangein
geneexpressionlevel:(cid:1)3>log2foldchange>3.
infection was affected by either herbivory or drought whichtimepointaGOtermbecomessignificantlyoverrep-
stress,weusedthebioinformaticstoolWigwams(Polanski resented in the B. cinerea-related DEG set. For this, we
etal.,2014).TheWigwamsalgorithmidentifiesgenemod- clusteredallB.cinerea-responsiveDEGsaccordingtotheir
ules showing evidence for co-regulation in multiple gene time point of first differential expression, divided them
expression time series and identifies signatures of condi- over activated and repressed genes, and performed GO
tion-dependent regulatory mechanisms in co-regulated term analysis on them. Figure7 shows the timing and
genesets.Wigwamsidentified35modulesofco-regulated strengthoftheonsetofsignificantGOtermenrichmentin
genes in the B. cinerea data sets (Figure S1). Analysis of the single and sequential double stress conditions. In the
theseclustersforco-expressionrevealedgenemodulesof B. cinerea single stress data set, GO terms related to
which the expression patters were clearly affected in one responsestoET,fungus,chitin,SA,andoxygen-containing
or both of the sequential stress treatments in comparison compound, or to processes such as systemic acquired
to the B. cinerea treatment alone (examples shown in resistance, respiratory burst, and defense appear early in
Figure6). These gene modules represent signatures of a the activated gene set, reflecting the importance of these
previous stress in the B. cinerea-induced transcriptome processesintheplantresponsetothisnecrotrophicpatho-
profile, and may thus be functionally related to the effect gen. Prior infestation with P. rapae clearly delayed the
of thefirststressonthe outcomeof theplantresponseto appearanceoftheseGOterms(becomevisibleat18haiin
B.cinereainfection.ThegenesintheseWigwamsmodules Figure7), while pre-treatment with drought stress did not
aregiveninTableS5alongwiththeirGOtermanalysis. have a dramatic effect on the phasing of the activated
Among the B. cinerea-responsive Wigwams modules of genes. For the repressed genes in the B. cinerea set of
which the co-expression pattern is different when plants DEGs, pre-infestation with P. rapae has clearly only minor
were previously exposed to herbivory or drought stress, effectsontheGOtermphasing.Bycontrast,pre-treatment
are gene modules with GO term enrichments for rather with drought stress noticeably affected the phasing of GO
general plant processes such as nucleoside biosynthesis terms relatedto responsesto fungus,JA, SA, ABA,chitin,
andmetabolism(modules6and9),andcellgrowth(mod- andoxygen-containingcompound,andtoauxinmetabolic
ule 25), but also modules related to more specific plant process, defense, systemic acquired resistance, and glu-
processes, such as response to chitin and nitrogen (mod- cosinolate biosynthetic process. Remarkably, biological
ule 15). Functional analysis of underlying candidate genes processes related to hormone action prevail in the B.
should reveal their importance for effects on the outcome cinerea-responsiveprocessesthataresensitivetomodula-
ofthesecondstressresponse. tionbypriorexposuretooneoftheotherstresses.
Further knowledge about the biological processes that
EffectofherbivoryordroughtstressonresistancetoB.
areaffectedwhenB.cinereainfectionisprecededbyeither
cinerea
droughtstressorherbivorycanbegainedbyanalyzingthe
phasing of gene expression under the different single and Both herbivory and drought stress imposed a first-stress-
sequential stress conditions. We did this by analyzing at signature in the dynamics of the B. cinerea-induced
©2016TheAuthors
ThePlantJournal©2016JohnWiley&SonsLtd,ThePlantJournal,(2016),86,249–267
256 SilviaCoolenet al.
Bc Pr+Bc Dr+Bc Bcmodule # and GO term top 5
1.5 1.5 1.5
• Pyrimidine ribonucleotide metabolic process
1 1 1
(cid:129) Pyrimidine ribonucleo(cid:2)de biosynthe(cid:2)c process
Log2 counts–00..505 Log2 counts–00..505 –00..505 #6 (cid:129)(cid:129)(cid:129) NRNiuubccolleenoou(cid:2)scilddeeeo (cid:2)bpdihoeos sybpniohthsayetne(cid:2)t cbh ipeo(cid:2)rsocyc npetrshosec(cid:2)ecs sprocess
–1 –1 –1
–1.51 2 3 4 –1.51 2 3 4 –1.51 2 3 4
1.5 1.5 1.5 (cid:129) RNA/macromolecule methyla(cid:2)on
1 1 1 (cid:129) Methyla(cid:2)on
Log2 counts–00..505 Log2 counts–00..505 Log2 counts–00..505 #9 (cid:129)(cid:129)(cid:129) NRNNiutcArol emgoeobnda cisfioec-maco(cid:2)pnootnuanindi nmge ctoambopliocu pnrdo cmeesstabolic
–1 –1 –1 process
–1.51 2 3 4 –1.51 2 3 4 –1.51 2 3 4
1.5 1.5 1.5 (cid:129) Response to chi(cid:2)n
1 1 1 (cid:129) Response to organic nitrogen
Log2 counts–00..505 Log2 counts–00..505 Log2 counts–00..505 #15 (cid:129)(cid:129)(cid:129) RRReeessspppooonnnssseee tttooo nomirtegrcaohngaiecnn si cucaobl msst(cid:2)apmnocuuenluds
–1 –1 –1
–1.51 2 3 4 –1.51 2 3 4 –1.51 2 3 4
1.5 1.5 1.5
1 1 1 (cid:129) Mul(cid:2)dimensional cell growth
Log2 counts–00..505 Log2 counts–00..505 Log2 counts–00..550 #25 (cid:129)(cid:129)(cid:129)(cid:129) PURPalena(cid:3)ginduetil rmatny(cid:2)e psonpens ec iooceifnlfi lac cwela lca(cid:2)l elosll ilnoz ge rprgoraowncitezhass(cid:2)on
–1 –1 –1
–1.51 2 3 4 –1.51 2 3 4 –1.51 2 3 4
Time points Time points Time points
Figure6. ExpressionpatternsofselectedWigwamsmodulesfromtheB.cinereasetofDEGsduringsingleandsequentialstressconditions.
AselectionofWigwamsmodulesofco-expressedgeneclustersisdepictedthatshowadifferentpatterninoneorbothofthesequentialstressesP.rapae-B.
cinerea(Pr+Bc)anddrought-B.cinerea(Dr+Bc)incomparisontothesinglestressB.cinerea(Bc).Themodulesrepresentstandardizedpatternsofdifferential
geneexpressionovertime(log2counts).Blue-coloredgraphsindicatemodulesofwhichthegenesaresignificantlyco-expressedovertimeinthegivenstress
condition.Timepoints1,2,3and4represent6,12,18and24hafterB.cinereainoculation.ThetopfiveGOtermswithhighestsignificanceintherespective
modulesaregiven(fulldatasetforallWigwamsmodulesispresentedinTableS5).
transcriptome profiles. Wigwams analysis gained insight Pierisrapaedataset:effectofB.cinereainfectionand
into the identity of candidate genes related to these first- droughtstressondynamicsofP.rapae-inducedgene
stresssignatures(FigureS1andTableS5),whereasanaly- expression
sisofGOtermenrichmentprovidedglobalinsightintothe
The same approach as described above for the B. cinerea
biological processes that were affected by the stress pre-
transcriptome data was taken to investigate the effect of
treatments(Figure7).Toinvestigatewhetherthetwoprior
prior B. cinerea infection and drought stress on the tran-
stressesaffectedtheresistanceleveltoB.cinereainfection
scriptionaldynamicsthat areinducedby P.rapae feeding.
we performeddisease resistance bioassays. Inoculationof
Analysis of the global expression profiles of the 3952 P.
5-week-old Arabidopsis Col-0 plants with B. cinerea
rapae-responsive DEGs yielded nine clusters of co-
resulted in the development of spreading lesions in about
expressed genes during single and sequential P. rapae
60% of the inoculated leaves (Figure8). Plants that were
stress (Figure9). As expected, P. rapae feeding induced
exposed to drought stress prior to B. cinerea inoculation
manygenesrelatedtotheGOterm‘responsetoJAstimu-
showed a similar percentage of leaves with spreading
lus’ (Figure9 (cluster 7) and Table S6), reflecting induced
lesions(approximately70%).Interestingly,plantsthatwere
defenses that are triggered by herbivory-inflicted wound-
exposed to herbivory prior to B. cinerea inoculation
ing (Wasternack, 2015). In addition, P. rapae feeding
showedasignificantlyenhancedlevelofresistanceagainst
repressedSA-relatedgenesassociatedwithGOterms‘de-
B.cinereainfection(averageapproximately35%spreading
fense response’ and ‘systemic acquired resistance’ (Fig-
lesions). Together these results indicate that a first stress
ure9 (cluster 3) and Table S6), reflecting the antagonistic
can have strong effects on the outcome of the adaptive
relationshipbetweenJA-andSA-dependentdefenses(Pie-
stress response to a second stress, depending on the nat-
terseetal.,2012).Inanalogywithwhatweobservedinthe
ureofthefirststress.
B. cinerea data sets, the general gene expression patterns
©2016TheAuthors
ThePlantJournal©2016JohnWiley&SonsLtd,ThePlantJournal,(2016),86,249–267
Transcriptomedynamicsduringsequentialstresses 257
Figure7. Timing of GO term overrepresentation
patternsinB.cinereasingleandsequentialstress
datasets.
HeatmaprepresentsthestrengthsofthePvaluesof
GO term overrepresentation in the B. cinerea-re-
sponsiveDEGsets(correspondingtothegenelists
inTableS1)thatbecomesignificantlyactivated(up)
orrepressed(down)forthefirsttimeatthegiven
stressconditionsandtimepoints.Colorindexrep-
resentslevelofsignificance(Pvalues).Ontheright,
overrepresented GO terms. Bc, B. cinerea; Pr, P.
rapae;Dr,drought;hai,hafterB.cinereainfection.
Figure8. EffectofherbivoryanddroughtstressonresistanceofArabidopsistoB.cinerea.
(a)QuantificationofB.cinereadiseasesymptomsonArabidopsisaccessionCol-0plants(%spreadinglesionsperplant).Oneachplant,sixleaveswereinocu-
latedwithonedropletofB.cinereaspores.Threedayslater,theaveragenumberofleaveswithspreadinglesionswasdeterminedperplant.Asteriskindicates
statisticallysignificantdifferencefromsinglestress(Bc)treatment(n=6plants;Student’st-test;P<0.05).
(b)PhotographsofB.cinereadiseasesymptoms3daysafterinoculation.Bc,B.cinerea-inoculatedplants;Pr+Bc,B.cinerea-inoculatedplantsthatpriortoinoc-
ulationwereexposedtoherbivorybyP.rapaelarvaefor24h;Dr+Bc,B.cinerea-inoculatedplantsthatpriortoinoculationreceivedadroughttreatmentfor
7days,followedbyare-wateringphaseofoneday.Reddots,B.cinerea-inoculatedleaves;purplearrow,damagecausedbyP.rapaefeeding.
over time overlapped greatly between the responses to P. pre-treatment. In general, these P. rapae-related results
rapae single and sequential double stress treatments, confirmpreviousfindings(DavilaOlivasetal.,2016).
againsuggestingthatArabidopsisiscapableofreprogram- To pinpoint co-regulated genes whose expression pat-
ming its transcriptome to the last stress encountered, terninresponsetoherbivoryisaffectedbypriorB.cinerea
thereby overruling the effects of the prior stresses. For infection or drought stress, the set of P. rapae-responsive
instance, while prior drought stress impacted the expres- DEGs was analyzed with the Wigwams algorithm. Wig-
sion of over 1000 genes in the leaf tissue just before the wams identified 93 modules of co-regulated genes in the
start of the P. rapae treatment (Table S1 and Figure3 last P. rapae set of DEGs. Analysis of these clusters for co-
lane),alreadyfromthefirsttimepoint(3h)afterherbivory expression under the single and sequential double stress
this effect was mostly vanished in the P. rapae-induced conditionsrevealedgenemodulesofwhichtheexpression
profiles, which readily followed a similar expression pat- patterns were clearly affected by one or both of the
ternasintheP.rapaesingletreatment(Figure9).Asimilar sequential double stress treatments in comparison to the
pattern is visible in the core set of 394 DEGs (Figure4c, P. rapae treatment alone (examples shown in Figure10;
middlepanel).Nevertheless,duringthesequentialstresses full set in Figure S2). The identities of the genes in the P.
first-stresssignaturescanbedetected,e.g.genesincluster rapae-related Wigwams gene modules are given in
3 of Figure9 and clusters 7 and 8 of Figure4(c) (middle Table S5 along with their GO term analysis. It is beyond
panel) show a weaker repression in the B. cinerea thescopeofthispapertodiscusstheidentityofthegenes
pre-treatment and a stronger repression in the drought in detail. However, among the P. rapae-responsive
©2016TheAuthors
ThePlantJournal©2016JohnWiley&SonsLtd,ThePlantJournal,(2016),86,249–267
258 SilviaCoolenet al.
Figure9. Dynamics of the expression of the P.
rapae set of DEGs during single and sequential
doublestresses.
Heatmap showing the expression patterns over
timeofthe3952P.rapae-responsiveDEGsduring
feedingofP.rapaeoncontrol(Pr),B.cinereapre-
infected (Bc+Pr), or drought pre-treated (Dr+Pi)
Arabidopsis plants. The P. rapae-responsive DEGs
wereclusteredusingmclustyieldingnineclusters
(colored bars on the left). On the right side, the
mostsignificantGOtermforeachcluster(fulldata
setinTableS6).Blue–redcolorkeyforchangein
geneexpressionlevel:(cid:1)3>log2foldchange>3.
Wigwams modules of which the co-expression pattern is different stresses interact via the hormone-regulated sig-
clearly different when plants were pre-exposed to nalingnetwork.
B.cinereainfectionordroughtstress,aregenemoduleswith
Droughtdataset:effectofB.cinereainfectionand
GOtermenrichmentsforxylem,phloemandtissuedevelop-
herbivoryondynamicsofdroughtstress-inducedgene
ment(module3),cellwallbiosynthesisprocesses(module
expression
45),andresponsestoJAandwounding(module69).
Also for the P. rapae data set,phasing of theregulation Also for drought stress we investigated the effect of the
of biological processes that are associated with herbivory other two stresses on the dynamics of the transcriptome
and influenced by prior B. cinerea infection or drought changes that are induced by this abiotic stress. We ana-
stresswasdetermined.Therefore, thelevel ofsignificance lyzedthedynamicsoftheglobalexpressionpatternsofthe
of overrepresentation of theGOterms forthe P.rapae set 4032 drought-responsive DEGs during single and sequen-
of DEGs at their first time of differential expression was tialstresswithdroughtasthesecondstress,whichyielded
assessed. In the P. rapae single stress data set, GO terms 10 clusters of co-expressed genes (Figure12). GO term
related to responses to chitin, wounding, JA, ET, SA, analysis of overrepresented biological processes in each
auxin,ABA,waterdeprivation,osmoticstress,andoxygen- cluster highlights the main differentially regulated biologi-
containing compound are already highly enriched at 3h cal processes. As expected, drought stress induced a rela-
after infestation in the activated set of DEGs (Figure11), tivelylargenumberofgenesrelatedtoGOterm‘response
reflecting the importance of these processes in the to water deprivation’ (Figure12 (cluster 6) and Table S7)
response of Arabidopsis to herbivory. In many cases, pre- and GO terms related to ‘response to oxygen-containing
infectionwithB.cinereastrengthenedthelevelofoverrep- compound’ (clusters 3 and 10, and cell wall-related pro-
resentationoftheseGOtermsatdifferenttimepointsafter cesses (cluster 7). Another feature that stands out is the
P. rapae infestation, while pre-treatment with drought association of drought stress with massive repression of
stressoftenweakenedthem.IntherepressedsetofDEGs, genes, many of which are associated with biological pro-
drought stress clearly enhanced the overrepresentation of cesses such as ‘photosynthesis’ and ‘defense response’
GO terms related to responses to chitin, JA, fungus, ABA, (clusters 1, 2, 5 and 8), reflecting the fact that drought-
SA, bacterium, and oxygen-containing compound, and to stressedplantsshifttheirstrategyfromenergy-demanding
processes such as systemic acquired resistance, defense processesrelatedtogrowthandimmunitytoadaptationto
response to fungus, and negative regulation of pro- theabioticstresscondition.Interestingly,after1dayofre-
grammed cell death, while B. cinerea infection had no watering (7+ 1day columns in Figure12), the drought-
major effect on the phasing of these GO terms. Overall, inducedtranscriptionalchangesthatintensifiedoverthe7-
these data indicate that B. cinerea infection and drought day period of water withhold, were for 77% (3106 of the
treatmentpriortoP.rapaeinfestationaffectsthetimingof 4032 DEGs; Table S1) reset towards basal levels within
several defense-related processes, in particular responses 24h, demonstrating the plant’s ability to swiftly redirect
to JA, ABA, SA and ET, corroborating the notion that transcriptional programming when drought stress is
relieved. Similar to what we observed for the B. cinerea
©2016TheAuthors
ThePlantJournal©2016JohnWiley&SonsLtd,ThePlantJournal,(2016),86,249–267