How Environmental Factors Affect the Production of Guanidine Alkaloids by the Mediterranean ...

15 Pages·2017·2.72 MB·English

marine drugs
Article
How Environmental Factors Affect the Production of
Guanidine Alkaloids by the Mediterranean Sponge
Crambe crambe
EvaTernon1,*,EricaPerino2,RenataManconi3,RobertoPronzato2andOlivierP.Thomas1,4,*
1 UniversitéCôted’Azur,CNRS,OCA,IRD,Géoazur,250rueAlbertEinstein,06560Valbonne,France
2 DipartimentodiScienzedellaTerra,dell’AmbienteedellaVita,UniversitàdiGenova,CorsoEuropa26,
16132Genoa,Italy;[email protected](E.P.);[email protected](R.P.)
3 DipartimentodiScienzedellaNaturaedelTerritorio,UniversitàdiSassari,ViaMuroni25,07100Sassari,
Italy;[email protected]
4 MarineBiodiscovery,SchoolofChemistry,NationalUniversityofIrelandGalway,UniversityRoad,
H91TK33Galway,Ireland
* Correspondence:[email protected](E.T.);[email protected](O.P.T.);
Tel.:[phone](O.P.T.)
AcademicEditor:PaulLong
Received:6December2016;Accepted:12June2017;Published:16June2017
Abstract: Most marine sponges are known to produce a large array of low molecular-weight
metaboliteswhichhaveapplicationsinthepharmaceuticalindustry. Theproductionofso-called
specializedmetabolitesmaybecloselyrelatedtoenvironmentalfactors. Inthiscontext,assessing
the contribution of factors like temperature, nutrients or light to the metabolomes of sponges
providesrelevantinsightsintotheirchemicalecologyaswellasthesupplyissueofnaturalsponge
products. ThespongeCrambecrambewaschosenasamodelduetoitshighcontentofspecialized
metabolites belonging to polycyclic guanidine alkaloids (PGA). First results were obtained with
field data of both wild and farmed specimens collected in two seasons and geographic areas of
the North-Western Mediterranean. Then, further insights into factors responsible for changes in
themetabolismweregainedwithspongescultivatedundercontrolledconditionsinanaquarium.
Comparative metabolomics showed a clear influence of the seasons and to a lesser extent of the
geographywhilenoeffectofdepthorfarmingwasobserved. Interestingly,spongefarmingdidnot
limittheproductionofPGA,whileexsituexperimentsdidnotshowsignificanteffectsofseveral
abioticfactorsonthespecializedmetabolomeataone-monthtimescale. Somehypotheseswere
finallyproposedtoexplaintheverylimitedvariationsofPGAinC.crambeplacedunderdifferent
environmentalconditions.
Keywords: sponge;abioticfactors;specializedmetabolome;metabolomics;spongefarming
1. Introduction
Marinespongesareknowntoproducealargearrayofsmallmoleculesalsocalledspecialized
metabolitesthatarelikelytobeinvolvedinallelopathicinteractions[1–6]. Fromabiotechnological
pointofview,suchmetabolitesarefrequentlyinvestigated,mostlyforpharmaceuticalapplications[7–9].
Several biotechnological approaches have been developed for the large-scale production of these
high value-added marine products which include in situ and ex situ farming, primmorph or cell
culture[10,11]. Farmingexplantsinsituhasbeenconsideredasoneofthemostcost-effectivewaysto
guaranteesufficientsustainablebioactivemetabolites[12]. Sincespatialandtemporalvariationsin
concentrationsofmajorspecializedmetaboliteshavebeendocumentedpreviously[13–15],theideal
farminglocationthatpromotestheirbiosynthesisaswellastheharvestseasonhavetobeassessed.
Mar.Drugs2017,15,181;doi:10.3390/md15060181 www.mdpi.com/journal/marinedrugs
Mar.Drugs2017,15,181 2of15
Thenaturalvariationsintheproductionofthespecializedmetaboliteshavebeenmostlyexplained
byenvironmentalandecologicalfactorslikebioticinteractions[16–19],abioticfactors[13,20–22]or
morerarely,theorganism’slifecycle[15,23]aswellasphysiologicalstate[24,25]. Overall,theeffects
ofabioticfactorsontheproductionofspecializedmetaboliteshaveproducedcontrastingresultsso
far [20,26–29]. As highly effective filter feeders, marine sponges are able to take up a large range
(<0.2–50 µm) of inorganic and organic particles from seawater [30]. Therefore, the environmental
conditions,suchastheseawatertemperatureortheavailabilityofnutrients,areexpectedtoinfluence
thespongemetabolism, andassessmentoftheirimpactonthebiosynthesisandconcentrationsof
specializedmetabolitesisofhighrelevanceinthecontextofspongefarming.
In contrast to previous studies where a targeted analysis of a small number of identified
compounds was performed, comparative metabolomics allows the monitoring of environmental
changes using a broader range of metabolites [31]. Metabolomics refers to the set of low
molecular-weightmolecules(<1.5kDa)biosynthesizedbythecellsinassociationwithitsmetabolism,
definedasthemetabolome[32]. Providingasnapshotofthemetabolicphenotypeatagiventime,
metabolomicshasthepotentialtodeterminebiomarkersindicativeoftheeffectsofabioticstressors
suchasnaturalandanthropogenicfactors[33]. Therefore,thecomparisonofmetabolomesforthe
investigation of organism-environmental interactions using high resolution analytical technology
constitutesapowerfultool. Inthisstudy,weusedametabolomicsapproachusingUHPLC-HRMSto
assesstheinfluenceofseasonalityandgeographicaldistributiononthespecializedmetabolomeofthe
MediterraneanencrustingspongeCrambecrambe(Porifera,Demospongiae)forwildpopulationsandin
situfarms. Inasecondsetofexperimentsperformedexsitu,weinvestigatedtheenvironmentalfactors
likelytoberesponsibleforthedetectedvariationsinsitu(temperature,light,andnutrientsavailability).
ThespecializedmetabolomeofC.crambehasbeenwellcharacterizedduringthepastdecades.Itis
composedoftwofamiliesofpolycyclicguanidinealkaloids(PGA,SupplementalinformationFigure1);
the crambescins (one or two cycles) and crambescidins (five cycles) that both exhibit substantial
biological activities [34–39]. From C. crambe, several crambescidins (named from their molecular
weight i.e., 800, 816 and 830) have been isolated and characterized [36,40,41] together with three
sub-families of crambescins A–C differing in the presence of a pyrrolidine ring (A), a spiroaminal
(B)andalinear3-hydroxypropylchain(C)[37,40,42]. Werecentlyshowedthatcrambescidinsmight
formachemicalshieldaroundthespongethatcontributestothesponge’sprotectionandecological
success[6]. Alterationofthischemicalshieldduetochangesinenvironmentalconditionswillaffect
theallelopathicinteractionsthatthespongemaintainswithitshabitat.
2. MaterialsandMethods
• Collectionofwildandfarmedsponges
TriplicatesofwildspecimensofC.crambewerecollectedatdifferentdepths(5,10and25m)by
SCUBA diving in the Ligurian Sea (Tigullio Gulf Punta del Faro 44◦29(cid:48)84.83” N, 9◦21(cid:48)86.02” E) in
April(spring)andSeptember2013(autumn)andintheSardiniaSea(PortoConteBay,40◦35(cid:48)43.69”N,
8◦12(cid:48)31.44”E)(Figure1(A-a,B-c))inApril2013(spring). Autumnandspringcorrespondrespectively
totheassumedmaximumandminimumproductionofguanidinealkaloidsbyC.crambe[17]. Farmed
specimensofC.crambewerecollectedbySCUBAdivingintheLigurianSea(TigullioGulf,PuntaPedale
44◦19(cid:48)15.35”N,9◦12(cid:48)57.00”E)andintheSardiniaSea(TramariglioBay,40◦35(cid:48)32.60”N,8◦10(cid:48)11.72”E)
(Figure1(A-b,B-c))inAprilandSeptember2013.Aftercollection,thespongeswereimmediatelyfrozen
at−20◦C,freeze-driedandmaintainedat−20◦Cuntilextraction.
MaricultureofC.crambewasperformedusingthemethodologyUSAMA®[43]usingunderwater
modular apparatus composed of different structures: (1) a stainless steel plant, composed of
1.5×1.5m2modulesthatcanbecombinedindifferentways(SupplementaryMaterialsFigureS2A,B).
Severalpivotsareusedtosettle, atequaldistance, theropeswiththespongeexplants(necklaces);
(2)aPVCmodularplant,madebyrectangularorsquareforms(SupplementaryMaterialsFigureS2C,D)
anchored by ropes to structures. Pivots are inserted to anchor the sponge necklaces. The sponge
Mar.Drugs2017,15,181 3of15
Mar. Drugs 2017, 15, 181  3 of 15 
explants were assembled in the ‘sponge necklaces’ [44]: (i) travertine tile method, composed of
m10ecthmod×, c1o0mcmpotsreadv eorft i1n0e ctmile ×s 1a0n cchmo rteradvteorttinhee triolepse asnacnhdortehde troo ptheed rroiplleesd ainndto ththee rocporen derrilolefdr oincktoy 
tthaeb lectosr.nTehr eosfp rooncgkeye xtapblalenttss. wTheree sppaosntegde oenxpealacnhtsti lweuersein gpaastneodn -otonx iecauchn dteilrew uatseinrgb i-ac onmopn‐otnoexnict 
uepnodxeyrwpauttetyr b(Siu‐cboCmopaot,nVeennte ezpiaonxiyY apcuhtttyS y(sStuembCs,oIatta,l yV)e(nSeuzpipanleim Yeancthatr ySyMsatetemrisa, lsItFaligyu) r(eSuSp2Ep)l;e(mii)ennytaloryn 
Mmaetsehrimales tFhiogdu,rme Sad2Ee)o; f(iai) pnoyslho-nli kmeenshet mweitthho2dc, mmamdees hoef sa fipxoesdh‐tloikreo pneest .wTihthe s2p comng meeesxhpelas nftixsewde troe 
reonpcelos.s eTdhein spthoenmgee eshxpesla(nStusp wpelerme eennctaloryseMd ainte trhiael smFeisghuerse (SS2uFp)p.lementary Materials Figure S2F). 

FFiigguurree 11.. CCoolllleeccttiioonn ssiitteess ooff CCrraammbbee ccrraammbbee iinn tthhee LLiigguurriiaann SSeeaa ((AA)) ((aa PPuunnttaa ddeell FFaarroo,, bb PPuunnttaa PPeeddaallee)); ;
aanndd SSaarrddiinniiaa SSeeaa ((BB)) ((cc CCaappoo CCaacccciiaa))..  
The abiotic sea parameters were obtained by the multiparameter probe ‘Idronaut’ in the 
Theabioticseaparameterswereobtainedbythemultiparameterprobe‘Idronaut’inthePortofino
Portofino Promontory area from 0 to 35 m (Povero pers. comm.). At both the Liguria and Sardinia 
Promontoryareafrom0to35m(Poveropers. comm.). AtboththeLiguriaandSardiniasites,water
sites, water temperatures were recorded by means of an underwater HOBO® Data Logger (Onset, 
temperatures were recorded by means of an underwater HOBO® Data Logger (Onset, MA, USA)
MA, USA) installed in the plants and in the wild. The temperature was recorded every 6 h. Monthly 
installedintheplantsandinthewild. Thetemperaturewasrecordedevery6h. Monthlyaverage
average temperature values were then considered. 
temperaturevalueswerethenconsidered.
 Aquarium experiments 
• Aquariumexperiments
Crambe crambe specimens were sampled in the bay of Villefranche‐sur‐mer (20 m depth along 
CrambecrambespecimensweresampledinthebayofVillefranche-sur-mer(20mdepthalong
the rocky coast) in October 2013 (Temperature & Light experiment) and in January 2014 (Nutrient 
therockycoast)inOctober2013(Temperature&Lightexperiment)andinJanuary2014(Nutrient
experiment). Specimens were carefully transferred to the Observatory of Villefranche‐sur‐mer and 
experiment). SpecimenswerecarefullytransferredtotheObservatoryofVillefranche-sur-merand
aacccclliimmaatteedd ffoorr oonnee mmoonnthth inin aqauqaurairai asuspupplpielide dbyb cyocnotinntuinouuosu sseasweaawteart eflroflwo (w2 L(2∙mLi·nm−1i)n p−u1m) ppuemd apte ad 
5 m depth in the bay of Villefranche‐sur‐mer with a natural light/dark cycle as described previously 
at a 5 m depth in the bay of Villefranche-sur-mer with a natural light/dark cycle as described
[45]. Each sponge specimen was cut into small replicates of similar size (~5 cm2) in order to perform 
previously[45].Eachspongespecimenwascutintosmallreplicatesofsimilarsize(~5cm2)inorder
the two following sets of experiments: (1) “Temperature and Light experiment” and (2) “Nutrient 
to perform the two following sets of experiments: (1) “Temperature and Light experiment” and
experiment” that investigated the effects of variation in those environmental parameters on the 
(2)“Nutrientexperiment”thatinvestigatedtheeffectsofvariationinthoseenvironmentalparameters
specialized  metabolism  of  C. crambe.  Being  collected  from  an  encrusted  rocky  substrate,  the 
on the specialized metabolism of C. crambe. Being collected from an encrusted rocky substrate,
specimens were carefully cut into different pieces with a hammer and a chisel to produce genetically 
thespecimenswerecarefullycutintodifferentpieceswithahammerandachiseltoproducegenetically
identical biological replicates.  
identicalbiologicalreplicates.
2.1. Temperature and Light Experiment  
2.1. TemperatureandLightExperiment
This first experiment was performed in November 2013, using a running seawater flow rate of 2 
ThisfirstexperimentwasperformedinNovember2013,usingarunningseawaterflowrateof
L2∙Lm·imn−in1. −T1h.eT hfievefi vfoellfoowlloinwgin tgretartematemnetsn twsewreer iemimplpelmemenetnetde:d (:i)( ic)ocnotnrtorol l(T(T == 161 6°C◦C; n;naatuturaral llilgighht tccyyccllee)), ,
((iiii)) lliigghhtt 2244//2244 hh ((330000 lluuxx)),, ((iiiiii)) ddaarrkk 2244//2244 hh (0(0 luluxx)), ,((iviv)) 1144 °◦CC, ,aanndd ((vv)) 2200 °◦CC. .TThhee ccoolllleeccttiioonn ooff ssppoonnggee 
specimens was performed after 0, 7 and 30 days, in each aquarium and in triplicate. Water 
specimenswasperformedafter0,7and30days,ineachaquariumandintriplicate. Watertemperature
temperature and light intensity were recorded in all aquaria by means of an underwater HOBO® Data 
Logger (Onset, MA, USA). The probes were set to record temperature 2 times per day (every 12 h).
Mar.Drugs2017,15,181 4of15
and light intensity were recorded in all aquaria by means of an underwater HOBO® Data Logger
(Onset,MA,USA).Theprobesweresettorecordtemperature2timesperday(every12h).
2.2. NutrientExperiment
ThesecondexperimentwasconductedinMarch2014,usingrunningseawaterataflowrateof
0.7L·min−1. Fiveaquariaeachcontaining4spongereplicatesweresetup: 2controlaquariawith
onlyrunningseawaterand3treatmentaquariawithrunningseawaterplustheadditionofanutrient
mixture. ThenutrientmixturewasmadeupofMQwaterandinorganicsalts,i.e.,phosphate(KH PO )
2 4
andnitrate(NaNO ),andwasbroughttothethreetreatedaquariabyaperistalticpumpataflow
3
rateof1mL·min−1 inordertoreachaconstantconcentrationofphosphate(PO 3− =2.5µmol·L−1)
4
andnitrate(NO − =18µmol·L−1). Theseconcentrationswerechosenastheyareconsistentwiththe
3
anthropogenizedsiteofCortiou, closetothecityofMarseilleintheNorthwesternMediterranean
Sea[46]. AtT0andafter5,11,18and24days,spongereplicateswerecollectedfromeachaquarium,
andseawaterwassampledfromonecontrolandonetreatedaquaria. Seawatersampleswerespiked
withmercuricchlorideandkeptat+4◦C.Dissolvedinorganicsalts(PO 3−,NO −,NO − andSiOH )
4 3 2 4
in seawater were then analyzed by a colorimetric method, using a Continuous Flow Automated
AnalyzerTechniconAutoAnalyzerIIfollowingamethoddevelopedpreviously[47]. Foreachsetof
experiments,spongesampleswereflash-frozenusingliquidnitrogenandmaintainedat−26◦Cuntil
chemicaltreatment.
• Metabolomicanalysis
Methanol, acetonitrile and dichloromethane (Chromasolv®, analytical grade), formic acid
(puriss.P.a. ~98%)andammoniumformate(LC-MSUltraforUHPLC-MS)werepurchasedfromSigma
Aldrich(Saint-QuentinFallavier,France). UltrapurewaterwaspreparedusingaMilli-Qwatersystem
(MilliporeLtd.,Billerica,MA,USA).
Allspongesampleswerefreeze-driedandgroundtoobtainadrypowderwhichwasextracted
threetimeswitha15mLmixtureofMeOH/CH Cl (1:1,v/v)inanultrasonicbath(35kHz)atroom
2 2
temperature. DuetotheencrustinggrowthformofthespongeC.crambe,therockysubstratewasburnt
at550◦Cfor24htoobtainthespongeorganicweight(Totalweight−rockweight=spongeweight).
Crude extracts were filtered through 8 µm paper filter (Whatman Grade 2V), dried using a rotary
evaporatorandfurtherfractionatedbysolidphaseextractionovera500mgC18silicaSPEcartridge
(Phenomenex, Torrance, CA, USA) with a step gradient of H O (20 mL), MeOH (20 mL), CH Cl
2 2 2
(20mL).ThemethanolfractionwasdriedusinganSPD111SpeedVac(Thermosavant,ModelRH12-28),
concentratedto10mg·mL−1,andstoredat−26◦CuntilfurtherdilutionpriortoLC-MSanalysis.
UHPLC-HRMSanalysis
PriortotheUHPLC-HRMSanalysis,allsamplesweresolubilizedin1mLofMeOHandwere
diluted1/1000. Inaddition,fiveQualityControl(QC)sampleswerepreparedbycombining10µLof
eachsampleina2mLvialandthesepoolswereinjectedeverysevensamplestoallowchromatogram
alignmentduringdatatreatment. In-lineUHPLC-UV-HRMSanalysiswasperformedusingaDionex
systemUltimate3000equippedwithanautosamplerandaDionexUltimate3000diodearraydetector
(210and280nmdetection),connectedtoaqToFmassspectrometerwithanelectrosprayionization
interface(BrukerImpactII).Massspectrawererecordedinthepositivemode. UHPLCseparationwas
achievedonananalyticalNucleodurPolarTeccolumn(100×2mm,1.8µm,MachereyNagel)using
a linear elution gradient of H O/CH CN/formic acid to which was added 10 mM of ammonium
2 3
formatefrom80:20:0.1(v/v/v,isocraticfrom0to2min)to40:60:0.1(v/v/v,isocraticfrom8to10min),
ataflowrate0.45mL·min−1foratotalof14min. Theinjectedvolumewassetat10µLanddetection
at 280 nm. The mass spectrometer analyzer parameters were set as follows: nebulizer sheath gas,
N (2.1bar);drygas,N (8L/min);capillarytemperature,200◦C;capillaryvoltage,2500V;endplate
2 2
offset,500V;collisiongas,He;collisionenergy,4eV.Datawereacquiredinthe50to1200m/zrange.
Mar.Drugs2017,15,181 5of15
UHPLC-HRMSdataprocessingforuntargetedanalyses
FollowingUHPLC-ESI-HRMSdataacquisition,MSchromatograms(BasePeakchromatogram—BPC)
wereexportedaslinespectraandconvertedintothenetCDFfileformattoprocessthedataincentroid
modewithXCMS[48]usingRsoftware(version3.2.2.). TheXCMSapproachinvolvedseveralsteps
necessary to generate a final matrix: (1) Peack picking (peakwidth = c(2, 20), ppm = 10) without
thresholdprefilter[49],(2)retentiontimecorrection(method=”obiwarp”),(3)grouping(bw=10,
minfrac=0.3,minsamp=1),(4)Fillpeaksusingaparameterof“noiselevel”thatsetstheminimum
intensityforacentroiddatapointtobeconsideredaspartofapeak(noiselevel=103),andfinally(5)
reportanddatamatrixgeneration(ions/Retentiontime×sample)thatwasexportedusingMicrosoft
Excel. Two successive filtering steps were applied to the matrix in R in order to suppress data of
high analytical variability: (i) signal/noise ratio > 10 between variables matching in both extracts
and blank samples (considers that ions are part of the noise if their mean intensity is less than
10-foldtheblankvalue)and(ii)coefficientofvariationintheintensityofthevariablesfortheQC
samples<20%. Eachpeakareawasfurthernormalized(accordingtothedriftofequivalentionof
pooled samples), mean-centered (the mean ion intensity of all samples was subtracted from each
sampleionintensity)andfinallylog-transformedtomakeindividualfeaturesmorecomparable,using
on-lineMetaboAnalyst3.0[50]. Multivariateanalysiswasconducted(Principalcomponentsanalysis
(PCA)andPartialLeastSquares-DiscriminantAnalysis(PLS-DA))andPCAwasfurtherchosensince
itgavethebestdiscrimination. Veryimportantpeaks(VIPs)weredeterminedaccordingtothePCA
loadingplots.
Metabolomictargetedanalysis
Chemical identification was performed only for compounds of interest (crambescins and
crambescidins, Supplementary Materials Figure S2) that have been extensively described.
Thechromatogramobtainedbetween5.5and8minofretentiontimewasconsideredourmetabolomic
window,andthemajorcompoundsofthisareawerethoroughlyidentifiedbycomparingtheirMS
patternwithastandard(SupplementaryMaterialsTableS1)[5,37]. Becauseseveralionadductscan
bedetectedforasinglecompoundwithelectrosprayionization,thefollowingmono-anddi-charged
adductswerealsotakenintoaccountinthemetaboliteannotation: [M+H]+,[M+Na]+,[M+NH ]+,
4
[M + 2H]2+/2, [M + 2NH ]2+/2. They are the most commonly observed in positive electrospray
4
ionization mode with ammonium formate and formic acid as eluent modifiers. The theoretical
m/z values for typical adduct species were compared with the experimental values to ensure the
identification of the compounds (once per treatment and fraction and once for the QC samples).
Forspectrapresentingmultipleadducts,wesummedtheareaofalladductstoobtainthetotalintensity
corresponding to one compound. Analytical blanks confirmed that no memory effects or sample
contaminationskewedresults. Theconcentrationofthemetaboliteswasassessedusingacalibration
withstandardsofcrambescidin-816andcrambescin-A2-462,assumingallcompoundsofthesame
familyhadsimilarionizationpotential. One-wayANOVAwasperformedonthedatatoinvestigate
thestatisticaldifferencesbetweenconditions.
3. Results
• Insitustudywithwildandfarmedsponges
Toinvestigatedepth,seasonalandgeographiceffectsonthespecializedmetabolomeofC.crambe,
wildspecimenswerecollectedintheLigurianSeaduringbothspringandautumnandintheSardinia
Seaduringspring. Afirstmetabolomicsapproachshowedasignificanteffectoftheseasononthe
metabolomeofwildspecimensofC.crambeasrevealedbythePrincipalComponentAnalysis(PCA)
plot (Figure 2). Variations of the metabolic content in relation with geography or depth were not
statisticallysignificant. CrambecrambewasalsofarmedintheLigurianSeausingbothmeshandtiles
protocols,andsampleswerecollectedinautumnandspring. Themetabolomicsapproachshowsthat
Mar.Drugs2017,15,181 6of15
thespecializedmetabolomeofspecimensfarmedarenotsignificantlydifferentfromwildspecimens,
andthehighestvariabilitywasattributedtoseasonalchanges(Figure2).Ontheotherhand,thefarming
protocolsdidnotshowsignificantdifferencesinthespecializedmetabolomeofC.crambe. Therefore,
growingconditionsandlocationhavesmallimpactonthecontentofthespecializedmetabolitesof
thesponge.
Mar. Drugs 2017, 15, 181  6 of 15 

m/z  RT [M+H]+
490.3373  6.90  979.6746 
483.3403  6.98  963.6806 
246.2087  7.05  489.4174 
223.1689  4.86  445.3378 
263.2011  6.18  525.4022 
Figure 2. PCA plot obtained from the metabolomics study on wild (Wi) and farmed (Fa, Tile or Mesh 
Figure2.PCAplotobtainedfromthemetabolomicsstudyonwild(Wi)andfarmed(Fa,TileorMesh
protocols) samples harvested at different locations (Sar = Sardinia, Lig = Ligurian) in both autumn 
protocols)samplesharvestedatdifferentlocations(Sar=Sardinia,Lig=Ligurian)inbothautumn
(Aut) and spring (Sp). The table lists the five most important ions contributing to the distribution of 
(Aut)andspring(Sp).Thetableliststhefivemostimportantionscontributingtothedistributionof
samples in the PCA plot.  
samplesinthePCAplot.
The  five  most  contributive  ions  to  this  PCA  were  identified  (Table  S1) using  full scan 
Tchheromfiavteogmraomsst fcoor nvtrailbiduattiivone. iIonntesretsotintghlyis, mPCosAt owf ethree iiodnesn teilfiuetedd (bTeatwbleeenS 14).8 uasnind g7 fmuliln, scan
corresponding to compounds slightly more polar than crambescins and crambescidins. None of 
chromatograms for validation. Interestingly, most of the ions eluted between 4.8 and 7 min,
these compounds were assigned to known metabolites. Therefore, we concluded that the variability 
corresponding to compounds slightly more polar than crambescins and crambescidins. None of
of the metabolic content is not driven by the major metabolites (i.e., crambescins and crambescidins) 
thesecompoundswereassignedtoknownmetabolites. Therefore,weconcludedthatthevariabilityof
but rather by minor metabolites.  
themetaboliccontentisnotdrivenbythemajormetabolites(i.e.,crambescinsandcrambescidins)but
To better investigate the variations in the concentrations of the two major families of specialized 
ratherbyminormetabolites.
metabolites, the crambescins and crambescidins, a targeted metabolomics analysis was undertaken. 
TForobmet ttehrisi nsvtuedsyti,g watee othuetlivnaerdi aat iloonwseirn ctohnececnotnracteinontr aotfi obnostho ffatmheilitews odumrianjgo rsfparmingil iceosmopfasrpeedc itaol ized
metabaoultiutemsn, th(3e9 craagmainbsets c3i9n1s aμnmdolc∙gra−1m sbpe, spc id=i n1s.5,5a t×a r1g0e−6t e*d**m, Letigaubroialonm siacmspalneas;l yFsiigsuwrea s3Aun).d Terhtea ken.
Fromctohniscesntutrdayti,own eoof ustpleincieadlizaeldo wmeertacboonlciteens trina tiaountuomfnb owthasf asmupilpieosrtdeud rbinyg bsoptrhi ntghec ocmrapmabreesdcitno aanudtu  mn
(39agcarainmsbte3s9ci1diµnm faoml·igli−es1 (supp, pto= 1718.5 a5nd× 21105− μ6m**o*l∙,gL−1i gspu,r riaesnpescatmivpellye)s. ;AF isgimurilear3 Ase)a.sTonhael ctroenncde nwtarsa tion
shared by farmed specimens for both the crambescins and crambescidins families (Figure 3A). Wild 
of specialized metabolites in autumn was supported by both the crambescin and crambescidin
familiaensd( ufaprmtoed1 7s8amanpldes2 1sh5oµwmedo la·glm−1osst ps,igrnesifpiceacntti v(epl y=) .0.A065s7im) idlaifrfesreenacseosn afolrt rbeontdh wfaamsilsiehsa roefd  by
compounds; these differences being mostly supported by samples harvested in autumn (93 against 
204 μmol∙g−1 sp for the crambescins and 169 against 284 μmol∙g−1 sp for the crambescidins).
Mar.Drugs2017,15,181 7of15
farmedspecimensforboththecrambescinsandcrambescidinsfamilies(Figure3A).Wildandfarmed
samples showed almost significant (p = 0.0657) differences for both families of compounds; these
differencesbeingmostlysupportedbysamplesharvestedinautumn(93against204µmol·g−1spfor
thecraMmabr. eDsrcugins 2s01a7n, 1d5, 116819  against284µmol·g−1spforthecrambescidins). 7 of 15 

FigureF3ig.uCreo 3n.c Ceonntrceantitorantisonosf ocfr carmambbeesscciiddiinnss aanndd crcarmambebsceisncsi nins Ci.n crCam.bcer ainm μbmeoinl∙gµ−1m spo lo·bgt−ai1nesdp froobmta ined
the targeted metabolomics study for (A) wild and farmed specimens, (B) Temperature and Light 
fromthetargetedmetabolomicsstudyfor(A)wildandfarmedspecimens,(B)TemperatureandLight
experiment, and (C) Nutrient experiment (*** p < 5 × 10−4; ** p < 0.005; * p < 0.05, one‐way ANOVA). 
experiment,and(C)Nutrientexperiment(***p<5×10−4;**p<0.005;*p<0.05,one-wayANOVA).
The spatial location of C. crambe slightly impacted the content of crambescidins since their 
Thcoenscpenattriaatliolnosc iant iSoanrdoinfiaC w.ecrrea smtabteistsilciagllhyt lhyigihmerp thaacnte tdhotshee inc tohne tLeingturoiafnc Sreaam (4b5e μscmidoli∙ngs−1 sspin, pc e= their
concen0t.r0a3t3i8o n*)s, ai nreSsualrtd minoisatlyw seurpepsotratetids tbiyc athlley chonigchenetrratthioanns tohf ocrsaeminbetshciediLnisg. uItr iisa nnoSteewao(r4th5yµ tmhaot lt·hge− 1 sp,
crambescins did not show any significant difference between the two regions (23 and 22 μmol∙g−1 sp 
p=0.0338*),aresultmostlysupportedbytheconcentrationsofcrambescidins. Itisnoteworthythat
in Liguria and Sardinia, respectively). 
thecrambescinsdidnotshowanysignificantdifferencebetweenthetworegions(23and22µmol·g−1
spinLig urEiax asnitud eSxapredriimniean,trse spectively).
Some experiments were also conducted in aquaria to study the variability of the specialized 
• Exsituexperiments
metabolome of C. crambe when submitted to diverse environmental factors.  
So me experiments were also c onducted in aquaria to study the variability of the specialized
metabolomeofC.crambewhensubmittedtodiverseenvironmentalfactors.
Mar.Drugs2017,15,181 8of15
3.1. TMemar.p Derruagtsu 2r0e17a, n15d, 1L8i1g htExperiments 8 of 15 
A3.1lt. hToemugpehrawtuerea atntde mLipghtte dExtpoerkimeeenptsfi xedconditionsoftemperatureinallaquaria,thiswasmade
difficultbyanopenrunningseawatersystemthatisaffectedbynaturalvariations. Thetemperatureof
Although we attempted to keep fixed conditions of temperature in all aquaria, this was made 
thecontrolaquarianaturallyfluctuatedbetween15and17◦C,andamaximumofone-degreevariation
difficult by an open running seawater system that is affected by natural variations. The temperature 
wasroefc othred ecdonftororlb aoqtuhatrhiae ncoatnutrraolllyle dflu“c1t4ua◦tCed” b(beetwtweeene n151 3a.n5d◦ C17a °nCd, a1n4d.5 a◦ Cm)aaxnimdutmhe o“f2 o0n◦eC‐d”e(gbreeetw  een
19.5◦vCaraiantidon2 0w.5as◦ rCec)oerxdpeder fiomr beontths .thLei gcohnttirnotlleends “it1y4 w°Ca”s (bseettwaete3n0 103L.5u °xC (a~n1d2 1040.5W °C/)m an2d) dthuer “in20g °tCh”e 12h
ofal(ibgehtwt-deeanr k19c.5y c°lCe ainndt h2e0.5te °mC)p eexrpaeturirmeeenxtsp. eLriigmhte nintt.enFsoirtyt hweasl isgeht taet x3p00e rLiumxe (n~1ts2,0t0h We/smam2) educryinclge was
maintthaei n1e2d hf oorf tah leigchotn‐dtarorkl ecxycplee riinm tehnet ,teamnpdetrhateurree seuxlptseroimftehnet. mFoert athbeo lloigmhti ceaxpnearliymseenstws, ethree csoammep ared
withcaypcleer wmaasn menatinltiagihntede xfpore rthime econnttaronld exppeerrmimaennetn, tandda rtkhee xrepseurlitms oefn tth.e metabolomic analyses were 
compared with a permanent light experiment and permanent dark experiment.  
A PCA plot revealed no clear difference between control, temperature and light experiments
A PCA plot revealed no clear difference between control, temperature and light experiments 
(Figure4). Itisnoteworthythattheinter-samplevariabilitywithinatreatmentincreasedwiththe
(Figure 4). It is noteworthy that the inter‐sample variability within a treatment increased with the 
experimentaltimeasshownbythedistributionofthereplicates.Tobenoted,thefivemostcontributive
experimental time as shown by the distribution of the replicates. To be noted, the five most 
ionstothePCApresentedmid-polarity,veryclosetothePGA(5.5<RT<6.5),butwerenotassigned
contributive ions to the PCA presented mid‐polarity, very close to the PGA (5.5 < RT < 6.5), but were 
toannyokt nasoswignnemd etota abnoyl iktenso.wn metabolites. 

m/z RT [M+H]+ Known 
251.2014 6.45 501.4028 no
273.1966 5.58 545.3932 no
260.2066 6.06 519.4132 no
256.1937 6.5 511.3874 no
265.1986 6.55 529.3972 no  
Figure 4. PCA plot obtained from an untargeted metabolomics study of the temperature/light 
Figure 4. PCA plot obtained from an untargeted metabolomics study of the temperature/light
expereixmpeernimt.eCn=t. Cco =n tcroonlt,rDol,= DD =a Drka,rLk, =L L=i Lghigth,t1, 414◦ C°C= =T Teemmppeerraattuurree 1144 ◦°CC aanndd 2200 °◦CC == TeTmempepreartautruer 2e02 0◦C.
°C. The table shows the five most important peaks that contribute to the distribution of samples on 
Thetableshowsthefivemostimportantpeaksthatcontributetothedistributionofsamplesonthe
the PCA plot, along with their retention time.  
PCAplot,alongwiththeirretentiontime.
Mar.Drugs2017,15,181 9of15
Thetargetedmetabolomicsstudyconfirmedtheabsenceofstatisticallysignificantdifferences
in the concentrations of the crambescins and crambescidins of C. crambe specimens across the
treatments. Thesummedconcentrationofthemajorspecializedmetaboliteswasveryclosetothe
Mar. Drugs 2017, 15, 181  9 of 15 
onemeasuredinsamplesharvestedinthefield,althoughtheproportionofcrambescidinrepresented
up to 50–70T%he otafrgtehteedt omteatlabcoolnomteincst s(tFuidgyu croenf3irBm).edD tihfefe arbesnecneces obfe sttwatiesetincaltlrye asitgmniefincatsn,t dfoifrfeirnenscteasn ce, the
in the concentrations of the crambescins and crambescidins of C. crambe specimens across the 
concentration of crambescidins in control at T2 (increase by a factor 2.7), were mainly driven by
treatments. The summed concentration of the major specialized metabolites was very close to the 
inter-individualvariability.
one  measured  in  samples  harvested  in  the  field,  although  the  proportion  of  crambescidin 
represented up to 50–70% of the total content (Figure 3B). Differences between treatments, for 
3.2. NutrientExperiment
instance, the concentration of crambescidins in control at T2 (increase by a factor 2.7), were mainly 
driven by inter‐individual variability.  
3.2.1. NutrientConcentrations
3.2. Nutrient Experiment 
Although only nitrate and phosphate were added to the aquaria, four inorganic salts were
measuredintheseawaterduringtheexperiment: nitrate,nitrite,phosphateandsilicate(SI,FigureS3).
3.2.1. Nutrient Concentrations 
No difference in the concentration of non-added salts was noticeable between control and treated
Although only nitrate and phosphate were added to the aquaria, four inorganic salts were 
aquaria; those salts being most probably controlled by the sponge. The nutrient mixture was
measured in the seawater during the experiment: nitrate, nitrite, phosphate and silicate (SI, Figure S3). 
efficientNlyo daidffdereedncteo int hthee tcroenacteendtraatqiouna orifu nmon‐aasddsehdo wsalnts bwyast hneotinceitarbaltee baetnwdeepnh coosnptrhola atendc otrneacteendt rations
(23anda3quµamriao;l ·tLh−os1e astaTlt4s ,breeisnpge cmtiovste lpyr)o.bHabolwy ecvonetrr,oalltetdem bpy tethde hsopmonogge.e nTehoe unsuctroienncte nmtirxattuiroen wsians treated
aquariaefwficeireentolyn aldydreeda tcoh tehde taretadteady aq1u8araifutmer aism shpoowrnta bnyt tvhea rniiatrtaioten asn(dP pOho3s−ph:a0te. 7c–o1n.c5enµtrmatoioln·Ls −(213; NO −:
4 3
6–13µmanodl ·L3 −μ1m).olT∙Lh−1i sati nTs4t, arbeislpiteyctimvealyy). bHeoewxepvleari, naettdembpytetdh ehotmecohgneniceaolusis csounecsenetnractoiounns tienr etrdeadteudr ingthe
aquaria were only reached at day 18 after important variations (PO43−: 0.7–1.5 μmol∙L−1; NO3−: 6–13 
deliveryofthenutrientmixturebytheperistalticpump.
μmol∙L−1). This instability may be explained by the technical issues encountered during the delivery 
of the nutrient mixture by the peristaltic pump.  
3.2.2. MetabolomicsStudy
3.2.2. Metabolomics Study 
The distribution of the samples on the PCA showed no clear difference in the specialized
metabolomeThoef cdoisntrtriboultaionnd onf utthrei esnatm-tprleeast eodn stahme PpCleAs (sFhiogwuerde 5n)o. Hcleoawr edvifeferr,eancceh ainn gtheei nsptehceiamlizeetda bolome
atT4sememeteabdotloombee ionf itcioantetrdola nadnds hnouutrlidenbt‐etrceoatnesdi dsearmedpl.eIst  i(sFiwguorret h5)n. oHtiocwinegvetrh, aat thcheamngoe stinc otnhter ibutive
metabolome at T4 seemed to be initiated and should be considered. It is worth noticing that the most 
ionsdifferedinthisexperimentfromthepreviousone.Inthiscase,wewereabletoidentifycrambescin
contributive ions differed in this experiment from the previous one. In this case, we were able to 
C1withalowerconcentrationatT4innutrient-treatedsamples.
identify crambescin C1 with a lower concentration at T4 in nutrient‐treated samples.  

Figure5.Cont.
Mar.Drugs2017,15,181 10of15
Mar. Drugs 2017, 15, 181  10 of 15 
m/z RT [M+H]+ Known 
280.1541 7.03 559.3082 no
408.1482 8.13 408.1482 no
326.2444 5.8 326.2444 no
358.2473 1.15 358.2473 no
234.1909 6.5 467.3818 yes  
Figure 5. PCA plot obtained from an untargeted metabolomics study of the nutrients. C = control and 
Figure5.PCAplotobtainedfromanuntargetedmetabolomicsstudyofthenutrients.C=controland
N = nutrient‐treated samples and triplicates of biological samples are represented by a colored point. 
N=nutrient-treatedsamplesandtriplicatesofbiologicalsamplesarerepresentedbyacoloredpoint.
The table shows the five most important ions contributing to explain the sample distribution on the 
ThetablPeCsAh polwots atlohnegfi wviethm thoesirt riemtepntoiornta tnimteio annsd cthoenirt reixbpuetcitnedg [tMo +e xHp]+l.a inthesampledistributiononthe
PCAplotalongwiththeirretentiontimeandtheirexpected[M+H]+.
The targeted metabolomics analysis showed that the differences observed between control and 
nutrient‐treated samples were not statistically different for the major metabolites (Figure 3C). The 
The targeted metabolomics analysis showed that the differences observed between control
concentrations measured in this experiment were very close to those measured in in situ samples.  
and nutrient-treated samples were not statistically different for the major metabolites (Figure 3C).
Thecon4c.e Dntisrcautisosinosn measuredinthisexperimentwereveryclosetothosemeasuredininsitusamples.
The Mediterranean Sea is considered an oligotrophic region where spatial distribution of 
4. Discussion
biogeochemical elements [51] as well as plankton species [52] is dependent on meso‐scale circulation 
Theof Mwaetderi tmerarsasense. aVnariSaetiaoniss ofc osuncshi dbeiorteicd anadn aoblioigtioc tpraorpahmiecterrse gmioignht wcohnetrriebustep atot ivaalridatiisotnrsib inu tion of
the production of the specialized metabolites in common organisms like sponges. This aspect is of 
biogeochemicalelements[51]aswellasplanktonspecies[52]isdependentonmeso-scalecirculation
interest as sponges are recognized as important producers of metabolites with pharmaceutical 
ofwatermasses. Variationsofsuchbioticandabioticparametersmightcontributetovariationsin
interest. We selected the Mediterranean sponge C. crambe as a model due to the presence of a high 
the prodquuacnttiiotyn oof fbtihoaectsivpee cailaklailzoeidds mnaemtaedb oPloitlyecsyicnlicc Gomuamnidoinnico rAglaknaliosimdss frloikme tshpeo cnragmesb.escTinh iasnads pect is
of interecrsatmabsesscpidoinng faemsialirees. recognized as important producers of metabolites with pharmaceutical
interest. WeIns eolredcetre tdo tahseseMss ethdei tvearrriaabnieliatyn, ssppeocnimgeenCs .ofc rbaomthb ewailsd aanmd ofadremleddu Ceratmobteh cerapmrbees sepnocnegeosf ahigh
were harvested from two distinct Mediterranean sub‐basins (Ligurian Sea and Sardinia Sea), located 
quantity of bioactive alkaloids named Polycyclic Guanidinic Alkaloids from the crambescin and
from ~500 kilometers along a latitudinal cline (44°18′ N vs. 40°35′ N, respectively), thus differing by 
crambescidinfamilies.
the intensity of the solar radiation they receive as well as by their hydrological characteristics 
In order to assess the variability, specimens of both wild and farmed Crambe crambe sponges
(temperature, salinity) and primary productivity [53]. The production of specialized metabolites in 
werehawrvileds tsepdecfirmoemnst wwoasd sihsotiwnnc ttMo ebde ivteerrrya nsiemainlars ubbet-wbaeesnin tsh(eL tiwguo risaanmpSleead alnocdatSioanrsd i(nFiigauSree a2)),. located
from~5T0h0erkeifloorem, ettheer sroalleo npglayaedla tbityu deninviarloncmlineneta(l4 4p◦a1r8am(cid:48) Netevrss .in4 0t◦h3e5 (cid:48)coNnc,ernetsrpateiocnti voef lysp),etchiauliszeddi ffering
by the imnteetanbsoiltiyteso ifs tlhimeitseodl ainr trhaisd ciaasteio. Tnhtish leimyirteedc einivtreasapsecwifiecl ml aestabboylict hveairriahbiylidtyr ooblosegrivceadl bcehtawreaecnt eristics
the two locations might be related to the genetic diversity as previously suggested [1,33,54–57]. In 
(temperature, salinity) and primary productivity [53]. The production of specialized metabolites
accordance with previous studies [13–15,20,23,58], a marked seasonal trend of the metabolic content 
in wild specimens was shown to be very similar between the two sampled locations (Figure 2).
showed a maximum in autumn and a minimum in spring (specimens of the Ligurian Sea, Figure 2). 
Therefore,theroleplayedbyenvironmentalparametersintheconcentrationofspecializedmetabolites
Interestingly, the same seasonal pattern was observed in both wild and farmed specimens cultured 
is limiteind thinis athreias, scuagsgee.stTinhgi sa rloimle iotfe idnteinrntarla pshpyescioifilocgimcael tfaacbtoorlsic asv maraiianb dirliivtyerso obfs tehrev veadriabtieotnws eofe tnhet he two
locationsspmeciiaglhiztebd emreetlaabtoeldismto. Sthucehg feancteotrisc cdouivlde resitihtyera bse pgrreovwitohu [s2l1y,24su,2g5]g oers rteepdro[1d,u3c3t,i5on4 –[5577,]5.8]I.n Ita hcacso rdance
with prbeeveino uosftesnt uhdyipeosth[e1s3iz–e1d5 ,t2h0a,t2 t3h,5e 8r]e,leaasme aofr klaerdvasee ians olantea lsutrmemnder osifgtnhaels mthee teanbdo loifc tchoe netneenrgtys howed
investment in reproduction, allowing a switch towards an enhanced production of specialized 
amaximuminautumnandaminimuminspring(specimensoftheLigurianSea,Figure2).Interestingly,
metabolites during autumn and throughout the winter [17]. This trade‐off between the reproduction 
the same seasonal pattern was observed in both wild and farmed specimens cultured in this area,
and the production of specialized metabolites was hypothesized by [24] and [17] for C. crambe. Our 
suggestdinagta aserto laelsoo frienvteearlns aal spighnyifsiicoalnot gdicifafelrfeanccteo rins athsem maeitnabdorliicv ecorsntoefntt hbeetvwaereina twioinlds aonfdt hfaermspeedc ialized
metabolsipsemci.mSeuncs hinf aaucttuomrsnc (oFuigludree i2t)h, weritbhe hgigrhoewr ctohn[c2e1n,t2ra4t,i2o5n] oof rthree ptorxoicd curcatmiobnes[c5id7i,n5 8fa]m. Iitlyh ians wbielde noften
hypothespsioznegdest (hFaigtutrhee 3Are)l. eTahsues,o af cloamrvbaineatiinonla otfe instuermnaml fearctsoirgsn aactlisngth teogeenthdero wfitthh ebieontiec rfgacytoirnsv seuscthm  entin
as competition for space [1,16,18,19] may also explain the seasonal variability of the production of 
reproduction,allowingaswitchtowardsanenhancedproductionofspecializedmetabolitesduring
the specialized metabolites. Abiotic factors such as temperature or nutrient availability might trigger 
autumnandthroughoutthewinter[17]. Thistrade-offbetweenthereproductionandtheproduction
other processes like sponge reproduction or other constraints of primary metabolism [25,58,59], 
ofspecializedmetaboliteswashypothesizedby[24]and[17]forC.crambe. Ourdatasetalsoreveals
rather than directly affecting the specialized metabolome.  
a significanTthdei fmfeerteanbocleominicst haeppmroeatcahb oallilocwceodn teenncotmbpeatwssienegn owf itlhde avnardiabfailritmy eodf tshpee cwihmoelen sseitn ofa utumn
(Figures2p)e,cwialiitzhedh migehtaebrocliotensc pernotdruacteiodn byo fCt. hcreamtobxe iwcicthr abmotbh etshcei mdiinnofra amndil ythien mwajiolrd mseptoabnoglietess(. FIti gisu re3A).
Thus, aincotemrebstiinnga ttioo nnotoef thinatt enronnea loff athcet omrasjoarc stpinecgiatloizgedet mheetrabwoilitthesb bieolotincg ftaoc tthoer fsivseu mcohsta csoncotrmibupteivteit ion for
space [1,16,18,19] may also explain the seasonal variability of the production of the specialized

metabolites. Abioticfactorssuchastemperatureornutrientavailabilitymighttriggerotherprocesses
likespongereproductionorotherconstraintsofprimarymetabolism[25,58,59],ratherthandirectly
affectingthespecializedmetabolome.
The metabolomics approach allowed encompassing of the variability of the whole set of
specializedmetabolitesproducedbyC.crambewithboththeminorandthemajormetabolites. Itis