The Multifaceted Inhibitory Effects of an Alkylquinolone on the Diatom Phaeodactylum tricornutum

Abstract The mechanisms underlying interactions between diatoms and bacteria are crucial to understand diatom behaviour and proliferation, and can result in far‐reaching ecological consequences. Recently, 2‐alkyl‐4‐quinolones have been isolated from marine bacteria, both of which (the bacterium and isolated chemical) inhibited growth of microalgae, suggesting these compounds could mediate diatom–bacteria interactions. The effects of several quinolones on three diatom species have been investigated. The growth of all three was inhibited, with half‐maximal inhibitory concentrations reaching the sub‐micromolar range. By using multiple techniques, dual inhibition mechanisms were uncovered for 2‐heptyl‐4‐quinolone (HHQ) in Phaeodactylum tricornutum. Firstly, photosynthetic electron transport was obstructed, primarily through inhibition of the cytochrome b 6 f complex. Secondly, respiration was inhibited, leading to repression of ATP supply to plastids from mitochondria through organelle energy coupling. These data clearly show how HHQ could modulate diatom proliferation in marine environments.


Introduction
Diatoms are ac lass of unicellular algae found worldwide in aquatic environments, in which they are one of the chief primary producers. [1] They persist in both benthica nd pelagic habitats anda re surrounded by ad iverse array of other microbes, in particular, bacteria. [2] Although aw ide range of diatom-bacteria interactions have been identified, characterisation of the molecules and the corresponding modes of action that drive these interactions remainss carce.H owever,aclass of bacterial secondary metabolites, 2-alkyl-4-quinolones, some of which are used by bacteria as quorum sensing (QS) signals, [3] have been identified as possessinga lgicidal effects on a varietyo fm icroalgae. [4] Although not as ubiquitous as other QS compounds, such as N-acyl homoserinel actones( AHLs), previouss tudies have reported that marineb acteria, particularly Pseudoalteromonas and Alteromonas species, [5] but also freshwater and soil bacteria of the genera Pseudomonas and Burkholderia,produce quinolones. [6] Quinolones reported most frequently from marineb acteria are 2-pentyl-4-quinolone (PHQ) [5c] and the closely related 2heptyl-4-quinolone (HHQ). [4,5,7] Ar ecent study by Harvey et al. demonstrated that the marine bacterium Pseudoalteromonas piscicida was toxic to the microalga Emiliania huxleyi,d ue to the excretion of HHQ, with ah alf-maximal growth inhibitory concentration (IC 50 )i nt he nanomolar range. [5a] In addition, two quinolones (2-undecen-1'-yl-4-quinolone and 2-undecyl-4-quinolone) were detected in extracts of another Alteromonas strain (KNS-16), both of which inhibited the growth of ar ange of microalgae, with IC 50 values varying between 1.6 and 200 mm,d ependingo nt he alga. [8] Remarkably,t he growth of algae in both of these studies was not only inhibited by living P. piscicida or Alteromonas KNS-16 cells, but also if treatedw ith the respective isolated quinolone. Furthermore, Alteromonas KNS-16 was isolatedd irectly from an algal bloom. [8] This informations uggests that 2-alkyl-4-quinolones mediateb acteriaalgae interactions through growth inhibition, and indeed may accumulate in the diffusion boundary that surroundsm icroalgae and associatedb acteria, reaching locally high concentrations. [2a, 9] The mechanismsu nderlying interactions between diatoms and bacteria are crucial to understandd iatom behaviour and proliferation, and can result in far-reaching ecological consequences. Recently,2 -alkyl-4-quinolones have been isolated from marine bacteria,b oth of which (the bacterium and isolated chemical) inhibited growth of microalgae, suggesting these compounds could mediate diatom-bacteria interactions. The effects of several quinolones on three diatoms peciesh ave been investigated. The growth of all three was inhibited, with half-maximal inhibitory concentrations reachingt he sub-micromolar range. By using multiple techniques, dual inhibition mechanisms were uncovered for 2-heptyl-4-quinolone (HHQ) in Phaeodactylum tricornutum. Firstly,p hotosynthetic electron transport was obstructed, primarilyt hrough inhibition of the cytochrome b 6 f complex. Secondly, respiration was inhibited, leadingt or epression of ATPs upply to plastidsf rom mitochondria through organellee nergy coupling. These data clearly show how HHQ could modulate diatom proliferation in marine environments.
Our understanding of the physiological effects of quinolones on diatoms remains scarce.I nr ecent works treating diatoms with quinolones, it was shown that PHQ inhibited growth in Cylindrotheca fusiformis, Thalassiosiraw eissflogii and natural phytoplankton assemblages. [5c] The same compound was also found to inhibitg rowth and/or motilityo ft he benthicd iatoms Amphora coffeaeformis, Navicula sp. and Auricula sp. [10] However,v ery few other alkylquinolones have been tested on diatoms, despite the diversity of alkylquinolones produced by marine bacteria. Furthermore, it is still not clear what causes quinolones to inhibitthe growth of microalgae at all.
Nevertheless,t he effects of alkylquinolones have been studied on other organismsp reviously. [11] For instance, studies by Reil et al. found that synthetic quinolones, including 2-alkyl-4quinolones, were inhibitors of complex I( NADH:ubiquinoneoxidoreductase;N ADH:r educed nicotinamide adenine dinucleotide) and complex III (cytochrome bc1 complex) in mitochondria. [11b] In addition, the same authors tested synthetic quinolones on isolated spinach thylakoids and reported that 2-alkyl-4-quinolone N-oxides were strong inhibitors of photosystem II (PSII), whereas 2-alkyl-4-quinolones (such as HHQ or PHQ) were only weak inhibitors;t his suggestst hat 2-alkyl-4-quinolone Noxides are more potent inhibitors of photosynthesis than that of the corresponding 4(1H)-quinolones in vascular plants. [11c] In addition, both compound groups were only weaki nhibitors of the cytochrome b 6 f complex. [11c] Furthermore, the alkylquinolones HHQ and PQSh ave been tested on ar ange of bacteria and yeastsu pon which they had distinct effects on cell proliferation, motility and biofilm formation,a nd thus, indicating that their effects are speciesspecific. [11a] The observations that quinolones produced by marine bacteria can inhibit growth in certain microalgae has prompted us to investigate their effects on diatomsi ndetail. We aimed to study how structural analogues would affect diatomg rowth, and whetheramode of action could be observed with diatoms in vivo. Herein, we present work regarding an umber of native bacterial quinolones, namely,2 -heptyl-4-quinolone Noxide (HQNO), as well as the Pseudomonas QS signals HHQa nd 2-heptyl-3-hydroxy-4-quinolone (PQS), and the 2-nonyl congeners 2-nonyl-4-quinolone (NHQ) and 2-nonyl-4-quinolone Noxide (NQNO). To account for the different environments in which quinolones have been detected, we selected three diatoms from different aquatic ecosystemsf or our initial screening: Cylindrotheca closterium is am arine biofilm-forming diatom often found in the benthoso ft he intertidal zone. In contrast, Phaeodactylum tricornutum is ap lanktonic diatom, which was originally isolated in coastal water and is commonly used as am odel organism. [12] Additionally, Achnanthidium minutissimum represents ab iofilm-forming freshwater diatom.B ecause Reil et al. identified an inhibition of photosynthesisb y certain quinolones in isolated spinacht hylakoids, as well as an inhibition of respiration in isolated mitochondria of non-photosynthetic organisms, [11c] we used av ariety of experiments to probe not only photosynthesis, but also respiration. In doing so,w ed emonstrate in vivo howt he reported growth impairment of diatomsb yq uinolonesi sa chievedt hrough as imultaneouss pecific inhibition of both photosynthesis andr espiration.

Inhibition of diatom growth by quinolones
Intrigued by the reported bioactivity of quinolones on microalgae, we trackedt he growth of three diatoms treated with quinolones at ar ange of concentrations (0.125-100 mm). The quinolones used varied with regard to their N-oxidation, 2alkyl chain length and 3-hydroxyl group (Figure1). Five quinolones, HHQ, NHQ, PQS, HQNO and NQNO, were appliedt oc ultures of P. tricornutum, C. closterium and A. minutissimum. Of the three quinolones tested with ah eptyls ide chain (HHQ, PQS and HQNO), all diatomsw ere mosts ensitive towards HHQ, with IC 50 valuesb etween 1a nd 5 mm for each respective species (Figure 2A,D ,G ). HHQ completelyi nhibited growth at concentrations as low as 3 mm in C. closterium (Figure 2A)a nd A. minutissimum ( Figure 2D), and at 10 mm in P. tricornutum ( Figure 2G). Relativet oH HQ, the IC 50 values of PQS were foundt ob e3to 16 times higher ( Figure 2B,E ,H ). A. minutissimum and P. tricornutum were less sensitive to the N-oxide HQNO compared to that of PQS ( Figure 2F and I), whereas C. closterium was more sensitive ( Figure 2C).
Diatoms were more sensitivet ob oth NHQ and NQNO (Figure S1 in the Supporting Information) compared with that of their hept-2-yl counterparts (HHQ and HQNO), which indicated that al ongera lkyl side-chain length increased the toxic effect of these compounds. This observation substantiates aq uantitative structure-activity relationship study to test quinolones on spinach thylakoids, which found that quinolones reached their maximum photosynthetic inhibitory potentiala ta na lkyl chain length of 11 carbon atoms. [11c] However,u nlike experimentsw ith spinach thylakoids, diatoms displayed decreased sensitivity towards N-oxide quinolones:j ust as HHQ was more potent than that of its N-oxide counterpart, so too was NHQ more potent than that of NQNO.
Ta ken together,t hese observations showedt hat structural features of quinolones influencedt he respective growth response of diatoms; N-oxide-or hydroxyl-functionalised quinolones were less potent than that of non-functionalised quinolones.T his trend was consistent among all diatoms tested, although sensitivity varied between species, of which C. closterium was the mosts ensitive, followed by A. minutissimum and P. tricornutum.

Blockingofe lectron flow between PSII and PSI by HHQ
The strong inhibitory effect of HHQ on all three diatom species, and its relevance towards other interactions between microalgae and bacteria, prompted us to investigate the mode of action of HHQ in P. tricornutum in detail. To this end, we first investigated the effect of HHQ on the activities of photosystems II andI(PSII and PSI, respectively)b ym easuring the fluorescence induction of PSII and absorbance changes of P 700 (the special pair of chlorophyll in the reactionc entre of PSI) simultaneously,u sing dual pulse amplitudem odulated (Dual-PAM) fluorometry.F luorescencec omesm ainly from PSII and its intensity depends on the redox state of Q A -the primary quinone electron acceptoro fPSII-which takes up electrons originating from the water-splitting reaction at the oxygen-evolving complex. Meanwhile, changes in absorbance (l = 875 nm minus that at l = 830 nm) provide information about the redox state of P 700 within PSI. [13] Using thesep arameters, the PSII and PSI kinetics were measured in dark-adapted and low-lightadaptedc ells. HHQ was tested at ar ange of concentrations;a s an example, Dual-PAMd ata at 5 mm (the IC 50 value for P. tricornutum)a re showni nF igure3 (the quantum yield, Y(II), from the other concentrationst ested are shown in Figure S2). DMSO and DCMU (a potent PSII inhibitor), serveda sr espective nega-tive and positive controls ( Figure 3). Typically,t he change in absorbance of P 700 (Figure 3, grey line) consists of three phases duringt he saturating pulse, in both dark-and light-adapted control cells:afast increase in absorbance in the first 30 ms, which is indicative of photo-oxidation of PSI;f ollowed by a partial decrease in absorbance between 30 and 200 ms, which indicates ar eduction of PSI;a gain followed by an increase in absorbance, showingt he re-oxidationo fP SI (> 200 ms). In the presence of HHQ, the transientr eduction of P 700 between 30 and 200 ms was still present in dark-adapted cells, but was suppressed in light-adapted cells. This reduction transient has previously been assignedt oe lectron flow from PSII in P. tricornutum, [14] which is in agreement with its suppressioni nt he presence of DCMU in ourd ata ( Figure 3). In this regard, the absence of aP 700 reduction transientinlight-adapted cells treated with HHQ suggests that these molecules inhibita ne lectrontransfer step between PSII and PSI. However,i nt he DCMU treatment, the transientr eductiono fP SI was also abolished in dark-adapted cells, which suggested that HHQ had ad ifferent mode of action to DCMU. This observation was supported by the maximum quantum yield of PSII, which wasnot significantly affected in dark-adapted cells by HHQt reatment (Figure S2), whereas DCMU treatment induced ac lear decrease in PSII quantum yield. To confirmt his, and to identify the molecular target of HHQ in the electron-transfer chain of P. tricornutum,w eu sed a Joliot-type spectrometer (JTS) and saturatingc oncentrationso f HHQ (50 mm,s ee Figure S2). Firstly,w em easured the light dependency of the quantumy ields of PSII and PSI (Y(II) and Y(I), respectively), in the presence and absence of HHQ ( Figure 4). Again, the maximal yields of PSI and PSII were not significantly inhibited by HHQ in the dark ( Figure 4A,B ). However,a ta ll light irradiances, the quantum yields of both photosystems were significantly lower in the presence of HHQ relative to that of the control ( Figure 4A,B ). This translates into al ower electron-transfer rate through both PSII and PSI in the presence of HHQ, regardless of the irradiance ( Figure S3). Under the same conditions, we also measured the acceptor-and donor-side limitations of PSI (see the Experimental Section). The data clearly show that the decrease of Y(I) is paralleled by an increase of the donor-side limitation (Y(ND)), that is, P 700 is more oxidisedi nH HQ-treated samples ( Figure 4C). In addition, PSIi s not limited by the acceptor side because the proportion of non-photo-oxidisableP 700 (Y(NA)) is almostn on-existent, regardlesso fl ight irradiance ( Figure 4D). These data rule out the possibility that HHQ inhibits the PSI acceptors ite or beyond (e.g.,t he ferredoxin NADP reductaseo rt he Calvin-Benson-Bassham cycle). The higherf ractiono fo xidised P 700 reveals a limitation of the electron flow "uphill" in PSI, confirming that the site of inhibition of the photosynthetic electron-transfer chain takes place between PSII and PSI.
Effecto fH HQ on the activity of cytochrome b 6 f To probe the exact target of HHQ in the photosynthetic apparatus of P. tricornutum, three complementary approaches were  Fast fluorescencet ransients are finely time-resolved measurements of chlorophyll fluorescenced uring as aturating multi-turnover pulse that provide specific information about processesi nP SII, but also beyond. [15] Information is derived from stepwise transitions in fluorescence levels, referred to as the Ja nd Is teps at 2a nd 30 ms, respectively,a nd the Ps tep, which describes the time pointw hen maximum fluorescence (F m )i sr eached ( Figure 5). Whereas DMSO-treatedc ontrol cells showedt ypical shoulders at the Ja nd Is teps of the fluorescence transient, DCMU-treated cells reached F m at the Js tep, which was atypical indicator for PSII inhibition ( Figure 5). However,t reatment with 5 mm HHQ, the half-inhibitory concentration of HHQ, only induced as mall increasea tt he Js tep (2 ms) of the fluorescencet ransient. This indicated that the primary target of HHQ was not PSII, and thus, led to the hypothesis that the target of HHQw as likely to be downhill of the plastoquinone pool. [15,16] In parallel, we analysed the ECS of photosynthetic pigments, [17] which was the change in the absorption spectra of some photosynthetic pigments due to the electric field generated across the thylakoid by the photosynthetic process. The ECS, which can be seen as an in vivo voltmeter,i sapowerful and widespread technique to investigate photosynthetic physiology.I nP. tricornutum,s imilar to that in other diatomsa nd stramenopiles, [18] the ECS is the sum of al inear electric field strength (proportional to the electric field across the thylakoid) and aq uadratic component (proportionalt othe square of the electric field strength). The kinetics of the linear ECS following as aturating laser flash can be used to evaluate the activity of each photosynthetic complex( PSI, PSII, cytochrome b 6 f, ATPase). In theory,t hese kinetics possesst hree distinct phases: 1) af ast rise of the electric field, representing charge separation due to PSI and PSII activity (< 0.1 ms); [19] 2) as econd rise, corresponding to the turnovero fc ytochrome b 6 f ( % 10 ms), which pumps additional protons into the lumen;a nd 3) ar elaxation of the electric field as ATPase consumes protons from the lumen to the stroma andc onvertsA DP into ATP( > 10 ms). [19] We also measured the redox state of c-type cytochromes, comprising cytochrome f in cytochrome b 6 f and cytochrome c 6 ,w hichs huttles electrons between cytochrome b 6 f and PSI (see the Experimental Section).
As mentioned above,t he first phase of ECS kinetics represents charge separation by PSI and PSII immediately after the absorption of ap hoton.I no ur measurements, this charge-separation effect was decreased by 30 %i nt he presence of HHQ (represented by the first data point after illumination (= 0) in Figure 6A). The charge separations in PSII and PSIl ead to electron transfer from water to plastoquinones, andf rom c-type cytochromes to ferredoxins, respectively.A ccordingly,t his fast rise of ECS is concomitant with the oxidation of c-type cytochromes by PSI ( Figure 6B), which is similarw ith and without HHQ, and thus, indicates that PSI photochemistry and electron transfer from cytochromes to P 700 is unaffected. The 30 %d ecreaseo ft he ECS fast rise could reflect as light decrease of PSII activity in the presence of HHQ;h owever,t his cannot explain the almostc omplete inhibition of photosynthetic activity in the light.
After this fast phase, generating reduced quinols and oxidised c-type cytochromes, the turnover of the cytochrome b 6 f complex catalyses the transfer of electrons from the reduced  quinolst ot he oxidised c-type cytochromes. This process is coupled to protonpumping across the thylakoid. Theoutcome is ap hase of reduction of the c-type cytochromes( Figure 6B) and as econd rise in the trans-thylakoid electric field and ECS ( Figure 6A). However,i nt he presence of HHQ, only as mall increaseo ft he electric field in this time frame was observed, and the reductiono fc-type cytochromes was 20-fold slower. This observation identifies the cytochrome b 6 f as the main target of HHQ, which explains the overall inhibition of the photosynthetic electron-transfer rate.

Suppression of thylakoidprotonmotive force in darkadapted cells by HHQ
The last phase of ECS measurements (> 10 ms;F igure 6) shows the decay of the electric field and corresponds to the movement of protons from lumen to stroma, as catalysed by ATP synthase.T his decay was retarded by HHQ ( Figure 6A). HHQ treatment also decreasedt he amplitude of the quadratic ECS contribution by about 90 %( i.e.,E CS proportional to the square of electric field strength, as shown in Figure 6C). These two observations hinted at as econd effect of HHQ:s uppression of the pre-existinge lectric field in the dark (DY d ), as observed in Bailleul et al. [18a] To quantify the possible effect of HHQ on the electric field across the thylakoids in dark-adapted diatoms, we measured the kinetics of the relaxation of the linear and quadratic ECS generated after as aturating pulse of light in untreated P. tricornutum,a sw ell as upon treatment with HHQ (50 mm). We also used the membrane potential un-coupler carbonyl cyanide m-chlorophenyl hydrazone (CCCP) as ac ontrol to artificially suppress DY d . [18a] The amplitude of the quadratic versus linear ECS signals were plotted ( Figure 7A-C), giving the same characteristicp arabolic function for all treatments, with the vertex indicating the electric field strength in the dark, precedingt he light perturbation (expressed in number of charge separations per photosystem;F igure 7D). The datai ndicated that ap rotonm otivef orce (PMF) was maintained across the thylakoid membrane of untreated P. tricornutum cells in the dark, with a DY d correspondingt o5 .0 AE 0.6 charge separations per photosystem, similar to previously measured values. [18a] However,t he electric fieldi nt he dark was clearly suppressed in the presence of HHQ (DY d = 1.4 AE 0.3 charge separations/photosystem), almost as much as with the uncoupler CCCP (DY d = 0.8 AE 0.2 charge separations/photosystem).
In the presence of cellular ATPi nt he plastid in the dark (which comes from mitochondrial respiration activity),p lastidic ATPase is able to hydrolyse ATPt oA DP,w hich generatesa PMF across the thylakoid membrane. [20] It is well known in plants, green algae and diatoms that the inhibition of respiratory activity (with uncouplers, mitochondrial inhibitors or under anaerobic conditions) leads to ad ecreaseo ft he PMF acrosst he thylakoid. [15,20] Thus,H HQ could suppress the PMF by two means: through an uncoupler effect or through an inhibition of respiration. For that reason,w em easured the effect of HHQ on the respiratory activity of P. tricornutum.
Respiration rates of P. tricornutum were measuredint he dark and derived before and after the addition of HHQ by using a Clark electrode. Respiration rates were more than halved after HHQ addition compared with the respiration rates before the addition of the compound (Figure 8). Accordingly,t he observed decrease in electric field strength in the dark was attributed to an inhibition of respiration, rather than through uncoupling of thylakoid charges eparation.

Discussion
Ad iverse array of interactions between diatomsa nd bacteria have been documented;h owever,t he physiological mechanisms underlying these interactions are rarely characterised. This study demonstrated the growth inhibitory effects of five different2 -alkyl-4-quinolones on three diatom species, and identified am ode of action for HHQ in P. tricornutum. Of all five tested quinolones (Figure 2), HHQ and NHQ hadl ower IC 50 concentrations than those of their functionalised homologues (HHQ = 1.2-4.9 mm;N HQ = 0.16-1.38 mm), which illustrates that in diatoms non-functionalised quinolones are more potent than that of their functionalised analogues. Although previous studies have shownt he toxic effects of HHQ on the coccolithophore E. huxleyi [5a] and of PHQ on other diatoms, [5c, 10] this study builds on previousf indings by testing ab roaderr ange of quinolonesw ith structural variations. These experiments not only show that diatom growth is inhibited by aw ide range of quinolones, but also that some appear to be considerably more potent than that of PHQ. Interestingly,a lkylquinolones were more active than that of their corresponding N-oxide derivatives by approximately an order of magnitude, whereas in bacteria-bacteria interactions the N-oxides were much more potent than that of the corresponding non-functionalised quinolones. [21] The effects of HHQ were characterised exhaustively by using aw ide variety of spectroscopict echniques. These experiments indicatet hat HHQ inhibits both cytochrome b 6 f and, to al esser extent,P SII in plastids, ando xygen consumption and ATP production in mitochondria. Under actinic light, Y(II) values ( Figure S2) decreased andt he transientr eductiono fP 700 was absent ( Figure 3);t his provides evidencet hat electron transport in thylakoids is inhibited. This phenotype was clearly visible with 5 mm treatments of HHQ, which was also the IC 50 value derived from growth experiments and suggested that the inhibition of photosynthesis was the chief mode of growth inhibition. These resultsw ere confirmed at all light irradiances under steady-state illumination (Figure 4), showingt hat the quantum yield andr elativee lectron-transport rates of PSI and PSII werei mpaired by HHQ. The observedi nhibition of PSI activity was due to ah ighero xidation of P 700 and not acceptorside limitations, whichi ndicated that electron transfer was hampered between PSI andP SII.
Subsequently,t he binding site of HHQ was identified as the cytochrome b 6 f complex,t he activity of which was slowed down 20-fold ( Figure 6). These experimentsa lso showed that charges eparation due to the activity of PSII was slightly decreased. These observations are supported by the fluorescence transients( showni nF igure 5), which show as lightly higher amplitude of the O-J phase, indicating an inhibition of electron transfer towards the plastoquinone pool. These data give conclusive evidence that HHQ hinders photosynthesis primarily throught he inhibition of cytochrome b 6 f and, to al esser extent, PSII (most likely at the Q B site). Such ar esult is reinforcedb ys tructurals imilarities between HHQ and plastoquinone/plastoquinol (the mobile electron carrier between PSII and cytochrome b 6 f), along with other structurally related moleculesw ith similar inhibitory effects, such as stigmatellin and aurachines. [22] HHQ also inhibited mitochondrial respiration, as demonstrated by oxygen consumption experiments (Figure 8). The inhibition of respiration leads to as upplementary phenotype. That is, ATPp roduced by mitochondrial respiration in the dark can be hydrolysed by the chloroplastic ATPase, working "in reverse" andp umpingp rotons into the lumen. Because of this, diatoms, similar to other photosynthetic organisms, generate aP MF across the thylakoidi nt he dark. [18a, 20] Here, the inhibition of mitochondrial respiration by HHQ, and, in turn, the dark PMF (Figure 7), can be visualised by the slower ATPase activity and lower quadraticE CS following as aturating laser flash ( Figure 6C). Complete inhibition of respiration was not achieved, nor was as pecifics ite of inhibition identified. It is nevertheless plausible that HHQ inhibits respiration by hindering electron transport at complex Io rI II, between which ubiquinone shuttlese lectrons, similart ot he PSII-plastoquinone-cytochrome b 6 f system in plastids. Indeed, quinolones have been identified as inhibitors of these complexes in other organisms. [11b, 23] Whereas previouss tudies have shown the effect of 2-alkyl-4-quinolones on respiration in prokaryotesa nd nonphotosynthetic eukaryotes,t his study providese vidence of their inhibition in photosynthetic eukaryotes, and shows that these compounds can simultaneously hindert he functiono f both photosynthesis and respiration.
Previous studies, mostly on vascular plants, have demonstratedt he inhibition of photosynthesis by NQNO, which is knownt ob ind to cytochrome b 6 f. [24] In comparison, much less is known regarding other quinolones.T his work suggests that HHQ is am ore potent inhibitor of photosynthesis in diatoms than that of NQNO and mayb eam ore favourable molecule for future diatom photo-physiology studies. For example, a screening of quinolones by Reil et al. with spinach thylakoids showedt hat non-functionalised alkylquinolones( e.g.,H HQ and NHQ) had only very weak effects on PSII and cytochrome b 6 f activity. [11c] In contrast, this study demonstrates that HHQ strongly inhibits oxygen evolution ( Figures S5 and S6), mainly due to the inhibition of cytochrome b 6 f. However,i ts hould be noted that many previous studies utilised thylakoid preparations to test quinolones. This presentsaproblem when comparing them to experiments on whole cells, in which diffusion across membranes and accumulation withincell compartments may influence the effect of quinolones. This is particularly important when considering the unique plastid, thylakoid and cell wall architectures of diatoms. [25] Nevertheless, physiological differences between diatoms and other photosynthetic organisms appear to define the activity of 2-alkyl-4-quinolones. Such ah ypothesis is supported by fluorescencet ransients from two other microalgae treated with HHQ ( Figure S4). In the coccolithophore E. huxleyi,f or example, treatment with 25 mm HHQ led to an increase of the fluorescence transient at the Ja nd I steps, which could indicate inhibition at PSII and the cytochrome b 6 f complex.I nc ontrast, HHQ treatment of the green alga Dunaliella tertiolecta only induceda ni ncrease of the fluorescence transient at the Js tep, whichs uggestedt hat the effect was primarily related to PSII. Indeed,t he effecti nD. tertiolecta was nearly identical to the effect inducedb yn on-saturating treatments of DCMU.T aken together,t hese resultsi llustrate how differentp hotosynthetic organismsr espond to HHQ in diverse manners, ands uggest that the potent effect of HHQ on E. huxleyi observed by Harvey et al. was due to the inhibition of photosynthesis. [5a] Conclusion This study adds detailed physiological data underlying the strong growth inhibitory effect of 2-alkyl-4-quinolones on diatoms, expanding the diverser epertoire of their bioactivity.F urthermore, this study builds on the increasinge vidence that 2alkylquinolones from bacteria have major roles beyondt hat of QS as importantm ediators of interspecies and even interkingdom interactions. This work parallels those on AHLs, which have also been shown to mediate interkingdomi nteractions in marinee nvironments. For example, AHLs mediate the settling of zoospores in the green macroalga Ulva, [26] whereas tetramic acids, spontaneously generated from certain AHLs, have also been shown to impair photosynthesis in diatoms. [27] All tested quinolones inhibited photosynthesis, while detailed physiological experiments identified cytochrome b 6 f as the chief binding site of HHQ, along with the less severe inhibition of PSII ( Figure 9). With the isolation of quinolone-producing bacteria from marine sourcesinprior studies, [5a, c, 7, 8] this study highlights how HHQ could modulate respiratory and photosynthetic activity (Figure 9), and subsequently the proliferation of diatoms in marine environments.

Experimental Section
Diatom strains and culture conditions: P. tricornutum ("wild-type 8", NEPCC 640) was obtained from the Canadian Centre for the Culture of Microorganisms (CCCM, http://cccm.botany.ubc.ca). Another strain of P. tricornutum ("wild-type 1" Pt 1.8.6) was used for the identification of the site of inhibition of HHQ because the deconvolution of the ECS and c-type cytochromes signals (see below) were previously made on this strain, and we could not rule out that this deconvolution procedure would be as correct in wild-type 8. A. minutissimum (Kützing) Czarnecki was isolated from epilithic biofilms of Lake Constance, Germany. C. closterium strain WS3_7 (DCG 0623) was obtained from the Belgian Coordinated Collections of Microorganisms (BCCM, http://bccm.belspo.be). Prior to the experiments, P. tricornutum and C. closterium cultures were made axenic by treating them with an antibiotic mix (500 mgmL À1 penicillin, 500 mgmL À1 ampicillin, 100 mgmL À1 streptomycin and 50 mgmL À1 gentamicin) for aw eek and replacing the antibioticsupplemented medium every second day. A. minutissimum was axenified according to ap rocedure reported by Windler et al., [28] and was cultured in am odified bacillariophycean medium, [29] which instead of soil extract contained F/2 multivitamins, trace metal and silicon/selenium nutrients. P. tricornutum and C. closterium were cultured in Artificial Seawater Medium (ASW; 34.5 gL À1 Tropic Marin, 0.08 gL À1 NaHCO 3 )s upplemented with Guillard'sF /2 (Sigma-Aldrich). [30] Due to different culture facilities, diatoms were incubated at 18 8Ci na12:12 hl ight/dark regime at 25 mmol photons m À2 s À1 for growth experiments, and at 20 8C, with a1 6:8h light/dark regime with al ight intensity of 70 mmol photons m À2 s À1 for PAMf luorometry and Clark electrode experiments. For JTS-10 experiments, P. tricornutum was incubated at 20 8Cw ith a1 2:12 hl ight/dark regime with alight intensity of 70 mmol photons m À2 s À1 .
Cell counts were conducted by using aM ultisizer 4e Coulter Counter (Beckmann Coulter). Samples collected for chlorophyll content determination (3 mL of culture) were centrifuged (4500 g,5min) and the pellet was extracted with methanol (100 mL) and vortexed, followed by acetone (900 mL). The resulting suspension was vortexed and centrifuged once more (18 000 g,2min). The chlorophyll content of the resulting supernatant was determined according to ap rocedure by Jeffrey and Humphrey, [31] in quartz cuvettes on an Ultrospec 2100 pro UV/Vis spectrophotometer (Biochrom).
Preparation of quinolone solutions:T he 2-alkyl-4-quinolones (HHQ and NHQ) and 2-alkyl-4-quinolone-N-oxides (HQNO and NQNO) were synthesised as described previously. [21] Briefly,c orresponding 3-oxoalkanoic acid methyl esters generated from acyl chlorides with Meldrum's acid were used in condensation reactions with aniline to lead to methyl-3-phenylamino-2-enoates that were subsequently subjected to Conrad-Limpach cyclisation to give the 2-alkyl-4-quinolones. For the preparation of HQNO and NQNO, HHQ and NHQ were converted into their hydroxyquinoline tautomers as ethyl carbonates that were used in the subsequent N-oxidations. [32] The resulting ethyl carbonate N-oxides were deprotected to yield the corresponding HQNO and NQNO. PQS was prepared according to ap rocedure reported by Hradil et al. by using anthranilic acid that was converted into 2-oxononyl 2'-aminobenzoate and consequently cyclised in N-methyl-2-pyrrolidone (NMP) at 250 8Ct oPQS. [33] HHQ, NHQ, PQS, HQNO and NQNO were prepared in DMSO (Sigma-Aldrich) such that the final volume of DMSO added to the cultures or samples was always 0.5 %( v/v).
Growth assays:A tt he start of the experiment, axenic cultures of A. minutissimum and P. tricornutum were adjusted to an appropriate cell density by using aM ultisizer Coulter Counter.I nt he case of C. closterium,afixed minimum fluorescence was used with a PAMf luorometer (Walz, Germany), due to the cells rapid sinking rate. Growth assays were conducted in 48-well plates (Greiner Cellstar,S igma-Aldrich) and tracked for 6days following compound addition. Growth was measured by recording chlorophyll autofluorescence, which was normalised to ab lank well (filled with ASW), on aC ytation 5C ell Imaging Multi-Mode Reader (l ex = 425 nm/l em = 685 nm, Biotek). The treatments were randomised among well positions, with three replicates per treatment. The entire growth assay was repeated once more with similar trends. Growth rates were calculated as the logarithmic ratio of cell densities divided by the time interval. The exponential growth phase was thus identified, from which IC 50 values were derived by using the online AATBioquest tool. [34] Dual-PAM experiments:D ual-PAM experiments were performed by using aW alz Dual-PAM 100 fluorometer in dual-channel mode, equipped with aD ual-E and aD ual-DB detector and ac uvette holder with stirrer.C ultures of P. tricornutum in the exponential phase were concentrated to ac hlorophyll a concentration of 40 mgmL À1 ,s upplemented with sodium bicarbonate to prevent carbon limitation (16 mm), and adjusted to pH 8.0. For each test, this prepared suspension (2 mL) was treated with aq uinolone stock solution or DMSO control (for af inal DMSO concentration of 0.5 %, v/v)a nd incubated in very low light (resting in the cuvette holder,e quivalent to no higher than 10 mmol photons m À2 s À1 at the surface) for 2min with stirring, after which the holder was closed. After 10 si nt he dark, each sample was exposed to one saturating multi-turnover pulse (intensity 8000 mmol photons m À2 s À1 , width 800 ms), then low actinic red light was switched on (68 mmol photons m À2 s À1 )f ollowed by as aturating pulse after 30 s. Stirring was switched off immediately before each pulse and restarted immediately afterwards. PSII fluorescence and P 700 absorbance (l = 875 nm minus that at l = 830 nm) were recorded in such am anner for each quinolone for each concentration in at least two biological replicates. The initial P 700 absorbance values were set to equal zero. The PSII quantum yield was also calculated from each of these measurements, by using the same method as that used in JTS fluorescence experiments (see below). JTS-10 experiments:P hotosynthetic parameters of P. tricornutum were measured with aJ TS (JTS-10, Biologic, Grenoble, France) equipped with aw hite probing light-emitting diode (LED;L uxeon; Lumileds) and as et of interference filters (3-8 nm bandwidth) and cut-off filters. The device combines absorbance and fluorescence spectroscopy measurements, allowing the activities of PSI and PSII to be studied under the exact same conditions and on the same sample. The actinic light was provided by ac rown of red LEDs (l = 639 nm, intensities used in this study:5 6, 135, 340, 800, and 1500 mmol photons m À2 s À1 ). For photosynthesis measurements, cultures of P. tricornutum in exponential growth phase were concentrated tenfold by centrifugation (4500 rpm, 4min) and resuspended in its own supernatant to reach af inal concentration in the range of 5-10 10 6 cells mL À1 .T he centrifuged samples were then left for about 30 min under low light to allow the cells to recover from centrifugation.
Fluorescence spectroscopy:I nf luorescence spectroscopy mode, the JTS-10 was equipped with aw hite probing LED (Luxeon;L umileds) and ab lue filter for detecting pulses. PSII parameters were calculated as reported by Genty et al. [35] In brief, the maximum quantum yield of PSII and quantum yields in light-adapted samples were calculated as F v /F m = (F m ÀF 0 )/F m and Y(II) = (F m 'ÀF)/F m '', respectively,i nw hich F 0 is the fluorescence of the dark-adapted sample, F m is the fluorescence when as aturating pulse is applied on ad ark-adapted sample, F is the fluorescence of the sample adapted to the actinic light and F m '' is the fluorescence when a saturating pulse is applied on light-adapted sample. The relative electron-transport rate through PSII (rETR PSII )w as calculated as rET-R PSII = Y(II) I I,i nw hich I is the actinic light irradiance, and then values were all normalised to the value at 1500 mmol photons m À2 s À1 .
Absorption spectroscopy: P 700 measurements:T he redox state of the PSI primary donor (P 700 )w as calculated as the difference between the kinetic absorption changes at l = 705 and 735 nm, to eliminate spectrally flat contributions due to diffusion. Quantum yields of PSI were calculated from the measurements of the absorption changes between l = 700 and 735 nm, in the dark (P 0 ), in the light-adapted condition (P stat )a nd after as aturating pulse (P sp ). P max ,c orresponding to 100 %o xidised P 700 ,w as measured as the light minus dark absorption difference in the presence of the PSII inhibitor DCMU (10 mm in ethanol, final concentration of 10 mm). Once normalised to P max ,t his allowed the percentage of oxidised P 700 to be calculated in each light condition.
The quantum yields of PSI (Y(I)) and the donor-(Y(ND)) and acceptor-side (Y(NA)) limitations were calculated according to ap rocedure reported by Klughammer and Schreiber,[13a] as Y(I) = (P sp ÀP stat )/ (P max ÀP 0 ), Y(ND) = (P stat ÀP 0 )/(P max ÀP 0 )a nd Y(NA) = (P max ÀP sp )/ (P max ÀP 0 ). The relative electron transport rate through PSI (rETR PSI ) was calculated as rETR PSI = Y(I) I,i nw hich I is the actinic light irradiance and then values were all normalised to the value at 1500 mmol photons m À2 s À1 .
ECS and c-type cytochromes measurements:B ased on previous P. tricornutum ECS spectra, [18a] we measured the absorption changes (DI/I;t he relative difference of intensity between sample and reference photodiodes) at three wavelengths (l = 520, 554, 566 nm). To separate ECS and c-type cytochromes (cytochrome f and cytochrome c 6 )c ontributions, and to eliminate flat contributions due to diffusion, we used the following equations:c ytochrome c = [554]À0. 4  To follow the kinetics of those signals following as aturating laser flash, we used al aser dye (LDS 698) pumped by af requency-doubled Nd-YAG laser (Quantel). Before the saturating laser flash was applied, cells were dark adapted for 1min. For ab etter comparison, all data in Figure 6w ere normalised to the linear ECS value measured immediately after the flash in the control. The photochemical event ("a phase" of ECS kinetics) finished well before 100 ms [19] and the first experimental point was measured 150 ms after the laser flash.
The dark-adapted electric field (DY d )m easurements were attained from the dark relaxation of the linear and quadratic ECS after a 10 ms pulse of saturating red light (4500 mmol photons m À2 s À1 ). ECS data were normalised to the increase of the linear ECS generated after as aturating laser flash (that is, one charge separation per photosystem). We plotted the amplitude of the quadratic versus linear ECS signals during the relaxation of al ight-induced PMF and obtained the parabolic function, which allowed the calcu-lation of the dark electric field, DY d .T his experiment was then conducted with cells treated with CCCP (10 mm in ethanol, final concentration 15 mm)a nd HHQ (10 mm in DMSO, final concentration 50 mm) Fast fluorescence transients:T oo btain fast fluorescence transients, an axenic P. tricornutum culture was adjusted to 2 10 6 cells mL À1 and supplemented with 40 mm sodium bicarbonate. Quinolone stocks were added to 1mLo fd iatom culture in 1.5 mL cuvettes to af inal volume of 0.5 %( v/v). After compound addition, the treated cultures were incubated for 2min in very low light (resting in the cuvette holder). Finally,t he fast fluorescence transients of the cultures were measured with an Aqua Pen instrument (AP-C 100, Photon Systems Instruments, Drasov,C zech Republic) during as aturating multi-turnover blue light flash. Fast fluorescence transients were normalised according to ap rocedure reported by Strasser et al.: [36] V t = (FÀF 0 )/(F m ÀF 0 ), in which V t is the fluorescence at time t, F 0 is the initial fluorescence and F m is the maximum fluorescence reached.
Oxygen electrode measurements:O xygen measurements were performed by using aC lark-type electrode (Hansatech Instruments Ltd.) at 20 8C, with stirring. DMSO and quinolone stocks were added such that the final DMSO concentration never exceeded 0.5 %( v/v). All experiments began with a2min period of complete darkness, followed by addition of HHQ. In the case of photosynthesis measurements, this was then followed by red-light illumination of 200 mmol photons m À2 s À1 for the remainder of the test. Measurements were normalised based on the oxygen evolution observed in the dark period. Respiration rate (= oxygen consumption) of intact diatom cells was measured in the dark by calculating the rate of oxygen consumption over time (in 60 s), before and after adding HHQ (n = 8). Subsequently,t he values were divided by the average value before HHQ addition to obtain ar atio in which the average respiration rate of untreated cells was one.
For oxygen measurements on intact diatom cells, exponentially growing P. tricornutum (wild-type 8) cultures were concentrated by centrifugation (4500 g,5min) to af inal concentration of 5 mgchlorophyll a mL À1 and supplemented with 16 mm NaHCO 3 .F or oxygen measurements with thylakoids, thylakoids at af inal chlorophyll a concentration of 5 mgmL À1 were dissolved in 1mLof50mm tricine pH 7.8, 5mm MgCl 2 ,5m m K 2 HPO 4 and 1mm ATP. In addition, ADP (0.24 mm)a nd potassium ferricyanide (1.5 mm)w ere applied. Thylakoid membranes were isolated from exponentially growing P. tricornutum wild-type 8c ultures, following the procedure outlined by Lepetit et al., [37] with the following modifications:cells were broken at 13 000 psi, and thylakoid fragments were pelleted by centrifugation at 4 8Cf or only 10 min at 30 000 g (Sorvall) to harvest only larger thylakoid fragments and to reduce the time thylakoids were exposed to the centrifugation forces.