Calcium sensor kinase activates potassium uptakesystems in gland cells of Venus flytrapsSönke Scherzera,1, Jennifer Böhma,1, Elzbieta Krola, Lana Shabalab, Ines Kreuzera, Christina Larischa, Felix Bemmc,Khaled A. S. Al-Rasheidd, Sergey Shabalab, Heinz Rennenberge, Erwin Neherf,2, and Rainer Hedricha,2

aInstitute for Molecular Plant Physiology and Biophysics, University of Wuerzburg, D-97082 Würzburg, Germany; bSchool of Land and Food, University ofTasmania, Hobart TAS 7001, Australia; cDepartment of Molecular Biology, Max Planck Institute for Developmental Biology, 72076 Tübingen, Germany;dZoology Department, College of Science, King Saud University, Riyadh 11451, Saudi Arabia; eInstitute of Forest Sciences, University of Freiburg, 79085Freiburg, Germany; and fDepartment for Membrane Biophysics, Max Planck Institute for Biophysical Chemistry, D-37077 Goettingen, Germany

Contributed by Erwin Neher, April 24, 2015 (sent for review January 30, 2015; reviewed by Colin Brownlee and Stephen D. Tyerman)

The Darwin plant Dionaea muscipula is able to grow on mineral-poor soil, because it gains essential nutrients from captured animalprey. Given that no nutrients remain in the trap when it opensafter the consumption of an animal meal, we here asked the ques-tion of how Dionaea sequesters prey-derived potassium. We showthat prey capture triggers expression of a K+ uptake system in theVenus flytrap. In search of K+ transporters endowed with adequateproperties for this role, we screened a Dionaea expressed sequencetag (EST) database and identified DmKT1 and DmHAK5 as candi-dates. On insect and touch hormone stimulation, the number oftranscripts of these transporters increased in flytraps. After cRNAinjection of K+-transporter genes into Xenopus oocytes, however,both putative K+ transporters remained silent. Assuming that cal-cium sensor kinases are regulating Arabidopsis K+ transporter 1(AKT1), we coexpressed the putative K+ transporters with a largeset of kinases and identified the CBL9-CIPK23 pair as the majoractivating complex for both transporters in Dionaea K+ uptake.DmKT1 was found to be a K+-selective channel of voltage-depen-dent high capacity and low affinity, whereas DmHAK5 was identi-fied as the first, to our knowledge, proton-driven, high-affinitypotassium transporter with weak selectivity. When the Venus fly-trap is processing its prey, the gland cell membrane potential ismaintained around −120 mV, and the apoplast is acidified to pH3. These conditions in the green stomach formed by the closed fly-trap allow DmKT1 and DmHAK5 to acquire prey-derived K+, reduc-ing its concentration from millimolar levels down to trace levels.

Dionaea muscipula | CIPK | HAK5 | AKT | transporter

The carnivorous plant Dionaea muscipula actively traps smallanimals. To this end, it forms leaves that develop into a bi-

lobed capture organ, with the upper surface of the modified leafequipped with three sensory hairs per lobe. Flies, ants, and spi-ders, attracted by volatiles, frequently visit the traps, accidentallytouching the sensory hairs, thus triggering action potentials (APs)in the trap (1). After two APs have been elicited, the trap closeswithin a fraction of a second (2). The prey, moving around andtrying to escape, repeatedly activates the mechanosensors, whichin turn, stimulate the secretion of digestive enzymes from glandslining the inner surface of the trap. The end result is that the preyin Dionaea’s green stomach is completely decomposed, releasingnutrients and minerals that are then available to the plant.An uptake system for NH4 has already been described (3).

However, little is known about the uptake of potassium, which forplants, is a macronutrient required for turgor-based growth andmovements (review in ref. 4). From energetic considerations, it hastraditionally been proposed that high-affinity K+ uptake in plants isan active process mediated by a K+/H+ symporter, whereas low-affinity, passive uptake is operated by channels (5, 6). The major K+

uptake systems in roots, Arabidopsis thaliana high affinity K+

transporter 5 (AtHAK5) and Arabidopsis K+ transporter 1 (AKT1),were identified by studying Arabidopsis thaliana Transfer-DNAinsertion lines, in which these genes are knocked out. AtHAK5

was recognized as the only high-affinity uptake system at ex-ternal K+ concentrations <10 μM, whereas both AtHAK5 andAKT1 were shown to contribute to K+ acquisition between 10 and200 μM. At external concentrations >500 μM, AtHAK5 is notrelevant, and K+ uptake is dominated by AKT1 (7, 8). Althoughactivation by phosphorylation through the Ca2+-dependent pro-tein kinase complex calcineurin B-like (CBL)/CBL-interactingprotein kinases (CIPK) has enabled the characterization of thetransport function of AKT1, no detailed electrophysiologicalcharacterization of the high-affinity K+ transporter has beenpossible so far, because HAK5 remains silent in the heterologousexpression system of Xenopus oocytes (9). Consequently, theselectivity and regulation of HAK5 are still under debate (10).Here, we ask the question of how the flytrap acquires prey-

derived potassium ions. Our in planta experiments suggested thatDionaea glands operate an insect-induced, dual-affinity potassiumuptake system. We, therefore, screened an expressed sequence tag(EST) collection containing RNA from glands of insect- andtouch hormone-stimulated flytraps and identified orthologs ofAKT1 and AtHAK5, which we name DmKT1 and DmHAK5.Activation of these proteins in Xenopus oocytes by coexpressionwith a calcium sensor kinase allowed us to provide a detailed

Significance

The Venus flytrap Dionaea muscipula has been in the focus ofscientists since Darwin’s time. Carnivorous plants, with theirspecialized lifestyle, including insect capture, as well as di-gestion and absorption of prey, developed unique tools togain scarce nutrients. In this study, we identified the molecularnature of the uptake machinery for prey-derived potassiumand the posttranslational regulation. For the first time, to ourknowledge, we functionally characterize DmHAK5 here—aKUP/HAK/KT family member as activated by a CBL-CIPK kinasecomplex. Detailed electrophysiological characterization iden-tified DmHAK5 as a proton-driven, high-affinity potassiumtransporter with a weak selectivity. Working hand-in-hand withthe low-affinity, high-capacity K+-channel DmKT1 activated bythe same kinase, the transporters allow the Venus flytrap to takeup prey-derived potassium.

Author contributions: K.A.S. Al-R., H.R., E.N., and R.H. designed research; S. Scherzer, J.B.,E.K., L.S., I.K., C.L., S. Shabala, H.R., and R.H. performed research; S. Scherzer, J.B., E.K.,L.S., I.K., C.L., F.B., S. Shabala, H.R., E.N., and R.H. analyzed data; S. Scherzer, J.B., K.A.S. Al-R.,S. Shabala, E.N., and R.H. wrote the paper; and F.B. performed bioinformatic analysis.

Reviewers: C.B., Marine Biological Association of the United Kingdom; and S.D.T., Univer-sity of Adelaide.

The authors declare no conflict of interest.

Freely available online through the PNAS open access option.1S. Scherzer and J.B. contributed equally to this work.2To whom correspondence may be addressed. Email: eneher@gwdg.de or hedrich@botanik.uni-wuerzburg.de.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1507810112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1507810112 PNAS | June 9, 2015 | vol. 112 | no. 23 | 7309–7314

PLANTBIOLO

GY

electrophysiological characterization of the high- and low-affinitytransporters DmHAK5 and DmKT1, respectively. Both pro-teins, when expressed in oocytes, displayed the gross featuresas observed in transport studies on whole Dionaea glands.

ResultsFlytrap Ingests Prey-Derived Potassium. To study the capacity of theflytrap for intake of prey-derived potassium, we used an insectpowder stock for feeding experiments (11). In a previous study,this well-defined nutrient source was successfully used to trace theintake of amino acids and ammonium (3). The insect fodder paste,which contains a K+ concentration of 30 mM, was applied to theinner trap surface, whereupon the sensory hairs were stimulated toclose the capture organ and commence secretion of digestive fluid.This protocol, designed to mimic living prey, caused the two traplobes to form a hermetically sealed green stomach (2). By sam-pling the nutrient intake organs and removing any remainingfodder paste and stomach contents, we determined their time-dependent changes in potassium content. When monitoring thetotal trap K+ content by inductively coupled plasma MS, we foundthat the level of this major plant cation started rising 4–6 h afterinitiating feeding and reached a steady state at 12–24 h. Thehigher level lasted for at least 2–3 d (Fig. 1A). The steady-state K+

content of the fed traps was about two times that of control traps(t = 0 and 2 h after feeding). These results indicate that Dionaeahas the capacity to take up and store prey-derived potassium in thecapture organ. To confirm this finding and study the potassiumintake at a higher resolution, we complemented the insect pastewith the K+-tracer Rb+ at concentrations of 30 (Fig. 1A), 3, and0.3 mM (Fig. S1). Apart from starting from a zero level, the Rb+

concentration of traps showed the same kinetics of increase as that

determined for K+. Rubidium uptake was already visible at a foodstock concentration of 0.3 mM and increased with increasing Rb+

concentration (Fig. S1).

Gland K+ Uptake Displays Both Low- and High-Affinity TransportComponents. To study the K+ uptake, we monitored the K+

fluxes into the gland-covered inner trap surface using the vi-brating ion-selective electrode technique microelectrode ion fluxmeasuring (MIFE) (12, 13). Single-trap lobes were fixed to achamber, and ion-selective microelectrodes were placed close tothe trap surface, which is densely packed with about 37,000glands per trap (2), monitoring K+ fluxes under different exter-nal K+ concentrations. Unstimulated traps did not take up anyexternal K+ in the high-affinity range. In contrast, when trapswere pretreated with the touch hormone mimic coronatine[COR; a substitute for insect stimulation (2)], macroscopic in-ward K+ fluxes were detected (Fig. 1B and Fig. S2A). When weincreased K+ concentration stepwise from 0.01 to 1 mM withmedia buffered to pH 4, electrodes registered inward fluxes of upto 200 nmol m−2 s−1 in COR-stimulated traps. Fluxes tended tosaturate, reaching a steady state at ∼500 μM external K+ (Fig. 1Band Fig. S2A). By fitting a Michaelis–Menten equation to the K+

dose–response curve in this low-concentration range, a half-sat-uration concentration of about 60 μM was calculated (Fig. 1B).This apparent Km value is in line with a high-affinity potassiumuptake system (ref. 14 and references therein). Such a systemrequires energy from either an electrical and/or a chemical gra-dient. Uphill carrier transport of many solutes is coupled to thedownhill movement of protons (15–17). When we shifted the pHfrom 4 to 8, COR-induced potassium fluxes were completelysuppressed, indicating that high-affinity trap K+ uptake is drivenby a proton gradient (Fig. S2B).Increasing the external K+ concentration to values much

higher than the Km of the high-affinity transport further in-creased the fluxes. At 10 mM K+, inward fluxes as high as or evenabove 1,000 nmol m−2 s−1 (10 times the maximum value of thehigh-affinity transport) were detected. These high-capacity fluxespoint to the presence of a low-affinity transport system in addi-tion to the low-capacity, high-affinity component (Fig. S2C). Thefact that even unstimulated traps respond to high external K+

concentrations points to a basal expression of the low-affinitycomponent compared with the high-affinity uptake system.Extracellular electrodes provide a noninvasive assay of trap K+

fluxes, but they do not report the membrane potential of the cellsengaged with K+ transport, a quantity required for thermody-namic considerations. Influx of potassium ions and protons intogland cells is expected to depolarize their membrane potential.To monitor the predicted membrane response of gland cells, weinserted sharp impalement electrodes into cells of layer 1 or 2 ofthe dome-shaped glands. At pH 4 and 0.1 mM external K+, werecorded a resting potential around −140 mV for both insect andCOR-stimulated gland cells (cf. 3). The addition of 2 mM K+

depolarized the membrane potential of stimulated glands (Fig.1C, red line). Given a gland cell cytoplasmic potassium load of100 mM and a green stomach concentration above 1 mM K+, anelectrical force of ≤−120 mV would theoretically suffice to driveK+ influx through an acid pH-activated potassium channel (18).At an external potassium level lower than 1 mM, however, thegland cell membrane potential would not be able to energizetransport against a potassium concentration gradient greaterthan 100-fold if it was mediated by a channel-like mechanism.However, the application of 0.6 mM K+ at pH 4 still resulted in apotassium-dependent depolarization in the same range. Thisfinding points toward a high-affinity K+ transport mechanism, asystem that is not only energized by the membrane potentialgradient (Fig. 1C, black line). By plotting the depolarizationagainst the applied K+ concentration, a biphasic characterbecomes evident (Fig. 1D, red). The high-affinity component in

Fig. 1. K+ uptake in Dionaea traps. (A) Traps were fed with standardizedinsect powder, and both time-dependent K+ and Rb+ uptakes were analyzed.After 12 h of digestion, both the K+ and Rb+ concentrations in fed trapsdoubled compared with unfed traps (time = 0). (B) Using MIFE experimentswith unstimulated (black) and COR-treated (red) Dionaea traps, K+ influxeswere measured in response to the indicated external K+ concentrations. High-affinity K+ influx was obtained in COR-treated traps only. After the normali-zation of K+ fluxes to 1 at 1 mM K+, data points could be best fitted with aMichaelis–Menten function (solid line), revealing a Km value of 0.06 ± 0.01 mM(n ≥ 7; mean ± SD). (C) K+ uptake triggered membrane depolarizations inDionaea glands on application of 0.6 or 2 mM external K+. (D) High- and low-affinity response curves of gland cell membrane depolarizations (red andblack, respectively) and normalized currents (blue) of DmKT1/DmHAK5/CBL9/CIPK23-coexpressing oocytes at various external K+ concentrations. Curves inthe two concentration ranges were independently fitted with Michaelis–Menten equations to obtain the indicated half-maximal depolarizations(EC50 values; mean membrane depolarization ± SD; 3 ≤ n ≤ 8).

7310 | www.pnas.org/cgi/doi/10.1073/pnas.1507810112 Scherzer et al.

K+ uptake reaches its half-maximum activity at 0.48 ± 0.38 mM,whereas the low-affinity transport does so at 65.48 ± 33.7 mMexternal K+ (Fig. 1D). Performing the same experiment withunstimulated gland cells (Fig. 1D, black), we find almost nodepolarization in the high-affinity range. When we further in-crease the K+ concentration to the low-affinity range, a distinctdepolarization in unstimulated gland cells becomes obvious.This investigation points to a different expressional regulationof high- and low-affinity transporters during digestion.Thus, our impalement electrode recordings together with the

MIFE measurements indicate that gland cells of insect and COR-stimulated traps possess both a high-affinity, pH-dependent K+/H+

symport and a low-affinity, high-capacity K+ transport system.

Stimulated Glands Express AKT-Like Potassium Channel and HAK-TypeK+-Transporter Genes. To elucidate the genes encoding Dionaea’sgland potassium uptake system, we searched an EST database(cf. 3, 19, and 20), identifying the AKT1-like, Shaker-type po-tassium channel ortholog DmKT1 and the HAK-type trans-porter DmHAK5 as potential candidates. Both proteins werefound to be expressed and up-regulated in digesting glands of theVenus flytrap. Detailed information is in SI Text, section 1 andFig. S3 A and B.

Dionaea’s K+ Transport Requires Activation by a Calcium SignalingKinase. Xenopus laevis oocytes are widely used to express genesfrom different taxa. The function of Arabidopsis AKT1 whenexpressed in this system depends on phosphorylation by a cal-cium sensor kinase of the CBL/CIPK type (21, 22). We also foundorthologs of CIPK23 and CBL9 expressed in Dionaea. In contrastto AKT1 (21, 22), however, HAK5 transporters do not seem to befunctionally expressed in this animal expression system, which isplant background-free. Consequently, no detailed electrophysio-logical characterization of HAK transporters has been possible(9, 23). Given that, in Arabidopsis, the AKT1 potassium channelfunction depends on a CBL/CIPK pair, we reasoned that, inDionaea, AKT1 and probably, HAK5 orthologs could also dependon a similar calcium sensor kinase complex. Indeed, expression ofDmKT1-mediated macroscopic inward K+ currents is only in thepresence of the Arabidopsis CBL9/CIPK23 pair, which also acti-vates its Arabidopsis ortholog AKT1 (Fig. S4 A–C) (21, 22).Regarding HAK-type transporters, we first performed an ex-

periment similar to that in the work by Kim et al. (9), whichstudied a high-affinity KUP/HAK/KT family member. Just like inthe work by Kim et al. (9), we were not able to observe macro-scopic K+ currents in Xenopus oocytes expressing DmHAK5cRNA alone (Fig. 2A). However, we then expressed DmHAK5together with various CBL/CIPK combinations and found sub-stantial inward currents (Fig. 2) when using the CBL9/CIPK23pair and K+-containing buffers at a pH of 4. The requirement ofpH 4 is not surprising given that, on prey processing, Dionaea’sstomach acidifies. The notion that both DmHAK5 and DmKT1may interact with the CIPK23 was supported by bimolecularfluorescence complementation-based protein–protein interactionassays (Fig. S4 D–G) (cf. 24). Thus, the same calcium sensor ki-nase, CIPK23, regulates the transport function of both DmKT1and DmHAK5. We conclude that the calcium-dependent kinaseacts as a key regulator for both the low- and high-affinity potas-sium uptakes in Dionaea glands.

DmKT1 Is a K+-Selective Channel with Voltage- and pH-DependentGating. As with other members of the Shaker-like K+-channelfamily, DmKT1 is classified as a voltage-dependent inward-rec-tifying K+ channel. When the AKT1 ortholog from Dionaea wascoexpressed with CBL9/CIPK23, a typical hyperpolarization-activated, inward-rectifying K+-channel activation was observedunder 100 mM external K+ (Fig. S4 A–C). Animals contain po-tassium and sodium at about similar concentrations (25, 26).

When prey are decomposed in Dionaea’s green stomach, theK+ uptake system is exposed to both monovalent cations. Wetherefore investigated the cation selectivity of the flytrap AKT1ortholog. Oocytes were incubated with solutions containing Li+,Na+, K+, Rb+, Cs+, NH4

+, or N-methyl-D-glucamine (NMDG+) at100 mM concentration. However, DmKT1-mediated cation inwardcurrents were observed in potassium-based buffers only (Fig. 3A).On a 10-fold change in bath K+ concentration from 10 to 100 mM,a near-Nernstian shift in reversal potential (53.80 ± 1.06 mV; n =6 ± SD) was observed, indicating that DmKT1 is a K+-selectivechannel. With the acidification during prey digestion, this K+

channel is additionally confronted with a high external H+ con-centration. When oocytes were superfused with 100 mM KCl,a distinct increase in K+ inward current was observed when thepH was lowered stepwise from 6 to 3 (Fig. 3 B and C). Inter-estingly, any shift in reversal potential in response to this change inproton concentration by three orders of magnitude was in-significant from zero (−0.43 ± 3.06 mV; n = 6 ± SD; P > 0.5 byone-way ANOVA). This result documents that H+ is not trans-ported by the channel. A Boltzmann analysis of DmKT1 channelactivity at an external pH of 6 revealed a voltage dependence witha half-maximal activation potential (V1/2) of −129.64 ± 3.06 mV.Increasing the external H+ concentration shifts this potential tomore positive values, reaching a V1/2 of −78.29 ± 2.35 mV at pH 3(Fig. 3D). These findings are in line with an acid-induced shift inthe open probability (PO) of the channel (cf. 27).

DmHAK5 Is an H+-Driven K+ Transporter. The fact that even ru-bidium, which is commonly used as a potassium tracer, does notpermeate DmKT1 (Fig. 3A) raises questions about the transportof rubidium in experiments with insect powder feeding (Fig. 1Aand Fig. S1). We therefore superfused oocytes expressingDmHAK5 and CBL9/CIPK23 with a sorbitol-based solution ofpH 4 supplemented with 2 mM Li+, Na+, K+, Rb+, Cs+, NH4

+, orNMDG+ (Fig. 4 A and B). In the presence of potassium andrubidium ions, DmHAK5-mediated inward currents were ofsimilar amplitudes, whereas the currents with caesium and am-monium as charge carriers were reduced by 65–50%. Lithium,sodium, and NMDG ions at 2 mM concentration did not evokeDmHAK5-mediated currents (Fig. 4 A and B).In the green stomach, the selective K+ uptake will decrease the

external K+ concentration to very low levels, but Na+ might re-main high. Consequently, we asked whether DmHAK5 wouldstill be impermeable to Na+ under such conditions and moni-tored the currents of DmHAK5-expressing oocytes at a Na+

concentration 50 times higher compared with the K+ concen-tration of 2 mM. Currents were the same as those recordedduring replacement of Na+ by sorbitol (Fig. 4C). Also, in the

Fig. 2. Transport activation of DmHAK5. (A) Normalized current responserecorded at −140 mV of Xenopus oocytes injected with H2O, cRNA ofDmHAK5, or cRNA of DmHAK5 supplemented with CBL9/CIPK23. Macroscopicpotassium currents were only detectable in DmHAK5/CBL9/CIPK23-coexpress-ing oocytes in solutions containing 2 mM potassium (n ≥ 6; mean ± SD).(B) Typical whole-oocyte currents recorded at −60 mV of an H2O-injected(black) and a DmHAK5/CBL9/CIPK23-expressing (red) Xenopus oocyte. Addi-tion of 2 mM K+ resulted in macroscopic inward currents of ∼400 nA only withDmHAK5-expressing oocytes. The dotted line represents zero current.

Scherzer et al. PNAS | June 9, 2015 | vol. 112 | no. 23 | 7311

PLANTBIOLO

GY

absence of K+, currents were small and similar in amplitude,regardless of whether 100 mM Na+ or sorbitol was present. Thisfinding shows that, even under conditions of high sodium back-ground, which might be found in the green stomach, DmHAK5remains Na+-impermeable and that transporter activity is un-affected by Na+.As mentioned above, gland acid secretion can lower the pH in

the green stomach to a value of 3 or even lower (2), and MIFEexperiments displayed a pH dependency of K+ transport. Tostudy these effects in more detail, we exposed oocytes expressingthe activated DmHAK5 to a saturating K+ concentration, whilevarying the H+ concentration. Compared with control oocytes,an increasing K+ current at pH ≤ 5 was observed (Fig. 4D). Thisfinding is well in line with the in planta recordings and supportsthe hypothesis of an H+-driven, high-affinity K+ symport.

DmKT1 Has High Capacity and Low Affinity, Whereas DmHAK5 HasLow Capacity and High Affinity. To explore whether both Dio-naea gland K+ uptake systems act complementary to each other,we analyzed their affinity to their major substrates. With oocytesexpressing the calcium sensor kinase-activated DmKT1 orDmHAK5, we stepwise increased the external potassium con-centration in buffers of pH 4. In the presence of 5 mM externalK+ and at −140 mV, DmKT1-expressing oocytes supported aninward current of 0.15 ± 0.06 μA, which increased up to 6.42 ±0.32 μA when the K+ concentration was raised to a maximum of100 mM (Fig. S5A). A plot of current against [K+] could be fittedwith a Michaelis–Menten function, with a Km value of 108.29 ±7.34 mM (Fig. 5A).Likewise, a rising inward current was observed when super-

fusing DmHAK5-expressing oocytes with increasing K+ concen-trations under similar conditions. However, the dose–responsecurve in this case could be fitted with a Michaelis–Menten

equation with a Km value of 127.38 ± 2.77 μM (Fig. 5B) (850 timeslower than that of DmKT1). These features and the temperaturedependence (SI Text, section 2 and Fig. S6 A and B) of DmHAK5are well in line with a low-capacity, high-affinity, proton-driven K+

transporter, which was observed by MIFE and impalement ob-servations of whole Dionaea glands. When we coexpressedDmKT1 together with DmHAK5 in oocytes, we observed a bi-phasic K+ uptake (Fig. 1D, blue). This combined behavior of thetwo transporters is reminiscent of the gland K+ response (Fig.1D, red).

Low Extracellular Potassium Leads to Decreased Conductance ofDmKT1. To gain additional insights into the molecular basis ofchannel activity at different external K+ concentrations, we per-formed a quantitative comparison of DmKT1 activation curves byanalyzing the channel’s cord conductance GK (cf. 28) (Methods).This quantity was shown to be a good estimation for the voltage-dependent channel conductance in Arabidopsis AKT1 and KAT1.Plots of GK vs. voltage, obtained at various K+ concentrations,were well-described by Boltzmann functions (Fig. 5C). Theyrevealed that GK of DmKT1 decreased on lowering the externalK+ concentration. Decreasing the external K+ concentrationsfrom 100 to 5 mM, the cord conductance decreased by nearly 82%(Fig. 5C). The deduced maximal cord conductance (GK−max),which was derived from the Boltzmann fittings, was plotted againstthe corresponding K+ concentrations (Fig. 5D). The resultingsaturation curve could be best described by a Michaelis–Mentenequation and yielded a half-maximal value (K0.5) of 34 mM ex-ternal potassium. How can this behavior be explained? Usingbiophysical approaches, it had been shown that the Arabidopsispotassium channel KAT1 is K+-sensitive in the low millimolarrange (K0.5 of 1.4 mM), whereas Arabidopsis AKT1 is K+-sensitive

Fig. 4. Selectivity and pH dependency of DmHAK5. Xenopus oocytes wereeither injected with H2O (control) or cRNA of DmHAK5/CBL9/CIPK23(DmHAK5). (A) Representative current recordings at −60 mV of a control(black) and a DmHAK5-expressing (red) Xenopus oocyte in response to 2 mMLi+, Na+, K+, Rb+, Cs+, NH4

+, or NMDG+ (as indicated using sorbitol-basedsolutions). Both K+ and Rb+ resulted in macroscopic inward currents withDmHAK5-expressing oocytes. Application of NH4

+ and Cs+ induced weakerDmHAK5-mediated inward currents. The dotted line represents zero current.(B) Average results from experiments as shown in A, except that the holdingpotential was −140 mV (n = 7; mean ± SD). (C) Normalized steady-statecurrents of Xenopus oocytes injected with DmHAK5 were recorded at −140 mVin response to sorbitol- or Na+-based solution (200 mOsm/kg). Addition of2 mMK+ resulted in identical K+ influx irrespective from the presence or absenceof sodium in the bath solution (n = 7; mean ± SD). (D) ISS recorded at −140 mVwith oocytes injected with either H2O (gray bars) or DmHAK5 (red bars) in thepresence of 100 mM K+ at different pH values (as indicated; n = 5; mean ± SD).Compared with control oocytes, K+ currents increased at pH values ≤5.

Fig. 3. Selectivity and pH dependency of DmKT1. (A) Currents recorded at−140 mV of DmKT1/CBL9/CIPK23-expressing oocytes in the presence of bathsolutions containing various monovalent cations (each 100 mM). Large cat-ion influx was only obtained with potassium in the external bath solution(n = 4; mean ± SD). (B) Inward K+ currents elicited by a DmKT1/CBL9/CIPK23-expressing Xenopus oocyte markedly increased on acidification of the ex-ternal solution (Vm = −100 mV). Exchanging 100 mM external Li+ for 100 mMK+ at pH 6 resulted in potassium influx, which was enhanced by increasingthe external H+ concentration. The dotted line represents zero currents.(C) Normalized currents of DmKT1/CBL9/CIPK23-expressing Xenopus oocytesat −140 mV. Stepwise acidification from pH 6 to 3 increased the potassiumcurrents through DmKT1 (n = 6; mean ± SD). (D) The relative open proba-bility (rel. PO) of DmKT1-expressing oocytes at the indicated H+ concentra-tions was plotted against the applied test voltage. Note the prominentpositive shift of the half-maximal activation potential (V1/2) with increasingacidification. The data points were fitted with a Boltzmann function (solidlines; n = 6; mean ± SD).

7312 | www.pnas.org/cgi/doi/10.1073/pnas.1507810112 Scherzer et al.

at higher concentrations (K0.5 of 16.2 mM) (28, 29). In bothstudies, the inhibition of KAT1 and AKT1 has been associatedwith a pore collapse at low external K+ concentrations. We foundthat the open probability of DmKT1 is only weakly affected by K+

concentration changes (SI Text, section 3 and Fig. S5B). There-fore, the majorGK reduction of DmKT1 is also most likely causedby the reduction of the number of activatable channels at mod-erate external potassium concentrations. The higher sensitivityto low external K+ of DmKT1 compared with Arabidopsis K+

channels very likely represents an adaption to the needs withinthe Dionaea digestive organ. By inactivating DmKT1 underconditions of outward-directed K+ driving forces, Dionaea pre-vents K+ leakage from the gland cells.

DiscussionThe carnivorous Darwin plant Dionaea is equipped with a uniquesensory system and information management through APs tocapture prey and then, digest its nutrients (30). Because Dionaeais a nonmodel, nontransformable plant, we used a recentlyestablished gain-of-gene function approach (3, 31) and revealedthe trap transporters engaged in potassium uptake. We docu-ment that transcription of a Dionaea HAK5-type transporter inglands is induced by prey capture and the presence of jasmonate-type touch hormones. In addition to DmHAK5, we found that aShaker-type DmKT1 channel is also expressed in Dionaea glands.We cloned and expressed both DmHAK5 and DmKT1 in theXenopus oocytes system, which is free of any plant background.Nevertheless, in this heterologous system, both transportershave the same fingerprint as they do in their native flytrap.In these specialized organs, like in roots of the model plant

A. thaliana, the HAK5 K+ transporter and AKT1 K+ channelrepresent the two major complementary systems for K+ uptake

(8, 14). Interestingly, we found that the DmHAK5 K+/H+

cotransporter is posttranslationally activated by the same CBL/CIPK complex as the AKT1-type Shaker channel (21). TheHAK-type transporter is apparently the only system capable ofoperating at micromolar K+ concentrations (14). The first high-affinity K+-transporter HAK1 was cloned from Schwanniomycesoccidentalis, an ascomycete yeast that is able to grow in a nu-trient-poor environment and can accumulate potassium whenthe available external concentrations are in the low sub-micromolar range (32). Dionaea faces a very similar situation,and it is able to obtain K+ from the captured animal nutrientsource, even when concentrations are at trace levels.During digestion of Dionaea’s prey, acidification in the ex-

ternal stomach shifts the open probability of DmKT1 to morepositive values, thus allowing this channel to activate. Given thatthe cytoplasm of the gland cells contains 100 mM potassium andthat the membrane potential of a digesting Dionaea gland isabout −120 mV, this K+-selective channel can acquire externalK+ passively down to a value of 1 mM. Below this concentration,the K+ flux would be reversed if the channels would remainopen, the result of the outwardly directed electrochemical driv-ing force. Nonetheless, such K+ leakage is prevented by a re-duction of channel activity under low external K+.In this study, we were able to activate an HAK-type trans-

porter for the first time, to our knowledge, in Xenopus oocytes.We found that DmHAK5 is of low selectivity, transporting K+ =Rb+ ≥ NH4

+ ≥ Cs+. Interestingly, Na+ is neither a substrate nor amodulator of the HAK5-type transporter in Dionaea. The in-fluence of this cation without having the transporter functionallyexpressed in oocytes was still unclear and a matter of debate(33–38).By reporting a Km of 127 μM, we clearly classify DmHAK5 as

a high-affinity transporter that is active at K+ concentrations toolow for K+ uptake by DmKT1. During the absorption of a prey’smineral deposits, K+ concentration drops to 10 μM, which leadsto the question of how DmHAK5 energizes an uphill transportagainst a 10,000-fold concentration gradient under the assump-tion of a 100 mM cytosolic K+. To overcome such a high gra-dient, a passive K+ channel would only generate an influx at amembrane potential (electrical potential) more negative than−240 mV. However, the membrane potential of gland cells, evenduring prey digestion and nutrient uptake, is maintained around−120 mV (cf. this study and 3). To overcome this problem,DmHAK5, acting as a K+/H+ symporter, uses the chemical po-tential of the proton in addition to the electrical potential. As-suming a cytosolic pH of the gland cells of 7 and a pH of 3 orlower for the prey-digesting acidic fluid in the green stomach, theproton-motive force would add the equivalent of an additional−240 mV to the membrane potential. From thermodynamicconsiderations, the combined force would allow the flytrap toaccumulate K+ more than 100,000-fold. Other than this ther-modynamic consideration and the pH dependence of DmHAK5-mediated transport, final proof for K+/H+ coupling and stoichi-ometry requires measurements on the K+- and H+-dependentshift in reversal potential.Thus, both proteins are activated by the same calcium-

dependent kinase and work together to provide an orchestratedacquisition of potassium. With the high capacity of DmKT1 andthe high affinity of DmHAK5, this K+ uptake module within thegreen stomach is able to sequester the prey-derived potassiumrapidly when the potassium concentration is relatively high aswell as reduce it to very low remnant levels.

MethodsCloning of DmKT1 and DmHAK5. In brief, we used an available EST databasefrom D. muscipula (19) and identified coding sequences related to potassiumtransporter homologs. Putative transporter genes were cloned into oocyte

Fig. 5. Biophysical analysis of DmKT1- and DmHAK5-mediated K+ currentsat varying extracellular K+ concentrations. (A) Dose–response curve ofDmKT1/CBL9/CIPK23-expressing Xenopus oocytes at −140 mV. The ionicstrength of the solutions with varying K+ concentrations was adjusted to100 mM with Li+. The saturation curve could be best fitted with a Michaelis–Menten equation (solid line), documenting a Km value of 108.29 ± 7.34 mMfor DmKT1 (n = 5; mean ± SD). (B) Dose–response curve of DmHAK5/CBL9/CIPK23-expressing Xenopus oocytes clamped to a membrane potentialof −140 mV. TheMichaelis–Menten fit revealed a Km value of 127.38 ± 2.77 μMfor DmHAK5 (n = 6; mean ± SD). (C) GK/V relationship of DmKT1-expressingoocytes in a K+ concentration range from 100 to 5 mM. Data points werenormalized to GK at −140 mV in 100 mM K+ and fitted with a Boltzmannfunction. Decreasing the potassium concentration leads to a reduction in thecord conductance but not a change of the voltage dependence of the relativeopen probability (see also Fig. S5B) (n = 5; mean ± SD). (D) The relative max-imal cord conductance was calculated with the equation: GK−max = GK−max(XmM)/GK−max (100 mM). Data points were plotted against the applied K+ con-centration and fitted with a Michaelis–Menten equation. Note that GK−max

increased with increasing potassium concentrations with a half-maximal ac-tivity (K0.5) of 34 mM (n = 5 ± SD).

Scherzer et al. PNAS | June 9, 2015 | vol. 112 | no. 23 | 7313

PLANTBIOLO

GY

expression vectors (39), and cRNAs were injected into Xenopus oocytes forfunctional analysis. Detailed information is in SI Methods.

Plant Material and Tissue Sampling. D. muscipula plants were purchased fromCRESCO Carnivora and grown in plastic pots at 22 °C in a 16-h light/8-h darkphotoperiod. The material for the expression analyses was treated as de-scribed elsewhere (3).

Feeding Experiments. For rubidium and potassium uptake measurements,traps were fed with 20 mg insect powder and dispensed in 100 μL water orRbCl solution to obtain the desired Rb+ concentration. Insect powder wasproduced from adult moths of Vanessa cardui (Lep.: Nymphalidae), whichhad K+ content that was ∼30 mM; 24 h after feeding, traps were harvested,the remaining insect material was removed, and the plant tissue was com-pletely dried. Subsequently, the ground plant powder was digested withHNO3 in a pressure ashing device (Seif). Samples diluted with H2O wereanalyzed by inductively coupled plasma optical emission spectroscopy (JY 70Plus; Devision d’Instruments S.A.).

Expression Analyses. RNA was separately isolated from each sample andtranscribed into cDNA using M-MLV Reverse Transcriptase (Promega).Quantifications of the actin transcript DmACT (GenBank accession no.KC285589) and DmHAK5 transcript were performed by real-time PCR asdescribed elsewhere (40). DmHAK5 transcripts were normalized to 10,000molecules DmACT.

In a second approach, transcript levels of DmHAK5 and DmKT1 afterstimulation with insects and CORwere measured with RNA sequencing (RNA-seq). RNA-Seq by expectation-maximization (RSEM) (10.1186/1471-2105-12-323) was used to quantify expression levels based on an assembledtranscriptome. DNA-sequencing (DNA-Seq) (10.1186/gb-2010-11-10-r106)

was used to normalize between samples and experiments for the finalvisualization.

Noninvasive Ion Flux Measurements. Net K+ fluxes were measured using non-invasive MIFE technique (University of Tasmania). The theory of noninvasiveMIFE measurements and all specific details of microelectrode fabrication andcalibration are described in SI Methods and prior publications (12, 13, 41).

Intracellular Measurements. Before measurements, the lobe of a cut trap wasglued to the bottom of a chamber and left for recovery (30 min) in a standardsolution containing 0.1 mM KCl, 10 mM CaCl2, and 150 mM MES adjustedwith Tris to pH 4. Osmolarity was kept at 240 mOsm/kg with D-sorbitol whenneeded. KCl supplementing standard solution was at concentrations of 0.15,0.3, 0.6, 3, 6, 30, 60, and 120 mM. Detailed information is in SI Methods.

Oocyte Recordings. Oocyte preparation and cRNA generation and injectionhave been described elsewhere (3, 24, 28, 42). In two-electrode voltage-clamp studies, oocytes after 2 or 3 d of expression were perfused with ND 96-based solutions. Detailed information regarding solutions, pulse protocols,and data analysis is in SI Methods.

ACKNOWLEDGMENTS. We would like to thank Prof. Ingo Dreyer for se-quence analysis. F.M.B. was supported by a grant of the German ExcellenceInitiative to the Graduate School of Live Sciences, University of Würzburg.This work has been supported by the European Research Council (ERC) underthe European Union’s Seventh Framework Programme FP/20010-2015/ERCGrant Agreement 250194-Carnivorom. This work was also supported by In-ternational Research Group Program IRG14-08, a Deanship of Scientific Researchat King Saud University (to R.H., E.N., and K.A.S. Al-R.), and grants from Austra-lian Research Council Project DP150101663 and the Grain Research and Devel-opment Corporation (to S. Scherzer).

1. Kreuzwieser J, et al. (2014) The Venus flytrap attracts insects by the release of volatileorganic compounds. J Exp Bot 65(2):755–766.

2. Escalante-Pérez M, et al. (2011) A special pair of phytohormones controls excitability,slow closure, and external stomach formation in the Venus flytrap. Proc Natl Acad SciUSA 108(37):15492–15497.

3. Scherzer S, et al. (2013) The Dionaea muscipula ammonium channel DmAMT1 providesNH4

+ uptake associated with Venus flytrap’s prey digestion. Curr Biol 23(17):1649–1657.4. Hedrich R (2012) Ion channels in plants. Physiol Rev 92(4):1777–1811.5. Maathuis FJM, Sanders D (1996) Mechanisms of potassium absorption by higher plant

roots. Physiol Plant 96(1):158–168.6. Rodríguez-Navarro A (2000) Potassium transport in fungi and plants. Biochim Biophys

Acta 1469(1):1–30.7. Chérel I, Lefoulon C, Boeglin M, Sentenac H (2014) Molecular mechanisms involved in

plant adaptation to low K(+) availability. J Exp Bot 65(3):833–848.8. Alemán F, Nieves-Cordones M, Martínez V, Rubio F (2011) Root K(+) acquisition in

plants: The Arabidopsis thaliana model. Plant Cell Physiol 52(9):1603–1612.9. Kim EJ, Kwak JM, Uozumi N, Schroeder JI (1998) AtKUP1: An Arabidopsis gene en-

coding high-affinity potassium transport activity. Plant Cell 10(1):51–62.10. Kronzucker HJ, Britto DT (2011) Sodium transport in plants: A critical review. New

Phytol 189(1):54–81.11. Kruse J, et al. (2014) Strategy of nitrogen acquisition and utilization by carnivorous

Dionaea muscipula. Oecologia 174(3):839–851.12. Shabala SN, Newman IA, Morris J (1997) Oscillations in H+ and Ca2+ ion fluxes around the

elongation region of corn roots and effects of external pH. Plant Physiol 113(1):111–118.13. Shabala S, et al. (2006) Extracellular Ca2+ ameliorates NaCl-induced K+ loss from

Arabidopsis root and leaf cells by controlling plasma membrane K+ -permeablechannels. Plant Physiol 141(4):1653–1665.

14. Rubio F, et al. (2014) A low K+ signal is required for functional high-affinity K+ uptakethrough HAK5 transporters. Physiol Plant 152(3):558–570.

15. Fischer WN, et al. (2002) Low and high affinity amino acid H+-cotransporters forcellular import of neutral and charged amino acids. Plant J 29(6):717–731.

16. Schumacher K (2014) pH in the plant endomembrane system-an import and exportbusiness. Curr Opin Plant Biol 22:71–76.

17. Carpaneto A, et al. (2005) Phloem-localized, proton-coupled sucrose carrier ZmSUT1mediates sucrose efflux under the control of the sucrose gradient and the protonmotive force. J Biol Chem 280(22):21437–21443.

18. Brüggemann L, et al. (1999) Channel-mediated high-affinity K+ uptake into guardcells from Arabidopsis. Proc Natl Acad Sci USA 96(6):3298–3302.

19. Schulze WX, et al. (2012) The protein composition of the digestive fluid from the venusflytrap sheds light on prey digestion mechanisms. Mol Cell Proteomics 11(11):1306–1319.

20. Paszota P, et al. (2014) Secreted major Venus flytrap chitinase enables digestion ofArthropod prey. Biochim Biophys Acta 1844(2):374–383.

21. Xu J, et al. (2006) A protein kinase, interacting with two calcineurin B-like proteins,regulates K+ transporter AKT1 in Arabidopsis. Cell 125(7):1347–1360.

22. Li L, Kim BG, Cheong YH, Pandey GK, Luan S (2006) A Ca2+ signaling pathway reg-ulates a K+ channel for low-K response in Arabidopsis. Proc Natl Acad Sci USA 103(33):12625–12630.

23. Hong JP, et al. (2013) Identification and characterization of transcription factorsregulating Arabidopsis HAK5. Plant Cell Physiol 54(9):1478–1490.

24. Geiger D, et al. (2010) Guard cell anion channel SLAC1 is regulated by CDPK proteinkinases with distinct Ca2+ affinities. Proc Natl Acad Sci USA 107(17):8023–8028.

25. Finke MD (2013) Complete nutrient content of four species of feeder insects. Zoo Biol32(1):27–36.

26. Ajai AI, B M, Jacob JO, Audu UA (2013) Determination of some essential minerals inselected edible insects. African Journal of Pure and Applied Chemistry 7(5):194–197.

27. Hoth S, et al. (1997) Molecular basis of plant-specific acid activation of K+ uptakechannels. Proc Natl Acad Sci USA 94(9):4806–4810.

28. Geiger D, et al. (2009) Heteromeric AtKC1middle dotAKT1 channels in Arabidopsisroots facilitate growth under K+-limiting conditions. J Biol Chem 284(32):21288–21295.

29. Hertel B, et al. (2005) KAT1 inactivates at sub-millimolar concentrations of externalpotassium. J Exp Bot 56(422):3103–3110.

30. Burdon-Sanderson J (1873) Note on the electrical phenomena which accompany stimu-lation of the leaf of Dionaea muscipula. Proc R Soc Lond B Biol Sci 21(139-147):495–496.

31. Brownlee C (2013) Carnivorous plants: Trapping, digesting and absorbing all in one.Curr Biol 23(17):R714–R716.

32. Bañuelos MA, Klein RD, Alexander-Bowman SJ, Rodríguez-Navarro A (1995) A potassiumtransporter of the yeast Schwanniomyces occidentalis homologous to the Kup system ofEscherichia coli has a high concentrative capacity. EMBO J 14(13):3021–3027.

33. Rubio F, Alemán F, Nieves-Cordones M, Martínez V (2010) Studies on Arabidopsis athak5,atakt1 double mutants disclose the range of concentrations at which AtHAK5, AtAKT1and unknown systems mediate K uptake. Physiol Plant 139(2):220–228.

34. Qi Z, et al. (2008) The high affinity K+ transporter AtHAK5 plays a physiological role inplanta at very low K+ concentrations and provides a caesium uptake pathway inArabidopsis. J Exp Bot 59(3):595–607.

35. Spalding EP, et al. (1999) Potassium uptake supporting plant growth in the absence ofAKT1 channel activity: Inhibition by ammonium and stimulation by sodium. J GenPhysiol 113(6):909–918.

36. Bertl A, Reid JD, Sentenac H, Slayman CL (1997) Functional comparison of plant in-ward-rectifier channels expressed in yeast. J Exp Bot 48(Spec No):405–413.

37. Qi Z, Spalding EP (2004) Protection of plasma membrane K+ transport by the salt overlysensitive1 Na+-H+ antiporter during salinity stress. Plant Physiol 136(1):2548–2555.

38. Aleman F, Nieves-Cordones M, Martinez V, Rubio F (2009) Differential regulation ofthe HAK5 genes encoding the high-affinity K+ transporters of Thellungiella halophilaand Arabidopsis thaliana. Environ Exp Bot 65(2-3):263–269.

39. Nour-Eldin HH, Hansen BG, Norholm MH, Jensen JK, Halkier BA (2006) Advancinguracil-excision based cloning towards an ideal technique for cloning PCR fragments.Nucleic Acids Res 34(18):e122.

40. Escalante-Pérez M, et al. (2012) Poplar extrafloral nectaries: Two types, two strategiesof indirect defenses against herbivores. Plant Physiol 159(3):1176–1191.

41. Shabala S, Shabala L (2002) Kinetics of net H4, Ca2+, K+, Na+, NH4+, and Cl− fluxes

associated with post-chilling recovery of plasma membrane transporters in Zea maysleaf and root tissues. Physiol Plant 114(1):47–56.

42. Becker D, et al. (1996) Changes in voltage activation, Cs+ sensitivity, and ion perme-ability in H5 mutants of the plant K+ channel KAT1. Proc Natl Acad Sci USA 93(15):8123–8128.

7314 | www.pnas.org/cgi/doi/10.1073/pnas.1507810112 Scherzer et al.

Supporting InformationScherzer et al. 10.1073/pnas.1507810112SI Text1. Transcriptional Regulation of Dionaea’s Potassium Uptake Modules.Expression pattern of DmKT1 and DmHAK5. In our expression studies,we find DmKT1 as well as DmHAK5 expressed in the petiole,trap, and glands, with the DmKT1 expression level higher in glandsand traps compared with the DmHAK5 expression level undernonstimulated conditions (Fig. S3). Thus, the basal expression levelof DmKT1 might be responsible for the K+-dependent depolar-ization of unstimulated gland cells in the low-affinity range (Fig.1D). On trap stimulation, either by an insect or with the touchhormone mimic COR, transcript numbers of the transporterDmHAK5 increased 6-fold in traps and 18-fold in glands. Incontrast, the K+-channel DmKT1 was only slightly induced bystimulation (Fig. S3B).DmHAK5 expression and high-affinity K+ uptake show a similar time-dependent behavior. To study the time dependency in Dionaea’sK+ uptake system, we analyzed the expression level of DmHAK5after COR stimulation. As we already observed (Fig. S3B),DmHAK5 expression under unstimulated conditions was verylow (Fig. S1B). Interestingly, its expression reaches a maximum3.5 h after stimulation by COR and stays at this level for 2 d.When we analyzed the time dependency of high-affinity K+

uptake in the Dionaea gland system, we were technically not ableto detect any response to 0.03 mM K+ during the first 3.5 h afterstimulation. First depolarization was detected after 6 h, reachingthe maximal K+ uptake after 24 h (Fig. S1B). The delay betweentranscriptional induction and plasma membrane response is prob-ably explained by the posttranslational machinery, which is neces-sary to incorporate the active protein into the plasma membrane.

2. DmHAK5 Acts as a Transporter. Carrier-mediated transportprocesses can be distinguished from a Nernstian-type ion channelbased on their activation energy Q10. To determine the Q10 ofDmHAK5, we calculated the temperature dependence between10 °C and 30 °C of the K+/H+ current amplitude. This tem-perature change affected the DmHAK5-dependent currentsstrongly (Fig. S6), and the Q10 value of 6.87 ± 1.06 (activationenthalpy was 38 kJ/mol) derived from Arrhenius plotting clas-sifies the gland cell high-affinity K+ transport activity as carrier-like. This result supports the notion of a K+/H+ symport mech-anism by DmHAK5.

3. Estimation of the Relative Open Probability. Voltage-gated ionchannels can exist in three functional states: closed, open, andinactivated (1). Although the closed and inactivated states arenonconductive states, the open conformation is conductive,mediating ion currents. In the case of DmKT1, prolonged hy-perpolarization of the cell membrane shifts the channel into itsopen state (Fig. S4C). The relative open probability (rel. PO) is ameasure that determines the fraction of channels in the open/conductive state (1, all channels are open; 0, all channels areclosed). The voltage-dependent gradual shift of rel. PO from zeroto one obeys a Boltzmann distribution. To determine rel. PO,oocytes were successively challenged with voltage pulses rangingfrom +60 to −140 mV. When the channel-mediated currentsreached equilibrium (steady state), the membrane potential wasinstantaneously changed to a constant voltage of +10 mV. Be-cause the change in membrane potential is faster than the decayof the channels’ open probability, the instantaneous currents(Iinst) right after the voltage jump represent a measure for theopen probability of the steady-state conditions at the precedingmembrane potential. When plotting Iinst as a function of voltage,

a Boltzmann equation (below) can be used to describe the dataand determine channel characteristics, like the voltage where thehalf-maximal open probability is reached:

PPO +Pc

=1

1+ exp�ðV −V0.5Þ× z× e0

k×Τ

�.

PO and Pc are the fractional occupancies of the open and closedstates, respectively. V is the membrane potential, and V0.5 is thevoltage at which one-half of the channels are in the open state; kis Boltzmann’s constant, and z is the number of the moved elec-tric charges e0 = 1.6 × 10−19 C necessary for the channel to open.The temperature is T = 290.6 K.

SI MethodsCloning of DmKT1 and DmHAK5. Inspection of the available ESTdata from Dionaea muscipula (2) revealed coding sequencesrelated to potassium transporter homologs. PolyA-mRNA wasisolated from 10 μg total RNA using Dynabeads (Invitrogen)according to the manufacturer’s instructions. cDNA was gener-ated from D. muscipula whole-plant mRNA. The cDNA ofpotassium transporters was amplified using gene-specific oli-gonucleotide primers, which comprised the coding DNA se-quence (CDS) of DmKT1 (GenBank accession No. LN715171)and DmHAK5 (GenBank accession no. LN715172) and Ad-vantage cDNA Polymerase Mix (Clontech), revealing a sequenceidentical to that expected from the EST data. The full-lengthclones were inserted into pGEM-T Easy using the pGEM-T EasyVector System I (Promega). For heterologous expression inXenopus oocytes, the generated cDNAs were cloned into oocyteexpression vectors (based on pGEM vectors) by an advanceduracil excision-based cloning technique described by Nour-Eldinet al. (3). For functional analysis, cRNAs were prepared usingthe mMessage mMachine T7 Transcription Kit (Ambion). Oo-cyte preparation and cRNA injection have been described else-where (4). For oocyte electrophysiological experiments, 25 ngpotassium transporter cRNA was injected.

Noninvasive Ion Flux Measurements. Microelectrodes with externaltip diameters of ∼2 μm were pulled, silanized, and filled with anappropriate ion-selective mixture [K+ 60031; Sigma; mixture wasprepared as described in the work by Jayakannan et al. (5)].Electrodes were then calibrated in an appropriate set of stan-dards and mounted on a 3D micromanipulator (MMT-5; Nar-ishige). An immobilized lobe was placed into a 90-mm Petri dishcontaining 40 mL basic salt media (BSM; 0.2 mM KCl, 0.1 mMCaCl2 adjusted to required pH using MES/Tris) and then, placedinto a Faraday cage. The K+-selective microelectrode was posi-tioned with the tip aligned 50 μm above the lobe surface using a3D hydraulic manipulator. During the measurements, a com-puter-controlled stepper motor moved electrodes in a slow (10 s)square-wave cycle between the two positions: close to (50 μm)and away from (150 μm) the trap surface. The potential differ-ence between two positions was recorded by the MIFE CHARTsoftware (6) and converted into electrochemical potential dif-ference using the calibrated Nernst slopes of the electrodes. Netion fluxes were calculated using the MIFEFLUX software forcylindrical diffusion geometry (7).To study the dose dependence of K+ uptake, intact Dionaea

leaves were pretreated with COR for 24 h as described above. Asingle lobe was cut and immobilized in a measuring chamber by

Scherzer et al. www.pnas.org/cgi/content/short/1507810112 1 of 7

using medical adhesive (VH355; Ulrich AG). The lobe was leftfor adaptation in BSM solution before flux measurements started.Net K+ fluxes were measured in response to consecutive increasesin KCl content in the external BSM (KCl: 10, 50, 100, 200, 1,000,and 10,000 μM) during 3 min at each KCl concentration. The pHof the media was maintained at pH 4. Traps without CORtreatment were used as controls. To address pH dependence ofthe studied channel, K+ fluxes were recorded at pH 4 and 8 attwo KCl levels (10 and 200 μM).

Intracellular Measurements. In the course of experiments, leaveswere constantly perfused with the standard solution (1 mL/min).For stimulation, a 1-min exposure to the respective KCl con-centration was sufficient. For impalements, microelectrodes fromborosilicate glass capillaries with filament (Hilgenberg) werepulled on a horizontal laser puller (P2000; Sutter InstrumentsCo.). They were filled with 300 mM KCl and connected throughan Ag/AgCl half-cell to a head stage (1 GΩ; HS-2A; Axon In-struments). Tip resistances were about 30 MΩ. Reference electrodeswere filled with 300 mM KCl as well. An IPA-2 Amplifier (Appli-cable Electronics, Inc.) was used. The cells were impaled using anelectronic micromanipulator (NC-30; Kleindiek Nanotechnik).

Oocyte Recordings. In double-electrode voltage-clamp studies,oocytes were perfused with citric acid/Tris-based buffers. Thestandard bath solution contained 10 mM citric acid/Tris (pH 4),1 mM CaCl2, 1 mM MgCl2, and 1 mM LaCl3. Osmolality of eachsolution was adjusted to 220 mOsm/kg using D-sorbitol. To bal-ance ionic strength for measurements under varying cationconcentrations, we compensated with lithium for a total con-centration of 100 mM. Solutions for selectivity measurements ofDmKT1 were composed of standard bath solution supplementedwith 100 mM Na+, NMDG+, NH4

+ Li+, K+, Rb+, or Cs+ chloridesalts. For selectivity measurements of the high-affinity DmHAK5transporter, respective cations were used at a concentration of2 mM, and osmolarity was adjusted to 220 mOsm/kg using

D-sorbitol. For analyzing the pH dependency, the pH value ofstandard solutions was adjusted between pH 7 and 6 with theMES/Tris buffer system. For pH values between pH 5.5 and 3,10 mM citric acid instead of MES was used. Temperature ex-periments were performed under standard solution conditions asdescribed by Scherzer et al. (8). For current–voltage relations(IV curves), single-voltage pulses were applied in 10-mV dec-rements from +60 to −140 mV starting from a holding potential(VH) of 0 mV. Steady-state currents (ISS) were measured at theend of test pulses lasting 100 ms in the case of DmHAK5 and 3 sin the case of DmKT1, and the cord conductance (GK) wascalculated according to

GK =Iss

ðV −VrevÞ.

Here, Vrev is the reversal potential, which was determined fromtail currents after a preactivation of −100 mV. Because currentswere small in 5 mM K+, we extrapolated the measured Vrev from10 to 100 mM to calculate the Vrev for 5 mM K+.The calculation of the channel cord conductance eliminates the

influence of the chemical driving force of the potassium gradientand was shown to result in a voltage-dependent conductanceaccording to Geiger et al. (9):

GK− =

Gmax

ðV −VrevÞ0@1+ exp

0@zF

�V −V1=2

�RT

1A1A.

V1/2 is the half-maximal activation voltage, z is the apparentgating charge, and GK−max is the maximal cord conductance.Data were analyzed using Igor Pro and Origin Pro-9.0G.

1. Karmazinova M, Lacinova L (2010) Measurement of cellular excitability by whole cellpatch clamp technique. Physiol Res 59(Suppl 1):S1–S7.

2. Schulze WX, et al. (2012) The protein composition of the digestive fluid from the venusflytrap sheds light on prey digestion mechanisms. Mol Cell Proteomics 11(11):1306–1319.

3. Nour-Eldin HH, Hansen BG, Norholm MH, Jensen JK, Halkier BA (2006) Advancinguracil-excision based cloning towards an ideal technique for cloning PCR fragments.Nucleic Acids Res 34(18):e122.

4. Escalante-Pérez M, et al. (2012) Poplar extrafloral nectaries: Two types, two strategiesof indirect defenses against herbivores. Plant Physiol 159(3):1176–1191.

5. Jayakannan M, Babourina O, Rengel Z (2011) Improved measurements of Na+ fluxes inplants using calixarene-based microelectrodes. J Plant Physiol 168(10):1045–1051.

6. Shabala SN, Newman IA, Morris J (1997) Oscillations in H+ and Ca2+ ion fluxes aroundthe elongation region of corn roots and effects of external pH. Plant Physiol 113(1):111–118.

7. Shabala L, Ross T, McMeekin T, Shabala S (2006) Non-invasive microelectrode ion fluxmeasurements to study adaptive responses of microorganisms to the environment.FEMS Microbiol Rev 30(3):472–486.

8. Scherzer S, et al. (2013) The Dionaea muscipula ammonium channel DmAMT1 providesNH4

+ uptake associated with Venus flytrap’s prey digestion. Curr Biol 23(17):1649–1657.

9. Geiger D, et al. (2009) Heteromeric AtKC1middle dotAKT1 channels in Arabidopsisroots facilitate growth under K+-limiting conditions. J Biol Chem 284(32):21288–21295.

Scherzer et al. www.pnas.org/cgi/content/short/1507810112 2 of 7

Fig. S1. Time-dependent activation of DmHAK5 in Dionaea traps. (A) Traps were fed with standardized insect powder, which was supplemented with RbCl atthe given concentrations. Rb+ uptake was analyzed 24 h after feeding onset. (B) The expression of DmHAK5 in Dionaea traps reaches a maximum 3.5 h afterinduction by COR (right axis). This expression rate stays at this level for 2 d. The high-affinity K+-dependent depolarization in corresponding traps wasmeasured by application of 30 μM K+ (left axis). First depolarization was detected after 6 h, reaching the maximal K+ uptake after 24 h (mean membranedepolarization ± SD; n ≥ 5).

Scherzer et al. www.pnas.org/cgi/content/short/1507810112 3 of 7

Fig. S2. Ion flux measurements across the membrane of Dionaea glands. (A) Time-dependent net K+ fluxes measured with COR-stimulated (red) and non-stimulated (black) Venus flytrap glands in response to the indicated external K+ concentrations. Stimulated traps showed K+ influxes increasing with increasingexternal K+ application. Nonstimulated traps showed no significant K+ influx under identical conditions (n ≥ 7; mean ± SD). (B) K+ influx into Dionaea glands isH+-dependent. Normalized K+ fluxes from COR-stimulated Venus flytraps were measured in response to low (0.01 mM; left bar) and high (0.2 mM; center bar)KCl concentrations at pH 4. Increasing the external pH to 8 made the K+ flux vanish at a K+ concentration of 0.2 mM (n ≥ 8; mean ± SD; right bar).(C) Representative experiment of a COR-stimulated Venus flytrap measured at K+ concentrations in the high- and low-affinity ranges. With external K+ up to1 mM, a high-affinity uptake system was active, mediating K+ influx of ∼200 nmol m−2 s−1. With 10 mM external K+, high-capacity influx of up to1,800 nmol m−2 s−1 was detected, indicating that, in addition to the high-affinity uptake, there is a low-affinity/high-capacity uptake system expressedin Dionaea glands.

Scherzer et al. www.pnas.org/cgi/content/short/1507810112 4 of 7

Fig. S3. Expression pattern of the K+ uptake system in Dionaea. (A) Quantification of DmHAK5 transcript levels in different aerial tissues. DmHAK5 transcriptswere normalized to 10,000 molecules of DmACT1 (Dionaea Actine1). Expression of the high-affinity K+ transporter was strongly up-regulated by COR in trapsand glands of Dionaea (n = 3; mean ± SE). (B) Quantification of DmHAK5 and DmKT1 transcript levels after stimulating traps of Dionaea with COR and feedinginsects. Expression values are variance-stabilized for graphical exploratory analysis. Both DmHAK5 and DmKT1 are being induced during stimulation. Foldchange and overall expression level are much higher for DmHAK5 after stimulation compared with DmKT1.

Scherzer et al. www.pnas.org/cgi/content/short/1507810112 5 of 7

Fig. S4. CIPK23 interacts with and activates the Dionaea K+ uptake system. Whole-oocyte voltage-clamp measurements of DmKT1 at membrane potentialsranging from +60 to −140 mV in 20-mV decrements in the presence of 100 mM KCl are shown. (A–C) Macroscopic K+ currents of up to 12 μA were onlydetectable in oocytes coexpressing DmKT1 with CBL9 + CIPK23. (D–G) Bimolecular fluorescence complementation in Xenopus oocytes revealed physical in-teraction between (D and E) DmHAK5, (F and G) DmKT1, and (D and F) CIPK23/CBL9 but not (E and F) CIPK24/CBL9. Pictures taken with a confocal laser-scanning microscope show one-quarter of an optical slice of an oocyte. (D) DmHAK5::YFPC coexpressed with CBL9 + CIPK23::YFPN. (E) DmHAK5::YFPC coex-pressed with CBL9 + CIPK24::YFPN. (F) DmKT1::YFPC coexpressed with CBL9 + CIPK23::YFPN. (G) DmKT1::YFPC coexpressed with CBL9 + CIPK24::YFPN.

Scherzer et al. www.pnas.org/cgi/content/short/1507810112 6 of 7

Fig. S5. Biophysical analysis of DmKT1-mediated K+ currents at varying extracellular K+ concentrations. (A) Steady-state currents of oocytes expressingDmKT1/CIPK23/CBL9 at pH 4 under different external K+ concentrations recorded at −140 mV. The ionic strength of the solutions with varying K+ concen-trations was adjusted to 100 mM with Li+. Inset shows currents of H2O-injected control oocytes under the same conditions (n = 5; mean ± SD). (B) The relativeopen probability (rel. PO) of DmKT1/CIPK23/CBL9 expressed in oocytes at an external pH of 4 was plotted against the test voltages. The data points were fittedwith a Boltzmann function (solid lines; n = 6; mean ± SD). Note that K+ is not affecting the half-maximal activation potential (V1/2).

Fig. S6. Temperature dependency of DmHAK5. (A) DmHAK5/CIPK23/CBL9 functions as a transporter. Arrhenius plot of normalized currents recorded at 2 mMK+ and a membrane potential of −140 mV at pH 3 and varying temperatures (as indicated). A mean Q10 value of 6.87 ± 1.06 was calculated at membranepotentials of −100 and −140 mV. Numbers represent a mean Q10 value of the currents recorded at 10–20 °C, 20–30 °C, and 10–30 °C (n = 6; mean ± SD). In asecond approach, the linear regression in the Arrhenius plot with a slope of −4,588 ± 98 corresponds to an activation enthalpy Ea (Ea = −R × slope; R is the gasconstant) of 38 ± 4 kJ/mol (n = 6; mean ± SD). (B) Temperature dependency of DmHAK5/CIPK23/CBL9-mediated K+ currents. Representative measurementswere performed at pH 3 and a membrane potential of −40 mV. Currents were induced by 2 mM K+. Increasing the bath temperature from 20 °C to 30 °C (blueline) resulted in rapidly increasing inward currents (red line), which behavior was reversed by lowering the temperature to 20 °C again. The dotted linerepresents the zero current value.

Scherzer et al. www.pnas.org/cgi/content/short/1507810112 7 of 7