The Plasmodial Surface Anion Channel Is Functionally Conserved in Divergent Malaria Parasites

The plasmodial surface anion channel (PSAC), a novel ion channelinduced on human erythrocytes infected with Plasmodium falciparum,mediates increased permeability to nutrients and presumablysupports intracellular parasite growth. Isotope flux studiesindicate that other malaria parasites also increase the permeabilityof their host erythrocytes, but the precise mechanisms are unknown.Channels similar to PSAC or alternative mechanisms, such asthe upregulation of endogenous host transporters, might fulfillparasite nutrient demands. Here we evaluated these possibilitieswith rhesus monkey erythrocytes infected with Plasmodium knowlesi,a parasite phylogenetically distant from P. falciparum. Tracerflux and osmotic fragility studies revealed dramatically increasedpermeabilities paralleling changes seen after P. falciparuminfection. Patch-clamp of P. knowlesi-infected rhesus erythrocytesrevealed an anion channel with striking similarities to PSAC:its conductance, voltage-dependent gating, pharmacology, selectivity,and copy number per infected cell were nearly identical. Ourfindings implicate a family of unusual anion channels highlyconserved on erythrocytes infected with various malaria parasites.Together with PSAC's exposed location on the host cell surfaceand its central role in transport changes after infection, thisconservation supports development of antimalarial drugs againstthe PSAC family.

INTRODUCTION

Top

Introduction

Red blood cells (RBCs) infected with Plasmodium falciparum,the most virulent cause of human malaria, exhibit markedly increasedpermeabilities to a diverse collection of solutes, includinganions, sugars, amino acids, purines, some vitamins, and organiccations, all via the plasmodial surface anion channel (PSAC)(8). PSAC is induced some hours after parasite invasion andeventually reaches a functional copy number of between 1,000/celland 2,000/cell, as estimated from comparisons of single-moleculeand whole-cell patch-clamp measurements. Several functionalproperties suggest it is a novel ion channel. First, its rankorder selectivity for anions (SCN^– > I^– >Br^– > Cl^–) differs from nearly all human chloridechannels and corresponds to the theoretical weakest bindingsite, Eisenman series I for anions (40). Another surprisingselectivity feature is PSAC's ability to exclude Na⁺ by morethan 100,000-fold relative to Cl^– despite broad permeabilityto organic solutes of either net positive or negative charge(5).

Second, although infected RBCs are not excitable, PSAC exhibitsvoltage-dependent gating with significantly more frequent openingevents at negative membrane potentials than at positive membranepotentials. This voltage dependence is also apparent in whole-cellrecordings, in which current-voltage profiles exhibit "inwardrectification." Thus, it has become an important marker of PSAC.An important debate has evolved around this phenotype becausesome groups report mixed voltage dependences that suggest thation channels other than PSAC might also be induced by the parasite(21, 34).

Third, PSAC has unique pharmacological properties. In additionto inhibition by phloridzin (25), furosemide, NPPB [nitro-2-(3-phenyl-propylamino)benzoicacid] (23), and glybenclamide (24)—nonspecific antagonistsof channels and carriers—PSAC is unexpectedly blockedby dantrolene, a specific antagonist of sarcoplasmic reticulumCa²⁺ channels without measurable activity against five otherclasses of anion channels (G. Lisk and S. A. Desai, submittedfor publication). A small secondary screen of dantrolene derivativessubsequently achieved further specificity for PSAC by identifyinga nitrophenyl-furan derivative that has higher affinity forPSAC inhibition and no measurable activity against the sarcoplasmicreticulum Ca²⁺ channels (22). Consistent with an essential rolefor PSAC, this higher-affinity derivative inhibits in vitroparasite growth at lower concentrations than dantrolene.

An important unanswered question is whether other plasmodialspecies induce similar channels on their host membranes. Previoustransport studies using parasites other than P. falciparum haveproduced mixed answers to this question. One study measured[³H]choline uptake by Plasmodium knowlesi-infected rhesus monkeyRBCs and concluded that rather than inducing novel ion channels,this parasite upregulates an endogenous carrier to fulfill theincreased demands for this precursor of phospholipid biosynthesis(3). In contrast, another study found that Plasmodium vinckeiinduces a broad specificity, furosemide-sensitive pathway similarto PSAC and that it also upregulates the endogenous mouse RBCcholine carrier (33). Electrophysiological studies of Plasmodiumgallinaceum-infected chicken RBCs (36) and of Plasmodium berghei-infected mouse RBCs (20) appear to support upregulation of endogenouschannels, although achieved signal-to-noise ratios—calculatedfrom solution compositions and reported seal resistances—werenot adequate to rule out PSAC-like channels.

Here, we examined the transport properties of rhesus monkeyRBCs before and after in vitro infection with P. knowlesi, aparasite species chosen in part for its phylogenetic distancefrom P. falciparum. Our osmotic lysis and radioisotope uptakestudies revealed increases in organic-solute permeability parallelingthose documented with P. falciparum. Then, whole-cell and cell-attachedpatch-clamp methods identified a small-conductance anion channelon P. knowlesi-infected rhesus RBCs but absent from uninfectedcells. This anion channel showed marked similarities to theP. falciparum PSAC, with nearly indistinguishable single-channelconductance, gating behavior, selectivity, and pharmacologicalproperties. We propose that PSAC is highly conserved on theseand other plasmodia and that this family of unusual channelsis an essential adaptation for survival in the intraerythrocyticenvironment.

MATERIALS AND METHODS

Top

Materials and Methods

Parasite culture and osmotic lysis assays. P. knowlesi was cultured under in vitro conditions with RBCscollected from healthy rhesus monkeys (Macaca mulatta) and RPMI1640 supplemented with 50 mg/liter hypoxanthine and 20% humanserum. In contrast to P. falciparum's 44 to 48 h cycle in humanRBCs, the erythrocytic cycle averaged 24 h. For both parasites,the kinetics of osmotic lysis in various solutions was followedas described previously (38). In brief, trophozoite-infectedRBCs were enriched to 95 to 99% by percoll-sorbitol separation(1), washed in phosphate-buffered saline (PBS) (150 mM NaCl,20 mM Na-phosphate, pH 7.5), and resuspended to a 0.25% hematocritin buffer (20 mM Na-HEPES, 0.1 mg/ml bovine serum albumin, pH7.5) with 280 mM of sorbitol, isoleucine, or alanine. Na⁺ permeabilitywas evaluated in PBS with 0.1 mg/ml bovine serum albumin. Thetransmittance of 700 nm light, %T, through these cell suspensionswas then continuously followed at 37°C; these measurementstrack changes in turbidity that result from osmotic swellingand lysis of cells. This method produces estimates of permeationrates through PSAC that quantitatively match estimates obtainedwith both radioisotope flux and patch-clamp (38).

Electrophysiology. Single-channel and whole-cell patch-clamp recordings of rhesusmonkey RBCs were obtained with quartz glass pipettes pulledto tip diameters of <0.5 µm and resistances of 1 to4 M ${Omega}$ . Most experiments utilized symmetrical bath and pipettesolutions of (in mM) 1,000 choline-Cl, 115 NaCl, 10 MgCl₂, 5CaCl₂, and 20 Na-HEPES, pH 7.4. Some experiments replaced choline-Clwith equimolar NaCl and identified channel activity with indistinguishableproperties. Recordings were filtered at 5 KHz with an eight-poleBessel filter and digitized at 100 KHz. Seal resistances weretypically >100 G ${Omega}$ , as required for detection of these small-conductancechannels (2).

Whole-cell patch-clamp to determine anion selectivity was performedsimilarly but with a bath solution of (in mM) 500 NaCl, 10 MgCl₂,5 CaCl₂, 20 Na-HEPES, pH 7.4, and pipette solutions of either500 NaClO₄ or 500 Na-lactate with 10 MgCl₂, 5 CaCl₂, and 20Na-HEPES, pH 7.4. Under these conditions, the permeability coefficientof ClO₄^– or the lactate anion relative to Cl^– canbe determined by measuring the reversal potential (E_rev), themembrane voltage that produces zero current.

Single-channel amplitudes, open probabilities, and dwell distributionswere measured with home-written code as described previously(2, 6).

Radioisotope uptake. P. knowlesi-infected or uninfected rhesus monkey RBCs were washedand used to measure uptake of [³H]alanine (2 µCi/ml) ata 10% hematocrit in PBS supplemented with 1 mM unlabeled alanineat 22°C. At 5-min intervals, uptake was terminated by transferof 330-µl aliquots to 1.5 ml of tracer-free PBS with 200µM furosemide at 4°C and centrifugation at 14,000x g through dibutyl phthalate. Each time point was measuredin duplicate. Cell pellets were digested and counted as previouslydescribed (11). Uptake was corrected for extracellular trappingby subtracting the uptake measured at zero time in the presenceof furosemide.

RESULTS

Top

Results

Broad increases in permeability after P. knowlesi infection. PSAC-mediated permeability changes after infection of humanRBCs by P. falciparum can be readily identified by an increasedosmotic fragility in isotonic solutions of permeant solutes(17). In these solutions, there is net uptake of the permeantsolute via PSAC, cell swelling, and selective osmotic lysisof infected RBCs. Lysis of P. falciparum-infected human RBCscan be quantitatively and continuously tracked with a simplelight-scattering assay that measures the turbidity of the cellsuspension (38).

Here, we found that this assay can also be used to follow osmoticlysis of rhesus monkey RBCs with or without P. knowlesi infection.These kinetic measurements revealed marked increases in sorbitol,alanine, and isoleucine permeabilities after infection, whichcould be largely inhibited by 200 µM furosemide (Fig.1A through C, upper panels), a well-studied antagonist of theP. falciparum PSAC (2). Uninfected rhesus RBCs exhibited negligiblepermeability to each of these solutes (Fig. 1A through C, lowerpanels). Both uninfected and infected RBCs were stable in isotonicNaCl solutions (Fig. 1D), consistent with sustained low Na⁺permeability despite increases in organic solute permeability.We also confirmed these permeability changes with more conventional[³H]alanine radioisotope accumulation experiments (Fig. 1E).Both this list of solutes and estimates of their absolute permeabilitiesparallel previous measurements in the P. falciparum-human RBCsystem (5, 17, 38), suggesting PSAC-like channels on P. knowlesi-infectedrhesus RBCs.

View larger version (11K):
[in this window]
[in a new window]
FIG. 1. Increased organic solute permeability after P. knowlesi infection. (A through D) Osmotic lysis of trophozoite-infected rhesus RBCs in buffered sorbitol, isoleucine, alanine, and NaCl solutions (black curves, upper panels). Each lower panel represents an identical experiment with uninfected rhesus monkey RBCs; their lack of measurable lysis reflects their low intrinsic permeability to these solutes. In all panels, red traces reflect the lysis time course in the presence of 200 µM furosemide. Transmittance of 700 nm light though cell suspensions was converted to percent lysis with measurements on identical but fully lysed suspensions of sorbitol-treated infected RBCs or detergent-treated uninfected RBCs. Notice that each of the P. knowlesi-induced permeability changes is furosemide-sensitive. (E) [³H]alanine uptake in uninfected RBCs (triangles) or infected RBCs (circles) with or without 200 µM furosemide (red or black symbols, respectively). Uptake was not corrected for parasitemia (6% in this experiment); such corrections would increase the difference between infected and uninfected cells by more than 15-fold. [³H]isoleucine exhibited similar increases in uptake (not shown).

Identification and characterization of the P. knowlesi PSAC. We used electrophysiological methods to probe the mechanisticbasis of these increases. Although technically complicated bythe somewhat smaller size of rhesus RBCs when compared to humanRBCs (35), we successfully achieved high-resistance whole-cellpatch-clamp recordings on these cells. In symmetrical bath andpipette solutions containing Cl^–, P. knowlesi-infectedRBCs exhibited large whole-cell currents that were absent fromuninfected RBCs (Fig. 2). Infected RBC whole-cell measurementsrevealed a steep voltage dependence with significantly largercurrents at negative voltages than at positive voltages despiteequal but opposite driving forces. Furosemide abolished thesecurrents, consistent with a broad-specificity channel that mediatesthe transport of both inorganic anions and the organic solutesevaluated for Fig. 1. In identical whole-cell experiments withuninfected rhesus monkey RBCs, the measured currents were significantlysmaller and voltage independent, indicating low permeabilitiessimilar to those of human RBCs (8). (With uninfected RBCs ofboth species, it is not possible to precisely calculate conductiveCl^– permeabilities from whole-cell measurements becauseof errors in measuring their small currents.)

View larger version (16K):
[in this window]
[in a new window]
FIG. 2. Whole-cell patch-clamp measurements. (A) Ensemble of current responses to V_m values of –100 to +100 mV in 10 mV increments for uninfected rhesus monkey RBC in the whole-cell configuration with 1.0 M choline-Cl in both bath and pipette. (B) Current responses of P. knowlesi-infected RBC to the same protocol before or after addition of saturating concentrations of furosemide (left and right groups of traces, respectively). Dashes on the left and right of traces represent the baseline current with zero applied voltage. Notice the currents at negative V_m are greater than those at corresponding positive V_m and that furosemide largely abolishes these currents. Scale bars represent 25 ms (horizontal) and 3 nA (vertical) in both panels A and B. (C) Comparison of whole-cell chord conductances between P. knowlesi-infected rhesus RBCs (black bar, mean ± standard error of the mean [SEM]) and P. falciparum-infected human RBCs (white bar). (D) Anion selectivity determined with bi-ionic experiments with 500 mM Na⁺Cl^– in the bath and either 500 mM Na⁺ClO₄^– or 500 mM Na⁺-lactate in the pipette. Black and white bars represent mean reversal potentials ± SEM for experiments with P. knowlesi-infected rhesus and P. falciparum-infected human RBCs, respectively. The similar reversal potentials indicate similar anion selectivity ratios for these parasite-induced currents and yield mean estimates of permeability ratios relative to Cl^– of 3.8 and 0.3 for ClO₄^– and the lactate anion, respectively.

We wondered how these P. knowlesi-induced currents compare quantitativelywith those induced by P. falciparum trophozoites on human RBCs.Because both exhibit larger and more ohmic whole-cell currentsat negative imposed membrane potentials (V_m), we tallied chordconductances between –100 and 0 mV for a number of cellsin each system (20 and 21 RBCs for P. knowlesi and P. falciparum,respectively) (Fig. 2C). These chord conductances were similardespite the phylogenetic distance between both the two parasites.We did not perform statistical analyses of these chord conductancesbecause of possible systematic differences in parasite maturityand in pipette access resistances.

We also compared the ion selectivity properties of these parasite-inducedcurrents with bi-ionic experiments that created anion gradientsacross the RBC membrane while maintaining symmetrical cationconcentrations. Under these conditions, anion channels exhibitaltered current-voltage profiles with reversal potentials (E_rev)quantitatively linked to relative anion permeabilities (18).With Cl^– as the predominant extracellular anion, we foundthat intracellular ClO₄^– or lactate produced marked andopposing changes in E_rev (Fig. 2D), indicating significantlygreater permeabilities for these anions than for Na⁺. Our whole-cellstudies with both parasites indicate similar anion selectivities(ClO₄^– > Cl^– > lactate), consistent with oneor more weak binding sites for anions as they cross the hostRBC membrane (8).

We next investigated the various similarities between P. knowlesi-and P. falciparum-induced permeability changes with single-channelpatch-clamp. In P. falciparum, these changes are adequatelyaccounted for by PSAC, whose small conductance requires hypertonicsalt solutions for unambiguous detection (2). Because P. knowlesimay also induce small-conductance anion channels, we performedcell-attached patch-clamp of infected rhesus monkey RBCs in1,145 mM Cl^– solutions. Under these conditions, one ormore functional copies of a $~$ 20-pS channel were observed in approximately50% of patches on infected cells (Fig. 3A); this channel wasnever seen on uninfected rhesus monkey RBCs (23 cells). We measuredthis new channel's single-channel amplitudes, open probability,and net ion flux at a range of voltages (Fig. 3B through D)and found that its voltage dependence parallels that of themacroscopic whole-cell currents (Fig. 2B). We also measuredthe durations of open and closed events at a V_m of –100mV in prolonged single-channel recordings (Fig. 4). These dwelltime analyses indicated that the gating behavior of the P. knowlesi-inducedchannel is indistinguishable from that of the P. falciparumPSAC. Because gating, single-channel conductance, and voltagedependence constitute a functional signature, we concluded thatthe ion channels induced by these two parasites are membersof a highly conserved family. We suggest that the members ofthe PSAC family be distinguished by using the parasite speciesname as a prefix.

View larger version (26K):
[in this window]
[in a new window]
FIG. 3. Single-channel recordings on P. knowlesi-infected rhesus RBCs. (A) Cell-attached recording from an infected RBC patch with one functional channel molecule. V_m values were as indicated to the right of each trace. Scale bars represent 150 ms (horizontal) and 2.0 pA (vertical). Seal resistance was 160 G ${Omega}$ . Closed-channel levels are indicated with black dashes at both ends of each trace. Notice the fast-flickering transitions at negative V_m and the significantly fewer openings at positive voltages. (B) Single-channel current amplitudes (i) versus V_m for the P. knowlesi and P. falciparum channels (open and filled circles, respectively). Each symbol represents the mean ± SEM of multiple measurements. The solid line is the best fit of the measurements between V_m of –100 and +20 mV and corresponds to a 23-pS channel. (C) Open probability (P_o) versus V_m for the P. knowlesi channel, measured from 10 to 12 s of single-channel recording at each voltage. The solid line represents the best fit to a single exponential decay. (D) Time-averaged current (I) for a single P. knowlesi channel, calculated at each voltage as the product of measurements in panels B and C. The inward-rectifying behavior of this calculated current parallels whole-cell measurements in Fig. 2, suggesting that the P. knowlesi PSAC is the predominant conductive pathway in the infected rhesus RBC membrane.

View larger version (11K):
[in this window]
[in a new window]
FIG. 4. Open (A) and closed (B) dwell time distributions for single-channel recordings on P. knowlesi- and P. falciparum-infected cells (red and black traces, respectively). The number of events tallied in each histogram ranged from 9,700 to 35,000. The P. knowlesi and P. falciparum channels exhibit indistinguishable gating in the absence of inhibitors.

Comparative pharmacology of P. falciparum and P. knowlesi channels. Our osmotic lysis experiments suggest that the P. knowlesi PSACis inhibited by furosemide (Fig. 1). The effects of this inhibitoron the P. falciparum PSAC have been extensively studied, revealinga single nonluminal binding site with a K_m of 2.7 µM onthe channel extracellular face (2, 7). Strangely, the P. falciparumPSAC is also inhibited by dantrolene, a specific Ca²⁺ channelantagonist that does not inhibit human anion channels (Liskand Desai, submitted). We wondered whether the P. knowlesi PSACshares these pharmacological properties and used single-channelrecordings to explore possible mechanisms. With low concentrationsof each of these agents, we found clear inhibitory effects onthe P. knowlesi PSAC (Fig. 5A) that resemble the inhibitionwith intermediate kinetics described for the P. falciparum PSAC,indicating another level of conservation for this family ofchannels.

View larger version (18K):
[in this window]
[in a new window]
FIG. 5. Single-channel pharmacology and a lower furosemide affinity for the P. knowlesi PSAC. (A) Traces reflect current flow through a single P. knowlesi PSAC at V_m of –100 mV without inhibitor or with 4 µM furosemide, 10 µM furosemide, or 10 µM dantrolene (top to bottom traces, respectively). Red dashes on each side of traces reflect the closed-channel level. Scale bars represent 100 ms (horizontal) and 2.0 pA (vertical). (B and C) Osmotic lysis time courses for P. falciparum-infected human and P. knowlesi-infected rhesus RBCs in the buffered sorbitol lysis solution with 0, 4, and 10 µM furosemide (left to right traces in each panel). Direct comparison of these two groups of unprocessed transmittance measurements (%T) is possible because both panels used identical hematocrits and a common blank. Notice that lysis kinetics are inhibited to a greater extent for the P. falciparum channel at both nonsaturating furosemide concentrations. (D) Furosemide dose-response for inhibition of the P. knowlesi PSAC. Sorbitol permeability coefficients (P_s) were calculated from up to eight lysis time courses at each concentration as described previously (38) and normalized to a value of 1.0 without inhibitor. The solid line represents the best fit to y = K_m/(K_m + x). (E) Closed dwell distributions from single-molecule recordings of the P. falciparum and P. knowlesi channels (black and red traces, respectively) in the presence of 10 µM furosemide. Each histogram tallies approximately 80,000 closing events. For each profile, the peak at a time of >10 ms corresponds to the population mean for furosemide-induced closings; this peak is a direct measure of furosemide unbinding rates from the inhibited channel. Furosemide unbinds more quickly from the P. knowlesi channel, accounting for its lower affinity there.

Are there any differences between these two PSAC channels thatmay have resulted from the phylogenetic distance between P.falciparum and P. knowlesi? We used the simple osmotic lysisassay to survey channel properties and found modest but reproducibledifferences in their levels of inhibition by furosemide. Pairedexperiments with 4 µM and 10 µM furosemide consistentlyproduced less robust inhibition of sorbitol-induced osmoticlysis of P. knowlesi-infected rhesus RBCs (Fig. 5B and C). Thisdifference was statistically significant at both concentrations(seven trials for each parasite at 4 µM and at 10 µMwith P values of 0.004 and 0.007, respectively, in two-tailedStudent's t tests), suggesting a lower furosemide affinity forthe P. knowlesi PSAC. A furosemide dose response for inhibitionof lysis revealed Michaelian kinetics with an estimated K_m of5.1 ± 0.5 µM for the P. knowlesi PSAC (Fig. 5D),approximately twice that previously reported for the P. falciparumortholog. We explored this difference further with single-channelrecordings of the two channels in the presence of furosemide.Because the differences in the furosemide affinity are modest,visual examination of individual traces could not readily identifydifferences between the two channel types (not shown). Nevertheless,analysis of open and closed durations from single-channel recordingswith furosemide should be able to identify even modest differencesin affinity by quantifying a large number of channel-inhibitorinteractions. As previously described for the P. falciparumPSAC (2), furosemide did not affect the durations of the P.knowlesi PSAC openings, consistent with inhibition at a nonluminalsite (not shown). Analysis of closed durations, however, revealeda marked difference between these two related channels. Thedistribution of furosemide-induced closings exhibited a substantiallyshorter mean duration for the P. knowlesi PSAC than the P. falciparumPSAC (Fig. 5E). This shorter closed duration suggests that theP. knowlesi channel's lower affinity results from a faster rateconstant for furosemide unbinding from related but not identicalsites on the extracellular face of these channels.

DISCUSSION

Top

Discussion

Plasmodial survival and growth in the cytosolic milieu of RBCsrequires adaptations that allow the trafficking of nutrientprecursors from host plasma into parasite compartments. Workfrom multiple groups suggests that several nutrients have insufficientpermeability in uninfected RBCs to sustain the known high ratesof utilization by the intracellular parasite. Because PSAC isbroadly permeable to these precursors and is the predominantconductive pathway in the P. falciparum-infected human RBC membrane(2), it may function as an essential first step in parasitenutrient acquisition. In our model (8), required nutrients enterRBC cytosol via diffusion through PSAC, cross the parasitophorousvacuole membrane through distinct nonselective channels localizedin that membrane (9, 10), and are taken up via one of severalhighly specific carriers in the parasite plasma membrane (4,29, 39). With the identification of a lactate carrier in theparasite plasma membrane (13), this and other soluble metabolicwaste products should be able to exit the infected RBC via diffusionin the opposite direction through the nonselective parasitophorousvacuole membrane channel and PSAC.

Other plasmodial species may infect primate, rodent, or avianRBCs but still face a similar dilemma because each of thesehost cells has low intrinsic permeability to metabolic precursors.Here, we explored this dilemma with single-channel and whole-cellpatch-clamp methods and identified, for the first time, an unusualion channel induced by a parasite other than P. falciparum.Because the gating, conductance, selectivity, and pharmacologicalproperties of the P. knowlesi-induced ion channel show markedsimilarities to those of the P. falciparum PSAC, these channelsconstitute a family of functional PSAC orthologs that can beeasily distinguished from unrelated ion channels.

Although these PSAC channels are highly conserved, we identifieda modest but statistically significant phenotypic differencein the furosemide affinity of the P. knowlesi and P. falciparumorthologs. Previous functional studies with furosemide implicatea binding pocket on the extracellular face of the channel thatis distinct from the pore used by permeating solutes (2, 7).The difference in affinity presumably results from differencesat one or more residues in the channel proteins of these parasites.

Studies from multiple groups have produced conflicting observationson whether PSAC represents a parasite-encoded protein traffickedout to the host RBC membrane (2) or a host protein activatedby specific or nonspecific parasite activities (12, 21, 37).How do our findings with P. knowlesi-infected rhesus erythrocytescontribute to this debate? Although they do not directly addressthe genetic basis of PSAC functional activity, our findingsrequire a high level of structural and functional conservationin this family of channels because estimates of the time tothe most recent common ancestor are $≥$ 25 million years for humansand rhesus monkeys (28) and for P. falciparum and P. knowlesi(14). This level of conservation in spite of the large phylogeneticdistances provides compelling yet circumstantial evidence againsta protein with an inconsequential physiological role.

Consistent with this proposal, a number of studies from multiplegroups used isotope uptake and osmotic lysis measurements todemonstrate increased permeabilities of primate (26, 27), avian(32), and rodent (16, 19, 33) erythrocytes after infection withvarious plasmodia. Known PSAC inhibitors, whenever tested inthese other host-parasite pairs, invariably abrogated theseincreases, further implicating high-level conservation withinthis channel family (Fig. 6).

View larger version (27K):
[in this window]
[in a new window]
FIG. 6. Phylogenetic tree for major plasmodial species. The tree was drawn by combining several genetic analyses (15, 30, 31). The relative phylogeny of these parasites is indicted by the connector lines, the lengths of which do not reflect bootstrap scores. Parasites with electrophysiological evidence of PSAC are marked with arrows. Asterisks reflect parasites known to induce increased tracer uptake in previous studies (16, 19, 26, 32, 33), which are consistent with PSAC orthologs in these species as well. The distance between P. knowlesi and P. falciparum, combined with reports of increased host RBC permeability for various other parasites, suggests that all plasmodia have functional PSAC orthologs.

A detailed study of [³H]choline uptake by P. knowlesi-infectedRBCs (3) failed to detect PSAC-like uptake, seemingly in directconflict with our findings using the same parasite and hostspecies. In that study, parasite-induced increases in [³H]cholineuptake were instead attributed to upregulation of endogenouscarriers. Experimental differences may well account for thediscrepancies with our findings. Their uptake measurements wererestricted to choline concentrations of $≤$ 100 µM, wherehigh-affinity carrier-mediated uptake predominates over uptakevia lower-affinity channels such as PSAC. Moreover, they usedsubtraction of uptake in 1.2 mM [³H]choline to correct for extracellulartrapping. Because flux through ion channels generally saturatesonly at much higher solute concentrations, this subtractionsystematically excluded detection of channel-mediated uptakein their study. An additional experimental difference, use ofin vivo infected monkeys in contrast to our use of parasitescultured in erythrocytes from uninfected monkeys, may also influencethe observed upregulation of the endogenous choline carrier.

It is debated whether anion channels other than PSAC exist inP. falciparum-infected human RBCs (12, 21, 32). Some of theseproposed other channels, collectively known as "new permeationpathways," exhibit markedly differing voltage dependences andother biophysical properties. We found no evidence for these"new permeation pathways" on P. knowlesi-infected cells, suggestingthat this parasite induces only PSAC-like anion conductances.

PSAC has been recognized as a potentially important and noveltarget for future antimalarial drug development. Our demonstrationof a functional PSAC ortholog on P. knowlesi-infected rhesusRBCs has two important ramifications for drug development programsthat target PSAC. The similar pharmacology and mechanisms ofchannel inhibition indicate that animal malaria models willbe suitable for in vivo evaluation of PSAC antagonists identifiedusing laboratory isolates of P. falciparum. These findings alsosuggest that the PSAC protein is a highly constrained integralmembrane protein because even the biophysical properties ofpermeation are highly conserved between divergent parasites.We suspect mutations in the channel are constrained by the needto maintain PSAC's unparalleled selectivity properties and rateof nutrient uptake. In light of these constraints, resistanceto antimalarials targeting PSAC may be harder to acquire thanresistance to drugs against less constrained targets insidethe parasite compartment.

ACKNOWLEDGMENTS

We thank A. Thomas for providing the P. knowlesi isolate andfor assistance with establishing culture. We thank B. Keeganand L. Lambert for regular provision of blood from rhesus monkeys.

This research was supported by the Intramural Research Programof the NIH, NIAID.

FOOTNOTES

* Corresponding author. Mailing address: NIAID/NIH, Room 3W-01, 12735 Twinbrook Parkway, Rockville, MD 20852. Phone: (301) 435-7552. Fax: (301) 402-2201. E-mail: sdesai{at}niaid.nih.gov.

REFERENCES

Top