Plasmodial Surface Anion Channel-independent Phloridzin Resistance in Plasmodium falciparum*

The plasmodial surface anion channel (PSAC) is an unusual ion channel induced on the human red blood cell membrane after infection with the malaria parasite, Plasmodium falciparum. Because PSAC is permeant to small metabolic precursors essential for parasite growth and is present on red blood cells infected with geographically divergent parasite isolates, it may be an ideal target for future antimalarial development. Here, we used chemically induced mutagenesis and known PSAC antagonists that inhibit in vitro parasite growth to examine whether resistance mutations in PSAC can be readily induced. Stable mutants resistant to phloridzin were generated and selected within 3 weeks after treatment with 1-methyl-3-nitro-1-nitrosoguanidine. These mutants were evaluated with osmotic lysis and electrophysiological transport assays, which indicate that PSAC inhibition by phloridzin is complex with at least two different modes of inhibition. Mutants resistant to the growth inhibitory effects of phloridzin expressed PSAC activity indistinguishable from that on sensitive parasites, indicating selection of resistance via mutations in one or more other parasite targets. Failure to induce mutations in PSAC activity is consistent with a highly constrained channel protein less susceptible to resistance mutations; whether this protein is parasite- or host-encoded remains to be determined.

through the combined effects of extracellular pore mouth cationic amines to repel extracellular cations and at least one weak binding site in its pore to favor bulky dehydrated ions over smaller ions (9).
Another unusual feature of PSAC is its small single channel conductance of ϳ20 pS in 1.15 molar Cl Ϫ solutions (10); most other channels with broad selectivity profiles exhibit larger single channel conductances. A channel's chord conductance, calculated from measured current amplitudes at a range of imposed membrane potentials, quantitatively corresponds to the number of ions passing through the open channel per unit time. In terms of Eyring rate theory, this channel's small conductance reflects the one or more energetic barriers that each solute molecule must overcome to traverse the length of the channel pore. These barriers may function to additionally fine tune the selectivity properties of PSAC.
Finally, despite the nearly invariant membrane potential of human RBCs after infection, permeation through PSAC is voltage-dependent. Both whole cell and single channel recordings reveal significantly smaller absolute currents at membrane potentials (V m ) of ϩ100 millivolts than of Ϫ100 millivolts despite equal but opposite driving forces for ion movement. Historically, channels exhibiting this phenomenon were often referred to as "inward rectifying" voltage-dependent channels. Although voltage-dependent channels are most well known as mediators of electrical signals on excitable cells, they are also present on many non-excitable cells (11). The voltage dependence of PSAC results from a reduced frequency and duration of openings at increasingly positive V m values, presumably because of charged residues on the channel protein that move during transitions between the open and closed states. Although the physiological relevance of this voltage dependence is unknown, it represents an important functional marker of PSAC activity. There are other complex and unusual features of PSAC gating, the process of opening and closing, that suggest additional structural constraints on the protein (12).
Because of the unusual properties of PSAC and because of polymorphisms in channel gating apparent when RBCs from a single human donor are infected with genetically distinct parasite isolates (10), we proposed that PSAC is a parasite-encoded ion channel. Although several reports have suggested that PSAC activity results from the modification of host RBC proteins (13)(14)(15), none of these mutually exclusive proposals has been independently confirmed. In particular, we were unable to reproduce these reports despite intensive efforts (10).
Homology searches using known anion channel genes have not revealed obvious candidates for PSAC gene(s) in the completed Plasmodium falciparum genome sequence (16). It appears that PSAC genes differ sufficiently from known channels to render bioinformatic approaches ineffective. Here, we used an alternative approach with the goal of identifying these elusive genes. We used the potent mutagen 1-methyl-3-nitro-1-nitrosoguanidine (MNNG) to increase the frequency of random mutations in parasite genes. We followed mutagenesis with the selection of in vitro resistance to growth inhibition by known PSAC antagonists. These antagonists are believed to kill parasites by preventing PSAC-mediated uptake of essential nutrients. This approach selected for unambiguous and stable resistance to growth inhibition by 100 M phloridzin, a nonspecific antagonist of PSAC (17). Osmotic lysis and electrophysiological studies with phloridzin revealed that resistance did not result from either an altered mechanism or an affinity for PSAC inhibition. Effects of phloridzin on other parasite activities are implicated.
We also used single channel studies to determine that phloridzin inhibition of PSAC is complex, with at least two separate mechanisms. Comparison to inhibition by furosemide revealed additional important insights into the conformational changes required for PSAC gating. These detailed studies should help guide the design of novel and specific PSAC antagonists that may become future antimalarial drugs.

MATERIALS AND METHODS
Chemical Mutagenesis-Phloridzin-resistant mutants were generated and selected using a protocol based on that developed by Inselburg (18). Up to 2 ϫ 10 9 ring stage parasites (Indo 1 strain) were synchronized with the sorbitol lysis method (19) and cultured for 24 h in the presence of 150 g/ml aphidicolin to produce a DNA replication block and to maximize the mutagenic effect of MNNG. After washing, these trophozoite-infected RBCs were incubated in culture medium (RPMI 1640 supplemented with 50 mg/liter hypoxanthine and 10% human serum) for 2 h before adding 0.5-1.5 g/ml MNNG from a freshly prepared Me 2 SO stock of 1.5 mg/ml. Parasites were cultured at 37°C for 24 h, washed to remove MNNG, and followed with daily microscopic examination of Giemsa-stained smears. Under our experimental conditions, MNNG treatment killed Ͼ90% of the parasites through the random and high-frequency introduction of mutations throughout the genome. Surviving parasites were cultured for 3-14 days to allow clonal increases in desired mutants. When parasitemias had recovered to Ն1%, 100 M phloridzin was added as a selective pressure. Phloridzin was typically removed or applied at lower concentrations for several days when parasites were not visible upon microscopic examination of smears to allow for the selection of incompletely resistant mutants. In separate experiments, other PSAC antagonists such as 100 M furosemide (10), 100 M glibenclamide (20), or 50 M 5-nitro-2-(3-phenylpropylamino)-benzoic acid (2) were also used for selection but have not, to date, generated unambiguous drug-resistant mutants. In light of the probable role of PSAC in nutrient acquisition (1), some selections were also performed in media with reduced concentrations of essential nutrients and only 2% dialyzed serum to improve the likelihood of identifying PSAC mutants (not shown).
Growth Inhibition Studies-Wild-type and the phloridzin-resistant mutant parasites were synchronized by sorbitol lysis, subcultured to 0.5% parasitemia, and grown at 2% hematocrit in culture medium with and without 100 M phloridzin. At 24-h intervals the culture medium was replaced. At these same intervals an aliquot was removed, washed, and grown without phloridzin in culture medium containing [ 3 H]hypoxanthine (25 Ci/ml) and no unlabeled hypoxanthine. Incorporation of [ 3 H]hypoxanthine into parasite DNA and RNA, a direct measure of parasite viability (21), was measured in 96-well format using a plate reader and standard ␤-scintillation counting.
Osmotic Lysis Assays-The kinetics of infected RBC osmotic lysis in sorbitol solutions were followed as described previously (22). In brief, trophozoite-infected RBCs were grown by standard culture methods, enriched to 95-99% by Percoll/sorbitol separation (23), washed in phosphate-buffered saline (150 mM NaCl, 20 mM sodium phosphate, pH 7.5), and resuspended to 0.5% hematocrit in sorbitol lysis solution (280 mM sorbitol, 20 mM sodium HEPES, and 0.1 mg/ml bovine serum albumin, pH 7.4) at 37°C. The transmittance of 700-nm light through this cell suspension was then followed continuously as a marker of osmotic swelling and lysis. This method produces estimates of permeation rates through PSAC that quantitatively match those obtained with both radioisotope flux and patch clamp (22). Because phloridzin inhibits host RBC membrane permeability through one or more binding sites on the intracellular face of PSAC (1), some experiments included a preincubation in the final phloridzin concentration before the addition of sorbitol lysis solution; the resulting levels of inhibition were comparable with those without pre-incubation, indicating that phloridzin uptake into intact infected RBCs is not rate-limiting.
Electrophysiology-Cell-attached patch clamp recordings of infected RBCs were obtained as described previously (1). Unless otherwise indicated, these experiments utilized symmetrical bath and pipette solutions of 1000 mM choline chloride, 115 mM NaCl, 10 mM MgCl 2 , 5 mM CaCl 2 , and 20 mM sodium HEPES, pH 7.4. Pipettes with tip diameters of Ͻ0.5 m and resistances of 1-4 megaohms were pulled from quartz glass. Recordings were filtered at 5 kHz with an eight-pole Bessel filter and digitized at 100 kHz. Seal resistances were typically Ն100 gigaohms.
To determine mean single channel open probabilities (P o ) for wildtype and phloridzin-resistant mutants, we collected single channel molecule recordings from patches on RBCs infected with each parasite in the presence of 0, 25, or 100 M phloridzin added to both bath and pipette. For each condition, we used a home-written code that identifies transitions between channel open and closed states by the 50% threshold-crossing technique and normalized the results to the measured P o in the absence of phloridzin, 43 Ϯ 2% (10).
Dwell time distributions were obtained using a code that detects mid-threshold crossings, uses linear interpolation of adjacent sample times (24), and corrects for a Gaussian filter rise time of 66.4 s as described (25). Histogram ordinate values were normalized to the percentage of the total number of events detected under each condition and are displayed on a square root logarithmic plot (24).
Burst analysis was performed with these corrected durations by defining inhibitor-induced closings as periods of inactivity Ն10 ms. Using this threshold, we tallied the durations of opening bursts defined as clusters of openings between consecutive inhibitor-induced closings. The resulting distributions were normalized and displayed in the same manner as the dwell time distributions. This approach to defining opening bursts is justified by the Ͼ100-fold difference between the mean durations of block by phloridzin or furosemide and those of intrinsic closings, as analyzed previously (12). Varying the threshold that defines inhibitor-induced closings within the range of 5 and 20 ms did not produce significant changes in the burst distributions (not shown). Fig. 1A shows our algorithm for generating and selecting mutants resistant to PSAC antagonists. As reasoned previously (18), ring stage parasite cultures were synchronized by growth with aphidicolin, which blocks DNA replication and stalls parasites at the early trophozoite stage. When it is removed, DNA replication ensues with a more synchronized population of parasites. At this point, we added MNNG (Fig. 1A, green-shaded area). Because MNNG is unstable in aqueous solutions and may have other deleterious effects on cell cultures (26), timing its addition with the onset of DNA replication, achieved through the use of aphidicolin, should maximize its mutagenic potential. We varied the MNNG concentration to obtain ϳ95% killing of parasites, as judged by microscopic examination after 24 h. At this level there should be both a high frequency of introduced mutations and yet sufficient numbers of viable parasites to allow identification of mutants. Surviving parasites were then cultured for several days before applying selective pressure with 100 M phloridzin (Fig. 1A, red-shaded areas). With this algorithm, resistant parasites that exhibited replication rates of 5-7-fold every 48 h in the continued presence of phloridzin were identified within 3 weeks.

RESULTS
We tested whether these mutant parasites represented stable genetic changes by culturing them under standard conditions without phloridzin for 5 weeks before performing the experiment in Fig. 1B. In the absence of phloridzin, growth rates of the selected mutants and the parent wildtype parasites were indistinguishable. In contrast, the addition of 100 M phloridzin to the culture revealed a marked difference; within 3 days wild-type parasites were essentially eliminated, but the growth of the mutant continued unabated. The prolonged culture of these parasites in the absence of phloridzin before this experiment indicates that phloridzin resistance resulted from the acquisition of stable mutations in one or more parasite genes.
We then evaluated whether resistance resulted from a lowered phloridzin affinity for PSAC inhibition. We used a recently developed a light-scattering method that tracks the kinetics of osmotic lysis in sorbitol and produces quantitative estimates of PSAC permeability (22) . Fig. 2, A and B, shows the resulting lysis time courses for wild-type and mutant parasites, respectively. The rates of lysis without and with a range of phloridzin concentrations were not statistically different within the two populations of parasites.
Because there may be subtle changes in the gating, conductance, or interaction of PSAC with phloridzin that are not detectable with the macroscopic lysis assay, we performed cell-attached patch clamp of infected cells and obtained single PSAC recordings on both parasite isolates. We successfully identified single PSAC activity on mutant parasites (Fig. 2C); these PSAC molecules exhibited voltage-dependent gating and conductance resembling those seen previously on unmutagenized parasites (1). Based on the levels of inhibition seen in the osmotic lysis experiments (Fig. 2, A and B), we chose two phloridzin concentrations (25 and 100 M) for inclusion in separate single channel studies with both wild-type and mutant parasites (Fig. 2D, left and right columns, respectively). Phloridzin produced concentrationdependent reductions in PSAC gating on RBCs infected with each of these two parasites.
We tallied our osmotic lysis experiments with a range of phloridzin concentrations and normalized them relative to the lysis rates without inhibitor (Fig. 2E, open symbols). We also collected between 94 and 254 s of single PSAC recordings at each phloridzin concentration on both wild-type and mutant parasites, determined channel open probability at a V m of Ϫ100 mV, and normalized these results to the open probability without phloridzin. When these independent transport measurements were plotted together, the effects of phloridzin on both sorbitol-mediated lysis and single PSAC recordings were concordant, consistent with a single channel for the uptake of diverse solutes (10). We also found that the dose response for the inhibition of phloridzin-resistant mutants was not measurably changed from that of wild-type parasites (Fig. 2E, compare triangles to circles); both were adequately fitted by a decaying Hill equation with a dissociation constant, K 0.5 , of 24 M. Thus, resistance to the growth inhibitory effects of phloridzin was mediated through mutations in one or more parasite genes not responsible for the expression of PSAC.
Previous whole cell patch clamp with phloridzin added to either the cytosolic or extracellular compartments revealed that inhibition by phloridzin occurs primarily at the intracellular face of PSAC (1). Here, we analyzed our single channel recordings to examine the mechanism of this inhibition. The addition of either 25 or 100 M phloridzin to both the bath and pipette solutions did not affect the durations of PSAC openings (Fig. 3A), which are adequately described by a single exponentially decaying state with a mean duration of 0.2-0.4 ms (12). Instead, phloridzin altered the distribution of closed times by adding a population of closings with a mean duration of 46 Ϯ 2 ms (Fig. 3B, smooth curve). The mean duration of these imposed closings was not affected by changes in phloridzin concentration. This collection of observations is conservatively explained by simple allosteric inhibition of PSAC through a single phloridzin binding site on the channel protein's intracellular face.
Our analysis of dwell durations in the presence of phloridzin shows marked similarities to the recent characterization of PSAC inhibition by furosemide, which also produces allosteric inhibition at a single site with a mean block duration of 85 ms. Because furosemide has an ϳ10-fold higher affinity for PSAC inhibition (K 0.5 ϭ 2.7 M) (10), we were surprised by the similar mean block durations for furosemide and phloridzin. It is generally assumed that access to binding sites on ion channels is diffusion-limited. If furosemide and phloridzin both satisfied this assumption, then the mean closed duration of phloridzin should have been some 10 times shorter than that of furosemide because of the direct inverse relationship between the K 0.5 for an inhibitor and the measured mean closed duration in single channel recordings.
If the model of a single reversible binding site for each inhibitor is correct, the only explanation for this discrepancy would be that access to the binding site on PSAC is not diffusion-limited for at least one of the two inhibitors. Specifically, the rate constant for phloridzin binding would need to be significantly lower than that for furosemide binding to compensate for their similar unbinding rates. Visual examination of recordings in the presence of either 25 M phloridzin or 25 M furosemide (Fig. 4A, left and right columns, respectively) reveals that both compounds produce relatively long and similar periods of inhibition, apparent as regions devoid of downward channel openings. Although these compounds impose similar closed channel durations, channel openings in their presence differ markedly. Clusters of channel openings in the presence of phloridzin were visibly longer than those with the same concentration of furosemide. This observation, made in the FIG. 1. Selection of P. falciparum parasites resistant to the growth inhibitory effects of phloridzin. A, algorithm for mutagenesis and selection of phloridzin resistance. Parasite cultures were mutagenized by the sequential overnight application of 150 g/ml aphidicolin and 0.75 g/ml MNNG (blue-and green-shaded areas, respectively). When the surviving parasites had recovered to Ͼ1% parasitemia (filled circles), intermittent exposure to 100 M phloridzin (red-shaded areas) was used to select for resistance. By day 30, parasitemias increased rapidly despite continued exposure to phloridzin; the culture parasitemia was reduced by the addition of fresh RBCs on day 33 to prevent overgrowth (black arrow). B, growth rates of unmutagenized or selected mutant parasites (open and filled symbols, respectively) in the presence or absence of 100 M phloridzin (circles and triangles, respectively). Because this experiment was performed after culturing the mutant parasites for 5 weeks in the absence of phloridzin, resistance reflected stable changes in the parasite genome. context of the identical concentrations of the two inhibitors, is qualitatively consistent with slower binding rates for PSACphloridzin interactions than for PSAC-furosemide interactions.
We next quantified this observation to test whether it can account for the differing macroscopic affinities of phloridzin and furosemide. We defined bursts of PSAC activity as clusters of openings flanked by closings of Ն10 ms. This cutoff value is some 10-fold greater than the longest mean duration for intrinsic PSAC closings (12) and substantially less than the mean inhibition durations imposed by phloridzin or furosemide. Thus, this cutoff value minimizes the erroneous assignment of intrinsic and inhibitor-associated bursts. Fig. 4B shows the resulting distribution of burst durations; each of the 8,534 bursts tallied here results from measurable PSAC activity between consecutive inhibition events by phloridzin or furosemide (red and green histograms, respectively).
The simplest scheme for interaction between an inhibitor (I) and PSAC is shown in Reaction 1, where P represents the unbound channel and IP represents a bound and inhibited channel; k on and k off represent the forward and reverse rate constants for the binding process. If the interactions of the two inhibitors with PSAC obey this scheme, then the burst distributions shown in Fig. 4B should follow Equation 1, in which t represents the duration of individual bursts and a and b are constants. The distribution's mean value is given by exp(b) and is visible as a peak on this plot (12,24). We found that the burst distribution for 25 M phloridzin was well fitted by Equation 1 (Fig. 4B, smooth red line) with a mean value of 21 Ϯ 0.6 ms. The forward rate constant, k on , is then inversely related to this mean value. With 25 M furosemide, our algorithm for identifying bursts between consecutive inhibition events produced a distribution peaking at substantially shorter durations (Fig. 4B, green histogram). Here, the best fit to Equation 1 (Fig. 4B, smooth green line) yielded a mean burst duration of 2.9 ms. Because this estimate is nearly 10-fold less than that for phloridzin, access to their respective binding sites predominantly accounts for the differing affinities of furosemide and phloridzin. In contrast, the mean durations of inhibitor-induced closings (Fig. 3B), a direct measure of k off , plays a relatively minor role in determining the relative affinities of these two agents. Our model for interactions between PSAC and these two inhibitors, which incorporates open and closed durations in single channel recordings as well as previous whole cell measurements, is illustrated in Fig. 5.
Interestingly, the bursts of activity seen with 25 M furosemide were not well fitted by Equation 1 (Fig. 4B, green histogram and  smooth curve); the bursts instead required two or more on-rate constants with values too similar to one another to be easily resolved (not shown). Because unbinding is adequately fitted by a single k off value (12), these separate but similar forward rate constants for binding presumably represent independent routes by which furosemide can reach a single pocket on the channel protein to achieve its inhibitory effect. Fig. 6 shows a separate complex effect of phloridzin on PSAC activity. We found that continued recording in the presence of either 25 or 100 M phloridzin invariably leads to an altered single channel phenotype. The already small conductance of PSAC was abruptly reduced by ϳ65% (Fig. 6). This clearly different activity typically lasted for seconds to 1-2 min before complete and irreversible loss of all channel activity from these patches. The possibility that this activity represents a channel distinct from PSAC was excluded by the tight link between its appearance and the disappearance of typical PSAC gating and conductance, its PSAC-like inward-rectifying voltage-dependence, and its transitory nature. Additional experimentation will be needed to characterize this behavior.

DISCUSSION
The killing of parasites by nearly every known PSAC antagonist supports an essential role for this unique parasite-induced channel. In light of the high permeability of PSAC to diverse solutes needed for parasite growth, the most appealing proposal is that this channel serves as the first step in a sequential diffusive pathway for nutrient acquisition by the intracellular parasite (1). In this model, waste products such as lactate Ϫ , a toxic metabolite made by the parasite, can also be removed from the cell by transport in the reverse direction, first through a lactate carrier on the parasite plasma membrane (27) and then through giant non-selective channels on the parasitophorous vacuolar membrane (28,29) and across the host membrane through PSAC. Because all of the steps in this sequential pathway are passive, the direction of flow is governed by solute electrochemical gradients weighted by the gradients of coupled solutes at the parasite membrane if carriermediated. Nutrients, present at higher concentrations in serum, would enter the cell; metabolites, more abundant at their site of production in parasite cytosol, would move in the opposite direction.
Alternative proposals for PSAC function also exist. In one study (30), a role in facilitating parasite egress from the RBC at the end of its 48-h intracellular cycle was proposed based on mathematical modeling of PSAC-mediated Na ϩ uptake, which eventually leads to swelling and osmotic lysis of the infected RBC. Using several assumptions, this study calculated that the low but finite Na ϩ permeability of PSAC would produce osmotic lysis at 48 h, matching the duration of the intraerythrocytic parasite cycle. In this model, the physiological role of PSAC would be to promote parasite egress from the RBC via a narrowly defined Na ϩ permeability; its permeability to nutrient solutes such as amino acids, purines, and vitamins would then be merely an epiphenomenon. In addition to other problems with this proposal, various studies of rupture and merozoite release indicate that proteases are required and that parasitophorous vacuolar membrane lysis precedes lysis of the host RBC membrane (31,32), findings that are inconsistent with a role for Na ϩ -mediated osmotic lysis.
Although a PSAC role in nutrient acquisition seems the most reasonable (1), there is still no direct evidence for any proposal. Here, we used chemical mutagenesis and selection of resistance to PSAC antagonists to search for such evidence. Although we attempted to find mutants resistant to four separate PSAC antagonists, only experiments with phloridzin have produced successful selections to date. Because these mutants exhibited PSAC activity with unaffected gating, conductance, and affinity for phloridzin, in vitro killing of parasites by phloridzin is likely mediated by one or more other targets within the infected RBC complex.
We consider phloridzin to be the least specific of the commonly studied PSAC antagonists. Not only is it or its congener, phloretin, known to inhibit a diverse collection of channels and carriers, but these compounds induce unambiguous changes in artificial lipid bilayer permeability due to membrane dipole effects and disordering of lipid tail groups (33). With this level of poor specificity, it is not surprising that phloridzin has other essential targets within the infected RBC.
Our study raises some important caveats about the interpretation of parasite growth inhibition studies. Prior to these findings, workers in the PSAC field accepted that parasite killing by phloridzin was primarily mediated via its actions on PSAC; there was concordance between its K 0.5 for channel inhibition (Fig. 2E) and its IC 50 for parasite growth inhibition (17). Because the selected mutants had growth rates unaffected by 100 M phloridzin despite an unchanged K 0.5 for PSAC inhibition, this concordance appears to be only coincidental. In light of the nonlinear relationship between nutrient uptake and the successful production of viable progeny, the quantitative concordance of these half-maximal inhibitory values should have been considered weak evidence even without our studies.
Random mutagenesis, as performed here, represents an important check on the drug development concept that a given compound lacks off-target effects. In principal, it can also be used to gauge how easily important parasite targets can acquire resistance mutations. For example, resistance to chloroquine, a safe and inexpensive antimalarial drug, can be primarily attributed to mutations in PfCRT, a membrane protein of unknown function. Mutations in a single locus of FIG. 5. Model of PSAC inhibition by furosemide (F) and phloridzin (P). Both agents inhibit PSAC through allosteric mechanisms that do not involve direct block of the channel pore. Whereas furosemide's binding site is on the channel's extracellular face, phloridzin's is on the intracellular side (bottom). The similar off-rate and markedly differing on-rate constants for these agents are represented by the lengths of the arrows. This altered state is visible for the remainder of the upper trace and throughout the lower trace. This phenomenon was seen with both wild-type and phloridzin-resistant parasites; it was seen with phloridzin, but never with furosemide. Closed channel levels are marked with dashes to the right and left of each trace; scale bars represents 100 ms and 2 pA. this protein have been successfully generated by selection with lethal doses of chloroquine applied to in vitro cultures (34); mutations at this locus (Lys-76) proved sufficient to convert at least one chloroquine-sensitive parasite into isolates resistant to clinically achievable doses. Targets such as PSAC that consistently fail to be mutated with this approach may have highly constrained structures and be more desirable to drug development programs.
Our inability, to date, to generate mutants with chemical mutagenesis could also be explained if PSAC activity results from a parasite-activated human protein on the RBC membrane. Although the unusual functional properties of PSAC and the presence of biophysical polymorphisms linked to parasite genotype (10) both suggest a parasite-encoded PSAC, methods other than our mutagenesis algorithm may be needed to definitively resolve this uncertainty.
PSAC inhibition by 100 M phloridzin is not sufficient to produce growth inhibition, because growth of the mutants was unabated despite unchanged action on PSAC. Under our nonphysiological experimental conditions, this concentration produces a 75% reduction in sorbitol and Cl Ϫ fluxes. Inhibition under culture conditions may be substantially lower depending on phloridzin's degree of adsorption by serum lipids and its metabolism by RBC or parasite activities. Although we anticipate that work with higher phloridzin concentrations applied to resistant parasites would reveal growth inhibition due to loss of PSAC-mediated nutrient uptake, such experiments would probably be complicated by the additional off-target effects of this nonspecific inhibitor. Instead, antagonists that are highly specific for PSAC are needed. Indeed, the inability to generate parasites resistant to 5-nitro-2-(3-phenylpropylamino)-benzoic acid, furosemide, or glybenclamide, considered to be somewhat more specific PSAC antagonists, support the position of PSAC as an attractive drug target.
Our chemical mutagenesis algorithm offers a powerful way to study parasite-encoded phenotypes in the infected RBC complex. Because the mature human RBC lacks DNA and is not affected by MNNG, mutagenesis should be able to dissect the relative contributions of host and parasite factors to a phenotype. It may have more potential utility in P. falciparum than in other model organisms for two important reasons. First, gene-specific transformations remain difficult in P. falciparum despite extensive efforts by various groups (35). Reasons for this include the parasite's relatively slow replication rate (44 -48 h) and the incalculably low transformation efficiency (36). Second, when compared with other model organisms, parasite cell biologists have identified an unsatisfactorily small number of phenotypes unique to P. falciparum. Without such clear phenotypes and knowledge of their genetic basis, random mutagenic approaches offer a tractable alternative.
We have also used single channel recordings with two PSAC antagonists to probe how the inhibition of PSAC occurs. Inhibition by phloridzin is complex, with at least two different components. One component superficially resembles the recently described allosteric inhibition by furosemide. A second component of inhibition was apparent with prolonged recording of single channels; similar complexity has not been seen with other PSAC antagonists previously. The presence of two components of inhibition may account for the failure of simple Michaelis-Menten kinetics to fit phloridzin's dose response for PSAC inhibition (Fig. 2E).
Analysis of the primary allosteric component with measurement of both the durations of single channel inhibition and the intervening bursts of PSAC activity indicate that phloridzin's access to its binding site is slower than that expected from diffusion-only considerations. In light of its ability to intercalate into lipid bilayers (37), one explanation for this slow access would be that its intracellular binding site can only be reached via lateral diffusion in the inner membrane leaflet. As successfully carried out for other ion channels, additional functional studies with inhibitors should not only provide further insights into PSAC structure but may play a central role in the design and implementation of improved chemotherapeutic agents.