Human trypanolytic factor APOL1 forms pH-gated cation-selective channels in planar lipid bilayers: Relevance to trypanosome lysis

Significance

African trypanosomes are parasites that can cause African sleeping sickness in humans. Host defense against some of these is provided by the human serum factor apolipoprotein L-1 (APOL1), which causes swelling and lysis of susceptible trypanosomes. Lysis follows uptake of APOL1 into acidic parasite endosomes and is thought to involve ion influx across the plasma membrane. In this paper we show that, after interaction of APOL1 with lipid bilayers at acidic pH, subsequent pH neutralization triggers the opening of pH-gated channels that selectively conduct cations across the bilayer. Based on these results, we propose a mechanism of trypanosome lysis that involves endocytic recycling of APOL1 and the opening of APOL1-induced cation-selective channels, at neutral pH, in the parasite plasma membrane.

Abstract

Apolipoprotein L-1 (APOL1), the trypanolytic factor of human serum, can lyse several African trypanosome species including Trypanosoma brucei brucei, but not the human-infective pathogens T. brucei rhodesiense and T. brucei gambiense, which are resistant to lysis by human serum. Lysis follows the uptake of APOL1 into acidic endosomes and is apparently caused by colloid-osmotic swelling due to an increased ion permeability of the plasma membrane. Here we demonstrate that nanogram quantities of full-length recombinant APOL1 induce ideally cation-selective macroscopic conductances in planar lipid bilayers. The conductances were highly sensitive to pH: their induction required acidic pH (pH 5.3), but their magnitude could be increased 3,000-fold upon alkalinization of the milieu (pK_a = 7.1). We show that this phenomenon can be attributed to the association of APOL1 with the bilayer at acidic pH, followed by the opening of APOL1-induced cation-selective channels upon pH neutralization. Furthermore, the conductance increase at neutral pH (but not membrane association at acidic pH) was prevented by the interaction of APOL1 with the serum resistance-associated protein, which is produced by T. brucei rhodesiense and prevents trypanosome lysis by APOL1. These data are consistent with a model of lysis that involves endocytic recycling of APOL1 and the formation of cation-selective channels, at neutral pH, in the parasite plasma membrane.

The serum of some primates, including humans, can lyse several African trypanosome parasites such as Trypanosoma brucei brucei (T. b. brucei), whereas the human-infective pathogens T. b. rhodesiense and T. b. gambiense are resistant to human serum. Trypanolytic activity is split between two subfractions of high-density lipoprotein, called trypanosome lytic factors (TLF1 and TLF2), that both contain haptoglobin-related protein (Hpr) and apolipoprotein L-1 (APOL1) (1–4). Lysis follows the endocytosis of TLF by the parasite and is blocked by weak bases, indicating that acidification of endosomes and/or the lysosome is required (5, 6). Recent evidence confirms that the parasite vacuolar H⁺-ATPase endosomal acidification machinery is required for lysis by human serum (7). Receptor-mediated endocytosis of TLF1 is driven by a parasite haptoglobin–hemoglobin receptor, which binds Hpr–hemoglobin complexes, whereas TLF2 uptake occurs via an unknown, Hpr-independent mechanism (4, 6, 8, 9).

Recombinant APOL1 is a trypanolytic substitute for whole TLF particles, although its uptake must presumably rely on pinocytosis, and lysis occurs with delayed kinetics (10). In addition, APOL1 transgenesis in mice (which normally lack both APOL1 and Hpr-encoding genes) is sufficient to prevent infection with human serum-sensitive, but not human serum-resistant, trypanosomes (11). Hpr–hemoglobin, in the context of TLF1 particles, may also contribute directly to killing via the generation of oxygen free-radicals in the parasite lysosome, particularly under conditions of intravascular hemolysis (12).

Whereas multifactorial mechanisms were proposed to account for resistance of T. b. gambiense to human serum (13–16), resistance in the case of T. b. rhodesiense is determined by the parasite’s serum resistance-associated (SRA) protein, which localizes to parasite endosomes and binds to APOL1 under acidic conditions (10, 17). Primate APOL1 orthologs, and two African human APOL1 variants, contain mutations in the C-terminal SRA-binding site that reduce binding affinity for SRA, allowing these variants to lyse SRA-producing trypanosomes (18–20). Two such human variants have been identified: the first, G1, is defined by two point mutations (S342G and I384M), one of which (I384M) was found to reduce binding affinity for SRA; the second, G2, is defined by a two-residue deletion (N388 and Y389) that prevents SRA binding (19). However, the G1- and G2-coding variants, found only among individuals of recent African ancestry, are strongly associated with chronic kidney disease in African Americans (20, 21) and are thought to increase the propensity of APOL1 to cause tissue damage and human cell death (19, 22).

The exact mechanism by which APOL1 causes trypanosome lysis is currently unclear. Exposure of trypanosomes to human serum caused a rapid increase in cation flux across the plasma membrane, followed by cytoplasmic swelling and lysis that could be prevented if sucrose was added to the growth medium (23). Swelling was not a secondary result of energy crisis, as cells retained ATP up until the point of lysis. In addition, lysis induced by TLF-1 could be prevented if sodium chloride in the medium was replaced with the larger tetramethylammonium⁺ or gluconate⁻ ions (24). These observations constitute compelling evidence for a classic colloid-osmotic mechanism of lysis, driven by cytoplasmic swelling, due to an increase in the small-ion conductance of the plasma membrane. It was further suggested that, after endocytic uptake of TLF1, the lytic component (later confirmed as APOL1) might associate with endosomal membranes at low pH, before recycling back to the cell surface, where it could directly generate ion-permeable pores in the plasma membrane (24). Alternatively, other authors have proposed that APOL1 forms ion-permeable pores in the lysosomal membrane and that lysis is driven by osmotic swelling not of the cytoplasm, but of the lysosome (25, 26).

Although proteins of the APOL family (six members in humans) share some characteristics with Bcl-2 family proteins, as well as bacterial α-helical pore-forming toxins, there is scant and conflicting evidence in the literature of membrane pore formation by APOL1, and little is known regarding possible functions of the other APOL proteins (27). An APOL1 fragment was reported to form single anion-selective channels in planar lipid bilayers at pH 5.5; however, the fragment was later determined to be trypanolytically inactive because it was missing the C-terminal domain (11, 25, 28). In contrast, it was reported that whole TLF1 lipoprotein particles were associated with the formation of single cation-selective channels, also at pH 5.5 (24).

Here we demonstrate that nanogram quantities of purified trypanolytic APOL1 dramatically increase the cation conductance of planar lipid bilayers in a highly pH-sensitive manner. Consistent with this phenomenon, we find that APOL1 forms cation-selective channels that are opened in response to pH neutralization. Potential implications for the mechanism of trypanolysis are discussed.

Results

To generate full-length human APOL1, in the absence of serum contaminants, we purified recombinant APOL1 reconstituted from the insoluble fraction of bacterial expression lysates (Materials and Methods and Fig. S1). The protein eluted as a single peak from a size-exclusion column (Fig. S1) and was highly active: concentrations of 200 ng/mL (4.6 nM) were sufficient to lyse >90% of human serum-sensitive T. brucei parasites within 24 h (strain 427, Fig. 1A). The SRA protein, which is produced by T. b. rhodesiense, prevents trypanosome lysis via binding to the C terminus of APOL1 (10). We confirmed that the human serum-resistant T. brucei strain 427-SRA, which produces the SRA protein, is completely resistant to lysis by recombinant APOL1 (Fig. 1B). Resistance was apparently dependent on the SRA/APOL1 interaction, as recombinant APOL1 variants defined by amino acid changes to the C-terminal domain that are known to ablate (APOL1-G2) or reduce SRA binding (APOL1-G1) were capable of lysing human serum-resistant 427-SRA trypanosomes [LD₅₀: APOL1-G2, 100 ng/mL (2.3 nM); APOL1-G1, 4 μg/mL (92.0 nM)]. Of note, the APOL1-G2 protein, defined by the deletion of residues N388 and Y389, may be structurally different from the other APOL1 variants, as it eluted earlier from the size-exclusion column (Fig. S1).

Fig. 1.

SRA-producing trypanosomes are resistant to lysis by recombinant APOL1, but not APOL1-G1 and -G2. Trypanosomes susceptible to human serum, strain 427 (A), and resistant to human serum, strain 427-SRA (B), were exposed to dilutions of the indicated recombinant APOL1 variants for 20 h at 37 °C. Note that SRA-producing trypanosomes were resistant to lysis by APOL1 (WT), but susceptible to lysis by APOL1-G2, which does not bind SRA, and to a lesser extent APOL1-G1, which binds SRA with reduced affinity. Percentage viability was determined by alamar blue fluorescence relative to cells not treated with APOL1 (Materials and Methods). Results are mean ± SD from two independent experiments. 100 ng/mL APOL1 = 2.3 nM.

An essential step in the lytic process is APOL1 uptake into acidic endosomes. We therefore monitored APOL1-induced ion conductance across planar lipid bilayers separating pH 5.3 (cis) and pH 7.2 (trans) solutions to mimic, respectively, the endocytic and cytoplasmic compartments. During a typical experiment, the addition of APOL1 to the cis compartment at pH 5.3 resulted in only a modest increase in membrane conductance, on the order of 50–100 pS, that was sustained after removal of soluble protein by perfusion with cis buffer (Fig. 2A, Inset). However, we were able to amplify the APOL1-induced conductance by some several hundred-fold with the addition of Hepes buffer (pH 7.3) to the cis side, an effect that was fully reversible upon perfusion of the cis side with pH 5.3 buffer (Fig. 2A and Fig. S2). In control experiments, no increase in bilayer conductance was observed either at pH 5.3 or pH 7.25 with the addition of n-dodecyl-β-d-maltoside (DDM), the detergent in which APOL1 was solubilized. The effect of pH on the APOL1-induced conductance was recapitulated during repeated neutralization and acid perfusion cycles, indicating that the interaction of APOL1 with the membrane was essentially irreversible (Fig. 2A and Fig. S2). In contrast, addition of APOL1 to the cis side at pH 7.3 caused no increase in conductance either before or after removal of soluble APOL1 by perfusion of the cis side at pH 5.3, suggesting that acidic pH is necessary to promote the initial interaction of APOL1 with the membrane (Fig. S2). We considered that the APOL1-induced conductance might be biologically relevant, as truncated APOL1, which cannot lyse trypanosomes due to the lack of C-terminal residues 341–398 (11, 28), failed to generate a neutral-pH–dependent conductance in planar lipid bilayers (Fig. S3).

Fig. 2.

pH-dependent conductance response to APOL1 and its prevention by SRA. The voltage was held at −20 mV, except for a brief excursion to 0 mV (denoted by an asterisk; the current did not return to zero due to a small electrode offset). The solutions were 0.5 M KCl, 5 mM CaCl₂, 0.5 mM EDTA, supplemented with either 5 mM K-succinate, pH 5.3 (cis), or 5 mM K-Hepes, pH 7.2 (trans). When indicated, the cis side was adjusted to pH 7.3 with 20 μL 1 M Hepes, pH 7.5. Horizontal dashed bars indicate perfusion of the cis side with cis solution. A downward deflection in the current record represents an increase in the magnitude of the current. (A) Cis addition of 40 ng APOL1 caused a minor increase in current magnitude at pH 5.3 (Inset: current and time-scale expansion), and then soluble protein was removed by perfusion. Upon cis pH adjustment to 7.3, there was a large (∼200 pA) increase in current magnitude, which was reversed upon readjustment to pH 5.3. A stirring defect explains the delayed current response to the second Hepes addition. (B) When 5 μg SRA were added to the cis side, the subsequent addition of 40 ng APOL1 still caused a minor increase in current magnitude at pH 5.3 that was resistant to cis perfusion; however, the current actually decreased in magnitude upon adjustment of the cis side to pH 7.3 and remained low when the cis pH was returned to 5.3 (Inset: current scale expansion on the same time line). This procedure was then repeated on the same membrane with 5 μg SRA and 40 ng APOL1-G2 (which does not bind SRA), resulting in a normal current response (minimal increase in the current magnitude at low cis pH, followed by a large increase upon cis neutralization).

We reasoned that if the APOL1-induced conductance was truly relevant to the mechanism of trypanosome lysis, then it should be prevented by preadding recombinant SRA to the cis compartment at pH 5.3, a condition that would mimic the endosomal lumen of T. b. rhodesiense (17). In fact, when APOL1 was added to the cis side in the presence of SRA at pH 5.3, we still observed a minor conductance increase, which was retained after perfusion of the cis side at acidic pH, but the usual amplification upon raising the cis pH was completely prevented (Fig. 2B). [On the contrary, the conductance actually decreased upon cis neutralization and was not restored by readjustment to pH 5.3, suggesting that the interaction of APOL1 with the membrane may be rendered reversible by the preaddition of SRA (Fig. 2B, Inset)]. This effect depended specifically on the APOL1–SRA interaction, as evidenced by the fact that the G2 APOL1 variant, defined by a two-residue deletion that prevents SRA binding and allows killing of SRA-producing trypanosomes (Fig. 1B), induced a normal conductance response on the same membrane, even in the presence of SRA (Fig. 2B). These observations suggest that the formation of a neutral-pH–dependent conductance by APOL1 may be relevant to the mechanism of trypanosome lysis.

We next examined in detail the effect of pH on the APOL1-induced conductance. After APOL1 was added to the cis side at pH 5.3, the trans side was acidified to pH 5.3, and the cis side was perfused (also at pH 5.3) to remove soluble protein. Again, APOL1 induced a minor conductance of about 50–100 pS, which was resistant to cis perfusion, but, with alternating titration of first the cis and then the trans sides with KOH, there was a sigmoidal increase in conductance with a pK_a of 7.1 and a maximal conductance increase of around 3,000-fold, which was reversed by alternating titration of first the trans and then the cis sides with HCl (Fig. 3A). The change in conductance was almost entirely dependent on KOH additions to the cis side; KOH additions to the trans side had no significant effect on the conductance (Fig. S4).

Fig. 3.

pH dependence of the APOL1-induced conductance. The initial pH (5.3 cis, 7.2 trans) was established in 0.5 M KCl, 5 mM CaCl₂, and 0.5 mM EDTA containing universal buffer (10 mM MES, 10 mM Mops, 10 mM TAPS) titrated with KOH. (A) A total of 120–180 ng APOL1 was added to the cis side; after 20–30 min, the trans pH was titrated to 5.3 with 1 M HCl and the cis side was perfused with 10 vol of cis buffer (pH 5.3). With the voltage held at +10 mV, the titration curve was obtained by alternately adding 1 M KOH to the cis and trans sides and then the titration was reversed with alternating additions of 1 M HCl to the trans and cis sides. The current at each value of pH (pH_cis = pH_trans) was normalized to the maximal current (3,167 pA on average) to obtain relative current (I_rel). Plotted is the average I_rel of three independent experiments, ±SD, with the forward titration fitted to the Hill equation (pK_a = 7.1; Hill coefficient = 2.54). The same data are plotted on a log scale (Inset) to show more clearly the magnitude of the pH-dependent change in conductance. (B) Several nanograms of APOL1 were added to the cis side and allowed to incubate for 10 min at pH 5.3. The cis compartment was then adjusted to the indicated pH with KOH; the total time spent at each pH was at least 280 s. The voltage was alternated between −80 and +80 mV. Channel opening is indicated by downward (−80 mV) or upward (+80 mV) deflections in the current record, which was filtered at 30 Hz; the dotted line represents the current when all channels were closed. Almost absent at pH 6.50, the channel opening was observed on raising the cis pH. Typical single-channel conductances were in the range 14–18 pS, although multiple additional open channel states (opened at a numbered arrowhead) were also observed [3.75 pS (1), 8.75 pS (2), 27.50 pS (3), and 32.50 pS (4)]. (C) From the same experiment, and plotted as a histogram, is the relative channel current (I_rel) versus pH. At a given voltage, I_rel was defined as the average current (1/T ${{\displaystyle \int }}_0^{T}Idt$) during the total time T (at least 75 s) spent at each pH, normalized to the average current at pH 7.65.

Such a large change in macroscopic conductance might be explained by pH gating of APOL1-induced channels. To resolve pH-gated channels, we added just a few nanograms of APOL1 at cis pH 5.3 and then titrated the cis pH with KOH. Channel openings were barely detectable at pH 6.5, and channel open probability increased with increasing pH (Fig. 3 B and C), mirroring the pH-dependent increase in the macroscopic conductance (Fig. 3A). pH gating was independent of voltage polarity (Fig. 3C), although at a given pH, channel opening was more likely at −80 mV than at +80 mV (Fig. 3B). Typical single-channel conductances ranged between 14–18 pS, although several other conductance states, including larger conductances (27.5 and 32.5 pS) and smaller conductances (3.75 and 8.75 pS), were also observed (Fig. 3B). It is unclear at present if these observations relate to multiple conductance states of the same channel, or if different channel forms exist. (Not shown in Fig. 3B is that, even when all of the channels were “closed,” there existed a residual current ranging, at −80 mV, from 0 pA at pH 6.5 to 1.35 pA at pH 7.65; this current may reflect a residual conductance of the channels in their “closed” state.)

To determine ion selectivity of the APOL1-induced conductance, reversal potentials were obtained at pH 7.2 in the presence of KCl gradients. The macroscopic, as well as the single-channel conductance, was ideally selective for K⁺ over Cl⁻ (Fig. 4). We also obtained a bi-ionic potential (SI Materials and Methods) of −1.5 mV for the macroscopic conductance (0.5 M NaCl cis, 0.5 M KCl trans), confirming that the APOL1-induced conductance was as permeable to Na⁺ as it was to K⁺.

Fig. 4.

The APOL1-induced conductance is ideally cation selective. (A) The starting solutions were 0.5 KCl, 2 mM CaCl₂, and 0.5 mM EDTA, supplemented with 5 mM K-succinate, pH 5.3 (cis), or 5 mM K-Hepes, pH 7.2 (trans). With the voltage held at +10 mV, a total of 120–180 ng APOL1 was added to the cis side. After 15–20 min, the cis pH was raised to 7.3 with 1 M K-Hepes, pH 7.5, to give a conductance of 10–50 nS. The reversal potential (the voltage at which the current registered zero, E_rev) was determined after the cis side was perfused with 10 vol of 0.1 M KCl, 2 mM CaCl₂, 0.5 mM EDTA, 5 mM K-Hepes, pH 7.2, to give a fivefold trans/cis concentration gradient of KCl and after each of multiple 10-μL additions of 3 M KCl to the cis side. Plotted in the figure is E_rev versus the activity ratio of KCl (α_trans/α_cis). (Activity coefficients were obtained from appendix 8.10, table 11 in ref. 48, where we took the KCl concentration to be equal to the K⁺ concentration; K-Hepes and K-EDTA contributed about 4 mM K⁺). The ideal K⁺-selective line is defined by the Nernst equation (E_rev = 59.1 × log₁₀ α_trans/α_cis). Data points are mean ± SD of three independent experiments (error bars smaller than data points). (B) The starting solutions were as above (pH 5.3 cis, 7.2 trans), except that the trans side contained 0.1 M KCl. Sufficient APOL1 was added to allow the evolution of single channels after cis perfusion with 10 vol of 0.5 M KCl, 2 mM CaCl, 0.5 mM EDTA, and 5 mM K-Hepes, pH 7.2, to give a fivefold cis/trans KCl concentration gradient/4.12-fold activity gradient. The single-channel current was plotted against voltage and fitted to a straight line. E_rev is estimated as −36.3 mV, indicating ideal cation selectivity. The slope of the line gives a single-channel conductance of 14 pS. Electrode offsets were less than 1 mV.

Discussion

Bacterial pore-forming toxins, such as diphtheria, botulinum, and anthrax toxins, are water-soluble at neutral pH, yet they are capable of inserting into lipid bilayers to form ion-permeable channels upon exposure to the acidic pH characteristic of eukaryotic endosomes (29–33). In this paper, we show that full-length, recombinant APOL1 induces a minor conductance in planar lipid bilayers at acidic pH that can be reversibly amplified by up to 3,000-fold by raising the cis pH (Figs. 2 and 3A, and Fig. S2; pK_a = 7.1). Moreover, we demonstrate that this phenomenon can be attributed to the pH gating of APOL1-induced cation-selective channels (Figs. 3 B and C and 4). The APOL1-induced conductance remained after soluble protein was perfused out of the solution and after subsequent neutralization/acidification cycles (Fig. 2), yet no conductance formed when APOL1 was added to the chamber at neutral pH and then washed out at acidic pH (Fig. S2). Together, these data imply that acidic pH stimulates APOL1 binding and insertion into the membrane, but, unlike the channels formed by acid-dependent pore-forming toxins of bacteria, significant opening of APOL1-induced channels is permitted only upon neutralization of the cis pH.

Channel opening by pH titration in the neutral pH range was previously observed among members of the two-pore-domain and inward-rectifier potassium channel subfamilies and can result from the deprotonation of basic residues at anomalously low pH values. For example, the pH-dependent opening of the so-called TASK-2 and Kir1.1 channels involves the respective deprotonation of an arginine (pK_a = 8) and a lysine (pK_a = 7) residue (34, 35), which may lead to large conformational changes or local perturbations of the selectivity filter (36). Thus, it appears that APOL1 combines the pH-dependent solubility characteristics of bacterial pore-forming toxins, with the complex pH-gating behaviors more typical of eukaryotic channels.

The neutral-pH–dependent APOL1-induced conductance is likely relevant to the mechanism of trypanosome lysis for the following reasons. First, with just nanogram quantities of full-length trypanolytic APOL1 (Fig. 1), we readily generated, in a pH-dependent and reversible manner, macroscopic conductances of up to 300 nS (∼18,000 channels of 17 pS each, Fig. 3A). In contrast, a truncated version of APOL1, which was previously shown to be nontrypanolytic due to loss of the C-terminal domain (residues 341–398) (11, 28), did not induce a neutral-pH–dependent conductance in planar lipid bilayers (Fig. S3). Second, the neutral-pH–dependent conductance (but not the minor conductance at pH 5.3) was prevented by the SRA protein of T. b. rhodesiense (Fig. 2B), which prevents trypanosome lysis by APOL1 via an interaction with the APOL1 C-terminal domain (Fig. 1) (10, 18, 19). We speculate that this interaction prevents the acid-dependent membrane insertion of a putative transmembrane sequence (residues 332–354) (19), which could conceivably form part of a pH-gated channel or serve to anchor APOL1 in the membrane upon pH neutralization (Fig. 5). Finally, a two-residue deletion near the APOL1 C terminus (APOL1-G2), which prevents SRA binding and allows killing of SRA-producing trypanosomes (Fig. 1) (19, 20), also allowed formation of the neutral-pH–dependent conductance in the presence of SRA (Fig. 2B).

Fig. 5.

Recycling model of trypanosome lysis by APOL1. After endocytosis by trypanosomes (top left), APOL1 encounters acidic endosomes, where it may insert into the endosomal membrane (shading indicates decreasing pH). At acidic pH (cis pH < 6), APOL1 forms an essentially nonconductive state (polypeptide backbone represented as a closed coil), but in the case of human serum-susceptible trypanosomes (e.g., T. b. brucei), it may generate cation-selective channels if it is recycled to the plasma membrane and exposed to neutral pH conditions (open coil; cis pH > 7), but not if it is trafficked to the lysosome (cis pH < 5). Cytoplasmic swelling results from APOL1-induced sodium influx, as well as chloride influx via endogenous channels (24, 25). In this model, membrane insertion of APOL1 is accomplished by three putative transmembrane sequences, one of which resides in the APOL1 C-terminal domain (19). The SRA protein of human serum-resistant T. b. rhodesiense (oval), which is thought to be membrane-anchored via a glycosylphosphatidylinositol group (not shown for clarity), is reported to cycle between the plasma membrane and the endocytic system, where it can bind to the C terminus of APOL1 at acidic pH (17). SRA may prevent insertion of the C-terminal transmembrane sequence of APOL1, allowing dissociation of APOL1 from the membrane upon recycling to neutral pH conditions of the cell surface, where APOL1 also dissociates from SRA.

These data are therefore consistent with a model of trypanosome lysis first conceptualized by Molina-Portela and colleagues (24) and elaborated here (Fig. 5), whereby APOL1 first inserts into endosome membranes at acidic pH before being recycled to the cell surface, where exposure to neutral pH allows for the opening of APOL1-induced cation-selective channels and the influx of sodium (Figs. 3–5), with the accompanying potassium efflux, chloride influx, and membrane depolarization. As with most cell types, this would be facilitated in trypanosomes by a relatively low internal sodium concentration (13.7 mM) and a negative plasma-membrane potential (−76 mV) (37). Given the extraordinary rate of endocytic recycling in trypanosomes, it would not be surprising if some fraction of endocytosed APOL1 were recycled to the plasma membrane to generate cation conductances (38). This would allow for classical colloid-osmotic lysis: cytoplasmic swelling would occur as a result of ion and water influx across the plasma membrane, from the extracellular milieu, driven by the osmotic pressure of cytoplasmic macroanions (Fig. 5) (39). A prediction of this model is that swelling and lysis should be prevented simply by balancing the internal osmotic pressure with plasma-membrane impermeant solutes added to the extracellular fluid. Indeed, it was reported that sucrose could prevent lysis by human serum (22) and that large ions including tetramethylammonium⁺ and gluconate⁻ could prevent lysis by TLF1 (23, 24). Such a model would also explain the increase in cation permeability of the parasite plasma membrane, observed within 15 min of exposure to human serum (23).

The proposed model would also explain the morphological changes that are consistently demonstrated to precede trypanolysis, whether it is initiated by recombinant APOL1, TLF1/2, or normal human serum: these changes are characterized by a rounding of the cell body and swelling of the cytoplasm (12, 23, 25, 26, 40). In cases where swelling of the cell body proceeded slowly, and completion of lysis required several hours, swelling of the lysosome was also demonstrated (25, 26), but lysosomal swelling was not obviously apparent when lysis was driven rapidly to completion in 2 h or less (12, 23, 40). Therefore, it is our opinion that cytoplasmic swelling, as a result of osmolyte influx across the plasma membrane, is the defining feature of trypanosome lysis by APOL1 and that lysosomal swelling can occur as a consequence, but only if sufficient time is allowed before lysis occurs. Indeed, it was reported that 4,4-diisothiocyanatostilbene-2, 2-disulfonic acid, which blocks channel-mediated chloride flux across the plasma membrane, could prevent lysosomal swelling (as well as lysis) induced by human serum (25). Taking the current data into account, we favor a model in which cation influx via APOL1-induced plasma-membrane channels is coupled to anion and water influx into the cytoplasm (Fig. 5); this, in turn, would dilute cytoplasmic macromolecules with saline and alter the osmotic balance between the cytoplasm and lysosome. Lysosomal swelling might then occur, but only—assuming significant resistance of the lysosomal membrane to ion movement—if there were sufficient time for ions (and water) to cross into the lysosomal lumen.

Notwithstanding the above considerations, it was proposed by Pérez-Morga et al. (25) that lysis is initiated not by cation-selective channel formation in the plasma membrane, but by the formation of APOL1-induced anion-selective channels in the lysosomal membrane and results from uncontrolled osmotic swelling of the lysosome. Because this view contradicts our own, but is by now almost dogma, we will respond to it here directly. First, we must point out that anion selectivity was inferred from the reversal potential of single channels formed in planar lipid bilayers treated with a nontrypanolytic fragment of APOL1 (lacking the essential C-terminal domain). In contrast, single cation-selective 17 pS channels, similar to those reported here, were generated in planar lipid bilayers treated with trypanolytic TLF1 (24) (the very low activity described in that report may be explained by the acidic, cis pH 5.5 conditions). Second, and as elaborated above, a review of the published morphological as well as biochemical evidence indicates that lysis is initiated by electrolyte flux across the plasma membrane, not the lysosomal membrane. Finally, it is difficult to reconcile the results in the current report with the formation of an APOL1-induced conductance in the lysosomal membrane, where the luminal pH, estimated at <pH 5.0 (41), would preclude the opening of APOL1-induced conductances (Fig. 3). [What little conductance we did obtain at pH 5.3 may be irrelevant to trypanosome lysis, as it was not preventable by SRA (Fig. 2B, Inset)].

In conclusion, we have demonstrated that APOL1 forms cation-selective channels in planar lipid bilayers in a manner that requires first acidic and then neutral pH. These data are allied to a cytoplasmic-swelling model of trypanosome lysis that is contingent with endocytic recycling of APOL1 and the formation of cation-selective channels in the parasite plasma membrane. Confirmation of this model will require additional studies; nevertheless, the possibility that APOL1 generates cytolytic lesions in plasma membranes may inform speculation about the potential role of APOL1 in mediating human cell death (42), as well as hypotheses regarding the potential pathophysiological impact of human APOL1 variants that are associated with kidney disease (19, 22).

Materials and Methods

Please refer to SI Materials and Methods for additional details.

Purification of Recombinant N-Terminally His-Tagged APOL1 and SRA.

Full-length, mature-form APOL1 (residues 28–398, accession no. O14791), G1, G2, and truncated APOL1 variants as well as SRA (residues 24–267 (18), accession no. Q8T309) were purified from the insoluble material of an Escherichia coli BL21 (DE3)-RIL (Agilent Technologies) auto-induction lysate as follows. APOL1 proteins were solubilized in 1% zwittergent 3–14 (SB3-14, EMD-Millipore), and SRA was solubilized in 2% (wt/vol) 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS, Calbiochem) and 0.1 mM DTT. In each case, solubilization was facilitated by briefly adjusting to pH 12 (1–2 min) with 10 mM NaOH and 150 mM NaCl before reducing the pH (to ∼8.0) with 30 mM Tris⋅HCl, pH 7.4. Solubilized APOL1 proteins were bound to nickel (HisTRAP, GE Life Sciences) equilibrated in 20 mM Tris⋅HCl, pH 8.5, 150 mM NaCl, and 1% SB3-14 and then eluted with a 0–300 mM imidazole gradient prepared in the same buffer. Further purification was achieved using a Superdex-200 pg size-exclusion column (GE Life Sciences), equilibrated in 50 mM Tris⋅HCl, pH 8.5, 150 mM NaCl, and 0.1% DDM (EMD-Millipore), and purified proteins were stored in aliquots (60–200 μg/mL; 1.4–4.6 μM) at −80 °C (Fig. S1). SRA was bound to nickel equilibrated in 20 mM Tris⋅HCl, pH 8.3, 150 mM NaCl, and 1% CHAPS in the presence of 20 mM imidazole and then washed and eluted with the same buffer containing, respectively, 150 and 300 mM imidazole. Size exclusion was then performed as above, except the column was equilibrated in 50 mM Tris⋅HCl, pH 7.4, and 150 mM NaCl (no DDM). Purified SRA was stored in aliquots (1.3 mg/mL) at −80 °C.

Trypanosome Lysis Assays.

Human serum-sensitive T. b. brucei (427) was derived from Lister-427 and the human serum-resistant, SRA-expressing line 427-SRA was generated by the G. Cross laboratory (The Rockefeller University, New York) (43). Cells were grown in HMI-9 medium (44) and exposed to serial dilutions of APOL1 or APOL1 variants in 96-well plates at a density of 5 × 10⁵ cells per milliliter. The parasites were incubated for 20 h at 37 °C (5% CO₂), and cell viability was determined using the alamar blue assay (45).

Planar Lipid Bilayers.

Bilayers were formed at room temperature from soybean asolectin [lecithin type IIS (Sigma Chemical) from which nonpolar lipids had been removed (46)] across an 80- to 120-μm hole in a Teflon partition separating two aqueous solutions of 1 mL vol, as described previously (47). One percent asolectin in pentane (20 μL) was layered on top of both cis and trans solutions; the hole was pretreated with 3% (vol/vol) squalene (3 μL, Sigma) in petroleum ether, and the solvent was allowed to evaporate. The level of each solution was raised above the hole; a change in capacitance was used to monitor membrane formation. Perfusion was done with a BPS-2 Bilayer Perfusion System (Warner Instruments) and a pair of coupled syringes. Agar salt bridges [3 M KCl, 3% (wt/vol) agar] linked Ag/AgCl electrodes to the cis and trans compartments, which could be stirred by magnetic fleas. The buffers used were 5 mM K-succinate, pH 5.3 (cis), 5 mM K-Hepes, pH 7.2 (trans), or, alternatively, a universal buffer was used, consisting of 10 mM MES, 10 mM Mops (3-[N-morpholino]propanesulfonic acid), and 10 mM TAPS [N-Tris(hydroxymethyl)methyl-3-aminopropanesulfonic acid], which allowed for the adjustment of pH in the range 5.3–9.0 with KOH (pH values were determined by parallel titrations). Unless otherwise stated in the Fig. 4 legend (reversal potential) or SI Materials and Methods, Bi-Ionic Potential, the solutions also contained 0.5 M KCl, 5 mM CaCl₂, and 0.5 mM EDTA. Voltages were maintained using the BC-535C bilayer clamp (Warner Instruments) and are given as the voltage of the cis solution (defined as the side to which protein was added) with respect to the trans solution. The current response was filtered at 30 Hz by a low-pass eight-pole Bessel filter (Warner Instruments) and recorded using a chart recorder as well as digitally using IGOR NIDAQ Tools MX 1.0 and IGOR software (WaveMetrics) via an analog-to-digital converter (NI USB-6211; National Instruments). The reversal potentials of macroscopic bilayer conductances were obtained by adjusting the voltage until the current registered zero in the presence of a cis:trans salt gradient (Fig. 4 legend).