Aquaglyceroporin 2 controls susceptibility to melarsoprol and pentamidine in African trypanosomes

Abstract

African trypanosomes cause sleeping sickness in humans, a disease that is typically fatal without chemotherapy. Unfortunately, drug resistance is common and melarsoprol-resistant trypanosomes often display cross-resistance to pentamidine. Although melarsoprol/pentamidine cross-resistance (MPXR) has been an area of intense interest for several decades, our understanding of the underlying mechanisms remains incomplete. Recently, a locus encoding two closely related aquaglyceroporins, AQP2 and AQP3, was linked to MPXR in a high-throughput loss-of-function screen. Here, we show that AQP2 has an unconventional "selectivity filter." AQP2-specific gene knockout generated MPXR trypanosomes but did not affect resistance to a lipophilic arsenical, whereas recombinant AQP2 reversed MPXR in cells lacking native AQP2 and AQP3. AQP2 was also shown to be disrupted in a laboratory-selected MPXR strain. Both AQP2 and AQP3 gained access to the surface plasma membrane in insect life-cycle-stage trypanosomes but, remarkably, AQP2 was specifically restricted to the flagellar pocket in the bloodstream stage. We conclude that the unconventional aquaglyceroporin, AQP2, renders cells sensitive to both melarsoprol and pentamidine and that loss of AQP2 function could explain cases of innate and acquired MPXR.

Fig. 1.

Cells lacking the unusual AQP2 are MPXR. (A) Schematic representation of AQP2 (Tb927.10.14170) and AQP3 (Tb927.10.14160). The selectivity filter residues are indicated; NSA/NPS/IVLL in AQP2 and NPA/NPA/WGYR in AQP3. An AQP2-specific insertion is also indicated. (B) Homology models for AQP2 and AQP3 showing the key amino acids that line the channel (colors as in A). The models were generated using SWISS-MODEL (51) and the Protein Data Bank coordinates 1ldfA. (C) Schematic map (Left) and Southern blot (Right) showing AQP2 gene knockout. Δ indicates the region deleted and “flank” indicates the 671-bp probe. Transcription is from left to right. Thicker black lines indicate the predicted untranslated regions. Genomic DNA was digested with SacII (S) before Southern blotting. (D) Bloodstream-form cells. Dose–response curves for melarsoprol and pentamidine. Wild-type (WT) cells are compared with aqp2 null cells. EC₅₀ values are shown. (E) Insect-stage (procyclic form) cells. Other details as in D above. (F) Bloodstream-form cells. EC₅₀ values for other related drugs. Average and SEM are shown (n ≥ 10). *P < 0.0001 as determined using a one-way ANOVA test.

Fig. 2.

AQP2 is required for drug sensitivity, whereas AQP3 is neither necessary nor sufficient. (A) EC₅₀ values for melarsoprol following recombinant AQP expression in bloodstream-form cells. aqp2/aqp3 null cells with a tetracycline-regulated (Tet-on) copy of either AQP2 or AQP3 were grown in the absence (black bars) or presence (gray bars) of tetracycline. Error bars, SEM derived from two independent triplicate assays. *P < 0.0001 as determined using nonlinear regression analysis. (B) EC₅₀ values for pentamidine. Other details as in A. (C) Tet-regulated expression of GFP-tagged AQPs. Western blots reveal tightly regulated inducible expression of both ^GFPAQP2 and ^GFPAQP3. Both proteins migrated below their predicted molecular mass of ∼60 kDa but proteins with extensive transmembrane regions often display aberrant migration. (D) Both ^GFPAQP2 and ^GFPAQP3 are membrane associated. Western blots show supernatant (S), wash (W), and pellet (P; membrane-fraction) following hypotonic lysis. ^GFPAQPs are expressed in an aqp2/aqp3 null background. (E) EC₅₀ values for melarsoprol. aqp2/aqp3 null cells with a Tet-on copy of ^GFPAQP2 or ^GFPAQP3. Other details as in A. (F) EC₅₀ values for pentamidine. Other details as in E. (G) The KO-B48 strain encodes an AQP2/AQP3 chimeric sequence. Schematic illustrates the AQP2/AQP3 locus before and after the emergence of pentamidine resistance. Selectivity filter residues are indicated; six residues are specific to the AQP3 sequence (underlined) and four of these are found in the predicted chimeric protein (Fig. S1).

Fig. 3.

AQP2 is restricted to the flagellar pocket specifically in bloodstream-form cells. (A) Fluorescence microscopy reveals the location of ^GFPAQP2 and ^GFPAQP3 in bloodstream-form cells. DNA is stained with DAPI and the nucleus (N) and mitochondrial genome (kinetoplast, K) are indicated. (B) Fluorescence microscopy reveals the location of ^GFPAQP2 and ^GFPAQP3 in insect-stage cells. (C) Immunofluorescence microscopy reveals the location of ^GFPAQP2 colocalized with the flagellar pocket (FP) marker, ISG65 in bloodstream-form cells. ^GFPAQPs are expressed in an aqp2/aqp3 null background. (Scale bars, 10 μm.)

Fig. 4.

Model to explain melarsoprol and pentamidine transport in T. brucei. Four T. brucei transporters have been linked to the control of melarsoprol and or pentamidine susceptibility. Both drugs enter the cell through AT1/P2 (Tb927.5.286b) and AQP2. HA1-3 (Tb927.10.12500–10, only two annotated in the reference genome) is specifically linked to pentamidine susceptibility and generates the proton motive force required for pentamidine symport via AQP2. Efflux of the toxic melarsoprol adduct, Mel T, is via MRPA (Tb927.8.2160). The AT1/P2 and AQP2 channels transport the drugs with different efficiencies as indicated by the weighted arrows. AT1/P2 and MRPA have not been localized. FP, flagellar pocket; Mel T, melarsoprol–trypanothione adduct. The homology model for AT1/P2 is restricted to residues 340–454.