Export of malaria proteins requires co-translational processing of the PEXEL motif independent of phosphatidylinositol-3-phosphate binding

Abstract

Plasmodium falciparum exports proteins into erythrocytes using the Plasmodium export element (PEXEL) motif, which is cleaved in the endoplasmic reticulum (ER) by plasmepsin V (PMV). A recent study reported that phosphatidylinositol-3-phosphate (PI(3)P) concentrated in the ER binds to PEXEL motifs and is required for export independent of PMV, and that PEXEL motifs are functionally interchangeable with RxLR motifs of oomycete effectors. Here we show that the PEXEL does not bind PI(3)P, and that this lipid is not concentrated in the ER. We find that RxLR motifs cannot mediate export in P. falciparum. Parasites expressing a mutated version of KAHRP, with the PEXEL motif repositioned near the signal sequence, prevented PMV cleavage. This mutant possessed the putative PI(3)P-binding residues but is not exported. Reinstatement of PEXEL to its original location restores processing by PMV and export. These results challenge the PI(3)P hypothesis and provide evidence that PEXEL position is conserved for co-translational processing and export.

Figure 1. Exported P. falciparum proteins do not bind PI(3)P.

(a) Recombinant proteins expressed in this study as fusions to GST or GFP/6His. (b) Recombinant proteins stained with Coomassie blue stain. Molecular weights (kDa) are shown. (c) Binding of p40PX and HRPII with native PEXEL (WT) or mutant PEXEL (RLE>A) to PI(3)P liposomes measured by SPR. Binding to control PC/PE liposomes was subtracted to give the sensorgrams. BSA was used to measure nonspecific binding to the liposome-coated chip (red). Inset: a similar level of low binding of HRPII^WT and HRPII^RLE>A to PI(3)P liposomes. Experiments were performed in triplicate. (d) Binding of proteins to PI(3)P liposomes or control PC/PE liposomes, determined by ultracentrifugation. Pellet and supernatant fractions were resolved by SDS–PAGE and Coomassie blue stain. Proteins were ultracentrifuged in buffer to remove potential aggregates before incubation with liposomes, explaining why input and sum of pellet and supernatant sometimes differ. Experiments were performed in triplicate. Full-length gels shown in Supplementary Fig. 6. (e) Binding of p40PX and P. falciparum proteins with native PEXEL (WT) or mutant PEXEL (M) to PI(3)P-coated beads (+) or control beads lacking lipid (−). Bound protein was eluted in SDS–PAGE sample buffer and detected by immunoblot with anti-GST (p40PX) or anti-GFP (P. falciparum proteins). Ten per cent of unbound fraction volume was loaded to visualize protein inputs. Densitometry (histogram shown) of the p40PX bands shows that 68% of p40PX input bound to PI(3)P-coated beads (PI3P+) compared with 18% of input to control beads (PI3P−). No binding was detected for P. falciparum exported proteins. Experiments were performed in triplicate. Full-length gels shown in Supplementary Fig. 7. (f) Binding of recombinant proteins to phospholipid head group of PI(3)P (inositol 1,3 bisphosphate; I1,3P2) in solution, measured by isothermal calorimetry. The titration did not differ between I1,3P2 (2.4 mM) into dialysis buffer (150 mM NaCl, 20 mM HEPES pH 7.4) (negative control) and the same buffer containing either GBP130_66–88, HRPII^WT or HRPII^RLE>A. As controls, binding of Htp1 from S. parasitica to Fmoc-Tyr(SO₃)-OH and also p40PX binding to I1,3P2 are shown. Experiments were performed in duplicate.

Figure 2. PI(3)P is not localized in the ER of P. falciparum.

(a) Chimeric proteins expressed in P. falciparum. p40PX-GFP, no signal sequence; STEVORss-p40PX-GFP, STEVOR signal sequence; STEVORss-p40PX-GFP_SDEL, STEVOR signal sequence and ER retention signal; Hrs-GFP, no signal sequence; STEVORss-Hrs-GFP, STEVOR signal sequence; STEVORss-Hrs_SDEL-GFP, signal sequence and ER retention signal; STEVORss-Hrs_C215S-GFP, STEVOR signal sequence and mutation of lipid-binding residue C215S in both FYVE fingers. (b) p40PX and Hrs reporters are expressed in P. falciparum-infected erythrocytes. Blots probed with anti-GFP antibodies. First panel: p40PX-GFP, STEVORss-p40PX-GFP and STEVORss-p40PX-GFP_SDEL. A ‘GFP core' band was observed for STEVORss-p40PX-GFP. This is observed for all GFP chimeras secreted to the PV and does not indicate degradation specific to this protein. Note the absence of ‘GFP core' when SDEL ER-retention signal was attached, which prevented secretion to PV (lane 3). Second panel: Hrs-GFP, STEVORss-Hrs-GFP, STEVORss-Hrs-GFP_SDEL and STEVORss-HrsC_215S-GFP. Note the lack of ‘GFP core' for STEVORss-Hrs fusions, confirming they were not secreted. (c) p40PX-GFP localizes to the cytoplasm and to membranes of the food vacuole (labelled with anti-CRT) and apicoplast (labelled with anti-ACP). STEVORss-p40PX-GFP localizes to the PV in P. falciparum-infected erythrocytes, as shown by co-localization with anti-EXP2, and not in the ER, which was labelled with anti-PMV. If PI(3)P was located within the ER, this chimera would be expected to remain within the ER via p40PX binding. Fusion to an ER-retention signal (STEVORss-p40PX-GFP_SDEL) retains the protein in ER, co-localization with anti-ERC. Scale bar, 5 μm. (d) Localization of Hrs-GFP to the cytoplasm and membranes of food vacuole (labelled anti-CRT) and apicoplast (labelled anti-ACP). STEVORss-Hrs-GFP localizes to the ER as shown by co-localization with anti-ERC; however, mutation of residues necessary for lipid binding in both Hrs FYVE fingers (STEVORss-Hrs_C215S-GFP) does not abrogate ER localization, demonstrating PI(3)P-independent retention in ER. Addition of ER retention signal (STEVORss-Hrs-GFP_SDEL) also causes ER localization. Scale bar, 5 μm. (e) Quantification of GFP localization in the PV (black), ER (grey), or mix of both (white) determined using 40 P. falciparum-infected erythrocytes per construct (20 cells per experiment, performed twice).

Figure 3. Oomycete RxLR does not mediate export in P. falciparum.

(a) KAHRP_1–96 is exported to the P. falciparum-infected erythrocyte; however, HRPIIss-AVR3a, HRPIIss-PH001D5, REX3ss-AVH5 and REX3ss-AVR1b are not exported but accumulate in the parasite and PV. Scale bar, 5 μm. (b) Quantification of cells with exported GFP determined using 20 P. falciparum-infected erythrocytes per construct (10 cells per experiment, performed twice). (c) Immunoblot with anti-GFP antibodies of each chimera from the tetanolysin pellet (P) and supernatant (S) shows that KAHRP_1–96 is exported but RxLR chimeras are not. Tetanolysin inserts pores in the erythrocyte membrane, allowing sampling of the host cell cytosol. Aldolase was included as a control, as described previously. Full-length gels shown in Supplementary Fig. 8. (d) RP-HPLC shows the KAHRP peptide is cleaved when incubated with PMV (red). AVR3a and PH001D5 peptides are not cleaved by PMV. PEXEL/RxLR sequences are labelled green. (e) MS/MS-extracted ion chromatograms of KAHRP and PH001D5 peptides after incubation with PMV shows that KAHRP peptides were cleaved after the PEXEL Leu (GNGSGDSFDFRNKRTL−) but PH001D5 was not cleaved.

Figure 4. PEXEL-independent export of PEXEL proteins does not occur in P. falciparum.

(a) HRPII 4A is exported to the P. falciparum-infected erythrocyte. HRPII 4A RLE>A and HRPII 4A RLE>A Down are not exported. Scale bar, 5 μm. (b) PfEMP3_1–82 is exported by P. falciparum to the erythrocyte, but PfEMP3 4A RLQ>A is not. Scale bar, 5 μm. (c) Quantification of cells with exported GFP determined using 20 P. falciparum-infected erythrocytes per construct (10 cells per experiment, performed twice). (d) Immunoblot with anti-GFP antibodies of chimeras from the tetanolysin pellet (P) and supernatant (S) shows that HRPII 4A and PfEMP3_1–82 are cleaved at the PEXEL (blue arrow) and exported. Signal peptide-cleaved HRPII 4A was also visible (red spot). HRPII 4A RLE>A, HRPII 4A RLE>A Down and PfEMP3_1–82 RLQ>A were not exported and bands corresponding to signal peptide-cleaved (various positions possibly due to insertion of four alanines; red spots/arrow) and uncleaved protein (black spots/arrow) are present. Full-length gels shown in Supplementary Fig. 9.

Figure 5. Correct PEXEL positioning is essential for export.

(a) KAHRP_1–96 and KAHRPΔ38–48 are exported to the P. falciparum-infected erythrocyte but KAHRPΔ38–53 and KAHRPΔ38–53 RLQ>A accumulate in the PV and are not exported. KAHRP16Ala, KAHRP10Ala and KAHRP5Ala are exported to the red blood cell. Scale bar, 5 μm. (b) Quantification of export in a. Data are the mean (±s.e.m.) GFP fluorescence intensity in the host cell from 20 cells per construct (20 replicates), shown as a percentage of KAHRP_1–96. Data were analysed by t-test (***P<0.0001 and *P<0.003, compared with KAHRP_1–96). (c) Immunoblot with anti-GFP antibodies of chimeras from the tetanolysin and saponin pellet (P) and supernatant (S). This shows KAHRP_1–96 at a size corresponding to cleavage at the PEXEL and it was exported (32 kDa; blue arrows), KAHRPΔ38–53 and KAHRPΔ38–53 RLQ>E were cleaved to a size corresponding to cleavage by signal peptidase (32.5 kDa; red spots) and were not exported, but were present in the saponin supernatant, confirming they were secreted to the PV, and that KAHRP16Ala was of a size corresponding to cleavage at the PEXEL and it was exported (32 kDa; blue arrow); however, cleavage was inefficient as bands consistent with signal sequence cleaved (red arrow) and uncleaved (black) species were present in the pellet fraction. HSP70 was used as a control. Full-length gels shown in Supplementary Fig. 10.

Figure 6. The PEXEL is cleaved co-translationally.

(a) Top: schematic of PEXEL proteins and their cleavage positions. (b) ³⁵S-methionine/cysteine-labelling (³⁵S-Met/Cys) of HRPII 4A and HRPII 4A RLE>A in P. falciparum for the indicated times shows bands corresponding to uncleaved (black arrow), signal peptide-cleaved (red arrow) and PEXEL-cleaved (blue arrow) species. Two signal peptide-cleaved species are present for RLE>A (red arrows), possibly due to insertion of four alanines following the signal sequence. (c) Immunoblot with anti-GFP antibodies of the same membrane shows the bands are GFP specific. (d) ³⁵S-Met/Cys-labelling of KAHRP_1–70 and KAHRP_1–70-RLQ>A in P. falciparum. Bands corresponding to uncleaved (black arrow), signal peptide-cleaved (red arrow) and PEXEL-cleaved (blue arrow) species are indicated. (e) Immunoblot with anti-GFP antibodies of the same membrane. Note the absence of a signal peptidase cleaved product for KAHRP_1–70 RLQ>A in the autoradiography that is present in the immunoblot as the latter reveals protein produced over hours. This shows signal peptidase activity was inefficient. (f) ³⁵S-Met/Cys-labelling of PfEMP3_1–82 in P. falciparum. Bands corresponding to cleavage of the indented signal peptide (red arrow) and PEXEL (blue arrow) are indicated. (g) Immunoblot with anti-GFP antibodies of the same membrane. A PEXEL mutant (PfEMP3_1–82 L>A) was included as a size control. Most protein detected with anti-GFP was present before radiolabelling; thus, quantity is not comparable to ³⁵S-Met/Cys. Incorporation of ³⁵S-Met/Cys into HRPII and KAHRP constructs did not increase appreciably over time, in contrast to PfEMP3. The reason for this is unknown but may be due to saturation of anti-GFP agarose in the former experiments, which was used to enrich the GFP proteins. Full-length gels for all panels shown in Supplementary Fig. 11.