Sequence elements within the PEXEL motif and its downstream region modulate PTEX‐dependent protein export in Plasmodium falciparum

Abstract The parasite Plasmodium falciparum causes the most severe form of malaria and to invade and replicate in red blood cells (RBCs), it exports hundreds of proteins across the encasing parasitophorous vacuole membrane (PVM) into this host cell. The exported proteins help modify the RBC to support rapid parasite growth and avoidance of the human immune system. Most exported proteins possess a conserved Plasmodium export element (PEXEL) motif with the consensus RxLxE/D/Q amino acid sequence, which acts as a proteolytic cleavage recognition site within the parasite's endoplasmic reticulum (ER). Cleavage occurs after the P1 L residue and is thought to help release the protein from the ER so it can be putatively escorted by the HSP101 chaperone to the parasitophorous vacuole space surrounding the intraerythrocytic parasite. HSP101 and its cargo are then thought to assemble with the rest of a Plasmodium translocon for exported proteins (PTEX) complex, that then recognises the xE/D/Q capped N‐terminus of the exported protein and translocates it across the vacuole membrane into the RBC compartment. Here, we present evidence that supports a dual role for the PEXEL's conserved P2′ position E/Q/D residue, first, for plasmepsin V cleavage in the ER, and second, for efficient PTEX mediated export across the PVM into the RBC. We also present evidence that the downstream ‘spacer’ region separating the PEXEL motif from the folded functional region of the exported protein controls cargo interaction with PTEX as well. The spacer must be of a sufficient length and permissive amino acid composition to engage the HSP101 unfoldase component of PTEX to be efficiently translocated into the RBC compartment.


| INTRODUCTION
The pathogenesis of malaria relies upon the ability of Plasmodium parasites, the causative agents of the disease, to propagate asexually within red blood cells (RBCs) while avoiding the immune mechanisms of their host.2][3][4][5] The intraerythrocytic stage of the parasite resides within a membranous compartment called the parasitophorous vacuole (PV). 6,7Consequently, parasite proteins destined for export must traverse two membranes, the parasite plasma membrane and the encasing parasitophorous vacuole membrane (PVM), to reach the host RBC.The Plasmodium translocon of exported proteins (PTEX) mediates the translocation of proteins across the PVM 8 and is the only protein channel known to reside at the PVM.2][13] The cleavage step occurs within a pentameric amino acid motif near the N-terminus of the protein destined for export, termed the Plasmodium export element (PEXEL). 1,2,14The resulting mature proteins then travel via the vesicular transport pathway to the parasite plasma membrane where they are secreted into the PV. 15 Here, the exported proteins are unfolded and translocated across the PVM and into the host cell compartment by PTEX. 168][19] However, the information that is contained within proteins destined for export that allows HSP101 to specifically recognise these proteins prior to the unfolding step is still unknown. 20,213][24][25][26] This N-terminal region usually contains a recessed signal peptide to promote entry of the proteins into the parasite's ER 27 or a hydrophobic stretch of amino acids that serves the same function. 28The discovery of the five-amino acid PEXEL motif in the N-terminal region of many exported proteins was the first export-specific signal 1,2 that was proven to be predictive for exported proteins 29 and allowed the identification of >450 putative exported proteins in P. falciparum. 4 Interestingly, however, the presence of the PEXEL motif alone does not guarantee export, as its location within a protein's primary structure 30 and the presence of $12 amino acids downstream of the motif have also been shown to be required to achieve efficient export. 24Furthermore, PEXEL-negative exported proteins (PNEPs) are also present in the Plasmodium exportome, 5,31,32 suggesting that PEXEL is not strictly required for passage through PTEX.
The function of the PEXEL motif has been relatively well studied and its consensus sequence is RxLxE/D/Q where x can be almost any amino acid.3][14] Although most exported proteins utilise P 3 R and P 1 L residues some proteins utilise other amino acids with similar properties. 33It is assumed therefore that P 1 and P 3 facilitate PEXEL cleavage and thus help to release the exported proteins from the ER membrane where they are originally anchored following ER import. 11In comparison, the purpose of the last conserved P 2 0 glutamic acid/glutamine/aspartic acid (E/Q/D) residue of PEXEL is less certain.Some evidence suggests that following cleavage, the P 1 0 and P 2 0 residues, which now cap the mature PEXEL protein, play a role in promoting export across the PVM. 1,14,34ter cleavage, the exposed xE/Q/D motif becomes N-terminally acetylated 35 resulting in an Ac-xE/Q/D cap that was thought to serve as a 'barcode' or a recognition motif for HSP101 engagement in the PV. 36[39] Here, we present evidence that the P 2 0 residue has a dual function.For some PEXEL proteins, P 2 0 mutations greatly reduce PMV cleavage increasing protein retention in the ER and reducing export.
For other proteins, P 2 0 mutations do not inhibit PMV cleavage as much and PEXEL proteins can reach the PV but are translocated less efficiently.We also provide evidence that the length of the region downstream of the PEXEL motif regulates the degree of cargo interaction with HSP101 and ultimately affects protein export across the PVM.

| RESULTS
2.1 | A systematic review of PEXEL motif mutants reveals that a single point mutation of P 2 0 into a positively charged residue produces the most consistent export-blocking effect To better understand the contribution of the PEXEL P 2 0 position on protein export, we conducted a review of P 2 0 mutagenesis experiments that have been performed on various PEXEL proteins (Table S1).We sorted the mutants based on the amino acids incorporated into the P 2 0 position and linked this to the export phenotype, PEXEL processing pattern and N-terminal acetylation (Table S1 'PMV cleavage' and 'N-acetylation') reported for the corresponding mutant.
This in general revealed that alanine (A) mutations in the P 2 0 position have a limited or uncertain effect on export (Table S1).For example, Tarr et al. reported that when the P 2 0 position of KAHRP (knobassociated histidine-rich protein, PF3D7_0202000) was changed to an A, the protein was exported into the RBC. 37This was contrary to a previous study 14 where the reporter was mostly trapped in the PV, despite both groups using very similar constructs (i.e., the first 69aa of KAHRP leader sequence fused to GFP).The same discrepancy was also observed with P 2 0 A mutations of STEVOR and GBP130 (PF3D7_1016300) (Table S1). 1,14,34,39,40These apparently contradictory findings may be attributable to variations in the age of the parasites being studied and consequent differences in promoter activity, as well as to differences in the module(s) appended downstream of the PEXEL motif. 37,39P 2 0 A mutagenesis studies of other PEXEL proteins, such as RESA (PF3D7_0102200), PfEMP3 (PF3D7_0201900), REX3 (PF3D7_0936300) and murine P. berghei CP1 (PBANKA_1246500), did not result in any export defect 4,37,38 (Table S1).
Mutation of the P 2 0 PEXEL position to positively charged amino acids appears to produce the most robust export-blocking effect on exported proteins tested in two different studies (Table S1). 37,40tation to basic amino acids would usually reverse the charge of the P 2 0 position, given that the two most common P 2 0 residues are glutamic and aspartic acids.Curiously, the P 2 0 R mutant of REX3 (PF3D7_0936300) PEXEL also moderately inhibited PMV cleavage of the motif, 37 suggesting that P 2 0 may be involved in PEXEL processing, albeit to a lesser extent than P 1 and P 3 .

| P 2 0 lysine mutations reduce export efficiency in multiple PEXEL proteins by inhibiting plasmepsin V cleavage
To clarify the effects of P 2 0 mutations of different PEXEL proteins expressed under the same conditions we synthesised several fluorogenic peptides containing wildtype PEXEL motifs and mutations thereof of three different PEXEL proteins and determined how well they were cleaved by recombinantly-expressed P. vivax plasmepsin V (PvPMV) (Table S2). 41PvPMV was employed as this protease was more experimentally amenable than the P. falciparum equivalent. 41,42peptide containing the KAHRP (PF3D7_0202000) PEXEL sequence (RTLAQ) with P 3 R to A and P 1 L to A mutations (ATAAQ) served as a control and was not efficiently cleaved by PvPMV compared to the wildtype sequence (Figure 1A).We then introduced P 2 0 mutations into the PEXEL motifs of Hyp1 (PF3D7_0113300) and STEVOR (subtelomeric variable open reading frame, PF3D7_0200400) (RLLTE and RLLAQ, respectively), changing the P 2 0 residue from E to lysine (K) for Hyp1 and from Q to K for STEVOR.In both cases, the P 2 0 K fluorogenic peptides were not efficiently cleaved indicating that P 2 0 K strongly inhibits PEXEL processing (Figure 1A).
Previously, a GFP-tagged STEVOR with P 2 0 A mutation was shown to be trapped at the parasite periphery by microscopy. 34However, the PEXEL processing status of this mutant was not investigated.
We therefore included P 2 0 A mutants of both Hyp1 and STEVOR in our cleavage assay.Compared to the wildtype peptides, cleavage of the P 2 0 A mutants of Hyp1 and STEVOR was moderately inhibited, but this was only statistically significant for Hyp1 (58%).The P 2 0 A mutant is thus less efficiently cleaved than WT and but more efficiently cleaved than the charge reversal P 2 0 K mutant (Figure 1A).
We next sought to determine how the P 2 0 K mutations would affect the trafficking and proteolytic processing of these proteins in parasite-infected RBCs.To generate reporters specific for these proteins we fused the first 113aa of Hyp1, containing the wildtype PEXEL motif RLLTE or a P 2 0 K mutant version, to a reporter cassette comprising nanoluciferase (Nluc), murine dihydrofolate reductase (mDH) and three FLAG epitopes (FL). 42For KAHRP, we fused the first 105aa, including the wildtype PEXEL motif RTLAQ or a P 2 0 K mutant version, to Nluc-mDH-FL.For STEVOR, we fused the first 99aa, including the wildtype PEXEL motif RLLAQ or the corresponding P 2 0 K mutant version, to the same reporter.All six constructs were transfected into the HSP101-HAglmS or the HSP101-HA parasite background line to enable direct comparison between the wildtype and P 2 0 K mutants.Although microscopy and western blot data of wildtype and P 2 0 K Hyp1-Nluc-mDH-FL have been reported previously 42 we have presented new images and blots here for comparison to KAHRP and STEVOR.All P 2 0 wildtype (P 2 0 WT) reporters were exported with For the P 2 0 K mutants, Hyp1 differed from STEVOR and KAHRP with Hyp1 being largely trapped around the nucleus in the ER and P 2 0 K STEVOR and P 2 0 K KAHRP being trapped in the ER and around the parasite circumference in the PV (Figures 1B, S1, and S2).As efficient protein export was observed for some P 2 0 K KAHRP and STEVOR reporter parasites (Figures S1 and S2, respectively), the percentage of reporter signal inside the parasite compartment (including the PV) compared to the whole infected RBC was quantified.This indicated that although the efficiency of export varied widely, there was generally more reporter within the parasite compartment for the P 2 0 K reporters than for the P 2 0 WT reporters (Figure 1C).
We have previously shown that the Hyp1 P 2 0 K reporter was not efficiently cleaved at the PEXEL motif by PvPMV and that this could be why this reporter was not exported and remained trapped in the ER. 42To determine if incorrect cleavage of the KAHRP and STE-VOR reporters could also be responsible for the increased trapping of the reporters in the parasite we performed western blot analysis on parasite lysates expressing WT and P 2 0 K Hyp1-, STEVOR-and KAHRP-Nluc-mDH-FL constructs.In the western blot analysis of Hyp1 parasite lysates probed with anti-FLAG and anti-Nluc IgGs, the predominant correctly cleaved species of WT Hyp1 migrates at 50 kDa with the full-length pre-processed species, expected at 70 kDa, below detection (Figure 1D, lane 2, single asterisk).In contrast, cleavage of the Hyp1 P 2 0 K reporter appears to be upstream of the PEXEL motif, probably near the transmembrane domain (Figure 1D, lane 3, double asterisk). 42Incorrect cleavage correlates with the high ER retention observed by immunofluorescence analysis (IFA), although the mechanism for this is unknown (Figure 1B, D). 42 In contrast, P 2 0 K STEVOR and KAHRP constructs migrated predominantly at the same size as their WT counterparts on a western blot (Figure 1D, lanes 5-8, single asterisk), indicating that the P 2 0 K STEVOR and KAHRP reporters are mostly processed within their PEXEL motif which would explain why they visually appear more efficiently trafficked to the PV and RBC relative to Hyp1 P 2 0 K (Figure 1B).However, we noted additional low abundance miscleaved bands for the KAHRP and STEVOR P 2 0 K reporters that were approximately 3-4 kDa bigger than the PEXEL cleaved species (Figure 1D, lanes 6 and 8, double asterisks).We were able to detect these mis-cleaved species using anti-FLAG antibody, suggesting they are not C-terminally truncated forms of the full-length proteins (Figure 1D, Table S3) and may represent aberrant N-terminally processed forms of the P 2 0 K KAHRP and STEVOR reporters arising from less efficient PEXEL processing, which may account for the small reduction in export.We could also detect the pre-processed form for KAHRP and STEVOR (Figure 1D, triple asterisk).Collectively, both western blot and biochemical analyses suggest that a P 2 0 mutation, particularly to a positively charged residue, can reduce efficient cleavage and cause ER retention for Hyp1 but not for KAHRP and STEVOR.For the latter two proteins, cleavage of the P 2 0 K PEXEL is much more efficient and the proteins traffic to the PV but are translocated less effectively into the RBC than WT reporters.
2.3 | P 2 0 K STEVOR and KAHRP are more soluble than P 2 0 K Hyp1 Consistent with our earlier findings, 42 we observed that the ER-trapped mis-cleaved Hyp1 P 2 0 K reporter was poorly soluble, whereas the low-abundance correctly processed Hyp1 P 2 0 K reporter was readily extractable by hypotonic lysis (Figure 2B, lanes 5-8).The poor solubility of the mis-cleaved species may partly account for its ER retention.To determine if PV trapping of STEVOR and KAHRP P 2 0 K reporter proteins was likewise due to reduced solubility, we performed protein solubility assays on parasite lines expressing the WT and P 2 0 K Nluc-mDH-FL reporters (Figure 2A).The correctly cleaved forms of P 2 0 K KAHRP-(Figure 2B, lanes 13-16, single asterisk) and STEVOR-Nluc-mDH-FL (Figure 2B, lanes 21-24, single asterisk) were largely found in the soluble fraction (Tris Sn), although some was present in the other fractions as well.Additionally, we noticed that the putative low-abundance higher molecular weight forms of P 2 0 K KAHRP and STEVOR (Figure 1D) were again present and evenly distributed in all fractions, including the soluble fraction (Figure 2B, lanes 13-16 and 21-24, double asterisks).These data suggest that mis-cleaved P 2 0 K reporters may remain trapped in the ER because they are less soluble than the correctly cleaved reporters, although the mechanism behind this is not obvious since both the size of the mis-cleaved P 2 0 K proteins and previous mass spectrometric analysis suggest they lack their hydrophobic signal peptides. 42When correctly processed, however, the P 2 0 K reporters are more soluble, which may facilitate their traffic beyond the ER to at least the PV.

| The length of the spacer region is essential for protein translocation across the parasitophorous vacuole membrane
Thus far our data have indicated that the amino acid at the P 2 0 position is important for the correct cleavage of Hyp1 but less so for KAHRP and STEVOR indicating that other residues within and/or bordering the PEXEL motif may also be important for accurate cleavage.
Earlier work has shown that truncation of the amino acid sequence (termed spacer region) that separates the PEXEL motif from a downstream folded protein, such as GFP, influences export. 22,24Interestingly, the N-terminal regions of PNEPs are functionally exchangeable with this spacer region of PEXEL proteins 39 and replacement of the spacer region with the N-terminal sequence of a PV-resident protein inhibits export, 4 suggesting that this region may comprise a bona fide export signal.To investigate whether the spacer region has a role in binding to PTEX, the spacer region of the Hyp1-, STEVOR-and KAHRP-Nluc-mDH-FL constructs were C-terminally truncated, from their original lengths of $50aa, down to 13aa and 3aa preceding the folded domain of Nluc (Figure 3A, B).IFAs of trophozoite-infected RBCs expressing the truncation constructs showed reduced export with reduced spacer length in all three constructs.Quantification of the exported signal across the cell population further revealed that truncation from $50aa to 13aa reduces export by $10%-20%, while export was strongly reduced in 3aa spacer constructs, showing a marked $80% reduction in fluorescence signal relative to the control (Figure 3A-C).This observation contrasts with the mutations of the P 2 0 PEXEL motif alone, performed in the previous section (Figure 1), and other studies, which displayed variable export-blocking phenotypes with different PEXEL protein sequences. 1,14,34,37Co-labelling of microscopy images with EXP2 (PV marker) and PfERC (ER marker) further indicated that Hyp1-Nluc-mDH-FL with a 3aa spacer accumulated mainly in the PV with some signal in the ER overlapping with HSP101-HA (Figure 3A panels 3-7).
IFAs using anti-Nluc with STEVOR-Nluc-mDH-FL and KAHRP-Nluc-mDH-FL parasites showed that the 13aa and 3aa spacer reporters behaved similarly to the Hyp1-reporter (Figure 3B, panels 2, 3 and 5, 6) and also displayed the highest co-localisation with the HA-tagged translocon component HSP101, which we have shown resides within the ER in addition to the PV. 42,43uncation of the spacer did not appear to reduce processing of the PEXEL motif in this context as western blots of the Hyp1-Nluc-mDH-FL truncation constructs showed that each reporter protein migrated according to a predicted mass consistent with PMV-processed versions of the proteins (Figure 4A, B, lanes 2-4, Table S4).Taken together, these results show that the length of the spacer is important for export, post-PEXEL processing.

| Truncation of the spacer region reduces cargo binding with HSP101
The observed co-localisation of all spacer constructs with the translocon components EXP2 and HSP101 in the PV was perplexing as the truncated proteins appeared to have processed PEXEL N-termini but were unable to be exported, suggesting a failure to correctly engage with PTEX.We therefore sought to determine if the truncated PEXEL proteins could bind HSP101 by co-immunoprecipitation.To do this, the Hyp1 truncation constructs (51aa, 13aa and 3aa Hyp1-Nluc-mDH-FL) were transfected into the HSP101-HAglmS parasite line. 424][45] The whole trophozoite-infected RBCs were lysed and incubated with anti-HA-IgG agarose to immunoprecipitate the HA-tagged HSP101 from the sample.Western blot analysis of the eluates revealed a significantly reduced amount of Hyp1-Nluc-mDH-FL co-eluted with HSP101 with decreasing length of the spacer, with 60% and 90% reduction (n = 3) observed in the 13aa and 3aa spacer, respectively, relative to the 51aa spacer Hyp1-Nluc-mDH-FL (Figure 4A, lanes 6-8 and 4C).Importantly, the experiment was performed in the presence of a stabilising ligand WR99210, suggesting that the 13aa and 3aa spacer truncation mutants did not proceed to the unfolding step within PTEX.
The same samples were also subjected to a reciprocal co-immunoprecipitation using anti-Nluc antibodies to pull down the region regulates cargo engagement with PTEX, particularly with HSP101 which is possibly the first point of contact cargo has with PTEX. 19,466 | The sequence requirements of the 13aa spacer for export are relatively unconstrained Having observed that the 13aa Hyp1 spacer still permitted a reasonable amount of export compared to the full-size 51aa spacer we decided to use this as a basis for subsequent mutagenesis experiments due to its small size.The spacers of PEXEL proteins appear to possess little obvious sequence information for putative PTEX recognition apart from appearing relatively unstructured with no conserved domains.To assess whether the 13aa spacer may contain some cryptic trafficking information, we decided to remove it from the original 51aa spacer and fuse this new ΔNT13aa spacer to the Nluc-mDH-FL reporter and express this in parasites (Figure 5A).
Western blots of the Hyp1 ΔNT13aa-Nluc-mDH-FL parasites indicated the reporter protein was of a size consistent with correct PMV cleavage (Figure S3).Microscopy of the ΔNT13aa reporter showed it was efficiently exported and quantification of the fluorescence intensity implies a higher degree of export of the ΔNT13aa reporter than the 13aa spacer (Figure 5B, C).This result indicated that the ΔNT13aa spacer was competent for export and that the information contained in the 13aa spacer is not essential for export of the Hyp1-Nluc-mDH-FL reporter.
Next, we investigated whether certain amino acids within the 13aa spacer were necessary for export.As protein chaperones tend to recognise misfolded proteins via their hydrophobic residues, 47 the four hydrophobic amino acids tyrosine (Y), L, isoleucine (I), and valine (V) in the 13aa Hyp1 spacer TEYKDTLQIKVEQ were mutated to determine if this reduced export.In the absence of predictive tools for HSP100 chaperones, we used a binding predictor for the ER HSP70 chaperone BIP 47 that scans 7aa peptide windows and found The sequence requirements of the 13aa spacer for export are relatively unconstrained.(A) Alignment of the sequences of Hyp1 spacers (including the PEXEL motif) whose trafficking was investigated.(B) Fluorescence microscopy of Hyp1 spacers detected with rabbit anti-Nluc antibody, using mouse anti-EXP2 as a PVM marker, indicated that the ΔNT13aa, 13aa and 13aa.Ser reporters were exported efficiently whereas the Hyp1 13aa.Glu reporter appeared to be more strongly retained in the parasite.(C) Quantification of the % of exported protein signal beyond the EXP2 boundary into the infected RBC compartment indicated that the degree of export was progressively reduced in the Hyp1 13aa.Ser, Hyp1 13aa.Glu and Hyp1 3aa parasites relative to the Hyp1 13aa parasites.The Hyp1 ΔNT13aa parasites exported more efficiently than the Hyp1 13aa parasites.Quantification of export was performed using pooled images from three independent experiments.The following numbers of cells were used for the analysis: 64 (ΔNT13aa), 60 (13aa), 64 (13aa.Ser), 64 (13aa.Glu) and 77 (3aa).Box and whisker plot represents 25th-75th and 10th-90th percentiles, respectively.Statistical significance was determined using one-way ANOVA with Brown-Forsythe test followed by Dunnett's multiple comparisons test for comparison to the 13aa reporter.(*, p-value, <0.05).(D) Nanoluciferase export assay with the same parasite lines produced a similar trend to the microscopic imaging but specifically indicated that most of the reduction in export in the mutant Hyp1 13aa and Hyp1 3aa spacers was due to retention in the PV rather than the parasite.Statistical significance was determined using one-way ANOVA with Brown-Forsythe test followed by Dunnett's multiple comparisons test for comparison to the 13aa reporter.(*, p-value <0.05, **, p-value <0.01).
that mutation of the hydrophobic amino acids to the polar amino acid serine, produced lower scores than for other amino acids (Figures 5A and S4A-C).Therefore, we replaced the hydrophobic amino acids in the 13aa spacer with serine and transfected this Hyp1 13aa.Ser reporter into parasites where, by western blot, the Hyp1 13aa.Ser reporter was of a size consistent with correct PMV cleavage (Figure S3).Microscopic analysis of the Hyp1 13aa.Ser parasiteinfected RBCs indicated that this mutant was exported less efficiently than the WT 13aa spacer (Figure 5A-C).
Next, the hydrophobic amino acids of the 13aa spacer were mutated to glutamic acids (Hyp1 13aa.Glu) as these mutations were scored most poorly by the BIP binding predictor (Figure S4A-C).
Western blots of Hyp1 13aa.Glu-Nluc-mDH-FL parasites indicated PMV processing was correct and microscopic analysis of parasiteinfected RBCs expressing the reporter indicated the Hyp1 13aa.Glu protein was exported significantly less well than the Hyp1 13aa reporter (Figure 5B, C).These results indicated that although there is probably a great deal of flexibility in the amino acid sequences of spacers, some attributes are required such as the presence of hydrophobic amino acids that could be important for general chaperone binding.
As it can be difficult to visibly discern cargo trapped in the PV versus that inside the parasite we employed a recently-developed protein export assay based on the release and detection of Nluc bioluminescence from differentially lysed cellular compartments. 43,48We saw that the Nluc bioluminescence signal exported into the RBC compartments closely followed the same trend as the spacer reporter parasites whose export was measured visibly by microscopy (compare Figure 5C, D).However, the % exported in the Nluc bioluminescence export assay was however often 10%-20% greater than the microscopy measurement, probably because bioluminescence is highly sensitive and can detect lower protein levels dispersed in the large RBC compartment (Figure 5D).Additionally, microscopy does not measure RBC signal in front of or behind the parasite which may result in underestimation of the total signal.Interestingly, the bioluminescence signal of the reporters secreted into the PV compartment increased as export into the RBC compartment decreased indicating that as the spacer length shrank, or its hydrophobic residues were mutated, the proteins were still able to traverse the parasite plasma membrane but were less efficiently translocated into the RBC by PTEX, leaving them to accumulate in the PV.It is also worth noting that the signal retained in the parasite was relatively constant indicating spacer length or mutations did not reduce secretion into the PV.

| DISCUSSION
Here, we sought to understand what element(s) of PEXEL proteins govern recruitment of PTEX to facilitate the export of proteins into the RBC compartment.It was previously hypothesised that the last conserved residue of the PEXEL, which remains on the mature protein, is responsible for interacting with PTEX. 1,2,14Our data partly support this in that mutation of the P 2 0 residue to K can lead to increased trapping of some PEXEL proteins in the PV, most notably for the KAHRP and STEVOR reporters.The P 2 0 K mutation can also greatly reduce the efficiency of PMV cleavage and as such increase protein trapping in the ER, as shown for the Hyp1 reporter.Apart from the efficiency of cleavage of the PEXEL motif for governing PTEX's interaction with the cargo proteins, the spacer region downstream of the PEXEL that separates the PEXEL from the folded functional Nluc region of the reporter protein is also important for trafficking.As the length of spacer increased, so did the reporter's binding to PTEX and the efficiency with which it was exported into the RBC compartment.A small spacer length of 13aa was found to still result in efficient export with hydrophobic residues in the spacer important for export efficiency.Reporters with short or mutated spacers appeared to be still efficiently cleaved by PMV but interacted weakly with PTEX, leaving them trapped in the PV compartment and unable to be exported.One caveat of this study is that our reporter had a folded Nluc domain a defined distance from the upstream PEXEL cleavage site, and we are unsure if similar rules will generally apply to native PEXEL proteins, many of whose structures and functions are unknown.
Other studies have shown that the amino acid in the last position of export motifs is important for export.For example, reporter proteins containing the oomycete effector motif RxLR, 30,39 which lack the P 2 0 residue, were not cleaved by PMV and failed to promote export into the host cell. 30,39Of all possible mutations of P 2 0 , only charge reversal P 2 0 mutations, either to R or K, strongly inhibit PMV cleavage 37 whilst a single A mutation was observed to cause a variable export phenotype. 4,37is was evident from the small reduction of cleavage observed in the in vitro cleavage assay with P 2 0 A mutation compared to the K mutation.
Despite this, the strong inhibition of in vitro cleavage of STEVOR P 2 0 K did not translate to what was observed in parasites as the P 2 0 K reporter protein was efficiently cleaved.It seems likely that suboptimal amino acid variants at PEXEL P 2 0 are still permissive for PMV cleaving in the cellular context, possibly because of the fast kinetics of the proteolytic reaction in vivo or due to concentrated substrate in the ER that permits cleavage of substrates with suboptimal cleavage sites.It is also possible that PfPMV in parasites is more active against the P 2 0 mutants than the PvPMV used in the peptide cleavage assay. 49,50ving previously observed that the ER-trapped Hyp1 P 2 0 K mutant, which was mis-cleaved by an unknown protease upstream of the PEXEL, became quite insoluble and remained in the ER, 42 we explored how the solubility of cleaved PEXEL reporters can influence their export.In contrast to the mis-cleaved Hyp1 P 2 0 K reporter, 42 the solubility of KAHRP and STEVOR P 2 0 K reporter proteins did not appear to decrease relative to their WT counterpart.Given that the KAHRP and STEVOR mutants (a) were efficiently cleaved and trafficked to the PV and (b) did not show a marked decrease in solubility, the reduction in export of these mutants relative to their WT counterparts, albeit slight, is curious.It is plausible that the P 2 0 residue of these proteins is not important for the initial interaction with HSP101 but rather for commitment to stronger downstream interactions with the whole of PTEX required for cargo unfolding and export.Further investigation will be required to substantiate this hypothesis.
Our results also argue that PTEX, more specifically HSP101, recognises a wider region in the cargo than the PEXEL motif and that this is important for export.Truncation of the spacer region in the three different reporter constructs used in this study consistently blocked export without apparently affecting PEXEL processing.Microscopic analysis of the 3aa spacer Nluc-mDH-FL constructs clearly showed that the protein is trapped within the PV area where PTEX is located, consistent with previous reports. 4,24Despite this co-localisation, we have shown that spacer mutants that bind less strongly to HSP101 are exported less efficiently.
While the molecular mechanism of PTEX cargo recognition remains to be elucidated, our data have shed some light into how this process may occur.We found that while a 13aa spacer was sufficient to facilitate modest export, a longer 51aa spacer promoted stronger cargo binding to HSP101, suggesting that perhaps the increased length of the unstructured N-terminal polypeptide increases the likelihood of the cargo stably binding to HSP101.Clp/HSP100 chaperones generally require a recognition signal of at least 10-20 broadly diverse amino acids to initiate polypeptide unfolding and translocation. 51,52r data are consistent with this model and suggest that HSP101 also requires a region within the N-terminal portion of the cargo protein to initiate cargo translocation.This could explain why single P 2 0 K point mutation in the mature PEXEL motif is not enough to prevent the interaction altogether and why the N-termini of PNEPs, despite lacking a mature PEXEL motif, can still act as an export signal. 39Cargo recognition in AAA+ ATPases, particularly the family of Clp/HSP100 chaperones, begins with the pre-unfolding step that is initiated by a low-affinity probabilistic binding of the chaperone to a loosely folded or aggregated region of a protein, followed by a commitment step where the ATPase binds more stably to the cargo protein before unfolding and threading the protein through the chaperone's central cavity. 53It has been shown for the AAA+ ATPase ClpXP, that the length of the cargo polypeptide bound to the inner cavity of the ClpXP affects the commitment step, such that longer polypeptides seem to promote more successful commitment and subsequent unfolding. 54,55Clp/HSP100 chaperones, particularly ClpB and HSP104, are thought to have a similar pre-unfolding step. 52,53,56We therefore propose a model whereby a longer spacer region may increase accessible areas for the initial probabilistic binding step, or stabilise association of exported proteins to HSP101, subsequently leading to less frequent dissociation from the unfoldase (Figure 6A).
Consistently, the Hyp1-Nluc-mDH-FL reporter with a short 3aa spacer region exhibited low-level affinity to HSP101 that greatly reduced export, suggesting that the cargo may have initially associated with HSP101 but later dissociated from the unfoldase because there was insufficient net affinity to proceed to the commitment step.
We also explored which amino acids of the spacer region could promote or inhibit HSP101 binding.The subunits of the second nucleotide-binding domain (NBD2) of HSP101 contain conserved tyrosine residues which are thought to bind the unfolded cargo protein via hydrophobic interactions to help ratchet the cargo through HSP101. 19As the HSP101 subunits undergo allosteric changes powered by ATP hydrolysis, the tyrosines move up and down to help grip and pull on the cargo (Figure 6B).Since it is possible that the tyrosine residues interact with the hydrophobic residues in the spacer region, we mutated the four hydrophobic residues in the 13aa Hyp1 spacer and this was shown to reduce export.Export reduction was particularly strong for the mutation of hydrophobics to charged residues (E) compared to polar residues (S).The deletion of the 13aa spacer region from the 51aa acid spacer still resulted in strong export as the next 13aa downstream from the first 13aa still contained four hydrophobic residues.One possible reason why the 13aa spacer preceding the globular Nluc region was exported much better than the 3aa spacer was that the longer spacer could project further into HSP101 0 s central cavity, down into NBD2 where it could engage the cargobinding tyrosine residues (Figure 6B).
In conclusion, our data suggest dual functions for the P 2 0 position of PEXEL proteins.The first is that it forms part of the PMV recognition sequence for cleavage.In some proteins such as Hyp1, mutation  The expression of Hyp1/ STEVOR/ KAHRP-Nluc-mDH-FL reporters was driven by a bidirectional Plasmodium berghei EF1α promoter that also controlled expression of the blasticidin deaminase drug resistance cassette.The plasmid pEF-Hyp1-Nluc-mDH-FL was derived from plasmid pEF-Hyp1-Nluc-DH-APEX. 43The Hyp1 component of this plasmid contained the first 113aa of Hyp1 (PF3D7_0113300), including the RLLTE PEXEL motif. 43 S5).The overlap region between these two PCR products contained the K mutations.PCR fragments were then sewn together with Generation of the Hyp1-Nluc-mDH-FL plasmids with a truncated spacer region between the PEXEL motif and the start of the Nluc gene was carried out as follows: PCRs were performed with primer 1 paired with primer 7, 8 or 9 to produce spacers of 3aa, 13aa and 51aa, respectively (Table S5).Note that primers 7, 8 and 9 produced Hyp1 PCR products that contained one additional amino acid between the last Hyp1 residue and the start Met of Nluc.The PCR products were ligated into pJET1.2/bluntand screened as above.

| Indirect immunofluorescence analysis
IFA was performed essentially according to Tonkin et al. 60 where infected RBCs were settled onto a poly-L-lysine (Sigma, P8920) coated coverslip and fixed with 4% paraformaldehyde/0.0075%glutaraldehyde.Following fixation, the cells were permeabilised with 0.1 M glycine/0.1% Triton X-100 for 12 min at room temperature.
Coverslips were probed overnight with primary antibodies (Table S6).
The cells were washed and probed with fluorescent-labelled secondary antibodies (Alexa Fluor 594 and 488 nm) (Table S6) for 1h at room temperature.Fixed material was mounted in VECTASHIELD with DAPI and imaged on Zeiss Cell Axio-Observer (Carl Zeiss).Image acquisition was performed with Zen Blue imaging software.

| Quantification of Imaging Data
All Images were visualised and analysed using ImageJ/FIJI essentially according to Gabriela et al. 42 Briefly, a region of interest (ROI) was selected by tracing the periphery of the infected RBC on the DIC channel.The parasite area (including the PV) was selected using the outermost staining of anti-EXP2 as a guide.

| Western blotting
Proteins were transferred from the gels to a nitrocellulose membrane using iBlot ® Blotting System (Invitrogen).The blots were blocked in 1% casein in PBS and probed with primary antibody (Table S6) diluted in the blocking buffer overnight at 4 C.The blots were washed and probed with fluorescent-labelled (Alexa Fluor 700 and 800 nm) or horseradish peroxidase (HRP)-labelled secondary antibodies (Table S6) in blocking buffer for 1 h at RT followed by three washes with 1Â PBS.The fluorescent secondary antibodies were visualised with a LI-COR Odyssey FC imaging system.Densitometry analysis was performed with Image Studio v. 1.0.

| Biochemical PEXEL cleavage assays
The cleavage assay was performed as described by Hodder et al. 41 2 nM of P. vivax PMV in buffer (25 mM Tris-HCl pH 6.

E 1 P 2 0
Alanine and Lysine mutations reduce the proteolytic processing of the PEXEL motif.(A) 5 μM fluorogenic peptides were incubated with 2 nM recombinant PvPMV and assayed at 20 C. Fluorescence data was normalised to the WT substrates (n = 3).PEXEL motifs are indicated below the x-axis with mutated resides indicated in red.Statistical significance was determined using ordinary one-way ANOVA with Brown-Forsythe test followed by Šídàk's multiple comparisons test.(****, p-value<0.0001;***, p-value<0.001).(B) Representative IFA images (n = 3 independent replicates) of HSP101-HAglmS parasites expressing WT and P 2 0 K PEXEL motifs of Hyp1, STEVOR, and KAHRP-Nluc-mDH-FL reporter proteins.Parasite cells were probed with anti-EXP2 and anti-HA antibodies to visualise the PTEX components within the cell.The Nluc-mDH-FL proteins were localised using anti-Nluc antibody.The red bar in the schematic picture of the construct indicates the location of the PEXEL motif.Scale bars, 5 μm.DIC, Differential Interference Contrast.DAPI (4 0 ,6-diamidino-2-phenylindole; blue) was used to stain parasite nuclei.(C) Quantification of the Nluc fluorescence signal of STEVOR and KAHRP-Nluc-mDH-FL constructs within the boundary of the EXP2 staining (including the parasite cell and its PV).Quantification was performed using pooled images from three independent experiments.The following numbers of cells were used for the analysis: KAHRP P 2 0 WT (n = 151) and K (n = 161), STEVOR P 2 0 WT (n = 135) and K (n = 145).Box and whisker plot represents 25th-75th and 5th-95th percentiles, respectively.Statistical significance was determined using ordinary one-way ANOVA.(****, p-value<0.0001).(D) Representative western blot (n = 3) of lysates made from mid-stage trophozoites expressing Nluc-mDH-FL constructs probed with anti-FLAG and anti-Nluc IgGs.The identity of various protein species was based on their observed sizes.***, full length proteins., **, mis-cleaved P2' K Hyp1 (58.1 ± 1.2 kDa, n = 10), KAHRP (51.4 ± 1.2 kDa, n = 3) and STEVOR (53.8 ± 1.7 kDa, n = 3)-Nluc-mDH-FL.*, correctly PEXEL-cleaved species of the Nluc-mDH-FL proteins.< is a cross-reactive protein.Hyp1-Nluc-mDH-FL blots are shown separate to KAHRP and STEVOR blots as they were run on different gels.The Neg lane contains protein from parasites not transfected with the Nluc-mDH-FL reporter and were part of the same blot as the KAHRP and STEVOR samples but on non-adjacent lanes indicated by the white vertical line.Blots were probed with anti-PTEX150 antibody as a loading control.most of the signal in the RBC compartment (Figures 1B, C, S1, and S2).
Figure4B, D).Taken together, these results demonstrated that the spacer

F I G U R E 4
Truncation of the spacer region of Hyp1-Nluc-mDH-FL reduces its interaction with HSP101.(A) Western blot of anti-HA immunoprecipitation (IP) of HSP101-HAglmS parasites expressing various truncations of the Hyp1 spacer region (n = 3) as indicated above each lane.HSP101-HAglmS parasites not transfected with a Hyp1-Nluc-mDH-FL reporter are represented by '-'.Immunoblots were performed to detect other PTEX components (EXP2 and PTEX150) as a positive control and GAPDH as a negative control.Hyp1 spacer mutants were visualised using anti-FLAG antibody.(B) Western blot of the reciprocal anti-Nluc IP of the truncated spacer reporters (n = 4).Immunoblots were performed using anti-HA to detect HSP101 and GAPDH as a negative control.(C) and (D) Densitometry of the IP performed on (A; left) and (B; right).For anti-HA IP, the intensities of co-immunoprecipitated 13aa and 3aa spacer Hyp1-Nluc-mDH-FL were normalised to the value of the 51aa spacer.For anti-Nluc IP, the intensities of co-immunoprecipitated HSP101 bands in the 13aa and 3aa spacers were normalised to the 51aa spacer.Error bars, ±SD.Statistical significance was determined using one-way ANOVA with Brown-Forsythe test followed by Tukey's multiple comparisons test.(****, p-value<0.0001).

of P 2 0 2 0 4 . 1 |
to K greatly reduces PMV binding and/or successful proteolytic activity but in other PEXEL proteins such as STEVOR and KAHRP P is not as critical for PMV cleavage in the cellular context.A lack of PMV cleavage probably results in ER retention as evidenced by the substantial retention for poorly cleaved Hyp1 versus efficiently cleaved STEVOR and KAHRP.Successful cleavage of the PEXEL P 2 0 K mutant appears to permit trafficking to the PV as STEVOR and KAHRP more efficiently reach the PV than the poorly cleaved Hyp1.Once in the PV, however, the STEVOR and KAHRP P 2 0 K reporters were not exported into the RBC compartment as efficiently as their wild-type counterparts.This could be due to the reporters not being efficiently recognised by PTEX or because the reporters were in a PTEX-free sub-compartment of the PV.The latter, however, is unlikely, because recent split-GFP experiments indicate that PVresident proteins have full access to PTEX. 57It is possible therefore that the charge reversal P 2 0 mutant somehow binds to PTEX less efficiently leading to less engagement, unfolding and eventual export into the RBC compartment.4 | MATERIALS AND METHODS Culture of P. falciparum transfectants Asexual blood-stage P. falciparum (3D7 background) was cultured according to the established protocol. 58Cultures were routinely maintained in complete RPMI media containing RPMI-1640 base medium supplemented with 2.5 mM HEPES, 367 μM hypoxanthine, 31.25 μg/ mL Gentamicin, 25 mM NaHCO 3 and 0.5% (w/v) Albumax II (Invitrogen).Prior to transfection, 100 μg of plasmid DNA was resuspended in TE and cytomix (25 mM HEPES, 120 mM KCl, 0.15 mM CaCl 2 , 2 mM EGTA, 5 mM MgCl 2 , 10 mM K 2 HPO 4 /KH 2 PO 4 pH 7.6) and mixed with packed RBCs as per Hasenkamp et al.59After electroporation using Gene Pulser XCell System (BioRad), the uninfected RBCs were mixed with 20 μL of HSP101-HAglmS trophozoite-stage parasites42 which were allowed to invade the transfected RBCs for 2 cell cycles before starting selection with 2.5 μg/mL blasticidin S.

4. 2 |
Generation of PEXEL P 2 0 mutants and spacer truncation constructs A synthetic murine dihydrofolate reductase (mDH) gene fragment with C-terminal 3x FLAG epitopes (Bioneer Pacific) was ligated into the Nluc-DH-APEX plasmid using SpeI and MluI enzymes to remove the previous mDH-APEX gene cassette.Generation of the P 2 0 lysine (K) mutation of the Hyp1-Nluc-mDH-FL was performed as follows: Hyp1 region was first amplified as two overlapping PCR fragments with the primer pairs 1 & 2 and 3 & 4 (Table

primer pair 1
& 4 via overlapping PCR and ligated into the pJET1.2/blunt plasmid (ThermoFisher Scientific).Mutagenesis was confirmed via standard Sanger sequencing (service provided by Monash Micromon Genomics) of the isolated plasmids.The mutant P 2 0 K Hyp1 fragment was released from pJET1.2/blunt using XhoI and NcoI and ligated into pEF-Hyp1-Nluc-mDH-FL to replace the wildtype Hyp1 fragment.

(
PF3D7_0202000) in the Nluc-mDH-FL reporter plasmid, synthetic leader sequences encoding the first 99aa of STEVOR and 105aa of KAHRP containing XhoI and NcoI restriction sites were first obtained F I G U R E 6 Hypothetical model for the HSP101-cargo recognition.(A) Side and top views of a scale model of PTEX based on cryoelectron microscopy structure showing how the Hyp1spacer with attached Nluc (red) could extend into PTEX's central cavity.19(B) Enlarged view of the central HSP101 cavity showing the spacer length required to engage the helical protein binding-regions of HSP101-nucleotide binding domain 2 (NBD2) which could apply an unfolding force to the cargo proteins for translocation.PTEX model was adapted from the published CryoEM structures (PDB: 6E10 and PDB: 6E11).19 as string oligos from GeneART (ThremoFisher Scientific).P 2 0 K mutations for both STEVOR and KAHRP leaders were also synthesised as above.The synthetic DNA sequences were ligated into pJET1.2/bluntand validated by sequencing prior to transfer via XhoI and NcoI sites into the pEF-Hyp1-Nluc-mDH-FL plasmid replacing the Hyp1 sequence.

4 . 5 |
Fluorescence intensity within the parasite area was then recorded (parasite fluorescence; PF).Fluorescence intensity within the ROI of the same infected cell was also recorded (total fluorescence; TF).The estimated '%Fluorescence in Parasite' is the PF as a percentage of the TF after both values were corrected for the background fluorescence.Chemical cross-linking of P. falciparum culture and immunoprecipitation RBCs infected with the HSP101-HAglmS /PEXEL-Nluc-mDH-FL parasites were enriched through magnetic purification and their proteins were solubilised in 20Â pellet volume of IP lysis buffer (1% Triton X-100, 0.1% SDS, 150 mM NaCl, 10 mM Tris-HCl pH 7.4) supplemented with cOmpleteTM Protease Inhibitor Cocktail (Roche) and subjected to 2 freeze and thaw cycles.The lysate was clarified by centrifugation and incubated overnight at 4 C with anti-HA monoclonal agarose (Sigma-Aldrich).Following incubation, the agarose beads were washed 5Â with 1 mL IP lysis buffer and the proteins were eluted with 50 μL 2Â NRSB (100 mM Tris-HCl pH 6.8, 4 mM EDTA, 4% SDS, 0.01% bromophenol blue, 20% (v/v) glycerol).For Nluc IgG immunoprecipitations, the parasite lysates were incubated overnight at 4 C with 10 μg IgG-purified anti-Nluc antibody.Following incubation, protein A-Sepharose 4B (Invitrogen) was added to bind the immune complexes and samples were incubated for an additional 1 h at RT.The beads were washed, and proteins eluted as above.In all cases, both input and elution fractions were reduced in 200 mM DTT at 70 C for 5-10 min prior to SDS-PAGE separation and western blotting.
4 and 25 mM MES pH 6.4) was incubated with 5 μM FRET peptide substrates representing WT and mutant KAHRP, STEVOR and Hyp1 sequences (Table S2) in a total volume of 20 μL.Samples were incubated at 20 C for 20 h and measurement was carried out using Envision plate (PerkinElmer) reader (ex.340 nm; em.490 nm).Biochemical P. vivax PMV inhibitory assays (20 μL total volume) were performed using 2 nM P. vivax PMV in buffer (25 mM Tris-HCl and 25 mM MES, pH 6.4) with 5 μM FRET KAHRP_WT fluorogenic peptide.Assay reactions were incubated at 37 C for 2 h in the presence of peptides (10 points dose-response, 1 in 2 dilution series starting at 100 nM) representing STEVOR and Hyp1 sequences (