To enable functional studies and accurate analysis of HSP101’s subcellular localisation, we generated a transgenic parasite line expressing a triple hemagglutinin (HA) tagged-HSP101 for antibody detection and appended the gene with a glmS riboswitch so its expression could be inducibly knocked down [44]. To create the PfHSP101-HAglmS line, P. falciparum 3D7 parasites were transfected with the pHSP101-HAglmS plasmid and were cycled on/off WR99210 until the plasmid had integrated into the HSP101 locus (Fig 1A). PCR analysis of two clonally derived lines using oligonucleotides that distinguish between wildtype and integrant genotypes revealed that integration of the HAglmS sequence into the hsp101 locus had occurred in both clones (Fig 1B). Western blot analysis using anti-HA IgG, further confirmed the presence of HA-tagged HSP101 protein (Fig 1C). Addition of 2.5 mM glucosamine (GlcN) at ring-stage resulted in ~75% (SD ± 22.3%, n = 4) knockdown of HSP101 (Fig 1D), leading to parasites stalling at the ring-trophozoite transition the following cycle and parasite death (Fig 1E) as well as a failure to export proteins (Fig 1F). The effects of glmS-induced knockdown of PfHSP101 that we observed were consistent with the previously reported phenotypes of the knockdown of P. berghei HSP101 [36] and conditional inactivation of HSP101 in P. falciparum [35].

We have previously shown that in addition to the fully formed PTEX complex, HSP101, EXP2, and PTEX150 also exist in subcomplexes [30, 31]. The functional significance of these subcomplexes, and their relations to the full PTEX complex are still not known. A study by Beck et al. showed that HSP101 still binds to exported proteins when dissociated from EXP2 and PTEX150 [35], suggesting that HSP101 may recognise its cargo prior to forming the PTEX complex with the remaining core components for the subsequent translocation process. To investigate this model, we sought to assess the impact of losing individual PTEX components on complex formation. We employed our HSP101-HAglmS line (Fig 2A), along with the previously published PTEX150-HAglmS (Fig 2B) and EXP2-HAglmS (Fig 2C) parasites [36, 37], to enable knockdown of all core PTEX components via GlcN-induced mRNA degradation. Native protein complexes from erythrocytes infected with late ring stage parasites that had been treated +/- 2.5 mM GlcN for one cell cycle were fractionated by blue native PAGE (BN-PAGE) and analysed via western blotting. In the absence of GlcN treatment, a >1236 kDa band, consisting of all three core components of PTEX [30, 31], was common to all parasite lines (Fig 2A, 2B, and 2C odd numbered lanes). In the same lanes, PTEX150 and EXP2 formed smaller oligomers of ~500 kDa and ~700 kDa, respectively, whilst HSP101 produced oligomers of >720 kDa, consistent with what has been previously described (Fig 2A) [30, 31, 45]. Strikingly, we observed that following the knockdown of the HSP101 component, PTEX150 and EXP2 were still present in the >1236 kDa, although this complex was now devoid of most HSP101 (Fig 2A, lanes 4 and 6). Knockdown of HSP101 also revealed that anti-HSP101 antibody cross reacted with ~1000 and ~200 kDa bands that were not recognised by the anti-HA antibody (Figs 2A lane 2 asterisks and S1A). In contrast, individual knockdown of both PTEX150 and EXP2 resulted in the loss of the >1236 kDa complex (Fig 2B lanes 8 and 12, and 2C lanes 14 and 16).

One could anticipate that the PTEX150-EXP2 subcomplex would have a lower molecular weight relative to the full translocon due to the loss of HSP101. If PTEX150 and EXP2 retained the heptameric-heptameric arrangement of the full PTEX complex in the absence of HSP101 [32], the resulting tetradecameric subcomplex of PTEX150-EXP2 would have a theoretical size of approximately ~1300 kDa. The difference size between HSP101-PTEX150-EXP2 and PTEX150-EXP2 subcomplex might still be too large for the BN-PAGE system to resolve accurately [32]. Nevertheless, we still observed a small but reproducible reduction in the size of the >1236kDa EXP2 and PTEX150 subcomplex after HSP101 knockdown, consistent with the presence of HSP101-free EXP2-PTEX150 subcomplex (Fig 2A lanes 5 vs 6). These data collectively suggest that PTEX150 and EXP2 are still able to associate and form a large subcomplex that is not dependent on the HSP101 association (Fig 2D).

Notably, the loss of the >1236 kDa complex upon EXP2 knockdown caused a remarkable accumulation of PTEX150 ~500 kDa oligomer (Fig 2C lane 16) that did not label with anti-HSP101 antibody (Fig 2C lane 14). This observation suggests that in the absence of EXP2, PTEX150 alone cannot associate with HSP101, and forms a different oligomeric state. There was no equivalent increase of the EXP2 ~700 kDa oligomer upon destabilisation of the >1236 kDa complex by knockdown of PTEX150 (Fig 2B lane 12) or HSP101 (Fig 2A lane 6). The EXP2 knockdown, however, clearly reduced the level of the ~700 kDa oligomer suggesting that perhaps the EXP2 oligomer is not a part of the PTEX complex (Fig 2C lane 18). The ~700 kDa EXP2 species may therefore represent a dedicated non-translocon EXP2 pool which serves other functions, such as forming the PVM nutrient channel [38]. As observed with the EXP2 knockdown, we did not see the appearance of subcomplexes consisting of EXP2 and HSP101 upon PTEX150 knockdown, suggesting that EXP2 similarly is unable to interact with HSP101 in the absence of PTEX150.

To validate this finding, we performed co-immunoprecipitations to examine direct interactions between each of the core PTEX components following the knockdown of each subunit (S1B and S1C Fig). As expected, knockdown of HSP101-HAglmS with 2.5 mM GlcN followed by immunoprecipitation of PTEX150 from parasite lysates revealed a non-significant change (p = 0.9319, n = 3) in PTEX150 interaction with EXP2, despite little HSP101 being present (S1B lanes 3 and 4, and S1C Fig). In contrast, knockdown of EXP2 resulted in less interaction between PTEX150 and HSP101 (p = 0.0036, n = 3, S1B lanes 7 and 8, and S1C Fig). Interestingly, although there was a trend for reduction in EXP2 and HSP101 interaction upon knockdown of PTEX150 (S1B lanes 11 and 12, and S1C Fig), we obtained noticeably varied results, even after multiple repeats (p = 0.2570, n = 5), S1C Fig). This may be due to the residual amount of PTEX150 (as a result of inefficient PTEX150 knockdown relative to HSP101 and EXP2), permitting the reconstitution of PTEX complex during the prolonged incubation time when performing immunoprecipitation. BN-PAGE does not have this confounding factor and hence the knockdown effect can perhaps be more readily visualised, even when PTEX150 knockdown is not complete. Nevertheless, the co-immunoprecipitations support the BN-PAGE data and suggest that following reduction of HSP101, PTEX150 can still engage with EXP2 to form the large >1236 kDa subcomplex. Collectively, our data indicate that PTEX assembles in two broad steps where EXP2 and PTEX150 first assemble to form a stable subcomplex, which then enables HSP101 docking to form the fully functional translocon.

We have previously reported that HA-tagged HSP101, displays dual localisation in the PV and inside the parasite [42]. Interestingly, an independent study also showed that the 3xFLAG-tagged HSP101 displayed a similar localisation pattern [43]. We therefore decided to investigate this phenomenon in more detail. We first confirmed the dual localisation of HSP101 in our HSP101-HAglmS line by performing immunofluorescence assays (IFAs) on paraformaldehyde/glutaraldehyde fixed cells. Anti-HA staining indeed showed labelling at two main sites: the parasite’s periphery, and the perinuclear area inside the parasite (Fig 3A panels 1 and 2). This internal staining pattern can be observed throughout ring to early-mid trophozoite stage parasites and becomes less pronounced at the later stage (Fig 3A panel 3), coinciding with the duration of active protein export [31, 35]. To confirm that this staining is an accurate representation of HSP101 localisation rather than non-specific staining of our anti-HA antibody, PTEX150-HAglmS and EXP2-HAglmS parasites were also labelled using the same anti-HA antibody and imaged under the same conditions. In contrast to the anti-HA labelling that localised HSP101-HA at the parasite’s periphery and internally (Fig 3A panels 1–3), both PTEX150-HA and EXP2-HA localised solely at the parasite’s periphery, consistent with previous observations (Fig 3A panels 4–7) [29], confirming that the internal anti-HA labelling observed in the HSP101-HAglmS parasite line was specific.

To quantify the relative proportion of cells exhibiting dual localisation of HSP101-HA in our parasite population, the intensity of internal and peripheral fluorescence signals was measured separately for more than 20 mid-trophozoite cells (where the internal signal is most visible, and the ER-PV separation is clear) and expressed as an internal/PV ratio. A higher internal/PV ratio would therefore indicate that the measured cell has more internal signal. Indeed, HSP101-HA staining displayed a higher overall internal/PV ratio relative to EXP2-HA and PTEX150-HA, suggesting that HSP101-HA localises within the parasite in a significant number of cells in the population (Fig 3B).

We also sought to validate this finding using a biochemical approach. For this, we utilised a proteinase K protection assay to assess the susceptibility of HSP101 and EXP2 to proteolytic degradation upon erythrocyte and PVM permeabilisation by equinatoxin II and saponin (Fig 3C) [46, 47]. We have previously used this assay to show that the C-terminal region of EXP2 resides in the PV lumen and can be readily degraded by proteinase K upon PVM permeabilization to leave only the N-terminal transmembrane domain intact [33]. We assumed that if an internal parasite pool of HSP101 exists, it would be protected against protease degradation at a concentration sufficient to fully degrade the C-terminal part of EXP2. Quantitative proteolysis of full-length EXP2 (probed with rabbit serum against the full-length EXP2) after equinatoxin II and saponin treatment indicated that in the supernatant and pellet fractions, EXP2 was degraded into smaller fragments that when combined, resulted in >90% degradation (Fig 3D and 3E) and complete degradation of PTEX150 at 20 μg/mL proteinase K (S2 Fig). In comparison, treatment of HSP101 with 20 μg/mL proteinase K only degraded 40% of the total full-length protein for combined soluble and pellet fractions, with the degraded and resistant proportions likely representing vacuolar and internal parasite HSP101, respectively [29, 30].

When parasites were treated with equinatoxin II and 0.25% Triton X-100 to disrupt all parasite membranes, EXP2, HSP101, and PTEX150 were completely degraded by treatment with 20 μg/mL proteinase K, indicating the three proteins were susceptible to proteinase K treatment when completely solubilised (Figs 3E and S2). Thus, both microscopy and biochemical data support the presence of an internal pool of HSP101.

Having confirmed the presence of an internal pool of HSP101 in our parasites, we next sought to determine HSP101’s subcellular localisation. By IFA, the perinuclear HSP101 signal was reminiscent of ER localisation. Indeed, P. berghei HSP101 has also been reported to localise to the ER in ring and trophozoite stages [41]. To test whether the internal HSP101 we observed in P. falciparum also localised to the ER, IFAs were first performed on tightly synchronised HSP101-HAglmS parasites at the late ring stage, where the parasite cell size is sufficiently large to clearly visualise dual localisation of HSP101 (Fig 3A panels 1 and 2). As expected, immunolabelling with anti-HA and ER-resident calcium-binding protein PfERC showed co-localisation of the internal pool of HSP101-HA with PfERC, indicative of ER-localisation of PfHSP101 (Fig 4A panels 1 and 2).

There are at least two possibilities to explain the ER-localisation of HSP101. First, it is possible that the ER-located HSP101 represents newly synthesised HSP101 which was recently shown to occur up to 20–24 hours post infection [38, 48] although, this does not explain why we could still observe ER staining at mid to late trophozoite stage (>24 hours) when HSP101 was not actively being translated [38, 49]. The second possibility is that HSP101 has a unique localisation and function in the ER independent of PTEX. To explore these alternatives, we first investigated whether secretion of HSP101 occurs throughout the late ring to mid trophozoite stage where we still observed ER HSP101. To do this, we first treated early ring stage HSP101-HAglmS parasites (0–4 hours post infection) with an increasing concentration of Brefeldin A (BFA), which has been shown to inhibit ER-to-Golgi secretion of Plasmodium protein [50, 51]. Parasites were then fixed and labelled with anti-HA and PfERC antibodies 20 hours post treatment, when they reached late ring-early trophozoite stage. As expected, there was a dose-dependent accumulation of HSP101 inside the parasite’s ER as measured by increased co-localisation with anti-PfERC signal by IFA. At 2 and 10 μM BFA concentration, the parasite’s ER increasingly collapsed into a single HSP101 and ERC staining body (Fig 4A panels 3–6). This accumulation appeared to also coincide with the disappearance of the anti-HA staining at the parasite’s periphery resulting in greater co-localisation of HSP101-HA and ERC (Fig 4B).

We next performed the BFA trapping experiment on mid stage trophozoites (24–28 hours post infection) in which there was little new HSP101 being synthesised [38]. Here, parasites were treated with a higher concentration of BFA (18 μM) but for a shorter period (5 hours) to enable visualisation of parasite cells prior to the parasites reaching schizogony. With anti-HA labelling, increased accumulation of HSP101 was still observed in the ER, as well as the disappearance of HSP101 signal from the PV (Fig 4C panel 1–2 versus 3–4). Anti-EXP2 and PTEX150 co-labelling was also performed to gauge the impact of BFA treatment on the localisation of the remaining PTEX components during mid trophozoite stage. Although the transcription profile of PTEX150 matches that of HSP101 [38, 49], BFA-induced ER retention was much less pronounced for PTEX150 and in most cases, PTEX150 remained localised predominantly at the parasite’s periphery (Fig 4C, panels 1–2 versus 3–6). On occasion, a faint signal of PTEX150 inside the parasite cell can be observed (Fig 4C panel 3 white arrow) that could represent a small amount of trapping of newly synthesised protein. In contrast, EXP2 labelling displayed perinuclear staining more visibly in BFA-treated parasites (Fig 4C panels 5–6 white arrows), likely because synthesis of new EXP2 peaks during the trophozoite stage [29, 38]. EXP2 staining could also be observed around the parasite’s periphery, but sometimes in the form of PTEX150-free blebs (Fig 4C panel 5), perhaps due to the stress caused by failure to secrete proteins important for maintaining proper PVM organisation, or the trapping of cargo proteins [38, 42]. Quantification of the Internal/PV ratio of the anti-HA, anti-PTEX150, and anti-EXP2 signals further revealed that a significant increase can only be observed for anti-HA (HSP101-HA) staining upon BFA treatment, consistent with the qualitative observation described above (Fig 4D). Overall, the data collectively highlights that active secretion of HSP101 still occurs during active protein export, and this observation cannot be sufficiently explained by secretion of newly synthesised HSP101 protein alone since PTEX150, although transcribed at the same time as HSP101, had mostly been secreted into the PV [30, 38]. Whether all ER-localised HSP101 will eventually be secreted into the PV, or there exists a pool of HSP101 that is stably localised in the ER is unclear at this point.

It has been previously proposed that HSP101 might chaperone exported proteins from the ER into the PV [39, 40]. The internal localisation of HSP101 and the assembly model of PTEX we have proposed here are consistent with this model. We therefore investigated the possible involvement of HSP101 in the early protein export pathway in the ER where we first sought to determine whether HSP101 in the ER co-localises with proteins destined for export. For this, we generated a reporter construct called Hyp1-Nluc-mDHFR-3xFLAG. This construct contains the first 113 amino acids of Hyp1 (PF3D7_0113300) including 52 amino acids downstream of the plasmepsin V cleavage site [52]. The reporter used here was modified from an earlier version [52] to contain a murine dihydrofolate reductase (mDHFR) domain to stabilise the Hyp1-Nluc cargo within PTEX and a triple FLAG tag [53] to facilitate downstream purification of the protein. Reporter constructs were transfected either to the HSP101-HAglmS or HSP101-HA parasite lines [29] and as anticipated, parasite IFAs demonstrated that Hyp1-Nluc-mDHFR-3xFLAG was efficiently exported into the host-cell when expressed in early to late trophozoite parasites under the P. berghei EF1a promoter (Fig 5A panels 1 and 3). Treatment of parasites with the anti-folate compound WR99210, resulted in a stabilised mDHFR domain and trapping of the reporter protein within the parasite and with EXP2 in the PV (Fig 5A panel 2).

To test whether HSP101 recognises the Ac-xE/D/Q N-terminus generated after cleavage of the PEXEL motif in the ER, we created a single Glu to Lys charge reversal mutation of the conserved 5^th residue of the Hyp1 PEXEL motif as this type of radical mutation has been previously reported to block export [54]. The resulting P5 Lys Hyp1-Nluc-mDHFR-3xFLAG reporter protein failed to be exported as anticipated, but interestingly appeared to co-localise with ERC in the ER regardless of WR99210 treatment (Fig 5A panels 4–6). We used this ER-trapped construct and assessed its co-localisation with the ER-located HSP101, which revealed a qualitatively high degree of co-localisation around the perinuclear area of the parasite (Fig 5A panel 7).

To further improve the resolution of subcellular proteins, we imaged HSP101-HA parasites expressing P5 Lys Hyp1-Nluc-mDHFR-3xFLAG using 3D structured illumination microscopy (3D-SIM). Co-labelling of paraformaldehyde/glutaraldehyde-fixed cells with anti-HA and Nluc IgGs revealed partial co-localisation between the P5 Lys Hyp1-Nluc-mDHFR-3xFLAG and the internal pool of HSP101-HA in the perinuclear region and in close apposition to the periphery of the parasite (Fig 5B top panel). But whilst P5 Lys Hyp1-Nluc-mDHFR-3xFLAG signal was distributed evenly (perhaps due to saturation of signal caused by accumulation of the protein in the ER), internal HSP101 sometimes formed small puncta or concentrated to a certain area (Fig 5B white arrows) that overlapped with P5 Lys Hyp1-Nluc-mDHFR-3xFLAG throughout the z-stack. In contrast, co-labelling of anti-EXP2 with anti-Nluc antibodies revealed almost no co-localisation, consistent with EXP2 residing only at the PVM (Fig 5B lower panel). Pearson’s coefficients of anti-HA (HSP101) and anti-Nluc co-labelling on maximum projection images yielded an average value of 0.67 (SEM ±0.02, n = 24) (Fig 5C), which is significantly higher than the co-localisation value for anti-Nluc and anti-EXP2 (0.12 ±0.02, n = 21) (Fig 5C). This data suggests that the ER-localised HSP101 and the PEXEL reporter protein localise in proximity in the ER.

To gain further insights into the export blocking phenotype of P5 Lys mutation in Hyp1, protein extract from parasites expressing the wildtype (WT) or P5 Lys Hyp1-Nluc-mDHFR-3xFLAG were separated on SDS-PAGE and analysed by western blot using anti-FLAG and anti-Nluc antibodies. WT Hyp1-Nluc-mDHFR-3xFLAG protein migrated at a predicted size of 50 kDa, consistent with efficient plasmepsin V cleavage within the PEXEL site, although a small amount of larger full-length protein could be observed (Fig 5D). In contrast, P5 Lys Hyp1-Nluc-mDHFR-3xFLAG runs predominantly at a higher molecular weight (~56 kDa), indicative of miscleaved protein, followed by a small proportion of PEXEL cleaved protein (87% and 13% (SD ± 5.3%, n = 6), respectively (Fig 5D). To determine if plasmepsin V was able to cleave the mutant reporter, a synthetic Hyp1 PEXEL peptide containing the P5 Lys mutation was treated with recombinant P. vivax plasmepsin V since the P. falciparum protease is more difficult to express (S3 Fig) [55] Relative to the WT Hyp1 peptide, cleavage of the P5 Lys reporter was reduced >90%, which is consistent with the proportion of miscleaved protein observed in western blots (S3 Fig). As the miscleaved 56 kDa P5 Lys Hyp1-Nluc-mDHFR-3xFLAG band is smaller than the full-length band it might be processed upstream of the PEXEL motif (Fig 5D). Mass spectrometry of immunoprecipitated forms of both Hyp1 reporter proteins (S4A Fig) indicated peptide coverage was only present downstream of the PEXEL cleavage site for the WT protein (S4B Fig) whereas peptides corresponding to the region upstream of the PEXEL were detected with P5 Lys protein (S4C Fig). This data, together with the in vitro plasmepsin V cleavage assay, suggests that PEXEL processing of the P5 Lys Hyp1-Nluc-mDHFR-3xFLAG is significantly reduced, leading to the formation of a higher molecular weight protein that has its PEXEL motif intact (S4C Fig). The implication of this observation will be discussed later.

There is a noticeable difference observed by IFA between the complete ER trapping of the P5 Lys Hyp1-Nluc-mDHFR-3xFLAG and other reported PEXEL mutants, which despite having poorly-cleavable PEXELs, are often still secreted to the PV as soluble proteins [21, 54, 56]. We therefore performed protein solubility assays [56] on WT and P5 Lys Hyp1-Nluc-mDHFR-3xFLAG. Whilst WT Lys Hyp1-Nluc-mDHFR-3xFL was readily extracted by hypotonic lysis (S5 Fig) indicative of a soluble protein, we found that the miscleaved species of P5 Lys Hyp1-Nluc-mDHFR-3xFLAG was mostly extractable only in the presence of Triton X-100 (S5 Fig), suggesting that it was strongly membrane-associated. This result suggests that complete ER retention of P5 Lys Hyp1-Nluc-mDHFR-3xFLAG could be due to the miscleaved protein remaining associated with the ER membrane and why this only occurs for P5 Lys Hyp1-Nluc-mDHFR-3xFLAG is not clear.

Overall, these results indicate that mutation of the P5 Hyp1 residue to Lys blocks efficient cleavage by plasmepsin V and results in nearly complete trapping of the reporter protein in the ER. Furthermore, a substantial amount of HSP101 co-localises with P5 Lys Hyp1-Nluc-mDHFR-3xFLAG in the parasite and not in the PV.

We next performed co-immunoprecipitations to investigate if ER-localised HSP101 forms specific interactions with WT and P5 Lys Hyp1-Nluc-mDHFR-3xFLAG. P5 Lys Hyp1-Nluc-mDHFR-3xFLAG’s exclusive ER-localisation would allow us to assess its interaction with internal HSP101 only. In addition, the fact that the PEXEL motif of P5 Lys Hyp1-Nluc-mDHFR-3xFLAG is intact (Figs 5D and S4), would enable us to determine if PEXEL processing is crucial for the interaction with ER-localised HSP101. For this experiment, HSP101-HAglmS parasites expressing the Hyp1 reporters were cultured in the presence of WR99210 to induce structural stabilisation of the murine DHFR domain present in the reporter cargo [28]. This strategy has been shown to stall the unfolding of reporter cargo by PTEX, enabling isolation of an otherwise transient cargo-PTEX interaction by co-immunoprecipitation (Fig 5A, panel 2) [42]. Parasites were then enriched by magnetic purification, lysed in 1% Triton X-100, and incubated with either anti-Nluc IgG, followed by antibody capture with protein A agarose, or anti-HA agarose beads (Fig 6A). Anti-Nluc immunoprecipitations of WT and P5 Lys Hyp1-Nluc-mDHFR-3xFLAG revealed that HSP101 interacted with WT Hyp1-Nluc-mDHFR-3xFLAG but not with the P5 Lys Hyp1-Nluc-mDHFR-3xFLAG, likely due to stabilised interactions between WT Hyp1-Nluc-mDHFR-3xFLAG and HSP101 in the PV where unfolding occurs (Fig 6B, lanes 13 and 16). To preserve weaker binding interactions, parasites were treated with 0.5 mM thiol-cleavable cross-linker dithiobis(succinimidyl propionate) (DSP) prior to lysis and immunoprecipitation. Under these conditions we found that HSP101 readily co-eluted with WT and P5 Lys Hyp1-Nluc-mDHFR-3xFLAG from parasites lysed in Triton X-100 (Fig 6B lanes 14 and 17) and the more stringent RIPA buffer (Fig 6B lanes 15 and 18). This data suggests that HSP101 can bind proteins destined for export in the ER prior to secretion into the PV.

Given most of the immunoprecipitated P5 Lys Hyp1-Nluc-mDHFR-3xFLAG was not cleaved within the PEXEL motif (Figs 5D and S4), our data also suggests that correct PEXEL processing may not be necessary for HSP101 binding. To rigorously exclude the possibility that small amounts of PEXEL-cleaved P5 Lys Hyp1-Nluc-mDHFR-3xFLAG (Figs 5D and S4) was the species that co-immunoprecipitated HSP101, we also performed reciprocal anti-HA immunoprecipitation to capture the HA-tagged HSP101 (Fig 6C). Western blot analysis of the eluates indicated that HSP101-HA could co-immunoprecipitate both PEXEL-cleaved WT Hyp1-Nluc-mDHFR-3xFLAG (Fig 6C lanes 11 and 13) and the miscleaved P5 Hyp1-Nluc-mDHFR-3xFLAG (Fig 6C lanes 12 and 14). Furthermore, HSP101 did not co-immunoprecipitate SERA5, a protein secreted into the PV, validating the specificity of the interaction (Fig 6C).

To explore if other PTEX components could bind ER-trapped cargo, PTEX150 was immunoprecipitated and could only co-purify the WT Hyp1-Nluc-mDHFR-3xFLAG reporter and not the P5 Lys Hyp1-Nluc-mDHFR-3xFLAG, suggesting that only ER-resident HSP101 can form an interaction with the ER-trapped mutant cargo (S6 Fig).

To further demonstrate that HSP101-cargo interaction occurred in the parasite, HSP101-HAglmS parasites expressing WT and P5 Lys Hyp1-Nluc-mDHFR-3xFLAG reporters were first permeabilised by saponin and treated with 100 μg/mL proteinase K to degrade the contents of the host cell and the PV, including PTEX components and associated HSP101 (Fig 6D). As was observed previously, HSP101 was still present in all treated samples, while the PV-located EXP2 was almost fully degraded by proteinase K (Fig 6E, lanes 1–3 vs lanes 4–6). Following digestion, the resulting pellets were similarly treated with 0.5 mM DSP and solubilised with 1% Triton X-100 lysis buffer prior to anti-Nluc co-immunoprecipitation assay (Fig 6E). Analysis of the eluted fraction revealed that both WT and P5 Lys Hyp1-Nluc-mDHFR-3xFLAG co-immunoprecipitated HSP101 (Fig 6E, lanes 8 and 9). We observed less HSP101 co-eluted with WT Hyp1-Nluc-mDHFR-3xFLAG relative to the P5 Lys Hyp1-Nluc-mDHFR-3xFLAG probably because much of the WT Hyp1-Nluc-mDHFR-3xFLAG had already trafficked to the PV and/or into the host cell and would have been susceptible to proteinase K digestion after saponin treatment. Consequently, ER-localised P5 Lys Hyp1-Nluc-mDHFR-3xFLAG, which is protected from protease, co-purified greater amounts of HSP101.

Collectively, our immunoprecipitation data indicates that HSP101’s interactions with cargo happens intracellularly as well as in the PV.

The mechanistic basis of ER trapping of the P5 Lys Hyp1-Nluc-mDHFR-3xFLAG is not clear. Our western blotting and mass spectrometry analysis suggests that P5 Lys Hyp1 forms two primary products, the predominant miscleaved species with intact PEXEL and the correctly cleaved PEXEL product (Fig 5D). One plausible model is that plasmepsin V may still be able to bind to the P5 Lys Hyp1-Nluc-mDHFR-3xFLAG but was unable to properly proteolyse the mutant PEXEL. This would potentially clog the Sec61-plasmepsin V import complex [26] leading to the aberrant processing of the P5 Lys Hyp1-mDHFR-3xFLAG as an attempt to free the complex. To test the model, we initially performed an anti-FLAG co-immunoprecipitation using HSP101-HA parasites expressing WT and P5 Lys Hyp1-Nluc-mDHFR-3xFLAG in the presence of 0.5 mM DSP. We found that the P5 Lys Hyp1, but not the WT Hyp1-Nluc-mDHFR-3xFLAG reporter, co-precipitated plasmepsin V, suggesting that P5 Lys Hyp1 cargo was indeed able to bind to plasmepsin V. Interestingly, HSP101 was also present in the P5 Lys Hyp1 eluate (Fig 7A), indicating that the ER trapping of P5 Lys Hyp1-Nluc-mDHFR-3xFLAG might stabilise HSP101-cargo-plasmepsin V interaction in the ER. This finding is of interest in light of previous work by Russo et al., who proposed a model of direct cargo handover from plasmepsin V onto HSP101 post cleavage [39, 40]. We therefore sought to perform a reciprocal co-immunoprecipitation of HSP101-HA in the presence of P5 Lys Hyp1-Nluc-mDHFR-3xFLAG. This revealed that while HSP101-HA co-purified the P5 Lys Hyp1-Nluc-mDHFR-3xFLAG reporter as previously demonstrated (Fig 6C), plasmepsin V was conspicuously absent (Fig 7B). This data suggests that HSP101 and plasmepsin V may separately bind to proteins destined for export in the ER and therefore, HSP101 may not receive proteins directly from plasmepsin V after PEXEL cleavage.

The expression of Hyp1-Nluc-mDHFR-3XFLAG reporters was driven by a bidirectional Plasmodium berghei EF1a promoter that also controlled expression of the blasticidin deaminase drug resistance cassette. The plasmid pEF-Hyp1-Nluc-mDHFR-3xFLAG was derived from plasmid pEF-Hyp1-Nluc-DH-APEX [42]. The Hyp1 component of this plasmid contained the first 113 aa of Hyp1 (PF3D7_0113300), including the RLLTE PEXEL motif [52] which when cleaved leaves 52 aa of Hyp1 remaining. A synthetic murine dihydrofolate reductase (mDHFR) 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 mDHFR-APEX gene cassette. This generated the final pEF-Hyp1-Nluc-mDHFR-3xFLAG plasmid.

Generation of the P5 lysine (lys/K) mutation of the Hyp1-Nluc-mDHFR-3xFLAG was performed as follows: Hyp1 region was first amplified as two overlapping PCR fragments. First fragment amplified with Hyp1K_1 (5’-TGCTTATAAATAAATAAAAATTTTATAAAACTCGAGCAAAATGA-3’) and Hyp1K_2 (5’-TGCTAGCAGGTTATTAACAAAATATAAAGACACATTACA-3’) and the second fragment amplified with Hyp1K_3 (5’-TGTAATGTGTCTTTATATTTTGTTAATAACCTGCTAGCA-3’) and Hyp1K_4 (5’-TCGAGTGTGAAGACCATGGTATCAACAACA-3’).

The overlap region between these two PCR products contained the lysine mutation indicated above. PCR fragments were then sewn together with primer pair Hyp1K_1 and Hyp1K_4 via overlapping PCR, ligated into the pJET1.2/blunt plasmid (ThermoFisher Scientific), and electroporated into E. coli XL-10 gold cells (Stratagene). After sequencing, the mutant P5 Lys Hyp1 fragments were released from pJET1.2/blunt using XhoI and NcoI and ligated into pEF-Hyp1-Nluc-mDHFR-3xFLAG to replace the wildtype Hyp1 fragment. This resulted in the final pEF-P5Lys Hyp1-Nluc-mDHFR-3xFLAG plasmid.

To generate plasmid pHSP101-HAglmS for creating a P. falciparum HSP101 conditional knockdown parasite line, the parent construct pPTEX150-HAglmS [36] was digested with BglII and PstI to remove the Pfptex150 targeting sequence and in its place, 920 bp of the C-terminus of Pfhsp101 sequence minus the stop codon was cloned. This Pfhsp101 targeting sequence was amplified from P. falciparum 3D7 gDNA using oligonucleotides DO610 (5’-GCGAGATCTCTAAATCCATTATTGGAAATGAAGAT-3’) and DO611 (5’-GCGCTGCAGCGGTCTTAGATAAGTTTATAACCAAG-3’). PCR genotyping was performed by amplifying parasite genomic region with a primer set DO561 (5’-TAAGTACATTAGAAAAGGATGTAGACA-3’) / DO562 (5’-GGTCTTAGATAAGTTT-ATAACCAAGTT-3’), DO561 / DO233 (5’-tccgcatgcGGTCTTAGATAAGT-TTATAACCAAGTT-3’), and DO561 / DO276 (5’-GTGATTTCTCTTTGTTCAAGGA-3’) as indicated in Fig 1A and 1B.

Asexual blood stage Plasmodium falciparum was cultured as per [67]. Cultures were maintained routinely in complete media comprising of RPMI-1640 base media supplemented with 2.5 mM HEPES, 367 μM Hypoxanthine, 31.25 μg/mL Gentamicin, 25mM NaHCO₃, and 0.5% Albumax II (Invitrogen). To generate HSP101-HAglmS parasites, 100 μg of pHSP101-HAglmS plasmid was electroporated into erythrocytes which were then mixed with 3D7 trophozoites and selected with 2.5 nM WR99210 [68, 69]. Parasites were cycled on and off WR99210 until a 100 kDa HA positive band appeared by western blot using anti-HA antibodies, indicative of integration into the hsp101 locus. Clonal HSP101-HAglmS positive parasites were selected by limiting dilution and correct integration verified by PCR.

To induce episomal expression of Hyp1-Nluc-mDHFR-3xFLAG and P5 Lys Hyp1-Nluc-mDHFR-3xFLAG in HSP101-HAglmS parasite line, 100 μg of pEF-Hyp1-Nluc-mDHFR-3xFLAG and pEF-P5 Lys Hyp1-Nluc-mDHFR-3xFLAG was used to transfect the HSP101-HAglmS parasite line. Transfections were performed as above except that 2.5 μg/mL blasticidin S was used for selection. Cargo trapping was performed essentially as described before [42] with young ring stage parasites treated with 10 nM WR99120 for 16–20 hours before being harvested for immunoprecipitation or IFA (described below).

Saponin lysed parasite pellets were solubilized in 1% digitonin (ThermoFisher) for 30 min at 4°C. Following incubation, the soluble component was separated from the insoluble pellet by centrifugation at 14,000 x g for 30 min. 5% Coomassie G-250 and 100% glycerol was spiked into the supernatant to reach final concentrations of 0.25% and 10% respectively. Samples were separated on a 3–12% NativePAGE Bis-Tris gel (ThermoFisher). Gels were run as per manufacturer prior to blotting onto a 0.2 μm polyvinylidene fluoride (PVDF) membrane using iBlot system (Invitrogen). To enable the detection of HSP101, the gel was incubated for 15 min at room temperature in 50 mM DTT and 0.1% SDS in the tris-glycine buffer prior to transfer. The membranes were subsequently fixed with 8% (v/v) acetic acid in water, air-dried, and incubated with 100% methanol to remove the excess Coomassie G-250 dye. The membrane was blocked in 1% casein in 1x PBS for 1 hour at room temperature (RT) prior to probing with primary antibody solutions (diluted in 1% casein in PBS) as following: rabbit anti-HSP101 (IgG purified; 30 μg/mL), rabbit anti-PTEX150 (r741, 1:500), mouse anti-EXP2 (10 μg/mL).

IFA was performed essentially according to [70]. Where iRBCs 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 minutes at RT. Alternatively, thin blood smears of parasites were fixed in 90% acetone/10% methanol for 5 mins where specified. Coverslips or slides were blocked with 3% BSA/0.02% Triton X-100/1x PBS and probed overnight at 4°C, with rabbit anti-Nluc (12.5 μg/mL), mouse anti-EXP2 (10 μg/mL), rabbit anti-ERC (1:1000), mouse anti-FLAG M2 (Sigma, 10 μg/mL), mouse anti-HA (Sigma clone HA-7; 1:500), rabbit anti-SBP1 (1:500) [36] and rabbit anti-STEVOR (PF3D7_1254100, 1:500) [71]. After washing goat anti-rabbit Alexa Fluor 594 and Goat anti-mouse Alexa Fluor 488 (1:2000) secondary antibodies were applied for one hour at RT. 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.

Analysis was performed essentially according to the “in vitro cyst DBA fluorescence quantification” protocol [72]. Images were first imported to the open java source ImageJ/FIJI for analysis and “set measurement” function was set to record area, integrated intensity, and mean grey values. Firstly, a region of interest (ROI) was drawn around the parasite cell (including the PV) using the freehand selection tool. DIC channel and the anti-HA/anti-EXP2 staining was used concomitantly as a guide to define the outermost part of the PV. Measurement data for total parasite fluorescence intensity (TPFI) were then recorded. An identical procedure was carried out to measure a ROI inside the parasite cell excluding the PV (Internal fluorescence, IF). The whole procedure was repeated 3 times for each cell to minimize errors. The background intensity was measured inside 3 uninfected erythrocytes within each captured field and was averaged to obtain mean background value (MBV). Total parasite and internal fluorescence intensity were corrected by subtracting the MBV from the averaged integrated density of TPFI or IF with the value of MBV multiplied by the mean area of the measured cell. The estimated PV fluorescence intensity for each cell was calculated by subtracting the cell’s corrected TPFI and IF integrated density values. Lastly, the calculated intensity for PV was divided by the IF value to calculate the amount of internal fluorescence intensity relative to the PV. The internal/PV ratios of >10 cells were measured for each category and plotted on GraphPad Prism 8.2 software to calculate the statistical significance by ordinary one-way ANOVA test.

3D-structured illumination microscopy (3D-SIM) was performed using a Nikon Ti-E inverted microscope (Nikon) equipped with a motorised piezo stage, Nikon Intensilight E, SIM illuminator, SIM microscope enclosure, and Perfect Focus System. Cells were prepared as per IFA protocol but then imaged using a Plan Apo VC 100x 1.4NA oil objective and using 2D/3D-SIM imaging modes with diffraction grating 3D 1 Layer for 2D/3D SIM. A Z-stack for each cell was taken for co-localisation analysis using the open-source java ImageJ’s JaCoP plugin.

The cleavage assay was performed essentially as described in Hodder et al [55]. 2 nM of P. vivax plasmepsin V in buffer (25 mM Tris-HCl and 25 mM MES, pH 6.4) was incubated with 5 μM FRET peptide substrates representing WT and mutant KAHRP (DABCYL-RNKRTLAQKQ-E-EDANS and DABCYL-RNKATAAQKQ-E-EDANS) and Hyp1 (DABCYL-G-KIRLLTEYKD-E-EDANS and DABCYL-G-KIRLLTKYKD-E-EDANS) sequences in a total volume of 20 μL. Samples were incubated at 20°C for 20 hour and were measured with an Envision plate (PerkinElmer) reader (ex. 340 nm; em. 490 nm). Raw fluorescence values were subtracted from fluorescence background values and then data for each substrate was normalized to 100% activity. Biochemical P. vivax plasmepsin V inhibitory assays (20 μL total volume) consisted of 2 nM of P. vivax plasmepsin V in buffer (25 mM Tris-HCl and 25 mM MES, pH 6.4) with 5 μM FRET fluorogenic peptides. Inhibitory values were determined with a nonlinear regression four-parameter fit analysis in which the parameters were not constrained using Domatics software (version 5.3.1612.8630).

Protein samples were electrophoresed on a NuPAGE 4–12% Bis-Tris gel (Invitrogen) and transferred onto a nitrocellulose membrane using iBlot system (Invitrogen). The membrane was subsequently blocked using 1% casein in 1x PBS and incubated with primary antibody solutions with following dilutions: mouse anti-HA (Sigma, 1:500), rabbit anti-HSP101 (r950, 1:500), rabbit anti-PTEX150 (r741, 1:500), mouse anti-EXP2 (5–10 μg/mL), rabbit anti-EXP2 (r1167, 1:1000), chicken anti-FLAG (Abcam,1:2000), mouse anti-FLAG (10 μg/mL), rabbit anti-Nluc (12.5 μg/mL), rabbit anti-plasmepsin V (1:2000), rabbit anti-GAPDH (1:2000), rabbit anti-SERA5 (1:1000) for 1 hour at RT or overnight at 4°C. Secondary antibodies were incubated for 1–2 hours in following dilutions: Goat anti-rabbit Alexa Fluor Plus 700 and 800, Goat anti-mouse Alexa Fluor Plus 700 and 800 (Invitrogen 1:10,000), and horseradish peroxidase-conjugated Goat anti-chicken IgY (Abcam, 1:10,000). For enhanced chemiluminescence detection, membranes were probed with SuperSignal West Pico PLUS substrate (Pierce). All imaging and densitometry analysis were performed using Odyssey Fc system (LI-COR).

Protease protection assays were completed as outlined in [33]. Briefly, magnet purified trophozoites (24–30 hours post infection) were pelleted (500 g/ 5 min) and washed twice with digestion buffer 50 mM Tris, 150 mM NaCl, 1 mM CaCl₂, pH 7.4. The resultant pellet was split into 12 equal fractions. Four tubes were subsequently treated with recombinant equinatoxin made in house and empirically determined to lyse ~100% of erythrocytes. Another four tubes were treated with equinatoxin and 0.03% saponin; and the remaining four tubes were treated with equinatoxin and 0.25% Triton X-100. Parasites were incubated at room temperature for 10 min/ shaking at 1000 rpm. Proteinase K was then added to selected tubes to final concentrations of 0, 1, 5 and 20 μg/mL and digested for 15 min at 37°C. The reaction was stopped by addition of 200 μL of protease inhibitor cocktail (Roche) made up at a high concentration of 2 tablets per 3 mL digestion buffer with 1 mM PMSF. Parasite material was centrifuged (1000 x g, 5 min) and pellet material was separated from supernatant. Pellet and supernatant materials were resuspended in sample buffer and electrophoresed via NuPAGE Bis Tris SDS-PAGE 4–12% prior to analysis by western blotting.

For HA co-immunoprecipitations, trophozoite iRBCs were treated with 0.09% saponin to remove haemoglobin and solubilised in 20x cell pellet volume of lysis buffer (1% Triton X-100, 0.1% SDS, 150 mM NaCl, 10 mM Tris-HCl pH 7.4) supplemented with protease inhibitor cocktail (Roche) and subjected to 2 freeze and thaw cycles. The lysate was clarified by centrifugation (16,000 x g, 10 minutes, 4°C), and total protein concentration was measured by Bradford assay. Lysates were adjusted to 1–2 mg total protein in 1 mL and incubated overnight at 4°C with 25 μL of packed Monoclonal Anti-HA Agarose (Sigma-Aldrich). Following incubation, agarose beads were washed 5x with 1 mL the lysis buffer and eluted with 50 μL 2x protein sample buffer (100 mM Tris-HCl pH 6.8, 4 mM EDTA, 4% SDS, 0.01% bromophenol blue, 20% (v/v) glycerol).

For Hyp1-Nluc-mDHFR-3xFLAG co-immunoprecipitations, trophozoite iRBCs were magnetically purified and solubilised as above. Lysates were adjusted to 0.5–1 mg total protein in 1 mL and incubated overnight at 4°C with 10 μg IgG-purified anti-Nluc antibody. Following incubation, 25 μL packed Recombinant Protein A-Sepharose 4B (Invitrogen) was used to capture the immune complexes for 1 hour at RT. Beads were subsequently washed 5x with 1 mL lysis buffer and eluted with 50 μL protein sample buffer.

FLAG immunoprecipitations, iRBCs were magnetically captured and were solubilised in 20x pellet volume of 0.5x RIPA buffer (0.5% Triton X-100, 0.5% sodium deoxycholate, 0.05% SDS, 25 mM Tris pH 7.4, 150 mM NaCl). The rest of the immunoprecipitations was performed as above except that anti-FLAG M2 agarose beads (Sigma-Aldrich) were used to capture the reporter protein and all washes were performed in RIPA buffer.

For co-immumoprecipitations with PTEX antibodies, saponin lysed late ring-stage parasites that had been cross-linked with 1 mM DSP were solubilised with 0.5% Triton X-100 in 1x PBS or 1% Triton X-100, 0.1% SDS, 150 mM NaCl, 10 mM Tris-HCl pH 7.4 supplemented with protease inhibitor cocktail (Roche) at 4°C. Clarified lysates were incubated overnight at 4°C with 10 μg rabbit anti-HSP101 IgG or 20 μL of rabbit polyclonal anti-PTEX150 serum [29]. Following overnight incubation, 50 μL packed protein A-Sepharose 4B (Invitrogen) was added to the immune complexes and samples were incubated for an additional 1 hour at RT. Beads were subsequently washed 5x with 1 mL 0.5% Triton X-100 in 1x PBS and eluted with 60 μL of protein sample buffer.

Coomassie stained protein bands were excised and destained with 50% acetonitrile (ACN) in 100 mM ammonium bicarbonate (ABC) pH 8.5. The proteins were reduced and alkylated after treatment with 10 mM DTT (Astral Scientific) and 20 mM chloroacetamide (Sigma). The gel was dehydrated using 100% ACN and rehydrated with digestion solution, containing 100 ng/μL trypsin (Promega) or 100 ng/μL GluC (Promega) in 100 mM ABC. After overnight digestion at 37°C, the tryptic peptides were extracted from the gel and subject LC–MS/MS analysis.

Enzyme digests were analysed by LC-MS/MS using the QExactivePlus mass spectrometer (Thermo Scientific, Bremen, Germany) coupled online with an Ultimate 3000 RSLCnano system (Thermo Scientific, Bremen, Germany). Samples were concentrated on an Acclaim PepMap 100 (100 μm × 2 cm, nanoViper, C18, 5 μm, 100 Å; Thermo Scientific) trap column and separated on an Acclaim PepMap RSLC (75 μm × 50 cm, nanoViper, C18, 2 μm, 100 Å; Thermo Scientific) analytical column by increasing concentrations of 80% acetonitrile/0.1% formic acid at a flow of 250 nL/min for 90 min.

The mass spectrometer was operated in the data-dependent acquisition mode to automatically switch between full scan MS and MS/MS acquisition. Each survey full scan (m/z 375–1575) was acquired in the Orbitrap with 60,000 resolutions (at m/z 200) after accumulation of ions to a 3 × 106 target value with maximum injection time of 54 ms. Dynamic exclusion was set to 15 s to minimize repeated selection of precursor ions for fragmentation. The 12 most intense multiply charged ions (z ≥ 2) were sequentially isolated and fragmented in the collision cell by higher-energy collisional dissociation (HCD) with a fixed injection time of 54 ms, 30,000 resolution and automatic gain control (AGC) target of 2 × 105. The raw files were analysed Proteome Discoverer v2.5 (Thermo Scientific) and searched against a custom database of the recombinant sequence appended to the Plasmodium falciparum UniProtKB using the Byonic v3.0.0 (ProteinMetrics) search engine to obtain sequence information. Only proteins falling within a predefined false discovery rate (FDR) of 1% based on a decoy database were considered further.