2.6 The sequence requirements of the 13aa spacer for export are relatively unconstrained.
Since the 13aa Hyp1 spacer still permitted a reasonable amount of export compared to the full size 51aa spacer we decided to use this as basis for subsequent mutagenesis experiments due 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 minimal spacer may contain some cryptic trafficking information, we decided to remove it from the original 51aa spacer and fuse this new ∆NT13aa spacer to the Nluc-mDH-FL reporter and express this in parasites (Figs 5A). Western blots of the Hyp1 ∆NT13aa-Nluc-mDH-FL parasites indicated the reporter protein was of a size consistent with correct PMV cleavage (Fig S1). Microscopy of the ∆NT13aa reporter showed it was efficiently exported and quantification of the fluorescence intensity implies a higher degree of export of the ∆NT13aa reporter than the 13aa spacer (Fig 5B and C). This result indicated that the ∆NT13aa spacer was competent for export and did not rely on targeting information contained within the 13aa spacer for export essential for export of the Hyp1-NlucmDH-FL reporter.
Next, we investigated whether certain amino acids within the minimal spacer were necessary for export. As proteins chaperones tend to recognise misfolded proteins via their hydrophobic residues47, the four hydrophobic amino acids Y, L, I and V in the minimal Hyp1 spacer TEY KDTL QI KV EQ 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 that mutation of the hydrophobic amino acids to the polar amino acid serine, produced lower scores than for other amino acids (Figs 5A and S2). We therefore replaced the hydrophobic amino acids in the 13aa spacer with serines and transfected this Hyp1 13aa.Ser repoter into parasites where, by western blot, the Hyp1 13aa.Ser reporter was of a size consistent with correct PMV cleavage (Fig S1). Microscopic analysis of the Hyp1 13aa.Ser parasite-infected RBCs indicated that this mutant was exported less efficiently than the WT minimal 13aa spacer (Figs 5A-C and S1).
Next, the hydrophobic amino acids of the minimal 13aa spacer were mutated to glutamic acids (Hyp1 13aa.Glu) as these mutations were scored most poorly by the BIP binding predictor (Fig S2). Western blots of Hyp1 13aa.Ser-Nluc-mDH-FL parasites indicated PMV processing was correct and microscopic analysis of parasite-infected RBCs expressing the reporter indicated the Hyp1 13aa.Glu protein was exported significantly less well than the Hyp1 13aa parasites (Fig 5B and 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 43,48. We 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 Fig 5D to 5C). However, the signal for the Nluc bioluminescence export assay was however often 10-20% greater than the microscopy signal probably because bioluminescence is highly sensitive can detect lower protein levels dispersed in the large RBC compartment (Fig 5D). Additionally, microscopy does not measure RBC signal in front or behind the parasite which may underestimate 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 shrunk, 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 to 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 erythrocyte 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 1,2,14. Our data partly support this in that mutation of P2’ residue to K can lead to increased trapping of some PEXEL proteins in the PV, most notably for the KAHRP and STEVOR reporters. The P2’ K mutation can also greatly reduce the efficiency of PMV cleavage in the case of Hyp1 but not so for KAHRP and STEVOR 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 oomycete effector motif RxLR 30,38, which lack the P2’ residue, were not cleaved by plasmepsin V and failed to promote export into the host cell 30,38. Of all possible mutations of P2’, only charge reversal P2’ mutations, either to arginine or lysine, strongly inhibit plasmepsin V cleavage 36 whilst a single alanine mutation was observed to cause a variable export phenotype4,36. This was evident from the small reduction of cleavage observed in the in vitro cleavage assay with P2’ A mutation compared to the K mutation. Despite this, the strong inhibition in vitro cleavage of STEVOR P2’ K did not translate to what was observed in parasites as the reporter protein was efficiently cleaved. It seems likely that suboptimal amino acid variants at PEXEL P2’ are still permissive for PMV cleaving in the cellular context, presumably because of the fast kinetics of the proteolytic reactionin vivo or due to concentrated substrate in the ER that favours cleavage even under less cleavable substrate. It is also possible thatPf PMV in parasites is more active against the P2’ mutants than the Pv PMV used in the peptide cleavage assay49,50.
Having previously observed that the ER-trapped Hyp1 P2’ K mutant, which was mis-cleaved by an unknown protease upstream of the PEXEL, became quite insoluble and remained in the ER42, we explored how the solubility of cleaved PEXEL reporters can influence their export. In contrast, to the mis-cleaved Hyp1 P2’ K reporter, the solubility of the mis-cleaved KAHRP and STEVOR P2’ K reporter proteins did not appear to decrease, and it is unclear whether the mis-cleaved species accumulated in the ER or were trafficked to the PV like the correctly cleaved P2’ K species. We have shown previously that the ER-trapped Hyp1 P2’ K reporter can still bind to ER-resident HSP101 despite its insolubility 42. The fact that the KAHRP and STEVOR P2’ K mutants were efficiently cleaved and trafficked to the PV but were not strongly exported suggests the P2’ residue might not be important for the initial binding to HSP101 but rather for commitment to stronger downstream interactions with the whole of PTEX required for cargo unfolding and export.
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 at least 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 4,24. Despite 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 has 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 an increase the length of the unstructured N-terminal polypeptide increases the likelihood of the cargo will stably binding to HSP101. Clp/HSP100 chaperones generally require a recognition signal of at least 10-20 broadly diverse amino acids to initiate the polypeptide unfolding and translocation51,52. Our data is consistent with this model and suggest that HSP101 also requires a large region within the N-terminal portion of the cargo protein to initiate cargo translocation. This could explain why single P2’ K point mutation in the mature PEXEL motif is not enough to prevent the interaction altogether and why the N-termini of PNEPs, despite the latter lacking a mature PEXEL motif, can still act as an export signal 38. Cargo 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 cavity53. It 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 unfolding54,55. Clp/HSP100 chaperones, particularly ClpB and HSP104, are thought to have a similar pre-unfolding step52,53,56. We 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 (Fig 6A). Consistently, 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 19. As the HSP101 subunits undergo allosteric changes powered by ATP hydrolysis, the tyrosines move up and down to help grip and pull on the cargo (Fig 6B). Since it is possible that the tyrosine residues may 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 (E) residues 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 minimal 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’s central cavity, down into NBD2 where it could engage the cargo binding tyrosine residues (Fig 6B).
In conclusion, our data suggests dual functions for the P2’ 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 of P2’ to K greatly reduces PMV binding and/or successful proteolytic activity but in other PEXEL proteins such as STEVOR and KAHRP P2’ is not as critical for PMV cleavage. This indicates that other amino acids within and bordering the PEXEL probably have a bearing on how dependent PMV activity is on the P2’ residue. 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 P2’ K mutant appears to permit trafficking to the PV as STEVOR and KAHRP more efficiently reach the PV than poorly cleaved Hyp1. Once in the PV, however, the STEVOR and KAHRP P2’ K reporters were not efficiently exported into the RBC compartment. 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 PV-resident protein have full access to PTEX 57. It is possible therefore that the charge reversal P2’ mutant somehow binds to PTEX less efficiently leading to less engagement, unfolding and eventual export into the RBC compartment.
Materials and Methods
4.1. Culture of P. falciparum transfectants
Asexual blood-stage Plasmodium falciparum (3D7 or CS2 background) was cultured according to the established protocol 58. Cultures were routinely maintained in complete RPMI media containing RPMI-1640 base media supplemented with 2.5 mM HEPES, 367 µM hypoxanthine, 31.25 µg/mL Gentamicin, 25 mM NaHCO3, 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 CaCl2, 2 mM EGTA, 5 mM MgCl2, 10 mM K2HPO4/KH2PO4pH 7.6) and mixed with packed RBCs as per 59. After electroporation using Gene Pulser XCell System (BioRad), the uninfected RBCs were then mixed with 20 µL of HSP101-HAglmStrophozoite-stage parasites 42 and allowed to invade the transfected RBCs for 2 cell cycles before starting a selection with 2.5 µg/mL blasticidin S.