To identify candidate proteases responsible for PEXEL processing we used features of the cleavage site to predict the type of enzyme involved. We noted that HIV-1 aspartic acid protease cleaved substrate proteins within a motif similar to PEXEL 19,20
, suggesting the Plasmodium
enzyme shares similarity. To test this we incubated inhibitors of HIV-1 protease in culture with P. falciparum
-infected erythrocytes and monitored PEXEL cleavage. PEXEL processing was reduced in the presence of lopinavir, nelfinavir, ritonavir and saquinavir for PfEMP3- and knob associated histidine-rich (KAHRP)-GFP proteins (). Whilst the inhibition was inefficient it suggested PEXEL processing might involve a protease similar to HIV-1 aspartic acid protease. Sensitivity of signal peptidase was assessed using the inhibitors with acyl carrier protein signal sequence (ACP)-GFP 21
. This transgenic protein lacks a PEXEL and is secreted into the parasitophorous vacuole 21
. In contrast to the PEXEL-containing proteins, we observed no detectable reduction in processing of the signal sequence from the ACPs-GFP chimera with inhibitors (). We conclude that the PEXEL protease is not the parasite signal peptidase but involves a unique protease belonging to the aspartic acid family.
PEXEL processing is sensitive to HIV protease inhibitors
A list of candidate aspartic acid proteases was generated from the P. falciparum
genome based on; firstly, expression in blood and liver stages of the parasite 22,23
, secondly, presence of a signal sequence for ER entry; and thirdly, specificity to Plasmodia spp.
because the PEXEL is absent from closely related apicomplexans 14
. Plasmepsin V (PF13_0133), a putative aspartic acid protease localising to the ER that is functionally divergent to other members of the Plasmepsin family, satisfied our criteria 24
. To investigate the function of Plasmepsin V, we introduced a hemagglutinin (HA) tag at the C-terminus (3D7-PMVHA) (). HA-tagged Plasmepsin V localised to the ER 25
, as has been shown previously for the endogenous protein 24
To determine if Plasmepsin V could cleave PEXEL substrates we immuno-precipitated the HA-tagged protease (ipPMVHA) and incubated KAHRP peptides containing either a wild-type PEXEL or mutations that interfere with cleavage () 17
. The resulting peptides were separated by reversed-phase high-pressure liquid chromatography (RP-HPLC) and analysed by nano liquid chromatography tandem mass spectrometry (LC-MS/MS) (). Incubation of the KAHRP peptide with ipPMVHA produced two major peptide species (; first panel) derived from cleavage within the PEXEL after leucine (). In contrast, mutation of the conserved amino acids R>A, L>A or RLQ>A resulted in no cleavage (), consistent with the properties of the PEXEL cleaving protease in vivo
. Furthermore, mutation of the conserved fifth amino acid (Q>A) did not inhibit cleavage (), corroborating our previous data that this residue is not required for PEXEL processing 17
. To verify the PEXEL cleavage was due to Plasmepsin V, and not other co-precipitated proteases, we constructed P. falciparum
transgenic parasites that either conditionally, or constitutively, expressed an inactive form of Plasmepsin V tagged with HA (PMVmutHA and PMVmutHA2, respectively). We confirmed expression and ER localisation of PMVmutHA () and the immuno-precipitated proteins (ipPMVmutHA or ipPMVmutHA2) were incubated with KAHRP peptides and no processing occurred ( and Supplementary Fig. 2
). Transgenic parasites were generated that expressed HA-tagged Plasmepsin IX (PF14_0281) (3D7-PMIXHA), another member of the aspartic acid protease family in P. falciparum.
Immuno-precipitated Plasmepsin IX (ipPMIXHA) could not process the KAHRP peptide (). ipPMVHA also cleaved a recombinant GBP130 protein to the expected size () as well as a quenched fluorescent substrate of 9 amino acids (), both of which contained a PEXEL. Mutation of conserved amino acids in the PEXEL blocked cleavage of both substrates ( and ). The HIV protease inhibitor lopinavir reduced Plasmepsin V activity by 19.1% (P=0.0145; ) consistent with in vivo
inhibition of PEXEL cleavage of PfEMP3- and KAHRP-GFP (). Taken together, these results show that Plasmepsin V specifically cleaves the PEXEL and has the functional specificity expected of the PEXEL protease.
Activity of Plasmepsin V and cleavage of PEXEL
To provide further evidence that Plasmepsin V was responsible for PEXEL cleavage we expressed amino acids 37 to 521 in Escherichia coli
with an N-terminal Hexa-His fusion tag. The recombinant Plasmepsin V was extracted from inclusion bodies and purified on NiNTA under denaturing conditions and further fractionated using RP-HPLC (). Individual fractions were analysed using SDS-PAGE () or transferred to nitrocellulose and probed with anti-Plasmepsin V antibodies (). A major band was observed at approximately 55 kDa (fraction #3, 4 and 5) and smaller processed forms of Plasmepsin V were also detected (fractions 2–3) suggesting N-terminal processing () occurred either by the protease itself or other E. coli
proteases present in the purified inclusion bodies. Processed forms of ipPMVHA are also detectable in immunoblots suggesting one may be the activated form (Supplementary Fig. 3
). Recombinant protein fractions was assayed for PEXEL cleavage using KAHRP peptides and PEXEL-specific activity was observed with fractions 2 and 3, which contained the processed form of Plasmepsin V ( and Supplementary Fig. 4
). This data suggests that a recombinant, N-terminally truncated form of Plasmepsin V can cleave the PEXEL motif.