|Home | About | Journals | Submit | Contact Us | Français|
Small ubiquitin-related modifier (SUMO) is implicated in the regulation of numerous biological processes including transcription, protein localization, and cell cycle control. Protein modification by SUMO is found in Plasmodium falciparum; however, its role in the regulation of the parasite lifecycle is poorly understood. Here we describe functional studies of a SUMO-specific protease (SENP) of P. falciparum, PfSENP1 (PFL1635w). Expression of the catalytic domain of PfSENP1 and biochemical profiling using a positional scanning substrate library demonstrated that this protease has unique cleavage sequence preference relative to the human SENPs. In addition, we describe a novel class of small molecule inhibitors of this protease. The most potent lead compound inhibited both recombinant PfSENP1 activity and P. falciparum replication in infected human blood. These studies provide valuable new tools for the study of SUMOylation in P. falciparum.
Small ubiquitin-related modifier (SUMO) is a posttranslational modifier in eukaryotes that is structurally related to ubiquitin yet functionally distinct. SUMO modification has been implicated in the regulation of numerous biological processes including transcription (Desterro, et al., 1998), protein localization (Matunis, et al., 1996), and cell cycle control (Ulrich, 2009). Like other reversible posttranslational modifications (i.e. ubiquitination, acetylation, methylation and phosphorylation), SUMOylation is highly dynamic with diverse and complex functional implications. SUMOylated proteins were recently identified in Plasmodium falciparum, a unicellular eukaryotic parasite that causes human malaria (Issar, et al., 2008), suggesting that SUMOylation may play similarly important roles in parasite growth and survival.
P. falciparum is the causative agent of human malaria, infecting 250 million people per year and causing nearly 800,000 deaths (Aregawi, et al., 2008). The parasite has a complex and highly regulated lifecycle in both the human and mosquito hosts. Numerous transcriptional and proteomic profiling studies have demonstrated that patterns of gene and protein expression are highly regulated during the blood stage lifecycle (Bozdech, et al., 2003; Khan, et al., 2005; Le Roch, et al., 2003). P. falciparum lacks canonical eukaryotic transcription factors, but novel transcription factors and epi-genetic and post-transcriptional regulatory factors have recently been identified (Painter, et al., 2011). Beyond regulation at the transcriptional level, multiple unexpected protein isoforms in proteomic analyses suggest that posttranscriptional and posttranslational modifications may play important and unique roles in the regulation of parasite survival inside the host (Coulson, et al., 2004; Foth, et al., 2008; Shock, et al., 2007). Thus, SUMOylation may represent one of several key mechanisms that the parasite uses to control gene expression profiles. Proteomic analysis of SUMOylated proteins from both P. falciparum and Toxoplasma gondii, demonstrated that proteins from numerous essential biological pathways, including proteins that are unique to these parasites, are modified by SUMO (Braun, et al., 2009; Issar, et al., 2008). In P. falciparum, PfSir2, an epigenetic regulator of parasite antigenic variation, is modified by SUMO (Issar, et al., 2008). Although a preliminary list of SUMO modified proteins has been generated, the importance of SUMO modification and its regulatory machinery in P. falciparum remains unclear.
SUMOylation of target proteins is regulated by dedicated enzymatic machinery, including a family of SUMO-specific proteases (SENPs; for review see Yeh, 2009)). SENPs play two primary roles in SUMO regulation: they process SUMO precursors to reveal a C-terminal di-glycine prior to conjugation, and they cleave the isopeptide bond between the C-terminal glycine of SUMO and the lysine side chain of a target protein. Bioinformatic analysis has predicted two P. falciparum SENPs (Issar, et al., 2008; Ponder and Bogyo, 2007; Wu, et al., 2003). However, functional studies of the P. falciparum SENPs have posed many challenges. Traditional genetic disruption of P. falciparum SENPs will likely not be possible given the essential role of SENPs in yeast and mammals (Li and Hochstrasser, 2003; Yeh, 2009). Knockdown of gene expression by RNAi is not possible in P. falciparum, and, although two promising conditional gene knockout systems have been developed, the inherent leakiness of both systems limits their applications (Armstrong and Goldberg, 2007; Baum, et al., 2009; Meissner, et al., 2005). As an alternative to genetic disruption, small molecule inhibitors have been successfully used to characterize the function of several P. falciparum proteases (Arastu-Kapur, et al., 2008; Yeoh, et al., 2007). However, there are currently no small molecule inhibitors that target SENPs from any organism.
In order to understand the dynamics of SUMOylation in P. falciparum, we aimed to characterize the predicted SENPs and to identify small molecule inhibitors that perturb their function. In this study, we identified two putative SENPs of P. falciparum using bioinformatics analyses: PfSENP1 (PFL1635w) and PfSENP2 (MAL8P1.157). We were able to recombinantly express PfSENP1 and determine that it has robust SUMO processing activity and a unique cleavage site preference relative to the human SENPs. Using a SUMO processing assay, we screened a library of irreversible cysteine protease inhibitors and identified a single compound, JCP-666, that inhibits PfSUMO processing both by parasite lysates and by recombinant PfSENP1 and also blocked parasite replication in human blood. We subsequently synthesized a more potent analog, VEA-260, with improved stability. Together, our data suggest that PfSENP1 activity is essential for parasite growth, and that the novel inhibitor scaffold identified here may be useful for the development of inhibitors of PfSENP1 that do not inhibit SENPs of the human host.
A search of the P. falciparum genome using sequence homology identified two genes encoding putative SENPs: PFL1635w and MAL8P1.157 (Ponder and Bogyo, 2007; Wu, et al., 2003). Alignment of the catalytic domain of these proteins with the human and yeast SENPs suggested that the essential catalytic residues are conserved (Figure 1A). The catalytic domain of PFL1635w, which we refer to as PfSENP1, is more closely related to the human SENPs whereas MAL8P1.157, referred to as PfSENP2, is divergent and found within a branch that is only loosely related to the yeast Ulp1 and highly related to a similar sequence in other Plasmodium species (Figure 1B). Analysis of microarray data from the P. falciparum transcriptome indicates that both genes are expressed during the intraerythrocytic lifecycle with peak expression in the late trophozoite phase (approximately 25 hr post invasion), and lowest expression in the early ring stage, immediately after invasion (Bozdech, et al., 2003; Le Roch, et al., 2003).
Although the predicted SENP mRNAs appear to be transcribed in P. falciparum, it is unclear if the resulting proteins are expressed or catalytically active. One function of the previously characterized SENPs in higher eukaryotes is to prepare SUMO for conjugation by proteolytically removing the C-terminal amino acids from the precursor SUMO protein (SUMO-pro) to reveal a C-terminal di-glycine motif. By adding a His6-tag after the C-terminal extension of SUMO-pro, processing can be monitored by gel shift assay (Mikolajczyk, et al., 2007). Using this assay we found that P. falciparum lysates efficiently processed SUMO-pro (PfSUMO-pro; Figure 2A). Furthermore, this processing activity could be blocked by pre-incubation of parasite lysates with either N-ethylmaleimide (NEM) or human SUMO protein modified with a vinyl sulfone reactive group after the C-terminal di-glycine (hSUMO-VS). NEM blocks SENP activity by acting as a general alkylating agent that modifies the active site cysteine, while hSUMO-VS is an irreversible inhibitor of SENP proteases that contains the full length SUMO protein fused to a reactive vinyl sulfone group (Borodovsky, et al., 2001). Inhibition by these agents suggests that the PfSUMO-pro processing observed in parasite lysates is the result of active PfSENPs.
P. falciparum development in the red blood cell is highly regulated, with many genes and proteins varying in expression and activity over the course of replication (Bozdech, et al., 2003; Le Roch, et al., 2003). In order to determine when PfSENPs are active during the parasite blood stage lifecycle, we harvested synchronous cultures of P. falciparum every 8 hr for 48 hr starting at approximately 14 hr post invasion. Incubation of lysates with PfSUMO-pro resulted in a peak in protease activity at 30 hr post-invasion, corresponding to the late trophozoite stage. Processing activity was lowest during the ring stage (14 and 48 hr post invasion; Figure S1; Figure 2B), in agreement with the gene expression patterns of both of the predicted SENPs (Bozdech, et al., 2003; Le Roch, et al., 2003).
In order to assess the specificity of the SUMO processing activity in parasite lysates, we monitored processing of multiple related human SUMO sequences (hSUMO-1-pro, hSUMO-2-pro, or hSUMO-3-pro). Recombinantly expressed human SENP1 protease (hSENP1) processed both PfSUMO-pro and all three human SUMOs. P. falciparum lysates, on the other hand, processed PfSUMO-pro and hSUMO-1-pro but not hSUMO-2-pro or hSUMO-3-pro (Figure 2C). Thus, the parasite enzyme differs from the human enzyme in its specificity towards its substrate. Since recent studies have demonstrated that processing of SUMO-pro by SENPs is sensitive to the sequence of the C-terminal amino acid extension after the di-glycine (Mikolajczyk, et al., 2007), we wanted to determine if specific sequence differences in the C-terminal extensions of the SUMOs could explain the substrate specificity observed in parasite lysates. We generated chimeric SUMOs by swapping the C-terminal amino acid tails to create PfSUMO-h2pro, PfSUMO-h3pro, and hSUMO2-Pfpro. Whereas P. falciparum lysates failed to cleave chimeric PfSUMO-h2pro and -h3pro, chimeric hSUMO2-Pfpro was cleaved by parasite lysates (Figure 2D). Therefore, differences in C-terminal amino acid sequences rather than differences in the core SUMO sequence impact substrate recognition by parasite SENPs.
Although these results confirmed SENP activity in P. falciparum extracts, it was not clear if PfSENP1 and/or PfSENP2 were responsible for this activity. Therefore, we attempted to express the catalytic domains of both PfSENP1 and PfSENP2 in E. coli. Despite multiple attempts, we could only express and purify the catalytic domain of PfSENP1. The resulting recombinant PfSENP1 efficiently cleaved PfSUMO-pro, and cleavage was inhibited by both NEM pretreatment and by mutation of the predicted catalytic cysteine to alanine (Figure 3A). In addition, we found that recombinant wild-type but not the catalytically dead mutant PfSENP1 cleaved hSUMO-1 from the lysine side chain of RanGAP1, a well characterized SUMO-modified protein (Figure 3B).
PfSENP2 homologs (yeast Ulp2 and hSENP6 and 7) primarily function in the processing of poly-SUMO chains (Kroetz, et al., 2009; Lima and Reverter, 2008), on hSUMO-2, −3, and yeast SUMO (SMT3) at canonical SUMOylation motif sites. However, P. falciparum SUMO, like hSUMO-1, lacks canonical SUMOylation sites (Figure S2), suggesting that the parasite does not poly-SUMOylate substrates. In order to determine if PfSUMO can itself be SUMOylated, we used a previously described system (Uchimura, et al., 2004) in which E. coli is co-transformed of with a tri-cistronic plasmid for the overexpression of the SUMOylation machinery. Expression of the human SUMOylation proteins resulted in the formation of poly-SUMO chains that could be enriched using the C-terminal His6 tag on the unprocessed hSUMO-2 (Figure 3C). In contrast, co-transformation with PfSUMO-pro resulted in the recovery of only unprocessed PfSUMO (Figure 3C). These data suggest that PfSUMO cannot be SUMOylated and therefore cannot form poly-SUMO chains in the parasite. Given the lack of poly-SUMOylation and our inability to express recombinant PfSENP2, it is unclear what role, if any, PfSENP2 plays in the regulation of SUMO in P. falciparum.
The SUMO regulatory machinery of P. falciparum differs from its human host in that P. falciparum has only two predicted SENP proteases while humans have six SENPs. Differences in the ability of parasite lysates to cleave a panel of human SUMOs suggested that PfSENP1 might be biochemically distinct from the SENPs of its human host. We therefore analyzed the cleavage site specificity of PfSENP1 using a positional scanning substrate library. This library was previously used to characterize the preferred amino acid cleavage sequences of five human SENP proteases, hSENP8, and three human de-ubiquitinating proteases (DUBs), a distinct class of proteases that can cleave after the C-terminal diglycine of ubiquitin (Drag, et al., 2008; Drag, et al., 2008). Specifically, we used the P2, P3, and P4 positional scanning libraries to compare the specificity profiles of PfSENP1 to the human SENPs (Figure 3D). Not surprisingly, PfSENP1 only tolerated glycine in the P2 position. In the P3 position, PfSENP1, like hSENP1 and hSENP2, tolerated a wide range of amino acids. In the P4 position, PfSENP1 showed the greatest preference for glutamine, leucine, and norleucine. Glutamine is the native P4 amino acid of P. falciparum SUMO and all three human SUMOs; it also is the preferred P4 amino acid of hSENP1, 2, and 5. Leucine is the native P4 position of ubiquitin and is therefore strongly preferred in this position for DUBs. Among the human SENPs, only hSENP6, 7, and 8 preferentially cleave leucine over glutamine in the P4 position. Our findings also demonstrate that PfSENP1 shares P4 specificity with both groups of human SENPs but also accepts several additional amino acids in the P4 position that are not well tolerated by the hSENPs, including arginine, isoleucine, and threonine.
In order to determine if differences observed in peptide substrate specificity translate into differences in specificity in the context of the native SUMO substrate, we mutated glutamine 95 (the equivalent of the substrate P4 position) of PfSUMO-pro to threonine (PfSUMOQ95T-pro). The PfSUMOQ95T-pro mutation resulted in a 2-fold decrease in substrate processing by PfSENP1 relative to the wildtype PfSUMO-pro substrate (Figure S3A). In contrast, hSENP1 processing of PfSUMOQ95T-pro was decreased by more than 25-fold relative to processing of the wildtype PfSUMO-pro (Figure S3B). These data suggest that the P4 position could be used to drive selectivity away from the closest human homologs of PfSENP1, hSENP1 and 2.
The only reported SENP inhibitors are general cysteine alkylating agents and the full-length SUMO protein modified with a reactive group, such as a vinyl sulfone (Borodovsky, et al., 2002). Although effective for some biochemical assays, these inhibitors cannot be used to study SENPs in live cells. In addition, biotinylated probes consisting of 5–13 amino acids from the C-terminus of SUMO modified with a vinyl sulfone reactive group have been reported as probes of SENP proteases (Borodovsky, et al., 2005). However, these probes have not been shown to have SENP inhibitory activity. Therefore, we decided to screen a highly focused library of 508 irreversible cysteine proteases inhibitors (Arastu-Kapur, et al., 2008) using the parasite lysate PfSUMO-pro processing assay to identify novel small molecule SENP inhibitors (Figure S4). This library was used previously in P. falciparum to identify novel inhibitors of host cell rupture and includes general scaffolds of covalent irreversible cysteine protease inhibitors (Arastu-Kapur, et al., 2008). Since we were unable to express PfSENP2, we screened the inhibitor library against the native SUMO processing activity in parasite lysates rather than recombinant PfSENP1 alone. Interestingly, of the 508 compounds screened, only one inhibitor, JCP-666, reproducibly blocked SUMO-pro processing (Figure 4A). JCP-666 contains an aza-epoxide reactive group with a P1 aspartic acid side chain. Unlike the majority of aza-epoxide-based inhibitors in the library, JCP-666 lacks a P2 amino acid side chain. Chemical characterization of the compound indicated the primary species in the library was a mixture of chlorohydrin forms of the ring opened epoxide that formed during synthesis as a result of the removal of a BOC group with HCl (Figure 4B; Supplemental compound characterization). Because we could not isolate the epoxide form of JCP-666, we used the mixture of the chlorohydrin isomers in all our subsequent studies and refer to these simply at JCP-666.We did not find any evidence of similar chlorohydrin species for the related analogs JCP-665, JCP-667, JCP-668 even though these compounds were prepared using the same synthesis methods (Scheme S1). Interestingly, both the epoxide and chlorohydrin forms of JCP-666 were able to inhibit PfSENP1 activity (Figures S5). This suggests that either the chlorohydrin isomers inter-convert to the epoxide in reaction buffer or the chlorohydrin forms are also active as inhibitors (Scheme S2). Thus, the chlorohydrin from of JCP-666 may inhibit the target SENP by SN2-like displacement of the chloride by the active site cysteine. This is consistent with the fact that the chlorohydrin form of JCP-666 irreversibly inhibited PfSENP1. In addition, the chlorohydrins also inhibited human SENP1 activity against Pro-SUMO and also the SUMOylated RanGAP substrate (Figure S5). Regardless, the overall instability of the JCP-666 epoxide and the presence of multiple isomers of the chlorohydrin prompted us develop and alternative inhibitor for use in all subsequent studies.
For our first analog of JCP-666, we removed the aspartic acid sidechain on the main aza-epoxide scaffold. We reasoned that changes in this region would likely alter the stability of the epoxide group. Furthermore, removal of this group would simplify the synthesis by reducing steric hindrance for the key coupling reaction between the peptide scaffold and the epoxide of the molecule. We therefore synthesized VEA-260, which is identical to JCP-666 but lacks the P1 aspartic acid sidechain (Figure 4C). This new analog was as potent as the original JCP-666 against both recombinant PfSENP1 (Figure 4D; Figure S5A) and PfSENP activity in parasite lysates (Figure 4E; Figure S5B) suggesting that this aspartic acid sidechain is not essential for binding to the target enzyme. Furthermore, VEA-260, like JCP-666, acted as an irreversible inhibitor of PfSENP (Figure 4F; Figure S5C). Importantly, VEA-260 was not susceptible to epoxide ring opening during synthesis and could be isolated in the epoxide form.
To assess the selectivity of VEA-260 for P. falciparum SENPs, we examined whether it inhibited other P. falciparum cysteine proteases. Specifically, we incubated mixed-stage P. falciparum lysates with various concentrations of VEA-260 and measured the residual activity of falcipains 1/2/3 using radiolabeled probes as described previously (Arastu-Kapur, et al., 2008; Greenbaum, et al., 2002). Importantly, VEA-260, like JCP-666 (Figure S5D, E) did not inhibit any of the falcipains (Figure 4G), suggesting that it is likely to be sufficiently selective to allow its use for studies in live parasite cultures.
In order to assess whether VEA-260 was selective for parasite SENP activity over other SENP proteases and the related deubiquitinating proteases (DUBs), we monitored inhibition using the QTGG-AFC and LRGG-AFC fluorogenic substrates previously optimized for these proteases (Table 1;Drag, et al., 2008; Drag, et al., 2008). PfSENP1 and its closest human homologs (hSENP1 and 2) were all potently inhibited by VEA-260 and JCP-666. In contrast, hSENP6 (a homolog of PfSENP2) and UCH-L3 (a DUB) were not inhibited by VEA-260. Human SENP8 (a deNEDDylase) was inhibited by VEA-260 (Albrow et al Submitted) but not JCP-666, suggesting that removal of the aspartic acid P1 side chain may reduce selectivity to some extent. However, neither P. falciparum nor human red blood cells express hSENP8.
In order to determine the effects of inhibition of PfSENP activity on P. falciparum growth, we used VEA-260 for phenotypic studies in parasites cultured in human red blood cells. Although only JCP-666 was identified from the library screen, a further search of the library identified three additional compounds (JCP-665, 667, and 668) that are highly related in structure to VEA-260 and JCP-666 (Figure 5A). All three analogs have the aspartic acid P1 side chain like JCP-666. JCP-665 and JCP-667 further differ from VEA-260 only in the size of the aromatic groups on the terminal amide, while JCP-668 differs from VEA-260 in the reactive electrophile group. Interestingly, none of these analogs were obtained as the chlorohydrin form even though they were originally synthesized using the same chemistries. This suggests that the combination of the large di-naphthyl groups and the P1 aspartic acid sidechain cause the reduced stability of the epoxide of JCP-666 to HCl. All three analogs in the library did not inhibit PfSUMO-pro processing in P. falciparum lysates at the concentration originally used in our screen. However, we evaluated their ability to inhibit PfSENP1 at higher concentrations and found that JCP-668 also has weak activity (Figure 5A). The lower potency of JCP-668 suggests that the reactive electrophile affects potency, while the reduced potency of JCP-665 and JCP-667 suggest that bulky aromatics are required for efficient binding to PfSENP1. Treatment of synchronous ring stage cultures with VEA-260, JCP-665, JCP-667, and JCP-668 demonstrated that VEA-260 most potently inhibits parasite replication (EC50 70 ± 20 µM) followed by JCP-668 (EC50 220 ± 30 µM; Figure 5B). JCP-667 had no effect that could be measured within the solubility range of the compound. Interestingly, JCP-665 showed a shallow cell killing curve but was unable to affect complete clearance of the parasites (EC50 could not be determined). This non-dose-dependent activity for JCP-665 may result from general toxicity of the compound. Regardless, these data confirm that potency against PfSENP1 correlates with parasite killing, suggesting that PfSENP1 is likely to be an essential protease.
Finally, to more carefully monitor the effects of VEA-260 on parasite growth, we treated synchronously grown cultures of infected RBCs with 100 µM concentrations of the drug at various points in the blood stage lifecycle. We then analyzed thin blood smears of the parasites 12 hours later (Figure 5C). These results indicate that only parasite treated at the late schizont stage (i.e. 3 hours prior to rupture from the host (− 3hr)) looked morphologically altered and unable to release from the host RBC. The fact that parasites treated at earlier lifecycle stages did not show any significant change in growth progression or cell morphology confirms that the compound is not generally toxic to the parasites. It also suggests that blocking deSUMOylation of proteins may be most detrimental for the parasite as it prepares to exit the host cell. Further studies into the reason for this stage specific effect of the inhibitor are currently ongoing.
SENPs play an essential role in the regulation of SUMOylation, a posttranslational modification that modulates numerous biological processes. The essential role of SENPs in other eukaryotes suggests these proteases may also be essential for P. falciparum. In this study, we demonstrated that P. falciparum has at least one active SENP protease, PfSENP1. Using a positional scanning substrate library, we demonstrated that PfSENP1 has distinct cleavage specificity relative to its human homologs. By screening a library of cysteine protease inhibitors we identified one compound, JCP-666, that inhibited SUMO processing by parasite lysates and recombinant PfSENP1. Because of the instability in the epoxide electrophile of JCP-666, we focused our efforts on a novel analog VEA-260 that showed increased potency and overall stability compared to JCP-666. This compound inhibited parasite replication in a dose-dependent manner, suggesting that PfSENP activity is essential for parasite survival.
The ability of VEA-260 to inhibit PfSUMO processing by both recombinant PfSENP1 and soluble parasite lysates, the selectivity of VEA-260 for hSENP1/2-like proteases over PfSENP2 human homologs, and the ability of VEA-260 to block parasite replication all suggest that PfSENP1 is the major SENP protease in P. falciparum lysates. Although the precise effect of VEA-260 on PfSENP2 is unknown, these data suggest one of three possible scenarios: VEA-260 inhibits both PfSENP1 and 2, inhibition of PfSENP1 alone is sufficient to block parasite replication, or PfSENP2 is not an active protease. P. falciparum, like yeast, has only two predicted SENP proteases. PfSENP1 is more closely related to yeast Ulp1 (hSENP1/2-like) and PfSENP2 is more closely related to yeast Ulp2 (hSENP6/7-like). Ulp1 is cytosolic, lethal upon deletion, and more highly related to the hSENP1/2 that are susceptible to VEA-260 inhibition. In contrast, Ulp2 is localized to the nucleus and can be genetically disrupted (Kroetz, et al., 2009; Li and Hochstrasser, 2003). Defects resulting from Ulp2 disruption can be partially overcome by over-expression of the catalytic domain of Ulp1. Ulp2 is more closely related to the subset of human SENPs that includes hSENP6 and 7. As demonstrated here, VEA-260 is not an effective inhibitor of hSENP6, suggesting PfSENP2 is a less likely to be a target of this inhibitor. Therefore, we propose that PfSENP1 is the primary protease responsible for the SENP activity in P. falciparum soluble lysates and that inhibition of this protease alone is sufficient to block parasite replication.
Although we cannot rule out the possibility that VEA-260 inhibits PfSENP2, we speculate that PfSENP2 may not be an active protease. The catalytic domain of PfSENP2 has undergone a significant expansion relative to human SENPs, with nearly 200 additional amino acids relative to hSENP6 and 7, which themselves contain a 200 amino acid insertion relative to hSENP1 and 2 (Table S1; Lima and Reverter, 2008). Furthermore, the expansion within PfSENP2 contains long stretches of amino acid repeat sequences that may have disrupted protease function and may have also prevented expression of PfSENP2 in E. coli. In addition, PfSENP2 homologs (yeast Ulp2 and hSENP6 and 7) primarily function in the processing of poly-SUMO chains (Kroetz, et al., 2009; Lima and Reverter, 2008). We demonstrated here that PfSUMO cannot form poly-SUMO chains; therefore, it is unclear what role, if any, PfSENP2 plays in SUMO regulation in the parasite.
Regardless of whether or not PfSENP2 is a functional protease, VEA-260 is an inhibitor of PfSENP1. Most classes of protease inhibitors that target cysteine proteases in P. falciparum show some level of cross-reactivity with the falcipains (Arastu-Kapur, et al., 2008; Greenbaum, et al., 2002). Our lead SENP inhibitor, VEA-260, did not cross-react with the other major parasite cysteine proteases, including the falcipains, at nearly double its EC50 value in the replication assay. Furthermore, JCP-668 showed dose dependent inhibition of parasite growth proportional to the decreased potency of this compound for PfSENP1, and JCP-667 showed no toxicity to parasite growth correlating with its inability to inhibit PfSENP1. Although JCP-665 showed toxic effects not related to PfSENP1 inhibition, these effects were not sufficient to fully block parasite replication. These data suggest that the complete block in parasite replication by VEA-260 and JCP-668 was due to inhibition of the target SENP. Modulation of PfSENP1 expression would be useful to validate that VEA-260 susceptibility is proportional to the level of PfSENP1. However, previous studies in yeast have demonstrated that Ulp1 overexpression exerts a dominant-negative effect and is therefore likely to result in lethality in P. falciparum (Mossessova and Lima, 2000). Although VEA-260 also inhibits hSENP1 and 2, we demonstrated that it does not inhibit all SENPs nor does it inhibit highly related cysteine proteases, such as the DUB UCH-L3. These data indicate that VEA-260 may be a promising lead compound with a high degree of selectivity for PfSENP1 over other cysteine proteases.
Identification of a small molecule SENP-selective inhibitor also provides a new tool for the analysis of the SUMOylation pathway in P. falciparum. Recent proteomic analysis of SUMO modified proteins in P. falciparum identified a preliminary list of 23 proteins (Issar, et al., 2008). This is a surprisingly small number compared to proteomic studies in other organisms including yeast (>200 proteins identified) and the highly related parasite Toxoplasma gondii (>100 putative modified protein identified; (Braun, et al., 2009; Denison, et al., 2005). For example, SUMOylation of PfSir2 was not observed by mass spectrometry in P. falciparum, despite western blot and co-immunoprecipitation evidence that this protein is in fact modified by SUMO. This is likely due to the low levels of SUMO modified protein at any given time and the high level of activity of SENPs in P. falciparum lysates. With the discovery of small, cell permeable inhibitors of PfSENPs it should be possible to generate more complete profiles of SUMOylated proteins by mass spectrometry analysis. Furthermore, these inhibitors can be used to specifically block removal of SUMO at specific stages of the parasite lifecycle to provide a more complete picture of the dynamics of SUMOylation.
SUMOylation is a dynamic process that is involved in the regulation of a number of key biological processes. In particular, SUMO modification is required for correct transcriptional regulation. In human parasite pathogen Plasmodium falciparum, the causative agent of malaria, a large percentage of its genes are tightly regulated at the transcriptional level during its 48 hour blood stage lifecycle inside the human host. However, very little is known about the molecular mechanisms used by the parasite to control this transcriptional program. Therefore, tools that will allow a more detailed biochemical analysis of posttranslational processes such as SUMOylation are likely to help shed light on how the parasite regulates its blood stage processes. In this study, we find that only one of two P. falciparum SENP genes produce a functional enzyme. Biochemical studies of this protease reveal that it has specificity for the parasite SUMO-pro and some degree of unique specificity for primary substrates compared to the human SENPs. In order to identify new classes of small molecule inhibitors of SENPs we screened a small focused library of protease inhibitors in a newly developed assay in crude parasite extracts. This screen identified a class of non-peptidic aza-epoxides that covalently inhibit the PfSENP1protease. Using the lead compound and various analogs with reduced potency for PfSENP1, we were able to demonstrate that inhibition of PfSENP1correlates with parasite killing. Together, these studies suggest that PfSENP1 may be a viable drug target and provide valuable new tools for studying SUMOylation in both parasites and other eukaryotic systems. These studies should also enable future studies of the molecular mechanisms of SUMOylation as it relates to the regulation of the blood stage life cycle of this important human pathogen.
The catalytic domains of hSENPs 1, 2, 3, 5, 6, and 7, yeast Ulp1 and Ulp2, and the predicted SENPs of P. falciparum, P. berghei, and P. yoelii were aligned using ClustalW (MacVector 10.0.2). A phylogramwas created based on this alignment using the PhyML program with a bootstrapping procedure (number of bootstraps = 100) available on Laboratoire d’Informatique, de Robotique et de Microélectronique (LIRM) at Montpellier (France) (http://www.phylogeny.fr).
For details see Supplemental Experimental Procedures.
Strains, constructs, and primers for PfSUMO, chimeric SUMOs, PfSENP1, and PfSENP2 are described in the Supplemental Experimental Procedures. All constructs were expressed in BL21(DE3) E. coli (OD600=0.6; 0.1 mM IPTG) for 3 hr at 30°C, lysed in ML1B, and immobilized on Ni2+-NTA agarose. His6-tagged PfSENP1 was eluted with imidazole while untagged PfSENP1 was eluted from the CPD fusion with 100 µM InsP6 (Lupardus, et al., 2008). PfSENP2 was not successfully expressed in either system.
The SUMO-pro cleavage assay was modified to assess the effects of the PfSUMOQ95T-pro mutation. PfSUMOQ95T-pro or PfSUMO-pro were incubated with 2-fold serial dilutions of PfSENP1 or hSENP1 (400 nM – 3.1 nM) for 1 hr at 37°C. Proteins were separated by SDS-PAGE and visualized by Gelcode Blue protein stain reagent (Pierce), and the protease concentration required for 50% cleavage activity of the substrate (AC50) was estimated.
The SUMO-pro cleavage assay was also used to evaluate the reversibility of VEA-260 and JCP-666. PfSENP1 (200 nM) was treated for 2 hr at 37°C with VEA-260 or JCP-666 (200 µM), NEM (10 mM), or DMSO in a final reaction volume of 200 µL. An input sample was taken, and then the inhibitor treated protease was diluted, enriched with Ni2+-NTA, and eluted with 300 mM imidazole. The input and elution samples were subjected to the SUMO-pro processing assay described above.
A library of 508 cysteine protease inhibitors was screened for inhibition of PfSUMO-pro processing. For the screen, 1 µL of a 10mM DMSO stock solution of each inhibitor (100 µM final concentration) was added to 10 µg of soluble P. falciparum lysate in reaction buffer (50µL reaction) and incubated for 30 min at room temperature in a 96-well plate format. The PfSUMO-pro substrate was then added to the reaction (10 µM final concentration) and incubated for one hour at 37°C. PfSUMO-pro without lysate and lysate pretreated with DMSO alone were used as positive and negative controls for inhibition on each assay plate. The addition of 10µL SDS loading dye quenched the reaction, and cleavage was analyzed as described for the SUMO processing assays.
The lead compound JCP-666 was initially assessed by LC-MS (liquid chromatography, Agilent 1100 Series, and mass spectrometry, API 150EX, Applied Biosystems; see Supplemental Compound Characterization). However, the expected mass was not found. Subsequent chemical analyses and re-synthesis of JCP-666 suggested that the epoxide of the reactive warhead was unstable and modified by hydrochloric acid (see Supplemental Compound Synthesis and Characterization, Supplemental Results, and Schemes S1 and S2). VEA-260 was subsequently synthesized as described in the Supplemental Compound Synthesis and Characterization. The general scaffold of JCP-666 was used to perform a structure search of the library to identify related analogs. The purity and identity of the analogs JCP-665, JCP-667, and JCP-668 was assessed by LC-MS, HRMS, and proton NMR (see Supplemental Compound Synthesis and Characterization). Characterization of all compounds is described in detail in the Supplemental Compound Characterization.
The ability of VEA-260 to compete for labeling of common cysteine proteases of P. falciparum was measured using two general activity-based probes as described previously(Arastu-Kapur, et al., 2008; Greenbaum, et al., 2002) and in more detail in the Supplemental Experimental Procedures.
The ability of VEA-260 and JCP-666 to inhibit a panel of ubiquitin-like protein deconjugating proteases, including PfSENP1, hSENP1, hSENP2, hSENP6, hSENP8 (a deNEDDylase), UCHL3 (DUB), and IsoT (USP5; DUB), was tested using a previously described fluorogenic substrate assay (Drag, et al., 2008; Drag, et al., 2008). See Supplemental Experimental Procedures for additional details.
Synchronous P. falciparum ring-stage parasites (~9 h post invasion; ~1% parasitemia; 0.5% hematocrit) were treated with VEA-260, JCP-666, JCP-668, −665, or −667 at (0–300 µM) for 75 h at 37°C in a 96-well-plate format. Cells were fixed with 0.05% glutaraldehyde (Sigma) in phosphate-buffered saline (PBS) overnight at 4°C, permeabilized with 0.25% Triton X-100 for 5 min at room temperature (25°C), and stained with a 1:100 dilution of a 5 mg/ml working solution of propidium iodide (Sigma) in 200 µL PBS. Parasite replication was monitored by observation of a propidium iodide-stained population using a BD FACScan flow cytometer (Becton, Dickinson and Co.) located in the Stanford Shared FACS Facility at Stanford University. The data obtained were analyzed to quantify the number of infected and uninfected red blood cells using Flowjo 8.7.3. Parasitemia was determined as the ratio of infected cells to total cells. Parasitemia was normalized to DMSO for each run. The assay was repeated three to six times per drug per dose, and the EC50 was calculated by non-linear regression using GraphPad Prizm software.
Synchronous parasites were treated for 12 hr with either DMSO or 100 µM of VEA260 at late schizont (3 hr prior to rupture), ring (9 hr post infection -p.i.-), late ring (21 hr p.i.), or trophozoite (33 hr p.i.) stages. The morphology of the VEA260-treated parasites was compared to that of the controls Geimsa-stained thin blood smears.
We would like to thank K. Wilkinson for providing the IsoT and UCH-L3 proteases and M. Verdoes and A. Guzetta for assistance with the compound characterization. E.L.P. was supported by the US National Science Foundation Graduate Research Fellowship 2005021299. This work was funded by the Burroughs Wellcome Trust New Investigators in Pathogenesis Award and by the NIH grants R21-AI088541 and R01-AI078947 (to M.B.) and R01-CA69381 (to GS).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.