Targeting the ERAD pathway via inhibition of signal peptidase for antiparasitic therapeutic design

Michael B. Harbuta,1, Bhumit A. Patela, Bryan K. S. Yeungb, Case W. McNamarac, A. Taylor Brightd, Jaime Ballardc, Frantisek Supekc, Todd E. Goldee, Elizabeth A. Winzelerc,f, Thierry T. Diaganab, and Doron C. Greenbauma,2

aDepartment of Pharmacology, University of Pennsylvania, Philadelphia, PA 19104; bNovartis Institute for Tropical Diseases, Singapore 138670; cGenomics Institute of the Novartis Research Foundation, San Diego, CA 92121; dBiomedical Sciences Program, University of California at San Diego, La Jolla, CA 92093; eDepartment of Neuroscience, University of Florida, Gainesville, FL 32610; and fDepartment of Pediatrics, University of California at San Diego, La Jolla, CA 92093

Edited by Thomas E. Wellems, National Institutes of Health, Bethesda, MD, and approved November 15, 2012 (received for review September 17, 2012)

Early secretory and endoplasmic reticulum (ER)-localized that redundant complexes. During periods of ER stress, ERAD are terminally misfolded or misassembled are degraded by a ubiq- and UPR work together to achieve protein homeostasis within the uitin- and proteasome-mediated process known as ER-associated ER (4–7). degradation (ERAD). Protozoan pathogens, including the causa- P. falciparum lacks conventional transcriptional regulation and tive agents of malaria, toxoplasmosis, trypanosomiasis, and leish- shows little coordinated response to internal or external perturba- maniasis, contain a minimal ERAD network relative to higher tions such as heat stress or drug toxicity (8). Intriguingly, the tran- scription factors that initiate the UPR (IRE1, ATF6) in mammalian eukaryotic cells, and, because of this, we observe that the malaria – parasite Plasmodium falciparum is highly sensitive to the inhibition cells are absent from the genome of P. falciparum (9 11). Lacking of components of this protein quality control system. Inhibitors that any transcriptional response, the down-regulation of , identification, and subsequent disposal of misfolded proteins would specifically target a putative component of ERAD, signal be the parasite’s major compensatory mechanisms to maintain ER peptide peptidase (SPP), have high selectivity and potency for P. homeostasis during periods of ER stress. falciparum. By using a variety of methodologies, we validate that Here we show through a bioinformatics analysis that the ERAD SPP inhibitors target P. falciparum SPP in parasites, disrupt the pro- pathway of protozoan pathogens, including P. falciparum, is highly ’ tein s ability to facilitate degradation of unstable proteins, and in- simplified relative to mammalian cells, and that P. falciparum is MICROBIOLOGY hibit its proteolytic activity. These compounds also show low therefore vulnerable to small molecules that have been established nanomolar activity against liver-stage malaria parasites and are also to inhibit proteins within the ERAD system. In particular, malaria equipotent against a panel of pathogenic protozoan parasites. Col- parasites within multiple life stages, along with other protozoan lectively, these data suggest ER quality control as a vulnerability of pathogens, are highly sensitive to the inhibition of one of these protozoan parasites, and that SPP inhibition may represent a suit- putative ERAD proteins, peptidase (SPP), which able transmission blocking antimalarial strategy and potential pan- we validate to act in this ERAD pathway through a variety of protozoan drug target. techniques, and further suggest that SPP inhibition may be a viable antiparasitic strategy. intramembrane | small molecule | target validation Results rotozoan pathogens, including the malaria parasite Plasmodium A Bioinformatics Approach Identifies Minimal ERAD Pathway in Protozoan Pfalciparum, constitute one of the most substantial global public Pathogens, of Which P. falciparum Shows Heightened Susceptibility to health problems faced today. The emergence and spread of drug- Inhibition. A recent analysis of the UPR machinery in protozoan resistant parasites has rendered many of the traditional chemo- parasites revealed a distinct UPR characterized by the absence of therapeutics clinically ineffective in many cases (1). Therefore, the transcriptional regulation and therefore entirely reliant on trans- identification and validation of novel Plasmodium molecular tar- lational attenuation in response to ER stress (12). As a result of gets would greatly facilitate the discovery of new antimalarial drugs. this, Leishmania donovani parasites have heightened sensitization In the pathogenic stage, P. falciparum resides within an eryth- to compounds that promote ER stress, such as DTT (reducing rocyte, which is elaborately remodeled by the parasite to allow the agent) (12). In yeast and mammalian cells, ER stress initiates UPR infected cell to escape immune detection and to facilitate nutrient and ERAD in an intimately coordinated fashion, whereby the uptake and waste disposal in a cell with normally low metabolic induction of one process increases the capacity of the other (5, 7). fi activity. A necessary component of the parasite’s capacity to in- Thus, we reasoned that the modi ed response to ER stress in habit the erythrocyte is the establishment of a unique parasite- protozoan pathogens also likely extends to the ERAD pathway. derived protein secretory network that allows protein trafficking to Our investigation of this hypothesis using standard orthologue destinations beyond the parasite, including a parasitophorous detection tools revealed a striking lack of putative ERAD proteins vacuole and erythrocyte cytosol and plasma membrane (2). in P. falciparum relative to the extensive mammalian network (Fig. The endoplasmic reticulum (ER) is the hub of the secretory 1A and Fig. S1). All functional modules of the ERAD pathway (as pathway, where secretory proteins are folded and targeted for their named in ref. 7), including protein recognition, translocation, respective destination. The ER is sensitive to changes in calcium flux, temperature, and exposure to reducing agents, and, in higher eukaryotes, these stressors elicit transcriptional and translational Author contributions: M.B.H., B.A.P., and D.C.G. designed research; M.B.H. and B.A.P. responses to stabilize already synthesized secretory proteins and performed research; B.K.S.Y., J.B., F.S., T.E.G., E.A.W., T.T.D., and D.C.G. contributed new reagents/analytic tools; M.B.H., B.A.P., C.W.M., A.T.B., E.A.W., and D.C.G. analyzed decrease the load of translocation into the ER, a network collec- data; and M.B.H. and D.C.G. wrote the paper. tively called the unfolded protein response (UPR). In addition to fl the UPR, there exists a coordinated and extensive monitoring The authors declare no con ict of interest. system in the ER to ensure that terminally misfolded proteins or This article is a PNAS Direct Submission. are quickly extracted from this compartment and then Freely available online through the PNAS open access option. degraded via the –proteasome system in the cytosol in a 1Present address: California Institute for Biomedical Research, La Jolla, CA 92037. process known as ER-associated degradation (ERAD) (3). Studies 2To whom correspondence should be addressed. E-mail: [email protected]. in yeast and mammalian cells have shown ERAD to be a complex This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. network that comprises compartmentally restricted and partially 1073/pnas.1216016110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1216016110 PNAS Early Edition | 1of6 Downloaded by guest on October 2, 2021 A Ub B Selectivity index OH O ER Stress HRD1 H Substrate Substrate Protein target Inhibitor Pf IC 50 ( M) HepG2 IC 50 ( M) (HepG2/Pf) F N N recognition UBE2G2 N RNF5 extraction H O BIP PDI SEL1 P97 ATX p97 DBeQ 0.31 3.8 12.3 O GRP94 ERDJ5 NPL4 USP13 ERO1 Dislocation PDI 16F16 7.5 2.5 0.33 F LY-411575 SPP UFD1 UBE4B DERLIN1/2 RAD23 DOA1 17DMAG 0.061 1.1 18.0

ERAD BiP Ub OH O misfolded Ub Proteasome Epoxomycin 0.0072 0.49 68 H Ub N protein N N Proteasome SPP (Z-LL) 2 1.5 >10 >6.6 H ER cytosol O O Dislocation Substrate Substrate SPP LY-411575 0.10 >10 >100 NITD679 UPR recognition ESTY1/2 Ub ligase extraction ERLEC1 OS9 TRAP gp78 HERP YOD1 PNG1 SPP NITD679 0.065 >10 >100 Translational PERK DERLIN3 SVIP VIMP ERFAD CPVL ERLIN1 TMUB1 SPP NITD731 0.017 >10 >100 regulation eIF2α LONP2 iRhom1 ERLIN2 BRI3BP TRAM1 OH O H H RHBDL4 UBAC2 FAM8A1 UBXD2 O N N Transcription IRE1 VCIP135 N N AUP1 H regulation ATF6 O O O present in P. falciparum absent in P. falciparum NITD731

Fig. 1. P. falciparum has a minimal ERAD pathway and shows heightened susceptibility to inhibition of constituent proteins. (A) A bioinformatics analysis of ERAD in P. falciparum reveals a reduced number of ERAD orthologues in each functional module. Components of ERAD not identified in P. falciparum are sectioned in the shaded boxes, and proteins in white boxes have orthologues identified in P. falciparum.(B) Treatment of P. falciparum and human HepG2

cells with inhibitors of ERAD proteins reveals increased sensitivity of parasites vs. host cells. This is quantified by the selectivity index, which is a ratio of the IC50 for HepG2 cells divided by the IC50 for P. falciparum.

ubiquitin ligation, and protein extraction, showed far fewer respectively (Fig. 1B), and not active at all against HepG2 cells, orthologues in P. falciparum relative to the corresponding mam- indicating a high degree of selectivity for malaria parasites over malian pathway. We expanded our inquiry to three other patho- mammalian cells. genic protozoans, Toxoplasma gondii, Leishmania infantum, and Trypanosoma cruzi, to assess whether the reduction of the ERAD PfSPP Is a Component of ERAD and Its Role Can Be Inhibited by Small proteome is a common feature among protozoa (Fig. 1A and Fig. Molecules. Recent studies in mammalian cells have established the S1). On average, each protozoan investigated showed a ∼50% to necessity of hSPP1 during the process of dislocation during virus- 60% decrease in orthologues shared with the mammalian ERAD mediated ERAD of MHC class I molecules (20–23). In addition, system. As each of the genomes remains incompletely annotated, hSPP1 associates with a number of proteins required for ERAD, it may also be possible that some components from each including TCR8 and protein disulfide , and are so divergent that they may have not been detected by our also assembles with misfolded membrane proteins (21, 22, 24, 25). analysis. On the whole, the reduction in protozoan ERAD pro- Recent studies have concluded that human hSPP1 and PfSPP teins suggests that the pathway as found in the parasites may be are ER-resident proteins (26–28). We corroborated these find- less dynamic than its mammalian counterpart, and that the loss of ings by localizing PfSPP in parasites to the ER by indirect function of individual components of the protozoan pathway immunofluorescence by using a PfSPP antibody (Fig. S2). would severely compromise parasite viability. Throughout the parasite life cycle, we observed strict perinuclear As a result of the inherent genetic intractability of P. falciparum, localization of PfSPP in the ER, in accordance with what has been we undertook a small molecule-based approach to initially per- previously observed. Co-indirect immunofluorescence assays form a small screen of well characterized inhibitors that target an performed with an antibody to the resident ER protease plas- array of ERAD pathway proteins. We focused on ERAD com- mepsin V or the generation of a parasite line expressing HA- ponents identified in P. falciparum with known associated inhib- tagged PfSPP also confirmed ER residency, in agreement with itors, including 90 (13–15), protein disulfide previously published literature (26–28) (Fig. S2). isomerase (16), cytosolic AAA (ATPase associated with diverse To assess the effect SPP inhibitors have on the ability of para- cellular activities) ATPase p97 (or Cdc48; valosin-containing sites to cope with ER stress, we treated P. falciparum parasites protein) (17, 18), ER intramembrane aspartyl protease SPP (19), simultaneously with thapsigargin and SPP inhibitors and analyzed and the proteasome. P. falciparum parasites were indeed suscep- the effects of the inhibitor combinations for evidence of synergy. tible to each of the inhibitors to varying degrees, and all com- Thapsigargin causes the release of calcium from the ER, com- pounds displayed IC50 values of less than 10 μM (Fig. 1B). In promising the ER’s ability to produce properly folded proteins, addition, each inhibitor was assayed against the human hepatocyte and is lethal to parasites (Fig. S3) (29). The combination treatment HepG2 cell line, and a selectivity index was produced to determine of thapsigargin with the SPP inhibitors (Z-LL)2, LY-411575, or the fold increase in potency for the inhibitor toward P. falciparum NITD731 produced synergistic parasiticidal effects beyond what vs. the human cell line. As predicted, the majority of compounds would be expected by simply adding the effects of the individual were more potent for P. falciparum vs. the human cell line (Fig. compounds together (Fig. 2A). The results are presented as an 1B). These data suggest that P. falciparum may be acutely sus- isobologram of the varying ratios of the two compounds, in which ceptible to inhibition of the ERAD pathway, or to compounds that points below the line of additivity indicate synergy. As a negative disrupt the homeostatic balance of the ER. control, no synergy was seen with atovaquone (Fig. 2A, Lower), an As P. falciparum was highly sensitive to the SPP inhibitor LY- antimalarial ubiquinone analogue whose mechanism of action 411575, and this inhibitor showed no observable toxicity to host cells, involves the mitochondria (30). This suggests that the synergy we decided to focus on further chemical validation of P. falciparum between SPP inhibitors and thapsigargin is caused by the ER stress SPP (PfSPP) as an antimalarial target. In addition, PfSPP is most (thapsigargin) and the parasite’s compromised ability to manage it similar to mammalian hSPP1, which is an aspartyl protease (∼40 (PfSPP inhibition), which further connects PfSPP to a potential kDa) in the same family as (PS). Because of the central function within the parasite’s ERAD network. role of PS in the pathology of Alzheimer’s disease, a large number To provide a more direct link between PfSPP and its role in of inhibitors have been developed against this intramembrane ERAD, we used an established ERAD assay, which makes use of aspartyl protease, and many of these inhibitors show cross-in- an intrinsically unstable protein, CD3δ, that has been used pre- hibition of SPP, including LY-411575. Therefore, we were able to viously to identify novel ERAD components and small molecules “piggyback” on these efforts to screen a series of PS inhibitors that inhibit the disposal of unstable proteins (31, 32). In this system, based on the scaffold of LY-411575 in parasite replication assays. CD3δ is naturally degraded via ERAD when expressed in most cell We show the top two hits, NITD731 and NITD697, which are lines. When ERAD is impaired, the CD3δ protein accumulates, potent against P. falciparum, with IC50 values of 17 nM and 65 nM, which can be monitored by Western blot analysis (Fig. 2B).

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1216016110 Harbut et al. Downloaded by guest on October 2, 2021 HA epitope was transfected and assayed by Western blot. A 1.0 B Upon treatment of transfected cells with the translation in- NITD731 + ERAD (Z-LL) Degradation hibitor cycloheximide, levels of CD3δ-HA decreased (Fig. 2C, 2 cyto LY-411575 lane 1). siRNA knockdown of hSPP1 abrogated degradation 0.5 of CD3δ-HA, which produced a concomitant accumulation of CD3δ lumen CD3δ-HA protein (Fig. 2C, lane 2). This corroborates previous HA - ERAD Accumulation literature that showed that hSPP1 mediates ERAD of CD3δ (22). FIC Thapsigargin The loss of ERAD-mediated degradation of CD3δ-HA in cells 0 in which hSPP1 was knocked down was rescued with the expres- 0 0.5 1.0 C ++ +++ +CD3 -HA sion of a codon-optimized PfSPP (Fig. 2C, lane 3). In addition to FIC SPPi - + +++ +hSPP1 RNAi δ – -- PfSPP the aforementioned genetic studies, treatment of the CD3 -HA 1.0 +++ + expressing cells complemented with PfSPP with SPP inhibitors also resulted in the inability of the cells to degrade CD3δ-HA (Fig. 2 Inhibitor C – fi Pf (Z-LL) LY-411575NITD731 2 , lanes 4 6), con rming that these inhibitors impaired SPP 0.5 CD3 -HA function in ERAD. This suggests that PfSPP performs a similar role as its mammalian counterpart, and that inhibition of PfSPP likely hSPP1 restricts its ERAD function in P. falciparum. FIC Atovaquone PfSPP 0 SPP Inhibitors Directly Block PfSPP Activity in Heterologous Proteolytic 0 0.5 1.0 Tubulin FIC SPPi Assay. We next endeavored to assess whether our SPP inhibitors blocked the proteolytic activity of PfSPP. Therefore, we developed Fig. 2. SPP inhibitors sensitize parasites to ER stress and PfSPP facilitates the a cell-based SPP protease assay to interrogate its activity (Fig. degradation of a canonical ERAD substrate. (A) Individual IC50 values against 3A). Initially, we envisioned that we would build this assay in parasites were determined for each SPP inhibitor and thapsigargin, and for P. falciparum; however, genetic tractability was severely limiting, each inhibitor in combination with thapsigargin, at four different fixed ra- and PfSPP is an essential , so perturbations to this protein are tios. The fractional IC50 value (FIC; IC50 of inhibitor in combination divided by lethal to the organism (33). To circumvent this difficulty, we IC50 of inhibitor alone) was determined for each inhibitor in each combi- designed the assay in the budding yeast Saccharomyces cerevisiae nation and plotted on the isobologram. The x axis indicates the FIC of the because of the genetic manipulability of this system and because SPP inhibitors. The average ΣFIC for the SPP inhibitors with thapsigargin

S. cerevisiae contains no PS orthologue (thus yielding lower po- MICROBIOLOGY were 0.80, 0.72, and 0.79 for NITD731, LY-411575, and (Z-LL)2, respectively. Σ tential background signal) and only a single SPP gene (rather than The diagonal line represents a FIC of 1, indicating additivity between the fi two inhibitors. Area below the line (i.e., ΣFIC of <1) indicates a synergistic ve as in mammalian cells). It is worth noting that previous reports combination. (B) Schematic of the CD3δ-HA ERAD assay. CD3δ-HA is de- indicated that there was no SPP in yeast; however, an SPP-like graded in the presence of a functional ERAD network and stabilized in its protein [YKL100c; i.e., S. cerevisiae SPP (ScSPP)] is encoded in absence or inhibition. (C) U-2 OS cells actively degrade CD3δ-HA via ERAD, the yeast genome that is homologous to human hSPP1 (Fig. S4). a process that can be abolished by the knockdown of hSPP1. This process can An ScSPP KO strain was made in the Δpdr1,3 strain (a double-KO be rescued by the coexpression of PfSPP. SPP inhibitors block the PfSPP- strain that lacks multidrug resistance pumps). The Δspp yeast KO mediated degradation of the substrate. was viable and had no gross phenotypic differences compared with the WT strain. The Δpdr1,3-Δspp strain was then transformed with the following plasmids: (i) a reporter consisting of glucocorticoid δ Initially, we attempted to use this system by expressing CD3 in response element (GRE) fused to LacZ; (ii) a galactose-inducible P. falciparum or T. gondii but were unable to derive parasites that expression plasmid with a truncated glucocorticoid receptor-fused could express CD3δ, which may indicate how sensitive these par- C-terminal to the transmembrane domain of human CMV gly- asites are to unfolded proteins in the ER. Therefore, we switched coprotein UL-40 (gpUL40), a canonical Type II transmembrane to a human cell line-based assay wherein CD3δ tagged with an SPP substrate; and (iii) PfSPP codon-optimized for eukaryotic

Protease Substrate A Type II B GR526 MYC Pf galactose 1. TM domain

SPP 3x HA pdr1,3 2. SPP wb: anti-HA wb: anti-myc GRE -galactosidase 3. * C * ER lumen C 12 * Fig. 3. PfSPP is an active protease inhibited by PS/SPP inhibitors. (A) In the yeast activity assay, SPP cleaves 10 the substrate and releases GR526, which translocates x ER Membrane to the nucleus and binds the GRE. Expression of lacZ is cleavage Type II Type 8 detected via a luminescent substrate. (B) Expression of TM domain HA-tagged PfSPP in the Δpdr1,3-ΔSPP yeast strain as 6 detected by Western blotting for HA. gpUL40 is cytosol MYC SPP detected only upon induction of galactose, as detec- Nucleus 4 ted by anti-myc Western blotting. (C) Δpdr1,3 ΔSPP

GR526 with an overexpressed PfSPP shows successful cleav- GR526 MYC 2 μ N age of the substrate, inhibition by 50 M LY-411575, N GRE -galactosidase NITD679, and NITD731, but no change with the PS- Fold Change (relative to uninduced) 0 specific inhibitor DAPT at 50 μM. The fold activity is a ratio of induced, induced plus 50 μM LY-411575, or DAPT induced plus 50 μM DAPT to uninduced. Mean IC50 ± Induced NITD679NITD731 Uninduced LY-411575 SD is shown (*P < 0.05).

Harbut et al. PNAS Early Edition | 3of6 Downloaded by guest on October 2, 2021 expression. Upon cleavage of the transmembrane domain by PfSPP of NITD731 by the generation and sequencing of resistant para- (Fig. 3A, Lower), the glucocorticoid receptor is no longer anchored sites. Identification of genetic changes in the resistant parasite line by the transmembrane domain of the substrate and translocates provides details as to the potential molecular target(s) of the into the nucleus, where it binds the GRE that drives expression of compounds in culture, and also other resistance mechanisms not β-gal. PfSPP activity is then determined by β-gal activity via the directly related to direct inhibitor binding (34, 35). Drug-resistant addition of a chemiluminescent substrate of β-gal. parasites were selected by the application of sublethal amounts of To test the activity of PfSPP, we used the engineered yeast the inhibitor over a period of months, with inhibitor concentration Δpdr1,3-Δspp strain containing the reporter and substrate plasmids increasing concomitantly as parasite resistance increased. We and overexpressed PfSPP. The expression of PfSPP and substrate successfully selected for resistant parasites to NITD731 (identified were detected by using a Western blot with anti-myc and anti-HA as NITD731r), and the resulting resistant line was found to be tag antibody, respectively (Fig. 3B). The results in Fig. 3C show that more than fivefold more resistant than the parental Dd2 clone PfSPP is indeed an active protease that cleaves the GR526–gpUL40 (Fig. 5A). Assays of the resistant line with the antimalarial agents substrate. We next confirmed that the SPP inhibitors LY-411575, chloroquine and atovaquone showed no significant differences in NITD679, and NITD731 inhibited PfSPP activity. In contrast, the sensitivities to these compounds (Fig. 5A). PS-specific inhibitor DAPT showed no inhibition of PfSPP (Fig. 3C To identify a genetic change in the pfspp gene (PF3D7_1457000) and Fig. S5), illustrating that proteolytic activity of PfSPP is spe- that conferred resistance to NITD731, we generated a clonal par- cifically inhibited by these small-molecule SPP inhibitors. asite line from the resistant parasite trial by limiting dilution. Se- quencing analysis of the pfspp coding sequence revealed a single SPP Inhibitors Directly Bind PfSPP in the Malaria Parasite Proteome. nonsynonymous mutation, L333F (Fig. 5B). From the PfSPP cod- To provide a more direct link between the action of these SPP ing sequence, it is known that the L333 residue resides in trans- inhibitors on parasites with the putative endogenous target PfSPP, membrane domain 8, near the highly conserved PAL motif, a we synthesized an activity-based probe (ABP) based on the SPP hydrophobic region necessary for activity in SPP and PS (Fig. 5C). Unfortunately, no crystal structural yet exists for PfSPP, hindering inhibitor scaffold (Z-LL)2. A similar (Z-LL)2 ABP was originally used by Weihofen et al. to discover the SPP class of (19). predictions on the role of the L333 residue for the mechanism of PfSPP proteolysis. The (Z-LL)2 ABP contains a benzophenone moiety to allow co- “ ” valent crosslinking of the probe to its target, as well as a biotin tag To ascertain if there were any off-target mutations resulting for affinity purification. Treating solubilized parasite lysates with our from the in vitro selection protocol, we obtained whole-genome sequencing (WGS) data by using the Illumina platform for (Z-LL)2 probe confirmed that PfSPP is being directly targeted by – fi NITD731r. For NITD731r, we obtained 42× bulk genomic cov- (Z-LL)2 and our LY-411575 based compounds (Fig. 4). The rst fi lane of the blot in Fig. 4 shows a successful immunoprecipitation/ erage with 90.8% of the genome covered by ve or more high quality bases. This resistant clone was compared with WGS data Western analysis of PfSPP, indicating that PfSPP is targeted by × (Z-LL) . Pretreatment with unbiotinylated (Z-LL) effectively for the parent Dd2 clone, for which we obtained 24 bulk ge- 2 2 nomic coverage with 71.0% of the genome covered by five or competes for labeling of PfSPP against the biotinylated (Z-LL)2 probe, providing confirmation that (Z-LL) is not nonspecifically more high-quality bases. In total, 7,834 high-quality variants were 2 found in the two sequenced strains. Of these variants, 7,831 are labeling PfSPP. Likewise, LY-411575, NITD679, and NITD731 found in the parental and the resistant strain, and represent the compete for labeling in manner that is consistent with the relative genetic differences separating Dd2 from the 3D7 reference to potency of each compound against the parasite, confirming that which the strains were compared. The remaining three variants these compounds do indeed bind PfSPP within the complex parasite represent genetic differences between the parental clone and the proteome. resistant clone. One of these is the L333F mutation originally fi Pf found in pfspp by Sanger sequencing. The other two mutations Identi cation of SPP Inhibitor Target by Selection of Resistant Parasites. consist of an intronic mutation in PF3D7_0103500 and a non- To complement the aforementioned ABP studies, we used also – synonymous mutation in PF3D7_1135400 (Table S1). PF3D7_ used a chemical genetic approach to identify the parasite target(s) 1135400 is annotated as a hypothetical gene in PlasmoDB 9.1, and may represent another direct binding target or may be strictly involved in a unknown resistance mechanism; however, that this gene is purely hypothetical limits our ability to analyze its role (Z-LL) 2 -biotin O in this resistance mechanism. O O O To confirm the role of the L333F mutation in PfSPP that con- H H H H H BIOTIN N N N N N N N fers resistance in P. falciparum, we expressed the mutant PfSPP H H O O O O O (L333F) in our yeast assay and analyzed the potency of NITD731. The IC50 in the mutant L333F PfSPP showed more than twofold 10 M (Z-LL) 2 -biotin increase vs. that of WT PfSPP (Fig. 5D). This change was not a result of different expression levels of either protein, as the total luminescence of episomally expressed PfSPP and PfSPP L333F was 2 Inhibitor pretreat the same (Fig. S6). - (Z-LL) LY-411575NITD679NITD731 We also wished to confirm the importance of this mutation in PfSPP generating resistance in live parasites. To do this, the L333F PfSPP fi Elution gene was ampli ed from cDNA of the resistant parasite line and Pf ligated into an expression vector that would allow for transposase- SPP mediated integration into a parasite genome (36). Transgenic Input parasites expressing L333F PfSPP or WT PfSPP were generated and were then assayed for replication while in the presence of in- Fig. 4. SPP inhibitors target PfSPP in a complex parasite proteome. Struc- – creasing inhibitor concentrations. The transgenic L333F-expressing ture of the (Z-LL)2 based ABP used for parasite lysate labeling. Lower: 3-[(3- fi cholamidopropyl)dimethylammonio]-2-hydroxy-1-propanesulfonate–solu- parasites showed a statistically signi cant increase in resistance to bilized membrane proteins were incubated with 10 μMof(Z-LL) -biotin for NITD731 relative to the parental Dd2 clone. The increase in re- 2 sistance was not attributed to increased WT PfSPP expression, as 1 h, then irradiated with UV light. (Z-LL)2-biotin–labeled PfSPP was iden- tified via immunoprecipitation/Western analysis using streptavidin aga- parasites transfected with a WT allele showed no increase in its IC50. rose for immunoprecipitation and anti-PfSPP for Western blotting. (Fig. 5E). The modest level of resistance observed in our transgenic Competition labelings were carried out by pretreating samples with 5× line relative to the original drug-selected resistant parasite line may concentration of the indicated inhibitors. potentially be a result of low expression of the mutant PfSPP and

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1216016110 Harbut et al. Downloaded by guest on October 2, 2021 * A B C D E 30 * 10000 Dd2 parental 2.2 25 NITD731r L P G L * 1000 M A LC L 1.6 20 * Y G Dd2 parental D D 1234567 89 15 100 1.2 50 50 50 IC ( M) IC (nM) IC (nM) 0.80 10 10 0.40 5 PfSPP 1 0 0 NITD731 Chloroquine Atovaquone Dd2 SPP SPP NITD731r Pf Pf

Pf SPP(WT)SPP(L333F) Dd2+ Dd2+ Pf (L333F)

Fig. 5. Identification of SPP inhibitor targets in parasites. (A) Parasites grown in sublethal concentrations of NITD731 developed resistance to the compound, resulting in multifold increase in sensitivity vs. that of the parental line. The same parasites did not develop resistance to chloroquine or atovaquone. (B) Sequencing of the PfSPP gene from clones of each of the resistant lines revealed a G-to-A base mutation (in the reverse primer sequence) resulting in the nonsynonymous change, L333F. (C) The location of the L333 amino acid maps to transmembrane domain 8, upstream of the highly conserved PAL motif. (D) The L333F PfSPP mutant was generated for use in the yeast activity assay and shows a greater than twofold resistance to NITD731. (E) Introduction of the L333F transgene into Dd2 parental parasites confers resistance to NITD731. Transgenic parasites expressing the WT gene were also introduced as

a control. Mean IC50 ± SD is shown (*P < 0.05).

simultaneous presence of the endogenous WT allele in our trans- characterized by an integrated and redundant network topology. genic parasite lines. However, the data remain statistically signifi- The potential lack of redundancy of the ERAD network in para- cant and also are commensurate with the results of other studies of sites may be manifested by their heightened sensitivities to the this kind in P. falciparum. Together, the resistance data from Fig. 5 inhibitors of ERAD components relative to mammalian cells.

along with direct binding data in Fig. 4 suggest that PfSPP is indeed Of the candidate parasite ERAD targets, we confirmed a role for MICROBIOLOGY a direct target of the lead compound NITD731 and is important to PfSPP in the turnover of an unstable membrane protein. The pre- survival of P. falciparum parasites in culture. cise function of PfSPP in the recognition and degradation of aber- rant membrane proteins remains unanswered. As SPP is a multipass NITD731 Is Potent Against Liver-Stage Malaria Parasites as well as transmembrane protein, it is tempting to think of SPP as a channel Nonmalarial Pathogenic Protozoan Parasites. Prevention of liver- through which ERAD substrates translocate into the cytosol, the stage development would lead to true causal prophylaxis and would molecular identity of which has not yet been conclusively identified interrupt transmission. Therefore, we assessed the efficacy of (38). However, this scenario is complicated by the protease active fi NITD731 against liver-stage parasites by using a high-content site that lls the hydrophilic intramembrane cavity of SPP. imaging assay that allows the investigation of compound efficacy We also show that PfSPP can be directly targeted by highly against P. yoelii liver schizonts (37). We found that NITD731 was potent small molecules in P. falciparum by using a chemical bi- ology approach, and that a single nonsynonymous point mutation potent against exoerythrocytic parasites, with a comparable IC50 to that of the antimalarial drug atovaquone, one of the few com- within PfSPP was selected for by passage of parasites at sublethal monly used antimalarial agents that has efficacy against this stage doses of compound. In the generation of this resistant parasite, (Table 1). there was hypothetical protein that was found to contain non- Because other apicomplexan and kinetoplastid parasites have synonymous mutation. Given the purely hypothetical nature of SPP orthologues, we also assessed NITD731 against T. cruzi and this gene, it is hard to predict whether the corresponding protein is a direct target of NITD731 or is playing an indirect role in T. gondii. NITD731 was especially effective against both parasite resistance, although, given its structure, it would not appear to be species (Table 2), suggesting that NITD731 represents a pan- a channel or transporter. antiprotozoan SPP inhibitor. Although an that may be essential to a pathological or- Discussion ganism does not guarantee it is a suitable drug target, we believe fi focusing on PfSPP is worthwhile for several reasons. Perhaps most The identi cation of novel targets and antiparasitic compounds is promisingly, there exists an extensive drug-discovery repositioning a pressing need for infectious diseases of the developing world, opportunity for potential SPP inhibitors in existing chemical li- and is made all the more urgent by the emergence of drug-resistant braries from pharmaceutical companies. PS and SPPs are part of the parasites. Here we show that the ERAD pathway represents an A22-family integral membrane aspartyl proteases; it is therefore exploitable vulnerability in P. falciparum and other protozoan likely that there exist numerous potent PfSPP drug-like inhibitors in parasites. The absence of many mammalian ERAD orthologues PS-directed chemical libraries, reducing cost barriers for devel- in protozoans contrasts with the mammalian system, which is opment (39–41). Indeed, LY-411575 was a product of previous a PS drug-discovery effort. Table 1. Effect of SPP inhibitors on P. yoelii liver-stage parasites Mean P. yoelli EEF Table 2. Potency of NITD731 against T. cruzi and T. gondii Compound IC ± SD, μM HepG2 IC , μM 50 50 Parasite Mean IC50 ± SD, μM Host cell IC50, μM

Atovaquone 0.0015 ± 0.00090 >10 T. cruzi 0.00081 ± 0.00065 3T3 >10 NITD731 0.0078 ± 0.0018 >10 T. gondii 0.071 ± 0.011 U-2 OS >10 NITD679 0.047 ± 0.015 >10 LY-411575 0.048 ± 0.032 >10 T. cruzii amastigote/trypomastigote viability was assayed using biolumi- nescent determination of β-gal enzyme activity through a coupled reaction SPP inhibitor activity was assayed against P. yoelii sporozoites and host with firefly luciferase in parasites that harbor a β-gal gene. T. gondii viability HepG2 cells. EEF, exo-erythrocytic form. was monitored using a transgenic line expressing a luciferase reporter gene.

Harbut et al. PNAS Early Edition | 5of6 Downloaded by guest on October 2, 2021 Worries about host toxicity by SPP inhibition are mitigated by Materials and Methods the fact that knockdown of SPP is nonlethal in human cell lines. Details of parasite culture, including transfections, IC50 determinations, re- Furthermore, SPP inhibitors showed no toxicity when assayed sistance generation, and WGS, as well as cell-based assays, are provided in SI against human cell lines (HepG2, 3T3, U-2 OS) as shown here, Materials and Methods. Data were retrieved from the following sources: (i) indicating a potential high therapeutic index for this class of com- γ National Center for Biotechnology Information (www.ncbi.nlm.nih.gov), (ii) pounds. Future screening against -secretase and PfSPP to identify PlasmoDB (http://plasmodb.org/plasmo), (iii)TriTrypDB(http://tritrypdb.org/ compounds with activity biased toward PfSPP would help reduce tritrypdb), and (iv)ToxoDB(http://toxodb.org/toxo). Database homology risks associated with inhibition of Notch processing and other ac- γ searching was performed by using OrthoMCL (www.orthomcl.org), BLASTP, tivities of -secretase. and profile hidden Markov algorithms (42). Full methods are described in SI Finally, our lead compound is lethal to chloroquine-resistant Materials and Methods. blood-stage P. falciparum, as well as liver-stage malaria parasites, with an IC50 against both stages that is comparable with only that of ACKNOWLEDGMENTS. We thank Aaron Gitler (Stanford University), the antimalarial drug atovaquone. This suggests that PfSPP repre- Michael Klemba (Virginia Polytechnic Institute and State University), and sents a causal prophylactic and transmission-blocking antimalarial Randall Pittman (University of Pennsylvania) for advice and reagents. This target. It is also highly potent against a spectrum of pathogenic work was supported by National Institutes of Health Grants T32AI007532 protozoan parasites. Thus, although further screening of drug-like (to M.B.H.) and 1R56AI081797-01 (to D.C.G.), the University of Pennsylvania molecules or medicinal chemistry optimization may be necessary Transdisciplinary Awards Program in Translational Medicine and Therapeu- fi tics Pilot Program (D.C.G.), the Penn Genome Frontiers Institute, Wellcome for species-speci c targeting and improved pharmacodynamic Trust Grant WT078285 (to D.C.G.), and support from the Medicines for parameters, inhibition of SPP may represent a valid pan-anti- Malaria Venture to the Genomics Institute of the Novartis Research Founda- protozoan drug discovery strategy. tion and the Novartis Institute for Tropical Diseases.

1. Dondorp AM, et al. (2011) The threat of artemisinin-resistant malaria. N Engl J Med 21. Stagg HR, et al. (2009) The TRC8 E3 ligase ubiquitinates MHC class I molecules before 365(12):1073–1075. dislocation from the ER. J Cell Biol 186(5):685–692. 2. Maier AG, Cooke BM, Cowman AF, Tilley L (2009) Malaria parasite proteins that re- 22. Lee SO, et al. (2010) Protein disulphide isomerase is required for signal peptide model the host erythrocyte. Nat Rev Microbiol 7(5):341–354. peptidase-mediated protein degradation. EMBO J 29(2):363–375. 3. Smith MH, Ploegh HL, Weissman JS (2011) Road to ruin: Targeting proteins for deg- 23. Bagola K, Mehnert M, Jarosch E, Sommer T (2011) Protein dislocation from the ER. radation in the endoplasmic reticulum. Science 334(6059):1086–1090. Biochim Biophys Acta 1808(3):925–936. 4. Walter P, Ron D (2011) The unfolded protein response: From stress pathway to ho- 24. Crawshaw SG, Martoglio B, Meacock SL, High S (2004) A misassembled trans- meostatic regulation. Science 334(6059):1081–1086. membrane domain of a polytopic protein associates with signal peptide peptidase. 5. Travers KJ, et al. (2000) Functional and genomic analyses reveal an essential co- Biochem J 384(pt 1):9–17. ordination between the unfolded protein response and ER-associated degradation. 25. Schrul B, Kapp K, Sinning I, Dobberstein B (2010) Signal peptide peptidase (SPP) as- Cell 101(3):249–258. sembles with substrates and misfolded membrane proteins into distinct oligomeric 6. Carvalho P, Goder V, Rapoport TA (2006) Distinct ubiquitin-ligase complexes define complexes. Biochem J 427(3):523–534. fi convergent pathways for the degradation of ER proteins. Cell 126(2):361–373. 26. Krawitz P, et al. (2005) Differential localization and identi cation of a critical as- 7. Christianson JC, et al. (2012) Defining human ERAD networks through an integrative partate suggest non-redundant proteolytic functions of the presenilin homologues – mapping strategy. Nat Cell Biol 14(1):93–105. SPPL2b and SPPL3. J Biol Chem 280(47):39515 39523. 8. Bozdech Z, et al. (2003) The transcriptome of the intraerythrocytic developmental 27. Casso DJ, Tanda S, Biehs B, Martoglio B, Kornberg TB (2005) Drosophila signal peptide – cycle of Plasmodium falciparum. PLoS Biol 1(1):E5. peptidase is an essential protease for larval development. Genetics 170(1):139 148. 9. Ward P, Equinet L, Packer J, Doerig C (2004) Protein kinases of the human malaria parasite 28. Marapana DS, et al. (2012) Malaria parasite signal peptide peptidase is an ER-resident fi – Plasmodium falciparum: The kinome of a divergent eukaryote. BMC Genomics 5:79. protease required for growth but not invasion. Traf c 13(11):1457 1465. 10. Fennell C, et al. (2009) PfeIK1, a eukaryotic initiation factor 2alpha kinase of the 29. Booth C, Koch GL (1989) Perturbation of cellular calcium induces of luminal ER proteins. Cell 59(4):729–737. human malaria parasite Plasmodium falciparum, regulates stress-response to amino- 30. Srivastava IK, Rottenberg H, Vaidya AB (1997) Atovaquone, a broad spectrum anti- acid starvation. Malar J 8:99. parasitic drug, collapses mitochondrial membrane potential in a malarial parasite. J 11. Zhang M, et al. (2012) PK4, a eukaryotic initiation factor 2α(eIF2α) kinase, is essential Biol Chem 272(7):3961–3966. for the development of the erythrocytic cycle of Plasmodium. Proc Natl Acad Sci USA 31. Tiwari S, Weissman AM (2001) Endoplasmic reticulum (ER)-associated degradation of 109(10):3956–3961. T cell receptor subunits. Involvement of ER-associated ubiquitin-conjugating 12. Gosline SJ, et al. (2011) Intracellular eukaryotic parasites have a distinct unfolded (E2s). J Biol Chem 276(19):16193–16200. protein response. PLoS ONE 6(4):e19118. 32. Zhong X, Pittman RN (2006) Ataxin-3 binds VCP/p97 and regulates retrotranslocation 13. Egorin MJ, et al. (2002) Pharmacokinetics, tissue distribution, and metabolism of 17- of ERAD substrates. Hum Mol Genet 15(16):2409–2420. (dimethylaminoethylamino)-17-demethoxygeldanamycin (NSC 707545) in CD2F1 mice 33. Li X, et al. (2009) Plasmodium falciparum signal peptide peptidase is a promising drug and Fischer 344 rats. Cancer Chemother Pharmacol 49(1):7–19. target against blood stage malaria. Biochem Biophys Res Commun 380(3):454–459. 14. Banumathy G, Singh V, Pavithra SR, Tatu U (2003) Heat shock protein 90 function is 34. Rottmann M, et al. (2010) Spiroindolones, a potent compound class for the treatment essential for Plasmodium falciparum growth in human erythrocytes. J Biol Chem of malaria. Science 329(5996):1175–1180. – 278(20):18336 18345. 35. Nzila A, Mwai L (2010) In vitro selection of Plasmodium falciparum drug-resistant 15. Pallavi R, et al. (2010) Heat shock protein 90 as a drug target against protozoan in- parasite lines. J Antimicrob Chemother 65(3):390–398. fections: Biochemical characterization of HSP90 from Plasmodium falciparum and 36. Balu B, Shoue DA, Fraser MJ, Jr., Adams JH (2005) High-efficiency transformation of Trypanosoma evansi and evaluation of its inhibitor as a candidate drug. J Biol Chem Plasmodium falciparum by the lepidopteran transposable element piggyBac. Proc – 285(49):37964 37975. Natl Acad Sci USA 102(45):16391–16396. fi 16. Hoffstrom BG, et al. (2010) Inhibitors of protein disul de isomerase suppress apo- 37. Meister S, et al. (2011) Imaging of Plasmodium liver stages to drive next-generation ptosis induced by misfolded proteins. Nat Chem Biol 6(12):900–906. antimalarial drug discovery. Science 334(6061):1372–1377. 17. Chou TF, et al. (2011) Reversible inhibitor of p97, DBeQ, impairs both ubiquitin-de- 38. Braakman I, Bulleid NJ (2011) and modification in the mammalian pendent and autophagic protein clearance pathways. Proc Natl Acad Sci USA 108(12): endoplasmic reticulum. Annu Rev Biochem 80:71–99. 4834–4839. 39. Wolfe MS (2008) Inhibition and modulation of gamma-secretase for Alzheimer’s 18. Chung DW, Ponts N, Prudhomme J, Rodrigues EM, Le Roch KG (2012) Characterization disease. Neurotherapeutics 5(3):391–398. of the ubiquitylating components of the human malaria parasite’s protein degra- 40. Weihofen A, et al. (2003) Targeting presenilin-type signal peptide dation pathway. PLoS ONE 7(8):e43477. peptidase with gamma-secretase inhibitors. J Biol Chem 278(19):16528–16533. 19. Weihofen A, Binns K, Lemberg MK, Ashman K, Martoglio B (2002) Identification of signal 41. Sato T, et al. (2006) Signal peptide peptidase: Biochemical properties and modulation peptide peptidase, a presenilin-type aspartic protease. Science 296(5576):2215–2218. by nonsteroidal antiinflammatory drugs. Biochemistry 45(28):8649–8656. 20. Loureiro J, et al. (2006) Signal peptide peptidase is required for dislocation from the 42. Finn RD, Clements J, Eddy SR (2011) HMMER Web server: Interactive sequence simi- endoplasmic reticulum. Nature 441(7095):894–897. larity searching. Nucleic Acids Res 39(Web server issue):W29–W37.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1216016110 Harbut et al. Downloaded by guest on October 2, 2021