Protein-Protein Interactions in Malaria: Emerging Arena for Future Chemotherapeutics

Malaria is one of the most deadly diseases infecting humans. Advances in elimination and vector control have reduced the global malaria burden in the past decade; however, the emerging threat of drug resistance and suboptimal vaccine efficacies threaten global eradication efforts. Unlocking novel drug and vaccine targets while simultaneously mitigating spread of resistant strains seems to be the need of the hour. Protein-protein interactions (PPIs), an integral part of host-pathogen cross-talk and parasite survival, have only recently emerged as promis-ing drug targets. Large PPI networks (interactome) are being developed to better our understanding of various parasite biochemical pathways. In this chapter, we throw light on several newly characterized protein-protein interactions between the host (humans) and parasite (plasmodium) in key processes such as hemoglobin degradation, enzyme regulation, protein export, egress, invasion, and drug resistance and further discuss their viability for development as novel chemotherapeutic targets.


Introduction
Malaria is one of the deadliest diseases to affect humans, with the latest WHO reports indicating ~445,000 deaths in 2017 alone [1]. More alarmingly, despite decades of advances in controlling the malaria epidemic, death rates caused by malaria seem to have plateaued in the past 3 years, indicating drug resistance and re-emergence. Drug resistance to the current frontline antimalarials have been confirmed by many recent studies and steadily observed over increasing geographic coverage [2]. Thus, it is of utmost importance to develop novel antimalarials, with different modes of action and distinct targets, if possible in conjunction, to check the onslaught of malaria. This chapter looks at such potential scope for antimalarial drug development: disruption of protein-protein interactions in the malaria parasite Plasmodium, crucial for survival and proliferation.

PPIs: the basics
Protein-protein interactions (PPIs) constitute the fundamental backbone required for occurrence of any biological event. They are defined as the residue level interactions between either the same protein (dimers, trimers, or other multimers) or diverse proteins (protein complexes). These basic interactions are necessary for a myriad of functions such as kinase signaling, receptor binding, proteolytic digestion, apoptosis regulation, and antigen-antibody interactions [3,4]. Disruptions in the protein interaction networks (PINs) as a result of PPI inhibition have been shown to cause several diseases where either single or multiple biochemical pathways are affected [5]. Owing to their fundamental roles in almost every process imaginable, PPIs have emerged as attractive therapeutic targets in several diseases. Several forms of cancer were also shown to have dysregulated protein interaction networks (PINs) [6]. Similarly, PPI disruptions have been observed in several autoimmune as well as parasitic diseases. Small peptides that infiltrate cellular defenses and specifically bind to target structures are already in development. Taken together, targeting PPIs though challenging can provide a novel understanding of biochemical processes as well as uncover new ways to combat diseases like malaria.
PPIs can be generally categorized into several groups depending on their function or the type of interactions. They include internal (hot-spots) or external (surface), obligate (permanent) or non-obligate (transient), stabilizing or destabilizing, ability to induce conformational changes in either of the partner molecules, peptide-protein or peptide-peptide interactions, and contiguous or discontiguous epitope binding [7] (Figure 1).
Some types of PPIs such as membrane PPIs can be difficult to characterize. While dedicated techniques like the split-ubiquitin membrane yeast two-hybrid (MYTH) system were developed to specifically detect membrane protein interactions [8], these techniques are still considered time-consuming and labor-intensive. Such bottlenecks make it hard to generate a complete picture of the membrane interactome. Even for reliable bioinformatic models for detection of membrane

PPIs in malaria
Malaria traditionally has been treated using inhibitors which target the broad spectrum proteasome offering several advantages as compared to specific protein inhibitors. Specific inhibitors had comparatively low efficacy in vivo. Also, inhibitors targeting a specific protein/ligand could potentially inhibit parasite growth only in stages when the target proteins are expressed. Broad spectrum antimalarials, such as the current frontline drugs artemisinins (ARTs) and their combination therapies (ACTs), target and break down various cellular pathways including but not limited to hemozoin formation, DNA repair, and mitochondria machinery, which make them highly potent within short exposure times [10,11]. However, exposure to various cellular targets leads to the rapid development of drug resistance. While resistance to chloroquine worldwide was observed after ~40 years of continued use, resistance to ARTs was achieved in a relatively short span of a decade, from its inception in late 1990s to the first reported resistance in 2008 [12]. While this rapid emergence resistance was partially attributed to suboptimal drug regimens and poor administrative practices, the same could be attributed to earlier drugs as well. Thus, compounds that are specific/flexible to the target protein are the need of the hour. This section deals with and summarizes current knowledge about crucial PPIs in various biochemical pathways of the malaria parasite Plasmodium falciparum.

Hemoglobin hydrolysis
Hemoglobin hydrolysis is one of the most targeted pathways for treatment as it is fundamental for parasite survival and involves numerous proteins [13]. Majority of earlier and currently used drugs disrupt multiple protein interactions. Several studies have been conducted recently that target individual PPIs and design inhibitors based on those interactions. Our lab has previously identified a "hot-spot" region in falcipains, the principal hemoglobinases of P. falciparum [14]. Falcipains contain a pro and a mature domain, with the pro domain bound to and blocking access to the active site in the mature domain. The interactions between these two domains, termed "hot-spot" interactions, dissociate under acidic conditions and are essential for hemoglobinase activity. This specific hot-spot was identified at the interface of pro and mature domains in falcipain-2 (FP2) and falcipain-3 (FP3). The study further demonstrated that synthetic compounds, NA01 and NA03, specifically bound to this hot-spot region and stabilized these interactions. Thus, even in the presence of an acidic environment (pH -5.5), pro-mature domains remained intact, rendering them inactive [15].
Falcipains, owing to their crucial role in Hb degradation, are considered as attractive chemotherapeutic targets. Several inhibitors were designed based on the interactions of the FP2 and the active site inhibitor E64. Molecular dynamics (MD) simulations indicated that two sets of residues, namely, recruiter groups A (rgA) and B (rgB) (rgA (D170, Q171, C168, G169, A151, and G230); rgB (K76, N77, and N81)) of FP2 are primarily involved in the initial binding with E64 about 80% and 14% of the time, respectively, before finally proceeding to bind with the active site residues [16]. Efforts elsewhere have focused on selective inhibition of falcipains rather than indiscriminate inhibition including its human host cathepsin isoforms.
While the S1 and S3 subsite residues have been conserved across Plasmodium and human cysteine proteases, the P2 residue in the S2 pocket in falcipains contained an acidic amino acid (Asp234 in FP2 and Glu243 in FP3) as compared to neutral residues in cathepsins (Leu209 in Cat-K, Ala215 in Cat-L, and Phe211 in Cat-S). A class of peptidomimetic amino-nitrile compounds known to inhibit cysteine proteases was further engineered to exploit the P2-region charge differences and inhibit falcipains specifically [17,18].
Falcipains also contain a domain at their C-terminal called the hemoglobin binding domain (Hb domain), a β-hairpin loop which protrudes away from the active site. Deletion of this 14 amino acid domain ablated the ability of falcipains to degrade Hb, thus indicating a necessary role of this domain in Hb capture prior to degradation at the active site [19]. Our lab recently published another study that identified crucial protease-substrate PPIs within this domain. A functionally conserved single amino acid position in both falcipains (Glu185 in FP2 and Asp194 in FP3) was found to be essential for Hb interactions, with activated falcipain mutants unable to degrade Hb even with accessibility to the active site. Molecular docking results indicated both the residues interacted with Hb-α as well as Hb-β subunits with interactions mediated primarily through this position (Figure 2). A specific inhibitor which could target this position could have potential applications in arresting the parasite hemoglobin degradome [20].
Hemoglobin degradation as a source for parasite growth was also shown to be dependent on the hemoglobin tetramer composition. Children (<5 years) have different Hb subunits (HbF, α 2 γ 2 ) as compared to adults (HbA, α 2 β 2 ), and malaria mortality rates have consistently indicated child mortality to be higher (61% of   [20]. The essential amino acid isoleucine (I) was found to be a main differentiating factor as it is absent in both α and β chains but present in γ subunit and makes up to 99% of encoded proteins in P. falciparum proteome [21]. Thus, compounds that interfered with the isoleucine acquiring pathways, such as targeting the two P. falciparum isoleucyl tRNA synthetases (IRSs), have been described. The compound thiaisoleucine, where γ-methylene of isoleucine was substituted by a sulfur atom, was found to potently inhibit cytosolic IRS, while another compound, mupirocin, a known inhibitor of methicillin-resistant S. aureus (MRSA) IRS, inhibited the apicoplast-localized IRS [22]. Of the 36 putative aminoacyl tRNA synthetases present in P. falciparum, nearly 5 have been targeted with potent inhibitors, and more are being tested [23,24].

Invasion/egress
Both invasion and egress are important events in the erythrocytic stage and are responsible for the malaria symptoms including chills and fever. The process of invasion requires a host of proteins to be secreted from its apical organelles including rhoptry bodies and micronemes, among others, and is precisely coordinated. The parasite initially aligns the merozoite apical region toward the host erythrocyte and forms a tight junction at the apex, progressing as the moving junction (MJ) pushes the parasite into the erythrocyte, with the erythrocyte surface forming a ring around the engorged parasite, which would later become the parasitophorous vacuolar membrane (PVM) [25].
While the process of moving junction (MJ) formation and important players involved in the process were well elucidated in P. falciparum, the other major causative agent of malaria, P. vivax, remains poorly understood, partially owing to difficulties in culturing P. vivax in vitro. Recombinant P. vivax rhoptry neck protein 2 (PvRON2), based on literature evidence of involvement of PfRON2 in MJ formation and invasion, showed that both rhoptry proteins PvRON2 and PvRON4 bound preferentially to CD71-labeled human reticulocytes rather than normocytes [26]. More importantly, the cysteine-rich C-terminal PvRON2 region strongly interacts with PvAMA1 domains II and III, similar to earlier reports of PfRON2-PfAMA1 interactions [27].
The P. falciparum genome encodes 10 aspartic proteases termed plasmepsins, a few of which including PMIX and PMX have remained functionally uncharacterized until recently. Conditional knockdown (KD) using TetR-aptamer regulators inserted at PMIX and PMX loci using CRISPR-Cas9 gene editing tools indicated a drastic decrease in parasite that could egress in PMIX-KD and role of PMX-KD in both invasion and egress. Further, PMX was found to be involved in the processing of a semi-proenzyme PfSUB1 to mature protein, a step crucial in both egress and invasion and also in the final processing step of a known egress protein, SERA5 [28]. Another study was simultaneously published that attempted to inhibit the activity of both PfPMIX and PfPMX. A hydroxyl-ethyl-amine (HEA) scaffoldbased compound, 49c, was shown to potently inhibit both the proteases in vitro and in vivo at nanomolar concentrations and reiterated the role of these two proteases in egress and invasion [29]. MD simulations indicated that the flap tip and hinge regions present in both the plasmepsins were very well stabilized by the rigid structure of compound 49c, leading to higher-binding free energy as compared to control plasmepsin inhibitor pepstatin [30].
For a Plasmodium parasite to successfully invade different types of cells at different life stages, it must display multiple families of receptor molecules on its surface to perform various functions such as gliding and traversal between hepatocyte cells or other motile functions. To successfully egress from erythrocytes, the parasite expresses a merozoite-thrombospondin-related anonymous protein (MTRAP) which interacts with the tetrameric glycolytic enzyme, fructose-1,6-bisphosphate aldolase, for powering the actomyosin motors required for movement [31]. However, continuous detachment and reattachment of these two enzymes are required for proper gliding motility. Therefore, compounds have been described that specifically stabilize MTRAP-aldolase interactions thereby rendering the parasite immotile. Of the 400 Medicines for Malaria Venture's (MMV) Malaria Box inhibitor library, a single compound (C4) was found to significantly inhibit hepatocyte invasion and led to abnormal gliding movements in a dose-dependent manner. A structure of C4 bound to the TRAP-aldolase complex could provide better insight into the key interactions of the inhibitor and help design more potent compounds [31].

Protein export
The protein export element (PEXEL) comprising of the conserved sequence RxLxE/Q/D is found in the N-terminus of ~300 proteins bound for export in the Plasmodium proteome. The aspartic protease plasmepsin V (PMV) was shown to be the sole protease involved in the cleavage of the PEXEL domain at the ER, specifically cleaving after the leucine residue (RxL ↓ xE/D/Q ). The first, third, and fifth conserved positions were thoroughly probed for role in protease activity and showed that the first Arg and third Leu are essential for PMV recognition and cleavage, while the fifth E/Q/D position was shown not to be essential for cleavage but for trafficking out of PVM [32]. Owing to the importance of PMV, peptidomimetic compounds that resembled PEXEL were developed. While statin-based compounds were the traditional inhibitors of aspartic proteases, inhibitors based on HEA moiety were also shown to be strong inhibitors of aspartic proteases. A compound named LG20 consisted of PEXEL-like motif R-L-[L-HEA-A]-E-A, where the L-HEA-A mimics the aspartic protease transition state, while HEA motif is proteolytically uncleavable by PMV. Overall, the compound successfully inhibited PMV activity with concentrations in the picomolar range [33].
PfEMP1 is one the most studied exported protein as it is one of the exported in abundant quantities to the outer surface and thus is an attractive target along with others such as circumsporozoite protein (CSP). Immunoprecipitation (IP) and mass spectrometry studies of a GFP-tagged minimal section of PfEMP1 (PfEMP1B) identified novel targets in different cellular components including parasitophorous vacuole (PV), Maurer's clefts, the plasmodium translocon of exported proteins (PTEX) translocon, and a novel exported protein-interacting complex (EPIC). Several new interacting partners including parasitophorous vacuole protein-1 (PV1), PV2, and exported protein-3 (EXP3) have been identified, all of whom localized to the newly described EPIC [34]. Finally, a comprehensive pathway of PfEMP1 export has been suggested, where PfEMP1 is initially translocated to the ER and then trafficked to PTEX machinery and out of PVM with the aid of EPIC, where finally it is received by host erythrocytic chaperonin complex, the TCP-1 ring complex (TRiC), and transported to erythrocyte surface [34].
IP assays coupled with truncated-construct interaction assays have helped identify few prominent PMV-partner PPIs. The P. falciparum ortholog of signal peptidase complex 25 (PfSPC25) was found to interact with PMV as well as PfSec61 and PfSec62 translocon. Together, the PfSPC25-PfSec61-PfSec62 interactome along with PMV was found to regulate the entry of the effector cargo into the ER [35]. Apart from export, protein-protein interactions during autophagy, where different cellular components are engulfed and fused with vacuoles or lysozymes, can be considered as important PPI targets. The P. falciparum autophagy-related protein 8 (PfAtg8) upon activation binds to PfAtg3 through a thioester bond, and this complex promotes the membrane assembly for the formation of autophagosomes. Solved crystal structure of the PfAtg3-PfAtg8 complex identified an additional region, called the A-loop, having very low sequence similarity to that of humans, thus providing a new avenue for drug development [36]. MD simulations revealed several crucial PPIs including H-bonds, van der Walls contacts, and other electrostatic interactions that are crucial for their interactions. Crucially, a peptidomimetic compound mimicking the PfAtg3 segment WLLP that interacts with PfAtg8 was identified to achieve maximum potency [37].

PPIs mediating drug resistance
The emerging threat of resistance to the current frontline drug artemisinin (ART), and its combination therapies (ACTs), is a cause of great concern. The mechanism of ART action, though generally agreed to generate free radical species which disrupt several essential pathways in the parasite, is highly debated. Immunoprecipitation studies with chemically tagged ART analogue (AP1) revealed that ART interacts with over a dozen proteins and is predominantly activated by free heme rather than free ferrous ions. P. falciparum Kelch 13 (PfK13) was identified as a key marker of ART resistance in several studies, most notably by Ariey et al., 2008 [38]. The studies have identified residues in the six Kelch propeller domains that are responsible for crucial protein-protein interactions, with the mutants failing to interact with its substrates P. falciparum phosphatidylinositol-3-phosphate (PfPI3P) [39]. PfPI3P has been shown to be involved in the unfolded protein response (UPR) pathway, thus helping the parasite cope with the free radical-induced stress (Figure 3). Localization studies indicated the PfK13 to co-localize with the P. falciparum erythrocyte membrane protein (PfEMP1) and majorly concentrated at various proteostasis system in the parasite including ER, cytoplasm, and the UPR. Specifically, PfK13 mutants were shown to elevate the PI3P-vesicle production and amplification, and their export throughout the parasite and into the infected erythrocyte [40].
Structural analysis and MD simulations of PfK13 indicated the presence of two evolutionarily highly conserved domains in the broad-complex, tramtrack, and bric à brac (BTB) domain and in the shallow binding pocket formed by the six propeller domain repeats. These domains displayed a different electrostatic surface potential unlike the rest of the bottom PfK13 face and are rich in highly conserved arginine and serine residues. While the BTB domain was shown to be involved in the recruitment of a scaffold protein Cullin, the propeller domain pocket binds to the substrate molecules for further ubiquitination. MD simulations showed that the validated PfK13 markers such as C580Y and R539T mutants induced a significant structural destabilization in the shallow pocket region as compared to wild type while maintaining the overall structural integrity. Specifically, the C580Y mutant disrupted a disulfide bridge (C532-C580) and additional H-bonds created with neighboring residues not present in wild type strains. In the case of R539T mutant, it was shown to have substantially less H-bond interactions along with a complete loss of a salt bridge interaction with E606 while also losing another inter-blade H-bond interaction [41]. Overall, these mutants lead to significantly diminished levels of functional protein which cause diminished substrate interaction, ultimately promoting the PfPI3P-mediated unfolded protein response pathway. Our own group, through Co-IP and mass spectrometric studies, has identified several novel proteins including Trx-like mero protein (PF3D7_1104400), pyridoxal kinase (PF3D7_0616000), trafficking protein particle complex subunit 3 (TRAPP), and putative (PF3D7_0418500) that interact with PfK13 and could potentially have a role in stress-mediated response (Atul et al., 2019, unpublished).

PPI inhibition: peptidomimetics and design
While several strategies have been employed for PPI inhibition, bioinformatics-based drug design has been at the forefront to design specific PPI inhibitors. Structure-based drug design, where the solved structure of an enzyme in complex either with an inhibitor or a natural substrate was used to design inhibitors, was popularized in the 1990s. However, rational design for PPI inhibitors needs to overcome some common hurdles such as low proteolytic stability, analyzing extensive libraries of candidate molecules, and in some cases low ligand efficiency when compared to standard active site inhibitors. Thus, several strategies are employed in modern drug synthesis to overcome these problems. Short linear peptides tend to have lower conformational stability; thus cyclization of  the peptide is preferred which rigidifies the structure in an active configuration. Certain modifications in the backbone of the peptide such as backbone extension, side-chain shifting to nitrogen atoms (peptoids), and altering the stereochemistry can also be applied. Peptoids can easily fold into helices or other structures as they consist of repeated nitrogen-substituted glycine units that give an added advantage of mimicking the peptide structure and function. Stereochemistry of a compound can be changed by using D-peptides instead of L-peptides as they are more susceptible to proteolytic degradation and are one of the most common strategies to develop potent bioactive compounds. Another modification involves using β-peptides; peptides with amino group bound at β-carbon instead of α-carbon for each amino acid, often called as foldamers, could confer additional proteolytic stability both in vitro and in vivo. A class of oligopeptides (<80 residues in length) called as miniproteins can also be utilized as they have a rigid, well-defined three-dimensional structure. The 19 kDa fragment of merozoite surface protein 1 (MSP1), the fragment that is finally displayed on the merozoite surface after several processing steps, was fused along with a glycosylphosphatidylinositol (GPI) tag used to create a miniprotein, which was successfully targeted by antibodies specific to the miniprotein and inhibited erythrocyte invasion [42]. Substantial effort has been made to develop rational strategies in designing PPI inhibitors for target proteins that have no well-defined binding site (so-called "hot-spot") and, thus, have previously been considered undruggable (Figure 4).   [44,45].
As described in the earlier section, Villa et al. designed and synthesized peptidomimetics belonging to 1,2,3-triazoles, specifically 1,4-disubstituted 1,2,3-triazoles. These compounds mimicked the contacts made by PfAtg3 template structure containing the residues W-L-L-P, as this template was shown to have the majority of interactions with PfAtg8. Of the four compounds synthesized, compound 2 (C2) exhibited prominent inhibition (IC 50 -3.8 μM) in vitro, while C1 had better inhibitory effects in vivo [37].
Natural inhibitors of proteases are one of the best studies substrate groups in malaria as they are highly specific, stable, and reversible. Plasmodium spp. contain inhibitors of cysteine proteases (ICPs) and endogenous macromolecular inhibitors, which regulate the activity of cysteine proteases. Orthologs of ICPs are also observed in other apicomplexan groups such as Trypanosoma cruzi, whose ICP is called as chagasin [43]. Chagasin was shown to potently bind to the active site of both falcipains, FP2 and FP3. More importantly, three loop regions termed BC-, DE-, and FG loops were shown to be involved in active site inhibition (Figure 5A,B) [44]. However, further studies with the P. falciparum ICP, falstatin, indicated that unlike other ICPs, just the BC loop was sufficient for falcipain inhibition and that Asn289 of falstatin formed stabilizing hydrogen bonds and hydrophobic interactions (Figure 5C,D) [45].

Conclusion
Protein-protein interactions play roles of utmost importance in the growth and survival of any organism. Thus, focused targeting of such interactions specific to parasite can help produce robust and effective drugs. Recent research has indicated a renewed interest in targeting PPIs in the field of malaria. Various PPIs in pathways essential for parasite survival, erythrocyte invasion/egress, drug resistance, and others have been elucidated. These new classes of peptidomimetic compounds would form the future defense against an ever-increasing resistant parasite threat. Targeting PPIs offers several advantages over active site inhibition as 'hot-spots' are more flexible as compared to the active site and thus can be more selective in terms of drug interactions. In contrary to active site, the interactions at allosteric sites and exosites in an enzyme occur away from the active site; thus they tend to fall under less drug pressure and are less likely to develop resistance.