Oligomeric protein interference validates druggability of aspartate interconversion in Plasmodium falciparum

Abstract The appearance of multi‐drug resistant strains of malaria poses a major challenge to human health and validated drug targets are urgently required. To define a protein's function in vivo and thereby validate it as a drug target, highly specific tools are required that modify protein function with minimal cross‐reactivity. While modern genetic approaches often offer the desired level of target specificity, applying these techniques is frequently challenging—particularly in the most dangerous malaria parasite, Plasmodium falciparum. Our hypothesis is that such challenges can be addressed by incorporating mutant proteins within oligomeric protein complexes of the target organism in vivo. In this manuscript, we provide data to support our hypothesis by demonstrating that recombinant expression of mutant proteins within P. falciparum leverages the native protein oligomeric state to influence protein function in vivo, thereby providing a rapid validation of potential drug targets. Our data show that interference with aspartate metabolism in vivo leads to a significant hindrance in parasite survival and strongly suggest that enzymes integral to aspartate metabolism are promising targets for the discovery of novel antimalarials.


| INTRODUC TI ON
The parasite Plasmodium falciparum is responsible for the most lethal form of human malaria (World Health Organization, 2017). The spreading of P. falciparum in the human host depends on the availability of specific metabolites during its blood stage (Kirk & Saliba, 2007;Lindner, Meissner, Schettert, & Wrenger, 2013). The metabolism of these external nutrients represents a key-step for parasite proliferation, and it is believed to be essential for its survivability.
Although these metabolic steps would open new avenues for drug discovering targeting P. falciparum, the validation of these metabolic steps remains challenging due to limitations of applicability of probe techniques in P. falciparum and dependence upon reverse genetics (Meissner et al., 2016). This highlights the necessity of development of novel validation techniques capable of simplifying the validation of potential targets in P. falciparum-as well as other parasitic organisms. The examination of interaction surfaces between subunits of oligomeric proteins might offer a relatively straightforward alternative to this process (Meissner et al., 2016).
Protein oligomerization, the assembly of two or more copies of a single protein into one object, is a feature shared by all organisms and is present in more than 60% of all protein structures currently available within the Protein Data Bank (PDB; Hashimoto, Nishi, Bryant, & Panchenko, 2011). The biological importance, physicochemical properties, and evolutionary aspects of protein oligomerization have been recently summarized (Hashimoto & Panchenko, 2010;Hashimoto et al., 2011;Nishi, Hashimoto, Madej, & Panchenko, 2013). Furthermore, lower degree of evolutionary conservation of the oligomeric interfaces (Caffrey, Somaroo, Hughes, Mintseris, & Huang, 2004;Valdar & Thornton, 2001), as well as high specificity and binding affinity between the cognate partners, could successfully be utilized in drug target validation (Lunev et al., 2018). Based on these features, we hypothesized that the introduction of functionally incompetent forms of an enzyme into the native oligomeric assembly could be exploited in the analysis of biochemical pathways in vivo, particularly in cases where standard techniques (e.g., RNAi/knock in/out) have a low success rate.
The aspartate metabolism pathway within P. falciparum contains a number of oligomeric enzymes making it an ideal system to test our hypothesis that oligomeric self-assembly can be used to modulate in vivo behavior. Aspartate interconversion is essential for nitrogen metabolism of all organisms. In Plasmodium species, it was also proposed to play a key role in de novo pyrimidine biosynthesis as well as energy metabolism (Wrenger et al., 2011). Plasmodial aspartate aminotransferase (PfAspAT) and malate dehydrogenase (PfMDH) catalyzes the reversible reaction from aspartate + 2-oxoglutarate to oxaloacetate + glutamate and malate + NAD to oxaloacetate + NADH, respectively. The crystal structure of PfAspAT has been previously solved (Wrenger et al., 2011). As reported, PfAspAT is a homo-dimer with a molecular weight of 48.42 kDa per monomer (Wrenger et al., 2011). Similarly to the previously described AspAT of E. coli (Jäger, Moser, Sauder, & Jansonius, 1994), each subunit consists of three major domains: an N-terminal arm (residues 1-14), a large coenzyme-binding domain (residues 36-321) and a smaller domain (residues 15-36 and 322-404). The Nterminal arm domain (residues 1-14) distal from either active site is thought to stabilize the PfAspAT dimer and is necessary for activity, as truncated species lacking the N-terminal extension showed significantly reduced activity while retaining dimeric structure (Wrenger et al., 2011). The two active sites of PfAspAT are formed in a cleft between the big and small domains near the oligomeric interface and each active site pocket is composed of residues contributed from both subunits. Previous experiments where PfAspAT was selectively inhibited in vitro using a polypeptide chain consisting of first 50 PfAspAT amino acids (Wrenger et al., 2011) confirm the hypothesis that oligomeric interfaces show significantly higher sequence divergence amongst homologs and thus offer potential in specific interference with a target protein.
The structure PfMDH has also been recently solved, and we have provided an insight into the role of oligomeric assembly in the regulation of PfMDH activity (Lunev et al., 2018). PfMDH possesses a tetrameric conformation where each monomer is comprised of 326 residues and is composed of two major domains: an N-terminal cofactor-binding domain containing a parallel structure of six betasheets (Rossmann-fold) and C-terminal substrate-binding domain.
Our previous results suggested that a correctly formed tetrameric assembly of PfMDH is essential for activity (Lunev et al., 2018).
Indeed, the introduction of a tryptophan residue at one of the interfaces facilitating oligomerization (V190W) results in disruption of the tetramer, breaking it down into two dimers, and a significant reduction in activity. Co-purification and western blot experiments with mixed lysates of recombinantly expressed wild-type PfMDH (Streptagged) and PfMDH-V190W (His 6 -tagged) mutant, with a predicted molecular mass of 35.28 and 36.74 kDa for each monomer, respectively, demonstrated that PfMDH-V190W was able to insert itself into a pre-formed wild-type PfMDH assembly (Lunev et al., 2018). As shown by subsequent activity assays, the isolated wild-type:V190W chimera possessed no detectable activity in either direction, while recombinant wild-type PfMDH displayed both reductive and oxidative activity. These data demonstrate that recombinant mutants can be used as specific modifiers of wild-type PfMDH activity in vitro, offering the potential to validate it as a drug target, without recourse to complex genetics or initial tool compounds that may display significant off-target effects (Lunev et al., 2018). However, neither PfAspAT nor PfMDH has as yet been validated as a drug target in vivo.
In this study, structural information of the enzymes PfMDH and PfAspAT was used to generate mutants for use in in vivo protein interference experiments following two different approaches.
In the first approach, we designed mutants that would incorporate within the native oligomer and disrupt the native oligomeric state, thereby inhibiting the function of the native assembly. In a second complementary approach, a mutant was designed to incorporate within the native oligomer and inhibit its function, without disruption to the native oligomeric state. In both approaches, the activity of the target enzymes was determined within the lysate of transgenic parasites, suggesting successful incorporation of the mutants within the targeted assemblies. The resulting data clearly indicate a significant dependence of the parasite on functional aspartate metabolism. Our data also provide proof-of-principle for protein interference assay (PIA) as a general approach to the use of oligomeric assemblies to obtain functional data in vivo.

| RE SULTS
2.1 | The oligomeric interface of PfAspAT shows higher sequence diversity than its active site BLAST (Altschul, Gish, Miller, Myers, & Lipman, 1990) analysis of the close homologs of PfAspAT showed overall 38.7% sequence conservation, while the residues comprising the active site of PfAspAT showed a sequence conservation of 100% (Table 1; Figure 1). PISA (Krissinel & Henrick, 2007) analysis of the structural assembly of PfAspAT identified 98 residues involved in the inter-oligomeric contact, showing sequence conservation of 34.7% (Table 1), where 10.2% accounts for the active site residues.

| Point mutations of the key active site residues abolish catalytic activity of PfAspAT in vitro while not disturbing the dimerization and overall fold
Based on the crystal structure of PfAspAT (3K7Y; Wrenger et al., 2011), point mutations were designed in order to interfere with the catalytic TA B L E 1 Sequence conservation across the different oligomeric interfaces of PfAspAT. Sequence conservation amongst close homologs (identity above 28%) was analyzed using BLAST (Altschul et al., 1990). Analysis of the residues supporting the oligomeric contact was performed using PISA (Krissinel & Henrick, 2007) (Krissinel & Henrick, 2007). (c) and (e) (side view) Evolutional diversity of the residues involved in the oligomeric contact: absolutely conserved (red), strongly conserved (orange) and slightly conserved (pale yellow). Sequence conservation amongst close homologs (identity over 28%) was analyzed using BLAST (Altschul et al., 1990). This represents an approx. 170-fold reduced catalytic activity of mutant ( Figure 5). These data suggest that the introduction of two-point mutations Y68A and R257A would have the desired inactivation effect on PfAspAT while not affecting its folding or ability to form dimers.

| Inactivated PfAspAT mutant copies can be incorporated into the native assembly during recombinant expression in E. coli
Assuming that double mutation Y68A/R257A affects both active sites of one PfAspAT dimer, we further hypothesized that formation of PfAspAT wild-type/mutant chimera would also affect both active sites and result in a significantly less active enzyme.
The PfAspAT-Y68A/R257A gene was sub-cloned into the pBM1 vector using sequence-specific primers (Appendix 1) and the Further activity measurements indicated that Strep-purified sample showed reduced activity compared to the wild type; while no activity could be detected from the His 6 -purified sample ( Figure 5c).
These data indicate that single His 6 -purification of the wild-type:mutant PfAspAT co-expression product is able to isolate a chimeric oligomer consisting of the His 6 -tagged PfAspAT-Y68A/R257A mutant and Strep-tagged wild type. The lack of detectable activity of the purified chimera, confirms the hypothesis that a mutant copy (Y68A/R257A) can be introduced into the native PfAspAT dimeric assembly through co-expression resulting in an inactivated chimeric protein.

| Introduction of PfAspAT and PfMDH mutants results in a significant reduction in parasitaemia in aspartate-limited culture media
The cytosolic localization of PfAspAT within the parasite has previously been reported (Wrenger et al., 2011). Similarly, the Furthermore, the protein production within the transgenic parasites was visualized by western blotting from parasite lysate ( Figure 9).

| Activity measurements of PfMDH and PfAspAT in parasite lysates confirm the formation of the heterocomplexes
In order to confirm the activity profile of both enzymes in the respective transfected parasites, we performed specific activity assays using parasite extracts ( Figure 10). As expected, proteins extracted from parasites transfected with plasmids encoding additional copies of wild-type PfMDH or PfAspAT genes presented a statistically higher enzymatic activity (100.41 ± 1.05 and 8.83 ± 0.28 mU/mg, respectively) compared to non-transfected parasites (3D7), confirming that both enzymes were not only being overexpressed ( of the tetramer into a pair of dimers but it also made the re-formation of the tetramer highly unlikely due to the introduction of molecular clashes. It was also previously reported that the mutated PfMDH-V190W was able to incorporate into the native PfMDH-WT assembly in vitro, disturbing the native oligomeric state of the target protein as well as inhibiting its activity (Lunev et al., 2018). In this work, we also demonstrate that the hypothesized PfMDH chimeras are formed inside the parasite in vivo, through measurement of PfMDH activity in the lysate of transgenic parasites.
We performed similar experiments with PfAspAT. In contrast with PfMDH where the oligomeric state is disturbed, in our PfAspAT mu-

tant, the native oligomeric state is maintained. Incorporation of mutant
PfAspAT showed that the enzyme's native oligomeric assembly could also be utilized in order to target the wild-type protein and show an effect on activity in vitro ( Figure 5c) and in vivo ( Figure 10). The inactive PfAspAT-Y68A/R257A mutant was shown to be able to incorporate into native PfAspAT-WT dimeric assembly during recombinant co-expression, as confirmed by western blot (Figure 5b), resulting in complete loss of activity (Figure 5c). The mutated PfAspAT protein was then expressed in P. falciparum blood-stage cultures, and we have demonstrated that the hypothesized PfAspAT chimeras are formed inside the parasite in vivo, through measurement of PfAspAT activity in the lysate of transgenic parasites. While no effect is seen in aspartate-rich media, the introduction of both mutant proteins in aspartate-limited media results in a clear phenotype (Figure 7), without recourse to complex genetic manipulations. We have termed this approach the oligomeric protein interference assay (PIA; Meissner et al., 2016).
Our data show that oligomeric surfaces can be used to specifically inhibit protein activity in vivo, especially in cases when opportunities for genetic manipulation are limited. The introduction of PfAspAT-Y68A/R257A or PfMDH-V190W alone did not result in any significant effect on parasite proliferation in blood-stage cultures (Figure 7a,b). Nonetheless, we believe that these findings demonstrate that the expression of mutant proteins in cultured parasites has overall no negative effects on parasite growth. This also indicates that, while the expression levels of the mutant proteins are significantly higher (up to fourfold) than their endogenous counterparts, they are likely not overexpressed at a level that would induce metabolic stress on P. falciparum through depletion of amino acids.
As the mutant proteins introduced are designed and demonstrated to be inactive, it is also highly unlikely that we have significantly (c) While no effect of double transfection with PfMDH-V190W and PfAspAT-Y68A/R257A was observed in aspartate-rich media, the parasite's viability was significantly hampered in the aspartate-limited culture (p < o.oo1). Two-way ANOVA analysis was performed using <>GraphPad Prism 5.0 Müller et al., 2009). The lack of negative effect on the proliferation of culture parasites is supported by the measurements of specific activity of both enzymes in lysates of parasites transfected with the mutant genes, which show that these parasites retain a partial activity of both enzymes (Figure 10a,b). Additionally, continuous expression of the inserted mutant reduces the risk of possible degradation of the mutant protein by cellular proteases before it reaches the intended targets. Further, our mutations were designed to result in inactive proteins (confirmed by our in vitro activity assays (Lunev et al., 2018), Figure 3c). In fact, the activity assay in lysates of parasites supports not only the hypothesis that the introduction of mutated proteins will cause a decrease in the activity, but also indicates that the formation of heterocomplexes occurs in vivo, thereby influencing the function of the native protein.
Finally, in contrast to current small molecule inhibition approaches, no limitations regarding drug or compound delivery to the cytosol are encountered as the mutant proteins are expressed directly within the parasite. Based on this analysis, we believe that the introduction of inactive proteins specific for oligomeric targets represents a minimally perturbing method to specifically inhibit metabolic pathways of interest in the human malaria parasite P. falciparum that will have a minimal off-target effect, and, in this manner, offers a possible tool to be used in the validation of target candidates.

While the insertion of the individual mutations of PfMDH or
PfAspAT results in no negative effect on parasitemia, the transfection of parasites with both plasmids results in a significant reduction in parasite proliferation in aspartate-limited media (Figure 7c). This is a clear change in phenotypic behavior that has been generated without recourse to complex genetic approaches. These data strongly suggest that while the correct function of either PfMDH or PfAspAT is sufficient to support parasite proliferation during the blood stage, simultaneous inhibition of both results in a significant reduction in parasite growth. While our data support the concept of oligomeric protein interference assays for cytosolic proteins, in principle the use of the native targeting sequence will also allow the approach to be used for proteins present in other cellular compartments (mitochondrial, apicoplast, membrane inserted, etc.).
A number of recent manuscripts have focused on elucidating the role of the mitochondria of the malarial parasite (Ke et al., 2015;Nixon et al., 2013). Amongst the pathways supported by the mitochondria are those involved in the biosynthesis of aspartate. While aspartate is essential in the formation of new proteins, it is also a key precursor in pyrimidine biosynthesis (Cassera, Zhang, Hazleton, & Schramm, 2011;Hyde, 2007), another promising pathway for drug discovery, as confirmed by validation of dihydroorotate dehydrogenase (DHODH) as a drug target (Vyas & Ghate, 2011). During proliferation, the malaria parasite catabolizes hemoglobin as an amino acid source. Although aspartate is available in hemoglobin, the host cell protein does not sufficiently suit the need of P. falciparum for the rapid proliferation within blood stage as demonstrated by the presented growth experiments of blood-stage cultures in aspartate-limited media. Thus, aspartate biosynthesis or uptake is highly likely to be a key element in supporting the rapid proliferation of P. falciparum in human red blood cells. It has long been known that aspartate is the least common of all the amino acids available within the human serum, with recent measurements suggesting that the concentration of L-aspartate in human sera is <20 µM (Psychogios et al., 2011). This strongly suggests that biosynthetic pathways would be the main source of the aspartate required by P. falciparum in the human host ( Figure 11).
The significant drop in proliferation upon oligomeric-based inhibition of both PfMDH and PfAspAT in aspartate-limited cell culture media suggest that aspartate biosynthesis in the malarial parasite depends upon the function of both of these enzymes and validates this metabolic pathway as drug target in P. falciparum. This is in accord with recent data that suggest the products of glycolysis (both pre-and post-mitochondrial) are used in biosynthesis in the malarial parasite (Ke et al., 2015). In another recent study (Zhang et al., 2018), mutagenic index scores (MISs) calculations have predicted PfMDH as an essential enzyme. In contrast, the PIA experiments revealed no effect on the proliferation of MDH mutant overexpressing parasites. However, this down-regulation of intra-cellular MDH activity will not warrant a total "knock-down" of the plasmodial MDH due to the presence of residual endogenous wild-type MDH. On the other hand, the MIS approach suggested PfAspAT as non-essential, corroborating with our data, which shows no effect on the knock-down of PfAspAT alone although an intra-cellular simultaneous over-expression of PfAspAT and PfMDH mutated proteins clearly causing a severe growth defect. While the simultaneous inhibition of two enzymes may be highly challenging for the development of novel antimalarials, our data strongly suggest that future drug targets to treat malaria infection may be found within downstream components of the aspartate metabolism pathway (Figure 11). Taken together, our data also show that oligomeric surfaces offer a highly promising opportunity to specifically influence protein behavior in vivo and offer a novel avenue in the validation of pathways for downstream drug development, particularly in the field of infectious diseases.

| Expression and purification of recombinant PfAspAT
The purification of PfAspAT has been previously reported (Jain et al., 2010). Wild-type PfAspAT open reading frame (ORF) was cloned into pASK-IBA3 expression plasmid with additional C-terminal His 6 -tag to facilitate purification via Ni-NTA chromatography. The generated pASK-IBA3-PfAspAT plasmid was transformed into the commercially F I G U R E 1 0 Specific activity values of PfMDH (a) and (b) PfAspAT, respectively, measured from Plasmodium falciparum cell lysates. The activity of the wild-type MDH (a) was significantly higher compared to the V190W mutant (p < 0.05). The activity of AspAT-WT (b) was significantly higher compared to both control (3D7) and double mutant (p < 0.05). Mutant PfAspAT transfectant showed lower activity compared to the control (3D7; p < 0.05). The activity was measured in triplicates, in two independent experiments. GraphPad Prism 5.0 was used for one-way ANOVA analysis F I G U R E 11 Importance of the aspartate metabolism in Plasmodium falciparum. Our experiments showed that inhibition of the de-novo aspartate biosynthesis via PfMDH and PfAspAT is a viable target for future antimalarial drug design available E. coli expression strain BLR (DE3; Novagen) for expression.
After expression, the cells were lysed using sonication and centrifuged to separate the lysate. Soluble His-tagged PfAspAT was purified using Ni-NTA agarose (Quiagen) according to the manufacturer's recommendations. PfAspAT was further purified via size-exclusion chromatography using HiLoad 16/60 Superdex S75 column (GE Healthcare).

| Site-directed mutagenesis
The single mutants PfAspAT-Y68A and PfAspAT-R257A, and the double mutant PfAspAT-Y68A/R257A were generated via site-directed mutagenesis using specific oligonucleotides containing the altered codons (Appendix 1) and the pASK-IBA3-PfAspAT plasmid (Jain et al., 2010) as a template. All constructs were verified by Sanger sequencing.

| Determination of oligomeric state
The analysis of the oligomeric state of recombinant WT PfAspAT and mutants was performed according to the previously described protocol (Wrenger et al., 2011).

| In vitro activity assays
The specific activity of the double mutant of PfAspAT-Y68A/R257A was measured following the same procedure as previously reported (Wrenger et al., 2011). The effect of incorporation of mutant His 6 -tagged PfAspAT-Y68A/R257A into the native Strep-tagged wild-type assembly was also analyzed using the samples of the coexpression experiments (described above).

| Cloning and transfection of PfMDH V190W and PfAspAT Y68A, R257A
In order to obtain transgenic parasites, the ORFs of WT-PfAspAT and PfAspAT-Y68A/R257A were amplified via PCR using sequence-specific primers (Appendix 1) and subsequently cloned into pARL 1a-with the hDHFR (human dihydrofolate reductase) resistance cassette (Wrenger & Müller, 2004). The resulting plasmids encoded for the full-length WT-PfAspAT and PfAspAT-Y68A/ R257A mutant with an additional C-terminal Strep-tag followed by the stop-codon before the GFP gene encoded on pARL 1a-.

| qRT and western blot
An asynchronous culture of transgenic 3D7 parasites was isolated via saponin lysis. The total RNA of these parasites was extracted using TRIZOL following the manufacturer's instruction. The cDNA synthesis was performed as described in (Butzloff, 2013;Chan et al., 2013;Knöckel et al., 2012;Müller et al., 2010 (Salanti et al., 2003) and the respective MOCK cell line like (Chan et al., 2013). The data were analyzed using the Corbett Rotor-Gene 6.1.81 software and the 2 −ΔΔCt method (Livak & Schmittgen, 2001).
The protein expression of the transgenic cell lines was verified via western blot analysis as described in (Knöckel et al., 2012). Briefly, isolated parasites were resuspended in 5x SDS-PAGE sample buffer, boiled for 5 min at 95°C and centrifuged for 5 min at 14,000 ×g. The supernatant was separated by 10% SDS-PAGE and transferred on a nitrocellulose membrane (Bio-Rad). The expressed proteins were detected via their Strep-or His 6 -tag by using a monoclonal anti Strepor anti-His antibody (1:5,000 dilution (IBA; Pierce) and a secondary anti-mouse HRP-labeled antibody (1:10,000 dilution; Pierce) and visualized on X-ray films using the SuperSignal West Pico detection system (Thermo Scientific).

| Activity assays from parasites lysate
In order to analyze the specific activity of PfAspAT and PfMDH, Bradford assay (Bradford, 1976).

| Proliferation assays
In order to analyze the long-term influence of the overexpressing cell lines pARL-PfAspAT-Y68A/R257A-Strep-hDHFR, pARL-PfMDH-V190W-his-BSD and the double transgenic cell line pARL-PfAs-pAT-Y68A/R257A-strep-hDHFR + pARL-PfMDH-V190W-his-BSD in comparison with their respective MOCK line, parasites growth was monitored over several days. The parasites were synchronized using sorbitol and a starting parasitaemia of 1% of ring-stage iRBC (infected red blood cells) was adjusted. Giemsa-stained thin smears were analyzed daily and the parasitaemia was determined by light microscopy in percentage of iRBC to total RBC. Cultures with more than 8% of iRBC were diluted and cumulative parasitaemias were calculated as described in (Knöckel et al., 2012). Triple repetition of the proliferation assay was performed and the growth curves were generated with GraphPad Prism 4.0. The slope of the respective curves was calculated through an exponential equation (Müller et al., 2009).

| Localization of PfMDH
The ORF of PfMDH-WT with no stop-codon at C-terminus was amplified by PCR and subsequently cloned into pARL 1a-using KpnI/AvrII restriction enzymes (Appendix 1). The resulting plasmid encoded for the wild-type PfMDH fused in front of the GFP gene and was transfected into P. falciparum 3D7 parasites as described above. The localization of the MDH-GFP chimera was analyzed via live cell fluorescent microscopy  using an Axio Imager M2 microscope (Zeiss) equipped with an AxioCam HRC digital camera (Zeiss). In order to visualize the nucleus, the parasites were incubated with 10 µg/ml HOECHST 33342 dye (Invitrogen).
The images were analyzed with the AxioVision 4.8 software.

ACK N OWLED G EM ENTS
The authors would like to acknowledge the Fundação de Amparo à

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

E TH I C S S TATEM ENT
None required.

DATA ACCE SS I B I LIT Y
All data supporting this study are provided in full in the results section of this paper.