Mitochondrial type II NADH dehydrogenase of Plasmodium falciparum is dispensable and not the functional target of putative NDH2 quinolone inhibitors

The battle against malaria has been substantially impeded by the recurrence of drug resistance in Plasmodium falciparum, the deadliest human malaria parasite. To counter the problem, novel antimalarial drugs are urgently needed, especially those that target unique pathways of the parasite, since they are less likely to have side effects. The mitochondrial type II NADH dehydrogenase of P. falciparum, PfNDH2 (PF3D7_0915000), has been considered a good prospective antimalarial drug target for over a decade, since malaria parasites lack the conventional multi-subunit NADH dehydrogenase, or Complex I, present in the mammalian mitochondrial electron transport chain (mtETC). Instead, Plasmodium parasites contain a single subunit NDH2, which lacks proton pumping activity and is absent in humans. A significant amount of effort has been expended to develop PfNDH2 specific inhibitors, yet the essentiality of PfNDH2 has not been convincingly verified. Herein, we knocked out PfNDH2 in P. falciparum via a CRISPR/Cas9 mediated approach. Deletion of PfNDH2 does not alter the parasite’s susceptibility to multiple mtETC inhibitors, including atovaquone and ELQ-300. We also show that the antimalarial activity of the fungal NDH2 inhibitor HDQ and its new derivative CK-2-68 is due to inhibition of the parasite cytochrome bc1 complex rather than PfNDH2. These compounds directly inhibit the ubiquinol-cytochrome c reductase activity of the malarial bc1 complex. Our results call into question the validity of PfNDH2 as an antimalarial drug target. Importance For a long time, PfNDH2 has been considered an attractive antimalarial drug target. However, the conclusion that PfNDH2 is essential was based on preliminary and incomplete data. Here we generate a PfNDH2 KO (knockout) parasite in the blood stages of Plasmodium falciparum, showing that the gene is not essential. We also show that previously reported PfNDH2-specific inhibitors kill the parasites primarily via targeting the cytochrome bc1 complex, not PfNDH2. Overall, we provide genetic and biochemical data that help to resolve a long-debated issue in the field regarding the potential of PfNDH2 as an antimalarial drug target.


Abstract 23
The battle against malaria has been substantially impeded by the recurrence of drug resistance in Plasmodium falciparum, 24 the deadliest human malaria parasite. To counter the problem, novel antimalarial drugs are urgently needed, especially 25 those that target unique pathways of the parasite, since they are less likely to have side effects. The mitochondrial type II 26 NADH dehydrogenase of P. falciparum, PfNDH2 (PF3D7_0915000), has been considered a good prospective 27 antimalarial drug target for over a decade, since malaria parasites lack the conventional multi-subunit NADH 28 dehydrogenase, or Complex I, present in the mammalian mitochondrial electron transport chain (mtETC). Instead, 29 Plasmodium parasites contain a single subunit NDH2, which lacks proton pumping activity and is absent in humans. A 30 significant amount of effort has been expended to develop PfNDH2 specific inhibitors, yet the essentiality of PfNDH2 has 31 not been convincingly verified. Herein, we knocked out PfNDH2 in P. falciparum via a CRISPR/Cas9 mediated 32 approach. Deletion of PfNDH2 does not alter the parasite's susceptibility to multiple mtETC inhibitors, including 33 atovaquone and ELQ-300. We also show that the antimalarial activity of the fungal NDH2 inhibitor HDQ and its new 34 derivative CK-2-68 is due to inhibition of the parasite cytochrome bc 1 complex rather than PfNDH2. These compounds 35 directly inhibit the ubiquinol-cytochrome c reductase activity of the malarial bc 1 complex. Our results call into question 36 the validity of PfNDH2 as an antimalarial drug target. 37

Importance 38
For a long time, PfNDH2 has been considered an attractive antimalarial drug target. However, the conclusion that 39

Introduction 48
The mitochondrial electron transport chain (mtETC) is an important, validated drug target in malaria parasites. The 49 mtETC is the primary generator of the electrochemical gradient across the mitochondrial inner membrane. In the asexual 50 blood stages of malaria parasites, however, the only critical function of the mtETC is the continuous reoxidation of 51 ubiquinol to sustain dihydroorotate dehydrogenase (DHODH) activity, which is required for de novo pyrimidine 52 biosynthesis (1). In contrast, in insect stages, mitochondrial oxidative phosphorylation appears to have increased 53 importance (2), likely requiring an intact central carbon metabolism (3) and increased mtETC activity to maintain the 54 electrochemical gradient that drives ATP synthesis. For decades, the mtETC of malaria parasites has attracted major drug 55 development efforts (4), ultimately resulting in antimalarials for clinical use and in preclinical/clinical stages of 56 development. Malarone TM , a combination of atovaquone and proguanil, has been used clinically since 2000. Recent drug 57 development efforts focused on the parasite DHODH led to the clinical candidate DSM265, which is currently undergoing 58 Phase II clinical trials (5, 6). ELQ-300, an inhibitor of the Qi site of the bc 1 complex (Complex III), has also reached 59 preclinical development (7,8). This underscores that the essential protein components of the parasite mtETC are attractive 60 antimalarial drug targets. 61 In the parasite mtETC, there are five dehydrogenases that donate electrons to ubiquinone producing ubiquinol (reduced 62 ubiquinone), which is subsequently oxidized by the bc 1 complex (Complex III). These five enzymes include NDH2, 63 malate quinone oxidoreductase (MQO), DHODH, glycerol 3-phosphate dehydrogenase (G3PDH), and succinate 64 dehydrogenase (SDH). As mentioned above, the parasite DHODH is a validated antimalarial drug target. NDH2 has also 65 been considered a promising antimalarial drug target for over a decade (9)(10)(11)(12). In general, NADH dehydrogenase is a 66 membrane bound flavoenzyme that catalyzes electron transfer from NADH to quinone producing NAD + and quinol. In 67 human mitochondria, a type I NADH dehydrogenase (Complex I) has 45 subunits and pumps protons across the 68 mitochondrial inner membrane concomitant with electron transfer (13). Mutations of Complex I subunits are responsible 69 for a significant portion of hereditary human respiratory chain disorders (14). In contrast, malaria parasites lack the 70 conventional multi-subunit Complex I. Instead, they have a type II NADH dehydrogenase (NDH2), which is a single 71 subunit, non-proton pumping protein, likely attaching to the mitochondrial inner membrane and facing the mitochondrial 72 matrix. Toxoplasma gondii, another apicomplexan parasite, has two isoforms of NDH2, which both face the mitochondrial 73 matrix, catalyzing oxidation of mitochondrial NADH (15). NDH2 is also present in bacteria (16), fungi (17) and plants 74 (18), but not in humans or other mammals. 75

76
The absence of NDH2 in humans suggests that the parasite enzyme might be a good antimalarial drug target (9)(10)(11)(12). In 77 1990, Fry and Beesley first measured NADH oxidation activities in isolated mitochondria of malaria parasites (P. yoelii 78 and P. falciparum) using two spectrophotometric methods (19). Briefly, in the first assay, NADH oxidation was coupled 79 to cytochrome c reduction and changes of cytochrome c absorption spectrum were measured at a wavelength of 550 nm; 80 in the second assay, NADH oxidation produced NAD + , directly leading to a reduced absorption at 340 nm. Using these 81 measurements, Fry and Beesley found that NADH oxidation in the mitochondrial samples was more robust than that of 82 other substrates and was not inhibited by rotenone, a classical Complex I inhibitor. The conclusion was that mitochondria 83 of malaria parasites were able to oxidize NADH, although it was not clear which specific enzyme(s) were responsible or 84 which pathway(s) were involved. In 2006, Biagini et al. also observed significant NADH oxidation activity (direct assay 85 at 340 nm) in P. falciparum extracts (9). Biagini et al. used atovaquone and potassium cyanide to block the activities of 86 Complexes III and IV individually, leading them to conclude that the observed NADH oxidation was due to PfNDH2 (9). 87 However, with the use of total cell extracts containing various NADH dependent enzymes, it seems questionable to 88 attribute all the observed NADH oxidation activity to PfNDH2 alone (9,12). Coincidentally at that time, the ubiquinone 89 analogue HDQ (1-hydroxy-2-dodecyl-4(1H) quinolone) was found to be a potent inhibitor of the fungal NDH2 in 90 Yarrowia lipolytica (20). Later HDQ was shown to be highly effective against P. falciparum and T. gondii parasites (10). 91 Based on these results (9,10,12), it became widely accepted that PfNDH2 could be an attractive antimalarial drug target. 92 As a result, a significant drug discovery campaign based on high throughput screening was undertaken to seek HDQ-like 93 inhibitors to specifically inhibit PfNDH2 (21-23), yielding the lead compound, CK-2-68 (22). Recently, the crystal 94 structure of PfNDH2 was resolved via x-ray crystallization (24), which could further encourage drug development efforts 95 towards PfNDH2 using approaches based on in silico docking and structure activity relationships of PfNDH2 inhibitors. 96

97
The rationale for targeting PfNDH2 for antimalarial drug development has, however, been controversial (25, 26). The 98 fact that the entire mtETC in asexual blood stages could be functionally bypassed by expression of the heterologous 99 yDHODH from Saccharomyces cerevisiae to support pyrimidine biosynthesis in the presence of mtETC inhibition raised 100 added 48 h post electroporation. For hdhfr (human dihydrofolate dehydrogenase) selectable marker, 5 nM WR99210 was 128 used. 129 130 2. Plasmid construction. 131 1) Removal of yDHODH from the pAIO pre-gRNA construct. The pre-gRNA construct pAIO was generously provided 132 by Dr. Josh Beck (30); the plasmid contains yDHODH and Streptococcus pyogenes Cas9 coding sequences (CDS) 133 connected by a 2A "self-cleaving" peptide. To remove yDHODH, pAIO was digested with BamHI and BglII to release 134 the entire sequence of yDHODH and the first 250 bp of Cas9, since there is no unique restriction site between the two 135 genes that could be used to release the yDHODH CDS alone. The first 250 bp of Cas9 were amplified from the original 136 pAIO vector with primers P1 and P2, which include short homologous sequences that match the ends of the pAIO vector 137 after its digestion with BamHI and BglII. The PCR product and the digested vector were then joined together using 138 NEBuilder® HiFi DNA Assembly (New England Biolabs®, Inc). A colony PCR was performed to screen colonies using 139 primers P1 and P2. Positive clones were grown up, and their plasmid DNAs were digested with BamHI and BglII to 140 confirm the loss of yDHODH. The positive plasmids were then sequenced using a primer upstream of Cas9 (P3) to 141 confirm the intactness of Cas9. These procedures yielded the pre_gRNA construct without yDHODH, namely pAIO-142 yDHODH(-). 143 2) PfNDH2 KO construct. PfNDH2 (PF3D7_0915000) is 1602 bp long with no introns. We cloned the 5' and 3' 144 homologous regions of PfNDH2 into a pCC1 vector bearing the hdhfr selectable marker (31). The 5'HR (934 bp) was 145 amplified with primers P4 and P5 and cloned into pCC1 digested by NcoI and EcoRI. Subsequently, the 3'HR (936 bp) 146 was amplified with primers P6 and P7 and cloned into the vector digested by SpeI and SacII. After cloning, both 5'HR 147 and 3'HR were sequenced (Genewiz LLC). The KO construct was named 5'3'PfNDH2_pCC1. Maxi prep DNA of 148 5'3'PfNDH2_pCC1 (Qiagen) was digested with HincII overnight to linearize the vector before transfections. 149 3) Guide RNA constructs. The sequence between the 5'HR and 3'HR of PfNDH2 (490 bp) was submitted to the gRNA 150 design tool (http://grna.ctegd.uga.edu/) to seek potential gRNAs. From the list of candidates, three sequences were chosen 151 based on their high scores and zero off-target predictions. For each of these sequences, a pair of complementary 152 oligonucleotides (60 or 61 bp) was synthesized and annealed in a mixture of NEB Buffers 2 and 4 by heating to 95°C for 153 5 minutes, then slowly cooling to room temperature. The vector, pAIO-yDHODH(-), was digested with BtgZI and joined 154 with the annealed oligonucleotide pair by gene assembly (New England Biolabs®, Inc), yielding a pAIO-yDHODH(-)-155 gRNA construct. Other gRNA cloning procedures followed our published protocol (32). 156 Primers used for cloning procedures are listed below. 157 with 10 mM HEPES) treatment. On day 0, parasites were inoculated into a 24 well plate with each well containing 2.5 ml 181 of culture at 1% parasitemia and 3% hematocrit. Cultures were fed daily and split every two days. At each split (1:5), a 182 sample of the parasitized RBCs was pelleted and fixed with 4% paraformaldehyde at 4°C overnight. After all samples 183 were collected and fixed, they were washed with 1x PBS and stained with SYBR green I at 1:1000 (Catalog S7567, Life 184 technologies by ThermoFisher Scientific). The samples were washed with PBS three times and analyzed on a C6 Flow 185 Cytometer (BD). A total of 250,000 events were collected for each sample. Unstained infected RBCs and stained 186 uninfected RBCs served as negative controls for gating. Growth curves were drawn using Graphpad Prism 6. 187 188 4. Growth inhibition assays using 3 H-hypoxanthine incorporation. Inhibitor compounds were diluted by a series of three-189 fold dilutions in 96 well plates in low hypoxanthine medium (2.5 mg/L). Parasites were washed three times with low 190 hypoxanthine medium, supplemented with fresh blood sufficient to make 1% parasitemia and re-suspended in the proper 191 volume of low hypoxanthine medium to make a suspension with 3% hematocrit. Aliquots of the diluted culture were 192 added to the 96-well plates containing the inhibitor dilution series. After 24 h incubation, 10 μl of 0.5 μCi 3 H-193 hypoxanthine was added to each well and the plates were incubated for another 24 h. After a total of 48 h incubation, the 194 parasites were lysed by freezing-and-thawing, and nucleic acids were collected onto a filter using a cell harvester (Perkin 195 Elmer). Radioactivity was counted using a Topcount scintillation counter (Perkin Elmer). Data were analyzed and graphed 196 using Graphpad Prism 6. 197 198 5. Ubiquinol-cytochrome c reduction assay. Mitochondria of ∆PfNDH2 and D10 WT were individually isolated using a 199 method published previously (32, 33). Briefly, a large volume of parasite culture of each line (~2 liter) was lysed with 200 saponin (0.05%) and disrupted in a N 2 cavitation chamber (Parr 4639 Cell Disruption Bomb) in an isotonic mitochondrial 201 medium. The total parasite lysate was spun down at 900 g for 6 min to remove large debris, and the cloudy supernatant 202 was passed through a MACS CS column (Miltenyi Biotec) in a Vario MACS magnetic separation apparatus to remove 203 most of the hemozoin. The eluted light yellow material was pelleted at 23,000 × g for 20 min at 4 °C, and the pellet was 204 re-suspended in buffer and stored at -80 °C. The cytochrome c reductase activity of the bc 1 complex was measured with a 205 modification of previous methods (32-34). The assay volume was 300 µl, containing mitochondrial proteins (~5-10 µl 206 mitochondrial preparation), 100 µM decylubiquinol, 75 µM horse heart cytochrome c (Sigma-Aldrich), 0.1 mg/ml n-207 docecyl-β-D-maltoside, 60 mM HEPES (pH 7.4), 10 mM sodium malonate, 1 mM EDTA, and 2 mM KCN, and was 208 incubated at 35°C in a stirred cuvette. Reduction of horse heart cytochrome c was recorded at 550 nm with a CLARITY 209 VF integrating spectrophotometer (OLIS, Bogart, GA). A Bio-Rad colorimetric assay was used to measure protein 210 concentrations of all mitochondrial samples. 211 6. NADH-cytochrome c reductase assay. Assay conditions were similar to those described above for the ubiquinol-212 cytochrome c reduction assay. The 300 µL assay mix contained 5-10 µl of mitochondrial proteins, 50 µM horse heart 213 cytochrome c, 60 mM HEPES (pH 7.4), 10 mM sodium malonate, 1 mM EDTA, 2 mM KCN and 300 µM NADH. The 214 assay buffer contained no detergent, since it was reported that detergents heavily interfere with assays of NADH oxidation 215 (35). 216 217

Results 218
PfNDH2 is not essential in asexual blood stages of Plasmodium falciparum. 219 Transcriptomics data indicate that the type II NADH dehydrogenase in P. falciparum (PF3D7_0915000) is expressed in 220 the asexual blood stages (PlasmoDB.org). It has been shown that the leader sequence of PfNDH2 was able to target GFP 221 into the mitochondrion (36), suggesting that PfNDH2 is a mitochondrial enzyme. To further confirm that, we genetically 222 tagged PfNDH2 with 3x HA and the tagged PfNDH2 was localized to the parasite mitochondrion by immunofluorescence 223 assays (37). To assess the essentiality of PfNDH2, we employed the CRISPR/Cas9 DNA repair technique. A KO plasmid 224 vector was constructed ( Figure 1A, Materials and Methods), containing a 5'HR (homologous region) mostly upstream of 225 the gene's coding sequence (CDS) (outside) and a 3'HR near the end of the CDS (inside). The 3'HR was chosen from the 226 coding region to circumvent inclusion of overly high AT content in the KO vector. Three gRNA sequences targeting the 227 PfNDH2 gene were individually cloned into a modified pre-gRNA-Cas9 plasmid construct, from which yDHODH had 228 been removed (Materials and Methods). Previous studies have shown that expression of yDHODH in malaria parasites 229 renders the entire mtETC nonessential by providing a metabolic bypass for pyrimidine biosynthesis (1). Therefore, to 230 assess the essentiality of PfNDH2 in the context of a normal mtETC, we removed yDHODH from the gRNA vectors. The 231 KO plasmid was linearized by restriction digestion and transfected into D10 parasites together with the three circular 232 gRNA vectors (Materials and Methods). Viable transgenic parasites were observed under WR99210 selection three weeks 233 post transfection. As shown in Figure 1B, a diagnostic PCR revealed that PfNDH2 was disrupted. We then tightly 234 synchronized both ∆PfNDH2 and WT lines and examined the growth rates over 4 intraerythrocytic developmental cycles 235 (IDCs) via flow cytometry. As shown in Figure 1C, ∆PfNDH2 and WT parasites grew equally well over this time period. 236 The PfNDH2 KO line was further maintained in culture for over one month, and no growth defects were noticeable (data 237 not shown). Deletion of PfNDH2 also did not appear to affect parasite health and morphology ( Figure 1D). Collectively, 238 our data indicate that PfNDH2 is not essential in asexual blood stages of P. falciparum, consistent with the KO study 239 carried out previously in the rodent malaria parasite, P. berghei (28). These results argue against the long-held assumption 240 that PfNDH2 is an attractive drug target (9)(10)(11)(12). 241

242
The ∆PfNDH2 parasite is equally susceptible to mtETC inhibitors. 243 The healthy growth of the PfNDH2 parasites in vitro (Figure 1) suggests that the parasite mtETC remains functionally 244 competent in the absence of PfNDH2. To challenge the KO parasites, we exposed them to mtETC inhibitors in growth 245 inhibition assays, measured as 3 H-hypoxanthine incorporation. As shown in Figure 2, in comparison to the WT, the 246 ∆PfNDH2 parasites were equally sensitive to atovaquone (a Q o site inhibitor of the bc 1 complex) and ELQ-300 (a Q i site 247 inhibitor) (8). Thus, these data suggest that deletion of PfNDH2 has little effect on the sensitivity of asexual parasites to 248 downstream inhibitors of the mtETC. The loss of NDH2, thus, does not appear to affect the function of the remainder of 249 the mtETC. As noted previously, HDQ and its newer derivative CK-2-68 were considered to be PfNDH2 specific 250 inhibitors (21-23) or, more recently, dual-targeting inhibitors of cytochrome bc 1 as well as PfNDH2 (38). In that case, 251 HDQ and CK-2-68 would be expected to lose potency in the ∆PfNDH2 parasite, since the putative primary target is 252 absent. However, HDQ and CK-2-68 were still highly potent in the KO parasite (Figure 2), suggesting that HDQ and CK-253 2-68 primarily target another site than PfNDH2. 254

255
The bc 1 complex of the mtETC is the target of HDQ and CK-2-68. 256 Our data above suggests that HDQ and CK-2-68 target an activity other than PfNDH2 (Figure 2). Since HDQ and CK-2-257 68 are ubiquinone analogs, we suggest that they kill malaria parasites by targeting the bc 1 complex, although Vallieres et 258 al. and Biagini et al. previously suggested that HDQ and CK-2-68 had a dual effect on both PfNDH2 and the bc 1 complex 259 (38, 39). To distinguish between these alternatives, we performed growth inhibition assays in the yDHODH transgenic 260 parasite line using HDQ and CK-2-68 in combination with proguanil. As shown previously, expression of the yDHODH 261 gene bypasses the need for mtETC function in asexual parasites (1). The yDHODH transgenic parasites have become a 262 handy tool to examine whether a compound targets the mtETC, as all mtETC inhibitors suffer a large loss of potency in 263 the yDHODH background, which applies to both bc 1 inhibitors and PfDHODH inhibitors (40). Further, a low 264 concentration of proguanil (1 µM) can restore sensitivity to bc 1 inhibitors in yDHODH transgenic parasites (1), but not for 265 PfDHODH inhibitors. As a control, we showed that yDHODH parasites were fully resistant to atovaquone but became 266 fully sensitive in the presence of 1 µM proguanil (Figure 3). Upon inhibition by atovaquone, the yDHODH parasites lose 267 their primary source of ∆ m generation, conveyed by the bc 1 complex and cytochrome c oxidase of the mtETC, and 268 become hypersensitive to proguanil, which targets a secondary generator of ∆ m (1). Using this system, we tested the 269 HDQ and CK-2-68 sensitivity of the yDHODH parasites with and without 1 µM proguanil. As shown in Figure 3, 270 yDHODH parasites were highly resistant to HDQ and CK-2-68, as expected; upon proguanil treatment, the yDHODH 271 parasites regained sensitivity to these compounds. HDQ and CK-2-68, thus, behaved in a very similar manner to 272 atovaquone against the yDHODH transgenic parasites, indicating that HDQ and CK-2-68 target the bc 1 complex. These 273 results are consistent with a previous report that found that parasites grow normally in the presence of 10 µM HDQ when 274 expressing the yDHODH gene (41). Furthermore, recent chemical mutagenesis experiments using CK-2-68 generated 275 mutations that were all in the cyt b locus, rather than in PfNDH2 (29). Collectively, these results indicate that HDQ and 276 CK-2-68 are potent cytochrome bc 1 inhibitors. 277

HDQ and CK-2-68 directly inhibit the enzymatic activity of the bc 1 complex. 279
In addition to growth inhibition assays as described above (Figures 2 and 3), we also directly investigated the effect of 280 HDQ and CK-2-68 on the enzymatic activity of the bc 1 complex in a preparation enriched in parasite mitochondria using a 281 spectrophotometric assay (Materials and Methods) (33). As shown in Figure 4, HDQ and CK-2-68 inhibited the 282 ubiquinol-cytochrome c reductase activity in the mitochondria of ∆PfNDH2 and WT in a dose dependent manner. 283 Importantly, the inhibitory potency of HDQ and CK-2-68 were equally robust in two types of mitochondria from WT and 284 ∆PfNDH2, respectively. This provides further evidence that the antimalarial mode of action of HDQ and CK-2-68 arises 285 from inhibition of the bc 1 complex, rather than PfNDH2. 286 287

In vitro measured NADH linked cytochrome c reductase activity is likely non-biological. 288
Previously data from Fry and Beesley (19) revealed a relatively strong NADH-cytochrome c reductase activity in parasite 289 mitochondrial preparations. Interestingly, the activity was not inhibited by rotenone (80 µM) or antimycin A (a Qi 290 inhibitor at 20 µM) (19). Rotenone insensitivity suggested that malaria parasites lack a conventional multi-subunit 291 Complex I, which was later interpreted as evidence that the type II NADH dehydrogenase was essential (9). While 292 antimycin A did not inhibit the NADH-cytochrome c reductase activity in Fry and Beesley's mitochondrial preparations,293 it did inhibit the cytochrome c reductase activity when other mitochondrial substrates were used, such as -294 glycerophosphate and succinate (19). In addition, antimycin A kills malaria parasites in whole cell assays with an EC 50 of 295 13 nM (42). Thus, the provenance of the apparent NADH-cytochrome c reductase activity observed in mitochondrial 296 preparations has been an unsettled issue. In intact parasites, NADH oxidized by NDH2 is presumed to pass electrons to 297 ubiquinone, which are then transferred on to the bc 1 complex, cytochrome c, cytochrome c oxidase and, finally, to O 2 . If 298 the in vitro assay were replicating the initial steps of the in vivo pathway, we should observe a much diminished NADH-299 cytochrome c reductase activity in the ∆PfNDH2 parasites since PfNDH2, missing in the knockout parasite, is the only 300 known enzyme donating electrons from NADH to the mtETC in the parasites (15). As shown in Figure 5, however, 301 deletion of PfNDH2 had no effect on NADH-cytochrome c reductase activity, suggesting that this in vitro assay is likely 302 non-physiological. Further, in both ∆PfNDH2 and WT mitochondria, NADH-cytochrome c reductase activity was not 303 inhibited by a mix of malaria parasite specific bc 1 inhibitors, including atovaquone (62 nM), ELQ-300 (62 nM), and HDQ 304 (3,100 nM), each at equal or greater than 100x EC 50 ( Figure 5). Thus, our data are consistent with the earlier observation 305 that antimycin A failed to inhibit the NADH-cytochrome c reductase assay (19) and suggest that the in vitro NADH-306 cytochrome c reductase activity is likely non-enzymatic (see Discussion). 307 308

Discussion 309
A common strategy for developing antimicrobial drugs is to target divergent proteins of the microbe to circumvent 310 potential toxicity against the host. Proteins unique to microbes are even more interesting as their inhibitors would 311 potentially have little to no side effects in the host. The type II NADH dehydrogenase is present in malaria parasites but 312 not in humans; thus, it has been considered an attractive prospective drug target for a long time (9,11,43). However, a 313 unique protein may not necessarily be an essential one. A valid drug target should normally be essential to the pathogen in 314 order that its inhibition will arrest growth and/or kill the pathogen. Initial failures to disrupt the NDH2 gene in P. 315 falciparum parasites suggested that the gene might be essential (35,37). On the other hand, data on the effect of mtETC 316 inhibitors in yDHODH transgenic parasites (1) and the reported knockout of NDH2 in P. berghei (28) suggested that 317 PfNDH2 should be dispensable in asexual parasites (as described above). Without conclusive data, however, it remains a 318 long-debated issue in the field whether PfNDH2 is a good antimalarial drug target. In this report, we have provided strong 319 evidence indicating that PfNDH2 is dispensable in asexual blood stages and, therefore, is unlikely to be an effective 320 antimalarial drug target. We note that our knockout result is consistent with the recent genetic screen of P. falciparum 321 growth phenotypes, in which a PiggyBac transposon insertion was recovered in the CDS of PfNDH2, suggesting non-322 essentiality of the gene (44). 323

324
Our results suggest that the parasite mtETC is functionally intact in the absence of PfNDH2. Not only is the growth of the 325 KO line closely similar to that of the WT parental line ( Figure 1C), but the response to cytochrome bc 1 inhibitors is 326 virtually identical (Figure 2). Evidently, in the PfNDH2 parasites, the other ubiquinone-dependent dehydrogenases-327 MQO, SDH, G3PDH, and DHODH-supply sufficient ubiquinol to maintain adequate function of the mtETC during 328 asexual development. DHODH is essential for the parasite's pyrimidine de novo synthesis pathway, since malaria 329 parasites cannot salvage pyrimidine precursors; the other dehydrogenases, however, may be functionally redundant as 330 electron donors to the mtETC. We have previously carried out a comprehensive genetic and biochemical study in the TCA 331 cycle of P. falciparum (3). In the asexual blood stages, the main carbon source of the TCA is glutamine, rather than 332 glucose. KDH (alpha-ketoglutarate dehydrogenase) is the entry point of glutamine derived carbons into the TCA cycle. 333 KDH converts alpha-ketoglutarate to succinyl-CoA, with the concomitant reduction of NAD + to NADH, and it is likely 334 that KDH is the principal producer of NADH in the mitochondrial matrix due to a relatively large TCA flux observed with 335 labeled glutamine (3). Yet, neither KDH nor the TCA flux contributed by glutamine is essential to the parasite in asexual 336 blood stages, which is consistent with the non-essential nature of PfNDH2 as a consumer of NADH. 337 338 Although HDQ and CK-2-68, and probably other related derivatives (29) do not primarily target PfNDH2 in parasites, as 339 shown by our results, they are potent antimalarial compounds via inhibition of the parasite bc 1 complex. Importantly, 340 HDQ and CK-2-68 retained their potency in atovaquone resistant parasites (38,39). Experiments with yeast cyt b mutants 341 suggested that HDQ likely bound to the Qi site of bc 1 complex, whereas atovaquone is a Qo site inhibitor (38). CK-2-68, 342 on the other hand, is likely to be a Qo site inhibitor, but, nevertheless, exhibited no cross resistance with atovaquone (29). 343 Biagini et al. have developed additional quinolone derivatives with more favorable pharmacological properties that were 344 predicted to bind at the Qo site (39). Combinations of non-cross resistant bc 1 inhibitors may be effective at slowing the 345 development and spread of resistance, since strong resistance mutations in cyt b may exert a significant survival fitness 346 cost (45), including blocking transmission (46). Thus, the development of additional antimalarial candidates targeting the 347 bc 1 complex may facilitate the future development of effective combination therapies. Indeed, atovaquone and ELQ-300, 348 Q o and Q i inhibitors respectively, were recently shown to be a highly effective and synergistic antimalarial combination 349 (47). 350

351
The results of our attempts to measure in vitro NADH-cytochrome c reductase activity spectrophotometrically provide a 352 cautionary tale for the design and interpretation of assays involving the oxidation of NADH, a reactive reductant. Neither 353 elimination of PfNDH2 nor strong inhibition of the cytochrome c reductase activity of bc 1 affected the observed reaction 354 (Fig. 5), implying that the reaction does not proceed through the mtETC. Fry and Beasley apparently observed the same 355 phenomenon when they measured apparent NADH-cytochrome c reductase activity in Plasmodium mitochondria with and 356 without antimycin A, a general Qi site inhibitor of the bc 1 complex (19). Given the report that detergents (which form 357 micelles) accelerate NADH oxidation (35), we speculate that it may be the presence of mitochondrial phospholipid 358 membranes in the mitochondrial samples that produce this effect. Cytochrome c is known to bind to phospholipids head 359 groups (48), so mitochondrial particles could provide a surface that concentrates cytochrome c for reaction with NADH (a 360 trimolecular reaction, requiring 2 cytochromes c to oxidize one NADH, as it is a 2-electron reductant). The non-enzymatic 361 reaction may also be facilitated by the relatively high concentration of cytochrome c used in spectrophotometric assays 362 (50-100 µM). At any rate, our results demonstrate that the apparent robust NADH-cytochrome c activity that has been 363 reported in Plasmodium mitochondrial preparations in vitro is not an indication of high NADH dehydrogenase activity in 364 intact parasites. hypoxanthine incorporation assays. With an EC 50 of ~60 µM, 1 µM of proguanil had no effect on WT parasites (1). 382 without and with addition of bc 1 inhibitors (atovaquone (62 nM), ELQ-300 (62 nM) and HDQ (3,100 nM)). Data shown is 391 mean ± s.d. of n=3 replicates. 392

Acknowledgements 393
We thank Dr. Kristin D. Lane and Dr. Thomas E. Wellems of NIH/NIAID for kindly providing the CK-2-68 compound. 394 The Cas9-gRNA construct was generously provided by Dr. Josh R. Beck (now at Iowa State University) and Dr. Daniel E.