The Plasmodium falciparum apicoplast cysteine desulfurase provides sulfur for both iron-sulfur cluster assembly and tRNA modification

Iron-sulfur clusters (FeS) are ancient and ubiquitous protein cofactors that play fundamental roles in many aspects of cell biology. These cofactors cannot be scavenged or trafficked within a cell and thus must be synthesized in any subcellular compartment where they are required. We examined the FeS synthesis proteins found in the relict plastid organelle, called the apicoplast, of the human malaria parasite Plasmodium falciparum. Using a chemical bypass method, we deleted four of the FeS pathway proteins involved in sulfur acquisition and cluster assembly and demonstrated that they are all essential for parasite survival. However, the effect that these deletions had on the apicoplast organelle differed. Deletion of the cysteine desulfurase SufS led to disruption of the apicoplast organelle and loss of the organellar genome, whereas the other deletions did not affect organelle maintenance. Ultimately, we discovered that the requirement of SufS for organelle maintenance is not driven by its role in FeS biosynthesis, but rather, by its function in generating sulfur for use by MnmA, a tRNA modifying enzyme that we localized to the apicoplast. Complementation of MnmA and SufS activity with a bacterial MnmA and its cognate cysteine desulfurase strongly suggests that the parasite SufS provides sulfur for both FeS biosynthesis and tRNA modification in the apicoplast. The dual role of parasite SufS is likely to be found in other plastid-containing organisms and highlights the central role of this enzyme in plastid biology.


Introduction 52
Malaria parasites contain a relict plastid organelle called the apicoplast that is required 53 for its survival (Köhler et al., 1997;McFadden et al., 1996). The essentiality of this organelle and 54 the unique biochemical pathways within, such as iron-sulfur cluster (FeS), and isoprenoid 55 precursor biosynthetic pathways offers a potentially rich source of new antimalarial drug targets 56 (Ellis et al., 2001;Jomaa et al., 1999;Seeber, 2002). Since its discovery, several inhibitors 57 targeting these and other pathways have been described, supporting the assertion that this 58 organelle represents a viable source of novel drug targets (Botté et al., 2012;Dahl & Rosenthal, 59 2008; Ke et al., 2014;Shears et al., 2015). 60 FeS serve as cofactors for an array of proteins across kingdoms and are involved in a 61 myriad of biological functions, including electron transfer, sulfur donation, redox sensing, gene 62 expression, and translation (Blahut et al., 2020;Przybyla-Toscano et al., 2018;Rouault, 2019). 63 FeS cofactors are found in a variety of forms, most commonly in the rhombic 2Fe-2S or the 64 cubic 4Fe-4S forms and are typically bound to proteins through covalent bonds with cysteine 65 side chains (Beinert, 2000;Lill, 2009;Lu, 2018). In Plasmodium falciparum, FeS cofactors are 66 formed within the apicoplast by the sulfur utilization factor pathway (SUF), for use by FeS-67 dependent proteins within the organelle (Charan et al., 2017;Gisselberg et al., 2013;Pala et al., 68 2018;Swift et al., 2022), while the mitochondrion houses the iron-sulfur cluster formation (ISC) 69 pathway Gisselberg et al., 2013;Sadik et al., 2021). The ISC 70 pathway generates FeS for use by FeS-dependent proteins within the mitochondrion, in addition 71 to transferring a sulfur-containing moiety to the cytosolic iron-sulfur protein assembly (CIA) 72 machinery for FeS generation and transfer to cytosolic and nuclear proteins (Dellibovi-Ragheb et 73 al., 2013;Lill, 2009). 74 epifluorescence microscopy ( Figure 1D). Despite the presence of intact apicoplast organelles, 144 both parasite lines required supplementation with exogenous mevalonate, demonstrating that 145 SufC and SufD are essential proteins ( Figure 1E). Taken together, these results show that SufC 146 and SufD are essential for parasite survival, although neither protein is required for apicoplast 147 maintenance. 148 Since SufC and SufD are essential, it is likely that the SufBC2D complex is essential. 149 Unfortunately, we cannot genetically modify the apicoplast genome-encoded SufB with any 150 available experimental techniques, but we can simultaneously delete both SufC and SufD to 151 remove the possibility of any type of complex forming with SufB. In Escherichia coli, pulldown 152 assays demonstrated that different types of FeS assembly complexes can be formed (SufBC2D 153 and SufB2C2), which suggests that there may be some redundancy between complex proteins 154 (Saini et al., 2010;Yuda et al., 2017). We generated ΔsufC/sufD double knockout parasites in the 155 PfMev line, under continuous supplementation of mevalonate ( Figure 1F). Consistent with the 156 phenotypes of the ∆sufC and ∆sufD parasite lines, ΔsufC/sufD parasites were also found to have 157 intact apicoplasts (Figures 1G and 1H) and were dependent on exogenous mevalonate 158 supplementation for survival ( Figure 1I). Collectively, these findings demonstrate that while the 159 FeS assembly complex is essential for parasite survival, it is not required for apicoplast 160 maintenance. 161 162

SufS is required for apicoplast maintenance while SufE is not 163
We next investigated the sulfur acquisition steps of the SUF pathway upstream of SufC 164 and SufD. In the apicoplast of P. falciparum, SufS (PF3D7_0716600), along with its partner 165 SufE (PF3D7_0206100), mobilize sulfur from L-cysteine, with SufE transferring the sulfur to the 166 SufBC2D complex . The cysteine desulfurase activity of the P. 167 falciparum SufS has been confirmed biochemically (Charan et al., 2014) and through 168 complementation in E. coli . It was also shown that the P. falciparum 169 SufE enhances the cysteine desulfurase activity of SufS by up to ~17-fold in an in vitro 170 biochemical assay (Charan et al., 2014). To disrupt sulfur acquisition in the SUF pathway, we 171 generated deletions of both sufS and sufE in the PfMev line (Figure 2A). In ∆sufE parasites, the 172 apicoplast remained intact as evidenced by successful PCR amplification of the sufB gene; 173 however, we were unable to amplify this gene from ∆sufS parasites ( Figure 2B). Consistent with 174 the PCR results, we observed an intact apicoplast in ∆sufE parasites by live microscopy, while in 175 ∆sufS parasites we observed multiple discrete api-SFG labeled vesicles -a hallmark of apicoplast 176 organelle disruption ( Figure 2C). Additionally, both parasite lines were dependent on 177 mevalonate for survival ( Figure 2D). Taken together, these results indicate that both SufE and 178 SufS are required for parasite survival, but only SufS is required for apicoplast maintenance. 179 180

MnmA is essential for apicoplast maintenance 181
While SufS is required for apicoplast maintenance, none of the other SUF pathway 182 proteins or any of the FeS-dependent proteins in the apicoplast are required for this process 183 (Swift et al., 2022). This suggests that the reliance on sulfur for organelle maintenance is likely 184 driven by a different sulfur-dependent pathway. Several biochemical pathways require sulfur, 185 including those involved in the biosynthesis of thiamine, biotin, lipoic acid, molybdopterin, and 186 thio-modifications of tRNA (Hidese et al., 2011;Leimkühler et al., 2017;Mihara & Esaki, 187 2002). Of these pathways, we found that the biosynthesis of lipoic acid and tRNA thio-188 modifications were the only ones that appeared to be present in P. falciparum with predicted 189 localization to the apicoplast (Ralph et al., 2004) (Supplementary table 1). In a recent study , we 190 showed that lipoic acid synthesis is dispensable (Swift et al., 2022), however, tRNA thiolation 191 has not been studied in malaria parasites. Based on sequence homology, P. falciparum parasites 192 appear to contain several enzymes capable of catalyzing tRNA thiolation reactions, but only one 193 appears to be a possible apicoplast protein. This protein (PF3D7_1019800) is currently annotated 194 as a tRNA methyltransferase (Aurrecoechea et al., 2008), but it shares 30% sequence identity 195 with an E. coli enzyme called MnmA (tRNA-specific 2-thiouridylase). MnmA inserts sulfur at 196 carbon-2 (C2) of uridine at position 34 (s 2 U34) of tRNA Lys UUU, tRNA Glu UUC, and tRNA Gln UUG 197 (Black & Santos, 2015a;Leimkühler et al., 2017;Shigi, 2014Shigi, , 2018. E. coli MnmA receives 198 sulfur from a series of five sulfur transfer proteins, TusA/B/C/D/E (Black & Santos, 2015a;199 Ikeuchi et al., 2006;Shigi, 2014), which ultimately acquire sulfur from the IscS cysteine 200 desulfurase, but cannot obtain sulfur from the E. coli SufS . It is not clear 201 whether the apicoplast contains any orthologs of the Tus proteins, and the parasite IscS has 202 already been localized to the mitochondrion instead of the apicoplast in malaria parasites 203 . 204 Multiple sequence alignments (MSA) of the putative P. falciparum MnmA (Pf MnmA) 205 with orthologs from E. coli, Bacillus subtilis, and Saccharomyces cerevisiae reveal that it has a 206 421 aa N-terminal extension (Figure 3-figure supplement 1), which is predicted to contain an 207 apicoplast transit peptide (Foth et al., 2003;Ralph et al., 2004). MSA also demonstrates that the 208 Pf MnmA has conserved catalytic cysteines as well as a highly conserved ATP-binding PP-loop 209 motif (SGGXDS) (Figure 3A, Figure 3-figure supplement 1) (Numata et al., 2006;Shigi et al., 210 2020). This PP-loop motif activates the C2 of nucleotide U34 of the target tRNA by adenylation 211 in an ATP-dependent manner (Mueller, 2006;Numata et al., 2006). The first catalytic cysteine 212 receives sulfur generating an MnmA-persulfide, while the second catalytic cysteine releases the 213 sulfur from the adduct and transfers it to the activated U34 (Čavužić & Liu, 2017;Shigi et al., 214 2020). Other pathogenic apicomplexans also seem to have an ortholog of MnmA with the 215 conserved cysteines and domains (Figure 3-figure supplement 2) and a putative MnmA from 216 Toxoplasma gondii has recently been described (Yang et al., 2022). 217 To validate the predicted apicoplast localization of the putative Pf MnmA, we generated a 218 parasite line with two C-terminal FLAG tags in tandem appended to the endogenous PfMnmA 219 contains an aptamer array in the 3' untranslated region (UTR) of mnmA to use with the TetR-221 DOZI system (Ganesan et al., 2016;Rajaram et al., 2020) for inducible control over protein 222 production ( Figure 3B). We showed that FLAG-tagged Pf MnmA colocalizes with the 223 apicoplast marker protein, acyl carrier protein (ACP) by immunofluorescence (Manders' 224 coefficient, M1 = 0.818, standard deviation = ± 0.187, n = 22), confirming Pf MnmA localization 225 to the apicoplast ( Figure 3E). We next attempted to knock down Pf MnmA using the TetR-226 DOZI system in the mnmA-flag parasite line (Figure 3-figure supplement 4A). We monitored 227 parasite growth in control (aTc added) and knockdown (aTc removed) conditions over eight 228 days. From day five onwards, the parasites showed a significant growth defect under the 229 knockdown condition ( Figure 3F). When parasites in the knockdown condition were 230 supplemented with mevalonate (rescue), the parasites grew similarly to parasites under the 231 control condition (Figure 3F), further confirming the apicoplast-associated activity of Pf MnmA. 232 During the growth assay, we also assessed the apicoplast morphology of parasites under control 233 and knockdown conditions via live epifluorescence microscopy every 48 h. We started to 234 observe multiple discrete api-SFG labeled vesicles at day four following aTc removal (Figure 235 ~70% of parasites contained an intact apicoplast. By day eight, only ~25% of parasites had an 237 intact apicoplast. Significant growth defects and disruption of the apicoplast following knock 238 down of Pf MnmA suggests that Pf MnmA is essential for parasite survival and apicoplast 239 maintenance. To further confirm these findings, we deleted the mnmA gene through Cas9-240 mediated genome editing in PfMev parasites under continuous mevalonate supplementation 241 ( Figure 3I). The deletion of mnmA resulted in apicoplast disruption, as evidenced by the 242 inability to detect the apicoplast genome encoded sufB gene ( Figure 3J) and the presence of 243 multiple discrete vesicles labeled by api-SFG ( Figure 3K). Additionally, ∆mnmA parasites were 244 dependent on exogenous mevalonate supplementation for survival ( Figure 3L) In E. coli, s 2 U biosynthesis starts with the acquisition of sulfur from L-cysteine by the 250 cysteine desulfurase IscS, which then relays the sulfur via the five proteins of the Tus system 251 (TusABCDE) to MnmA (Black & Santos, 2015a;Ikeuchi et al., 2006;Outten et al., 2003;Shigi, 252 2014). MnmA then uses that sulfur to modify the target tRNAs at the U34 position in an ATP-253 dependent manner (Mueller, 2006;Numata et al., 2006). Not all bacteria contain IscS or the Tus 254 system to relay sulfur to MnmA. In B. subtilis for example, a specialized cysteine desulfurase, 255 YrvO, provides sulfur directly to MnmA (Black & Santos, 2015a (Swift et al., 2021), through knock-in via mycobacteriophage integrase-261 mediated recombination (Figure 4-figure supplement 1) (Spalding et al., 2010). This parasite 262 line is hereafter referred to as bsmnmA + . Bs MnmA might not be functional in P. falciparum in 263 the absence of its cognate cysteine desulfurase, Bs YrvO, as previously demonstrated in E. coli 264 (Black & Santos, 2015a). To address this possibility, we also generated a parasite line expressing 265 a Bs MnmA-YrvO fusion protein (bsmnmA-yrvO + ) (Figure 4-figure supplement 2) using the 266 same knock-in method. 267 The bsmnmA and bsmnmA-yrvO expression cassettes encode a conditional localization 268 domain (CLD) at the protein N-terminus for inducible control over protein localization (Roberts 269 et al., 2019) and contain an mCherry tag on the C-terminus for visualization by live cell 270 fluorescence ( Figure 4A). The CLD directs the tagged protein to the apicoplast, but following 271 the addition of the ligand, Shield1, the tagged protein is redirected to the parasitophorous vacuole 272 (Roberts et al., 2019). An aptamer array was also included at the 3' UTR of these genes, for use 273 with the TetR-DOZI system (Ganesan et al., 2016;Rajaram et al., 2020)  Additionally, the expression and apicoplast localization of Bs MnmA-YrvO fusion protein was 291 also confirmed via live epifluorescence microscopy ( Figure 5D). 292 To demonstrate complementation of Pf MnmA with Bs MnmA-YrvO fusion protein more 293 conclusively, we next attempted to knock down Bs MnmA-YrvO in the bsmnmA-yrvO + ΔmnmA 294 parasite line by utilizing the TetR-DOZI and CLD systems ( Figure 4A). We monitored the 295 growth of bsmnmA-yrvO + ΔmnmA parasites under permissive (aTc added, Shield1 removed) and 296 non-permissive conditions (aTc removed, Shield1 added) for eight days. Under the non-297 permissive condition, the bsmnmA-yrvO + ΔmnmA parasites showed a significant growth defect 298 from day four onwards (Figure 5E). Live epifluorescence microscopy on day eight revealed that 299 parasites grown under the non-permissive condition exhibited a disrupted apicoplast phenotype 300 Deletion of sufS resulted in a disrupted apicoplast mevalonate-dependent phenotype 326 ( Figure 2). However, deletion of other SUF pathway components (Figure 1, 2) resulted in an 327 intact apicoplast mevalonate-dependent phenotype. These findings led us to hypothesize that 328 mevalonate-dependence results from loss of FeS cofactors needed for isoprenoid synthesis (Swift 329 et al., 2022), while apicoplast disruption results from loss of sulfur needed for tRNA 330 modification. By successfully complementing Pf MnmA with Bs MnmA we demonstrated that 331 Pf MnmA has the same enzymatic activity as the well-characterized bacterial enzyme, and that 332 this activity is essential for parasite survival and apicoplast maintenance (Figure 5, 6). However, 333 the complementation experiments in Figures 5 and 6 did not provide any direct evidence that Pf 334 MnmA is reliant on sulfur generated from the endogenous SufS for use in tRNA modification. 335 To probe whether SufS provides sulfur to Pf MnmA, we used the bsmnmA-yrvO + parasite line. 336 In these parasites, we attempted to delete sufS with continuous supplementation of mevalonate. 337 We expected to obtain a parasite line with intact apicoplasts and a mevalonate-dependent 338 phenotype ( Figure 7A), which would suggest that sulfur acquired by Bs YrvO is only transferred 339 to MnmA but not to the components of the parasite SUF pathway. We were successful in 340 generating the bsmnmA-yrvO + ΔsufS parasite line in the presence of mevalonate ( Figure 7B). As 341 anticipated, these parasites retained intact apicoplasts as confirmed by both PCR and live 342 epifluorescence microscopy ( Figure 7C, 7D). These parasites rely on mevalonate for growth 343 ( Figure 7E). Collectively, these results suggest that the parasite SufS provides sulfur for both the 344 SUF pathway and MnmA-mediated tRNA modifications. 345 346 Discussion 347 A recent survey of P. falciparum apicoplast proteins found that five proteins are known 348 or predicted to rely on FeS cofactors (Swift et al., 2022). Three of these proteins were found to 349 be essential for the growth of blood-stage parasites due to their roles in supporting the MEP 350 isoprenoid precursor pathway (Akuh et al., 2022;Swift et al., 2022). These essential proteins 351 could be deleted in PfMev parasites without a noticeable growth defect or loss of the apicoplast 352 organelle as long as the cultures were supplemented with mevalonate. These results implied that 353 the SUF pathway of FeS synthesis would also be essential for parasite growth and that deletion 354 of SUF pathway proteins would not result in loss of the apicoplast. In general, this has proven to 355 be the case with the deletion of SufE, SufC, and SufD. Deletion of SufS, however, led to 356 apicoplast disruption and indicated that this enzyme plays another essential role in parasite 357 biology. Similar to what we observed in P. falciparum, deletion of SufS in T. gondii also leads to 358 loss of the apicoplast organelle and parasite death (Pamukcu et al., 2021). Conditional deletion of 359 SufS in the murine malaria parasite P. berghei demonstrated that this enzyme is essential for the 360 development of mosquito-stage parasites, although the status of the apicoplast was not reported 361 in this study (Charan et al., 2017). Taken together, these studies suggest that SufS plays a central 362 role in apicoplast biology in other parasite species and other stages of parasite development. 363 The phenotype of the SufC deletion line contrasts with a previous study using a dominant 364 negative SufC mutant. In vitro studies showed that P. falciparum SufC participates in a SufBC2D 365 complex that hydrolyzes ATP and can form FeS cofactors (Charan et al., 2017;Kumar et al., 366 2011). The ATPase activity of SufC is thought to provide energy to drive conformation changes 367 to the entire SufBC2D complex required for iron binding and FeS assembly (Bai et al., 2018;368 Hirabayashi et al., 2015;Yuda et al., 2017). In E. coli, SufC and SufD are essential for SUF 369 pathway activity and acquire iron for FeS assembly; loss of either protein results in reduced iron 370 content in the complex (Saini et al., 2010) and the same may be true for the parasite proteins. We 371 generated ΔsufC, ΔsufD, and ΔsufC/sufD lines and found that sufC and sufD are essential for 372 parasite survival, but we did not observe an apicoplast disruption phenotype in these deletion 373 lines (Figure 1). This finding conflicted with previous results showing that expression of a SufC 374 mutant (K140A) lacking ATPase activity functions as a dominant negative and leads to 375 disruption of the apicoplast . Several factors could be responsible for the 376 apicoplast-disruption phenotype resulting from expression of SufC (K140A). In other organisms, 377 the intact SufBC2D complex enhances SufS desulfurase activity (Hu, Kato, et al., 2017;Hu, 378 Page, et al., 2017;Outten et al., 2003;Wollers et al., 2010) potentially leading to accumulation 379 of toxic S -2 if non-functional SufC (K140A) blocks further sulfur utilization. Alternatively, 380 dominant negative SufC could lead to dysfunctional iron homeostasis. ATP binding to SufC 381 elicits a conformational change in the SufBC2D complex, exposing sites required for iron 382 binding and enabling the formation of nascent clusters (Bai et al., 2018;Hirabayashi et al., 2015;383 Yuda et al., 2017). The dominant negative mutant SufC should be able to bind ATP, but not 384 hydrolyze it, locking the SufBC2D complex in an open position and exposing these sites to the 385 environment. Exposure and release of iron could lead to oxidative damage and loss of the 386 organelle. 387 Gene deletion studies exposed different roles for the two proteins (SufE and SufS) 388 involved in sulfur acquisition. We found that both SufE and SufS are required for parasite 389 survival, however, only SufS is required for apicoplast maintenance (Figure 2, 8A, 8B, 8C). 390 These phenotypes make sense if SufE is required for FeS synthesis but not for tRNA thiolation. 391 In other organisms, SufE has been shown to be capable of enhancing the cysteine desulfurase 392 activity of SufS (Murthy et al., 2007;Outten et al., 2003;Pilon-Smits et al., 2002;Wollers et al., 393 2010;Ye et al., 2006) and this appears to be the case with the P. falciparum proteins with ~17-394 fold rate enhancement reported (Charan et al., 2014). SufE proteins also facilitate the transfer of 395 sulfur to the SufBC2D complex, but SufS enzymes can often transfer sulfur directly to the 396 complex as is presumably the case in organisms lacking SufE (Huet et al., 2005). The 397 requirement for SufE in P. falciparum suggests that either SufS desulfurase activity or SufS 398 sulfur transfer activity (to the SufBC2D complex) is too low in the absence of SufE to support 399 FeS synthesis. The fact that SufE is not required for apicoplast maintenance (and presumably 400 tRNA thiolation) further suggests that SufS desulfurase activity in the absence of SufE is 401 adequate for tRNA thiolation and that SufS may transfer sulfur directly to MnmA. 402 A recent study in the asexual blood-stage of P. falciparum reported 28 different tRNA 403 modifications, including s 2 U and mcm 5 s 2 U (Ng et al., 2018). The s 2 U modification of tRNA is 404 ubiquitous and critical for a number of biological functions related to protein translation as 405 demonstrated in other organisms, including the recognition of wobble codons (Urbonavičius et 406 al., 2001), tRNA ribosome binding (Ashraf et al., 1999), reading frame maintenance (Black & 407 Santos, 2015a), and the reduction of +1 and +2 frameshifts (Black & Santos, 2015a;408 Urbonavičius et al., 2001). The dual activity of P. falciparum SufS and its direct interaction with MnmA are not 451 typical features of SufS desulfurases. In E. coli, IscS is required for MnmA-mediated tRNA 452 thiolation and the role of SufS is confined exclusively to FeS biosynthesis . 453 Thiolation of tRNA in E. coli is also an indirect process involving the five proteins of the Tus 454 system to transfer sulfur from the desulfurase to MnmA (Black & Santos, 2015a;Ikeuchi et al., 455 2006;Shigi, 2014Shigi, , 2018. Similarly, the cysteine desulfurase of S. cerevisiae (called Nfs1) is an 456 IscS homolog and accomplishes the same feat through four sulfur transferases (Nakai et al., 457 2004;Shigi, 2018). While we cannot rule out the possible presence of intermediate sulfur 458 transferases in P. falciparum, homology searches failed to detect them. The tRNA thiolation 459 mechanism found in B. subtilis is most similar to what we have observed in P. falciparum; a 460 SufS paralog (YrvO) transfers sulfur directly to MnmA (Black & Santos, 2015a). The unique 461 feature of the P. falciparum system is that SufS has dual activity and is essential for two 462 metabolic pathways, while B. subtilis YrvO is dedicated to tRNA thiolation (Black & Santos, 463 2015a). 464 In this work and a previous publication (Swift et al., 2022) we defined the roles of the 465 five known FeS-dependent proteins and six of the seven proteins involved in the SUF FeS 466 synthesis pathway (SufB was not studied because we do not have a way to target the apicoplast 467 genome). The combined information supports a model in which SufS is needed for both FeS 468 biosynthesis and tRNA thiolation in the apicoplast. In blood-stage parasites, the SUF pathway is 469 required solely for providing FeS cofactors to enable isoprenoid precursor synthesis, but FeS 470 cofactors should also be required for lipoic acid and fatty acid biosynthesis in mosquito-stage 471 and liver-stage parasites (Akuh et al., 2022;Shears et al., 2015). All stages of parasite 472 development may require tRNA modifications due to the role they play in codon recognition 473 (Urbonavičius et al., 2001) which may have elevated importance in the apicoplast due to the 474 unusually small number of only 24 tRNAs (Wilson et al., 1996). The dual roles of SufS may be a 475 feature of the apicoplasts found in other pathogens such as T. gondii. Deletion of T. gondii SufS 476 or MnmA results in apicoplast defects and parasite death (Pamukcu et al., 2021;Yang et al., 477 2022 PfMev (Swift et al., 2020) and PfMev attB (Swift et al., 2021). Both parasite lines can generate 484 isoprenoid precursors from an engineered cytosolic mevalonate-dependent pathway (Swift et al., 485 2020;Swift et al., 2021). In the presence of mevalonate, these parasite lines can replicate 486 normally even in the absence of an intact apicoplast. The PfMev attB parasite line has an attB site 487 in the P230p locus (Swift et al., 2021). In both parasite lines, the apicoplast is labeled by the 488 super-folder green fluorescent protein (api-SFG) which has been codon-modified for expression 489 in P. falciparum (Roberts et al., 2019). The api-SFG reporter contains the signal and transit 490 peptide (first 55 amino acids) of the P. falciparum acyl-carrier protein (ACP) appended to the N-491 terminal end of SFG to direct trafficking to the apicoplast (Swift et al., 2020;Swift et al., 2021). Cultures were incubated at 37°C and maintained in 25 cm 2 gassed flasks (94% N2, 3% O2, and 499 3% CO2). 500 501

Generation of P. falciparum plasmid constructs for gene deletion 502
We employed Cas9-mediated gene editing to delete genes of interest (Figure 1-figure  503 supplement 1). For gene deletion, the pRS repair plasmid (Swift et al., 2020) in combination 504 with the pUF1-Cas9 plasmid (Ghorbal et al., 2014) were used. Alternatively, the pRSng (Swift et 505 al., 2020) or pRSng(BSD) repair plasmid (Swift et al., 2022) in combination with the pCasG-506 LacZ plasmid  were used for targeted gene deletion. The plasmids used for 507 generation of the gene deletion lines are listed in Supplementary table 3.  508 To generate deletion constructs, ~300-500 bp homology arms (HA) were amplified from 509 P. falciparum NF54 genomic DNA with HA1 and HA2 forward and reverse primers 510 (Supplementary table 4). The HA1 and HA2 amplicons were inserted into the NotI and 511 NgoMIV restriction sites, respectively, of the repair plasmids by In-Fusion (Clontech 512 Laboratories, CA, USA) ligation independent cloning (LIC). The gRNA sequences 513 (Supplementary table 4) were inserted as annealed oligonucleotides into the BsaI sites of 514 pCasG-LacZ plasmids by LIC to generate pCasG-GOI plasmid or into the pRS plasmid. All the 515 plasmids were sequenced to confirm sequence fidelity. All the restriction enzymes used in this 516 section were sourced from New England Biolabs Inc, MA, USA. 517 518

Generation of plasmid constructs for knockdown and epitope tagging of Pf MnmA 519
To generate the knockdown construct of the endogenous Pf MnmA, we created the pKD-520 mnmA-2xFLAG-10xapt plasmid. This plasmid also allowed us to tag Pf MnmA C-terminally 521 with 2xFLAG. To make this plasmid, ~ 300-400 bp HA1 and HA2 of the pfmnmA gene were 522 PCR amplified with MnmAKD.HA1F + MnmAKD.HA1R and MnmAKD.HA2F + 523 MnmAKD.HA2R, respectively from P. falciparum NF54 genomic DNA, (Supplementary table 524 4). These HA PCR fragments were fused together to generate a combined HA2-HA1 fragment 525 with MnmAKD.HA2F and MnmAKD.HA1R in an additional PCR step. The HA2 and HA1 526 fragments in the combined HA2-HA1 fragment are separated by two EcoRV sites to facilitate 527 linearization of the plasmid. The HA2-HA1 fragment was inserted into the AscI and AatII sites of 528 the pKD-2xFLAG-10xapt plasmid to generate the pKD-mnmA-2xFLAG-10xapt plasmid by LIC. 529 The pKD-2xFLAG-10xapt plasmid was generated by replacing the 3xHA tag of the pKD 530 plasmid reported elsewhere . To replace the 3xHA tag with the 2xFLAG 531 tag, the FLAG tag sequence was synthesized as oligonucleotides (Supplementary table 4), 532 annealed, and then inserted into the AatII and PspMOI sites of the pKD plasmid. Prior to 533 transfection, the pKD-mnmA-2xFLAG-10xapt plasmid was linearized with EcoRV. The 534 MnmA.KD gRNA sequence (Supplementary table 4) was inserted as annealed oligonucleotides 535 into the BsaI sites of the pCasG-LacZ plasmid by LIC to generate the pCasG-mnmAKD plasmid. 536 537 Generation of plasmid constructs for B. subtilis mnmA and mnmA-yrvO knock-in 538 To generate the bsmnmA knock-in plasmid, we amplified the bsmnmA gene from B. 539 subtilis genomic DNA with the following primer pair: MnmA.BspEI.InF.F and 540 MnmA.BsiWI.InF.R. The amplified product was used to replace the EcDPCK locus from the 541 pCLD-EcDPCK-mCherry-apt (Swift et al., 2021) plasmid using the BspEI and BsiWI cloning 542 sites to generate pCLD-bsmnmA-mCherry-apt. For enhanced stability of the aptamer system, we 543 replaced the existing 10x-aptamer array (apt) with a redesigned 10x-aptamer array (10xapt) that 544 prevents aptamer loss . First, the 10xapt DNA was amplified with the 545 MY.Apt.PspOMI.F and MY.Apt.XmaI.R primer pair from a pKD plasmid (Rajaram et al., 546 2020). The apt locus was removed from the pCLD-bsmnmA-mCherry-apt plasmid by digestion 547 with PspOMI and XmaI, followed by insertion of the 10xapt PCR product using the same cloning 548 sites by LIC resulting in pCLD-bsmnmA-mCherry-10xapt plasmid. The fidelity of the Bs mnmA 549 and 10xapt sequence of the pCLD-bsmnmA-mCherry-10xapt was confirmed by sequencing. 550 In B. subtilis, the genes mnmA and yrvO are 31 bp apart from each other and are co-551 transcribed (Black & Santos, 2015a). Hence, we decided to generate the pCLD-bsmnmA-yrvO-552 mCherry-10xapt plasmid with bsmnmA and bsyrvO genes fused in one cassette with a 15 bp 553 linker. To make the fusion gene, we first amplified mnmA with MnmA.BspEI.InF.F and 554 MnmA.Link.R, and yrvO with YrvO.Link.F and YrvO.BsiWI.InF.R primer pairs. These two 555 PCR products were stitched together by PCR amplification (bsmnmA-yrvO, For generating the ∆sufC/sufD line, RBCs were transfected with the pCasG-sufC and 578 pRSng(BSD)-sufC plasmids as described above. Synchronized ∆sufD parasites were added to the 579 transfected RBCs and cultured in medium with 50 µM mevalonate. The transfectants were 580 selected with 2.5 µg/mL blasticidin (Corning Inc, NY, USA), 1.5 μM DSM1, 2.5 nM WR99210, 581 and 50 µM mevalonate, after which the culture was maintained in complete medium containing 582 2.5 nM WR99210, and 50 µM mevalonate until parasites appeared. Upon parasite appearance, 583 the culture was maintained in complete medium with 2.5 µg/mL blasticidin, 2.5 nM WR99210, 584 and 50 µM mevalonate. 585 For generation of the mnmA-flag parasite lines, the linearized pKD-mnmA-2xFLAG-586 10xapt and pCasG-mnmAKD plasmids were co-transfected into RBCs as mentioned above. 587 Following transfection, these RBCs were mixed with PfMev parasites and selected with 2.5 588 µg/mL blasticidin and 1.5 μM DSM1 along with 0.5 µM anhydrous tetracycline (aTc, Cayman 589 Chemical, MI, USA) for seven days. After initial selection for seven days, this culture was 590 grown in complete medium with aTc until parasite reappearance. Upon parasite reappearance, 591 the culture was switched to and maintained in complete medium containing 2.5 µg/mL 592 blasticidin and 0.5 µM aTc. 593 To generate the bsmnmA + and bsmnmA-yrvO + transgenic parasite lines, either the pCLD-594 bsmnmA-mCherry-10xapt or pCLD-bsmnmA-yrvO-mCherry-10xapt plasmids were co-595 transfected into RBCs with the pINT plasmid (Nkrumah et al., 2006) encoding the 596 mycobacteriophage integrase (this integrase mediates attP/attB integration into the target genome 597 locus). Transfected RBCs were mixed with PfMev attB parasites (Swift et al., 2021) and cultured 598 with 2.5 µg/mL blasticidin and 0.50 µM aTc for seven days. After seven days, these cultures 599 were grown in complete medium with aTc until parasites were observed, at which point the 600 cultures were maintained in complete medium containing 2.5 µg/mL blasticidin and 0.50 µM 601 aTc. 602 The bsmnmA + ∆mnmA, bsmnmA-yrvO + ∆mnmA, and bsmnmA-yrvO + ∆sufS transgenic 603 parasite lines were generated with the same Cas9 and pRSng repair plasmids that were used to 604 generate the ∆mnmA, and ∆sufS transgenic lines. For bsmnmA + ∆mnmA, bsmnmA-yrvO + ∆mnmA, 605 and bsmnmA-yrvO + ∆sufS, medium supplemented with 1.5 μM DSM1, 2.5 nM WR99210, 1.25 606 µg/mL blasticidin, and 0.50 µM aTc was used for the initial seven days of selection, after which 607 the cultures were switched to growth medium containing blasticidin and aTc. Upon parasite 608 appearance, all cultures were maintained in medium containing WR99210, blasticidin, and aTc. 609 The bsmnmA-yrvO + ∆sufS transgenic parasite line was supplemented with 50 µM mevalonate in 610 addition to WR99210, blasticidin, and aTc. The bsmnmA + ∆mnmA line was difficult to generate 611 with only one successful line out of four attempts. Between two and eight parasite lines from 612 independent transfections were obtained for all other gene deletions (Supplementary table 2). 613 614

Confirmation of gene knockout, C-terminal tagging, and gene knock-in 615
Lysates from parasite cultures were prepared from the transgenic parasite lines by 616 incubating at 90 °C for 5 min. These lysates were used as the template for all genotype 617 confirmation PCRs. For confirmation of gene knockouts, the 5'-and 3'-end of the disrupted (∆5' 618 and ∆3', respectively) and native gene loci (5' and 3', respectively) were amplified with 619 corresponding primers (Supplementary table 4). Expected amplicons for confirmation PCRs 620 are provided in Figure 1-figure supplement 1C. To confirm the successful C-terminal tagging 621 of MnmA and insertion of the aptamer array at the 3'UTR of MnmA, the 5'-and 3'-end of 622 modified genes (∆5' and ∆3', respectively) and the native gene locus (C) were amplified with 623 corresponding primers. The expected amplicon sizes for these PCR products are provided in 624 Hamamatsu, Japan) using a 100x/1.4 NA lens. A series of images were taken spanning 5 µm 690 along the z-plane with 0.2 µm spacing. An iterative restoration algorithm using the Volocity 691 software (PerkinElmer, MA, USA) was used to deconvolve the images to report a single image 692 in the z-plane. 693 The degree of colocalization between the red channel (anti-ACP or mCherry) and green 694 channel (anti-FLAG or api-SFG) signals was determined with Volocity software. The fluorescent 695 intensity thresholds were set using the region of interest (ROI) tool using Volocity. To set the 696 thresholds, the fluorescence intensity of a region of the cell with no staining for either signal 697 (background) was used. To measure the degree of colocalization, the Manders' coefficient (M1) 698 (Manders et al., 1993)was determined. M1 is defined as the percentage of total pixels from the 699 test channel (anti-FLAG, mCherry) that overlaps with the percentage of total pixels from the 700 organellar marker channel (anti-ACP, api-SFG). A value of M1=1 denotes perfect colocalization, 701 while M1=0 denotes no colocalization (Manders et al., 1993). Mean M1 (±standard deviation) 702 values were obtained by analyzing multiple images from at least two independent biological 703 replicates. 704 705 Parasite growth assay 706 Parasite growth was monitored using an Attune Nxt Flow Cytometer (Thermo Fisher 707 Scientific, MA, USA) as previously described (Swift et al., 2020;Tewari et al., 2022). For 708 determining the growth dependence on mevalonate, cultures were seeded in the presence or 709 absence of 50 µM mevalonate at 0.5% parasitemia and 2% hematocrit in a total volume of 250 710 µL, in quadruplicate for each condition. Parasite growth was monitored every 24 h over four 711 days following SYBR green I (Invitrogen, CA, USA) staining. For the growth curves shown in 712

Conflict of interests 738
The authors declare that they have no conflict of interest. 739  Table  779 showing the primer pairs and the template used for each PCR reaction. (C) Table showing the 780 expected amplicon sizes for the genotyping PCR reactions shown in Figures 1(B), 1(F), 2(A), 781