Experimentally engineered mutations in a ubiquitin hydrolase, UBP-1, modulate in vivo resistance to artemisinin and chloroquine in Plasmodium berghei

As resistance to artemisinins (current frontline drugs in malaria treatment) emerges in south East Asia (SEA), there is an urgent need to identify the genetic determinants and understand the molecular mechanisms underpinning such resistance. Such insights could lead to prospective interventions to contain resistance and prevent the eventual spread to other malaria endemic regions. Artemisinin reduced susceptibility in SEA has been primarily linked to mutations in P. falciparum Kelch13, which is currently widely recognised as a molecular marker of artemisinin resistance. However, 2 mutations in a ubiquitin hydrolase, UBP-1, have been previously associated with artemisinin resistance in a rodent model of malaria and some cases of UBP-1 mutation variants associating with artemisinin treatment failure have been reported in Africa and SEA. Here, CRISPR-Cas9 genome editing and pre-emptive drug pressures was used to test these artemisinin resistance associated mutations in UBP-1 in P. berghei sensitive lines in vivo. The data demonstrate that the V2721F UBP-1 mutation results in artemisinin resistance and some low-level resistance to chloroquine, while the V2752F mutation results in high-level resistance to chloroquine and moderate resistance to artemisinins. Genetic reversal of the V2752F mutation restored chloroquine sensitivity in these mutant lines while simultaneous introduction of both mutations could not be achieved and appears to be lethal. Interestingly, these mutations carry a detrimental growth defect, which would possibly explain their lack of expansion in natural infection settings. This is the first independent, direct experimental evidence on the role of UBP-1 in artemisinin and chloroquine resistance under in vivo conditions.


Introduction 52
Artemisinins (ARTs) in artemisinin combinational therapies (ACTs) remain the mainstay of malaria 53 treatment globally and thus far remain mostly effective in sub-Saharan Africa where most of the 54 disease burden occurs (WHO, 2018). However, ART (and even ACT) resistance has emerged in SEA 55 with a risk of spreading which is seriously threatening recent gains achieved in malaria control 56 (Hamilton et al., 2019, Dondorp et al., 2009. ART resistance is thought to be primarily conferred by 57 specific mutations in the Plasmodium falciparum Kelch 13 (PfKelch13) gene, and such mutations are 58 currently almost endemic in most parts of SEA (Mbengue et al., 2015, Ashley et al., 2014, WHO, 59 2018. Phenotypically, these mutations are associated with delayed parasite clearance rates in vivo 60 and reduced susceptibility of ring stage parasites in vitro in ring stage survival assays (RSA) 61 (Witkowski et al., 2013, Dondorp et al., 2009. Interestingly, the prevalence of PfKelch13 mutations 62 remains low outside SEA (Menard et al., 2016) where the few observed PfKelch13 polymorphisms in 63 sub-Saharan Africa do not associate with treatment failure and/or delayed parasite clearance rates 64 (Sutherland et al., 2017). Moreover, large-scale genome wide association studies have revealed that it was found that the UBP-1 V2728F mutation was common in the ART, CQ (30mg/kg) and 86 mefloquine resistant lines while the V2697F mutation only fixated upon artesunate selection 87 (Henriques et al., 2013, Hunt et al., 2010, Hunt et al., 2007. Due to the complexity of the selection 88 procedure with multiple drugs, it has been difficult to confidently associate these UBP-1 mutations 89 with ART and CQ resistance in the absence of appropriate reverse genetics approaches. More 90 interestingly, UBP-1 mutation variants have been associated with ARTs decreased effectiveness in 91 Africa and some parts of Asia (Henriques et al., 2014, Adams et al., 2018, Cerqueira et al., 2017, 92 Borrmann et al., 2013. 93 In our present study, we have successfully engineered UBP-1 candidate mutations in an independent 94 rodent model, P. berghei using a CRISPR-Cas9 genome editing system. We provide a direct causal 95 link to the ART and CQ resistance profiles of these mutant lines both in vitro and in vivo. We have 96 also characterised their relative fitness as compared to the wildtype non-mutant parasites. 97 98 DNA from the vectors described above was transfected by mixing with Nycodenz purified schizonts 164 and electroporated using the Amaxa Nucleofector Device II program U-o33 as previously described 165 (Philip et al., 2013). Parasites were then immediately IV injected into a tail vein of mice. Positive 166 selection of transfected parasites was commenced 24 hours later by inclusion of pyrimethamine 167 (Sigma) in drinking water. 168

Genotype analysis of mutant lines 169
Blood was collected from parasite infected mice by cardiac puncture under terminal anaesthesia and 170 lysed by resuspension in 1X E-lysis buffer (Thermo). Parasite genomic DNA was extracted using the 171 Qiagen DNeasy Blood and Tissue kit according to manufactures' instructions. Genotype analysis of 172 the transfected or cloned parasite lines was analysed, initially by a dual PCR-RFLP. PCR using exterior 173 primers (GU4894+GU4895 or GU5186+GU4895) was used to amplify fragments from the DNA of the 174 mutant lines followed by restriction digests with either BseYI or SnaBI restriction enzymes to verify 175 successful editing of the UBP-1 locus. Further confirmation of the mutations was carried out by 176 Sanger DNA sequencing. 177

P. berghei in vitro culture and drug susceptibility assays 178
For in vitro maintenance of P. berghei, cultures were maintained for one developmental cycle using 179 a standardised schizont culture media containing RPMI1640 with 25mM hypoxanthine, 10mM 180 sodium bicarbonate, 20 % foetal calf serum, 100U/ml Penicillin and 100μg/ml streptomycin. Culture 181 flasks were gassed for 30 seconds with a special gas mix of 5% CO2, 5% O2, 90% N2 and incubated 182 for 22-24 hours at 37 0 C with gentle shaking, conditions that allow for development of ring stage 183 parasites to mature schizonts. Drug assays to determine in vitro growth inhibition during the 184 intraerythroctic stage were performed in these standard short-term cultures as previously described 185 (Franke-Fayard et al., 2008). Briefly, 1 ml of infected blood with a non-synchronous parasitaemia of 186 3-5% was collected from an infected mouse and cultured for 22-24 hours in 120 ml of schizont 187 culture media. Schizonts were enriched from the cultures by Nycodenz density flotation as 188 previously described (Philip et al., 2013) followed by immediate injection into a tail vein of a naive 189 mouse. Upon IV injection of schizonts, they immediately rupture with resulting merozoites invading 190 new red blood cells within minutes to obtain synchronous in vivo infection containing >90% rings 191 and a parasitaemia of 1-2%. Blood was collected from the infected mice 2 hours post-injection and 192 mixed with serially diluted drugs in schizont culture media in 96 well plates at a final haematocrit of 193 0.5% in a 200µl well volume. Plates were gassed and incubated overnight at 37 0 C. After 22-24 hours 194 of incubation, schizont maturation was analysed by flow cytometry after staining the infected cells 195 with DNA dye Hoechst-33258. Schizonts were gated and quantified based on fluorescence intensity 196 on a BD FACSCelesta or a BD LSR Fortessa (BD Biosciences, USA). To determine growth inhibitions 197 and calculate half-inhibitory concentrations (IC50), quantified schizonts in no drug controls were set 198 to correspond to 100% with subsequent growth percentages in presence of drugs calculated 199 accordingly. Dose response curves were plotted in  In vivo drug assays 201 A modified Peters' 4 day suppressive test was employed to assess in vivo drug responses and or 202 resistance profiles in the wildtype and mutant lines as previously described (Vega-Rodríguez et al., 203 2015). Parasitaemia was initiated by IP inoculation of between 10 6 -10 7 parasites followed by three 204 daily consecutive drug doses initiated ~4 hours post inoculation. CQ was prepared at 50mg/ml in 1X 205 PBS and diluted to working stock in 1X PBS while ART was prepared at 12.5mg/ml in a 1:1 mixture of 206 DMSO and Tween ® 80 (Sigma) followed by a ten-fold dilution in sterile water to an injectable working 207 solution. All drugs were delivered by IP and were prepared fresh immediately before injection. 208

berghei in vivo 223
To experimentally demonstrate that UBP-1 mutations confer ART resistance, we introduced P. 224 chabaudi UBP-1 candidate mutation (V2697F and V2728F) equivalents (supplementary Figure 1) in 225 the P. berghei 820 line using the CRISPR-Cas9 system developed and optimised in our lab ( Figure 1A). 226 Two plasmids were initially designed to either introduce the single mutation, V2752F (V2728F P. 227 chabaudi equivalent) or both mutations, V2721F (V2697F P. chabaudi equivalent) and V2752F in an 228 attempt to generate a double mutant ( Figure 1A). Silent mutations to mutate the Cas9 cleavage site 229 and introduce a restriction site (BseYI) were also introduced to prevent re-targeting of mutated loci 230 by Cas9 for the former and diagnosis by RFLP for the latter ( Figure 1A, 1B). Transfections of these 231 plasmids into the 820 line yielded <1% mutants for the V2752F mutant line (G1807, pG945) and 232 ~20% mutants for the V2721F and V2752F double mutant line (G1808, pG946) as confirmed by RFLP 233 analysis (BseYI digestion) of the edited UBP-1 loci ( Figure 1B). Since the efficiency was too low to 234 clone out the mutant lines by serial dilution, we attempted a pre-emptive drug selection with CQ 235 and ART of the G1807 and G1808 lines to examine if selective enrichment of the mutant population 236 could be achieved. Indeed, after infecting mice with the G1808 line and treating for three 237 consecutive days with ART at 20mg/kg, the recrudescent parasite population on Day 9 was enriched 238 to >90% mutant population as confirmed by RFLP analysis ( Figure 1C). Furthermore, CQ at 15mg/kg 239 moderately enriched the same G1808 line, but not to the extent of ART challenge ( Figure 1C). On the 240 contrary, a very low level mutant enrichment of the G1807 line was observed with CQ at 15mg/kg 241 while ART did not produce any enrichment in the same line. Interestingly, cloning of the G1808 ART 242 enriched lines yielded six clones which were all single mutants positive for the V2721F mutation 243 despite coming from a plasmid with donor templates that carried both the V2721F and V2752F 244 mutations ( Figure 1E). This suggests that the V2721F mutation specifically confers ART resistance, 245 which explains the selective enrichment of this population upon ART pressure. These data also 246 suggested that the double mutant parasites could either be lethal or unfit and henceforth 247 outcompeted by single mutation carrying parasites in the presence of drug pressure. Sequence 248 analysis of the G1808 line isolated after CQ challenge at 15mg/kg also revealed an enrichment for 249 the V2721F mutation (supplementary Figure 2) suggesting that despite being principally enriched by 250 ART, the V2721F mutation also modulates some low-level resistance to CQ. Meanwhile, when we 251 challenged the G1807 line (V2752F single mutation) with CQ at a higher dose (30mg/kg), a 252 recrudescent population on Day 10 was enriched to >60% ( Figure 1D). Sanger sequencing of G1808 253 ART enriched and G1807 CQ enriched clones confirmed the presence of the single V2721F and 254 V2752F mutations respectively, as well as the Cas9 cleavage silencing mutations and the silent 255 mutations introducing the BseYI diagnostic restriction site ( Figure 1F). 256

257
The V2721F mutation confers observable phenotypic resistance to ARTs in vivo while the V2752F 258 mutation confers high level resistance to CQ and low-level resistance to ARTs 259 We next quantitated the drug response profiles of the G1808 V2721F and G1807 V2752F cloned lines in 260 vitro and in vivo using DHA, ART and CQ. In short term P. berghei in vitro drug assays, both the 261 G1808 V2721F and G1807 V2752F parasites show no difference in sensitivity to DHA compared to the 262 parental 820 line (Figure 2A, 2B). The lack of decreased drug sensitivity of both lines is consistent 263 with the failure of the standard 72 hour drug assays to differentiate similar Kelch 13 ART resistant 264 parasites from sensitive lines in P. falciparum (Dondorp et al., 2009, Witkowski et al., 2013. 265 Meanwhile, a 1.8 fold increase in IC50 was observed for the G1807 V2752F line when challenged with CQ 266 ( Figure 2C) and not the G1808 V2721F ( Figure 2D). However, rodent malaria parasites offer the 267 advantage of experimental drug resistance assessment in vivo. This approach applied to G1808 V2721F 268 demonstrated that this mutation does indeed confer resistance in vivo to ARTs compared to the 269 parental 820 line. G1808 V2721F parasites survive three consecutive doses of 75mg/kg ART with the 270 recrudescent population appearing on day 9 after last dosing while 820 wildtype parasites are 271 effectively suppressed up to day 17 of follow-up ( Figure 2E). Both the G1808 V2721F and 820 lines 272 survive 45mg/kg dose of ART with the former having a slightly faster recrudescence rate on day 7 273 while the latter recrudesces a day later ( Figure 2E). Both lines remain sensitive to 125mg/kg ART 274 dose with no recrudescence observed up to day 17 ( Figure 2E). In contrast, the G1807 V2752F line is 275 highly resistant to CQ in vivo ( Figure 2F), surviving three consecutive doses at 25mg/kg, with 276 recrudescent parasites coming up on day 4 after the last dose as compared to the parental 820 line 277 and the G1808 V2721F lines which are sensitive and are effectively suppressed up to day 17. 278 Interestingly, the G1807 V2752F line also displays low level ART resistance at 75mg/kg dose, with 279 parasites coming up on day 12, later than the G1808 V2721F line ( Figure 2F). These data confirm that 280 the V2721F mutation confers ART resistance while the V2752F mutation mediates resistance 281 primarily to CQ and to some extent to ARTs. The recrudescence of the wildtype 820 and G1808 V2721F 282 at 45mg/kg ART is also in agreement with our previous findings, that P. berghei is less sensitive to 283 ARTs, especially in the spleen and bone marrow which could be the source of recrudescent infection 284 at relatively lower doses (Lee et al., 2018). 285 286

Growth of parasites carrying UPB-1 V2752F and V2721F mutations is impaired 287
The spread of drug resistance as is the case in most microbial pathogens is partly limited by 288 detrimental fitness costs that accompany acquisition of such mutations in respective drug 289 transporters, enzymes or essential cellular components. The G1807 and G1808 lines carrying UBP-1 290 V2721F and V2752F mutations respectively were each grown in competition with a parental line 291 expressing mCherry in vivo and shown to be characteristically slow growing ( Figure 3A-C). 292 Comparatively, the G1807 V2752F mutation is severely impaired relative to the G1808 V2721F being 293 completely outcompeted by day 8. These data and the earlier failure to generate the double mutant 294 demonstrate that UBP-1 is an important (possibly essential) protein for parasite growth and that 295 acquisition of resistance through mutation of UBP-1 confers mutation specific fitness costs. 296 297 Reversal of the V2752F mutation restores CQ sensitivity in the G1807 V2752F line while introduction 298 of the V2721F in the same line appears to be lethal 299 Drug pressure can select in the long or short term for mutations in sensitive parasite populations 300 that would affect responses to the same drug. To further confirm that the resistance phenotypes 301 observed in our mutant lines were due to the V2721F or V2752F mutations and not possible 302 secondary mutations which may have been acquired during the pre-emptive drug pressure, we 303 attempted to reverse the V2752F mutation to see if wildtype drug phenotypes can be restored in the 304 G1807 V2752F line. Using a CRISPR Cas9 editing strategy similar to the one outlined above, a sgRNA 305 targeting a region ~50bp upstream of the V2721F mutation was designed and cloned in the Cas9 306 expressing vectors ( Figure 4A). ~710bp of donor DNA (GU5189 + GU4787) containing the V2721F (for 307 targeted mutation swap) or both the V2721F and V2752F mutations (for a forced introduction of the 308 V2721F in the G1807 V2752F background) was used to generate the vectors pG963 and pG962 309 respectively ( Figure 4A). Silent mutations mutating the PAM site as well as introducing a second 310 restriction site, SnaBI, for RFLP analysis were also included. Transfection of the G1807 V2752F line with 311 pG963 and pG963 vectors successfully edited the UBP-1 loci generating the G1918 and G1919 lines 312 respectively with >85% efficiency as confirmed by SnaBI RFLP analysis ( Figure 4A). Cloning and 313 sequencing of the G1918 lines revealed successful targeted mutation swap, introducing the V2721F 314 mutation and re-editing of the 2752F to 2752V wildtype genotype ( Figure 4B, 4C). Phenotype 315 analysis of the G1918 clone line revealed a similar ART resistance profile to ART at 75mg/kg as the 316 G1808 V2721F line while CQ sensitivity was completely restored ( Figure 4D). This provided further 317 experimental evidence, that the resistance profiles observed were due to the V2721F or V2752F 318 amino acid substitutions and not the introduced silent mutations or secondary mutations that may 319 have been acquired during the pre-emptive drug exposure. Interestingly, cloning and sequencing of 320 the G1919 ( Figure  limited studies into their detailed involvement in mode of action and or resistance phenotypes such 336 as those observed with ARTs. In this study, using a CRISPR-Cas9 mediated reverse genetics approach; 337 we have provided experimental evidence on the direct involvement of a DUB (UBP-1) in mediating 338 resistance to ART and CQ, more importantly under in vivo conditions. As the debate into the 339 mechanism of action and resistance to ARTs continues, a consensus understanding is converging 340 that ART resistance is more complex as several factors, genetic determinants and possibly 341 mechanisms of action appear to be involved. In P. falciparum, ART resistance is confined to early ring 342 stage parasites which has been translated in laboratory conditions to increased survival in ring stage 343 survival assays (Witkowski et al., 2013). Mutations in Pfkelch13, PfCoronin as well as transient (hypo-344 hyperthermic) temperatures have all been shown to enhance ring stage parasite survival in the RSAs 345 (Henrici et al., 2019b, Straimer et al., 2015, Demas et al., 2018. As demonstrated in this study, ART 346 and CQ resistance can also be mediated by mutations in UBP-1 underscoring a mechanism of cross-347 resistance and some commonality in mode of action between CQ and ART especially relating to 348 haemoglobin digestion and trafficking in malaria parasites (Klonis et al., 2011). 349

350
The UBP-1 V2728F mutation was previously designated as a principle determinant of ART resistance 351 despite its common fixation with mefloquine and higher doses of CQ (Hunt et al., 2010). Contrary to 352 this argument, ART did not enrich for this mutation (V2752F) in our study enriching for the V2721F 353 mutation instead which was fixed with artesunate in P. chabaudi. However, enrichment of the 354 V2752F mutation with a higher dose of CQ was achieved showing that this mutation does indeed 355 modulate resistance to CQ while the V2721F mutation is chiefly responsible for the ART resistance 356 phenotype in the P. berghei model in vivo. Interestingly, drug challenge of these mutant lines in vivo 357 revealed that both mutations also give low-level cross-resistance to ARTs and CQ. This confirms that 358 both of these UBP-1 mutations modulate some form of resistance to both ARTs and CQ albeit to 359 some differing degrees which is, therefore, in strong agreement with previous observations in P. 360 chabaudi (Hunt et al., 2010). This also demonstrates a plurality of pathways to resistance involving 361 the same target. Recently, the exact equivalent UBP-1 mutations in P. falciparum, V3275F and 362 V3306F have been successfully engineered (Henrici et al., 2019a). In P. falciparum UBP-1, the 363 V3275F mutation (V2721F P. berghei equivalent) shows enhanced survival to DHA in RSAs but 364 remains sensitive to CQ. However, unlike in P. berghei, the V3306F (V2752F P. berghei equivalent) 365 showed no enhanced survival to DHA in RSAs or resistance to CQ (Henrici et al., 2019a). Whilst not 366 entirely in agreement with the data reported here, this could be due to limitations in the ability of in 367 vitro assays to fully predict actual drug responses in vivo which these data highlight and has been the 368 concern with Kelch 13 mutations recently (Sa et al., 2018). These observations may also somewhat 369 be confounded by species specific differences in drug responses, pharmacodynamics, modes of 370 action and resistance that in part remain to be fully investigated. For example, previous and original 371 linkage studies in P. chabaudi identified additional mutations in an amino acid transporter (pcaat) as 372 being strongly associated with CQ resistance phenotypes in tandem with UBP-1 mutations (Hunt et 373 al., 2010). Even though this could partly explain the observed in vitro sensitivity of P. falciparum 374 V3275F mutants to CQ, our data suggests that UBP-1 mutations are sufficient to mediate resistance 375 to both ARTs and CQ as the reversal of the V2752F mutation performed in this study, for example, 376 completely restores observable CQ sensitivity. This has provided, therefore, additional independent 377 evidence on the direct causative role of UBP-1 mutations in mediating resistance not just to ARTs, 378 but CQ as well. The study also illustrates the potential of the P. berghei rodent model in proving 379 causality to antimalarial drug resistance phenotypes under in vivo conditions especially in light of 380 recent reported discrepancies between some in vitro RSA resistance profiles of P. falciparum Kelch 381 13 mutants and actual in vivo phenotypes using the Autos monkey model (Sa et al., 2018).  Figure 4). However, unlike their rodent counterparts which associate with ART 388 resistance, the natural reported E1528D and D1525E mutations occur towards the less conserved N-389 terminus of the protein and outwith the conserved, bioinformatically predicted UBP-1 catalytic 390 domain (Hunt et al., 2007) (supplementary Figure 1). This would suggest that acquisition of the 391 mutations at the well conserved C-terminal in P. falciparum has a potential growth defect as we 392 have observed with P. berghei in this study. However, as these upstream mutations are not 393 conserved between P. falciparum and P. berghei UBP-1, we cannot test the hypothesis in this model. 394 In fact, P. falciparum UBP-1 is highly polymorphic with over 480 reported single nucleotide 395 polymorphisms (SNPs) https://plasmodb.org all of which are in the N terminal region. PfUBP-1 has 396 also been recently shown to be undergoing a strong positive selection in SEA (Ye et al., 2019). UBP-1 397 mutations could, therefore, be an independent avenue to which ART or multidrug resistance 398 phenotypes could emerge in endemic regions like has been seen in Africa (Ghana and Kenya), 399 without actually requiring a permissive genetic background as seems to be the current landscape 400 with Kelch 13 mutations. However, there are constraints upon the evolution of drug resistance and 401 UBP-1. Whilst these data confirm that a single protein that does not transport drugs can mediate 402 resistance to two quite distinct drug entities, it was not possible to generate a P. berghei line that 403 simultaneously contained the two UBP-1 drug resistance mutations examined in this study. 404

405
In yeast, UBP-1 localises to the endoplasmic reticulum playing roles in protein transport specifically 406 internalisation of substrates across membranes (Schmitz et al., 2005). Mutations in UBP-1 could, 407 therefore, modulate endocytosis of important essential host derived products such as haemoglobin 408 to the digestive vacuole in a similar manner thereby reducing exposure of the parasite to activated 409 drug for both ARTs and CQ. Interestingly, mutations in the AP2 adaptor complex that is involved in 410 clathrin mediated endocytosis have also been implicated in ART resistance in rodent malaria 411 parasites (Henriques et al., 2013). One of the AP2 adaptor complex mutation (I592T) has been 412 recently engineered in P. falciparum and has been shown to enhance ring stage parasite survival in 413 RSAs (Henrici et al., 2019a). This further suggests that inhibition of the endocytic trafficking system is 414 a possible generic mechanism for the parasites to survive lethal doses of drugs that require transport 415 and activation in the digestive vacuole. This would further explain the multidrug resistance 416 phenotype observed with the UBP-1 mutations in P. chabaudi and P. berghei in this study. 417 Acquisition of the V2728F mutation in P. chabaudi was structurally predicted to reduce 418 deubiquitination (Hunt et al., 2007). In such a situation, the cellular increase in ubiquitinated 419 proteins would be anticipated to positively feedback to the cellular machinery to rapidly degrade 420 protein substrates at the 20s proteasome promoting a non-specific and rapid protein turnover or 421 impaired substrate trafficking. This would result in generally slow growing parasite with reduced 422 expression of, for example, multi-drug resistance transporters as well as reduced endocytosis of 423 host-derived products like haemoglobin, which would in turn modulate parasite responses to these 424 drugs. More recently, functional studies have revealed that PfKelch13 (known determinants of ART 425 resistance) localises to the parasite cytosomes and play a role in haemoglobin trafficking (Yang et al., 426 2019). Consequently, PfKelch13 mutations have been shown to lead to a partial loss of PfKelch13 427 protein function leading to decreased haemoglobin trafficking to the parasite digestive vacuole and 428 less DHA activation which in turn mediates parasite survival (Yang et al., 2019). This is indeed in 429 agreement with our hypothesis on the consequences of UBP-1 mutations which in a similar manner 430 could impair trafficking of haemoglobin leading to less activation of ARTs and CQ. The experimental 431 validation of these mutations in ART and CQ resistance, therefore, provide an additional 432 understanding of drug resistance in malaria parasites. Further work, to identify and characterise 433 proteins and pathways which interact with UBP-1 in modulating ART and CQ resistance will hopefully 434 further illuminate the unique involvement of the parasite UPS network in ART and CQ resistance 435 possibly through direct comparison with Kelch 13 mutation equivalents. Furthermore, the P. berghei 436 model provides a useful system in which to investigate the interplay and impact of simultaneous 437 mutations of both K13 and UBP-1 in vivo as well as assess whether PfKelch13 mutations would 438 modulate resistance to CQ under in vivo conditions. 439 In conclusion, the work presented here provides direct experimental evidence for the involvement 440 of conserved mutations in a polymorphic ubiquitin hydrolase protein that serves as a nexus for 441 resistance to two very diverse classes of drugs. The findings also underscore the potential difficulties 442 that in vitro assays may have in appropriately assigning mutant parasites with appropriate 443 phenotypes in absence of conclusive in vivo measurements. P. berghei should therefore, be a 444 suitable and adaptable in vivo model for the rapid evaluation and/or genetic engineering of 445 mutations associated with human-infectious Plasmodium drug resistance observed in the field for 446 concurrent assigning of drug resistance phenotypes under both in vitro and in vivo conditions. 447 Successful editing in the transfected parasites was observed on day 12 after transfection and 631 pyrimethamine drug selection. RFLP (BseYI digestion) analysis of the transformed lines PCR products 632 (primers GU4894 + GU4895, 713bp) revealed <1% and ~20% efficiency for the G1807 and G1808 633 lines respectively as indicated by 2 distinct bands (442bp, 271bp) as compared to 713bp bands in the 634 parent 820 line. C. pre-emptive challenge of the G1807 and G1808 lines with ART and CQ at 20mg/kg 635 and 15mg/kg respectively and RFLP analysis of recrudescent parasites. Mice were infected with ~2 x 636 10^7 parasites IP on day 0. Treatment was started ~4 hours post infection by IP for three 637 consecutive days. Parasitaemia was monitored by microscopy analysis until recrudescence was 638 observed. D. Pre-emptive challenge of the G1807 line with higher doses of CQ and RFLP (BseYI 639 digestion) analysis of the G1807 recrudescent population after challenge with 30mg/kg CQ. E. RFLP 640 analysis of the cloned G1808 and G1807 ART and CQ challenged recrudescent parasites. F. DNA 641 sequencing confirming successful nucleotide editing for the G1807 clone2 and G1808 clone1 line. 642 The top sequence represents the 820 WT unedited sequence with positions for sgRNA, protospacer 643 adjacent motif (PAM) and V2721F or V2752F mutations indicated. The bottom sequence illustrates 644 the nucleotide replacements at V2721F or V2752F mutation locus and silent mutations to prevent 645 Cas9 retargeting as well as introduce the BseYI restriction site for RFLP analysis in the G1807 V2752F 646 and G1808 V2721F lines. relative to the wildtype 820 line. CQ dose response curves and IC50 comparisons of the G1807 V2752F 655 (C) and G1808 V2721F (D) lines relative to the wildtype 820 line. Significant differences between mean 656 IC50s or IC50 shifts were calculated using the paired t-test. Significance is indicated with asterisks; 657 *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns; not significant. Modified Peters' 4 day 658 suppressive test to monitor resistance to ART and CQ in vivo in the G1808 V2721F (E) and the 659 G1807 V2752F (F) mutant lines. Groups of three mice were infected with 1 x 10 6 parasites on day 0. 660 Treatment started ~1.5 hours later with indicated drug doses every 24 hours for three consecutive 661 days (treatment days shown by arrows). Parasitaemia was monitored by microscopy analysis of 662 Giemsa stained blood smears up to day 18. Error bars are mean parasitaemia from 3 mice groups.