PfGCN5, a global regulator of stress responsive genes, modulates artemisinin resistance in Plasmodium falciparum

Plasmodium falciparum has evolved resistance to almost all front-line drugs including artemisinins, which threatens malaria control and elimination strategies. Oxidative stress and protein damage responses have emerged as key players in the generation of artemisinin resistance. In this study, we show that PfGCN5, a histone acetyltransferase, binds to the stress responsive and multi-variant family genes in poised state and regulates their expression under stress conditions. We have also provided biochemical and cellular evidences that PfGCN5 regulates stress responsive genes by acetylation of PfAlba3. Furthermore, we show that upon artemisinin exposure, genome-wide binding sites for PfGCN5 are increased and it is directly associated with the genes implicated in artemisinin resistance generation like BiP and TRiC chaperone. Moreover, inhibition of PfGCN5 in artemisinin resistant parasites, Kelch13 mutant, K13I543T and K13C580Y (RSA∼ 25% and 6%, respectively) reverses the sensitivity of the parasites to artemisinin treatment indicating its role in drug resistance generation. Together, these findings elucidate the role of PfGCN5 as a global chromatin regulator of stress-responses with potential role in modulating artemisinin drug resistance, and identify PfGCN5 as an important target against artemisinin resistant parasites. Author Summary Malaria parasites are constantly adapting to the drugs we used to eliminate them. Thus, when we use the drugs to kill parasites; with time, we select the parasites with the favourable genetic changes. Parasites develop various strategies to overcome exposure to the drugs by exhibiting the stress responses. The changes specific to the drug adapted parasites can be used to understand the mechanism of drug resistance generation. In this study, we have identified PfGCN5 as a global transcriptional regulator of stress responses in Plasmodium falciparum. Inhibition of PfGCN5 reverses the sensitivity of the parasites to the artemisinin drug and identify PfGCN5 as an important target against artemisinin resistant parasites.

159 copy variant genes (var, rifin and stevor), we found that PfGCN5 exhibits specific binding to 160 antigenic variation genes, as shown in the representative example for Chromosome 1 from 161 the trophozoite stage ( Fig 1B). These results corroborate our earlier findings where we have 162 shown that stress and stimuli dependent genes show enrichment of histone modifications at 163 the centre and towards the 3'-end of the genes [5], while genes belonging to other 164 housekeeping functions demonstrate uniform distribution of histone modifications.

165
166 Further, to validate the peaks obtained in ChIP-seq, we performed ChIP-qPCR on randomly 167 selected genomic loci enriched for PfGCN5 and confirmed its binding ( Fig 1C). Lastly, gene 168 ontology (GO) analysis of PfGCN5 bound genes indicated enrichment of terms such as 169 antigenic variation, stress response to heat, and response to unfolded proteins (Fig 1D),

174
175 PfGCN5 is not a general transcription co-activator; it is specifically associated with 176 stress/stimuli associated genes 177 Next, we investigated how PfGCN5 binding relates to transcriptional activity of a gene at the 178 trophozoite stage. We systematically calculated the enrichment levels of PfGCN5 and 179 H3K9ac, a general activation mark, at the gene body of all P. falciparum genes and compared 180 it to the relative expression levels of genes as evaluated by RNA-seq-based transcriptomic 181 analysis. As expected, we observed a positive correlation between H3K9ac enrichment and 182 the expression status of the downstream gene (Fig 2A; left panel). On the other hand, we did 183 not observe strong positive correlation between PfGCN5 gene-body occupancy and the 184 expression of nearby genes (Fig 2A; right panel). Genes with either high or low gene 185 expression levels (outlier points for log2 read density) showed high PfGCN5 occupancy ( Fig   186 2A), suggesting that PfGCN5 binds to both active and suppressed/poised genes. In order to 187 confirm this, we compared the expression levels of genes bound by PfGCN5 and contrasted 188 them with the expression of all the P. falciparum genes. The expression level of PfGCN5 189 bound genes spreads from high expression to low expression values ( Fig 2B) indicating its 190 presence on expressed as well as suppressed genes. Interestingly, many of the PfGCN5 bound 191 genes have both activation (H3K9ac) and repression (H3K9me3) marks (Fig 2C), indicating 192 suppressed yet poised for future activation. Thus, absence of global correlation with 193 transcription and occupancy on suppressed/poised as well as active genes, suggest that 194 PfGCN5 is not a general transcriptional co-activator rather it may specifically regulate stress 195 responsive genes in P. falciparum.

196
197 PfGCN5 is a specific regulator of stress responsive genes 198 Next, we decided to look into the role of PfGCN5 during different stress conditions. 199 Synchronized ring stage parasites were exposed to two different physiological stress  Table). To further dissect the functional 215 correlation between transcriptome deregulation and recruitment of PfGCN5 under different 216 stress conditions, we performed ChIP-sequencing for PfGCN5 using -peptide antibody 217 during both temperature and artemisinin stress conditions. Notably, most of these genes 218 which are bound by PfGCN5, are upregulated under artemisinin and temperature stress 219 conditions (Fig 3D), indicating that PfGCN5 is associated with the activation of stress 220 responsive genes.  Table), representing 292 four major pathways namely chromatin assembly, response to stimuli, metabolic pathways 296 Alba proteins are also known to play important role in stress response pathways in higher 297 eukaryotic system [51, 52]. As PfGCN5 was found to be majorly associated with stress 298 responsive genes, we decided to further study PfGCN5 and PfAlba3 interaction. 299 300 First, to validate the interaction of PfGCN5 with PfAlba3, we cloned, expressed and purified 301 recombinant His-tagged PfAlba3. As we were unable to express the full length PfGCN5 due 302 to its large size, we cloned and overexpressed GST-tagged HAT and bromo domains of 303 PfGCN5 (S1B Fig). Surprisingly, in vitro binding assay using recombinant His-tagged 304 PfAlba3 and GST-tagged PfGCN5-HAT did not show any interaction (Fig 5A). Thus, it is 305 possible that PfAlba3 either interacts with PfGCN5 outside of the HAT and bromodomain or 306 it interacts indirectly with the PfGCN5 complex in vivo. Next, we performed 307 immunoprecipitation using PfGCN5 peptide antibody and looked for PfAlba3 as its 308 interacting partner in the pulled down fractions by Western blotting. As shown in Fig 5B, 309 PfGCN5 co-elutes with PfAlba3 indicating an interaction with the PfGCN5 complex.
310 Furthermore, immunofluorescence analysis suggested a partial colocalisation of PfGCN5 and 311 PfAlba3 at trophozoite stages of P. falciparum (Fig 5C). Lastly, to understand the 312 physiological role of PfGCN5 and PfAlba3 interaction, we performed in vitro 313 acetyltransferase assays with the PfGCN5 complex and found that PfGCN5 indeed acetylates 314 PfAlba3 (Fig 5D). Together, these data suggest that PfGCN5 interacts with PfAlba3 and 315 mediates its acetylation. 385 Together, it suggests that PfGCN5 may play an important role in drug resistance generation 386 either by directly regulating the expression of the genes important for 387 emergence/maintenance of artemisinin resistance and/or by interacting with various key 388 stress-regulators involved in resistance generation in P. falciparum (Fig 6).  496 Each experiment included technical triplicates and was performed over three independent 497 biological replicates. Primers details for the RT qPCR are given in S4 Table. 498 499 Chromatin Immunoprecipitation 500 Infected RBCs were crosslinked using 1% formaldehyde (Thermo Scientific, 28908) for 10 501 mins at RT. 150 mM glycine was added for quenching the cross-linking reaction. The 502 samples were washed using 1X PBS (chilled) before proceeding with lysis. Sample 503 homogenization was performed using swelling buffer (25 mM Tris pH 7.9, 1.5 mM MgCl2, 504 10 mM KCL, 0.1% NP40, 1 mM DTT, 0.5 mM PMSF, 1x PIC) followed by cell lysis in 505 sonication buffer (10 mM Tris-HCl pH 7.5, 200 mM NaCl, 1 % SDS, 4 % NP-40,1mM 506 PMSF, 1X PIC). Sonication was performed using Covaris S220 to obtain the chromatin size 507 of 200-400 bp. Pre-clearing was performed for 1 hour at 4 0 C using recombinant protein G 508 conjugated sepharose beads with continuous gentle inverting. 30 g purified chromatin was 509 used per antibody (both -HAT and -peptide antibodies) and incubated for 12 h at 4 0 C. 510 Samples were then incubated with saturated Protein G Sepharose beads for 4 hours at 4 0 C. 511 Bound chromatin was finally washed and eluted using ChIP elution buffer (1 % SDS, 0.1 M 512 sodium bicarbonate). Both IP sample and input were reverse crosslinked using 0.3M NaCl 513 overnight at 65 0 C along with RNAse. Proteinase K treatment was performed at 42 0 C for 1 514 hour. Finally DNA was purified using phenol chloroform precipitation. Target sites identified 515 from ChIP sequencing analysis were further validated by ChIP-qPCR using the Biorad SYBR 516 Green Master Mix (Biorad). Primers details for the ChIP-qPCR are provided in S5 Table. 517 Gene ontology was performed using PlasmoDB (www. plasmodb.org). Gene ontology terms 518 along with number of genes in each category are given in S6 Table and S7 Table. 519 520 ChIP-sequencing Library preparation and sequencing  556 Parasites were subjected to heat and therapeutic (artemisinin treatment) stresses for 6 hours 557 from late ring (~17 hrs) to early trophozoite (~23 hrs) stage. Double synchronization was 558 carried out to achieve tight synchronization of parasite stages. Parasites were exposed to a) 559 Heat stress (40 0 C for 6 hours) and b) Therapeutic stress (30 nM artemisinin for 6 hours). 560 561 RNA sequencing and Data analysis 562 Parasites were harvested for RNA isolation after 6 hours of stress induction. Total RNA was 563 isolated using TRIzol reagent according to the protocol. DNAse treated RNA was used for 564 cDNA synthesis. Quality of the RNA was verified using Agilent Bioanalyzer 2100. Three 565 biological replicates were pooled together for performing RNA sequencing. The cDNA 566 libraries were prepared for samples using Illumina TruSeq RNA library preparation kit. 567 Transcriptome sequencing was performed using Illumina NextSeq 500 system (1x150 bp 568 read length) at BioServe Biotechnologies (India) Pvt Ltd. Hyderabad in replicate. Quality 569 control of the RNA-sequencing reads was performed using FASTQC and reads were trimmed 570 based on the quality estimates. The quality verified reads were then mapped onto the 571 reference genome (PlasmoDB_v37) using the HISAT2 software (New Tuxedo Suite). After 572 verification of the mapping percentage, the alignment data (SAM format) was converted into 573 its binary counterpart (BAM format) using samtools. The same step also sorts the aligned 574 reads positionally according to their genomic coordinates, making them easier to process 575 further. In order to quantify the reads mapped onto the genomic features (genes, exons, etc.), 576 the htseq-count feature was used. The count data was then used to perform differential gene 577 expression (DGE) analysis and statistical validation using the Deseq2 package in the R 578 computational environment. MA plot is generated using 'R' software (http://r-project.org/). 579 580 Immunoprecipitation 581 In order to harvest the parasites, infected RBCs were lysed using 0.15% saponin at 37 0 C. 582 Harvested parasites were then lysed using ice cold parasite lysis buffer (20 mM TRIS pH 8.0,