A single point mutation in the Plasmodium falciparum FtsH1 metalloprotease confers actinonin resistance.

The antibiotic actinonin kills malaria parasites (Plasmodium falciparum) by interfering with apicoplast function. Early evidence suggested that actinonin inhibited prokaryote-like post-translational modification in the apicoplast; mimicking its activity against bacteria. However, Amberg Johnson et al. (2017) identified the metalloprotease TgFtsH1 as the target of actinonin in the related parasite Toxoplasma gondii and implicated P. falciparum FtsH1 as a likely target in malaria parasites. The authors were not, however, able to recover actinonin resistant malaria parasites, leaving the specific target of actinonin uncertain. We generated actinonin resistant P. falciparum by in vitro selection and identified a specific sequence change in PfFtsH1 associated with resistance. Introduction of this point mutation using CRISPr-Cas9 allelic replacement was sufficient to confer actinonin resistance in P. falciparum. Our data unequivocally identify PfFtsH1 as the target of actinonin and suggests that actinonin should not be included in the highly valuable collection of 'irresistible' drugs for combatting malaria.

Introduction

Actinonin is an anti-bacterial and anti-parasitic antibiotic derived from streptomycete bacteria (Gordon et al., 1962; Wiesner et al., 2001). In bacteria, actinonin targets peptide deformylase (PDF) (Chen et al., 2000), an enzyme involved in prokaryotic post-translational modification and also present in the relict plastid (apicoplast) of apicomplexan parasites. Actinonin causes defects in malaria parasite apicoplast development (Goodman and McFadden, 2014) and inhibits recombinantly expressed Plasmodium falciparum PDF (PfPDF – PF3D7_0907900) in vitro (Bracchi-Ricard et al., 2001) at concentrations consistent with its anti-parasitic activity, all of which led to the general conclusion that actinonin targets the apicoplast-localized PfPDF in malaria parasites. However, the characteristics of actinonin —particularly the rapid mode of action and the unusual kinetics of apicoplast genome loss —are at odds with how all other drugs targeting apicoplast translation impact the parasite (Amberg-Johnson et al., 2017; Uddin et al., 2018). In a search for the target of actinonin, Amberg-Johnson et al., 2017 used the related apicomplexan Toxoplasma gondii and identified a point mutation in the putative metalloprotease TgftsH1 that confers a 3.5-fold resistance to actinonin. They also showed that actinonin inhibits recombinantly expressed human malaria parasite FtsH1 (PfFtsH1) in vitro at levels comparable to its antimalarial activity (Amberg-Johnson et al., 2017). Moreover, parasites with reduced PfFtsH1 expression were more sensitive to actinonin, all of which prompted the interim conclusion that PfFtsH1, rather than PfPDF, might be the target of actinonin and that PfFtsH1 is a potential new antimalarial target (Amberg-Johnson et al., 2017).

Despite repeated attempts, Amberg-Johnson et al., 2017 were not able to generate actinonin resistant P. falciparum parasites. Interestingly, the residue mutated from asparagine to serine (N805S) in TgFtsH1 identified as conferring actinonin resistance by Amberg-Johnson et al. (Amberg-Johnson et al., 2017) is already a serine in PfFtsH1 (Table 1), which begs the question of whether PfFtsH1 is already ‘resistant’ to actinonin. This might mean that actinonin kills malaria parasites through a mechanism not involving PfFtsH1, perhaps even inhibition of PfPDF. Compounding this uncertainty is a report that PfFtsH1 is localized in the mitochondrion (Tanveer et al., 2013), which is inconsistent with the demonstrated impact of actinonin on the malaria parasite apicoplast (Goodman and McFadden, 2014; Uddin et al., 2018). However, phenotypic evidence from PfFtsH1 knockdowns and changes in post-translational processing of PfFtsH1 in the absence of a functioning apicoplast (Amberg-Johnson et al., 2017) strongly suggest a relationship between PfFtsH1 and the apicoplast. These contradictory findings may result from the of differential targeting of the various processed forms of PfFtsH1 (Tanveer et al., 2013) or stem from the close physical and functional association between the mitochondria and apicoplast in malaria parasites (van Dooren et al., 2005). Given the complexity of FtsH1 processing and localization in P. falciparum, it is unlikley that cell biological studies alone will be able to definitively resolve the issue of whether PfFtsH1 is the primary target of actinonin.

Table 1.

The impact of mutations in ftsh1 on parasite resistance to actinonin.

Parasite	FTSH1 Peptidase Motif (partial amino acid sequence)	Actinonin IC50 (µM)
Tg FTSH1 WT (TGGT_259260)	804 FGRDALSNGASSDI 811	14a
Tg FTSH1 ActR	804 FGRDALSSGASSDI 811	44a
Pf 3D7 FTSH1 (PF3D7_1239700)	481 FGKSETSSGASSDI 494	1.99 (n = 1)b
Pf D10 FTSH1 WT	481 FGKSETSSGASSDI 494	2.0 ± 0.2 (n = 4)
Pf D10 FTSH1 ActR	481 FGKSETSSCASSDI 494	73.3 ± 2.7 (n = 4)
Pf D10 (apicoplast minus)	481 FGKSETSSGASSDI 494	43.1 ± 4.1 (n = 2)c

acalculated from data provided in Amberg-Johnson et al., 2017, b and c are both consistent with previously published data (Goodman and McFadden, 2014; Amberg-Johnson et al., 2017).

To determine if PfFtsH1 is the target of actinonin, we generated P. falciparum parasites with robust resistance to actinonin, identified a point mutation conferring resistance, and recapitulated the resistance phenotype by introducing a single amino acid change using CRISPrCas9 genome editing.

Results and discussion

P. falciparum strain D10 parasites were selected for resistance using stepwise increases in actinonin concentration. Ten million parasites were treated with 2 µM of actinonin, which resulted in no parasites being detectable in the culture by microscopy. Fresh, drug-containing media was regularly provided until parasites were again detectable by microscopy, and normal growth rate had resumed. Drug concentration was then increased two-fold and the process repeated until parasites were growing vigorously in media containing 20 µM actinonin. Both the parasite strain and selection methodology used differ from previous attempts to generate resistance (Amberg-Johnson et al., 2017), which may explain why we obtained resistance where others did not.

Several clones were generated from our actinonin resistant parasite line, and each showed consistent actinonin resistance, with IC50 values 18 to 35-fold higher than the parental line (Table 1, Supplementary file 1a). The clone with the highest level of resistance showed an IC50 of 73.3 µM actinonin (Table 1, Figure 1B,C). We genotyped four actinonin resistant clones and all retained wild type sequences of Pf pdf, Pf formyl-methionyl transferase (Pffmt - PF3D7_1313200), and Pf methionyl amino peptidase (Pfmap - PF3D7_0804400) suggesting that neither PfPDF nor the related apicoplast post-translational protein modifying enzymes are the primary target of actinonin. Similarly, all four actinonin resistant clones retained wild type sequence for PfRING (PF3D7_1405700), another actinonin target candidate (Amberg-Johnson et al., 2017). However, each of the clones harbors a single nucleotide polymorphism in Pfftsh1 that changed amino acid 489 (adjacent TgFtsH1 N805S) from glycine to cysteine (Table 1, Supplementary file 1a), strongly implying that PfFtsH1 is the primary target of actinonin.

Figure 1.

Allelic replacement in Pfftsh1 confers actinonin resistance.

(A) Genomic sequences of parasite lines. Upper line is 3D7 reference sequence with sgRNA (red arrow) and resistance mutation site (dark blue bar) marked. Bottom four lines are genomic sequence traces with shield and resistance mutations highlighted in light blue and predicted changes to amino acid sequence highlighted in yellow (B) Comparison of parasite growth inhibition (IC50) based on the presence of the G489C mutation. (C) Dose response curves of data presented in B. Data presented are the mean of 3–5 biological replicates. Error bars represent the standard error of the mean. P values represent two-tailed, unpaired t-test. (Full statistical analysis available in Supplementary file 1b).

To unequivocally confirm that PfFtsH1 is the primary target of actinonin, and that the G489C mutation is sufficient to confer resistance, we used CRISPr Cas9 mutagenesis to introduce the mutation (with minimal collateral genome disruption) into the native Pfftsh1 gene (Figure 1A). Accordingly, a parasite clone carrying synonymous ‘shield’ mutations in the Pfftsh1 coding sequence designed to prevent ongoing Cas9 cleavage but retaining glycine 489 (rFtsH1G489G) remained sensitive to actinonin (Figure 1B,C, Supplementary file 1b), whereas two independent clones (rFtsH1G489Ca/b) modified to have the G489C mutation (in addition to the ‘shield’ mutations) showed actinonin resistance levels comparable to the actinonin-selected line (wtACTR) (Figure 1B,C, Supplementary file 1b).

Robust actinonin resistance in P. falciparum resulting from the G489C mutation confirms that PfFtsH1 is indeed the primary target of actinonin. That the resistance levels in PfFtsH1 G489C parasites are of the same order of magnitude as that seen in lines that lack an apicoplast (Table 1), strengthens the conclusion that PfFtsH1 has a role in apicoplast biogenesis, the anomalous localization (Tanveer et al., 2013) notwithstanding. The greater levels of resistance achievable through prolonged selection, while modest, suggests that these lines may have acquired other mutations that compensate for reduced PfFtsH1 function and/or alter secondary actinonin targets, such as the other metalloproteases present in the genome (Amberg-Johnson et al., 2017). Our ability to generate resistance to actinonin in a relatively small starting population of P. falciparum parasites means actinonin is not an ‘irresistible’ drug (Cowell and Winzeler, 2018), which tempers enthusiasm for development of PfFtsH1 as an antimalarial target.

Materials and methods

P. falciparum D10 were cultured according to standard protocols (Uddin et al., 2018; Trager and Jensen, 1976). Apicoplast-minus parasites were generated according to previously described methods (Uddin et al., 2018; Yeh and DeRisi, 2011). To generate actinonin resistant parasites, 107 D10 parasites were treated with 2 µM actinonin (Sigma-Aldrich) and cultured until parasites began growing robustly. The actinonin concentration was then increased 2-fold and the culturing repeated until parasites grew normally at 20 µM actinonin. This selection procedure required 10 weeks of constant drug selection to recover resistant parasites and 10 months of selection to develop parasites with the highest levels of resistance. Resistant parasites were cloned by limiting dilution and retested to confirm the resistance phenotypes. Drug effects were assayed after 72 hr of drug exposure using the SYBR Green (ThermoFisher) method (Uddin et al., 2018; Goodman et al., 2007).

Genomic DNA was isolated using 200 µL of parasite culture (Isolate II Genomic DNA kit, Bioline). Candidate actinonin resistance genes were amplified using the primers listed in Supplementary file 1c. CRISPr edited FtsH1 clones were amplified using primers in Supplementary file 1d. Products were purified (Isolate II PCR and Gel kit, Bioline) and Sanger sequenced (Australian Genome Research Facility, Parkville). Alignment and analysis of sequenced genes was done using Sequencher (Gene Codes Corporation, Ann Arbor, MI USA) and Geneious Prime (www.geneious.com).

CRISPr-Cas9 mediated gene-editing utilized pAIO (Spillman et al., 2017) expressing Cas9 and the Pfftsh1-specific sgRNA 5’-GTAAATCAGAAACTAGTAG-3’ inserted according to standard protocols (Ghorbal et al., 2014) . Two allelic replacement vectors—pFtsH1G489G carrying two synonymous ‘shield’ mutations and pFtsH1G489C carrying a further G to T mutation at base 1465 (Figure 1A)—were created by cloning a PCR amplified segment of Pfftsh1 into pGEM-T Easy (Promega). Quickchange XL (Clontech) was used to make sequential modifications for allelic replacement constructs. The shield mutations were introduced first and then the plasmid carrying the confirmed shield mutations was modified to also include the putative resistance mutation (G1465T). All constructs were confirmed by sequencing.

Each allelic replacement vector was linearized by digestion with EcoRI and co-transfected with pAIO-Pfftsh1 using standard transfection methods (Waller et al., 2000). Transfected parasites were selected by including 10 µM DSM-1 (Sigma-Aldrich) in the culture media for 14 days (rFtsH1G489G and rFtsH1G489Ca) or 7 days (rFtsH1G489Cb) followed by 10–14 days of culture without drug until parasites grew normally in culture. Parasites were cloned by limiting dilution and three to five clones of each line were screened for actinonin sensitivity and successful modification of the Pfftsh1 allele. All clones from rFTSH1G489G and rFtsH1G489Ca had the expected gene modifications while only one of five clones from rFtsH1G489Cb did. Actinonin sensitivity was correlated to the presence of the G489C mutation in all clones tested. One clone from each recombinant line was selected for complete characterization of actinonin sensitivity.

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Thank you for submitting your article "A single point mutation in the Plasmodium falciparum ftsh1 metalloprotease confers actinonin resistance" for consideration by eLife. Your article has been reviewed by four peer reviewers, including Jon Clardy as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Gisela Storz as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Sean Prigge (Reviewer #2); Ellen Yeh (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, we are asking editors to accept without delay manuscripts, like yours, that they judge can stand as eLife papers without additional data, even if they feel that they would make the manuscript stronger. Thus the revisions requested below only address clarity and presentation.

Summary:

The original report revised the generally accepted model for actinonin's antimalarial activity from inhibition of a peptide deformylase to inhibition of PfFtsH1, but a drug-resistant PfFtsh1 mutant was not reported. This report describes a drug resistant mutant in PfFtsh1 and supporting experiments that tighten the argument. Taken together, the results establish that FtsH1 is the primary target of actinonin in malaria parasites and show that parasites can generated significant resistance to this drug (18-35x). The reviewers generally found that this report added materially to the original report and tightened up the evidence supporting the original conclusion.

Essential revisions:

1) Introduction. Please provide more background on uncertainty of FtsH1 localization. The authors cite Tanveer et al., 2013 for mitochondrial localization of FtsH1. In Amberg-Johnson et al., 2017, cleavage of FtsH1 dependent on presence of apicoplast also indicates possible apicoplast localization. [related but different reviewer] What are the authors' thoughts about FtsH1 localization? The isoprenoid supplementation in this report and experiments in Amberg-Johnson et al., 2017, link the function of FtsH1 to the apicoplast, but the reported localization is to the mitochondrion (Tanveer et al., 2013).

2) Figure 1, Please show dose-response curves for D10 ActS, D10 ActR, FtsH1G489G and rFtsH1G489Ca/b. The data points used to derive the IC50 and slope of the sigmoidal curve can be informative in interpreting the drug response.

3) The authors seem to reference the pAIO-Pfftsh1 using Waller et al., 2000, when Spillman et al., 2017, may have been meant. Does the DHOD enzyme function in pAIO to the extent that this plasmid confers DSM1 resistance? It did not look like DSM1 was used to select for pAIO in Spillman et al., 2017.

4) The authors describe the dose escalation process of generating drug resistance, but it would be nice if they provided a sense of how long it took to achieve high level resistance.

Reviewer #1:

This seems like a pretty straightforward case. Earlier studies on actinonin's activity on the malaria parasite Plasmodium falciparum (Pf) left the impression that actinonin's target was peptide deformylase, its target in both bacteria and mammalian mitochondria. An eLife paper using a different apicomplexan, T. gondii, argued pretty convincingly, at least for me, that the target was likely to be a different protease, a metalloprotease called PfFtsH1 found in the Pf apicoplast. However there were alternative explanations for the results, and the target of actinonin was in the strictest sense not unambiguously identified. This short report, which shares no authors in common with the earlier Amberg-Johnson eLife publication, shows that a single mutation in PfFtsH1 confers resistance to actinonin in Pf. I think that the arguments presented here even more firmly establish PfFtsH1 as the target. Ironically the significance of this report, which I think is sufficient for publication in eLife, is to unequivocally relegate actinonin to a lesser category in the long list of potential antimalarial drugs – a relegation that I believe had largely been accomplished by the identification of FtsH1 as the target in the Toxoplasma gondii study reported in eLife by Amberg-Johnson et al.

Reviewer #2:

Actinonin is a peptidomimetic natural product known to inhibit certain metalloproteases. In bacteria and plastid-containing organisms, it has been shown to inhibit the metalloprotein peptide deformylase and this was thought to be the mechanism of action in the malaria parasite Plasmodium falciparum. Experiments in Toxoplasma gondii identified a different target metalloprotease, called FtsH1, and showed that a mutation in this enzyme conferred modest resistance to actinonin. Actinonin was also shown to inhibit recombinant P. falciparum FtsH1, suggesting that this is the target in malaria parasites. However, it was reported that actinonin resistance mutations could not be generated in P. falciparum. An additional point of uncertainty was the subcellular location of FtsH1, with one report claiming mitochondrial localization and another linking FtsH1 to the apicoplast. In the current report, the authors used a slow, dose escalation method to generate actinonin resistance in P. falciparum parasites. Resistant parasites contained a mutation in FtsH1 and did not have mutations in peptide deformylase or other potential target proteins. To confirm the link between drug resistance and the mutation, genome editing was used to add the mutation to a wild type line, generating resistance in this line. These experiments were carefully conducted with a genome editing control to generate a synonymous mutation that did not confer actinonin resistance. Assuming FtsH1 functions in the apicoplast, the authors used isoprenoid supplementation to disrupt the apicoplast and observed a reduced sensitivity to actinonin. Taken together, the results establish that FtsH1 is the primary target of actinonin in malaria parasites and show that parasites can generated significant resistance to this drug (18-35x).

Reviewer #3:

The identification of a PfFtsH1G489C variant that confers actinonin resistance strongly supports the previous work indicating PfFtsH1 as the target of actinonin. The conclusion is well-supported by allelic replacement that recapitulated the resistance phenotype that matches well with actinonin resistance in apicoplast-minus parasites.

1) Introduction. Please provide more background on uncertainty of FtsH1 localization. The authors cite Tanveer et al., 2013, for mitochondrial localization of FtsH1. In Amberg-Johnson et al., 2017, cleavage of FtsH1 dependent on presence of apicoplast also indicates possible apicoplast localization.

2) Figure 1, Please show dose-response curves for D10 ActS, D10 ActR, FtsH1G489G and rFtsH1G489Ca/b. The data points used to derive the IC50 and slope of the sigmoidal curve can be informative in interpreting the drug response.

Reviewer #4:

This study confirms a previous report by Amberg-Johnson and colleagues that a point mutation in Plasmodium falciparumftsh1 confers resistance to a peptidomimetic antibiotic, actinonin. The authors were able to generate actinonin-resistant P. falciparum and identified a point mutation in Pfftsh1 from selected resistant clones by genotyping. The resistance mutation was then introduced into the P. falciparum genome to verify its role in actinonin resistance. This short report adds supportive information to the previous finding that actinonin may target the apicoplast-associated FtsH1, but its novelty is considered limited.

Essential revisions:

1) Introduction. Please provide more background on uncertainty of FtsH1 localization. The authors cite Tanveer et al., 2013 for mitochondrial localization of FtsH1. In Amberg-Johnson, 2017, cleavage of FtsH1 dependent on presence of apicoplast also indicates possible apicoplast localization. [related but different reviewer] What are the authors' thoughts about FtsH1 localization? The isoprenoid supplementation in this report and experiments in Amberg-Johnson, 2017, link the function of FtsH1 to the apicoplast, but the reported localization is to the mitochondrion (Tanveer et al., 2013).

We have expanded the discussion of this contradiction in the Introduction as follows.

“Compounding this uncertainty is a report that PfFtsH1 is localized in the mitochondrion (Tanveer et al., 2013), which is inconsistent with the demonstrated impact of actinonin on the malaria parasite apicoplast (Goodman and McFadden, 2014; Uddin, McFadden and Goodman, 2017). […] Given the complexity of FtsH1 processing and localization in P. falciparum, it is unlikely that cell biological studies alone will be able to definitively resolve the issue of whether PfFtsH1 is the primary target of actinonin.”

2) Figure 1, Please show dose-response curves for D10 ActS, D10 ActR, FtsH1G489G and rFtsH1G489Ca/b. The data points used to derive the IC50 and slope of the sigmoidal curve can be informative in interpreting the drug response.

These have been included as part C in this figure, as requested.

3) The authors seem to reference the pAIO-Pfftsh1 using Waller et al., 2000, when Spillman et al., 2017, may have been meant.

Thank you for pointing this out. The reference referred to the transfection method used, not the vector. We have modified this statement for clarity. It now reads

“Each allelic replacement vector was linearized by digestion with EcoRI and co-transfected with pAIO-Pfftsh1 using standard transfection methods (Ghorbal et al., 2014).”

Please note, we also added “µM” to the concentration of DSM-1.

Does the DHOD enzyme function in pAIO to the extent that this plasmid confers DSM1 resistance? It did not look like DSM1 was used to select for pAIO in Spillman et al., 2017.

While the authors of this paper did not report selection with DSM-1 in their study, they did report the presence of a functional DHOD enzyme in pAIO. From the Materials and methods in Goodman, Su and McFadden, 2007, “This CAM promoter‐ yDHOD‐2A‐Cas9 amplicon was inserted into the sgRNA expression vector between XbaI/XhoI, removing the hDHFR and thymidine kinase selection cassettes, and fusing Cas9 to the P. berghei dihydrofolate reductase/thymidylate synthase (DT) 3′ UTR (previously supporting expression of thymidine kinase). The resulting plasmid containing both Cas9 and sgRNA expression cassettes was named pAll‐In‐One (pAIO).” The method of vector construction and resulting drug resistance cassette were unremarkable and it is to be expected that transfection of this construct would confer DSM-1 resistance. Our results proved this to be true.

4) The authors describe the dose escalation process of generating drug resistance, but it would be nice if they provided a sense of how long it took to achieve high level resistance.

This has been included in the Materials and methods. “This selection procedure required 10 weeks of constant drug selection to recover resistant parasites and 10 months of selection to develop parasites with the highest levels of resistance.”

Please note, for clarity we have made some minor edits. These include changes to gene names to be consistent with Amberg-Johnson et al., 2017, correction of small typographical errors, and reporting the specific concentration of actinonin used in drug resistance selection rather than relating it to IC50.

Key resources table

Reagent type (species) or resource	Designation	Source or reference	Identifiers	Additional information
Gene (Plasmodium falciparum)	FtsH1	PlasmoDB (plasmodb.org)	PF3D7_1313200
Gene (Plasmodium falciparum)	PDF	PlasmoDB (plasmodb.org)	PF3D7_0907900
Gene (Plasmodium falciparum)	MAP	PlasmoDB (plasmodb.org)	PF3D7_0804400
Gene (Plasmodium falciparum)	FMT	PlasmoDB (plasmodb.org)	PF3D7_1239700
Gene (Plasmodium falciparum)	RING	PlasmoDB (plasmodb.org)	PF3D7_1405700
Strain, strain background (Plasmodium falciparum)	3D7	MR4 - BEI Resources (beiresources.org)	MRA-102
Strain, strain background (Plasmodium falciparum)	D10	MR4 - BEI Resources (beiresources.org)	MRA-201
Strain, strain background (Plasmodium falciparum)	D10 ActR	This paper		Table 1
Transfected construct (Plasmodium falciparum)	D10 rftsH1G489G	This paper		Figure 1
Transfected construct (Plasmodium falciparum)	D10 rftsH1G489Ca	This paper		Figure 1
Transfected construct (Plasmodium falciparum)	D10 rftsH1G489Cb	This paper		Figure 1
Software, algorithm	GraphPad Prism software	GraphPad Prism (graphpad.com)	RRID:SCR_002798
Chemical compound, drug	Actinonin	Sigma	Sigma: A6671
Chemical compound, drug	DSM-1	Sigma	Sigma: 533304