A Click Chemistry‐Based Proteomic Approach Reveals that 1,2,4‐Trioxolane and Artemisinin Antimalarials Share a Common Protein Alkylation Profile

Abstract In spite of the recent increase in endoperoxide antimalarials under development, it remains unclear if all these chemotypes share a common mechanism of action. This is important since it will influence cross‐resistance risks between the different classes. Here we investigate this proposition using novel clickable 1,2,4‐trioxolane activity based protein‐profiling probes (ABPPs). ABPPs with potent antimalarial activity were able to alkylate protein target(s) within the asexual erythrocytic stage of Plasmodium falciparum (3D7). Importantly, comparison of the alkylation fingerprint with that generated from an artemisinin ABPP equivalent confirms a highly conserved alkylation profile, with both endoperoxide classes targeting proteins in the glycolytic, hemoglobin degradation, antioxidant defence, protein synthesis and protein stress pathways, essential biological processes for plasmodial survival. The alkylation signatures of the two chemotypes show significant overlap (ca. 90 %) both qualitatively and semi‐quantitatively, suggesting a common mechanism of action that raises concerns about potential cross‐resistance liabilities.

Despite concerns about the recent emergence of drugresistance, [1] thea rtemisinins (1a-c,F igure 1) remain frontline agents for the treatment of malaria. [2] Understanding the mechanism of action of such an important class has been the subject of intense research over the last two decades. [2] The proposed mechanism of bioactivation of the class involves the cleavage of the endoperoxide bridge by as ource of Fe 2+ or heme.T his cleavage results in the formation of oxy-radicals that rearrange into primary or secondary carbon centered radicals (or electrophilic carbocations through single-electron transfer oxidation) (Scheme 1). [3] These reactive intermediates are proposed to alkylate proteins and form adducts with essential parasite macromolecules that result in the rapid death of the parasite. However,t he detail of these important alkylation reactions are sparse and the underlying hypothesis remains controversial. [2] Thedebate has broadened with the development of highly active fully synthetic endoperoxides based on the pharmacophore of artemisinin namely the trioxolanes (2a) [4] and the tetraoxanes. [5] From the perspective of the underlying chemical mechanism of activation of peroxides and from ac rossresistance risk it is important to establish if these different Scheme 1. Iron-mediated fragmentation of endoperoxides to reactive intermediates capable of reacting with parasite proteins. (Only the secondary carbon-centered radical derived from artemisinin is depicted.) endoperoxide chemotypes share ac ommon mechanism of action or not.
As tudy of 1,2,4-trioxolanes using monoclonal antibodies has demonstrated parasite protein alkylation with both OZ277 (2b)a nd OZ439 (2c). [6] However,t he methods employed in this work were unable to definitively identify the targeted proteins.
In ar ecent study,W ang et al. [7] used an on-optimised ART-alkyne activity-based protein-profiling probe that via click-chemistry reactions was associated with some 124 P. falciparum proteins. [7] In this study,afurther 125 proteins are reported as being identified in single replicate experiments only,r aising concerns of the specificity of the approach. [7] Concurrently,u sing both ART-alkyne and azide optimized probes we were able to identify 59 P. falciparum proteins with high confidence that were specifically alkylated by ART pointing towards apleiotropic mechanism of drug action. [9] In our study we deployed probes with reduced linker length and lipophilicity compared to the probe used by Wang et al.,and we used both copper-dependent and copper-free reactions over as hortened incubation time (optimized to 1h as originally described, [8] cf.W ang et al.,r eaction time of 3h) together with control non-peroxide probe partner equivalents.T hese methodological differences help to improve the specificity and pharmacological relevance of the alkylated proteins identified. Significantly in our study only 6p roteins were identified in less than two replicate experiments (cf.125 in Ref. [7]) with an azide probe,demonstrating the improved specificity of our approach using optimized active and control probe based methodology.
Here,using our refined approach, we describe the rational design of potent activity-based protein-profiling probes (ABPPs) based on a1 ,2,4-trioxolane antimalarial core in order to characterize their malaria parasite protein alkylation fingerprint. Significantly,w es how that these synthetic 1,2,4trioxolanes and as emi-synthetic ARTs hare an overlapping parasite protein-alkylation signature suggestive of acommon mechanism of action for the endoperoxide class of antimalarial.
Thep robes were designed with the alkyne/azide click handle sited within the adamantane ring system since this is the site (Scheme 1) of reactive C-radical/carbocation gener-ation post activation by Fe II .W ea lso deployed ab ioorthogonal copper-free "click" methodology via the use of an azide analogue along with its negative control Figure 2. Thea zide probes were included in our analysis to demonstrate that the protein alkylation was solely due to iron mediated activation with no role for the copper in peroxide activation during sample work-up as discussed previously. [9] This Cu-free click reaction possesses comparable kinetics to the Cu-catalyzed reaction and proceeds within minutes in live cells with no apparent cytotoxicity issues. [8b] Thecomplementary reporter tags used in our study can either be sourced commercially or synthesized by literature procedures (Figure S1 in the Supporting Information).
Scheme 2a provides an overview for the synthesis of alkyne probe P1 (6a)a nd azide probe P2 (7a)a long with control probes CP1 (6b)and CP2 (13 b). Thefirst step in the synthesis of the trioxolane probes involved the Koch-Haaf carbonylation of hydroxyl adamantanone to give the methyl ester 3.C o-ozonolysis of oxime 4 with 3 provided trioxolane 5;h ydrolysis of the methylester function of 5 followed by EDC-mediated coupling of propargyl amine provided probe P1 (6a)i ng ood overall yield. Coupling of 3-azido-1-propylamine to 6 provided the azide probe P2 (7a)a ss hown in Scheme 1. Thecontrol probe CP1 (6b)was made by asimilar approach using diol 9 in ac yclisation reaction with 3 to produce the corresponding carba ester analogue 10.H ydrolysis of 10 and coupling of the resultant acid with propargyl amine afforded 6b in good yield (70 %). Azide 7b was made in as imilar manner by hydrolysis and coupling as shown in Scheme 2b.
Thea ctive probes containing azide/alkyne functionality retained potent antimalarial activity as determined by their IC 50 in vitro against P. falciparum 3D7 parasites ( Figure S2 and Table S1). Then on-peroxidic negative control probes CP1 (6b)a nd CP2 (7b)h ad no appreciable activity (IC 50 > 10 mm)i nt hese assays, [10] confirming the essentiality of the endoperoxide-bridge for drug activity and further validating

Angewandte Chemie
Communications our probe plus control pair strategy for biologically relevant target protein identification.
As anext step,invitro cultures of P. falciparum 3D7 were incubated with 1 mm of the alkyne trioxolane based probe P1 (6a)orits corresponding control CP1 (6b), for 6hours,atime already shown to be pharmacologically relevant, causing irreversible parasite toxicity. [7,9,10] Following incubation with P1 (6a)a lkylated proteins were extracted from erythrocyte free parasites and tagged with Alexa Fluor 488 azide via ac lick reaction. This was processed for 1D-Gel analysis as described in the supporting information section. After fluorescence imaging of the 1D-Gel, the strongest labeling was observed in the region of 12-75 KDa (Figure S2 c). Importantly,labeling was not observed for the corresponding "negative control" probe CP1 (6b)s amples.A dditional experiments were carried out to investigate the lowest concentration of reporter azide required to obtain maximum labeling with minimum background on SDS gels (data not shown). As depicted in Figure S2 c, the Alexa Fluor 488 azide at aconcentration of 20 mm was able to distinguish the P1 (6a) labeling profile from its corresponding control CP1 (6b).
To rule out the possibility that Cu I may have led to parasite independent (artifact) protein alkylation and to further validate the results am ore stringent bio-orthogonal copper-free "click" methodology was employed using cyclooctyne reporter tags as depicted in Figure 3. [8b] Replacement of an alkyne with an azide group in P2 (7a)h ad no detrimental effect on the antimalarial potencyo ft he trioxolane azide probe as indicated in Figure S2, Figure 3a nd Table S1. However,the labeling profile intensity with P2 (7a) was much higher compared to P1 (6a) (Figure 3c,d) suggesting greater efficiency of the copper-free click reaction. [8b] Reduction of P2 (7a)c oncentrations to 100 nm (LC90) had no impact on the labeling pattern observed (Figure 3f).
After validating the importance of the endoperoxide bridge for protein alkylation using a1Dgel, we excluded the gel electrophoresis step and advanced the method to direct analysis of the alkylated protein matrix captured using an "on-bead" trypsin digestion protocol as shown in Figure 3b and Figure S3. Overall, multiple proteins critical to parasite life were identified as trioxolane targets with the P2 (7a) probe ( Figure S3). No labeling was evident with CP2 (7b)the Figure 3. Labeling of parasite proteins (P. falciparum,3D7 strain) using 7a.a)Chemical structure of ozonide azide probe (P2 (7a)) and deoxyether analogue (CP2 (7b)) and their antimalarialactivity against P. falciparum 3D7. b) General workflows of copper-free click methodology for in situ parasite protein identification using azide trioxolanes probes as detailed mentioned in methodology section. c) Fluorescenceimage of 1D gel for proteins labeled in situ with alkyne probes (P1 (6a)and CP1 (6b)) vs. azide probes (P2 (7a)a nd CP2 (7b)), note that no labeling occurs with negative control alkyne (CP1 (6b)) and azide control (CP2 (7b)). d) Arbitrary fluorescence intensity measurements of the major protein bands labeled and identified with 20 mm Alexa flour 488 azide for parasite proteins tagged with 1 mm of alkyne probe (P1 (6a)) vs. proteins tagged with 1 mm of azide probe (P2 (7a)) identified with 20 mm Click-IT Alexa Fluor 488 DIBO Alkyne. Fluorescencearbitrary units reveal higher sensitivity in case of bio-orthogonalc opper free click reaction, that is, P2 (7a)treatment. e) Gel image of P2 (7a)treatment vs. control, pre and post coomassie stain with equal protein loading. f) Fluorescenceimage representing probe titration from 1to0.1 mm P2 (7a)probe;proteins identified via copper free click reaction with Click-IT Alexa Fluor 488 DIBO Alkyne. No changes were observed in labeling profiles of the trioxolane-tagged proteins with concentrations relevant to pharmacological concentrationo fthe drug (100 nm). g) Titration of DIBO dye at various concentrations up to 20 mm for parasite treated in situ with 1 mm P2 (7a). h) Time dependent increase of fluorescence signal for proteins tagged with 1 mm P2 (7a)and 20 mm Click-IT Alexa Fluor 488 DIBO Alkyne indicatingthat the maximum band intensities could be achieved after 1-hour of click reaction incubation.

Angewandte Chemie
Communications negative control analogue or with the DMSO control (Figure S3). Having completed the analysis of the 1,2,4-trioxolane proteome we carried out ah ead-to-head comparison of an analogous ARTactivity based profiling probe (Figure 4). [9] Remarkably,asdepicted in Figure 4, both the trioxolanes probe P2 (7a)a nd ARTp robe P3 (11 a)s hare strongly overlapping protein-labeling profiles both qualitatively and semi-quantitatively.F rom at otal of 62 proteins confidently identified with the two probes 53 of the proteins were tagged with both P2 (7a)and P3 (11 a) (Figure 4a), with no labeling observed for control probes CP2 (7b)a nd CP3 (11 b) ( Figure S3). From am echanistic perspective it is important to note that ca. 70 %o ft he tagged proteins can be readily glutathionylated (Figure 4b), ap ost translational modification that can effect redox regulation and signal transduction. [11] Fori nstance,E XP1, the membrane glutathione Stransferase identified with both P2 (7a)a nd P3 (11 a)w as reported to efficiently degrade cytotoxic hematin in malaria parasite. [12] Interestingly,a rtesunate,awater-soluble derivative of ART, has been reported to inhibit the GST activity of EXP1 in ah ematin dependent manner at ca. 2nm (IC 50 ). [12] PfEXP1 facilitates the conjugation of GSH with artesunate in vitro. [12] We proposed that carbon centered radicals generated from the reductive scission of the endoperoxide bridge from either P2 (7a)o rP3 (11 a)a lkylate proteins by C-radical attack at the disulfide bridges of the glutathionylated proteins identified in this study; [11,13] further work is on-going to confirm this mode of reactivity.T he major proteins in the mass range 30 to 75 KDa identified for P2 (7a)a re, PfLDH, PfOATand PfHGPRT, similar to that seen earlier with the equivalent ARTprobes, [7,9] Collectively the data suggest that 1,2,4-trixolanes efficiently target plasmodial energy supply, the antioxidant defence system and DNAs ynthesis.I n addition, ARTh as been shown to modulate av ariety of signaling pathways in cancer cells. [14] In the present study many cytoskeletal proteins including a-tubulin, b-tubulin, and actin 1w ere labeled with trioxolanes and ARTp robes suggesting ap otential link between these endoperoxide drugs and parasite cell structure,p rotein trafficking systems and signal transduction. [9] Moreover,t he global analysis of protein alkylation generated through P2 (7a), for both classes,t hat is,A RT and trioxolanes,i sc onsistent with the "cluster bomb" hypothesis, [7,9,15] whereby Fe 2+ /heme-activated drug alkylates multiple redox-susceptible protein targets functioning in multiple cellular pathways ( Figure S4) including the food vacuole,as ite considered important for iron dependent activation, and also in the cytosol ( Figure S5).
To conclude,achemical proteomic approach has for the first time enabled formal identification of the key proteins that are alkylated by the 1,2,4-trixolane class of antimalarial. Significantly,t he proteomic profile of 1,2,4-trioxolanes is similar to the artemisinins suggesting that 1,2,4-trioxolanes  (P3 (11 a)). a) Venn diagram demonstrating overlap between proteins identified with the endoperoxide probes, P2 (7a)a nd P3 (11 a)r espectively. (b) Percentage of the glutathionylated proteins, which containst he GSH binding motif that was identified with endoperoxides probes P2 (7a)a nd P3 (11 a)i nlight of Kehr et al. [11] (c) Head to head comparison between proteins identified with P2 (7a)v s. P3 (11 a). Proteins sorted according to their molecular weight from high to low.Errors bars represented the standard deviation for protein quantity in each treatment calculated by dividing the exponentially modified protein abundancei ndex (emPAI) [16] for each protein by the total emPAI values (each treatment contain two replicate, for accuracy each replicate is the average of four injectionsi nto the Orbitrap LC-MS/MS instrument).
are multi-targeting like artemisinin and it remains to be seen if as imilar stress response and accumulation of ubiquinated proteins occurs for this class of antimalarial in PfKelch13 resistant parasites. [17] Clearly,o ur data raises concerns of the potential crossresistance [18] between these two different antimalarial chemotypes.O ur optimised endoperoxide-ABPPs strategy has generated aspecific and robust set of tools to study potential protein targets of the endoperoxide class of antimalarials. [19] We are currently further refining this approach to accommodate ab roader range of peroxide-based antimalarial chemotypes.W ork is also underway to establish the lifecycle-dependent "endoperoxome" patterns in asexual and sexual stages of P. falciparum parasite isolates with wellcharacterized artemisinin drug resistance phenotypes to assist in our understanding of this worrying clinical phenomenon.