Photoaffinity labeling of protein targets in a complex metazoan: proof-of-concept using a probe for Schistosoma mansoni tubulin

Abstract

Treatment of schistosomiasis, a prevalent neglected tropical disease, relies precariously on a single drug. The discovery and development of alternative anti-schistosomal small molecules most often relies on phenotypic (whole-organism) screening, whereas target- or “protein-first”-based discovery options are hampered by a paucity of genetic interrogation strategies and a complex biology. Here, we demonstrate the application of photoaffinity labeling (PAL) as a chemical biology strategy to probe protein target-ligand interactions in the schistosome. Using a triazolopyrimidine (TPD) probe that binds tubulin, we established a PAL workflow with living Schistosoma mansoni worms. The probe elicited deleterious phenotypic responses consistent with the TPD series and, upon UV light activation, covalently labeled tubulin as identified by proteomics. Specific concentration-dependent engagement of tubulin was confirmed using a photostable competitor TPD. When applied directly to worm lysates, the PAL workflow produced non-specific labeling, suggesting that the conformation of the protein target is important for ligand binding. The successful application of PAL for a metazoan is, to our knowledge, novel, and the platform should prove generally applicable to identifying potential drug targets, and exploring protein-ligand interactions, in schistosomes and other organisms.

Schistosomiasis, a neglected tropical disease (NTD) caused by parasitic flatworms of the genus Schistosoma, affects approximately 700 million people worldwide,^{1, 2} with over 250 million requiring chemotherapy.³ This chronic infection imposes a substantial health and socioeconomic burden in endemic regions, contributing to long-term morbidity and reduced productivity.^{4, 5} Notwithstanding its global impact, treatment relies on just one drug, praziquantel (PZQ).^{6, 7} Although safe and reasonably effective (50–90%^{8, 9} cure), it has notable limitations, particularly its lack of activity against juvenile worms^{10, 11} which allows infection and associated morbidity to continue as the parasites mature and lay eggs. In addition, PZQ displays suboptimal pharmacokinetics, characterized by extensive first-pass metabolism to therapeutically inactive products.^{12, 13} Finally, the reliance on PZQ heightens the risk of drug resistance.¹⁴ Collectively, these factors underline the need for new therapeutics, including those with novel mechanisms of action.

The discovery and structural-activity characterization of anti-schistosomal small molecules has most often employed phenotypic (whole-organism) screening of the pathogen in culture.^15–17 Target- or ‘protein-first’ based discovery methods are less common and have been hindered by both biological and technical constraints.¹⁸ For example, many of the reverse and forward genetics technologies to investigate gene product function in mammalian cells,^{19, 20} some protozoan parasites,^{21, 22} and the free-living nematode, C. elegans.²³ are not yet routine for the schistosome, in spite of considerable effort.^24–26 An exception is transient (non-germline) RNA interference which has demonstrated robust utility over the last 25 years to identify gene products that influence parasite growth and/or survival.^27–32 Added to these difficulties are the more general constraints regarding the pathogen’s complex life-cycle, involving invertebrate and vertebrate hosts, and the need for animal biosafety level (ABSL)-2 infrastructure.

In this context, we investigated a chemical biology approach to identify drug targets in S. mansoni. Specifically, we identified small molecule pyrimidines that engage microtubules (MTs) in mammalian cells and elicit deleterious phenotypic responses in ex vivo adult Schistosoma mansoni worms.³³ We subsequently repurposed these compounds against another pathogen, the protozoan, Trypanosoma brucei,³⁴ and designed and synthesized a photoaffinity labeling (PAL) probe (1, Table 1) that labeled T. brucei tubulin.³⁵ Inspired by these findings, we hypothesized that a PAL strategy could be applied to S. mansoni to identify molecular targets under conditions similar to those used in our phenotypic assays. Accordingly, we developed a protocol using the validated tubulin-targeting probe, 1,³⁵ to successfully label schistosome tubulin in a concentration-dependent manner. The data demonstrate that PAL can be effectively combined with phenotypic screens to investigate mechanisms of action, thus providing an exciting opportunity to identify protein targets in an experimentally challenging organism.

Table 1.

Bioactivity of PAL probe 1 and photostable competitor 2 vs. adult S. mansoni, MT-stabilizing Activity (AcTub), cytotoxicity in HEK293 cells, and binding energy ΔG values based on molecular docking.

Cpd	Structure	Phenotypic responsesa and severity score at 10 μM and 72 hb	MT-stabilization assayc,f		HEK293 CC50 (μM)d,f	Calc. Binding Energy ΔG (kcal/mol)e
			1 μM	10 μM		Vinca site S. mansoni (mammalian)	Seventh site S. mansoni (mammalian)
1		dark, unc, on sides (3)	10.3 ± 0.1 **	2.69 ± 0.20 **	0.0108 ± 1.5	−60.11 (−65.07)	−62.13 (−50.44)
2		dark, unc, on sides (3)	6.75 ± 0.32 **	2.99 ± 0.02 **	0.0646 ± 0.0094	−62.79 (−61.90)	−60.64 (−46.64)

^aAbbreviations for phenotypic response descriptors^{33, 41} are: Unc - uncoordinated movements; dark- worms are less translucent vs. control; on sides - worms are unable to adhere to the well floor with their oral and/or ventral suckers).

^bEach descriptor is given a value of 1 and these are then added to generate a severity score (max score = 4).

^cFold-changes in acetylated α-tubulin (AcTub) protein levels in HEK293 cells, relative to vehicle (DMSO)-treated cells, were determined after a 4 h incubation with compound at either 1 or 10 μM (*p<0.05 and **p<0.01 by one-way ANOVA).⁴⁰

^dCytotoxicity was measured using HEK293 cells and a resazurin cell viability assay.⁴³

^eCalculated binding energy from docking studies at the seventh and vinca sites of S. mansoni tubulin and mammalian tubulin (values in brackets). The mammalian vinca and seventh sites are from PDB: 5NJH³⁸ and PDB: 7CLD,³⁹ respectively, whereas the corresponding sites in S. mansoni tubulin are based on homology modeling.

^fPreviously published.³⁵

Previously, we evaluated MT-active phenylpyrimidine (PPDs) and triazolopyrimidine (TPD) compounds against ex vivo adult S. mansoni.³³ These induced various phenotypic responses that included uncoordinated movements, an inability of the oral and ventral suckers to adhere to the well and, subsequently, degeneration.³³ Given the high amino acid sequence identity between mammalian and S. mansoni α- and β-tubulins, we hypothesized that these small molecules exert their effects through binding to schistosome tubulin. Due to the lack of established functional assays to directly evaluate MT-targeting activity in S. mansoni, we employed the TPD PAL probe, 1 (see Table 1), previously validated in HEK293 cells and T. brucei, as a chemical probe to investigate target engagement.³⁵

Probe 1 was designed based on an established structure–activity relationship (SAR) for this compound class with respect to mammalian MT-stabilizing activity, specifically using TPD 2 (Table 1) as a structural template, and was synthesized as previously described.^34–37 Probe 1 incorporates a diazirine group which, upon UV activation, covalently binds its protein target. An alkyne handle allows conjugation to a bifunctional reporter that comprises a TAMRA group for fluorescence imaging and biotin for affinity enrichment.³⁵

In mammalian cells, TPD compounds bind to two MT sites: the vinca and the seventh sites, which are at the interface of β1-α2 and α2-β2 tubulin, respectively.^{38, 39} Molecular docking studies with probe 1 indicate comparable binding affinities for the vinca site of S. mansoni and mammalian tubulin, with predicted ΔG values of −60.11 and −65.06 kcal/mol, respectively (Table 1). Compound 2 displayed similar affinities, with ΔG values of −62.79 kcal/mol for S. mansoni and −61.90 kcal/mol for mammalian tubulin. Interestingly, both probe 1 and 2 exhibited greater predicted affinity for the seventh site of S. mansoni tubulin compared to mammalian tubulin, with ΔG values of −62.12 and −50.44 kcal/mol for 1, and −60.64 and −46.64 kcal/mol for 2 (Table 1).

Examination of the docking poses of 1 and 2 with mammalian and S. mansoni tubulin revealed some key similarities in their binding modes (Figure 1). At the vinca site, for both 1 and 2, π–π stacking is observed between the TPD core and β-Tyr222, while the dimethylamino chain is predicted to form contacts with β-Lys174, β-Val175 and β-Ser176. In contrast, at the seventh site, 1 and 2 display stronger binding affinity for S. mansoni, likely due to differences in interactions within surrounding residues. In both the mammalian and S. mansoni seventh sites, interactions include H-bonding with α-Asp211 and π–π stacking of the TPD and difluorophenyl rings with α-Tyr224 and α-Tyr210, respectively. In the mammalian seventh site, the 3,3-dimethylbutan-2-yl group is observed near α-Leu227, β-Leu246, β-Lys350 and β-Thr351, while the dimethylamino chain shows hydrophobic contacts with α-Val177 and α-Ser178. In contrast, within the S. mansoni seventh site, the 3,3-dimethylbutan-2-yl shows hydrophobic contacts with β-Leu246, β-Ile345, and β-Pro346, while the dimethylamino chain is surrounded by α-Gln176, α-Ile177, α-Ala178 and α-Thr179 (Figure 1).

Figure 1.

Docking of 1 (A, in grey) and 2 (B, in yellow) to the vinca (PDB: 5NJH³⁸) and seventh sites (PDB: 7CLD³⁹) of mammalian and S. mansoni tubulin (homology model): α-tubulin is shown in red and β-tubulin in blue.

Compounds 1 and 2 were evaluated for MT-stabilizing activity and cytotoxicity in HEK293 cells, as previously described (Table 1).³⁵ Both compounds exhibited comparable MT-stabilizing activity, as evidenced by increased levels of acetylated tubulin, which is a well-established marker of stabilized MTs.⁴⁰ The compounds were also assessed in our ex vivo S. mansoni phenotypic screening platform,^{33, 41, 42} which employs a constrained nomenclature of descriptors to adjudicate phenotypic changes in the parasite, and from which a severity score, on a scale of zero (no effect) to 4 (maximal effect), is calculated. After 72 h, both 1 and 2 elicited similar phenotypic responses in adult worms, characterized by darkening, uncoordinated movement, and the worm’s inability to use the oral and/or ventral sucker to adhere to the surface of the assay plate wells (Table 1).

Given the ability of 1 to engage MTs in mammalian cells and having recorded measurable phenotypic changes in S. mansoni, we developed a PAL workflow^{35, 44} (Figure 2A) with ex vivo S. mansoni adults to measure whether 1 binds to schistosome tubulin (see Supporting Information). Briefly, approximately five males and two females were incubated with 2 μM PAL probe for 2 h in a 24-well plate. Then, the probe was photoactivated for 30 min using a handheld UV lamp (365 nm) positioned 1–2 cm above the surface of the culture medium in the assay plate (minus the plate lid). Following photoactivation, worms were homogenized by mechanical sonication, and the lysate clarified by centrifugation. Proteins were subjected to a Click reaction^{44, 45} with TAMRA-biotin-azide to allow for in-gel fluorescence visualization of the labeled protein bands. This was followed by excision of the protein bands and neutravidin-based enrichment of the labelled proteins for mass spectrometry-based proteomics (see Supporting Information).

Figure 2.

(A) Summary of the PAL protocol for ex vivo S. mansoni adults. (B) PAL experiments with S. mansoni lysates and (C) living worms resolve different labeling profiles under the same conditions with probe 1. (D) Proteomics and densitometry analysis. Proteomics analysis of the protein band labeled specifically by 1 (red box in C) identifies S. mansoni tubulin. Densitometry analysis reveals concentration-dependent binding in presence of the photostable competitor, 2. Protein concentrations were measured using the BCA assay and normalized to 0.3 μg/mL to ensure equal protein loading. *p < 0.005 and **p < 0.005 by one-way ANOVA.

To evaluate the specificity of binding by probe 1, competition experiments were performed by pre-incubating worms with the photostable competitor, 2 (2, 10 and 50 μM), for 30 min prior to incubation with 2 μM 1. The labeling workflow was also applied to S. mansoni lysates (see Supporting Information) to understand the probe’s binding specificity under such conditions, and if determined, whether binding could be competed for with the competitor.

With worm lysates, several fluorescent bands were observed by PAL with probe 1, including a prominent band at ~45 kDa (Figure 2B). The intensity of labeling of these bands could not be diminished in the presence of increasing concentrations of the competitor, 2, suggesting non-specific binding. PAL performed with live worms also revealed several intense bands, with the most prominent at ~55 kDa (Figure 2C). In the presence of increasing concentrations of competitor 2, the fluorescence intensity of this band decreased markedly, indicating a specific competitive interaction (Figure 2C, D). Other bands, particularly at ~40 and ~28 kDa, were also labeled by probe 1; however, their competition profiles were less pronounced than that for the 55 kDa band, suggestive of non-specific binding. After lysis of the live worms that had been exposed to PAL, the proteins bound by probe 1 were enriched by neutravidin-affinity purification. The 55 kDa gel band was excised and subjected to proteomic analysis, which identified the sample to be primarily composed of α- and β-tubulin (Figure 2D).

Collectively, whereas, PAL with worm lysates using 1 in competition with 2 suggested non-specific binding, the data generated with living worms confirms a specific PAL reaction with tubulin at 55 kDa. In the former experiment, it is possible that MTs depolymerize during lysate preparation, resulting in the loss of the vinca and seventh binding sites at the α-β tubulin interface. If confirmed, then the finding underscores the importance of investigating target binding under native conditions.

The demonstration here of a specific PAL reaction is, to the best of our knowledge, novel for a metazoan, and underscores the utility of this technique to identify potential drug targets from phenotypic screens which are commonly employed for this parasite. Importantly, with the associated knowledge of a small molecule’s chemistry and structure-activity relationship (SAR), nearly any bioactive ligand can potentially be adapted as a PAL probe, thereby opening opportunities to investigate a broad range of protein targets. In addition to target identification, PAL can be integrated with chemoproteomic approaches to enable a quantitative analysis of probe-bound proteins. We will apply this strategy in the future to quantify target-engagement and -selectivity. Overall, the results reported here highlight the potential of PAL as a versatile platform to support target-based drug discovery in the schistosome and evaluate protein-ligand interactions in other metazoans.