N-Aryl-3-mercaptosuccinimides as Antivirulence Agents Targeting Pseudomonas aeruginosa Elastase and Clostridium Collagenases.

In light of the global antimicrobial-resistance crisis, there is an urgent need for novel bacterial targets and antibiotics with novel modes of action. It has been shown that Pseudomonas aeruginosa elastase (LasB) and Clostridium histolyticum (Hathewaya histolytica) collagenase (ColH) play a significant role in the infection process and thereby represent promising antivirulence targets. Here, we report novel N-aryl-3-mercaptosuccinimide inhibitors that target both LasB and ColH, displaying potent activities in vitro and high selectivity for the bacterial over human metalloproteases. Additionally, the inhibitors demonstrate no signs of cytotoxicity against selected human cell lines and in a zebrafish embryo toxicity model. Furthermore, the most active ColH inhibitor shows a significant reduction of collagen degradation in an ex vivo pig-skin model.

Introduction

The growing number of antibiotic-resistant bacteria represents one of the biggest risks to public health, leading to an increasing number of infections that are difficult to treat. Bacterial resistance to antibiotics is natural, yet overuse and misuse of antibiotics accelerate resistance development, bringing the world to the verge of the so-called “post-antibiotic era”. Of special importance are infections caused by multidrug-resistant bacteria on the WHO priority pathogen list,[1] such as carbapenem-resistant variants of the Gram-negative pathogen Pseudomonas aeruginosa,[2] which are responsible for many nosocomial,[3] eye and burn infections,[4,5] as well as fatal lung infections in cystic-fibrosis and bronchiectasis patients.[6,7]P. aeruginosa also affects injured tissue such as skin via surgical or wound infections.[8] The versatile pathogen is known to produce numerous virulence factors.[9] One of them is elastase, the metalloenzyme that shows hydrolytic activity toward connective tissue, which significantly contributes to the virulence of these bacteria.[10] Clostridiaceae represent a family of Gram-positive bacteria that are known as causative agents of numerous fatal diseases with high mortality rates worldwide, such as botulism (Clostridium botulinum), soft-tissue infections like gas gangrene and wound infections (Clostridium perfringens, Clostridium histolyticum) and tetanus (Clostridium tetani).[11,12]Bacillus cereus is another Gram-positive bacterium responsible for foodborne illnesses and traumatic wound infections in humans.[13,14] The high lethality of these bacteria is closely related to the production of collagenases, extracellular enzymes that enable the bacteria to colonize specific niches in the host, to evade the host immune response and to obtain nutrition from infected cells. Moreover, collagenases cause tissue destruction via collagen degradation, which plays a significant role in the infection process by allowing the bacteria to reach anaerobic sites in host tissue and spread the infection.[15,16] This especially affects the wound infection prognosis and results in a delayed healing process.[17,18]

Recently, particular emphasis has been put on targeting bacterial virulence as an alternative approach for fighting microbial infections. The pursued “pathoblockers” preserve the commensal microbiome and are expected to be less susceptible to the development of resistance than conventional antibiotics. In our work, we focus on two zinc metalloproteases that are secreted virulence factors: elastase (LasB) from P. aeruginosa and collagenase H (ColH) from C. histolyticum (recently renamed as Hathewaya histolytica).[19] Both enzymes have a major impact on the infectivity of P. aeruginosa and C. histolyticum.[11,20] Moreover, their extracellular localization makes these enzymes particularly attractive targets, considering the difficulties associated with crossing the Gram-negative bacterial cell wall in the former species.

Most LasB and ColH inhibitors found in the literature contain various metal-chelating warheads.[20−24] Among them, hydroxamates represent the most common structural motif.[21,25−27] The main problem with such hydroxamate-containing protease inhibitors is their lack of stability under physiological conditions and their lack of selectivity over human matrix metalloproteases (MMPs), which makes them unsuitable candidates for antibacterial treatment in vivo(26,28) and rationalizes why there is still no drug on the market that could effectively inhibit these virulence factors. A selection of LasB and ColH inhibitors described in the literature is shown in Figure .

Figure 1

Structural motifs of some LasB and ColH inhibitors described in the literature (Zn-binding groups are highlighted in blue).[21,22,24,26]

In our previous work, we discovered N-aryl mercaptoacetamide inhibitors with low micromolar and nanomolar affinities toward LasB and ColH, respectively.[29,30] To constrain the flexibility and freeze the active conformation of our previously published thiols, we designed a novel succinimide class (Figure ). The succinimide core has been reported for inhibitors of various enzyme targets, such as serine proteases,[31] human leukocyte elastase, cathepsin G and proteinase 3,[32] tumor necrosis factor, and phosphodiesterase.[33] It can also be found in several drugs, such as ethosuximide, phensuximide, methsuximide, and lurasidone, used to treat absence seizures, schizophrenia, and bipolar disorder.[34]

Figure 2

Our previous results and the design of new inhibitors.[29,30]

Here, we report new N-aryl-3-mercaptosuccinimides, showing low micromolar potencies against P. aeruginosa elastase and nanomolar potencies against Clostridium collagenases. The most active compounds were investigated for their cytotoxicity and selectivity for the bacterial over human metalloproteases. To validate collagenases as targets, we have established an ex vivo pig-skin model and demonstrated the impact of our most potent inhibitor on this human skin mimic.

Results and Discussion

Design of New Compounds

We designed the initial succinimide core based on our previously published N-aryl mercaptoacetamide inhibitors.[29,30] To expand the structure–activity relationships (SARs) further and provide more detailed information on the aromatic moiety’s influence on the activity, we designed a series of compounds bearing polar, lipophilic, electron-withdrawing or -donating functional groups. In order to prevent disulfide formation, we protected the free thiol group in the form of thioacetate. Finally, we explored the possibility of growing the structure further by introducing an additional carbon spacer between the succinimide and the free thiol.

Synthesis of New Compounds

Reaction of anilines with mercaptosuccinic acid at 120–160 °C afforded 21 new free thiol-containing succinimides 7–27 in 6−68% yield. Due to the presence of an ortho-substituent, compounds 16–18 and 21 were obtained as mixtures of atropisomers. Acetic anhydride in the presence of pyridine and DMAP at room temperature led to protection of the free thiol group to obtain derivatives 28–30 in moderate yield (65–78%). The general synthetic route is presented in Scheme .

Scheme 1

Synthesis of Novel Free Thiol Succinimides and Thioacetate Derivatives

Reagents and conditions: (a) mercaptosuccinic acid, 120–160 °C, Ar, neat, 3.5 h to overnight; (b) Ac2O, Pyr, DMAP, DCM, rt, overnight.

Synthesis of N-aryl-3-mercaptomethylsuccinimides started from itaconic anhydride and the corresponding anilines (Scheme ). Cyclization of intermediate α-itaconamic acids 31, 32, 34, and 35 in the presence of acetic anhydride and sodium acetate at 100 °C[35] afforded itaconimides 36–39 in relatively low yield (8–39%). Michael addition of thioacetic acid on obtained itaconimides in the presence of triethylamine in dimethoxyethane or dichloromethane at room temperature led to the final compounds 40–43 in high yield (61–99%). α-Itaconamic acids 32–34 under the same reaction conditions in THF as a solvent provided compounds 44–46 in quantitative yield. Hydrolysis of thioacetate using sodium hydroxide in methanol at room temperature afforded free thiol-containing carboxylic acids 47–49 in moderate to quantitative yield (61–99%). A final neat cyclization step furnished target compounds 50–52 in low yield (14–36%).

Scheme 2

Synthesis of N-Aryl-3-Mercaptomethylsuccinimides and Their Acetylthio Analogues

Reagents and conditions: (a) itaconic anhydride, CHCl3, 2 h, rt; (b) Ac2O, NaOAc, 1–2 h, 100 °C; (c) AcSH, Et3N, DME or DCM or THF, rt; (d) 2 M NaOH, MeOH, 1–2 h, rt; (e) 120 °C, Ar, neat, overnight.

Activity against Antivirulence Targets

We evaluated all compounds synthesized in this work for their inhibitory activity against both LasB and ColH. IC50 values and percentage of inhibition results were determined from at least two independent experiments performed in duplicate.

SAR of Novel Succinimides on LasB

To expand the SAR, we designed and synthesized 31 succinimide-based derivatives and investigated their inhibitory activity against LasB using a functional FRET-based in vitro assay (Table ).[36] Among the first group of compounds 7–27, electronegative substituents such as chlorine or fluorine were found to be favorable for the activity. In particular, compounds 13 and 15, both with a 3,4-dihalo pattern, displayed more potent inhibitory activities when compared to the ortho-, meta-, or para-monosubstituted analogues. Furthermore, several examples indicate that polar groups, particularly electron-donating substituents are detrimental for the activity. In fact, all mono and dimethoxy derivatives were less potent than their chlorine analogues, with 3,4-dimethoxy derivative 22 showing the most dramatic loss in activity (47-fold compared to 13). This is further supported by the 5-fold difference in activity between 3,4-dimethyl (9) and 3,4-dimethoxy analogues (22). Among both chlorine and methoxy isomers, ortho- (10 and 18) and meta-derivatives (11 and 19) proved to be more potent than the para-derivatives (12 and 20), most probably due to electronic effects or disruption of the planar structure caused by the vicinity of the substituents to the succinimide core. Compound 25 with its polar electron-withdrawing para-acetyl substituent was 1.7 times more active than its methoxy-analogue, but still much less potent than compounds bearing lipophilic chlorine substituents. The observation that polar, electron-withdrawing substituents are better tolerated than electron-donating ones is illustrated through the example of nitro-compound 24, being more active than its 3-methoxy analogue 19. Compounds with an additional phenyl ring in positions 3 (26) and 4 (27) showed comparable activity to the compound with a naked core, with 3-phenyl derivative even being 1.6-fold more active than compound 7. Although these derivatives were less potent than compound 13, the fact that they show inhibition of LasB paves the way for further optimization of this part of the structure. In addition, the fact that compounds 8 (R1 = 4-Me) and 27 (R1 = 4-Ph) show no significant difference in the activity suggests that there are no steric limitations in para-position. An additional carbon spacer next to the free thiol in compounds 50 and 51 did not improve the activities of the most potent derivatives 13 and 15, respectively. However, the IC50 values determined for 50 and 51, being in the range of 5–10 μM, open the possibility to further grow the structure in the direction of the free thiol group. All thioacetate derivatives proved to be inactive, with <50% of inhibition at 200 μM, which confirms that the free thiol is crucial for the activity against LasB.

Table 1

Structures and LasB Inhibition of a Series of Novel Succinimide Derivativesa

compd	R1	R2	n	IC50 (μM)	compd	R1	R2	n	IC50 (μM)
7	H	H	0	44.2 ± 2.3	23	3-Cl-4-OCF3	H	0	55.2 ± 4.4
8	4-Me	H	0	50.6 ± 1.6	24	3-NO2	H	0	22.2 ± 0.4
9	3,4-di-Me	H	0	29.4 ± 0.9	25	4-Ac	H	0	64.0 ± 7.5
10	2-Cl	H	0	8.5 ± 0.4	26	3-Ph	H	0	27.6 ± 4.0
11	3-Cl	H	0	8.1 ± 0.5	27	4-Ph	H	0	44.6 ± 1.1
12	4-Cl	H	0	16.5 ± 0.8	28	4-OMe	Ac	0	>200
13	3,4-di-Cl	H	0	3.4 ± 0.2	29	3,4-di-Cl	Ac	0	>200
14	2,6-di-Cl	H	0	16.0 ± 3.7	30	3-Cl-4-OCF3	Ac	0	>200
15	3,4-di-F	H	0	3.5 ± 0.2	40	H	Ac	1	>200
16	2-Me-5-Cl	H	0	15.0 ± 0.6	41	3,4-di-Cl	Ac	1	>200
17	2,6-di-Me-3-Cl	H	0	30.5 ± 5.4	42	4-Ac	Ac	1	>200
18	2-OMe	H	0	28.9 ± 1.6	43	4-OMe	Ac	1	>200
19	3-OMe	H	0	40.2 ± 1.4	50	3,4-di-Cl	H	1	5.4 ± 0.7
20	4-OMe	H	0	111.8 ± 8.9	51	3,4-di-F	H	1	10.1 ± 1.4
21	2,4-di-OMe	H	0	45.0 ± 0.8	52	4-Ac	H	1	>200
22	3,4-di-OMe	H	0	160.2 ± 10.1

3,4-Di-halo pattern in 13, 15, and 50 proved to be beneficial for the activity. Means and SD of at least two independent experiments.

SAR of Novel Succinimides on ColH

In our previous work, we have shown that there is a structural similarity between the inhibitors of LasB and ColH.[29,30] It was therefore of interest to investigate the activity of all new compounds against ColH and to compare the SAR with that observed for LasB. Figure represents the inhibition of the peptidase domain of ColH (ColH-PD) in the presence of 1 μM of the selected compounds. Previously, we reported that polar substituents in para-position have the most beneficial effect on the activity of N-aryl mercaptoacetamides.[30] Here, we observed the same trend with the new succinimide class, with compound 25, bearing a para-acetyl substituent, being the most active one with 95% of ColH-PD inhibition. The para-methoxy derivative 20 was slightly less active with 74% of inhibition, but still following the trend of polar substituents being more favorable for the activity than nonpolar, lipophilic substituents, such as chlorine. The protection of the free thiol functional group proved to be detrimental for the activity, as in the case of LasB. The two compounds showing >50% of inhibition of ColH-PD were further tested in a dose–response manner in the presence of the reducing agent TCEP. This experiment revealed nanomolar IC50 values for 25 (0.06 ± 0.01 μM) and 20 (0.32 ± 0.05 μM). Overall, since the inhibitors presented in this work have a similar structure and SAR we described for N-aryl mercaptoacetamides, we expect the interactions with LasB and ColH to be similar as those in our previously published co-crystal structures.[29,30]

Figure 3

Inhibition of the peptidolytic activity of ColH-PD in the presence of 1 μM of the respective compound in a FRET-based assay: para-Acetyl substituent in compound 25 proved to be most favorable for the activity. Data are presented as means and SD of three independent experiments.

We consider the difference observed in SARs between LasB and ColH inhibitors beneficial in terms of their selectivity for each particular target. Each hit can be further developed for the treatment of single infections caused by either P. aeruginosa or C. histolyticum. However, wound infections, due to their nature, are likely to be colonized with other bacteria.[37,38] Therefore, having a common structural motif that inhibits both targets is particularly interesting, and if the inhibitors were to be used for Clostridia and P. aeruginosa co-infections purpose, their structure could be further optimized and adapted as dual inhibitors of ColH and LasB.

Broad-Spectrum Inhibition of Other Bacterial Collagenases

In addition to ColH from C. histolyticum, other Clostridium and Bacillus species also secrete collagenases that play pivotal roles in the pathogenesis of these bacteria by destroying the connective-tissue components in the infected host.[16] We therefore tested the two most active ColH-PD inhibitors (20 and 25) on three additional collagenases, using the collagenase unit of ColG (ColG-CU) from C. histolyticum, the peptidase domain of ColT (ColT-PD) from C. tetani, and the collagenase unit of ColQ1 (ColQ1-CU) from B. cereus strain Q1. As anticipated, the succinimide-based scaffold retained the broad-spectrum inhibitory properties of the mercaptoacetamide-based inhibitors (Table ).[30]

Table 2

Inhibition of ColH-PD, ColT-PD, ColG-CU, and ColQ1-CU in the Presence of 100 μM of Compounds 20 and 25a

	% inhibition @ 100 μM
compd	ColH-PD	ColG-CU	ColT-PD	ColQ1-CU
20	95 ± 1	85 ± 3	89 ± 4	99 ± 1
25	96 ± 2	100 ± 2	102 ± 3	99 ± 2

Means and SD of at least two independent experiments.

Selectivity against MMPs, HDACs, and TACE as Human Off-Targets

Previously, we described N-aryl mercaptoacetamides with high selectivity for the bacterial over a broad range of human MMPs.[29,30] MMPs are calcium-dependent zinc metalloproteases that play a pivotal role in numerous biochemical processes in humans.[39,40] Based on the depth of their S1′ binding pocket, MMPs can be divided into three classes: deep (e.g., MMP-3 and -14), intermediate (e.g., MMP-2 and -8), and shallow (e.g., MMP-1 and -7). With the aim to explore the interactions of our inhibitors with all three pocket types, which could help us to assess potential effects on other not-tested representatives, we chose a panel of six MMPs, comprising two members of each class. In addition, HDAC-3, HDAC-8, and TACE (ADAM-17)—enzymes involved in gene expression and the processing of TNF-α,[41,42] respectively—were selected as important additional human off-targets. Our results showed that the most potent inhibitors of LasB and ColH (13 and 25, respectively) fortunately possess a high selectivity over most of the off-targets tested. While both compounds did not inhibit MMP-1, -3, and -7 as well as both HDAC enzymes, we observed certain inhibition of MMP-2, -8, and -14 at 100 μM. Inhibition of TACE, which was observed for both compounds, will be considered as high priority in the future optimization of the structures (Tables and 4).

Table 3

Inhibition of Six MMPs in the Presence of 100 μM of Compounds 13 and 25a

	% inhibition @ 100 μM
compd	MMP-1	MMP-2	MMP-3	MMP-7	MMP-8	MMP-14
13	n.i.	39 ± 32	n.i.	n.i.	84 ± 8	n.i.
25	n.i.	14 ± 4	n.i.	n.i.	94 ± 1	84 ± 8

Means and SD of at least two independent experiments. n.i. = <10% inhibition.

Table 4

Activity of Compounds 13 and 25 against HDAC-3, HDAC-8, and TACEa

	IC50 (μM)
compd	HDAC-3	HDAC-8	TACE
13	>100	>100	5.2 ± 0.6
25	>100	>100	3.4 ± 1.2

Means and SD of at least two independent experiments.

Cytotoxicity Assays

Broad-spectrum inhibition of bacterial collagenases and selectivity against a panel of off-targets supported the further evaluation of the compounds’ toxicity in vitro. In this context, we investigated 13 and 25, the two most active compounds against both targets, for their cytotoxicity against the three human cell lines: HepG2 (hepatocellular carcinoma), HEK293 (embryonal kidney), and A549 (lung carcinoma). Neither of the compounds showed cytotoxic effects, with IC50 values >100 μM (Table ), making them suitable for further investigation in in vivo model systems. Compared to our previous hits, 5 and 6, they displayed similar or even lower toxicities in most of the cell lines tested. Particularly, compound 25 proved to be even less toxic than 6, which showed an IC50 of 100 μM in HEK293 cells.

Table 5

Cytotoxicity of Compounds 13, 25, 5, and 6 against HepG2, HEK293, and A549 Cell Linesa

	IC50 (μM)
compd	HepG2	HEK293	A549
13	>100	>100	>100
25	>100	>100	>100
5	>100	>100	>100
6	>100	100	>100

Means and SD of at least two independent experiments

In Vivo Toxicity in Zebrafish-Embryo Model

Due to the promising in vitro activities against antivirulence targets LasB and ColH and the lack of cytotoxicity against three human cell lines, we subjected compounds 15 and 25 to a toxicity study based on zebrafish embryos. An advantage of this nonmammalian in vivo model is the high genetic homology to humans and that it provides follow-up information on the type of toxicity encountered (e.g., hepatic, cardiovascular, etc.). In addition, this model can also predict mammalian teratogenicity by evaluation of lethality and malformation during the development of embryonic zebrafish.[43,44] Both compounds tested showed a maximum tolerated concentration (MTC) of ≥30 μM, which is higher than for the corresponding mercaptoacetamide-based LasB inhibitor 5 we published previously (MTC = 10 μM) (Table S1).[29]

Ex Vivo Pig-Skin Model

We established an ex vivo model based on pig skin to address the impact of our inhibitors on living mammalian tissue and on the contained collagen as the natural substrate of collagenase. We challenged the skin, prepared from the ear of freshly slaughtered pigs, with pure ColQ1 from B. cereus to degrade collagen. We assessed the activity of ColQ1 by quantifying the formation of hydroxyproline as an indicator for collagen turnover (Figure ). Optimization of the assay conditions for the model consisted of examining different buffer conditions and different protein concentrations (Figures S1 and S2). To evaluate the potential effect of 25 on collagen turnover, we incubated the skin with ColQ1 in the absence and presence of defined concentrations of this compound. The subsequent quantification of hydroxyproline revealed that 25 inhibited the collagenolytic activity of ColQ1, as demonstrated by the significantly reduced amount of hydroxyproline released compared to control (Figure ). These results support our previous finding that 25 is an inhibitor of a broad range of bacterial collagenases. The good performance of our inhibitors in this model is a sound starting point for their subsequent testing under in vivo conditions. We expect that our inhibitors will accelerate the in vivo healing process, by preventing the distribution of infection. The immune system will clear the bacteria, promoting remodeling of collagen and skin regeneration.

Figure 4

Representation of the pig-skin model: The composition of the skin, the dermal layer, and the amino acids of collagen are illustrated. The concept of the hydroxyproline quantification assay is explained by mixing chloramine-T and Ehrlich’s reagent to obtain a product that can be detected with a spectrophotometer.

Figure 5

Amount of hydroxyproline at different concentrations of 25. Data shown represent the means ± SD from three independent measurements. One-way ANOVA followed by Tukey’s HSD test (** = p < 0.01, **** = p < 0.0001).

Antibacterial Activity

The aim of this study was to create “pathoblockers” that target bacterial virulence factors without directly affecting bacterial viability. These have the ability to disarm, rather than to kill pathogens in order to make them less pathogenic. In this context, it was of interest to test the antibacterial activity of the new derivatives against P. aeruginosa strain PA14 and C. histolyticum strain DSM 1126 to exclude growth inhibition by selected compounds. We therefore selected the four most potent compounds against LasB (13, 15, 50, and 51) and the most active succinimide against ColH (25). As shown in Table , it is clear that compounds 13, 15, 50, and 51 did not affect the growth of PA14 and 25 did not affect the growth of DSM 1126, with a minimal inhibitory concentration (MIC) > 100 μM. This assures that the activity of these compounds is only through targeting the secreted bacterial virulence factors.

Table 6

Antibacterial Activity of Selected Compounds against P. aeruginosa and C. histolyticuma

	MIC (μM)
compd	P. aeruginosa	C. histolyticum
13	>100	n.d.
15	>100	n.d.
50	>100	n.d.
51	>100	n.d.
25	n.d.	>100

n.d. = not determined.

Conclusions

In this study, we aimed to constrain the conformation of previously published N-aryl mercaptoacetamide inhibitors,[29,30] which led to a series of novel succinimide inhibitors of the antivirulence targets LasB and ColH. Among these compounds, we identified 13 and 25 to show the best potency against LasB and ColH, respectively. Compound 13 displays two-fold improved activity against LasB compared to our previous hit 5, while compound 25 was slightly less potent than 6 against ColH; however, it maintained the activity in the nanomolar range. Both compounds showed sufficient selectivity for the bacterial metalloproteases over human MMPs and three other off-targets. The two most active compounds against ColH, 25 and 20, showed in addition broad-spectrum inhibition of homologous bacterial collagenases. These most potent LasB and ColH inhibitors showed no signs of cytotoxicity in three human cell lines. Interestingly, compounds 15 and 25 showed a MTC of ≥30 μM in our zebrafish model. This is, for 15, a 3-fold lower toxicity compared to our published LasB hit 5. Moreover, we have established a pig-skin model to further characterize the most promising collagenase inhibitor. In this ex vivo assay, compound 25 showed a promising effect in preventing collagen degradation, which paves the way for this compound’s further evaluation under in vivo conditions. To investigate the compounds in such in vivo models, additional work should be carried out to improve their pharmacokinetic profiles. However, we strongly believe that the new succinimide inhibitors presented in this work have a great potential to be developed further and to be used as therapeutics.

Experimental Section

Chemistry

All reagents were used from commercial suppliers without further purification. Procedures were not optimized regarding yield. NMR spectra were recorded on a Bruker AV 500 (500 MHz) spectrometer. Chemical shifts are given in parts per million (ppm) and referenced against the residual proton, 1H, or carbon, 13C, resonances of the >99% deuterated solvents as internal reference. Coupling constants (J) are given in Hertz (Hz). Data are reported as follows: chemical shift, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, dd = doublet of doublets, dt = doublet of triplets, br = broad and combinations of these) coupling constants, and integration. Liquid chromatography-mass spectrometry (LC-MS) was performed on a LC-MS system, consisting of a Dionex UltiMate 3000 pump, autosampler, column compartment, and detector (Thermo Fisher Scientific, Dreieich, Germany) and ESI quadrupole MS (MSQ Plus or ISQ EC, Thermo Fisher Scientific, Dreieich, Germany). High-resolution mass was determined by LC-MS/MS using Thermo Scientific Q Exactive Focus Orbitrap LC-MS/MS system. Purity of the final compounds was determined by LC-MS using the area percentage method on the UV trace recorded at a wavelength of 254 nm and found to be >95%. Melting points were determined by using a Stuart melting point SMP30 device.

General Procedure A: Synthesis of Succinimides 7–27 and 50–52

Mercaptosuccinic acid (1.0 equiv) and the corresponding aniline (1.0 equiv) were mixed in a crimp vial under Ar atmosphere and heated at 120–160 °C from 3.5 h to overnight. The crude product was purified using column chromatography. In case of N-aryl-3-mercaptomethylsuccinimides, 4-(aminoaryl)-2-(mercaptomethyl)-4-oxobutanoic acid was heated at 120 °C overnight.

General Procedure B: Synthesis of Thioacetates 28–30 by Acetylation of Free Thiol

Succinimide (1.0 equiv) was dissolved in DCM, and the solution was cooled in an ice bath. Pyridine (2.0 equiv) and DMAP (0.1 equiv) were added, followed by dropwise addition of Ac2O (2.0 equiv). After 30 min at 0 °C, the reaction mixture was allowed to warm up to room temperature (rt) and stirred overnight. Volatiles were evaporated under reduced pressure, and crude product was purified using column chromatography.

General Procedure C: Synthesis of α-Itaconamic Acids 31–35

α-Itaconamic acids were synthesized following the procedure described in the literature.[35] Itaconic anhydride (1.0 equiv) was dissolved in CHCl3. The corresponding aniline (1.0 equiv) was added to the vigorously stirring solution. After 2 h, the product was collected by filtration and washed with a small amount of chloroform. The product was used in the next step without further purification.

General Procedure D: Synthesis of Itaconimides 36–39

Itaconimides were synthesized following the procedure described in the literature from intermediate α-itaconamic acids described in general procedure C.[35] α-Itaconamic acid (1.0 equiv) was mixed with NaOAc (0.5 equiv) and Ac2O (3.5 equiv) and heated at 100 °C for 1–2 h. The dark reaction mixture was cooled to rt, poured into ice-cold water, and extracted 3 times with EtOAc. Combined organic layers were washed with brine and dried over anh. Na2SO4, filtered and the solvent removed under reduced pressure. The crude product was purified using column chromatography. In all cases except in the case of the 4-OMe derivative, the corresponding citraconimides were isolated as a side product, and therefore the yield of obtained itaconimides was low to moderate.

General Procedure E: Synthesis of Thioacetates 40–46 Using Michael Addition

Corresponding itaconimide/α-itaconamic acid (1.0 equiv) was dissolved in DME/DCM/THF under Ar atmosphere. Thioacetic acid (1.1–1.5 equiv) was added, followed by Et3N (0.01–0.1 equiv). The reaction mixture was stirred at rt overnight. Crude product was purified using column chromatography or used in the next step without further purification.

General Procedure F: Thioacetate Hydrolysis to Obtain Compounds 47–49

Thioacetate (1.0 equiv) was dissolved in methanol under Ar atmosphere, and 2 M aqueous solution of NaOH (2.0–3.0 equiv) was added. The reaction was stirred 1–2 h at rt. After quenching with 1 M HCl, the reaction was extracted three times with EtOAc. Combined organic extracts were washed with brine and dried over anh. Na2SO4, filtered and the solvent was removed under reduced pressure. The crude product was purified using column chromatography or used in the next step without further purification.

1-(3,4-Dichlorophenyl)-3-mercaptopyrrolidine-2,5-dione (13)

Compound 13 was synthesized according to the general procedure A, using 3,4-dichloroaniline (162 mg, 1 mmol) and mercaptosuccinic acid (150 mg, 1 mmol), at 120 °C overnight. The product was purified using column chromatography (100% DCM). The final product was obtained as white solid (155 mg, 56%, Mp 141 °C). 1H NMR (500 MHz, DMSO-d6) δ ppm: 7.81 (d, J = 8.5 Hz, 1H), 7.62 (d, J = 2.0 Hz, 1H), 7.34 (dd, J = 2.0, 8.5 Hz, 1H), 4.10 (dd, J = 4.5, 9.0 Hz, 1H), 3.87 (s, 1H), 3.37 (dd, J = 9.0, 18.0 Hz, 1H), 2.73 (dd, J = 4.5, 18.5 Hz, 1H). 13C NMR (126 MHz, DMSO-d6) δ ppm: 176.2, 173.8, 132.4, 131.2, 131.1, 131.0, 128.9, 127.5, 39.2, 34.9. HRMS (ESI–) m/z calcd for C10H6Cl2NO2S [M – H]− 273.94963, found 273.94931.

1-(4-Acetylphenyl)-3-mercaptopyrrolidine-2,5-dione (25)

Compound 25 was synthesized according to the general procedure A, using 4-aminoacetophenone (200 mg, 1.48 mmol) and mercaptosuccinic acid (222 mg, 1.48 mmol), at 120 °C overnight. The product was purified using column chromatography (Hex/EtOAc = 7/3). The final product was obtained as pale yellow solid (117.1 mg, 32%, Mp 104 °C). 1H NMR (500 MHz, CDCl3) δ ppm: 8.11–8.04 (m, 2H), 7.50–7.44 (m, 2H), 4.14–4.08 (m, 1H), 3.38 (dd, 1H, J = 9.3, 18.8 Hz), 2.77 (dd, 1H, J = 4.3, 18.8 Hz), 2.73 (d, 1H, J = 4.4 Hz), 2.63 (s, 3H). 13C NMR (126 MHz, CDCl3) δ ppm: 196.9, 175.6, 172.7, 136.9, 135.5, 129.2, 126.2, 37.4, 34.2, 26.7. HRMS (ESI–) m/z calcd for C12H10NO3S [M – H]− 248.038687, found 248.03867.

Expression and Purification of LasB and ColH-PD

LasB and ColH-PD were expressed and purified as described previously.[29,45]

In Vitro Inhibition Assays (LasB, ColH, ColT, ColG, ColQ1, MMPs, TACE, and HDACs)

All in vitro inhibition assays were performed as described previously.[29,30] TACE and HDAC inhibitor screening kits were purchased from Sigma-Aldrich (Saint Louis, MO). The assays were performed according to the guidelines of the manufacturer. Fluorescence signals were measured using a CLARIOstar plate reader (BMG Labtech, Ortenberg, Germany).

Cytotoxicity assays on HepG2, HEK293 and A549 cells were performed as described previously.[46]

Zebrafish Embryo Toxicity

Toxicity testing was performed according to the procedure described in the literature[47] with minor modifications using zebrafish embryos of the AB wild-type line at 1 day post-fertilization. Embryos were collected and kept in a Petri dish at 28 °C until the next day in 0.3× Danieau’s medium (17 mM NaCl, 2 mM KCl, 1.8 mM Ca(NO3)2, 1.5 mM HEPES (pH 7.1–7.3), 0.12 mM MgSO4, and 1.2 μM methylene blue). The toxicity assay was performed using a 96-well plate with one embryo per well and 10 embryos per condition. To obtain compound concentrations between 2 μM and 100 μM, solutions of 15 and 25 were prepared freshly using 0.3× Danieau’s medium with a final DMSO concentration of 1% (v/v). Single zebrafish embryos were placed in wells and directly incubated in the corresponding compound solutions. Monitoring of developmental defects, heart rate, touch-evoked locomotion response, and survival rate was done daily (up to 120 hpf) via microscopy (Table S1). All of the described experiments were performed with zebrafish embryos <120 h post-fertilization (hpf) and are not classified as animal experiments according to EU Directive 2010/63/EU. Protocols for husbandry and care of adult animals are in accordance with the German Animal Welfare Act (§11 Abs. One TierSchG).

Ex Vivo Pig-Skin Model

Skin explants were obtained from freshly slaughtered pig ears, which were supplied by a local slaughterhouse. The explants were made using sterile medical biopsy punches (pfm medical, Cologne, Germany) with a diameter of 5 mm. The skin was washed once each with 70% isopropanol and sterile water and three times with Dulbecco’s modified Eagle medium (DMEM) (Thermo Fisher Scientific, Schwerte, Germany) containing 1% penicillin and streptomycin. The punches were stored in DMEM medium and 15% glycerol at −80 °C until the time of the experiment. To do the experiment, a mixture of 300 nM of ColQ1, 4 mM CaCl2, 10 μM ZnCl2, and DMEM medium was prepared. The compound was pre-incubated with the mixture for 1 h at 37 °C and 5% CO2. Afterward, one skin explant was added to each well in a 24-well plate and incubated in an incubator at 37 °C and 5% CO2 while shaking at 300 rpm. Aliquots of DMEM medium were taken at different time points in order to measure the formed hydroxyproline using a hydroxyproline assay kit (Sigma-Aldrich). This assay was performed according to the protocol of the manufacturer. Absorbance was measured using a PHERAstar plate reader (BMG Labtech). The absorbance values were converted into the hydroxyproline concentration (μg/mL) using the calibration curve of hydroxyproline as a reference (Figure S3).

Bacterial Growth Inhibition Assay

Assays regarding the determination of the MIC were performed as described recently for P. aeruginosa PA14.[48] MICs concerning C. histolyticum (Hathewaya histolytica (Weinberg and Séguin 1916) Lawson and Rainey 2016) DSM 1126 strain were performed in brain heart infusion (BHI) medium. The McFarland standard was adjusted to 2, followed by predilution of 1:100. The dilution series of the substances (100 μM, 50 μM, 25 μM, 12.5 μM, 6.75 μM, and 3.13 μM final concentration) was carried out in a 96-well plate in BHI and mixed with the bacterial suspension. The plates were subsequently incubated at 37 °C for 48 h under anaerobic conditions, followed by growth control and evaluation of MIC values. The given MIC values are means of at least two independent determinations.

Screening of the Compounds for PAINS and Prediction of BBB Penetration

All of the compounds that were tested in biological assays were screened for PAINS and the possibility of BBB penetration using StarDrop software, Optibrium Ltd., Cambridge, UK (Table S2).