High Throughput Screening of a Library Based on Kinase Inhibitor Scaffolds Against Mycobacterium Tuberculosis H37Rv

Summary

Kinase targets are being pursued in a variety of diseases beyond cancer, including immune and metabolic as well as viral, parasitic, fungal and bacterial. In particular, there is a relatively recent interest in kinase and ATP-binding targets in Mycobacterium tuberculosis in order to identify inhibitors and potential drugs for essential proteins that are not targeted by current drug regimens. Herein, we report the high throughput screening results for a targeted library of approximately 26,000 compounds that was designed based on current kinase inhibitor scaffolds and known kinase binding sites. The phenotypic data presented herein may form the basis for selecting scaffolds/compounds for further enzymatic screens against specific kinase or other ATP-binding targets in Mycobacterium tuberculosis based on the apparent activity against the whole bacteria in vitro.

Table 1.

Table 2.

Table 10.

2-Acylaminothiopene-3-carboxamide and related compounds

In the library of compounds evaluated, there were a relatively large number (1,376 out of 25,671) of amide derivatives of 2-aminothiophene-3-carboxylic acids. Most of these compounds also have a fused ring system such as a tetra or pentamethylene or a tetrahydropyridine system at the 4–5 positions of the thiophene moiety. The number of compounds possessing the 2-acylaminothiophene moiety with a primary amide (1, Figure 1) in the screened set was 487. Of these, in the primary assay, 41 compounds displayed >86% inhibition of the growth of Mtb at 10 μg/mL. From this group, confirmed dose-response data were available for 32 compounds. Most of these compounds, however, displayed either poor IC₉₀ values or displayed significant cytotoxicity against Vero cells thus yielding poor SI values. Only two compounds, 2 and 3 (Figure 1) emerged as compounds with moderate to good activity against Mtb coupled with greater than 10-fold selectivity index values.

Pyrazolo[1,5-a]pyrimidines

In the screening set, there were 37 compounds possessing the pyrazolo[1,5-a]pyrimidine framework. Most of these compounds possessed an amino substituent at the 7-position, an aryl group at the 3-position and an alkyl, primarily a methyl group, at the 5-position. The inhibition potencies of these compounds in the primary assay covered the entire range from 100% to 0%. Of these 37 compounds, 11 compounds displayed > 85% inhibition in the primary assay. Five of these 11 compounds (4a–e) that displayed activity in the dose response assay against Mtb without attendant cytotoxicity against Vero cells (SI about or greater than 10) are presented in Table 2.

Tetrahydrobenzo[1,4]diazepin-2-ones

There were 48 compounds within the evaluated set that were 4-phenyl or substituted phenyl 1,3,4,5-tetrahydrobenzo[e][1,4]diazepinones acylated on the nitrogen at the 4-position with various acyl groups (5). In the primary screen, 20 displayed inhibition potencies > 85%. In the dose response assay, however, only five compounds displayed IC₉₀s in the range of 5–10 μg/mL. Most of these compounds also displayed significant cytotoxicity against Vero cells, leading to modest SI values in the range of 2.7 to 7.9. Compound 6 gave the highest selectivity (Figure 2).

Substituted 1,2,3-benzotriazin-4(3H)-ones

There were 14 1,2,3-benzotriazin-4(3H)-ones, and, of these, five compounds had IC₉₀s ≤ 10 μg/mL. Only one of these compounds gave an SI of >10 after cytotoxicity screening. The most active sample, 7a (Table 3), gave an SI of >28 resulting from its IC₉₀ of 1.4 μg/mL and cytoxicity of >40 μg/mL. The three most active and selective examples of this class all contained R₁ = aryl. It is notable that this class contains a labile ester linkage that may serve as a prodrug form of the 1,2,3-benzotriazine-4(3H)-one core heterocycle.

A search of the literature did not reveal any compounds of this general structure with tuberculosis activity or kinase activity. Some 1,2,3-benzotriazin-4(3H)-ones have been prepared and evaluated as metalloproteinase inhibitors.³⁴ Similar compounds were also among derivatives reported as compounds with potential chemotherapeutic use.³⁵ Structurally similar 3-aryl-1,2,3-benzotriazin-4(3H)-ones were also found to have antimycobacterial activity that was weaker than that of some corresponding 2-aryl-2H-1,3-benzoxazine-2,4-(3H)-diones.³⁶ Similar compounds have also been shown to exhibit antimicrobial and marine antifouling activity in industrial and commercial applications.³⁷ An analog, azinphosmethyl (Figure 3), to which tufted apple budmoth larvae (P. idaeusalis) are susceptible, has been reported.³⁸

Substituted benzopyran-2-ones

The 2H-1-benzopyran-2-ones (commonly referred to as coumarins) 8a–i shown in Table 4 have consistent activity, low cytotoxicity as measured in Vero cells, and consequently high Selectivity Index values. These compounds, with alkoxy ester groups at C-7 of the coumarin ring are very similar to a class that we presented in an earlier publication.² The background for this class and the anti-tuberculosis activity are presented therein. In the kinase library examined, there were 11 compounds that possessed the core structure represented by a coumarin ring with a 3-phenyl moiety and a 7-carbomethoxymethoxy group. Of these, the nine compounds in Table 4 possessed significant activity. It is very clear that this class of compounds has consistent activity and warrants expanded research efforts in the search for new compounds with new mechanisms of action.

2-Aminobenzothiazoles

In general, the benzothiazole core is only poorly represented in the antibacterial literature, but is more commonly seen in the kinase inhibitor literature. The reported compounds inhibited a variety of kinases including SHP-2,³⁹ JNK kinases,^40–42 ROCK-II,⁴³ and FLT-3.⁴⁴ A smaller subset of reported compounds contained the 2-aminobenzothiazole moiety and has been reported to inhibit P38α MAP kinase,⁴⁵ Raf-1,⁴⁶ LCK,⁴⁷ and others.⁴⁸ While some similarities to the current screening set exist, it must be emphasized that beyond the library selection criteria it is not clear that these “kinase-like” inhibitor sets will affect any of these kinases, nor is it clear that structurally related kinases or specific, related ATP binding sites exist in Mtb that could be the target of active samples from the screening set.

There were several small clusters of active compounds enriched in the 2-aminobenzothiazole core. The great majority of samples include a 2-aryl- or 2-heteroarylbenzothiazole core. One notable singleton (non-clustered hit) that contained a 2-pyrrolidine, but not a 2-heteroaryl substituent, is given in Figure 4 (structure 9), suggesting that alternative 2-substituents on the benzothiazole core should be explored for activity.

There were a small number (three – see general structure 10, Figure 4) of 2-phenylamino-substituted benzothiazoles of limited diversity that gave modest IC₉₀ values and showed significant toxicity and poor overall selectivities. Other, heteroaryl substitutions include 2-amine-linked thiazoles, benzothiazoles, and tetrahydrobenzothiazoles, but, for the most part, the small numbers of compounds did not lend themselves to a clear structure activity relationship (SAR) pattern, were modestly active and did not show significant selectivity. Examples of 2-aminobenzothiazoles that showed good to high activity and some degree of selectivity are presented in Tables 5–7.

There were three examples of 4-(2-pyridyl)-2-aminothiazoles of the type depicted by 11, not a sufficient number to allow SAR discussion. On the other hand, modest differences in substitution in terms of electron withdrawing potency and substitution position can apparently have a significant impact on toxicity and the resulting selectivity (see 11a and 11b in Table 5). These effectors could alter hydrogen bonding and/or chelation ability for this class, but more information is needed to determine what requirements are necessary for optimal potency and selectivity. Two other examples of the class are given by 12 and 13 (Fig. 5), again both showing significant differences in activity and selectivity with relatively modest structural alterations.

Another cluster of five compounds contains the 2-aminobenzothiazole core linked to a coumarin-3-carboxylic acid as an acylhydrazide. The majority of these (4/5 – see Table 6) compounds showed modest activity (> 1.0 μg/mL) and modest selectivities (>9.5 to >19). Again, there was not sufficient information to ascertain a clear SAR pattern. Additionally, the acylhydrazide linkage can be labile in M. tuberculosis (e.g. isonicotinic acid hydrazide, the active antitubercular drug INH), and there is a distinct possibility that these compounds may act similarly as prodrugs. Testing against INH resistant strains that lack the activation enzyme (e.g. catalase/peroxidase) may shed some light on the mechanism of these compounds and yield important information relevant to their potential value as INH resistance is now commonplace. Two other structurally related active samples (15 and 16, Figure 6) that have similar activities and selectivities are shown below.

The remaining active cluster that contains the 2-aminobenzothiazole core showed significant activity (many IC₉₀s < 1.0 μg/mL) and good selectivities (as high as >83). Overall, there were 50 representatives in the cluster with 18 of these giving significant activity and selectivity as shown in Table 7. The active and selective samples in the cluster all contain a furan-2- (17a–j) or thiophene-2- (18a–e) carboxylic acid amide linkage with a small number of similar actives of structure 19 (a–c) that contain a 3-(2-thienyl)acrylic acid amide linkage. Within the cluster of 50 compounds, there were a significant number of samples that contained a substituted benzoic acid, phenylacetic acid, or 2-phenoxyacetic acid amide linkage, but these compounds were, for the most part, significantly less active (IC₉₀ range 1.6 to 10 μg/mL) and selectivities ranging from 0.9 to 4.0. Overall, while there were a greater number of analogs within this particular set than with the other 2-aminobenzothiazoles, a distinct SAR pattern was not clear, and, from the variety of substitutions screened, it was a clear indication that a larger diversity set, as well as specific examples for comparison analysis, need to be explored. The identification of a specific target or targets would help in profiling the activity of this compound class, and preliminary animal screening of an active analog to ascertain bioavailability and activity in an efficacy model would help prioritize the class.

Substituted 2-benzylidenebenzofuran-3(2H)-ones

These compounds, also known as aurones, have demonstrated some antibacterial or antifungal activity as reported in several citations (see below). Aurones as originally identified are yellow naturally occurring pigments derived from plants.⁴⁹ They are flavonoid compounds, thus identifying them with a class of compounds with significant biological activity. The synthetic compounds 20a–c in Table 8 demonstrated significant activity as evidenced by their IC₉₀ and SI values. The kinase-like inhibitor library contained a total of 63 compounds with the base aurone structure, and only these three showed reproducible anti-TB activity.

A series of aurones was prepared through the oxidative cyclization of 2′-hydroxychalcones, and these compounds were found to have moderate activity against both Staphylococcus aureus and Escherichia coli.⁵⁰ Several patents have focused on either the antibacterial activity or the inhibition of bacterial chorismate synthase, an enzyme shown to be essential for bacterial viability. In one case,⁵¹ a series of aurones was prepared and evaluated for inhibition of Streptococcus pneumoniae chorismate synthase. A series of (2Z)-6,7-dihydroxybenzylidenebenzofuran-3(2H)-ones with various substitutions on the phenyl ring was prepared, and data demonstrating significant inhibition of this enzyme were presented.⁵² In another patent, aurones substituted both on the benzofuranone ring and the phenyl ring were found to significantly inhibit the growth of Streptococcus aureus KLE820 at 10 μg/mL.⁴⁹ Finally, a substituted aurone demonstrated modest inhibition of glucan synthase, possibly suggesting the existence of antifungal activity.⁵³

Thus, this class of compounds appears to have significant biological activity, and some enzyme targets have been suggested as potential leads for new antibacterial discovery. In the case of tuberculosis, however, only a fraction of the compounds had activity, and a molecular target still remains to be identified. The key question of selectivity would also need to be carefully considered, but there is clear potential for new drug discovery in the aurone class of compounds.

Substituted 2-(benzimidazol-2-yl)acrylonitriles

The core scaffold was present in 20 compounds in the target library. Two compounds (21a,b) from that group are shown in Table 9 and were found to have reproducible activity. All the compounds had R₁ as an aryl or heteroaryl moiety, and some cytotoxicity was seen for three of the four compounds subjected to follow-up assays, the exception being compound 21a. It is therefore difficult to draw any solid conclusions as to the value of these compounds as potential leads.

This general class of compounds can be found in a series of literature references that mention antibacterial activity.^54–58 Specific antibacterial activity was reported several times,^54–57though when cytotoxicity data was reported,⁵⁴ it was clear that this series, with R₁ (see structure in Table 9) as a variety of heteroaryl moieties, had significant cytotoxicity and little, if any, selectivity. In the one report that focuses on specific enzyme inhibition,⁵⁸ two compounds with R₁ as a furanyl salicylate demonstrated reasonable inhibition of the Yersinia pestis tyrosine phosphatase YopH (2 μM and 14 μM). This enzyme is of potential therapeutic interest because Y. pestis strains lacking the protein are avirulent. No data against the organism itself, however, was presented. Other types of compounds in the library evaluated in this report⁵⁸ were more potent inhibitors of the enzyme, and there have been no further reports on the two compounds containing a benzimidazolyl acrylonitrile unit. A series of compounds closely related to the title compounds but with a benzotriazole moiety rather than a benzimidazole moiety have been reported.^59–60 These compounds were initially found to have some antitubercular activity, but investigation of the cytotoxicity profile indicates that in fact, there was no selectivity and that the compounds were significantly cytotoxic.

To summarize this small series, some selectivity has been seen, but because of the cytotoxicity profile of the series in general, it is critical that any exploration of SAR with M. tuberculosis or any other bacteria be accompanied by a careful examination of the cytotoxic effects of the new compounds.

1,3,4-Oxadiazoles

There are numerous literature examples of the 2,5-disubstituted-1,3,4-oxadiazole system that have shown biological activities including anticancer (apoptosis induction, mitotic arrest, kinase inhibition, cell proliferation arrest etc.), anti-inflammatory, antifungal, antiviral, and antibacterial. In particular, 1,3,4-oxadiazoles are known from the literature to possess potent antibacterial or antimycobacterial activity.^61–75 Among the evaluated compounds, one series in particular, the 2-carboxamido-1,3,4-oxadiazoles, showed high activity and good selectivity for a number of samples.

The library evaluated contains 1,045 compounds that share a 2-carboxamido-1,3,4-oxadiazole core scaffold (22, Table 10). Approximately half (522 compounds) of these samples contained R₂ = phenyl and R₅ = phenyl/pyridyl. Of these, 91 compounds have potencies IC₉₀ < 10 μg/mL, out of which 18 hits also show cytotoxicity > 40 μg/mL. Within this set of 91 hits, potent IC₉₀ and high SI values were associated with specific substituent groups at R₂ and R₅ compared to inactive compounds. For example, R₅ = pyridyl appears favored over phenyl (e.g. 22a is potent and relatively non-toxic while six close analogs of 22a all containing a substituted phenyl for R₅ are inactive). In another typical example, substituting phenyl for the pyridyl group in 22b rendered this compound inactive. 1,2,3,4-Tetrahydronaphth-6-yl at R₅ also appears favored, 22c being a representative of a series of such compounds. The tetrahydronaphthyl may be substituted for R₂ as well as in compound 22d, yielding active and selective analogs. R₂ substitutions that were associated with potent and non-toxic hits are: 4-t-butyl (22e and 22f), 3-or 4-O-n-butyl (22g and 22h), 3-O-phenyl (22i). Dimethyl substitutions on the R₅ ring were also well tolerated yielding potent analogs, such as compounds 22j and 22k, although these 16 compounds display a range of toxicities from relatively non-toxic compounds with SI > 97 (22l) to toxic analogs such as 22m with SI < 1.0. In general, using a selectivity index of 15 as the criteria between non-toxic and toxic compounds, half of this series were relatively active and non-toxic compounds. Certain substitutions were not favorable on R₂/R₅ such as 3,4,5-tri-OMe substituents on R₅, rendering all 18 such analogs inactive. In another such case, out of 43 analogs containing benzodioxin at R₅, there were only two hits that were potent, and both these samples were also considered relatively toxic. Further, all 53 members of the analog series with a p-sulfonyl-amine substitution on R₂ were either inactive or highly toxic. Of the remaining 523 carboxamido-oxadiazole compounds, 336 have R₅ = phenyl/pyridyl and R₂ a group other than phenyl. The following R₂ atom groups were associated with potent analog series. A furyl or thiophenyl at R₂ is present in 26 potent analogs out of 58 such compounds in the library, for example 22n, 22o, and 22p. A few thiophen-3-yl analogs were also active (22q). Alkyl groups at R₂ were well represented among potent and non-toxic active hits and 19 out of 43 such compounds had IC₉₀s < 10 μg/mL (see 22r – 22v in Table 10). Analog series with the most optimal R₂ alkyls (e.g. isopropyl or t-butyl) appeared sensitive to R₅ phenyl substitutions (e.g. -OMe at any position tends to decrease activity). Further R₂ groups that were tolerated though under-represented among actives were benzothiazol-2-yl (22w), thiophene-2-ethenyl (22×), dihydro-1,4-dioxin-5-yl (22y), and cyclohexyl (22z). Examples of unfavorable R₂ groups (largely associated with inactivity) were methylpyrrolidine-2,5-dione and phenylsulfonylpiperidin-4-yl. Out of the remaining 187 oxadiazoles in the screened set, 176 belong to one of the following two analog series: R₅ = dihydro-1,4-dioxin-5-yl (36 compounds) and R₅ = furan/thiophen-2-yl (140 compounds). Potent, nontoxic hits within the latter series showed preference for R₂ = phenyl where R₂ substituents include O-phenyl (22aa), 4-O-n-butyl (22ab), or 4-t-butyl (22ac), and these examples were associated with the most potent hits and are analogous to the R₅ = phenyl/pyridyl. Further preferred R₂ groups are furyl, thiophenyl, benzothiazolyl and others (e.g. 22ad, 22ae, 22af, 22ag). In the R₅ series (R₅ = dihydro-1,4-dioxin-5-yl, 36 compounds) the only potent analogs in that specific grouping were those containing a phenyl at R₂ (e.g. 22ah and 22ai).

Conclusions

Herein, we report the screening of a designed set of kinase-type inhibitors to identify compounds and scaffolds that show significant antitubercular activity in vitro. Although the specific molecular targets of the active compounds have not been identified per se, these active compounds and their basic core scaffolds may serve as a useful basis set for the community of tuberculosis drug design researchers to probe activities of these actives in order to identify new targets, potentially crucial kinase and ATP-binding targets, as well as others. Hopefully, this publicly available data will stimulate new drug design programs and the development of new agents to treat tuberculosis.