Introduction

Tuberculosis (TB) remains the most prevalent infectious disease worldwide caused by a bacterium. In 2020, 9.9 million patients developed TB and 1.5 million deaths were reported by WHO.1 Shorter treatment duration and better therapeutic outcome against multi- and extensively-drug resistant strains of Mycobacterium tuberculosis (Mtb) are important goals of TB drug development.2 The investigation of new classes of active substances acting on targets that have not been exploited so far is an important approach to combat TB.3

8-Nitrobenzothiazinones (BTZs) have proven to be a highly effective antimycobacterial class of substances4 that inhibit the growth of Mtb even in the nanomolar concentration range by interference with arabinan biosynthesis. BTZs act as mechanism-based inhibitors of the enzyme decaprenylphosphoryl-β-D-ribofuranose-2′-epimerase (DprE1), which is located in the periplasmic space of Mtb.5 DprE1 catalyses the oxidation of decaprenylphosphoryl-β-D-ribofuranose (DPR) to the keto-intermediate decaprenylphosphoryl-2-keto-D-erythropentofuranose (DPX) in the synthesis of decaprenylphosphoryl-β-D-arabinose (DPA), a crucial step in the biosynthesis of arabinans.6 Due to their high in vitro efficacy and in vivo activity, BTZs are promising drug candidates for future TB therapy. A wide variety of BTZ derivatives were synthesized and investigated in recent years,4a, 4b, 7 but none of these surpassed the in vitro efficacy of BTZ043 and macozinone (PBTZ169) (Figure 1).

Chemical diagrams of BTZ0434b and macozinone[4a], currently investigated in clinical trials.

Several synthetic routes to BTZs are described in the literature (Scheme 1).7a, 8 All start from substituted 2-chlorobenzoic acid derivatives. In the acylisothiocyanate pathway (A), 2-chlorobenzoyl chlorides are treated with potassium-, sodium-, or ammonium thiocyanate to form the intermediate acylisothiocyanates, which are after 120 minutes treated with the corresponding secondary amines to form a thiourea intermediate that undergoes ring closure by nucleophilic substitution. This pathway was initially used for the synthesis of BTZ043,8b, 8c which was adapted from previous procedures,9 and then further modified by Gao et al 20137a and Peng et al. 2015.7b In the two references mentioned, the procedures for BTZ synthesis were adapted by using phase transfer catalysis (PEG-400) for preparation of acylisothiocyanates and changing the reaction times. As alternatives, the dithiocarbamate pathway (B) and the alkylxanthogenate (C) pathway were developed.8b, 8c A drawback of both pathways is the use of toxic and highly flammable carbon disulfide in the synthesis of the dithiocarbamate and alkylxanthogenate reagents. The alkylsulfanyl pathway (D)8a can easily be adapted for combinatorial chemistry purposes since the amine moiety at position 2 is added to a stable 2-(alkylsulfanyl)-4H-1,3-benzothiazin-4-one intermediate. This method, however, requires carbon disulfide and carcinogenic methyl iodide. We developed a new synthetic pathway,10 which avoids toxic reagents and intermediates with regard to Good Manufacturing Practices for pharmaceuticals11 and green chemistry principles,12 and prepared a panel of new BTZs and derivatives with a broad range of activity and evaluated them against Mtb.

Overview of synthetic procedures for BTZ preparation. From top to bottom: acylisothiocyanate pathway (A), dithiocarbamate pathway (B), alkylxanthogenate pathway (C) and (D) alkylsulfanyl pathway.

Despite the intensive research on DprE1 inhibitors during the last years, surprisingly few data comparing the intracellular activity of a panel of BTZs were published. Intracellular models are known to better mimic the environment of Mtb during the natural course of the disease as they include factors such as compound (in)activation, membrane permeability, removal by efflux pumps, and cytotoxicity to mammalian cells.13 Therefore, we have studied activities of all synthesized compounds in bacterial growth medium and in a high content macrophage infection model.

Results and Discussion A new synthetic pathway to BTZs

In this study, we describe an efficient synthesis for the preparation of BTZs using thiourea intermediates to form the thiazinone ring system in one step, which is shown in Scheme 2.

Preparation of BTZs via the thiourea pathway: (i) toluene, SOCl2, 110 °C; (ii) toluene, N,N-dialkylthiourea, 110 °C.

The synthesis starts with 2-chloro-3-nitro-5-(trifluoromethyl)benzoic acid,14 activated with thionyl chloride to form the acid chloride which subsequently reacted with an N,N-dialkyl thiourea derivative to yield the thiazinone ring system. With the thiourea reagent, it is possible for the first time to introduce the nitrogen and sulphur atoms within one synthetic step and arrive at the thiazinone ring system. Compared to the alkylsulfanyl pathway D (Scheme 1), the most recent BTZ synthesis, fewer synthetic steps are needed and toxic reagents such as carbon disulfide and methyl iodide are avoided. The alkylsulfanyl pathway also starts with the activation of the carboxylic acid, but the carboxylic acid chloride is first converted to the benzamide and is then reacted with carbon disulfide and methyl iodide to the alkylsulfanyl intermediate. The reaction of 2-chlorobenzoic acid chloride with a thiourea reproducibly gives very high yields. For some of the BTZ derivatives described here, e. g. 3 c and 11 c, overall yields of 65–75 % were achieved, which was significantly higher than the literature data for the alkylsulfanyl pathway (30–40 %).8a The thiourea pathway was used to prepare a panel of 31 BTZs (Figure 2). Macozinone and BTZ043 were also synthesized via the thiourea route as a proof of concept.

BTZ derivatives with secondary amine derived substituents in position 2 of the BTZ scaffold.

Synthesis of the intermediate thiourea derivatives

The crucial intermediates in the thiourea pathway to BTZs, viz. the N,N-dialkylthioureas, were accessed by three different procedures. Thioureas were prepared from secondary amines via benzoyl isothiocyanate15 (Scheme 3, a). In the first step of the reaction, benzoyl chloride is reacted with sodium isothiocyanate as shown in Scheme 3 (a). The reactive isothiocyanate formed immediately reacts with the corresponding secondary amine to yield the N-benzoyl thiourea. After hydrolysis the actual thiourea is obtained. As this process needs highly reactive and toxic acyl isothiocyanates, it does not constitute a substantial improvement of the known BTZ syntheses.

Preparation of N,N-dialkylthioureas as key intermediates of BTZ synthesis (Synthesis from benzoyl isothiocyanates (a), Synthesis under use of thiocarbonyldiimidazole (b) or trimethylsilyl isothiocyanate (c): (i) NaSCN, acetone; (ii) secondary amine, acetone (iii) 36 % HCl, 95 °C; (iv) secondary amine, THF; (v) isopropanol/water (9 : 1), reflux; (vi) secondary amine, THF; (vii) 2 N ammonia in MeOH, THF.

Trimethylsilyl isothiocyanate16 can also serve as a reagent for the production of thioureas(Scheme 3, b). In the first reaction step, the secondary amine is reacted with trimethylsilyl isothiocyanate in anhydrous THF. This leads to the nucleophilic addition of the amino nitrogen to the CN double bond of the isothiocyanate. The N-trimethylsilylthiourea can be converted into the corresponding thiourea by brief heating in the presence of water. This process also needs a reactive isothiocyanate.

From thiocarbonyldiimidazole, thioureas can be prepared in an efficient two-step synthesis17 (Scheme 3, c). Thiocarbonyldiimidazole is reacted with the secondary amine in THF, followed by a second substitution reaction with ammonia. The intermediate product formed in the first substitution is used without further purification. This robust procedure tolerates a variety of different secondary amines, even if they contain other functional groups such as tertiary nitrogen atoms, hydroxy or methoxy groups. For comparison, 4-morpholinecarbothioamide was synthesized using the three methods mentioned above. The highest yield of 68 % was achieved using thiocarbonyldiimidazole, whereas methods a and b yielded 16 % and 41 %, respectively. Since the yields are optimised and the mild reaction conditions allow the application to a variety of secondary amines, procedure c obviously is the method of choice for the preparation of N,N-dialkylthioureas. The use of thiocarbonyldiimidazole is limited only by substituents in close proximity to the nitrogen atom of the secondary amine. It is likely that steric hindrance prevents the substitution reaction. For example, the preparation of thiourea from 2,6-dimethyl-piperidine is not possible by method c.

Features of the BTZs in the panel Two crucial sites

BTZ system substituents, most prominently the 8-nitro group18 known to be essential for antimycobacterial activity were assessed in this study.18 As previous studies4a, 7b, 19 clearly show that the side chain at position 2 of the BTZ ring system influences the efficacy of the compounds, we initially focussed on modifications of this site. Secondary amines prepared as substituents in position 2 included piperazine amides and sulfonamides, tetrahydroisoquinolines, diazabicyclononanes and piperidines. In addition, we exchanged the bivalent sulfur atom for oxygen and nitrogen (benzoxazinones BOZs, quinazolinones QZs) as discussed below.

Lipophilic to hydrophilic

Different carboxylic acids provided access to piperazine amides 1 c–9 a with straight and branched alkyl chains as well as amino acids and the drug substance ibuprofen. Thus, both lipophilic and polar BTZ derivatives were prepared to assess their influence on antimycobacterial activity. Two sulfonic acid amides (10 a and 11 c) further diversified the panel of 2-substituents. The spatial extent of this side chain was varied in the piperidine derivatives 12 b–18 b. Additional hydroxyl groups, sulfoxides and sulfones yielded BTZ derivatives of different polarity and spatial extension. Compound 19 b has a larger piperidine-derived side chain that is also present in the approved TB drug delamanid.20

Partially aromatic vs. aliphatic

The tetrahydroisoquinoline system – compounds 20 b–23 b – represents the effect of partially planar aromatic moieties in the 2-position on antimycobacterial activity as compared to the BTZs with piperidine and piperazine and open chain substituents in this position.

Basic vs. less basic

A panel (24 b–27 b) of test compounds with a secondary or tertiary nitrogen atom in the side chain was synthesized, following the lead compound macozinone, but with varying basicity of the respective nitrogen atoms. Heterocycles introduced included the diazabicyclononane moiety (compounds 26 b–27 b) and a diazabicyclooctane (25 b).

Activity assays

The BTZs prepared were expected to cover a range of MICs from nanomolar (e. g. BTZ043, macozinone) to micromolar. The expectation is based on literature data4a, 7, 21 in which structure variations were made on BTZs and a large influence of the substituent at position 2 was shown: Karoli et al.7c used different primary and secondary amines for the preparation of BTZ analogues by the alkylsulfanyl pathway. Some of the analogues prepared in this study had similar efficacy compared to BTZ043. Gao et al.7a and Peng et al.7b used the acylisothiocyanate pathway to prepare BTZ analogues with different substituents at position 2, using piperazines and spiropiperidine derivatives. In a recent study by Zhang et al.,21 mainly spiro cyclic side chains with another tertiary nitrogen were investigated. The compounds synthesized via the alkylxanthoxygenate pathway do not reach the efficacy of the reference compound macozinone, but show improved solubility. The studies mentioned investigate the inhibitory effect of BTZ analogues only in broth and not in a macrophage infection model. The panel of test compounds prepared for this study were synthesized via the new thiourea route described above, demonstrating the suitability of the synthesis method for a variety of different derivatives. A large variety of substituents were introduced into BTZ scaffold to discover the structure-activity relationships (in vitro/intracellular) and to find new derivatives with nanomolar activity. For compound 14 a, the required thiourea precursor was obtained in low yield by the new thiourea route using thiocarbonyldiimidazole due to steric hindrance; consequently, the traditional N-benzoyl isocyanate synthetic pathway was employed (Scheme 1, method A).

The thiourea pathway to BTZs was modified to exchange the sulphur atom with oxygen or nitrogen. For this purpose, the carboxylic acid chloride was treated with ureas or guanidines instead of the thiourea as shown in Scheme 4. The corresponding BTZ analogues – BOZs22 and QZs – were isolated from the reaction mixture, but in lower yields than the corresponding BTZs. For the synthesis of the BOZs, a base, in this case diisopropylethylamine (DIPEA), had to be added, and it proved advantageous to carry out the reaction under argon. The QZ 32 a was only formed when a stronger base was added; however, even with 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) the yield of isolated product was 8 %.

Synthesis of BOZs and QZs by the adapted thiourea pathway: (i) toluene, SOCl2, 110 °C; (ii) Synthesis of BOZs toluene, N,N-dialkylurea, DIPEA, 110 °C (iii); Synthesis of QZs toluene, N,N-dialkylguanidine, DBU, 110 °C.

Structural characterisation of BTZs and analogues

Figure 3 below shows the 1H NMR spectra for analogous BTZs, BOZs and QZs, with all three compounds having the tetrahydroisoquinoline side chain.

1H NMR data (400 MHz, CDCl3, S denotes the residual solvent signal) of 23 b, 31 b and 32 a.

It is noticeable that the aromatic protons of the BTZ 23 b exhibit the largest downfield shift, whereas these protons in the BOZ and QZ are shielded. While there are no differences in the 1H NMR spectrum of the aromatic ring system of tetrahydroisoquinoline, the CH2 groups of this part of the molecule show different characteristics. In the spectrum of the BTZ 23 b, these protons show signal broadening, which indicates a hindered rotation at the C−N bond. The BOZ and QZ analogues show significantly sharper signals for the CH2 groups in question, whereby a signal broadening is also recognisable for the BOZ 31 b.

Compound 1 c was structurally characterized also by X-ray crystallography.23 Figure 4 depicts the molecular structure in the crystal. The X-ray analysis confirmed the structure and ruled out the formation of the 4H-3,1-benzothiazin-4-one structural isomer.24 In the crystal, 1 c exhibits a virtually planar BTZ scaffold, as previously encountered in macozinone25 and other BTZs18b, 26, with the nitro group being nearly coplanar. As expected, the piperazine ring adopts a low energy chair conformation. The bond angles in the piperazine ring show some minor deviations from ideal tetrahedral values due to the planar structure at the two nitrogen atoms. It is worth noting that the opposite enantiomeric conformer of the molecular structure depicted in Figure 4 is also present in the centrosymmetric crystal structure of 1 c.

Molecular structure of 1 c in the crystal. Displacement ellipsoids are shown at the 50 % probability level. Rotational disorder of the trifluoromethyl group is omitted for clarity.

Activity profiling of BTZs, BOZs and QZs against Mtb H37Rv

BTZs and analogues prepared for this study were evaluated for their efficacy against Mtb. For the first time, a panel of numerous BTZ derivatives was investigated in a luciferase-based macrophage infection assay. This method was developed for high-throughput purposes27 and used here to determine the minimum inhibitory concentrations (MIC) resulting in greater than 90 % growth inhibition in liquid culture employing supplemented 7H9 medium as well an intracellular THP-1 macrophage infection model.

All test compounds were evaluated in vitro in a micro broth dilution assay against Mtb H37Rv and the MIC90 was determined. For this in vitro assay, a luciferase-expressing Mtb H37Rv strain was used and mycobacterial growth was quantified using luminescence data.27b, 28 To validate the assay, a z score was determined based on positive and negative controls.29 The reference compounds, BTZ043 and macozinone, showed MICs of 3 and 2 nM respectively, in agreement with literature values.4a, 4b

The BTZ 2-substituent determines whether nanomolar or micromolar MICs are achieved

A very wide range of MICs was observed for the BTZs prepared in this study as shown in Table 1 and Figure 5A. The most active compound 1 e reached a MIC90 of 6 nM, whereas 7 c had a significantly reduced activity of 14 μM. The strong influence of the secondary amine in position 2 of the BTZ scaffold was striking. In the BTZs investigated, this substituent shifted the MIC by a factor of up to 10,000 and is thus of decisive importance for the extraordinary efficacy of particular BTZs. Comparing the BTZ macozinone and 33 b, a benzoxazinone (BOZ) with the same 2-substituent, the strong contribution of the 2-substituent to activity is again highlighted: 33 b has an MIC of “only” 1 μM. However, another BOZ 31 b with a tetrahydroisoquinoline side chain, analogous to BTZ 23 b, was inactive. BTZs, BOZs and QZs are discussed in comparison below.

Table 1. Activity determination and cytotoxicity of BTZs and analogues.

ID

7H9 MIC90 [μM][a]

intracellular MIC90 [μM][b]

cytotoxicity [μM][c]

ID

7H9 MIC90 [μM][a]

intracellular MIC90 [μM][b]

cytotoxicity [μM][c]

1 c

0.2

0.2

21.8

17 b

1.3

1.3

28.8

1 e

0.006

0.1

20.6

18 b

5.0

27.7

2.5

2 a

0.3

0.7

16.0

19 b

0.05

0.4

4.7

3 c

0.05

0.2

20.6

20 b

0.01

0.2

2.5

4 c

0.02

0.2

20.6

21 b

0.2

0.9

>10

5 d

3.8

3.8

21.1

22 b

1.0

3.9

>21.4

6 d

6.9

13.7

>18.3

23 b

0.5

1.1

>24.5

7 c

14.1

28.1

>18.7

24 b

4.1

0.5

5.7

5 c

1.6

1.6

>17.4

25 b

0.2

0.9

2.5

8 c

0.8

1.6

9.1

26 b

1.8

7.5

>20.0

9 a

1.6

0.8

9.1

26 c

4.5

4.5

>25.0

10 c

0.2

0.9

>19.4

26 d

0.9

3.7

>20.4

11 c

0.01

0.2

>19.7

27 b

1.8

3.7

>20.4

12 b

1.3

1.3

29.0

30 b

>43.4

>43.4

>29.0

13 b

1.3

2.5

27.8

31 b

>38.3

19.2

3.2

14 a

0.3

0.5

25.8

32 a

>38.4

>38.4

25.6

15 b

1.2

2.4

13.3

33 b

1.0

0.2

22.7

16 c

4.6

38.1

25.4

macozinone

0.002

0.027

21.9

16 d

1.1

2.2

24.4

BTZ043

0.003

0.014

23.2

[a] Mtb H37Rv in 7H9+OADC; [b] Mtb H37Rv during macrophage infection; [c] MTT assay cytotoxicity against THP-1 cells.

Visualization of extracellular and intracellular MICs of BTZ analogues against Mtb H37Rv (red triangle: macozinone, blue circle: BTZ043). A MICs of BTZ analogues in 7H9+OADC, B MICs of BTZ analogues in macrophage infection assay, C Quotient of extracellular and intracellular MIC (green squares: compounds with an increased intracellular activity 9 a, 24 b and 33 b).

The 2-substituent should be rather lipophilic or weakly basic

The introduction of an N-acyl or sulfonyl piperazine side chain also yielded derivatives with antimycobacterial activity stronger than for reported N-acyl and N-sulfonyl BTZs.7b Of note, 1 e, 4 c and 11 c all had MICs in the lower nanomolar range, with 1 e showing comparable activity to BTZ043 or macozinone with a MIC of 6 nM. Therefore, lipophilic alkyl chains with 5–7 carbon atoms on a piperazine moiety seem to be particularly suitable for profiling the efficacy of BTZ. This might have to do with the ability of a BTZ to penetrate into the outer mycobacterial membrane to where DprE1 is located,5 but this was not investigated in this study. There is certainly an optimum because neither longer lipophilic side chains as in compound 2 a nor the introduction of heteroatoms to increase polarity are beneficial for efficacy, as reflected by significant increases in MICs to >1 μM. Secondary amines derived from piperidine only reached MICs >1 μM, with 14 a representing the optimum in this small series at 0.3 μM. It would appear that a lipophilic side chain is necessary for nanomolar MICs. In compound 24 b, the second piperidine moiety could fulfil this requirement, but the more basic second nitrogen atom counteracts the gain in lipophilicity in the side chain by being basic enough to be protonated at physiological pH. The conclusions are supported by the observation that 19 b with a lipophilic non-basic substituent showed strong inhibition of Mtb growth (MIC: 50 nM).

Aliphatic, not aromatic far end of 2-substituent for low nanomolarf MICs

BTZs with tetrahydroisoquinoline substituents – 20 b–23 b – show distinct activity against Mtb without achieving the outstanding inhibition values of the lead compounds. In particular, compound 20 b with a 6-methoxy-1,2,3,4-tetrahydroisoquinoline substituent stands out for its in vitro MIC of 10 nM. Comparison with its isomer 21 b (7-methoxy, MIC 0.2 μM), 22 b (6,7-dimethoxy, MIC 1.0 μM) and with 23 b (no methoxy substituent, MIC 0.5 μM) proves that subtle differences of the positioning of just a methoxy group at the “far end” of the 2-substituent can cause 100-fold MIC differences. They cannot presently be accounted for drug/target interaction since firstly in vitro MICs are controlled by a number of factors, and secondly, DprE1 structures cocrystallized with BTZ display the domain next to this part of BTZs to be flexible, leading to low resolution of this part of the enzyme.4a, 6c, 30

Annulated aliphatic aza heterocycles in the 2-position do not lead to nanomolar MICs

A series of test compounds with a diazabicyclooctane 25 b and diazabicyclononane 26 b, 26 c, 26 d and 27 b were evaluated for their antimycobacterial activity. Compound 25 b with the 3-benzyl-3,7-diazabicyclo[3.3.0]octane side chain and a MIC of 0.2 μM proved to be the most effective here, while the other derivatives showed a significant increase in MIC compared to the lead structures.

Benzoxazinones and quinazolinones have higher MICs than benzothiazinones

Three benzoxazinones (BOZs) were prepared, with only 33 b showing efficacy against Mtb. This compound contains the methylcyclohexylpiperazine side chain of macozinone, but has a MIC of 1 μM only, corresponding to a 500× reduction of activity on exchanging sulfur for oxygen. The QZ 32 a bears the same tetrahydroisoquinoline side chain as BTZ 23 b and B