Pentapeptide Boronic Acid Inhibitors of Mycobacterium tuberculosis MycP₁ Protease

Abstract

Mycosin protease-1 (MycP₁) cleaves ESX secretion-associated protein B (EspB) that is a virulence factor of Mycobacterium tuberculosis, and accommodates an octapeptide, AVKAASLG, as a short peptide substrate. Because peptidoboronic acids are known inhibitors of serine proteases, the synthesis and binding of a boronic acid analog of the pentapeptide cleavage product, AVKAA, was studied using MycP₁ variants from M. thermoresistible (MycP_1mth), M. smegmatis (MycP_1msm) and M. tuberculosis (MycP_1mtu). We synthesized the boropentapeptide, HAlaValLysAlaAlaB(OH)₂ (1) and the analogous pinanediol PD-protected HAlaValLysAlaAlaBO₂(PD) (2) using an Fmoc/Boc peptide strategy. The pinanediol boropentapeptide 2 displayed IC₅₀ values 121.6±25.3 μM for MycP_1mth, 93.2±37.3 μM for MycP_1msm and 37.9±5.2 μM for MycP_1mtu. Such relatively strong binding creates a chance for crystalizing the complex with 2 and finding the structure of the unknown MycP₁ catalytic site that would potentially facilitate the development of new anti-tuberculosis drugs.

Mycobacterium tuberculosis exerts a staggering human and economic toll: in 2012, an estimated 8.6 million people developed tuberculosis and 1.3 million died from the disease¹. M. tuberculosis secretes several highly immunogenic proteins across the cell wall using the ESX-1 transport system², and these virulence factors cause lung tissue inflammation and necrosis³. In vivo inhibition of MycP₁ protease, in a mouse model of infection⁴, led to a lower mortality rate than in untreated animals. In addition, a M. tuberculosis strain with a mutation affecting the catalytic activity of MycP₁ was less virulent than a wild type strain⁴. Inhibition of MycP₁ protease, which is one of the components of the ESX-1 transport system, is an attractive target for drug development^5-11

It was recently shown that M. smegmatis MycP₁¹¹ and M. thermoresistibile MycP₁¹² process M. tuberculosis EspB at positions Ala³⁵⁸ and Ala³⁸⁶. We confirmed that the octapeptide, (H)AVKAASLG(OH), mimicked the natural substrate in a fluorescent resonance energy transfer (FRET) experiment using an internally quenched peptide, (Abz)AVKAASLG(DNP) with an N-terminal, ortho-aminobenzoic acid (Abz) fluorescent group and a C-terminal, 2,4-dinitrophenyl (DNP) quencher¹². Cleavage of this octapeptide substrate by MycP₁ liberated the readily measured, fluorescent pentapeptide (Abz)AVKAA(OH), and this FRET system also provided a convenient means for screening potential MycP₁ inhibitors¹³. Peptidyl boronic acids and their cyclic boronic esters with 1,2-diols are known inhibitors of various serine proteases^14-17 in the nanomolar range, including peptidoboronic acids that show selectivity towards M. tuberculosis.¹⁸ The mechanism of action of peptidoboronate inhibitors involves the formation of tetrahedral complexes with active-site serines.^19-21 Variability in the activity of structurally related peptidoboronic acids and peptidoboronates in the literature²² prompted us to determine if either the boronic acid analog 1 or the boronate 2 (Fig. 1) of the pentapeptide (H)AVKAA(OH) would inhibit the MycP1 protease and provide lead structures for the development of still other, clinically useful inhibitors.

Figure 1. Boronic acid analogs, HAlaValLysAlaAlaB(OH)₂ (1) and the pinanediol PD-protected HAlaValLysAlaAlaBO₂(PD) (2).

Solution-phase synthesis of HAlaValLysAlaAlaB(OH)₂ (1) and the pinanediol PD-protected HAlaValLysAlaAlaBO₂(PD) (2) involved the initial coupling of FmocLys(Cbz)OH (3) to HAlaOtBu to afford the dipeptide, FmocLys(Cbz)AlaOtBu, Fmoc-deprotection, and an additional coupling to (Cbz)AlaVal(OH) to provide the tetrapeptide intermediate, (Cbz)AlaValLys(Cbz)AlaOtBu (4) (Fig. 2). Acid-catalyzed removal of the tert-butoxy group furnished the Cbz-protected tetrapeptide, and HATU-promoted condensation²³ with ((R)-boroalanine-(1S,2S,3R,5S)-(+)-2,3-pinanediol ester provided the protected pentapeptide 5, purified by a combination of flash silica gel and preparative layer silica gel chromatography. Hydrogenolysis²⁴ afforded 2 and acid-catalyzed hydrolysis of 2 provided (H)AlaValLysAlaAlaB(OH)₂ (1), albeit in low yield.

Figure 2. Synthesis of HAlaValLysAlaAlaB(OH)₂ (1) and HAlaValLysAlaAlaBO₂(PD) (2).

Legend : a, HAlaOtBu, hydroxybenzotriazole (HOBt), iPr₂NEt, EDC-HCl, N,N-dimethylformamide (DMF), 0°C to 25°C, 24 h; b, piperidine, DMF, 25°C, 1 h; c, CbzAlaValOH, HOBt, iPr₂NEt, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride, DMF, 0° to 25°C, 24 h; d, CF₃CO₂H, CH₂Cl2, 25°C, 24 h; e, ((R)-boroalanine-(1S,2S,3R,5S)-(+)-2,3-pinanediol ester hydrochloride, i-Pr₂NEt, 1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate (HATU), DMF, 0°C, 0.5 h; f, H₂, 10% Pd/C, 25°C, 17 h; g, 2-methylpropylboronic acid, 3M HCl, 25°C, 17 h.

We have previously characterized the activity of MycP₁ variants from M. thermoresistible (MycP_1mth) and M. smegmatis (MycP_1msm) using a quenched fluorescent peptide assay¹³. In addition to these variants, we also expressed and purified MycP₁ from M. tuberculosis (MycP_1mtu). We characterized the activity MycP_1mtu and found significant differences in enzyme activity relative to other MycP₁ homologs. In particular, the specific activity of MycP_1mtu was 28.2±2.0 nmol/min/mg, which was four times higher than that of MycP_1mth homolog (Table 1). This difference in enzyme activity was not surprising because the peptide substrate, (Abz)AVKAASLGK(DNP)OH was based on the cognate M. tuberculosis substrate EspB_mtu residues 354-362 (AVKAASLG). This recognition region displayed sequence variations in M. thermoresistible EspB_mth (i.e., SVKPAAGG) and in M. smegmatis EspB_msm (i.e., SLKPASAG), and consequently, the affinity determinants in the MycP_1mth and MycP_1msm binding sites differed from those of MycP_1mtu. Nevertheless, all three species variants had measureable activity using the quenched fluorescent octapeptide as a substrate.

Table 1. Michaelis-Menten parameters for the three MycP₁ variants.

MycP1 variant	Km (μM)	Vmax (nmol/min/mg)
M. tuberculosis	79±11	28.2±2.0
M. thermoresistible	60±12	6±0.6
M. smegmatis	86±42	1.8±0.4

The potency of the synthetic boronic acid analogs 1 and 2 was tested using the three MycP₁ variants. Because compound 1 exhibited poorer inhibition than compound 2 in preliminary testing, we focused on the characterization of compound 2. As expected, all were inhibited to some extent, but the inhibitor showed relatively tighter binding to the cognate MycP_1mtu enzyme (Fig. 3) than their binding to the two others. The concentrations necessary to achieve 50% inhibition of MycP₁ (IC₅₀ values) were as follows: MycP_1mtu = 37.9±5.2 μM, MycP_1mth = 121.6±25.3 μM, and MycP_1msm = 93.2±33.7 μM. It has been reported that the inhibition of boronic acid peptides could vary over several orders of magnitude depending on the chemical structure of amino acid at critical positions of peptide¹⁸. While the inhibition of MycP₁ variants by the boronic acid analog 2 is relatively moderate, it could be improved by using a combinatorial chemistry approach to analyze a library of peptides with variable sequence. For example, MycP₁ displays a higher activity against a substrate peptide with Met in P1 position.¹² therefore, boronic acid analogs with aliphatic side groups in this position could be explored in the future.

Figure 3. Determination of IC₅₀ of HAlaValLysAlaAlaBO₂(PD) (2) for MycP₁ variants.

The IC₅₀ of 2 was measured for MycP₁ variants from M. tuberculosis (MycP_1mtu), M. thermoresistible (MycP_1mth), and M. smegmatis (MycP_1msm), using a quenched fluorescent peptide (Abz)AVKAASLGK(DNP)OH). Activity of MycP₁ is plotted as a function of the logarithm of the concentration of 2. Calculated IC₅₀ values were: MycP_1mtu = 37.9±5.2 μM, MycP_1mth = 121.6±25.3 μM, and MycP_1msm = 93.2±33.7 μM.