The cathepsins comprise a family of lysosomal protease enzymes whose primary function (i.e., protein degradation) plays a critical role in normal cellular homeostasis.1
Over expression of cathepsin L and/or abnormal activity has been implicated in a number of disease states.2
For example, cathepsin L is responsible for bone resorption through degradation of collagen type I; this disregulation is believed to lead to osteo- and rheumatoid arthritis.3
In addition, several infective organisms, such as SARS and Ebola viruses, utilize cathepsin L-like proteins for replication in human cells.4
The large number of disease states associated with cathepsin L calls for an understanding of the biological function.2
Recently, the Penn Center for Molecular Discovery (PCMD),5
completed a high throughput screening (HTS) campaign of the NIH Molecular Libraries Small Molecule Repository (MLSMR) to identify inhibitors of members of the papain-like cysteine protease family, including cathepsins B, L, and S.6
In this Letter, we detail our continuing efforts to create a comprehensive, publicly available profile of small-molecule inhibitors of the cysteine protease class, and herein describe the identification of a novel class of potent Cathepsin L inhibitors.
Previously reported inhibitors of cathepsin L include the peptides, leupeptin and aprotinin, and the fluoromethyl ketone, Z-LLL-FMK.3,7
The few known, potent small molecule inhibitors are either peptidic and therefore likely to suffer from physiological instability and poor permeability, or are non-selective for cathepsin L.3,8,9
The identification of potent, selective, stable, and cell permeable small molecule inhibitors would therefore be a valuable tool to interrogate cathepsin L and cathepsin L-like function, as well as to provide potential starting points for drug discovery and development.10–15
Initial HTS results of our Cathepsin L screen, indicated that several structurally related oxadiazoles exhibited potent inhibitory activity ().16–18
Cathepsin L inhibitory activity of oxadiazole-containing library samplesa
The most potent hit, cataloged in the MLSMR as disubstituted oxadiazole 1
, exhibited an IC50
of 0.13 μM.17
Wells containing related analogs (2
), differing only in the amide nitrogen substitution, displayed a range of activities (IC50
= 0.16–0.51 μM), suggesting a nascent structure activity relationship. As is our practice, library samples containing the putative active oxadiazoles were evaluated for both purity and integrity by LC-MS analysis. This analysis indicated that the primary constituent of each sample was indeed the expected oxadiazole, in up to 60% purity, however numerous impurities were also present. To confirm the biological activity attributed to 1
, a synthetic sequence was developed to generate the oxadiazoles in pure form.
Although little literature precedence exists for the construction of compounds such as 1
, we realized that modification of the Woodward-Confalone19
approach to α-amino acid substituted oxadiazolethiones could provide an entry to this class of sulfur-based 2,5-disubstituted oxadiazoles. Toward this end, commercially available L-Boc-Trp-OH was converted in high yield to the corresponding hydrazide (+)-6
, on a 10 gram scale ().20
Synthesis of 2,5-disubstituted oxadiazoles.
Cyclization with carbon disulfide in ethanol at reflux afforded the expected substituted thione (+)-7
as a stable solid in excellent yield, which upon chemoselective alkylation with 2-bromo-N
) efficiently generated (+)-8
(94% yield). Further study indicated that isolation of (+)-7
was unnecessary, and (+)-8
could be prepared directly from (+)-6
, in a single flask. Both sequences furnished (+)-8
in high yield. In similar fashion, the antipode, (−)-8, was prepared starting from D-Boc-Trp-OH. High enantiomeric purities (>99%) were demonstrated via chiral supercritical fluid chromatography.
Initial attempts to remove the Boc-group utilizing HCl in dioxane (4 N) or HCl in water (6 N) were compromised by the poor solubility of (+)-8. However, small quantities of the HCl salt of 1 could be obtained by treatment with 6 N HCl over 45 minutes. Upon LC-MS analysis of synthetic 1, we recognized that the impurities formed during the acid promoted Boc-deprotection step were identical to those found in the library screening samples (vide infra). Optimal conditions to remove the Boc-group were eventually developed. Specifically, treatment of (+)-8 with 25% water in TFA for 15 minutes, followed in turn by adjustment of the pH to 8.0 with NaHCO3 and aqueous extraction with methylene chloride furnished the free-base of 1 in 89% yield and 99% after column chromatography. In the end, oxadiazole 1 was prepared on gram scale from L-Boc-Trp-OH; the overall yield was 76%.
To our surprise, the free base of 1
was found to be completely devoid of activity when assayed against cathepsin L.18
This result suggested that an impurity present in the original library sample was responsible for the observed activity. Based on LC-MS analyses of the biologically active samples, we hypothesized that the active component was likely the ring-opened product 10
, formed via acid-promoted addition of H2
O to (+)-8
Conversion of oxadiazole (+)-8 to thiocarbazate 10.
The presence of a molecular [M+1] ion equal to 440 amu in the LC-MS analysis of impure samples was consistent with the presence of 10, thereby providing strong support for the hypothesis.
Thiocarbazates such as 10
have not been described previously in the literature. However, these compounds do bear structural resemblance to aza-peptides22–24
examples of which have been reported to exhibit cysteine protease inhibitory activity through a mechanism involving attack by the active site cysteine on the carbamate carbonyl.26
Known aza-peptide cathepsin inhibitor A.
To test the hypothesis that 10
was indeed the active species, we devised an expedient synthesis of this structural class. We envisioned that introduction of the C2 carbonyl unit of 10
could be achieved via chemistry in parallel to that used to prepare 1
, beginning with hydrazide (+)-6
. We began the synthesis by converting tryptophan hydrazide (+)-6
to an intermediate thiosemicarbazide (not shown) employing carbonyl sulfide (S=C=O) gas dissolved in ethanol ().27
Synthetic procedure to prepare thiocarbazates.
The intermediate thiosemicarbazide did not precipitate from solution, therefore 2-bromo-N
) was added to the reaction flask to generate thiocarbazate (−)-11
Pleasingly, the yield for the two-step, one flask operation was 62%. Deprotection of (−)-11
, employing our previously developed conditions (25% water in TFA), generated the free base (−)-12
in 98% yield with >95% purity. This 3-step sequence permitted construction of (−)-12 on gram scale, starting from L-Boc-Trp-OH; the overall yield was 58%.
As anticipated, both (−)-11
) and (−)-12
) exhibited potent inhibitory activity against cathepsin L with IC50
values of 56 nM and 133 nM, respectively.29
was only modestly active against cathepsin L (IC50
= 34 μM).29
Some variability in IC50
values determined for (−)-12
prompted us to explore the stability of both (−)-11
in solvents relevant to the bioassay. While (−)-11
was found to be completely stable in DMSO, as well as in the buffer employed in the Cathepsin L assay [e.g., NaOAc (20 mM), pH 5.5; EDTA (1 mM); cysteine (5 mM)], the free-base (−)-12
proved unstable in DMSO, generating decomposition products 13
, and 15
after only 1 hour at room temperature ().
Decomposition products of (−)-12 in DMSO.
Presumably 13–15 are formed from (−)-12 via intramolecular attack of the primary amine on the C2 thiocarbazate moiety, and release of 13. These products, prepared either via synthesis (13 and 14) or by HPLC purification of decomposed material (15), were assayed for cathepsin L activity and found to be inactive. Due to the instability of (−)-12 under the assay conditions, IC50 values vary somewhat, rendering interpretation of the bioassay data difficult. We are however, confident in the accuracy of the bioassay results for (−)-11, as this compound is stable under all conditions evaluated. Thus, future efforts will be focused on the more stable thiocarbazate (−)-11 and related congeners.
As most cysteine protease inhibitors contain an electrophilic “warhead,” and work through a mechanism involving reaction with the active site cysteine,7,26
we strongly believe these novel thiocarbazates behave similarly, and are active by virtue of their electrophilic carbonyl moiety. The formation of 15
supports this mechanism, and is indicative of the electrophilicity of the C2 carbonyl moiety. Reactivity at the anilide carbonyl is also a possibility, and cannot be ruled out. However, we observe no evidence of side-products resulting from reaction at this position (cf. ). A complete biochemical characterization of this novel chemotype is underway.31
In summary, samples from the NIH MLSMR revealed promising cathepsin L inhibitory activity attributed to a series of 2,5-disubstituted oxadiazoles (1–5). Analytical analyses (LC-MS) of the library samples, however, indicated numerous impurities, thus requiring the development of an efficient synthesis for the putative active 2,5-disubstituted oxadiazoles. Synthetic samples of pure 2,5-disubstituted oxadiazoles were found to be completely devoid of cathepsin L inhibitory activity. Careful LC-MS investigation of both the library and synthetic samples revealed a thiocarbazate to be the active component. Bioassay of the synthetic thiocarbazates confirmed the hypothesis: (−)-11 and (−)-12 display potent inhibitory activity against cathepsin L, with IC50 values of 56 nM and 133 nM, respectively, however instability of (−)-12 was also noted. The design, syntheses, and biological evaluation of analogs of the highly potent lead thiocarbazate class are now ongoing in our laboratories, and will be reported in due course.