Rational In Silico Design of Selective TMPRSS6 Peptidomimetic Inhibitors via Exploitation of the S2 Subpocket

Abstract

TMPRSS6 is a potential therapeutic target for the treatment of iron overload due to its role in regulating levels of hepcidin. Although potent TMPRSS6 inhibitors have been previously developed, their lack of specificity requires optimization to avoid potential side effects before pursuing preclinical development with in vivo models. Here, using computer-aided drug design based on a TMPRSS6 homology model, we reveal that the S2 position of TMPRSS6 offers a potential avenue to achieve selectivity against other members of the TTSP family. Accordingly, we synthesized novel peptidomimetic molecules containing lipophilic amino acids at the P2 position to exploit this unexplored pocket. This enabled us to identify TMPRSS6-selective small molecules with low nanomolar affinity. Finally, pharmacokinetic parameters were determined, and a compound was found to be potent in cellulo toward its primary target while retaining TTSP-subtype selectivity and showing no signs of alteration in in vitro TEER experiments.

Figure 1

Comparison between matriptase and TMPRSS6 in complex with Cpd-8 ((H)F-ThiazolylAla-V-R-Kbt). (A) Surface representation of matriptase (PDB: 6N4T) in complex with Cpd-8 with focus on the lipophilic S2 pocket lined with Phe708, His666, and Gln801. The nature of each surface is color-coded ranging from hydrophilic (purple) and neutral (white) to lipophilic (dark green). (B) Surface representation of the TMPRSS6 homology model with the lowest energy conformation of Cpd-8 obtained by conformational search as listed in materials and methods. The nature of each surface is color-coded ranging from hydrophilic (orange), neutral (white) to lipophilic (dark green). The analogue Cpd-8 is represented in green stick mode with representative heteroatom colors. Hydrogen bonds (light blue) and proton-π interactions (cyan) are indicated as dashed lines and a cyan bar. Created with BioRender.com.

Figure 2

Overview of the design and examples of modifications of the P2 position based on Cpd-8.

Scheme 1. Synthesis Pipeline for the Kbt-Containing Peptidomimetics

R represents the side chains of the amino acids used in the P2 position.

Figure 3

In vitro inhibition of TMPRSS6 (black) and matriptase (gray) of each molecule synthesized at a concentration of 100 nM. Activity is shown relative to vehicle-treated proteases (%). MD23 = Cpd-8.

Figure 4

Comparison between matriptase and TMPRSS6 in complex with MD5. (A) Surface representation of matriptase in complex with the lowest energy conformation of MD5 obtained by conformational search. The nature of each surface is color-coded ranging from hydrophilic (purple) and neutral (white) to lipophilic (dark green). (B) Surface representation of TMPRSS6 homology model with the lowest energy conformation of MD5 obtained by conformational search. The nature of each surface is color-coded, ranging from hydrophilic (orange), neutral (white) to lipophilic (dark green). The analogue MD5 is represented in green stick mode with representative heteroatom colors. Hydrogen bonds (light blue) and proton–π interactions (cyan) are indicated as dashed lines and a cyan bar. Created with BioRender.com.

Figure 5

Conformational search results for MD5 in matriptase revealing contact clashes between (H)Phe in position P4 and Gln801 in S4. The analogue MD5 is represented in green stick mode with representative heteroatom colors. The electronegative mesh model represents the occupied matriptase volume by the ligand. The clash contacts are indicated as orange disks close to Gln810. Created with BioRender.com.

Figure 6

In cellulo IC₅₀ dose–response curves for selected compounds. IC₅₀ values were calculated by assessing proteolytic activity in the cell media of transfected HEK293 cells (matriptase in red, TMPRSS6 in black) treated with the indicated inhibitors (A left panel, B middle panel, C right panel). Relative activity is expressed as compared to vehicle-treated cells (100%). Each panel displays a representative IC₅₀ curve and associated mean ± SD data are reported in Table 4.

Figure 7

Normalized AUC of the TEER experiments on Caco-2 cells for 8 days. Results are expressed as a mean ± SEM of the AUC over 8 days of treatment using 10 μM of A, B, C, vehicle, or untreated of three independent experiments, conducted in duplicate. Comparisons between the vehicle and other conditions were performed using one-way analysis of variance (ANOVA) for repeated measures followed by multiple comparisons using the Kruskal–Wallis test in GraphPad Prism 10.1.2. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 compared to the vehicle.