Mechanism of Caspase-1 Inhibition by Four Anti-inflammatory Drugs Used in COVID-19 Treatment

Abstract

The inflammatory protease caspase-1 is associated with the release of cytokines. An excessive number of cytokines (a "cytokine storm") is a dangerous consequence of COVID-19 infection and has been indicated as being among the causes of death by COVID-19. The anti-inflammatory drug colchicine (which is reported in the literature to be a caspase-1 inhibitor) and the corticosteroid drugs, dexamethasone and methylprednisolone, are among the most effective active compounds for COVID-19 treatment. The SERM raloxifene has also been used as a repurposed drug in COVID-19 therapy. In this study, inhibition of caspase-1 by these four compounds was analyzed using computational methods. Our aim was to see if the inhibition of caspase-1, an important biomolecule in the inflammatory response that triggers cytokine release, could shed light on how these drugs help to alleviate excessive cytokine production. We also measured the antioxidant activities of dexamethasone and colchicine when scavenging the superoxide radical using cyclic voltammetry methods. The experimental findings are associated with caspase-1 active site affinity towards these compounds. In evaluating our computational and experimental results, we here formulate a mechanism for caspase-1 inhibition by these drugs, which involves the active site amino acid Cys285 residue and is mediated by a transfer of protons, involving His237 and Ser339. It is proposed that the molecular moiety targeted by all of these drugs is a carbonyl group which establishes a S(Cys285)-C(carbonyl) covalent bond.

Keywords: COVID-19; anti-inflammatory drugs; caspase-1; cytokine storm.

Figure 1

Mechanisms of inflammasome activation. The activation in general relies on two signals. First, pathogen-associated molecular patterns (PAMPs) bind to pattern recognition receptors (PRRs), resulting in the synthesis of NLRP3 and pro-IL-1β. Then, a second signal causes inflammasome assembly, leading to the activation of caspase-1 and the processing of pro-IL-1β into IL-1β and pyroptosis. The second signal may come from a variety of pathways, including K⁺ efflux, lysosomal rupture or mitochondrial dysfunction. This results in the release of ROS, Ca²⁺ and mitochondrial DNA (mtDNA), all of which have been shown to activate the inflammasome. Pyroptosis occurs as a result of caspase-1-mediated cleavage of GSDM-D at the linker region of GSDM-D. Following GSDM-D cleavage, the amino terminus of GSDM-D (GSDM-D-N) forms a non-selective pore at the cell membrane through which IL-1β is then released. (From [6], with permission.).

Figure 2

The compounds studied.

Figure 3

Dexamethasone pose 13 from docking into PDB 6PCP receptor. It has a H-bond between F(dexamethasone) and C(copper)-colored H₂N(Arg-B-341), 2.745 Å, which is absent in pose 2 and 14. Purple C-colored His237 has a H-bond to O(carbonyl) of dexamethasone, 2.165 Å. H(Cys285) has a H-bond to polypeptide O(Ser-B-339), 2.818 Å, suggesting potential cleavage of S–H. There is an interaction between S(Cys285) and C(carbonyl) of dexamethasone, 3.202 Å, that is almost perpendicular to the C=O plane.

Figure 4

Two-dimensional display of the amino acid interactions of docked dexamethasone pose 13 seen in Figure 3.

Figure 5

LUMO of dexamethasone suggests the preferential molecular quinone methide moiety (left) prone to reactivity, and a potential addition of a proton can be expected.

Figure 6

LUMO of protonated dexamethasone also suggests that the quinone methide ring may be prone to reactivity with S(Cys145) thiolate. The [C=OH]⁺ moiety is located on the left.

Figure 7

This is pose 8 from docking methylprednisolone into PDB 6PCP receptor. There is an interaction between S(Cys285) and C(carbonyl) of methylprednisolone, 3.135 Å. His237 shows a H-bond to O(carbonyl) of methylprednisolone, 1.926 Å. H(Cys285) has a H-bond to polypeptide O(Ser-B-339), 2.818 Å.

Figure 8

Colchicine pose 9 upon docking into PDB 6PCP receptor. It has H-bonds between one colchicine methoxy moiety and two NH₂ groups of purple C-colored Arg-B-341, 2.232 Å and 2.490 Å (center-right). Turquoise C-colored cationic His237 has a H-bond to O(carbonyl) of flat tropolone ring of colchicine, 1.956 Å (bottom). H(Cys285) has a H-bond to O(Ser-B-339), 2.818 Å (center left). S(Cys285) has an interaction of 3.363 Å to C(carbonyl) of flat tropolone ring of colchicine.

Figure 9

Two-dimensional display of amino acid interactions of docked colchicine pose 9.

Figure 10

Colchicine LUMO suggests that the aromatic seven-membered ring is more prone for reactivity than the six-membered one.

Figure 11

This is the LUMO of protonated colchicine by the tropolone O(carbonyl), showing O-H distance of 0.982 Å. The tropolone ring is more prone to further reactivity than other atoms in this cationic species, suggesting potential interaction with S(Cys145) thiolate. The acetamide carbonyl (bottom) does not appear to be involved in potential reactivity when compared with interaction, though the former tropolone carbonyl is now transformed in the [C=OH]⁺ moiety (right).

Figure 12

This is pose 2 from docking raloxifene into the PDB 6PCP receptor that shows an interaction between S(Cys285) and C(carbonyl) of raloxifene, 3.833 Å. Cationic His237 has a H-bond to O(carbonyl) of raloxifene, 2.484 Å. Additional interaction between His237 and a raloxifene Ph is seen due to π–π stacking (bottom-right). H(Cys285) has a H-bond to O(Ser-B-339), 2.818 Å. A π–π T-shaped interaction is seen between Trp-B-340 and the nine-membered raloxifene ring (omitted for clarity). Arg-B-341 has a H-bond to the HN by the saturated raloxifene ring.

Scheme 1

Mechanism of caspase-1 inhibitors: R₁ and R₂ are substituents in the five drugs. (1) Cationic [HN-His237] donates a proton to O(carbonyl) of inhibitor. (2) Polypeptide O(Ser339) captures a proton from Cys285, cleaving its S–H bond, and generating S(Cys285) thiolate. (3) The activated carbonyl (step 1) is a nucleophile target for S(thiolate), which generates a covalent S(Cys285)-C(carbonyl) bond.

Figure 13

DFT quinone methide ring distances of dexamethasone; F atom ball style.

Figure 14

Dexamethasone, F atom ball style, superoxide radical stick style. From an initial separation between ring and superoxide centroids (3.5 Å), geometry optimization converges to the superoxide moving away, and minor modifications in the quinone methide ring are observed. We conclude that there was no transfer of a superoxide electron into the ring.

Figure 15

Protonated dexamethasone, bond distances. This species may be suspected to form during the RRDE experiment; the proton, bound to O(carbonyl), may arise from DMSO hygroscopicity.

Figure 16

DFT geometry optimization of protonated dexamethasone—O₂ radical initially π–π separated, 3.50 Å. Ring distances are affected by superoxide action and so the superoxide electron seems to be captured by the quinone methide ring. In addition, d(O-O) 1.264 Å is significantly shorter than the original superoxide bond length of 1.373 Å, thus suggesting superoxide transformation to O₂, which then gets released from the dexamethasone, 5.463 Å. The C=C bond ring distances are lengthened, whereas C–C bonds become shortened, when compared to the previous picture. A similar behavior has been previously observed by us for other superoxide scavengers [58].

Figure 17

Cyclovoltammetry RRDE dexamethasone. The ratio between the current at the ring electrode and that at the disk electrode (vertical axis) vs. concentration of scavenger. The initial decrease of the ratio is associated with the existence of humidity in DMSO. Once this is consumed, the stable ratio indicates no scavenging upon increasing concentration of dexamethasone, and so dexamethasone is not able to scavenge superoxide.

Figure 18

Relevant ring distances for DFT colchicine minimum of energy.

Figure 19

Colchicine—superoxide radical π–π distances. The superoxide radical is stick style. DFT geometry optimization after an initial 3.5 Å separation between the tropolone ring and superoxide centroids. The tropolone ring shows lengthening of C=C bonds and shortening of C–C bonds after geometry optimization. This sort of aromatization indicates the superoxide electron captured by the tropolone ring and the molecular O₂ leaving, 6.895 Å [58].

Figure 20

Protonated colchicine relevant bond distances.

Figure 21

Protonated colchicine—superoxide radical π–π distances show lengthening of C=C bonds and shortening of C–C bonds, suggesting the capture of an electron from the superoxide. Interestingly, the separation between rings is shorter (2.984 Å) than it was initially (3.5 Å). In the literature, short bond distances for complexes containing π–π interacting rings in related compounds have been reported [59].

Figure 22

Voltammograms of colchicine. The lower part is associated with the disk electrode (formation of the superoxide radical after reduction of O₂). The upper part features the ring electrode (oxidation of superoxide).

Figure 23

Colchicine efficiency. A linear trend (y = −33862x + 14.306) has R² = 0.9303.