The MICALs are a Family of F-actin Dismantling Oxidoreductases Conserved from Drosophila to Humans

Cellular form and function – and thus normal development and physiology – are specified via proteins that control the organization and dynamic properties of the actin cytoskeleton. Using the Drosophila model, we have recently identified an unusual actin regulatory enzyme, Mical, which is directly activated by F-actin to selectively post-translationally oxidize and destabilize filaments – regulating numerous cellular behaviors. Mical proteins are also present in mammals, but their actin regulatory properties, including comparisons among different family members, remain poorly defined. We now find that each human MICAL family member, MICAL-1, MICAL-2, and MICAL-3, directly induces F-actin dismantling and controls F-actin-mediated cellular remodeling. Specifically, each human MICAL selectively associates with F-actin, which directly induces MICALs catalytic activity. We also find that each human MICAL uses an NADPH-dependent Redox activity to post-translationally oxidize actin’s methionine (M) M44/M47 residues, directly dismantling filaments and limiting new polymerization. Genetic experiments also demonstrate that each human MICAL drives F-actin disassembly in vivo, reshaping cells and their membranous extensions. Our results go on to reveal that MsrB/SelR reductase enzymes counteract each MICAL’s effect on F-actin in vitro and in vivo. Collectively, our results therefore define the MICALs as an important phylogenetically-conserved family of catalytically-acting F-actin disassembly factors.


Supplementary Figure 3. Purification and characterization of recombinant human MICAL-3 redoxCH protein.
(a-f) Coomassie stained gels are shown and the arrows point to the recombinant hMICAL-3 redoxCH protein in all gels. MW in kDa. (a) A cDNA encoding hMICAL-3 redoxCH was inserted into a His-tag containing bacterial expression vector, transformed into bacteria, and following the appropriate growth conditions, lysates were loaded on a Ni-NTA affinity column to enrich for the Nus/His-tagged hMICAL-3 redoxCH protein (arrowhead). (b) After desalting, samples were loaded on a MonoQ column to enrich for the Nus/His-tagged hMICAL-3 redoxCH (arrowhead). (c) Samples in collection tubes 3-9 (from b) were desalted and then digested with thrombin. A Coomassie-stained band corresponding in size to the uncleaved Nus-tagged hMICAL-3 redoxCH protein is readily seen in the absence of thrombin (arrowhead), while cleaved hMICAL-3 redoxCH protein is observed following thrombin digestion (arrow). (d-e) Ni 2+ -NTA affinity chromatography (d) was used again to remove the Nus-tag and thrombin, after which the contents of collection tubes 3-15 were combined and subjected to cation ion exchange chromatography (e) to further separate hMICAL-3 redoxCH protein (arrow) from contaminating proteins. (f) The contents of sample tubes 4 through 15 (from e) were combined, concentrated, and analyzed on a gel to determine the purity of the hMICAL-3 redoxCH protein (arrow). The purified hMICAL-3 redoxCH is also shown in a transparent tube, where its yellowish color is readily observed. Unprocessed original scans of gels/blots are shown in Supplementary Fig. 11.

Supplementary Figure 4. The redox region and flavin binding characteristics of MICAL family proteins.
Essential experimental procedures for defining a flavin-containing enzyme are 1) the recognition of the presence of a flavin co-factor, 2) the identification of the flavin co-factor, and 3) the determination of the stoichiometry of the bound flavin co-factor 1,2 . Therefore, we set out to characterize each member of the MICAL family in this regard (see also Figure 1c Heat-induced denaturation of each the MICAL proteins using standard approaches 1 was used to determine the each purified MICAL family protein bound FAD non-covalently. In particular, following heat-induced denaturation of each of the MICAL proteins and centrifugation, the supernatant (Supe) and not the pellet was yellow, indicating that the flavin is bound non-covalently (i.e., the flavin was released when the protein was unfolded by heat treatment/denaturation, showing that it was not covalently linked to the MICALs). The absorbance spectrum of the released flavin exhibited a peak at 450 nm, which is characteristic of FAD. FMN exhibits a peak at 446 nm. These results indicate that each of the MICALs, including hMICAL-1 DG (see Supplementary Figure 8), binds FAD non-covalently.
[MICALs]=20µM. (c) Our results also revealed a 1:1 stoichiometry of the bound FAD cofactor to each of the MICAL family members. Characterization of the absorption spectra of MICAL proteins (see Figure 1c-f) also allows the amount of the purified MICAL that is bound to FAD to be determined (see Materials and Methods and 3 ). Note that the MICAL-1 DG protein generated more protein in the Apo form (without FAD) than the other MICALs, indicating that it is functional but less able to bind and incorporate FAD when this amino acid change is introduced into it.

Supplementary Figure 5. Purified human MICAL redoxCH proteins show little to no association with microtubules and do not alter tubulin polymerization dynamics. (a)
Images of Coomassie blue stained gels are shown. Co-sedimentation analysis was used to examine the association between MICALs and microtubules. Notice that after high-speed centrifugation, each of the purified human MICAL redoxCH proteins, hMICAL-1 (hM-1), hMICAL-2 (hM-2), hMICAL-3 (hM-3), and human MICAL-1 DG (hM-1 DG ), like Drosophila Mical redoxCH (dM) 4 , is present in the soluble (S) fraction (left gel). Similarly, in the presence of microtubules, the majority of the purified MICAL proteins (and a negative control, BSA) are present in the soluble (S) fraction (right gel). Notice, however, that microtubule associated proteins (MAPs), known microtubule binding proteins that were used as a positive control, change their distribution from the soluble fraction to pellet fraction in the presence of microtubules. These results indicate that purifed human MICAL redoxCH proteins have little to no association with microtubules. The percentage (+ the standard error of the mean (SEM)) of different purified MICAL proteins in the pelleted fraction following incubation with microtubules (MTs) was quantified by densitometry (n > 2). The effect of MICALs on tubulin polymerization was examined. A fluorescence-based tubulin polymerization assay was employed using standard approaches, where the fluorescence intensity (a.u. (arbitrary units)) of microtubules is substantially higher than tubulin monomers. Drosophila Mical redoxCH (dM), each of the purified human MICAL redoxCH proteins, hMICAL-1 (hM-1), hMICAL-2 (hM-2), hMICAL-3 (hM-3), and human MICAL-1 DG (hM-1 DG ), and/or NADPH were added in the tubulin solution and polymerization was initiated by increasing the temperature from 4°C to 37°C. There is no appreciable difference between tubulin polymerization alone (+NADPH) and tubulin polymerization in the presence of the human MICALs with NADPH.
[MICALs]=600nM, [NADPH]=100µΜ, [tubulin]=2mg/ml. Unprocessed original scans of gels are shown in Supplementary Fig. 11. Pyrene-actin assays reveal that at low concentrations, hMICAL-1's ability to alter actin polymerization is dosage dependent. However, with increasing higher concentrations, MICAL-1 exhibits decreasing effects on Factin (see Figure 4d). These effects are consistent with hMICAL-1 exhibiting such rapid consumption of NADPH in the absence of its F-actin substrate that NADPH becomes limiting (as higher levels of hMICAL-1 are added to the assay) in allowing MICAL-1 to alter F-actin dynamics (see also main text).
[Actin] = 1.15 µM, [NADPH]=100 µM. (e-f) Further analysis of hMICAL-1's effects on actin dynamics. MICAL-1 does not disassemble F-actin when it is incubated with NADPH prior to the addition of F-actin (see Figure 4f-g). These results support that MICAL-1, because of its high-rate of basal activity, consumes/uses-up NADPH prior to the addition of F-actin, such that it can no longer use NADPH in its reaction to modify actin. To further test this hypothesis, we added more NADPH (additional NADPH in e-f) into the MICAL-1 pre-reaction condition tube (f, arrowhead) (from Supplementary Figure 6c Supplementary Fig. 11.

Supplementary Figure 8. Purification and characterization of human MICAL-1 redoxCH DG protein (hMICAL-1 DG ).
Compare also with the strategy used in Supplementary Figure 1 to purify human MICAL-1 redoxCH . Coomassie stained gels are shown and the arrows point to the recombinant hMICAL-1 redoxCH DG protein in all gels. (a) A cDNA encoding hMICAL-1 redoxCH DG was inserted into a His-tag containing bacterial expression vector, transformed into bacteria, and following the appropriate growth conditions, lysates were loaded on a Ni-NTA affinity column to enrich for the Nus/His-tagged hMICAL-1 redoxCH DG (arrowhead). (b) The Nus-tagged hMICAL-1 redoxCH DG (arrowhead) was digested (+) with a thrombin protease to cleave-off the Nus tag. The smaller size of the digested human MICAL-1 redoxCH DG (without the Nus-His tag) can be seen (arrow). (c) The digested (+) sample from (b) was then loaded again on a Ni-NTA agarose column to remove the Nus-tag. Fractions from 2-6 were combined and used for d. (d) Ion-exchange chromatography was then used to remove contaminating proteins since the hMICAL-1 redoxCH DG can bind with the MonoQ column (arrow). Samples within collection tubes 2-6 were then combined and concentrated (e) and analyzed on a gel to determine the purity of the hMICAL-1 redoxCH DG protein. The purified hMICAL-1 redoxCH DG mutant is also shown in a transparent tube, where its yellowish color is readily observed. Note that the yellow color is lighter than the other MICALs (Supplementary Figures 1-3), because of more protein being made without FAD bound to it (Supplementary Figure 4c). residues. Yet, previous work from others have revealed that when actin is present in filaments, the Met44 and Met47 residues of actin are poorly accessible to diffusible solvents, including oxidants such as hydrogen peroxide [6][7][8][9] . Therefore, these results suggest the hypothesis that MICALs do not modify F-actin through the general release of hydrogen peroxide or another diffusible oxidant. Likewise, direct experiments also support this hypothesis by revealing that hydrogen peroxide has no effects on F-actin disassembly (even when added at high millimolar concentrations), that hydrogen peroxide does not disassemble F-actin in combination with MICALs binding to F-actin, and that hydrogen peroxide scavengers do not alter MICAL-mediated actin disassembly ( 4,5,10-13 ; present study). Further, the general release of hydrogen peroxide or another diffusible oxidant would not be expected to modify an amino acid stereospecifically as the MICALs do (which selectively modify actin's Met44 and Met47 residues in a single stereo-specific conformation) -and this modification as well as the MICALs effects on F-actin are selectively reversed by the methionine sulfoxide reductase SelR/MsrB ( 10,14 ; present study). Moreover, when MICALs are separated from F-actin using a barrier/compartmentalized chamber system, MICALs do not exert effects on F-actin 5 . Thus, all of these results indicate that the active site of the MICALs needs to gain access to the poorly accessible Met44 and Met47 residues that are buried within F-actin. A model is now emerging based on previous work and the experiments conducted herein that the MICALs are under tight regulation: both precisely localized and maintained in an inactive conformation in the cell (i.e., without O 2 consumption/NADPH activity/effects on Factin/ hydrogen peroxide production, etc.) so as to not dismantle all F-actin structures 4,11,[15][16][17]  , 3b Upper (f, red box), 3b Lower (g, red box), 3c WT actin (h, red boxes), 3c M2L actin (i, red boxes), 4b (j, red box), S1e (k, red box), S2c (l), S2d (m, red box), S2e (n, red box), S3a (o), S3b (p), S3c (q), S3d (r), S3e (s), S3f (t, red box), S5a Left (u), S5a Right (v), and S7b (w, red box). All other gels in Figures/Supplementary Figures show the full uncropped gels.