Skin stem cells orchestrate directional migration by regulating microtubule-ACF7 connections through GSK3β

Abstract

Homeostasis and wound healing rely on stem cells (SCs) whose activity and directed migration are often governed by Wnt signaling. In dissecting how this pathway integrates with the necessary downstream cytoskeletal dynamics, we discovered that GSK3β, a kinase inhibited by Wnt signaling, directly phosphorylates ACF7, a > 500 kDa microtubule-actin crosslinking protein abundant in hair follicle stem cells (HF-SCs). We map ACF7's GSK3β sites to the microtubule-binding domain and show that phosphorylation uncouples ACF7 from microtubules. Phosphorylation-refractile ACF7 rescues overall microtubule architecture, but phosphorylation-constitutive mutants do not. Neither mutant rescues polarized movement, revealing that phospho-regulation must be dynamic. This circuitry is physiologically relevant and depends upon polarized GSK3β inhibition at the migrating front of SCs/progeny streaming from HFs during wound repair. Moreover, only ACF7 and not GSKβ-refractile-ACF7 restore polarized microtubule-growth and SC-migration to ACF7 null skin. Our findings provide insights into how this conserved spectraplakin integrates signaling, cytoskeletal dynamics, and polarized locomotion of somatic SCs.

Figure 1. Structural evidence that electrostatic interactions and phosphorylation may regulate associations between ACF7 and MT

(A) Ultrastructural analyses of negatively stained MTs either decorated with ACF7(CT) or naked. Bars=20 nm. Data were normalized and are shown superimposed in the upper right as one-dimensional projection profiles along the MT axis. A Fourier Transform of an ACF7-decorated MT is shown with the 40 Å and 80 Å layer lines denoted by white arrows. (B) Co-sedimentation assay using ACF7(CT) and 20→2.5 μg of MTs ± subtilisin. Note small downshift of tubulin bands (asterisks) and also decreased binding to ACF7(CT) that occurs following enzymatic treatment of polymerized MTs. MTΔC-tail: MT with C-tail removed by subtilisin. (C) Dissociation constants (Kd) of ACF7’s interactions with MT ± subtilisin. Binding assays were performed as in (B), but with a range of ACF7(CT) concentrations. Scatchard plot represents average of 3 data sets. (D) GST-tagged ACF7(CT) was purified from [³²P]orthophosphate-labeled keratinocyte lysates ± CIP (Calf Intestine Alkaline Phosphatase) and subjected to autoradiography and immunoblotting (IB). (E) Tandem mass spectrometry peptide mass map of AspN-generated peptides from purified ACF7(CT) protein. Stars (*) denote multiple signals corresponding to a serine-rich phosphopeptide, whose deduced sequence is shown. (F) Diagram of ACF7 with its many domains: CH, calponin homology F-actin binding; EF: EF hand motif mediating potential Ca²⁺ binding; GAR-GSR, Gas2-related and GSR-repeats for MT binding. GSR-repeats contain two potential GSK3β phosphorylation clusters, shown here, aligned with consensus GSK3β site (phospho-Ser in red; requisite Arg in blue).

Figure 2. GSK3β associates with and phosphorylates ACF7

(A) Bulge SC lysates were analyzed by SDS-PAGE and immunoblot (IB) before and after immunoprecipitation (IP) with ACF7 and control IgG Abs. Blots were probed with GSK3β or ACF7 Abs as indicated. (B) In vitro GSK3β kinase (IVK) assays were performed on C- or N-terminal ACF7. Phosphorylation was analyzed by SDS-PAGE and autoradiography. (C) Lysates are from cultured cells expressing HA-tagged ACF7(CT) ± constitutively-active (ca) GFP-tagged GSK3β (or Ctrl vector) ± CIP phosphatase. Lysates were analyzed ± anti-HA IP by SDS-PAGE and anti-phospho-serine/threonine (Pi-Ser/Thr) immunoblotting. (D) Cells were transfected with plasmids encoding WT or GSK3-site mutants (see Fig. 1F) of HA-tagged ACF7(CT) (or HA-tagged GST control) and caGSK3β or control vector. After labeling with [³²P]-orthophosphate, proteins were subjected to anti-HA IP, SDS-PAGE, staining and autoradiography. HC, IgG heavy chain. (E) Recombinant kinase was used to test GSK3β phosphorylation of ACF7(CT) and its different mutants in vitro. Phosphorylation of ACF7(CT) was detected by IB with Pi-Ser/Thr Ab.

Figure 3. GSK3β phosphorylation of ACF7’s GSR domain inhibits microtubule binding

(A) In vitro binding assays on purified ACF7 proteins. After incubating with 0–20 μg taxol-stabilized MTs, purified WT or mutant versions of full length, HA-tagged ACF7 were cosedimented and subjected to immunoblot (IB) analyses. Note that phosphomimetic S:D but not phosphorylation-refractile S:A ACF7 shows significantly reduced binding affinity. (B) In vivo MT-binding assays. Lysates were prepared from cells expressing HA-tagged ACF7 or ACF7 mutants, and co-sedimentation assays was performed as in (A). (C) Lysates were prepared from cells co-expressing caGSK3 with HA-tagged ACF7 or ACF7(S:A). Co-sedimentation assays and IBs were as in (A). (D) Lysates were prepared from untreated cells or cells treated with GSK3β inhibitors (1: LiCl, 2: AR-A01448). Co-sedimentation assays and IBs were as in (A).

Figure 4. Loss of ACF7 impairs migration of bulge stem cells in vivo and in vitro

(A) RT-PCR (left panel), immunoblot (IB, right panel) and quantifications (graph) of ACF7 mRNA (real-time RT-PCR) and protein from isolated bulge HF-SCs and non-SC extracts. Error bars denote standard deviation (SD). (B) Immunofluorescence of skin sections at different hair cycle stages. ACF7 is enriched in bulge (arrows), and although CD34 is present in dermis, its epithelial expression is specific to bulge. Color coding is according to secondary Abs used in detection. Nuclei were counterstained with DAPI (blue). Dashed lines denote basement membrane. HF, hair follicle; der, dermis; epi, epidermis; DP, dermal papilla. Bars=50μm. (C) Immunofluorescence of sections of HF bulges (arrows) for SC markers indicated. (D, E) Whole mount LacZ imaging and quantifications of bulge SCs and their progeny from perilesional follicles of WT and ACF7 cKO mice. Dashed lines denote wound boundary; dashed arrows denote trails of bulge SCs/progeny that migrated into wound. The inset is of unwounded K15-Cre-activated skin, where LacZ expression is confined to the bulge, not visible on the body surface. Blue cells migrated upward only after wounding. The length of the blue cell trails 4 or 6d postwounding (P4, P6) was quantified in (E) by box-whisker plots. (F) Wounded animals were topically treated with LiCl or Wortmannin (Wort). Bulge SC migration was quantified as in (E).

Figure 5. ACF7 is required to polarize MTs along F-actin-focal adhesion networks, which in turn is inhibited by GSK3β activity

(A) ACF7 protein and its phosphorylation status were evaluated by immunoblot of lysates from KO or WT bulge SCs treated with GSK3β inhibitors or Wnt3a, as indicated. (B, C) Cultured bulge SCs immunolocalized for α-tubulin (MT), ACF7, ACF7 phosphorylation (ACF7-Pi, with P2-specific Abs) or F-actin (phalloidin). Notes: 1) ACF7 but not phospho-ACF7 is enriched at polarized MT tips; 2) the diffuse cytoplasmic ACF7-Pi pattern of WT is absent in KO cells. (D) Immunofluorescence of WT bulge SCs expressing caGSK3β or vector control. Transfected cells are marked with arrows. Note: caGSK3β inhibits ACF7 localization at MT tips and leads to disorganized MT networks. (E) WT bulge cells were treated with LiCl to impair GSK3β activity and subjected to immunofluorescence as indicated. Note clusters of ACF7 at cell periphery. (F) Cultured ACF7 KO bulge SCs were microinjected with expression vectors encoding GFP alone, GFP-ACF7 (rescue), GFP-ACF7(S:A) (GSK3-refractile), or GFP-ACF7(S:D) (phosphomimetic). Cells were immunolabeled for MTs. Note that only WT and S:A mutant, and not GFP or S:D mutant, localized and rescued MT organization defects that occur in KO bulge SCs in low-Ca²⁺ medium. Wherever the field includes uninjected cells, injected ones are denoted by an asterisk. For B–F, boxed areas are magnified in insets. Bars=20 μm.

Figure 6. GSK3β Phosphorylation Governs ACF7 Functionality in Cell Polarity and Directional Cell Movement in Vitro

(A) Isolated HF-SCs were subjected to modified Boyden chamber assays, and a checkerboard analysis was performed to distinguish chemokinesis vs chemotaxis effects. Feeder-conditioned medium was used as chemoattractant, and was mixed with serum-free medium to load into upper and lower chambers. Migration was quantified as in Experimental Procedures, and presented as bar graphs. Color coding corresponds to the concentration of chemoattractants in the lower chamber. Concentrations of chemoattractants in upper chamber are indicated in bar graph legends. Note that only WT cells showed significant chemotactic behavior when exposed to a positive chemoattractant gradient. (B) Migration of treated or transfected WT HF-SCs was examined as in (A), with only 0 or 100% of chemoattractants used in lower chamber and no chemoattractants in upper chamber. Note that both global inhibition and artificial activation of GSK3β activity inhibit migration of HF-SCs. (C) Immunofluorescence detects Ser-9 GSK3β phosphorylation (inactivation), GM-130 (Golgi), E-cadherin (E-Cad) and chromatin (DAPI). Arrows denote enriched phospho-GSK3β at the leading edge. Dashed lines denote colony perimeters. θ represents angle between solid arrow, denoting outward direction of HF-SC colony and dashed arrow, denoting Golgi orientation. Note polarization of inactivated GSK3β at migrating edge of both WT and ACF7-null HF-SCs, and loss of Golgi polarization in ACF7-null HF-SCs. Bar=20μm. (D) Windrose plots of Golgi orientation (quantification of angle θ) in treated or un-treated WT and microinjected or uninjected ACF7-deficient cells. Note random distribution of θ in KO cells, and only WT ACF7 restores proper cell polarity. (E) Quantifications of chemotactic behaviors show that only WT ACF7 rescues migration defects in ACF7-null bulge SCs. Error bars represent SD.

Figure 7. GSK3β Phosphorylation of ACF7 Plays a Critical Role in Bulge Stem Cell Migration and Skin Wound Repair In Vivo

(A) Box and whisker plots (left panel) and bar graphs (right panel) show that in vivo migration of HF-SCs and overall skin healing (length of hyperproliferative epithelium in wound) were restored by WT-ACF7 transgene but not ACF7(S:A). Stars represent P value less than 0.05, and error bar represents SE. (B) Schematic of methodology used to activate and induce migration of quiescent HF-SCs in skin biopsy (panel 1) by ex vivo wounding. A typical telogen HF is shown in panel 2 with arrows marking potential direction of HF-SC migration upon wounding. HF-SCs are identified by C12-FDG (a fluorogenic substrate of LacZ), and outward migration is visualized by phase contrast (PhC) or fluorescence microscopy. Dashed lines denote basement membrane in panel 2, or migrating front in panel 3–4. Stars denote C₁₂FDG(+) cell in panel 4. (C) Representative examples of SC progeny after exiting and migrating out of the HF bulge. Immunofluorescence was used to visualize ACF7, MT and F-actin cytoskeletons (see also S6D). (D–E) Cells migrating out of the bulge SC niche were immunostained for phospho-GSK3β and phospho-ACF7. Note enrichment of inactive (phosphor) GSK3β at migrating front and corresponding lack of GSK3β-phosphorylation of ACF7 at plus ends of MTs localized there. (F) Live imaging of progeny which migrated out of the bulge SC niche during wound response. Cells were microinjected to express GFP-EB1, and MT +tip movements were then monitored by videomicroscopy and tracked automatically (upper panels). Corresponding MT behavior is color-coded and described in text and Experimental Procedures. Data for directionalities of MT growth were collected from movies and then quantified and presented as Windrose plots (lower panels). Arrow denotes direction toward wound front. Dotted lines denote migrating front of cells streaming from the wound-induced skin explant. Bar=20μm.