Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Cell. Author manuscript; available in PMC 2012 February 4.
Published in final edited form as:
PMCID: PMC3050560

Skin Stem Cells Orchestrate Directional Migration By Regulating Microtubule-ACF7 Connections Through GSK3β


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 directly phosphorylates ACF7, a >500kd 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, depending 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.

Keywords: stem cell, cell migration, microtubule, F-actin, spectraplakin, ACF7, GSK3β


Directional cell movement is essential for developmental morphogenesis, tumor metastasis and wound repair. A typical migrating cell adopts front-rear polarity with asymmetrical distribution of signaling molecules and cytoskeletal components. During establishment of polarity, temporal capture and stabilization of microtubules (MTs) occur near filamentous actin (F-actin)-enriched leading edges, which enable reorientation of the MT-organizing center and Golgi complex to ensure biased vesicular transport for directional migration (Etienne-Manneville, 2004; Siegrist and Doe, 2007). Among many regulatory molecules required for cell migration, the ubiquitously-expressed serine/threonine kinase GSK3β (glycogen synthase kinase 3β) is particularly important in transmitting upstream signaling necessary for establishing cell polarity and guiding directional movement (Sun et al., 2009).

In contrast to most other protein kinases, GSK3β activity is high in nonstimulated cells, but it is dampened at the leading edge of scratch-wounded astrocytes and endodermal cells in vitro and in developing neurons of brain slices ex vivo (Etienne-Manneville and Hall, 2003; Kodama et al., 2003; Yoshimura et al., 2006). In vitro, GSK3β inhibition and cell polarity are known to be provoked by CDC42 (Etienne-Manneville and Hall, 2003). In vivo, GSK3β activity is inhibited by Wnt signaling, which in turn leads to β-catenin stabilization and a critical and near universal role in stem cell (SC) activation and migration (Fuchs, 2009; Nusse et al., 2008). Despite this tantalizing connection, the possible physiological relevance between Wnt signaling, SC activation and cytoskeletal remodeling remains unclear. Another key unaddressed question is how GSK3β regulates the dynamic changes in MT organization and stabilization that transpire during polarized movements of somatic SCs.

Hair follicles (HFs) provide an ideal model system to address these issues. Adult HFs undergo homeostasis through cyclical bouts of active growth (anagen), regression (catagen) and rest (telogen) (Fig. S1A). They also participate in epidermal re-epithelialization during wound-healing (Blanpain and Fuchs, 2009; Ito et al., 2005; Tumbar et al., 2004). Both processes rely upon a resident population of SCs, which reside in a specific niche called the bulge, located at the base of the non-cycling portion of the HF (Fig. S1A). Wnt signaling emanating from polarized epithelial-mesenchymal cross-talk in the SC niche is critical for HF-SC activation during tissue homeostasis (Greco et al., 2009). In addition, localized Wnt signaling/β-catenin at a wound site initiates the migration of SCs after injury (Ito et al., 2005; Ito et al., 2007). Based upon these points, the potential exists for using skin as a model to dissect how signaling cues to SCs might regulate the cytoskeleton to orchestrate cell polarization and directional cell movement. In searching for potential cytoskeletal regulators in this process, we noticed that HF-SCs display significantly more ACF7 (Actin Crosslinking Factor-7; also called MACF1, MT/actin crosslinking factor-1) than other skin epithelial cells (Blanpain et al., 2004; Morris et al., 2004; Tumbar et al., 2004) (Fig. S1B).

Unique to multicellular organisms, spectraplakins such as ACF7 can bind both MT and actin networks (Jefferson et al., 2004; Roper et al., 2002). Although broadly expressed, their essential functions are most clearly revealed in muscle, neurons and skin epithelial cells, which maintain elaborate yet dynamic cytoskeletal networks. Mutations in the single Drosophila spectraplakin gene (Kakapo/short-stop/shot) cause a wide variety of cellular and tissue defects that include perturbations in actin-MT organization, cell-cell adhesion and integrin-mediated epidermal attachment to muscle. There are two evolutionarily conserved mammalian counterparts. Mice lacking BPAG1/dystonin display sensory neuron and muscle degeneration, and have gross defects in cytoskeletal organization and function. By contrast, mice lacking ACF7 exhibit early embryonic lethality (Chen et al., 2006; Kodama et al., 2003). Recent studies show that mice conditionally lacking ACF7 display defects in cell migration. This was true for both K14-Cre cKO animals, impaired in skin wound-healing (Wu et al., 2008) and for Nestin-Cre cKO mice, defective in brain development (Goryunov et al., 2010).

While the loss of function data underscore the importance of spectraplakins in coordinating the cytoskeletal dynamics necessary for cells to polarize and move in a directed fashion (Rodriguez et al., 2003), the mechanisms underlying the regulation of spectraplakin-mediated actin-MT connections remain unknown. Similarly lacking are the molecular details of the circuitry that must link upstream signaling pathways to cytoskeletal remodeling in order for SCs to migrate from their niche. In the present report, we make major inroads into understanding this process. Specifically, we (1) identify clusters of functional GSK3β phosphorylation sites in ACF7’s MT binding domain; (2) show that GSK3β-phosphorylation at these sites suppresses ACF7’s ability to bind MTs; (3) generate phosphospecific ACF7 antibodies (Abs) and show that both states co-exist in HF-SCs, but only the unphosphorylated state localizes to MT tips that track along F-actin fibers and converge at focal adhesions; (4) generate phosphorylation-refractile and phosphorylation-constitutive mutations in full-length ACF7 and show that in ACF7-null HF-SCs, only phosphorylation-refractile mutants restore overall MT architecture, but both mutants abrogate ACF7’s ability to rescue cellular polarity and directional cell migration (5) perform rescue experiments in ACF7 cKO mice to show that only WT and not phosphorylation-refractile ACF7 can restore efficient migration of HF-SC and normal wound repair in a physiological context.


C-Terminal Tail Interactions Govern Binding between ACF7 and Microtubules

ACF7’s carboxy-terminal tail (CT) contains a GAR (Gas2-related) domain and a GSR (GSR-repeat) domain. Previous studies suggest and we’ve confirmed that both domains are involved in the interaction with MTs (Sun et al., 2001; Wu et al., 2008). To obtain structural information on this interaction, we incubated ACF7(CT) with polymerized MTs and conducted ultrastructural analyses. Under saturating conditions [4:1 molar ACF7(CT):tubulin heterodimer], ACF7(CT) markedly enhanced the electron density along the MT surface (Fig. 1A). When compared with naked MTs, ACF7(CT)-coated MTs were ~10 nm thicker in diameter (projection profile in Fig. 1A). Fourier Transform analyses further indicated that whether decorated with ACF7(CT) or not, assembled MTs displayed a 40Å layer line corresponding to the packing of tubulin dimers. However, only ACF7(CT)-decorated MTs displayed discrete 80Å layer lines, suggesting that ACF7 might associate with MT lattice with a rather weak distinction between α- and β- tubulins (Fig. 1A).

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

Interaction with MT lattice usually involves the acidic C-terminal tails of tubulin subunits that protrude from the MT surface. To test this hypothesis, we performed binding assays between ACF7(CT) and increasing concentrations of taxol-stabilized MTs. Just prior to adding ACF7(CT), we exposed half the polymerized MTs to subtilisin to shave protruding tubulin tails (MTΔC-tail). Following ultracentrifugation, pellets were then analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) and Coomassie Blue (CB) staining. Subtilisin-treated MTs still pelleted after ultracentrifugation, confirming that MTs remained assembled after treatment. Only a slight increase was noted in tubulin’s electrophoretic mobility, consistent with tail removal (compare asterisked lanes in Fig. 1B). However, this modification markedly diminished ACF7(CT)’s binding to MTs (Fig. 1B). When ACF7(CT)’s concentration was progressively increased, its association with MTs became saturating, reflected by an increase in the soluble pool of free ACF7(CT). A Scatchard plot of the data is shown in Fig. 1C. ACF7(CT) binding to untreated MTs gave a Kd of ~1.4 × 10−7M, which was comparable to published results on GAR and GSR domains of ACF7 (Sun et al., 2001). By contrast, the affinity of ACF7(CT) for subtilisin-treated MTs was >10× weaker (Kd ~1.5 × 10−6 M).

Phosphorylation of ACF7’s C-Terminal Domain by GSK3β

Association with tubulin tails usually depends on electrostatic interactions between acidic tubulin C-termini and positively-charged surfaces of MT-binding proteins. Consistent with this notion, ACF7(CT) contains many lysine (K) and arginine (R) residues. Particularly, the GSR domain (202 amino acids) harbors 36 strongly basic residues, with a calculated isoelectric point at 11.8. Additionally, 32% of residues in ACF7’s GSR domain are serine or threonine, suggestive of the potential to regulate ACF7-MT’s electrostatic interactions through protein phosphorylation. To address this possibility, we transfected primary keratinocytes with ACF7(CT) and labeled them with [32P]-orthophosphate. Our results showed that ACF7(CT) is efficiently phosphorylated in vivo (Fig. 1D).

We next analyzed phospho-ACF7(CT) by mass spectrometry. We circumvented the difficulties posed by ACF7(CT)’s high percentage of basic residues by choosing protease AspN, which cleaves peptide bonds N-terminal to Aspartate. LC-MS/MS (liquid chromatography-tandem mass spectrometry) identified multiple (up to 6) phosphorylation sites within a peptide encompassing the GSR repeats (Fig. 1E, 1F). Interestingly, the characteristic GSR repeats within this sequence have 6 serines that match consensus GSK3β phosphorylation sites (phosphorylation cluster-1, P1). While in-silico analysis with different phosphorylation prediction algorithms (Obenauer et al., 2003; Schiller, 2007) reveal additional potential GSK3β sites (phosphorylation cluster 2, P2, Fig. 1F), their incompatible sequence context precluded efficient peptide retrieval for MS/MS analysis.

To determine whether ACF7 is specifically phosphorylated by GSK3, we first tested for an endogenous association between ACF7 and GSK3β proteins in HF-SCs. Immunoblot analyses revealed GSK3β in anti-ACF7 immunoprecipitates of WT but not ACF7-null cell lysates (Fig. 2A). In vitro kinase (IVK) assays further showed that ACF7 is a substrate for active GSK3β, and that GSK3β phosphorylates ACF7(CT) but not ACF7-NT (N-terminal domain of ACF7, serving as a control) (Fig. 2B). Moreover, co-expression of GSK3β with ACF7(CT) in cultured cells resulted in phosphorylation of ACF7 that was sensitive to treatment of phosphatase (Fig. 2C).

Figure 2
GSK3β associates with and phosphorylates ACF7

To assess whether our identification of GSK3β phosphorylation sites in ACF7(CT) was correct, we replaced the predicted GSK3β-targeted serines with alanines and repeated our phosphorylation assays in vitro and in vivo. Individually, mutations in P1 and P2 each reduced overall phosphorylation. Combinatorial mutations of both clusters abolished ACF7(CT) GSK3β phosphorylation (Fig. 2D, 2E).

Phosphorylation of ACF7 by GSK3β Inhibits Microtubule Binding

Our finding of functional GSK3β phosphorylation sites in ACF7’s GSR domain hinted a potential role of this signalling event in tempering ACF7’s MT connection by reducing their electrostatic affinity. To test this possibility, we had to first overcome the technical hurdles of ACF7’s enormous size (5380 amino acid residues) and engineer mammalian expression vectors encoding HA-tagged full-length ACF7 as well as point mutants that converted GSK3β phosphorylation sites at P1 and P2 to either a kinase-refractile version harboring Ser→Ala mutations (S:A mutant) or a phosphomimetic version, containing Ser→Asp mutations (S:D mutant).

We purified the proteins by affinity chromatography (Wu et al., 2008), and carried out in vitro co-sedimentation assays with polymerized MTs (Fig. 3A). Both HA-tagged ACF7 and its S:A mutant counterpart maintained a strong affinity for MT-binding similar to WT ACF7, while ACF7(S:D) exhibited significantly reduced affinity for MTs. Similar results were obtained when the binding assays were repeated with freshly-prepared lysates from cells expressing ACF7 or its mutants (Fig. 3B). In contrast to the effects of S:A and S:D mutations on MT-binding affinity, ACF7’s F-actin binding affinity and ATP-hydrolysis activity were unaffected (Fig. S2A, B).

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

To directly assess the effects of GSK3β phosphorylation, we next co-expressed constitutively-active GSK3β (caGSK3β, S9A mutant) with either WT or kinase-refractile ACF7. As predicted, the MT-binding affinity of WT-ACF7 was reduced when GSK3β was superactivated (Fig. 3C). This effect was specific for ACF7’s C-terminal GSR domain, since GSK3β did not alter interactions between ACF7(S:A) and MTs (Fig. 3C). Consistent with these results, inhibition of endogenous GSK3β activity by exposing HF-SCs to different GSK3-specific inhibitors significantly increased the MT-binding affinity of endogenous ACF7 (Fig. 3D). Together, these findings provided compelling support for GSK3β as an important regulator of ACF7’s association with MTs, and showed that GSK3β-mediated regulation was exerted exclusively at P1 and P2 of the ACF7’s GSR domain.

The Wound Healing Delay in ACF7-Deficient Skin Arises From Defective HF-SC Migration

Previously we showed that mice targeted for loss of ACF7 in skin were defective in wound-repair (Wu et al., 2008), a process known to involve HF-SCs (Ito et al., 2005; Tumbar et al., 2004). When coupled with the appearance of ACF7 on the list of genes upregulated in HF-SCs (Fig. S1B) and an emerging role for Wnt signaling in wound repair (Fathke et al., 2006; Ito et al., 2007; Stoick-Cooper et al., 2007), our finding of GSK3β as a potential regulator of ACF7 took on newfound importance, and merited further investigation.

To begin to evaluate how ACF7 functions in this pathway, we first purified HF-SCs (CD34hiα6-integrinhi) by fluorescence activated cell sorting (FACS) vs other basal cells (CD34negα6-integrinhi) (Blanpain et al., 2004). RT-PCR and immunoblot on these purified cell populations verified ACF7’s enrichment in HF-SCs at both mRNA and protein levels (Fig. 4A). Immunofluorescence further documented elevated ACF7 in this niche throughout the hair cycle (Fig. 4B).

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

Loss of ACF7 in HF-SCs did not alter bulge architecture (data not shown) nor did it affect expression of key bulge markers (Blanpain and Fuchs, 2009) (Fig. 4C). In addition, no significant changes were found in proliferation or apoptosis of homeostatic HF-SCs (more details below), or in hair growth or cycling (Wu et al., 2008). We therefore focused on the hypothesis that the associated defects in wound-repair originate from perturbations in HF-SC migration. To test this, we bred our ACF7fl/fl animals with mice expressing a progesterone-regulatable recombinase (K15-Cre-PGR) specifically in bulge SCs (Ito et al., 2005). To monitor HF-SC progeny in a wound response, we further bred these mice to Rosa26-lox-Stop-lox-LacZ reporter animals.

As expected, treatment of adult mice with RU486 (a progesterone antagonist) activated Cre and selectively marked HF-SCs. When challenged to a wound, activated LacZ+ bulge cells exited the niche and migrated upward to reepithelialize wounded epidermis (Fig. 4D). This was readily observed by whole-mount staining, which displayed trails of HF-SC-derived blue (LacZ+) cells emanating from perilesional follicles of wounded tissue (dashed arrows). By contrast, ACF7 cKO bulge cells were delayed in this process by ~40% compared with wild-type (WT) controls over 4–6d after injury (Figs. 4D, 4E). Importantly, since targeting was specific to bulge cells, the delay was directly attributable to an SC defect. Moreover, ACF7-deficiency did not affect proliferation or apoptosis of bulge SCs, indicating that the defect was rooted in cell migration (Fig. S3A, B).

To examine the contribution of GSK3β in this process, we took a pharmacological approach to manipulate GSK3β activity in vivo. Wounds on WT skin were treated with either lithium chloride (LiCl) which directly inhibits GSK3β, or Wortmannin, which activates GSK3β by inhibiting an upstream regulator, phosphoinositol-3 kinase (PI3K). Interestingly, both treatments impaired wound-induced cell migration out of the adult HF-SC niche (Fig. 4F), suggesting that spatiotemporal regulation of GSK3β’s activity is required to achieve efficient bulge SC migration in vivo.

The Polarization of MTs along Actin-Focal Adhesion Networks is Abrogated When ACF7 is Phosphorylated by GSK3β

Previously, we showed that when ACF7 is ablated, cultured epidermal keratinocytes cannot coordinate microtubule growth along F-actin filaments, a feature which in turn leads to overstabilization of focal adhesions (FAs) and defective cell movement (Wu et al., 2008). When taken together with our results thus far, we posited that in vivo, HF-SCs might respond to migratory stimulations such as Wnt signaling by spatiotemporally regulating ACF7-MT connections and promoting the cytoskeletal remodeling needed for polarized migration. To test this hypothesis, we first generated phospho-specific ACF7 Abs against two synthetic phospho-peptides corresponding to the ACF7 GSR sequences encompassing P1 and P2, respectively. Each Ab was specific for the phosphorylated state of its GSK3β target sequence: when ACF7 was not phosphorylated or when the sites were selectively mutated, the Abs failed to recognize the ACF7 protein (Fig. S4A, B).

Isolated HF-SCs can sustain long-term culture without losing stemness (Blanpain et al., 2004), allowing us to investigate the role of ACF7 phosphorylation in vitro. Both ACF7 Abs detected the expected sized band in immunoblots of cultured bulge SC but not ACF7 cKO lysates (Fig. 5A). Importantly, these signals were sensitive not only to chemical inhibitors of GSK3β, but also to recombinant Wnt3a. These data confirmed the specificity of our phospho-specific Abs, and further demonstrated the ability of Wnt signaling to repress ACF7 phosphorylation in its C-terminal tail.

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

We next examined the GSK3β phosphorylation status of endogenous ACF7. In vitro, ACF7 decorated the ends of MTs that are co-aligned with underlying F-actin cables (Fig. 5B). By contrast, phospho-ACF7 was diffuse and/or punctate throughout the cytoplasm and showed no association with these MT cables (Fig. 5C). Nevertheless this cytoplasmic staining was specific for ACF7, as it was abolished in ACF7 KO bulge cells (Fig. 5C). These findings were consistent with the phospho-dependent decreased affinity of ACF7 for MTs that we observed in vitro and revealed a marked correlation between GSK3β phosphorylation of ACF7 and a severing of the ACF7-MT connection.

To directly evaluate the effect of GSKβ phosphorylation on this process, we overexpressed caGSK3β in WT cultured bulge SCs. In contrast to empty vector alone (Ctrl), caGSK3β dramatically reduced ACF7 localization along MTs (Fig. 5D). Moreover and quite remarkably, expression of constitutively active GSK3β transformed the straight and radial MTs of WT cells into a network of bent and curly MTs, reminiscent of the aberrant MT network typifying ACF7 KO cells (Fig. 5D, compare with data in 5C).

If GSK3 activation and phosphorylation of ACF7 is responsible for severing the polarization of MTs at the migrating front, then inhibiting endogenous GSK3β activity might be expected to stabilize these connections. Indeed, when we treated HF-SCs with LiCl, under conditions which potently inhibited GSK3β activity and ACF7 phosphorylation, ACF7 clustered at the ends of MTs (Figs. 5A, E). Additionally, polarized sites of converging ACF7, MT and F-actin usually colabeled with Abs against focal adhesion (FA) proteins (Fig. S4C).

ACF7-deficiency stabilizes FA through inhibiting MT targeting to FA (Wu et al., 2008). However, when DsRed-Zyxin-expressing HF-SCs were subjected to videomicroscopy and quantified, the effects of LiCl on FA turnover were modest (Fig. S4D, Supplemental Movie S1). This was also the case for the average size of FAs and the level of focal adhesion kinase (FAK) activity, which influence FA dynamics as well (Figs. S4E, F). Overall, these results suggest that constitutive association between ACF7 and MTs may not elicit the alterations in FA stability that are seen when ACF7 is missing altogether.

GSK3β has many targets, which complicates the interpretation of LiCl experiments (Sun et al., 2009). To distinguish the specific effects of GSK3β on ACF7-mediated cytoskeletal dynamics, we performed rescue experiments with our ACF7 phosphorylation mutants. To more precisely control concentration and ensure comparable expression of encoded proteins, we microinjected our expression constructs into primary cultured HF-SCs null for ACF7.

Introducing GFP-tagged versions of either full length or S:A mutant ACF7 restored ACF7’s localization to the MT ends residing near or at the cortex (Fig. 5F). Consistent with their MT-binding capability, kinase-refractile ACF7(S:A) and full-length ACF7 also generated arrays of radial MT trajectories in individual ACF7-null cells (Fig. 5F). By contrast, the phosphomimetic ACF7(S:D) was diffusely localized, and its MT organization appeared no different than in uninjected or GFP-injected ACF7-null cells (Fig. 5F). Taken together, these experiments provide compelling evidence that GSK3β plays a critical role in regulating ACF7’s dissociation from MTs and that phosphorylation is sufficient to dramatically alter polarized organization of MTs along actin stress fibers converging at FAs.

GSK3β Phosphorylation of ACF7 Must Be Dynamic to Govern Polarization and Directed Movement in Cultured HF-SCs

Cultured HF-SCs further allowed us to examine their directional movement in vitro. When subjected to a modified Boyden chamber assay with conditioned feeder fibroblast medium as a chemo-attractant, WT HF-SCs showed a marked, dose-dependent migratory response, which was greatly diminished in HF-SCs lacking ACF7 (Fig. 6A). Moreover, stimulation of WT HF-SCs’ migration was achieved only when conditioned medium was administered in a positive concentration gradient. These data further documented the chemotactic nature of the response, and confirmed ACF7’s role in sustaining directional cell movement. Consistent with our in vivo observations, treatment of HF-SCs with GSK3β inhibitors or ectopic expression of caGSK3β inhibited the response (Fig. 6B).

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

Directionality of movement relies heavily on cell polarity, and prior studies indicated that embryonic endodermal cells lacking ACF7 cannot sustain cell polarity after scratch-wounding in vitro (Kodama et al., 2003). To evaluate how regulated ACF7-MT connections might contribute to cell polarity, we devised a method to polarize cultured bulge SCs by seeding them at low density on fibronectin-coated dishes, and then elevating Ca2+ levels to induce cell-cell adhesion. Under these conditions, the perimeter of isolated WT colonies exhibited significant polarity as determined by immunolocalization of phosphorylated (inactive) GSK3β, Par proteins and aPKC (Fig. 6C, S4G). Polarization of perinuclear Golgi was particularly prominent, enabling quantification by measuring its preferential localization around the axis bisecting nucleus and colony edge. Interestingly, ACF7-deficient bulge SCs displayed polarized GSK3β phosphorylation but not Golgi (Fig. 6C). Moreover, manipulating GSK3β activity in WT HF-SCs disrupted Golgi polarization (Fig. 6D). These data placed ACF7 midstream in the pathway that links polarized GSK3β inhibition at the HF-SC front and Golgi polarization in the perinuclear region.

Our findings were intriguing in light of prior data showing that MTs are essential for polarizing Golgi assembly (Miller et al., 2009; Siegrist and Doe, 2007). We therefore wondered whether rescuing the ability of MTs to polarize along actin cables might also rescue defective Golgi polarization. To test this possibility, we repeated the polarization assays, this time with bulge SCs microinjected with our GFP-tagged versions of WT and phosphorylation-altered ACF7. WT-ACF7 rescued polarization of HF-SCs (Fig. 6D). As expected from its failure to bind MTs (Fig. 5F), ACF7(S:D) also failed to effectively polarize Golgi (Fig. 6D). Interestingly however, even though ACF7(S:A) efficiently rescued MT organization (Fig. 5F), it failed to rescue Golgi orientation (Fig. 6D).

Finally, we tested the ability of our mutants to rescue the chemotactic migration defects seen in ACF7-null cells. As shown in Fig. 6E, only WT-ACF7 rescued the ability of bulge SCs to migrate efficiently. Together, these data showed that the ability of MTs to track along actin cables is not sufficient to achieve either polarization or effective migration of bulge SCs. Moreover, both of these processes require in addition the dynamic regulation of ACF7 phosphorylation, since neither the phosphomimetic mutant, the phosphorylation refractile mutant nor the two mutants combined (Fig. 6E) were able to rescue these defects in KO cells.

GSK3β Phosphorylation of ACF7 Regulates Bulge SC Migration and Skin Wound Repair In Vivo

We next focused on whether dynamic regulation of ACF7 phosphorylation coordinates polarized bulge SC migration during wound repair in vivo. We began by engineering transgenic mice expressing N-terminally GFP-tagged, full-length versions of ACF7 and ACF7 S:A under the control of K14 promoter/enhancer. Mice genotypic for K14-ACF7 or K14-ACF7(S:A) alleles were born in the expected Mendelian numbers and grew normally (Fig. S5A). Transgenic ACF7 and ACF7(S:A) GFP-tagged proteins of the correct size were expressed comparably, and exhibited the expected differential GSK3β phosphorylation states in vivo (Fig. S5A). Immunofluorescence confirmed skin-specific transgene expression (Fig. S5B).

To determine the ability of these transgenic proteins to compensate for loss of ACF7 in HF-SCs in vivo, we bred our transgenics to ACF7fl/fl:K15-Cre-PGR:Rosa26-LacZ mice, and then induced ablation of endogenous ACF7. Mice expressing these transgenes and not ACF7 in their HF-SCs were visibly normal, and no gross differences were noted in hair cycles (Fig. S5C; data not shown). Immunofluorescence for a variety of differentiation markers showed normal morphology and localization patterns (Fig. S5D, S5E).

Given the normal tissue architecture and homoeostasis, we next turned to investigating how ACF7’s GSK3β-phophorylation status affects the ability of HF-SCs to respond and migrate outward to repair epidermis upon skin wounding. In response to punch wounds, only ACF7 cKO mice expressing GFP-ACF7, and not GFP-ACF7(S:A), showed significant rescue of bulge SC migration defects, as measured by LacZ whole mount staining (Fig. 7A, left). This difference appeared to reflect a differential ability to restore directional cell movement, since bulge SC proliferation and apoptosis assays showed no such change (Fig. S3A,B). Similar results were obtained when the wounding challenge was broadened by using transgenic mice mated to ACF7fl/fl: K14-Cre mice. Once again, the area of hyperproliferative epithelium that typically migrates into the wound site was only significantly rescued by GFP-ACF7 and not GFP-ACF7(S:A) (Fig. 7A, right).

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

Finally we addressed whether changes in GSK3β activity alter ACF7’s ability to coordinate MT-actin dynamics during wound-induced directed migration of bulge SCs out of HFs. For this purpose, we induced a rapid wound response in HFs by cutting the skin at the resting (telogen) phase and placing it into rich medium. In a process analogous to wound healing, the outward migration of activated HF-SC progeny could then be imaged by videomicroscopy and immunofluorescence (Fig. 7B–F; Fig. S6).

Targeted bulge cell progeny were identified by their ability to cleave a fluorogenic substrate for lacZ (Fig. 7B). Consistent with our in vitro results, Golgi complex as well as Par6 and aPKC localized in a polarized fashion in WT bulge SCs at the leading front (Fig. S6A). These cells also contained FAs (Fig. S6A) and well-polarized MT bundles that co-aligned with F-actin, and ACF7 localized at the interface between these MT +tips and F-actin bundles (Fig. 7C). Loss of ACF7 resulted in disorganized MT architecture (Fig. 7C). Additionally, ACF7 deficiency or change in GSK3β activity led to perturbations in cell polarity and migration of the marked bulge SCs/progeny that were streaming from the HFs. Interestingly, while both WT and S:A mutant ACF7 rescued the disorganized MT network, only WT ACF7 rescued alterations in cell polarity and migration (Figs. 7C; S6B, C).

Similar to our in vitro results (see Fig. 6C) and those obtained from other model systems (Sun et al., 2009), Ser9-phosphorylated (inactive) GSK3β was enriched at the leading edge of migrating HF-SCs (Fig. 7D). Consistent with these data, the majority of ACF7 phosphorylated by GSK3β was localized in the cell body and not at the leading edge of HF-SCs (Fig. 7E).

As a +tip protein, ACF7 can guide movement of MT plus ends. To monitor potential polarity of MT dynamics during wound-induced migration, we microinjected migrating bulge SCs/progeny with a GFP-tagged EB1 expression vector. Videomicroscopy combined with automated tracking of this plus end (+tip) MT binding protein revealed polarized MT growth toward the migrating front of WT HF-SCs (Supplemental movie 2 and Fig. 7F). Directionality of MT growth was largely randomized in KO cells, and only WT-ACF7 and not ACF7(S:A) rescued it (Fig. 7F). Taken together, these findings underscore an essential physiological role for GSK3β-mediated phosphorylation of ACF7 in sustaining polarized MT growth and directed HF-SC migration during wound healing.


Aberrant mobilization of SCs in response to injury can delay wound repair and have dire consequences to animal survival (Fuchs, 2009). Exposed to frequent mechanical stresses, skin SCs have developed a unique and elaborate cytoskeletal system. However little is known about how cytoskeletal dynamics are coordinated in these or other SCs. In this report, we’ve demonstrated the role of ACF7 and ACF7-mediated cytoskeletal dynamics in SCs in vivo. Our studies show that ACF7 is required for efficient upward migration of bulge cells in response to wounding, and that this function is primarily rooted in ACF7’s ability to coordinate MT dynamics and polarize HF-SCs. Through a comprehensive approach encompassing biochemistry, molecular and cell biology, we further unveiled a hitherto unrecognized regulatory role for GSK3β-mediated phosphorylation of ACF7’s major MT binding domain. We show that the consequence of this phosphorylation is the attenuation of interactions between ACF7’s basic GSR domain and the acidic C-terminal tubulin tails.

Polarized cell movement is an essential component of diverse biological processes such as cancer metastasis, tissue development, and wound healing (Lauffenburger and Horwitz, 1996). All these processes share significant similarities regarding their mechanisms for sensing directional cues and translating them into directional locomotion. Cells under the influence of various chemotactic molecules exhibit polarized activation of signaling proteins, such as the CDC42 Rho GTPase and phosphatidylinositol-3-phosphate kinase (PI3K), that guide differential remodeling of cytoskeletons at the leading edge versus the back of cells (Etienne-Manneville, 2004; Siegrist and Doe, 2007).

MTs play a particularly important role in this process, delivering positional information to establish the proper site of cortical polarity (Siegrist and Doe, 2007). Once MTs and their associated proteins determine the polarity site, a positive feedback loop initiates interactions between the actin-rich cortex and growing (plus) ends of MTs, resulting in reinforcement and maintenance of polarity. Such a mechanism not only provides cells with the ability to sense and amplify small asymmetries in their field, but also buffers and maintains the polarity axis after it is established. Common components of MT-based polarity pathways are plus end-directed kinesin motors and MT plus end-stabilizing proteins, including CLIP-170, EB1, Clasps and adenomatous polyposis coli (APC) (Akhmanova and Steinmetz, 2008). Our prior studies added ACF7 to this list (Karakesisoglou et al., 2000; Kodama et al., 2003; Wu et al., 2008). Our current study lends physiological relevance for ACF7’s ability to polarize MTs, and identifies a specific role for this connection in enabling HF-SCs to polarize and migrate into a wound site. In addition, our results elucidate a hitherto unappreciated signaling pathway whereby ACF7’s function is regulated dynamically in HF-SCs by GSK3β in order to control the directionality of MT growth, cell polarity and migration.

GSK3β modifies various regulatory components of MT cytoskeleton, including APC and some neuronal MT-associated proteins (Sun et al., 2009). Intriguingly, a recent study showed that ErbB2-induced repression of GSK3 is required for MT capture and targeting of ACF7 to plasma membrane in breast carcinoma cells (Zaoui et al., 2010). Our results identify ACF7 as a GSK3β substrate in skin somatic SCs, and raise the possibility that GSK3β may function as a master regulator of MT polarity. Importantly, however, our findings reveal that although only phosphomimetic ACF7 mutants disrupt MT organization, non-phosphorylatable ACF7 mutants nonetheless impair global cell polarization and SC migration. These findings underscore the importance of regulating ACF7’s phosphorylation status in a spatially and temporally defined manner in order to establish proper directionality in cells.

Consistent with this notion, GSK3β activity is specifically inhibited by CDC42 signaling at the leading edge of migrating cells (Etienne-Manneville and Hall, 2003; Sun et al., 2009), and also by Wnt signaling, which is often polarized in tissues (Nusse, 2008). Notably, Wnts have been broadly implicated in wound-repair and are required for HF-SC activation at the start of the hair cycle (Blanpain and Fuchs, 2009). Although ACF7 has been implicated in recruiting the Axin-APC-GSK3β complex to the site of active Wnt signaling in the gastrulating embryo (Chen et al., 2006), an essential role in Wnt signaling would not explain why SC activation and hair cycling still occur in ACF7 cKO skin. Our findings now provide an alternative role for ACF7 in Wnt signaling, namely as a downstream sensor of GSK3β inhibition and a mobilizer of the polarized response necessary for SC migration. Collectively, our findings support a model in which upstream chemotactic cues, e.g. Wnt signaling, trigger cellular polarization and directional movement through spatiotemporal regulation of ACF7 phosphorylation by GSK3β (Fig. S7). In this model, some outcomes of Wnt signaling and GSK3β inhibition, e.g. SC activation, proliferation and fate commitment in normal homeostasis, might still be maintained, while others, e.g. SC migration during wound-repair, would rely upon the ability to polarize directed migration. Future studies will determine the extent to which this pathway can explain ACF7’s broad and essential presence in tissues.

In closing, our findings provide insights into the complex but important molecular machinery underlying the polarized SC migration that integrates injury-induced migratory signals, cytoskeletal remodeling and polarized cell movements in HF-SCs. Our studies also add a fascinating new twist to the repertoire of spectraplakin’s many diverse and critical functions, and now pave the way for probing more deeply into the role of spectraplakins in mammalian SCs.


Generation of Transgenic Mice and Skin Wound Healing

The transgenic expression cassette was constructed so that GFP-ACF7 or GFP-ACF7(S:A) cDNA sequence were inserted 3′ to the human K14 promoter/enhancer and β-globin 5′UTR, and 5′ to the K14 3′UTR. Mice harboring these transgenes were engineered in Fvb/n albino mice and selected for their comparable expression relative to endogenous ACF7.

Skin wound healing assays were performed as described (Wu et al., 2008). To monitor the migration of SCs and their progeny after wounding, we treated K15-Cre-PGR:R26-LacZ:ACF7fl/fl mice ± transgene with RU486 for 5d starting at their first telogen (~P21). At 8 wks, mice were anesthetized and full-thickness punch wounds (6 mm) were introduced on their backs. Skin samples were then collected at 4 or 6d after wounding and β-galactosidase activity was detected in whole-mount tissues by X-gal staining (Ito et al., 2005). For each genotype, ≥10 mice were analyzed for the wound response.

To monitor the activation and migration of SCs from resting HFs subjected to wound-response, 3 mm biopsies were taken from back skin of 2-month-old K15-Cre-PGR:R26-LacZ mice that were treated with RU486. Biopsies were coated with a thin layer of matrigel for adhesion and immediately transferred to a coverslip coated with 10 μg/ml fibronectin. Skins were then incubated at r.t. for 10 min. to solidify the matrigel and then exposed to rich E-media containing 15% serum and 0.3 mM calcium (Blanpain et al., 2004), which promoted HF-SC outgrowth within 2–3d, analogous to a wound-response. Live HF-SCs/progeny were identified with fluorogenic β-galactosidase substrates C12FDG or C12RG (Invitrogen, yield green or red fluorescence respectively) according to the manufacturer’s instruction.

Cell Migration Assays and Time-Lapse Videomicroscopy

Migration assays in 96-well chemotaxis chambers (Millipore, Billerica, MA) were carried out according to manufacturer’s instructions. Briefly, 3T3 fibroblast-conditioned medium or serum-free medium (control) was added to the lower chamber. Bulge SCs were added to the upper chamber, and after 10h at 37°C, cells that migrated through the filter into the bottom chamber were collected, lysed and stained with CyQuant GR dye (labels DNA). A fluorescence plate reader was used to quantify fluorescence intensities.

To track individual HF-SC movement, skin biopsies were imaged with an Olympus phase-contrast microscope for 1d and manually tracked with NIH Image J. Displacements along the direction facing the leading edge of the migrating front were recorded and quantified (Wu et al., 2008). To monitor MT plus end movement in live cells, HF-SCs were microinjected with plasmid encoding GFP-EB1. 6 hr post injection, cells were imaged with a confocal-spinning disk microscope (Wu et al., 2008) for 5 min. at 2 sec./frame. Plus Tip Tracker software package (Matov et al., 2010) was used to process and track EB1 movements at the leading front. Because of MT dynamic instability, some gaps appeared between MT growth (red solid tracks in the output image of Fig. 7G). Gaps could occur in either the forward or backward direction depending on underlying MT dynamics and detection performance. Forward gaps could be a MT pause (cyan dotted) or reclassified as growth (green solid). Backward gaps could be MT shrinkage/catastrophe (yellow dotted) or reclassified as pause events (blue dotted). The same color codes were used in the output movie (supplemental movie 2), and initiation of a new growth track or gap was marked by a circle with corresponding color. Angles between MT growth track and outward direction of explants were calculated with a linear fit function of Matlab and plotted as Windrose plots.

Microtubule cosedimentation assay, subtilisin treatment and Immunoblotting

MT-binding assays were performed as described (Wu et al., 2008). Equivalent amounts of pellet were analyzed by CB staining or immunoblotting. Cleavage of the unstructured tubulin C terminus was carried out by limited proteolysis of taxol-stabilized MTs with subtilisin (Knipling et al., 1999). The proteolysis reaction was stopped by adding freshly prepared 20mM PMSF in DMSO. Subtilisin-treated MTs were pelleted by ultracentrifugation at 60,000 × g and resuspended in MT-stabilizing buffer containing 20mM PMSF and taxol to ensure complete removal of active protease and cleaved C-terminal tubulin fragments. Immunoprecipitations and immunoblotting were performed as described (Wu et al., 2004).

Supplementary Material





We are grateful to M. Schober, Y.C. Hsu, E. Ezratty, and T. Chen for discussions and helpful comments. We are grateful to J. Fernandez and H. Deng (Proteomics Center) for their technical assistance in mass spectrometry analysis. Valuable technical assistance was provided by N. Stokes, L. Polak, E. Wong, M. Nikolova, J. Racelis, A. North and S. Bhuvanendran. All mice used in this study were bred and maintained at the Rockefeller University AAALAC-accredited Comparative Biology Center (CBC) in accordance with institutional and NIH guidelines. This work was supported by a grant R01-AR27883 from the National Institutes of Health. E.F. is an investigator of the Howard Hughes Medical Institute. X.W. was an AACR Anna D. Barker Fellow in Basic Cancer Research and the recipient of a postdoctoral fellowship from the Jane Coffin Childs Memorial Fund for Medical Research.


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


  • Akhmanova A, Steinmetz MO. Tracking the ends: a dynamic protein network controls the fate of microtubule tips. Nat Rev Mol Cell Biol. 2008;9:309–322. [PubMed]
  • Blanpain C, Fuchs E. Epidermal homeostasis: a balancing act of stem cells in the skin. Nat Rev Mol Cell Biol. 2009;10:207–217. [PMC free article] [PubMed]
  • Blanpain C, Lowry WE, Geoghegan A, Polak L, Fuchs E. Self-renewal, multipotency, and the existence of two cell populations within an epithelial stem cell niche. Cell. 2004;118:635–648. [PubMed]
  • Chen HJ, Lin CM, Lin CS, Perez-Olle R, Leung CL, Liem RK. The role of microtubule actin cross-linking factor 1 (MACF1) in the Wnt signaling pathway. Genes Dev. 2006;20:1933–1945. [PubMed]
  • Etienne-Manneville S. Cdc42--the centre of polarity. J Cell Sci. 2004;117:1291–1300. [PubMed]
  • Etienne-Manneville S, Hall A. Cdc42 regulates GSK-3beta and adenomatous polyposis coli to control cell polarity. Nature. 2003;421:753–756. [PubMed]
  • Fathke C, Wilson L, Shah K, Kim B, Hocking A, Moon R, Isik F. Wnt signaling induces epithelial differentiation during cutaneous wound healing. BMC Cell Biol. 2006;7:4. [PMC free article] [PubMed]
  • Fuchs E. The tortoise and the hair: slow-cycling cells in the stem cell race. Cell. 2009;137:811–819. [PMC free article] [PubMed]
  • Goryunov D, He CZ, Lin CS, Leung CL, Liem RK. Nervous-tissue-specific elimination of microtubule-actin crosslinking factor 1a results in multiple developmental defects in the mouse brain. Mol Cell Neurosci. 2010;44:1–14. [PMC free article] [PubMed]
  • Greco V, Chen T, Rendl M, Schober M, Pasolli HA, Stokes N, Dela Cruz-Racelis J, Fuchs E. A two-step mechanism for stem cell activation during hair regeneration. Cell Stem Cell. 2009;4:155–169. [PMC free article] [PubMed]
  • Ito M, Liu Y, Yang Z, Nguyen J, Liang F, Morris RJ, Cotsarelis G. Stem cells in the hair follicle bulge contribute to wound repair but not to homeostasis of the epidermis. Nat Med. 2005;11:1351–1354. [PubMed]
  • Ito M, Yang Z, Andl T, Cui C, Kim N, Millar SE, Cotsarelis G. Wnt-dependent de novo hair follicle regeneration in adult mouse skin after wounding. Nature. 2007;447:316–320. [PubMed]
  • Jefferson JJ, Leung CL, Liem RK. Plakins: goliaths that link cell junctions and the cytoskeleton. Nat Rev Mol Cell Biol. 2004;5:542–553. [PubMed]
  • Karakesisoglou I, Yang Y, Fuchs E. An epidermal plakin that integrates actin and microtubule networks at cellular junctions. J Cell Biol. 2000;149:195–208. [PMC free article] [PubMed]
  • Knipling L, Hwang J, Wolff J. Preparation and properties of pure tubulin S. Cell Motil Cytoskeleton. 1999;43:63–71. [PubMed]
  • Kodama A, Karakesisoglou I, Wong E, Vaezi A, Fuchs E. ACF7: an essential integrator of microtubule dynamics. Cell. 2003;115:343–354. [PubMed]
  • Lauffenburger DA, Horwitz AF. Cell migration: a physically integrated molecular process. Cell. 1996;84:359–369. [PubMed]
  • Matov A, Applegate K, Kumar P, Thoma C, Krek W, Danuser G, Wittmann T. Analysis of microtubule dynamic instability using a plus-end growth marker. Nat Methods. 2010;7:761–768. [PMC free article] [PubMed]
  • Miller PM, Folkmann AW, Maia AR, Efimova N, Efimov A, Kaverina I. Golgi-derived CLASP-dependent microtubules control Golgi organization and polarized trafficking in motile cells. Nat Cell Biol. 2009;11:1069–1080. [PMC free article] [PubMed]
  • Morris RJ, Liu Y, Marles L, Yang Z, Trempus C, Li S, Lin JS, Sawicki JA, Cotsarelis G. Capturing and profiling adult hair follicle stem cells. Nat Biotechnol. 2004;22:411–417. [PubMed]
  • Nusse R. Wnt signaling and stem cell control. Cell Res. 2008;18:523–527. [PubMed]
  • Nusse R, Fuerer C, Ching W, Harnish K, Logan C, Zeng A, ten Berge D, Kalani Y. Wnt signaling and stem cell control. Cold Spring Harb Symp Quant Biol. 2008;73:59–66. [PubMed]
  • Obenauer JC, Cantley LC, Yaffe MB. Scansite 2.0: Proteome-wide prediction of cell signaling interactions using short sequence motifs. Nucleic Acids Res. 2003;31:3635–3641. [PMC free article] [PubMed]
  • Rodriguez OC, Schaefer AW, Mandato CA, Forscher P, Bement WM, Waterman-Storer CM. Conserved microtubule-actin interactions in cell movement and morphogenesis. Nat Cell Biol. 2003;5:599–609. [PubMed]
  • Roper K, Gregory SL, Brown NH. The ‘spectraplakins’: cytoskeletal giants with characteristics of both spectrin and plakin families. J Cell Sci. 2002;115:4215–4225. [PubMed]
  • Schiller MR. Curr Protoc Protein Sci. Unit 2. Chapter 2. 2007. Minimotif miner: a computational tool to investigate protein function, disease, and genetic diversity; p. 12. [PubMed]
  • Siegrist SE, Doe CQ. Microtubule-induced cortical cell polarity. Genes Dev. 2007;21:483–496. [PubMed]
  • Stoick-Cooper CL, Weidinger G, Riehle KJ, Hubbert C, Major MB, Fausto N, Moon RT. Distinct Wnt signaling pathways have opposing roles in appendage regeneration. Development. 2007;134:479–489. [PubMed]
  • Sun D, Leung CL, Liem RK. Characterization of the microtubule binding domain of microtubule actin crosslinking factor (MACF): identification of a novel group of microtubule associated proteins. J Cell Sci. 2001;114:161–172. [PubMed]
  • Sun T, Rodriguez M, Kim L. Glycogen synthase kinase 3 in the world of cell migration. Dev Growth Differ. 2009;51:735–742. [PubMed]
  • Tumbar T, Guasch G, Greco V, Blanpain C, Lowry WE, Rendl M, Fuchs E. Defining the epithelial stem cell niche in skin. Science. 2004;303:359–363. [PMC free article] [PubMed]
  • Wu X, Kodama A, Fuchs E. ACF7 regulates cytoskeletal-focal adhesion dynamics and migration and has ATPase activity. Cell. 2008;135:137–148. [PMC free article] [PubMed]
  • Wu X, Suetsugu S, Cooper LA, Takenawa T, Guan JL. Focal adhesion kinase regulation of N-WASP subcellular localization and function. J Biol Chem. 2004;279:9565–9576. [PubMed]
  • Yoshimura T, Arimura N, Kaibuchi K. Signaling networks in neuronal polarization. J Neurosci. 2006;26:10626–10630. [PubMed]
  • Zaoui K, Benseddik K, Daou P, Salaun D, Badache A. ErbB2 receptor controls microtubule capture by recruiting ACF7 to the plasma membrane of migrating cells. Proc Natl Acad Sci U S A. 2010;107:18517–18522. [PubMed]