Integrin-linked kinase is required for radial sorting of axons and Schwann cell remyelination in the peripheral nervous system

Abstract

During development, Schwann cells (SCs) interpret different extracellular cues to regulate their migration, proliferation, and the remarkable morphological changes associated with the sorting, ensheathment, and myelination of axons. Although interactions between extracellular matrix proteins and integrins are critical to some of these processes, the downstream signaling pathways they control are still poorly understood. Integrin-linked kinase (ILK) is a focal adhesion protein that associates with multiple binding partners to link integrins to the actin cytoskeleton and is thought to participate in integrin and growth factor-mediated signaling. Using SC-specific gene ablation, we report essential functions for ILK in radial sorting of axon bundles and in remyelination in the peripheral nervous system. Our in vivo and in vitro experiments show that ILK negatively regulates Rho/Rho kinase signaling to promote SC process extension and to initiate radial sorting. ILK also facilitates axon remyelination, likely by promoting the activation of downstream molecules such as AKT/protein kinase B.

Figure 1.

Recombination of the conditional ILK allele in SCs of mutant mice. (A) Regulatory sequences of the Dhh promoter drive the expression of the Cre recombinase (Cre Rec.) in SCs. Partial map of the ILK allele depicts the location of the loxP sequences. Upon Cre-mediated recombination, the genomic region located between the two loxP sites, which includes exon 2 and part of exon 1, is excised, thereby inactivating the conditional ILK allele. (B) Western blot analysis of protein lysates obtained from P1 SNs reveals a significant reduction in ILK levels compared with those of controls (Ctrl; n = 3, P < 0.0001). The low residual ILK protein levels detected in mutant (Mut) lysates are likely a result of the presence of endoneurial fibroblasts and some unrecombined SCs. Error bar indicates ± SEM. ***, P < 0.001. (C) Immunostaining of acutely plated SCs obtained from P1 control and mutant mice using antibodies for ILK and S100 (SC marker) reveals loss of ILK in mutant SCs. (D) Immunohistochemistry in E17.5 SN cryosections demonstrates loss of ILK in the mutant SCs in vivo. (E) Control and mutant SNs from P14 littermates. Note that mutant mice have thinner and more transparent SNs. (F) Upon conditional ablation of ILK, other protein members of the IPP complex (PINCH1 and 2 and parvin-α and -β) are reduced in P5 mutant SN lysates compared with controls. Bars, 50 µm.

Figure 2.

Axonal sorting and myelination are impaired in ILK mutant SNs. (a–l) 0.5-µm SN cross sections stained with Toluidine blue. Early in development (a–f), SCs are in close association with bundles of naked axons (c–f; asterisks). (c and d) Progressively individual axons are sorted from these bundles in a process referred to as radial sorting (arrowheads). (g, i, and k) In control nerves, individual large caliber axons are wrapped and myelinated. (h, j, and l) In the mutant nerves, radial sorting and myelination are impaired, and axon bundles persist into adulthood. Bars, 10 µm.

Figure 3.

Process extension and BL organization are impaired in ILK mutant SCs. (A and B) EM micrographs of SN ultra-thin sections at P5 (A) and P24 (B). (A) At P5, immature SCs extend processes (white arrowheads) that fully envelop bundles of axons (a and c), which is a typical feature at this stage of maturation. In mutant nerves (b and d), SC processes often fail to completely enwrap axon bundles (black arrowheads). In controls, promyelinating SCs are surrounded by a tightly associated BL (e, white arrows). In contrast, mutant promyelinating SCs display cytoplasmic protrusions at the cell surface (f, gray arrows), frequently associated with a disorganized BL that forms loops detached from the plasma membrane (f, black arrows). Myelinating SCs could be found in controls, and some were also present in mutant nerves (asterisks). Bars: (b) 5 µm; (a and d) 2 µm; (c, e, f) 1 µm. (B) At P24, there are virtually no remaining axon bundles on the control nerves. Large caliber axons have been engaged and are myelinated (a, c, and e), whereas small caliber axons, which will not be myelinated during development, are engaged by nonmyelinating SCs (c, white arrowhead). In mutant nerves, axonal sorting is severely impaired, and many bundles persist. Mutant SCs associated with axon bundles fail to completely surround them (black arrowheads), resembling the phenotype observed at P5 (b and d). Unsorted axons are surrounded by loops of BL (black arrowheads). Control myelinating SCs reveal a tightly associated single BL (g, white arrows) in contrast to ILK-null cells in which multiple cytoplasmic protrusions (gray arrows) are visible (f), which are frequently associated with empty loops of BL (h; black arrows). Boxes indicate regions that are shown at a higher magnification in the panels below. Bars: (b) 5 µm; (a, c, d, and f) 2 µm; (e and h) 1 µm; (g) 0.2 µm.

Figure 4.

ILK regulates SC process extension in a Rho/ROCK-dependent manner. (A) Pull-down assays for active GTP-bound Rho GTPases using SN lysates obtained from P5 control and mutant mice. The activity of Cdc42 (n = 4, P = 0.0096) and Rac1 (n = 4, P = 0.0012) is significantly increased in mutant lysates, whereas their total expression levels are not. The amounts of activated Rac1 are more abundant in mutant than control SNs (n = 4, P = 0.0017), and although not statistically significant, the amounts of activated Cdc42 are also increased (n = 4, P = 0.0511). The aggregate activation of RhoA, B, and C (n = 4, P = 0.0015), termed Rho, is significantly increased in mutant lysates, whereas their total expression is significantly decreased (n = 4, P = 0.0017). Altogether, there is still more activated Rho protein in mutant than control SNs (n = 4, P = 0.0050). The Rho antibody used for Western blotting, after the coprecipitation with the recombinant rhotekin fragment, is not specific for individual Rho isoforms. (B, a and b, d and e, and g and h) Immunocytochemistry to reveal the cytoskeleton of control and mutant mouse SCs plated on laminin-2 for 24 h. Categorization of control and mutant SCs according to process length. (a–c) There is a higher frequency of SCs with shorter processes in mutant SC cultures as compared with control cultures (n = 3; P_26–50 = 0.026; P_51–75 = 0.016; P_76–100 = 0.0071; P_101–150 = 0.036; P_151–200 = 0.020; P_201–250 = 0.0048). Treatment of the SCs with the ROCK inhibitor Y27632 (d–f) or fasudil (g–i) restores process extension in mutant cells. Error bars indicate ± SEM. *, P < 0.05; **, P < 0.01. Bars, 20 µm.

Figure 5.

Inhibition of ROCK signaling facilitates axonal sorting in ILK mutant mice. ILK conditional control and mutant mice were injected i.p. with either PBS or 40 mg/kg/d fasudil over six consecutive days from P8 to P13. SNs were collected at P14. (A, a–d) Toluidine blue–stained 0.5-µm semithin sections of control and mutant animals injected with either fasudil or with PBS only. (a and b) There were no visible differences in myelination between fasudil- and PBS-injected control nerves. Bars, 10 µm. (B, a–d) Electron micrographs of mutant animals injected with either fasudil or with PBS only. PBS-injected mutant mice display large unsorted axonal bundles (a and c) as seen before, whereas bundles in fasudil-injected mutant mice are smaller (b and d; n = 3; P_0–5 = 0.012; P_6–10 = 0.0064; P_51–100 = 0.011; P_101–250 = 0.027). (e) Quantification of the unsorted bundle area reveals a higher frequency of small bundles in the fasudil-injected mice (6–25 µm²) and a higher frequency of large bundles in PBS-injected mice (51 to >250 µm²). Crosses indicate that no bundles within the size category were found in any of the animals analyzed. In the axonal bundle depicted in d, asterisks mark axons that have presumably been recently sorted and are larger than those remaining in the bundle. (f and g) There is a significant increase in the number of 1:1 relations in fasudil- versus PBS-injected mutant animals (f; n = 3, P = 0.0052); however, the total number of myelinated axons in the nerves analyzed does not differ between PBS- and fasudil-injected mice (g). Error bars indicate ± SEM. *, P < 0.05; **, P < 0.01. Boxes in a and b indicate regions that are shown at a higher magnification in c and d, respectively. Bars: (a and b) 5 µm; (c) 2 µm; (d) 1 µm.

Figure 6.

ILK regulates PKB/AKT phosphorylation. (A) AKT phosphorylation on Ser-473 is reduced in P1, P5, and P14 mutant nerves (n = 4, P_P1 = 0.00088; n = 4, P_P5 = 0.00090; n = 5, P_P14 < 0.0001) unlike phosphorylation on Thr-308, which is higher in the mutants (n = 6; P_P1 = 0.0018; P_P5 = 0.0062; P_P14 = 0.010). AKT total levels are slightly reduced in the mutants (n = 10, P_P1 = 0.027; n = 10, P_P5 = 0.010; n = 11, P_P14 = 0.019). (B) GSK3β phosphorylation levels are not reduced in the absence of ILK either at P1 or P5. Error bars indicate ± SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 7.

ILK regulates PNS remyelination. (A) Cre-mediated recombination was induced in 2-mo-old control and mutant mice by five daily consecutive tamoxifen injections. A period of 2 mo was allowed for recombination and protein depletion before assays were performed. Cre Rec., Cre recombinase; 2MPC, 2 mo post-CI; 4MPC, 4 mo post-CI. (B) Western blots performed on protein lysates obtained from contralateral and crushed nerves show a reduction in ILK levels in mutant (mut) nerves compared with those of controls (cont). PINCH1 and AKT phosphorylation levels are reduced in mutant nerves compared with controls. (C) Semithin sections of SNs collected 2 and 4 mo after CI. In contrast to controls (a, c, and g), mutant nerves show a significant reduction in the number of remyelinated axons at 2 and 4 mo after CI (b, d, and g; n = 3, P < 0.0001). In d, the low magnification EM insert shows that in mutant SNs, virtually all axons are engaged by SCs (arrowheads) but only very few went on to myelinate (arrow). (e) Tamoxifen (TAM) treatment does not affect myelination. (f) Noninjured mutant nerves collected 6 mo after tamoxifen induction show no structural abnormalities, implying that ILK is not necessary for the maintenance of the myelin membrane. (h) There was no difference in the axonal number in mutant crushed nerves when compared with control crushed nerves. (D) Oct6 immunohistochemistry on 2-mo post-CI cryosections. Oct6 is virtually absent in control nerves, whereas levels are still high in mutant nerves. Error bars indicate ± SEM. ***, P < 0.001. Bars: (C) 10 µm; (D) 50 µm.

Figure 8.

Proposed mechanism for the regulation of SC development by ILK. ILK plays distinct fundamental roles in the regulation of SC maturation. Early in development, ILK negatively regulates Rho/ROCK activity to trigger radial sorting. The phosphorylation of AKT on Ser-473 occurs in an ILK-dependent manner. On an adult animal after a nerve CI, ILK is also required for the transition from a promyelinating to a myelinating SC during remyelination. AKT activity is likely to also play an important role in this final differentiation step.