β1 integrins are required for normal CNS myelination and promote AKT-dependent myelin outgrowth

In the vertebrate nervous system, myelin sheaths insulate axons and limit membrane depolarization to the nodes of Ranvier, where the machinery propagating action potentials is concentrated(Sherman and Brophy, 2005). Integrins, a family of heterodimeric transmembrane receptors consisting ofα and β subunits, are thought to be important for myelination. Integrins containing the β1 subunit are expressed in Schwann cells, the myelinating glial cells of the peripheral nervous system, where they mediate interactions with laminin that are crucial for myelination(Berti et al., 2006). Integrins are also expressed in oligodendrocytes(Milner and ffrench-Constant,1994; Milner et al.,1997), the myelinating glial cells of the CNS, but less is known about their function in these cells. In vitro studies show that the integrinαvβ1 promotes migration of oligodendrocyte progenitors, whereasα6β1 regulates oligodendrocyte survival and myelin membrane formation in response to laminin 2 (Ln2; Lama2 - Mouse Genome Informatics)(Buttery and ffrench-Constant,1999; Corley et al.,2001; Frost et al.,1999; Milner et al.,1996; Relvas et al.,2001; Tiwari-Woodruff et al.,2001). Consistent with the in vitro studies, Ln2 is expressed in premyelinating axonal tracts in the CNS, and oligodendrocyte death is increased in the brainstem of mice lacking integrin α6(Chun et al., 2003; Colognato et al., 2002; Farina et al., 1998; Jones et al., 2001). However,since α6-deficient mice die at birth, it has remained unclear whether the defects in oligodendrocyte survival lead to persistent changes in myelination. In fact, genetic studies addressing the function of β1 integrins in oligodendrocytes have led to contradictory results. Whereas oligodendrocyte-specific expression of a dominant-negative (DN) β1 integrin in a transgenic mouse model was associated with CNS hypomyelination(Camara et al., 2009; Lee et al., 2006), conditional ablation of a floxed allele of the β1 gene (Itgb1) in oligodendrocytes using a CNPase-Cre (Cnp-Cre - Mouse Genome Informatics) driver caused no myelination defects(Benninger et al., 2006). The interpretation of these experiments is complex, as DN β1 integrins can also ectopically activate integrin signaling(Lukashev et al., 1994), and CNPase-Cre induces recombination in only a subset of oligodendrocytes by P0 (Benninger et al.,2006).

To circumvent the complications associated with the previous studies, we have analyzed mice in which β1 integrins were inactivated in the ventricular neuroepithelium, including in the oligodendrocyte precursors(referred to here as Itgb1CNS-ko mice). We show that in the absenceβ1 integrins, the thickness of myelin sheets was reduced in axonal tracts in the spinal cord, optic nerves and cerebellum. Defects in myelination were also observed in the spinal cord from mice in which β1 was inactivated solely in NG2 (CSPG4 - Mouse Genome Informatics)-positive oligodendrocyte precursors. Myelination defects were not caused by defects in oligodendrocyte differentiation or survival. Instead, studies with cultured oligodendrocytes show that β1 integrins regulate the outgrowth of myelin sheets. Interestingly, it has recently been shown that overexpression of a constitutively active form of the serine/threonine kinase AKT (AKT1 - Mouse Genome Informatics) in transgenic mice leads to enhanced myelin sheath formation in the CNS without affecting oligodendrocyte proliferation or survival (Flores et al.,2008). We show here that β1 integrins regulate AKT activity in oligodendrocytes, and demonstrate that constitutively active AKT rescues the defect in myelin sheet outgrowth in cultured β1 integrin-deficient oligodendrocytes. Our findings define the role of β1 integrins in CNS myelination and establish a link between β1 integrins and AKT in the control of CNS axonal ensheathment.

All mouse lines have been previously described(Graus-Porta et al., 2001; Stephens et al., 1995; Zhu et al., 2008). Western blots were carried out as described(Colognato et al., 2002; Graus-Porta et al., 2001). Densitometry and quantification of relative levels of protein expression were performed on scanned images of western blots using MetaMorph software(Universal Imaging). Antibodies used were β1 integrin(Graus-Porta et al., 2001),MBP (Boehringer), PLP (Serotec), actin (Sigma), AKT phospho-S473 and AKT (Cell Signaling).

Oligodendrocyte progenitors were isolated and cultured following published procedures (Colognato et al.,2007; Colognato et al.,2004), with the addition of a panning step to remove microglia(Cahoy et al., 2008). Briefly,oligodendrocyte progenitor cells (OPCs) were isolated from 10-days-in-vitro(DIV) mixed glial cultures using a 20-hour mechanical dissociation followed by a 30-minute differential adhesion step with uncoated petri dishes. Unattached cells were subjected to two rounds of panning on lectin-coated dishes to remove remaining microglia. Lectin-coated dishes were prepared by incubating petri dishes with 2.3 μg/ml BS-lectin-1 in Dulbecco's PBS (D-PBS) for 4 hours at room temperature, followed by four washes with D-PBS immediately prior to use (Cahoy et al.,2008). Purified OPCs were added to poly-D-lysine (PDL)- or laminin-coated Permanox chamber slides in a modified SATO's medium (SATO with 0.5% FCS for differentiation experiments). Human placental laminin (laminin 211, Sigma) was used to coat the surfaces of slides and dishes at 10 μg/ml in PBS for 4 hours at 37°C. Surface coating with PDL (Sigma) was performed similarly but was instead diluted in dH₂O to obtain 10 μg/ml. Mixed glial cultures from neonatal spinal cords were grown on PDL-coated chamber slides and switched to SATO with 0.5% FCS to differentiate at 10 DIV. Recombinant protein comprising the EGF-like domain of neuregulin 1 was used at 100 ng/ml (PeproTech). The PTEN inhibitor bpV(pic) was used at 31 nM, the IC₅₀ for PTEN inhibition (Calbiochem)(Schmid et al., 2004). Wild-type and β1-CNS-mutant mixed glial cultures were transfected using FuGENE (Roche) as described (Colognato et al., 2004), with either MYR-AKT-EGFP or control EGFP plasmids (a kind gift from Dr Bing-Hua Jiang, West Virginia University, Morgantown, WV,USA). Transfected oligodendrocytes were differentiated within the mixed glial cultures and identified 6 days post-transfection using combined GFP and/or MBP immunocytochemistry (see below). A minimum of 25 fields per transfection were evaluated by morphometric analysis.

Electron microscopy (EM) samples were prepared as described(Graus-Porta et al., 2001). EM images were acquired using a Philips CM100 microscope (FEI, Hillsbrough, OR,USA). Morphometric analysis of nerve fibers was performed using MetaMorph software. Immunohistochemistry followed our published procedures(Belvindrah et al., 2007; Blaess et al., 2004). Primary antibodies used were PDGFαR (a kind gift from W. Stallcup, Burnham Institute, La Jolla, CA, USA), CC1 (Calbiotech), OLIG2 (Chemicon) and cleaved caspase 3 (Cell Signaling). Secondary antibodies were Alexa Fluor 488 and 568-conjugated (Molecular Probes). Nuclei were stained with TO-PRO-3(Molecular Probes). TUNEL assays were carried using the ApopTag Red In Situ Apoptosis Detection Kit (Chemicon). For immunocytochemistry, cells were fixed in 4% PFA (15 minutes) or 100% methanol (5 minutes, -20°C). Blocking and primary antibody incubations were in PBS with 10% donkey serum (with 0.1%Triton X-100 for PFA-fixed cells). Cells undergoing GFP/MBP immunocytochemistry were blocked in PBS with 10% donkey serum and 0.05% Triton X-100. For GALC staining, live cells were labeled with antibody for 30 minutes in DMEM, 1% FCS, followed by washes and PFA fixation. Primary antibodies were NG2 (Chemicon), β1 integrin (Chemicon, MAB1997), GALC (Sigma), CNP(Sigma), MBP (Serotec), GFP (Molecular Probes) and OLIG2 (IBL). Secondary antibodies were FITC- or Texas Red-conjugated (Jackson). Nuclei were stained with DAPI. Images were collected on an Olympus Fluoview 500 confocal microscope, an Olympus AX70 microscope, and a Zeiss Axioplan microscope. MetaMorph software (Universal Imaging) was used for morphometric measures. The AxioVision Interactive Measurement Module was used for myelin sheet and process length cell measures. To ensure consistent analysis across different experiments, MBP-positive cells and MBP-positive sheet-bearing cells were determined using intensity thresholding relative to background intensity(AxioVision). Morphometric analysis to determine the area of coverage of myelin sheets was performed by tracing the outer perimeter of thresholded MBP-positive sheet-bearing oligodendrocytes (AxioVision, Zeiss). Process length measurements were performed on corresponding phase micrographs by tracing from the cell body/process border to the tip of the longest process(AxioVision). Examples of morphometric tracing can be observed in Fig. 7F,G.

Statistical analysis was performed using GraphPad Prism. Data show mean± standard error (mean±s.e.m.). Statistical significance for g-ratio analysis was performed using the Mann-Whitney (rank sum) test. The number of animals analyzed (n) was three unless otherwise indicated. Student's t-test was used for all other assays described. Statistical significance was set at P<0.05 (^*, significant), P<0.01 (^**, very significant) and P<0.001(^***, highly significant).

To define the functions of β1 integrins in CNS myelination, we inactivated β1 expression by crossing mice carrying a floxed integrinβ1 gene (Itgb1-flox) with nestin-Cre mice. In this nestin-Cre line, CRE-mediated recombination occurs in neural precursors in the ventricular zone as early as embryonic day (E) 10.5, leading to the inactivation of floxed alleles in developing neurons and glia,including cells at all stages of oligodendrocyte development(Graus-Porta et al., 2001). The mutant mice are viable and fertile and show no obvious behavioral abnormalities (Graus-Porta et al.,2001), and will be referred to as Itgb1-CNSko mice(Fig. 1A). We confirmed recombination of the Itgb1-flox allele by PCR and the absence ofβ1 protein by immunoblotting in the spinal cords and brains of mutant mice (Fig. 1B,C; C.S.B.,unpublished) (Graus-Porta et al.,2001). In all assays, Itgb1-CNSko mice were compared with littermates that did not express CRE but were homozygous for the Itgb1-flox allele or carried one Itgb1-flox and one Itgb1-null allele. Control mice are referred to as wild type (WT)because previous studies revealed that the loxP sites or heterozygosity for Itgb1-null do not affect β1 function(Graus-Porta et al., 2001; Stephens et al., 1995). Histological analysis showed that spinal cords from Itgb1-CNSko adult mice were not obviously different in overall morphology from WT(Fig. 1D).

To examine myelin sheaths in Itgb1-CNSko mice, we analyzed spinal cords by electron microscopy (EM). The amount of myelin sheath insulating axons increases in proportion to the axon diameter(Sherman and Brophy, 2005). In Itgb1-CNSko mutants, the thickness of myelin surrounding axons was reduced compared with WT (Fig. 2A). Morphometric analysis of the nerve fibers revealed a significant increase in the ratio of axon diameter to fiber diameter (g-ratio; Fig. 2B; see Table S1 in the supplementary material). Interestingly, the smallest caliber axons (of <0.7μm diameter) did not show decreases in myelin thickness(Fig. 2B; see Table S1 in the supplementary material). The periodicity of myelin sheets in the mutants was unaffected, suggesting that the overall reduction in myelin thickness resulted from a decrease in the number of wraps (insets in Fig. 2A, lower panels). We also observed no axonal swelling or degeneration, and no changes in the density or size of axons (Fig. 2A,C,D). Consistent with myelination defects, the levels of two major CNS myelin components, myelin basic protein (MBP) and proteolipid protein (PLP, PLP1 -Mouse Genome Informatics) (Simons and Trajkovic, 2006), were modestly but significantly reduced in Itgb1-CNSko mice (Fig. 2E).

To investigate whether other axonal tracts in the CNS were similarly affected, we analyzed the optic nerves and the cerebellum. As in the spinal cord, myelin sheaths enwrapping the larger caliber axons were reduced in thickness (Fig. 3A,B; see Table S1 in the supplementary material). We also attempted to analyze myelination in the corpus callosum, but its organization was significantly disrupted in Itgb1-CNSko mice due to the perturbations in cortical size and structure (Graus-Porta et al.,2001), preventing a meaningful analysis.

To determine the extent to which β1 integrins in oligodendroglia are required for myelination, we inactivated their expression using Ng2-Cre mice. In this mouse line, CRE specifically induces the recombination of floxed alleles in NG2-positive oligodendrocyte precursors(Zhu et al., 2008). The analysis of axonal tracts in the spinal cord of these mutants (named Itgb1-OL-ko mice) revealed a reduction in myelin thickness(Fig. 4A,B). As in Itgb1-CNSko mice, myelin thickness was affected in axons with diameters >0.7 μm, although the phenotype was less severe, and some of the largest diameter axons appeared unaffected(Fig. 4C). The less severe defect in myelination could be a consequence of less effective inactivation ofβ1 integrins by Ng2-Cre compared with nestin-Cre. To test this hypothesis, we purified oligodendrocytes from mutant mice and analyzed integrin expression by immunohistochemistry and western blotting. Staining for NG2 and OLIG2 confirmed that oligodendrocytes were purified to near homogeneity (see Fig. S1A-C in the supplementary material). Whereasβ1 integrins were almost completely absent in oligodendrocytes from Itgb1-CNSko mice (Fig. 6A,B,I), a low amount of β1 integrin expression could still be detected in oligodendrocytes from Itgb1-OL-ko mice (see Fig. S1D,E in the supplementary material), suggesting that the less severe myelination defect in Itgb1-OL-ko mice was a consequence of less efficient CRE-mediated recombination. However, it cannot be excluded that β1 integrins in axons, which were inactivated by nestin-Cre but not by Ng2-Cre, also contributed to myelination. Taken together, our findings demonstrate that β1 integrins are required for the myelination of axonal tracts in the spinal cord, optic nerve and cerebellum, and that they act, at least in part, in oligodendrocytes to carry out their function.

The hypomyelination of axonal tracts in the CNS of Itgb1-CNSkomutants could arise from defects in oligodendrocyte development. We therefore analyzed their development using stage-specific molecular markers. The number of PDGFαR-positive oligodendrocyte progenitors at postnatal day (P) 0,prior to axonal ensheathment, was not reduced in the mutants(Fig. 5A,B,G;β 1mt 17.88±1.09/μm², WT 17.95±1.21/μm², P>0.05, n=3). Similarly, there was no reduction in the numbers of CC1-positive oligodendrocytes at P19, when myelination is in progress(Fig. 5C,D,H;β 1mt 32.71±0.89/μm², WT 33.5±1.36/μm², P>0.05, n=3), or at P60, after myelin sheaths have formed (Fig. 5E,F,H; β1mt 34.17±1.05/μm², WT 34.89±0.78/μm², P>0.05, n=3). Detection of OLIG2, which labels oligodendroglia throughout their development,also revealed no defects in mutant animals (see Fig. S2 in the supplementary material). Finally, we analyzed cell death in the mutants and observed no differences in either the developing or adult spinal cords (see Fig. S3 in the supplementary material). Analysis of the developing cerebellum also revealed no significant increase in the numbers of dying cells (see Fig. S4 in the supplementary material). We conclude that the reduction in myelin ensheathment in Itb1-CNSko mice is probably not caused by the loss of oligodendrocytes.

To further define the cell-autonomous function of β1 integrins in oligodendrocytes, we purified their progenitors from WT and Itgb1-CNSko mice to near homogeneity (see Fig. S1A-C in the supplementary material) and followed their differentiation in vitro(Fig. 6). Immunostaining and western blots confirmed the efficient inactivation of β1 integrins in oligodendrocytes from Itgb1-CNSko mice(Fig. 6A,B,I). The percentage of NG2-positive oligodendrocyte precursors that were present 1 day after plating into differentiation medium was not significantly different between WT and β1-deficient cells (Fig. 6C,D,J; β1mt 46.83±3.02%, WT 50.90±5.32%, n=4, P>0.05). No difference was observed in the percentage of GALC-positive oligodendrocytes at days 2 or 4(Fig. 6E,F,K;β 1mt 67.05±2.1%, WT 66.89±3.72%, n=4, P>0.05 at day 2; β1mt 81.6±2.82%, WT 78.87±4.29%, n=4, P>0.05 at day 4), or MBP-positive mature oligodendrocytes at 4 or 6 days postplating in differentiation medium (Fig. 6G,H,L; β1mt 33.26±4.37%, WT 37.39±4.54%, n=4, P>0.05 at day 4;β 1mt 23.28±2.95%, WT 32.97±4.22%, n=4, P>0.05 at day 6). These data are in agreement with our in vivo analysis (Fig. 5; see Fig. S2 in the supplementary material) and confirm that β1 integrins are not essential for the formation of mature oligodendrocytes.

Myelin membrane sheets are formed by the extension of branched processes surrounded by a specialized plasma membrane enriched with myelin proteins such as MBP (Richardson et al.,2006). Myelin membrane sheet formation is therefore thought to mimic many of the morphological changes that take place during oligodendrocyte ensheathment of axons (Simons and Trotter,2007). While performing lineage analysis, we observed that MBP-positive oligodendrocytes derived from the cerebral cortex of Itgb1-CNSko mutants extended smaller myelin membrane sheets(Fig. 6G,H and Fig. 7A). Morphometric analysis further revealed that fewer mutant oligodendrocytes elaborated myelin membrane sheets (Fig. 7B;β 1mt 29.98±1.81%, WT 36.62±0.23%, n=4, P<0.05 at day 2), and that the β1-deficient sheets were significantly smaller than their WT counterparts(Fig. 7C; β1mt2409.12±225.79 μm², WT 4164.7±311.60μm², n=4, P<0.05 at day 2;β 1mt 3658.04±457.05 μm², WT 5557.27±806.51 μm², n=4, P<0.05 at day 4). The decrease in sheet outgrowth was also reflected by a decrease in the mean length of mutant oligodendrocyte processes(Fig. 7D; β1mt85.01±5.59 μm, WT 115.16±3.16 μm, n=4, P<0.05 at day 4).

To confirm that defects in the formation of myelin sheets were not confined to oligodendrocytes from the cerebral cortex, we evaluated mixed glial cultures from the spinal cord. Examples for the morphometric analysis to determine MBP sheet area and process length are shown in Fig. 7F,G. Spinal cord oligodendrocytes from WT and mutant mice on average elaborated larger membrane sheets than those prepared from cerebral cortices. However, sheets formed byβ1-deficient spinal cord oligodendrocytes were significantly smaller than those from WT controls (Fig. 7E; β1mt 9863.98±719.86 μm², WT 12,944.54±346.64 μm², n=4, P<0.05).

Several signaling molecules have been implicated in controlling myelination, including the serine/threonine kinase AKT(Flores et al., 2008). To determine whether β1-deficient oligodendrocytes exhibited defects in AKT activation, we plated oligodendrocytes on laminin and treated the cultures with neuregulin 1 to stimulate AKT phosphorylation(Fig. 8A). We made use of neuregulin 1 because integrin signaling in oligodendrocytes amplifies receptor tyrosine kinase signaling, where neuregulin 1 signaling is particularly sensitive to integrin engagement (Colognato et al., 2002; Colognato et al., 2004). Whereas WT oligodendrocytes showed a large increase in AKT phosphorylation at serine 473 (211±48.29% compared with untreated WT, n=8, P<0.05), β1-deficient oligodendrocytes did not show significant AKT activation (105.78±6.15% compared with untreated β1mt, n=8, P>0.05). Oligodendrocytes grown on control poly-D-lysine substrates showed low levels of AKT phosphorylation that were not significantly different between WT and mutants (see Fig. S5A in the supplementary material). In addition, a similar AKT stimulation assay was performed using β1-integrin-blocking antibodies on wild-type rat oligodendrocytes grown on laminin (Fig. 8B). Oligodendrocyte progenitor cells were attached for 2 hours,followed by differentiation for 22 hours in the presence ofβ1-integrin-blocking antibodies. Cells were treated with neuregulin for 30 minutes and AKT phosphorylation was evaluated(Fig. 8B). Neuregulin-treated oligodendrocytes showed a significant increase in AKT phosphorylation in the presence of the control antibody (183.52±12.03%, n=4, P<0.01), whereas cells treated with β1-integrin-blocking antibodies showed no significant increase (117.54±29.87%, n=4). To ascertain β1 integrin specificity, we also used antibodies that block the non-integrin laminin receptor dystroglycan, which is expressed in oligodendrocytes (Colognato et al., 2007). In the presence of the dystroglycan antibody, AKT phosphorylation was still significantly activated by neuregulin 1(150.65±10.25%, n=4, P<0.01). Thus, loss ofβ1 integrin protein (Fig. 8A) and function (Fig. 8B) affects the ability of oligodendrocytes to activate signaling pathways leading to AKT phosphorylation.

We next sought to investigate whether the defects in myelin sheet outgrowth in β1-deficient oligodendrocytes could be attenuated upon increased activation of AKT. Strikingly, treatment with a PTEN inhibitor (bpV) to enhance AKT phosphorylation caused mutant oligodendrocytes to grow myelin membrane sheets of an equivalent size to WT(Fig. 8C; β1mt +bpV 5026.72±90.34 μm² versus β1mt untreated 3612.06±97.31 μm², n=4, P>0.05;β 1mt + bpV 5026.72±90.34 μm² versus WT 4800.85±312.75 μm², n=4, P<0.05). WT myelin membrane sheets, however, did not become significantly larger in the presence of the PTEN inhibitor (see Fig. S5B in the supplementary material). To determine more directly whether AKT signaling can rescue the defect in myelin sheets in β1-deficient oligodendrocytes, we transfected them with a construct expressing a constitutively active AKT-GFP fusion protein (CA-AKT)or with control GFP (Fig. 8D). Transfected oligodendrocytes were visualized by staining for GFP and MBP and the membrane sheet area was determined. The expression of CA-AKT dramatically increased the myelin sheet area in β1 mutant oligodendrocytes and restored it to near wild-type levels (Fig. 8D; β1mt + control GFP 2174.65±280.15μm² versus β1mt + CA-AKT 3449.05±274.46μm²; WT + CA-AKT 3783.54±69.66 μm²; n=2, P<0.01). Collectively, these results suggest thatβ1 integrins are required for the regulation of AKT activity, and that defects in this process contribute to the abnormal myelin outgrowth observed in β1-deficient oligodendrocytes.

We show here that β1 integrins are required for the myelination of axonal tracts in the CNS and we establish a link between β1 integrins and AKT in oligodendrocyte function. Myelin thickness was reduced in the spinal cord, cerebellum and optic nerve of Itgb1-CNSko mice without a reduction in oligodendrocyte numbers. These findings suggest that the myelination defects in mutant mice were caused by perturbations in the formation of myelin membrane sheaths. Consistent with this finding, myelin outgrowth was substantially impaired in cultured β1-deficient oligodendrocytes. Myelination was also affected in the spinal cord of Itgb1-OL-ko mice, providing additional evidence that β1 integrins act, at least in part, cell-autonomously in oligodendrocytes to regulate myelination. Interestingly, activation of AKT signaling was affected in cultured β1-deficient oligodendrocytes. Furthermore, myelin membrane sheet formation in the β1-mutant cultured cells was restored by inhibiting PTEN or by overexpressing constitutively active AKT. Taken together, these findings provide strong evidence that AKT is crucial forβ1 integrin function during the myelination of axonal tracts in the CNS.

Previous studies have provided conflicting results regarding the function of β1 integrins in CNS myelination. Although the expression of a dominant-negative β1 integrin in oligodendrocytes has been reported to affect myelination (Camara et al.,2009; Lee et al.,2006), no such defect was observed in mice following β1 integrin inactivation in oligodendrocytes using CNPase-Cre(Benninger et al., 2006). We now report significant defects in CNS myelination in mice when β1 integrins are inactivated with nestin-Cre, supporting the view thatβ1 integrins are important for CNS myelination. In addition, we observed defects in myelination in Itgb1-OL-ko mice, in which integrins have been inactivated in oligodendrocytes using Ng2-Cre. We consider it likely that the differences in the results reported here and the earlier study(Benninger et al., 2006) can be explained by differences in the efficiency and timing of CRE expression. CNP is expressed relatively late during oligodendrocyte differentiation(Scherer et al., 1994), and CNP-CRE induces recombination in ∼65% of oligodendrocytes at around P0(Benninger et al., 2006). By contrast, NG2 is already expressed in oligodendrocyte precursors at the stage when PDGFαR is expressed (Nishiyama et al., 1996), which is a marker for the earliest stages of oligodendrocyte differentiation. NG2-CRE has also been reported to lead to recombination in approximately 90% of NG2-positive cells(Zhu et al., 2008). Since oligodendrocytes are generated in excess(Raff et al., 1998),sufficient progenitors probably escaped gene inactivation and were able to compensate for the loss of β1 integrin protein in some progenitors. Interestingly, myelination defects in Itgb1-CNSko mice were more severe than in Itgb1-OL-ko mice. As β1 integrins in Itgb1-CNSko mice were inactivated in both neurons and oligodendrocytes, β1 integrins in axons might have additional roles in myelination.

We provide here insights into the mechanisms by which β1 integrins regulate myelination in the CNS. Unlike in previous studies, which indicated that β1 integrins regulate oligodendrocyte survival(Benninger et al., 2006; Colognato et al., 2002; Lee et al., 2006), we only detected a small trend towards increased death in the developing cerebellum of Itgb1-CNSko mice. Although we cannot fully explain the difference with previous studies, our findings suggest that the subtle changes in oligodendrocyte death in Itgb1-CNSko mice were compensated for during development. The changes in myelin thickness that we observed are therefore probably not caused by a decrease in the number of oligodendrocytes but by defects in myelin membrane outgrowth. This interpretation is consistent with our in vitro data, which demonstrate that the formation of myelin membrane sheets was affected in β1-deficient cultured oligodendrocytes.

Interestingly, recent findings show that constitutively active AKT enhances myelination without affecting the number of oligodendrocytes(Flores et al., 2008). Mice lacking the P85α regulatory subunit (PIK3R1 - Mouse Genome Informatics)of PI3K, an activator of AKT, show hypomyelination in the CNS(Tohda et al., 2007), whereas knockout mice for PTEN, a negative regulator of PI3K/AKT signaling(Sulis and Parsons, 2003),have thickened CNS myelin sheaths (Fraser et al., 2008). The opposite effects of β1 inactivation and AKT activation on myelination prompted us to test whether the two proteins are functionally linked. Consistent with this model, the loss of β1 integrins in oligodendrocytes affected AKT activation. A second laminin receptor,dystroglycan, which also promotes myelin membrane outgrowth in vitro(Colognato et al., 2007), did not affect AKT activation, indicating that there is a specific link between AKT and β1 integrins. Furthermore, defects in myelin membrane outgrowth in β1-deficient oligodendrocytes were rescued upon expression of constitutively active AKT or by modulating endogenous AKT activity levels with a PTEN inhibitor. Based on these findings, we suggest that β1-deficient oligodendrocytes are unable to properly myelinate axons at least in part due to defective AKT signaling. Interestingly, focal adhesion kinase (FAK, PTK2 -Mouse Genome Informatics) and integrin linked kinase (ILK), mediators of integrin functions that control AKT, have also been implicated in the regulation of myelin sheet outgrowth by oligodendrocytes(Chun et al., 2003; Hoshina et al., 2007). Collectively, these data support a model in which extracellular ligands that activate β1 integrins induce AKT activity to control myelin outgrowth and axonal wrapping in the CNS. In the future, it will be important to define the downstream effectors of AKT and the extent to which they integrate biosynthetic and cytoskeletal pathways to control the outgrowth of myelin membranes that enwrap axonal processes in the CNS.