In our previous studies, we showed laminin binds α-dystroglycan in the dystrophin glycoprotein complex and initiates cell signaling pathways. Here, differentiated C2C12 myocytes serve as a model of skeletal muscle. C2C12 cells have a biphasic response to the laminin-α1 laminin globular (LG) 4–5 domains (1E3) dependent on the concentration used; at low concentrations of 1E3 (<1 μg/ml), myoblast proliferation is increased while higher concentrations (>1 μg/ml) cause apoptosis in myoblasts and differentiated myotubes. This alters the activation of the transcription factors activator protein-1 (AP-1) and NF-κB via laminin-dystrophin glycoprotein complex (DGC)-src-grb2-sos1-Rac1-Pak1-c-jun N-terminal kinase (JNK)p46 and laminin-DGC-Gβγ-phosphatidylinositol 3-kinase (PI3K)-Akt pathways, respectively. A specific antibody against Ser63 phosphorylated c-jun completely blocks or supershifts the AP-1-DNA binding resulting from laminin binding but only partially blocks or supershifts the AP-1-DNA binding resulting from 1E3. This suggests that AP-1 contains phosphorylated c-jun in the presence of hololaminin but contains a different composition in the presence of 1E3. Nuclear NF-κB was only upregulated by a low concentration of 1E3 and is then diminished by a higher concentration; it also has a biphasic response. Nuclear localization of NF-κB is affected by PI3K/Akt signaling, and DGC associated PI3K activity also shows a biphasic response to 1E3. Furthermore, our data suggest that activation of c-jun N-terminal kinase participates in the cell survival pathway and suggest that NF-κB is involved in both survival and cell death. A model is presented which incorporates these observations.
- nuclear factor-κB
- activator protein-1
- laminin globular (LG) 4–5 domains
in skeletal muscle, dystrophin, dystroglycan (DG), and syntrophins are found in a complex with other proteins and glycoproteins (11, 46), the dystrophin glycoprotein complex (DGC). Defects in the complex cause muscular dystrophies. Duchenne muscular dystrophy is the absence of dystrophin and the most common progressive muscle wasting disease in humans.
In the sarcolemma, the DGC provides a link between in the extracellular matrix laminin (15) through the sarcolemma to the cytoskeleton (12). Laminin forms complex polymeric structures, attaching to neighboring laminins by domains on the β,γ-chains and the N-terminal end of the α-chain of the cruciform laminin trimer (15, 18). Laminin binds sarcolemma receptors, the DGC, and integrins, primarily through the five laminin globular (LG) domains at the C terminus of the α-chain (7, 23, 30, 37, 49). The LG4–5 domains are called the E3 region of laminin, and in laminin-111 this region binds to the polysaccharide of DGC α-DG, which anchors the basal lamina matrix to the sarcolemma (38). This binding involves polymeric laminin binding, which in turn helps organize the DGC in the sarcolemma (6). In the sarcolemma of healthy skeletal muscle, the α7β1-integrin is the only integrin binding laminin (43).
In previous studies (48, 49), we showed that the binding of laminin-111, or the purified human laminin-α1 chain E3 domain protein, called here 1E3, also initiates DGC-mediated signaling. One pathway, laminin-DGC-src-grb2-sos1-rac1-Pak1JNKp46 causes ultimately the phosphorylation of c-jun, a subunit of the activator protein-1 (AP-1) transcription factor (31, 49). The other, laminin-DGC-Gβγ-phosphatidylinositol 3-kinase (PI3K)-Akt (45), has been less well characterized but it is known that disruption of the laminin-DGC interaction causes apoptosis in myotubes and that this involves Akt (26). Furthermore, the 2E3 protein, containing the analogous sequences from the mouse laminin-α2-chain, has been shown to allow attachment of myotubes, and this prevents anoikis, apoptosis, and ultimately necrosis that occurs when cells are not attached to an appropriate matrix (30). While isolated LG4–5 domain disrupts polymeric laminin binding (6), little is known about the consequence this has on the muscle cell.
Recently, laminins, perlecan, and agrin, all ligands of DG, have been investigated as potential therapies for muscular dystrophies (14, 16, 17, 28, 34). In some studies, “mini-protein” constructs, containing DG-binding domains fused to other sequences, have used sequences from agrin and perlecan. Other studies have employed laminin-α1-subunits or an intact hololaminin αβγ-heterotrimer. These therapies have proved somewhat effective. For example, muscle-specific overexpression of the chimera of the laminin-binding region of agrin with the specific DG binding domain of perlecan prolonged the life expectancy of the dyW/dyW mouse model of congenital muscular dystrophy ∼2.5-fold (28). Perlecan is a ligand of DG but not integrins, showing the importance of DG binding. Since engineered proteins containing only limited sequence regions of DG ligands are being used, we undertook a study of the 1E3 protein, containing the DG-binding site of laminin-α1 in isolation away from other laminin domains, which could have potential as a treatment for disease.
Here, we shown that low concentrations of 1E3 cause proliferation in myoblasts while higher concentrations cause apoptosis in myoblasts and myotubes. We investigated this biphasic response to determine if it may provide information about the roles of laminin-binding-induced cell signaling in myocytes. We found that 1E3 induces activation of c-jun and its N-terminal kinase-pJNK p46 regardless of the concentration of 1E3. The c-jun is typically a subunit of the AP-1 transcription factor. AP-1 is activated by 1E3 and a specific phospho-c-jun antibody only partially blocked or shifted its DNA-protein-binding complex. In contrast, hololaminin, which causes only proliferation of C2C12 myoblasts, results in a different form of AP-1 that is totally responsive to the phospho-c-jun antibody. Laminin is also known to alter the cell compartmentalization and DNA-binding activity of the NF-κB transcription factor (9). Within a certain range of 1E3 concentrations (<1 μg/ml) nuclear NF-κB protein and activity increases, and these diminish at higher concentrations of 1E3 (>1 μg/ml) and this biphasic response matches closely the effect on cell viability. PI3K and Akt, which are upstream of NF-κB, also show a biphasic response to 1E3, suggesting a role for this signaling in modulating nuclear NF-κB.
Overall, the results suggest that laminin engineering for therapy should proceed with caution.
Mouse laminin-111 was obtained from Collaborative Biomedical Products. All other chemicals, unless otherwise specified, were of the highest purity available commercially.
Antibodies against phosphorylated c-jun (on Ser-63), phospho- activating transcription factor 2 (ATF2; on Thr-71), ATF2, laminin-γ, NF-κBp65, phospho-JNK (on Thr-183 and Tyr-185), heterogeneous nuclear ribonucleoprotein (hnRNP; A2/B1), and β-actin were from Santa Cruz Biotechnology. Antibodies against α-DG, IIH6 and VIA4, were the generous gifts of Dr. Kevin Campbell (University of Iowa). β-DG antibody was the generous gift of Dr. Tamara C. Petrucci (Instituto Superiore di Sanita, Rome, Italy). Goat anti-mouse IgG (H + L)-horseradish peroxidase conjugate and goat anti-rabbit IgG (H + L)-horseradish peroxidase conjugate were from Southern Biotechnology.
1E3 was expressed from the hLNal-E3 293 (HEK293) cell line transfected with the LG4–5 domains of human laminin-α1 and purified as described previously (40). Briefly, the cultured medium was harvested at 24 and 48 h and diluted 1:1 with water. Diluted medium was loaded onto a 25 ml DEAE-Sepharose column the outlet of which was connected a 5-ml Hi-trap (Pharmacia) heparin column. The heparin bound protein was eluted and detected by absorption at 280 nm or by gel electrophoresis stained with Coomassie brilliant blue (25). The concentration of 1E3 was determined (2) using BSA as a standard.
Cell proliferation and viability assay.
Mouse C2C12 myoblasts were grown and maintained in DMEM containing 10% FBS at 37°C in a humidified atmosphere of 5% CO2. For experiments with differentiated C2C12 cells, myoblasts were cultured in 100 × 10 mm dishes to 80% confluence and then the medium was changed to DMEM with 1% FBS and cultured for a further 5 days to allow myoblast fusion and myotube differentiation. For cell growth experiments, myoblasts or myotubes were placed in sixwell plates in DMEM with 10% FBS or 1% FBS, respectively, including the concentrations of 1E3 shown in results for 24 and 48 h at 37°C. The cells were counted after trypsinization using trypan blue. To determine cell cycle progression, propidium iodide (PI) was incorporated to nuclei after using 70% methanol to fix the cells. To determine apoptosis and necrosis, cells that stained with FITC conjugate annexin V and incorporated PI were determined by flow cytometry.
C2C12 myotubes (100 μl; 5 × 104) were plated on each well of a Chamber Slide (NUNC) in culture medium without or with different concentrations of 1E3 overnight at 37°C. The cells were washed three times with PBS before the samples were fixed in 2% paraformaldehyde solution for 30 min at room temperature. The Chamber Slide was washed three times with PBS. The rabbit polyclonal antibody against laminin-γ (1:50) was diluted in PBS containing 3% BSA, added to the slide, incubated for 30 min at room temperature in a humidified container, and then washed with PBS. The goat anti-rabbit IgG (H + L)-Alexa Fluor 488 secondary antibody (1:100) was added, and the cells again were washed with PBS. Slides were mounted with 50% glycerol mounting medium. The immunofluorescence in the cell was observed with a Zeiss LSM 5 confocal microscope using Ar and HeNe lasers. The positive spots were counted by using Imaris X64 version 5.7.1 (Bitplane, St. Paul, MN).
EMSA was performed by incubating 50-μg nuclear extracts, derived from control and stimulated (with 1E3 or laminin) C2C12 cells, with 10 fmol [γ-32P]-ATP labeled double-stranded oligonucleotides in incubation buffer [10 mM Tris·HCl pH 7.5, 50 mM NaCl, 2.5 mM MgCl2, 1 mM dithiothreitol, 0.1% (vol/vol) Triton X-100, and 5% glycerol] containing 1 μg/ml polydI:dC. After incubation at room temperature for 30 min, the complex formed was separated on a preelectrophoresed 6% native polyacrylamide gel in 0.25 × TBE buffer (final concentration: 12.5 mM Tris, 1.25 mM boric acid, and 0.25 mM EDTA) at 10 V/cm for 1 h. The products were analyzed by autoradiography. Nuclear and cytosolic fractions from the C2C12 myotubes were prepared using the NE-PER nuclear and cytoplasmic extraction kit and instructions from Pierce Chemical.
Oligonucleotides used were as follows: AP-1 (consensus): 5′ -cgcttgatgactcagccggaa-3′; ATF: 5′-agagattgcctgacgtcagagagctag-3′; and NF-κB: 5′-agttgaggggactttcccaggc-3′. These represent the canonical response elements for the named transcription factors.
Endogenous laminin was depleted from cultured C2C12 cell using trypsinization (47). To the laminin-depleted cells, either nothing or 1E3 was added to the culture medium for an hour. Cell membranes in buffer K (20 mM HEPES pH 7.5, 10 mM MgCl2, and 100 mM KCl) were precleared with protein G-Sepharose and then incubated with 5 μg anti-syntrophin for 30 min at 4°C. Samples were solubilized by addition 1% Triton X-100, 0.5% IGEPAL, and 0.5% sodium deoxycholate, and incubation was continued for another hour. The immune complexes were then incubated with protein G-Sepharose for an hour, centrifuged, and washed extensively with buffer K. The bound proteins were detected by Western blot with the anti-laminin-γ1 antibody (5).
For the antibody blockade experiments, the anti-α-DG antibodies VIA4 and IIH6 nonblocking and blocking, respectively, were added (1:200) in the culture medium for an hour before cells were treated with 1E3.
The DGC was immunoprecipitated with VIA4 antibody and assayed for PI3K activity as described previously (9, 45). Phosphatidylinositol-3-phosphate (PI3P) was included in the analysis as a standard marker.
1E3, but not hololaminin, shows a biphasic effect on myoblast growth.
In Fig. 1A, undifferentated C2C12 myoblasts were cultured for 24 h with laminin-111. The cells proliferated nearly 1.6-fold at 40 μg/ml of laminin-111 compared with cells cultured in the absence of laminin-111 (Fig. 1A), followed by a plateau up to 100 μg/ml of laminin. In contrast to laminin-111, low concentrations of 1E3 (<1 μg/ml) caused a significant proliferation to nearly twofold in the myoblasts in Fig. 1B followed by a marked inhibition at higher concentrations; the number of cells actually decreases below the number originally plated indicating that myoblasts died. Here, we will further characterize this biphasic effect of 1E3. Hololaminin has 20 times the molecular mass of the 1E3 protein and thus on a molar basis, 40 μg/ml laminin-111 should be roughly equivalent to 2 μg/ml 1E3, and yet they clearly have quite different effects.
The simplest explanation of the toxicity of 1E3 at high concentrations would be that it contains a toxic contaminant. However, the 1E3 is purified from cell culture media using ion exchange chromatography and heparin chromatography and the protein-stained gel of the resulting protein shows a high state of purity (Fig. 1C). We explored other possibilities.
Our hypothesis is that low doses of 1E3 are proliferative (Fig. 1B) because of increased mitosis while high doses of 1E3 are toxic not because they inhibit mitosis (Fig. 2) but rather for another reason. To test this, we next performed a more limited study of the effect on the cell cycle of no 1E3; 0.8 μg/ml, a dose that increased cell numbers; and 4 μg/ml, a does that should be toxic (Fig. 2). For this experiment, >40,000 cells were counted for each condition. The number of cells in G2-S and M phases (shown as the M2-M3 bar in Fig. 2) significantly (P < 0.001) increased from 20 to 25% with 0.8 μg/ml of 1E3. Thus the data show that the low dose of 1E3 causes myoblasts to enter mitosis, which is consistent with the proliferation seen in Fig. 1B. In other experiments not shown, we also showed a similar increase in mitosis with intact laminin-111 in agreement with others (18, 32). With 4 μg/ml 1E3, mitosis returns to the level found in the absence of 1E3. Thus 1E3 is not inhibiting mitosis at toxic doses.
1E3 also causes a biphasic effect on myotube viability and apoptosis.
We next investigated C2C12 myoblast cell viability by microscopy and flow cytometry and observed dying and apoptotic cells (full data not shown, but see Table 1). To investigate this further by flow cytometry, myotubes were used (Fig. 3A), which were used in all further experiments to extend these results to more differentiated cells. Briefly, 5 × 105 myotubes were incubated with 0, 0.5, 2.5, 5.0, and 10 μg/ml of 1E3 for 48 h. The cells were then stained with FITC-annexin-V and propidium iodide. Regions S1–4 of the histogram show the portion of the cells binding the different stains. The percentage of the cells in each region are also listed in Table 1. These regions are as follows: S1, annexin-V- and PI-negative cells, which are alive; S2, only PI-positive cells, which are dead cells; S3, only annexin-V-positive cells, which are apoptotic cells in an early stage; and S4, annexin-V- and PI-positive cells, which are apoptotic cells at a later stage of cell death. As 1E3 is increased from 0 to 10 μg/ml, the number of viable S1 cells decreases from 95 to 6.5%, and the cells begin to appear in the other quadrants as they progress through apoptosis to ultimate death. The lowest dose of 1E3 (0.5 μg/ml) caused a mild increase in viable cells. While myoblasts undergo mitosis, myotubes do not. This small increase by a low dose of 1E3 may be because it prevents spontaneous apoptosis in myotubes rather than because of a proliferative effect.
For comparison, the data for myotubes and myoblasts are summarized in Table 1. For both myoblasts and myotubes, low doses of 1E3 (<1 μg/ml) result in a higher proportion of viable cells while higher doses increase apoptosis.
The resulting myotube morphology resulting from growth in 1E3 is shown in Fig. 3B. Microscopic inspection shows cells with internal apoptotic bodies and enlarged bleb-like structures upon growth at the highest 1E3 concentration. Counting the cells (Fig. 3C) shows that the higher dose of 1E3 was toxic and the cell number decreases to ∼50% of the number of cells initially plated (3 × 105). Since myotubes do not divide, the small increase (to 3.4 × 105) seen at 0.8 μg/ml 1E3 may result from better plating efficiency in 1E3 or may indicate that the myotubes also contained some myoblasts, which proliferate.
1E3 also affects the amount of hololaminin bound to the DGC and its cellular distribution in myotubes.
In muscle, cruciform hololaminin forms a polymeric matrix, which causes clustering of the DGC, and this receptor clustering is prevented by the isolated laminin-α E3 sequences (6). To observe the effect of 1E3 on hololaminin binding in myotubes, confocal microscopy with an antibody directed against laminin-γ1-chain was performed. This allowed us to discern hololaminin in the presence of 1E3 binding. The experiments in Fig. 4A show the result when myotubes were cultured overnight in medium either lacking or containing 1E3, a time too short to observe any major toxic effect. Staining with trypan blue (data not shown) confirmed that there was no change in the number of viable cells. In the absence of 1E3, laminin staining is seen throughout the length of the myotube with some greater staining near the nucleus. At low 1E3 (0.85 μg/ml), the staining of laminin is enhanced and most of the enhancement occurs in the perinuclear region. At a higher dose (8.5 μg/ml), the staining along the length of the myotube diminishes while it remains in the perinuclear region. When the images were analyzed to count the number of fluorescently labeled spots of laminin over many microscopic fields, these data (Fig. 4B) confirm that staining is increasing in low 1E3 and decreased in high 1E3. In Fig. 4C, the DGC was immunoprecipitated with a syntrophin antibody and Western blotted with the laminin-γ1 antibody (Fig. 4C). Densitometry (Fig. 4D) from three experiments confirms that the amount of laminin was significantly increased by low 1E3 and decreased at high 1E3, although not to the level seen in the absence of 1E3. Since laminin staining is biphasic (Fig. 4B) as is the amount of laminin bound by the DGC (Fig. 4D), we conclude that the DGC may be involved in this biphasic response to 1E3, although we cannot exclude that laminin binding to an alternate receptor (e.g., integrins) is not also affected.
Laminin and 1E3 increase signaling through the JNK pathway, but this is not biphasic.
Since laminin binding to the DGC causes JNK1-p46 activation, which in turn phosphorylates c-jun (31, 47), we next investigated whether c-jun and ATF2 are phosphorylated in C2C12 myotubes in response to laminin. ATF2 is also a substrate of JNK (39). Both are phosphorylated in response to laminin while total c-jun and ATF2 are not altered and provide loading controls (Fig. 5, A–D). When a low dose of 1E3 is added, c-jun phosphorylation is increased and this increases further as greater 1E3 is added in Fig. 5E. The expression of actin was detected as a loading control in Fig. 5F. Thus the biphasic effect of 1E3 observed in cell viability and proliferation (Fig. 1) is not observed with c-jun phosphorylation.
Laminin-DGC initiated signaling through JNK1 is initiated by a c-src family kinase, and in a previous study (47) of C2C12 myoblast proliferation in response to low doses (<1 μg/ml) of 1E3, we showed that the src inhibitor PP2 could prevent 1E3-induced proliferation while the inactive analog PP3 was without effect. Thus the laminin-DGC-src-JNK1 signaling may be responsible for the proliferative effects of low doses of 1E3 but probably does not account for the toxic effect of higher 1E3.
However, the effect may be somewhat more complex. In Fig. 6A, we used the EMSA with the consensus AP-1 element DNA. When laminin is added, increased DNA-protein complex is observed, the p-c-jun antibody supershifts all of the complex to a lower mobility, and no unshifted complex is observed. Another c-jun antibody, reactive with both phosphorylated and nonphosphorylated c-jun, blocks DNA binding, diminishing the observed complex. Thus laminin is altering the majority of the AP-1 binding, and this binding clearly results from c-jun phosphorylation. However, when 1E3 was used instead of laminin, the phosphorylated c-jun antibody does not block the DNA-AP-1 complex but does shifts a small amount of the complex to lower mobility, as shown in Fig. 6B. This suggests that in response to 1E3, other variants of AP-1 are binding to the DNA. AP-1 is a complex transcription factor, which can be composed of c-jun dimers or can be composed of homo- and heterodimers with other members of the basic leucine zipper family, including JunB, JunD, c-fos, ATF-family members, and others. 1E3 may be causing alternative forms of AP-1 to bind in addition to the phospho-c-jun-rich AP-1 resulting from laminin. These alternative AP-1 forms may contribute to 1E3 toxicity.
To exclude the possibility that c-jun phosphorylation resulted from some other, non-DGC-mediated signaling, we used two specific monoclonal antibodies against DGC α-DG and C2C12 myotubes in Fig. 7. When cells were cultured in the presence of the VIA4, an antibody that binds α-DG but does not block laminin binding, 1E3 addition still results in increased c-jun phosphorylation. However, this is blocked by the IIH6 antibody, which also binds α-DG but blocks laminin binding. Also, JNK1-p46 is inhibited by IIH6 but not by VIA4 (Fig. 7C). Neither antibody is altering the amount of nuclear hnRNP, which provides a loading control (Fig. 7B). Thus it is 1E3 binding to the α-DG that causes c-jun phosphorylation, as was previously shown for laminin (31). Unpublished data from our laboratory has suggested that MKK4 is immediately upstream of JNK1 and causes its laminin-dependent activation. This is confirmed for 1E3 in Fig. 7D, since IIH6 but not VIA4 is also preventing MKK4 activation. This is also observed with EMSA and AP-1 DNA binding (Fig. 7E). In the absence of 1E3, little AP-1 binding is detected and this is blocked by IIH6. Increased amounts of 1E3 protein increase AP-1-DNA binding, which is ablated by the blocking antibody. 1E3 is binding to the DGC to cause these effects.
Nuclear NF-κB and its DNA binding do show a biphasic effect of 1E3.
NF-κB is a transcription factor important to muscle atrophy and inflammation (3, 22). In Fig. 8A, we investigated nuclear NF-κB in myotubes in response to 1E3. The amount of NF-κB in the nucleus is enhanced in low 1E3, 3.5-fold based on densitometry relative to “−”, and diminishes at a higher dose. The amount of hnRNP in the nucleus is little affected and provides a loading control (Fig. 8B). In fact, hnRNP does go up slightly, by 1.4 fold in “+” 1E3 relative to “−” but cannot account for the much larger effect seen in Fig. 8B. Nuclear NF-κB is responding in a biphasic fashion in response to 1E3 in a way that could explain the biphasic effects shown earlier on proliferation and cell viability. This biphasic effect is also seen in the DNA-binding activity assessed by EMSA (Fig. 8C). In this case, the gel shift is complex due to the presence of more than one protein-DNA complex, presumably due to the NF-κBp65 (top) and the NF-κBp55 (bottom) isoforms. The complex resulting from NF-κBp65 is clearly shown by the ability of the NF-κBp65 antibody to super shift this complex.
Thus nuclear NF-κB shows a biphasic response to 1E3 concentration. Whether this response is resulting from 1E3 binding to the DGC was investigated using the IIH6 and VIA4 α-DG antibodies. IIH6 diminished the nuclear NF-κB and abolishes the biphasic response (Fig. 9A) and the NF-κB-DNA complex becomes nearly undetectable (Fig. 9B) One of the many cell signaling pathways known to affect NF-κB is the PI3K/Akt pathway (9, 10), which is also a signaling pathway of laminin binding to the DGC (26, 45). To determine if this DGC pathway is also showing the same biphasic response, the PI3K activity was investigated using the PI3K immunoprecipitated with the DGC by the VIA4 antibody (Fig. 9C). In this experiment, the DGC was purified by immunoprecipitation with the VIA4 antibody and used as the source enzyme for the PI3K assay, following the phosphorylation of phosphatidylinositol using γ-32P-ATP. The position of PI3P, determined by applying a standard to an adjacent lane (not shown) is indicated. The autoradiogram shows the amount of PI3P formed. Clearly, the PI3K activity iscreases with low 1E3 (+) and is nearly completely absent at the higher 1E3 (++). DGC-associated PI3K activity is also showing a biphasic response to 1E3 suggesting strongly that this signaling pathway is involved in the NF-κB biphasic response.
The amount of activated pAkt follows a similar pattern as PI3K and reveals other differences. Figure 10 shows the Western blot of pAkt (A), total Akt (B), and the activity of the Akt kinase activity in phosphorylating histone 2B (C). Without 1E3, pAkt and its activity are low, addition of a moderate amount of 1E3 increases both and the higher toxic 1E3 amount reduces pAkt activity. Interestingly, the total Akt also shows an additional, more slowly migrating form of Akt that may also be phosphorylated though the resolution on the pAkt Western blot is too poor to be certain.
Here, we investigate the biphasic effect of 1E3 in myoblast proliferation, apoptosis, mitosis, and the myotube cell survival in low 1E3 concentrations and the toxic effects of high 1E3 (Figs. 1⇑–3) and then investigate some of the cell signaling associated with this binding (Figs. 4⇑⇑⇑⇑⇑–10). A model that can account for the data presented is shown in Fig. 11.
The most relevant conclusion is that a DG-binding domain, in isolation, can have a quite different effect when taken out of the context of the holoprotein. Certainly, the mini-agrin gene expression is beneficial in a mouse model of congenital muscular dystrophy, improving muscle function and repair and animal longevity (28). The agrin-perlecan chimera was nearly as effective, and since perlecan binds DG but not muscle integrins, it clearly shows that binding to α-DG is necessary to improve dystrophy. These constructs include an agrin region that also binds laminin, and this is likely to be an important feature of the success (28). The 1E3 protein used here does not contain these laminin-binding sequences, and this lack of binding to neighboring laminins is likely to be necessary for effective therapy.
Congenital muscular dystrophy is not the only muscular dystrophy to show benefits from laminin. Intramuscular injection of laminin-111 improves muscle strength and diminishes pathology in the mdx mouse as well and additionally increases expression of integrin-α7β1, another receptor for laminin (17, 34, 35). Furthermore, laminin has proved to be an effective adjuvant for myoblast transplantation in the mdx mouse (17), a potential therapy for muscular dystrophy in mice, which has proven unsuccessful in humans. Overexpression of laminin-111 in the congenital muscular dystrophy mouse model also improved life span, and muscle strength and deletion of the LG4–5 domain (1E3) region, which binds α-DG but not integrin, made this ineffective in limb skeletal muscle, although interestingly it remained effective in heart and diaphragm, showing that muscles differ in their laminin requirement (13, 14). The normal muscle laminin is laminin-211, not laminin-111, but both bind DG and integrin, support myotube growth and migration, and for many experiments can be used interchangably. In laminin-111, the LG4 domain binds DG while LG1–3 binds integrin, making it an easier model for distinguishing these receptors, while for laminin-211, DG binds both the LG4–5 and LG1–3 regions (37, 41).
Whether discrete LG4–5 is of physiological relevance is another interesting question. In laminin-α2, there is a site for furin cleavage in the LG3 domain and this cleavage occurs in vivo and is physiologically relevant (36). Thus LG4–5 detached by furin cleavage from muscle laminins is present in vivo and likely has different effects than hololamin.
Our study shows that while matrix proteins such as laminin show potential for therapy, truncation experiments will require some caution and unexpected effects do occur.
Other conclusions concern the interplay between cell signaling pathways, which may account for these effects. First, the background of the model (Fig. 11) requires discussion.
Low 1E3 enhances hololaminin binding.
1E3 represents the region of laminin-111 that binds strongly and specifically to α-DG (37). That it has such a different effect, being biphasic causing both viability and apoptotic effects, while laminin is purely increasing viability, was surprising. One of the simplest means by which 1E3 could have its biphasic effect would be if it bound to unoccupied laminin binding sites, on the DGC or integrins, to induce additional growth signaling but at higher concentrations displaces hololaminin to diminish this positive signaling. However, this simple explanation does not appear to be the case. Low concentrations of 1E3 (0.85 μg/ml) actually enhanced hololaminin binding instead by 1.75-fold followed by a small diminution by ∼25% relative to control at a high concentration (8.5 μg/ml; Fig. 4). However, at this high concentration, the localization of the laminin along the myotube is altered and this may by be an important effect. Laminin binding to the DGC also shows a biphasic response to 1E3, although it remains at levels higher than seen in the absence of 1E3. Thus the amount of laminin-bound alone cannot account for the dramatic effects seen on myocyte viability and so the cell signaling accompanying these changes was investigated.
c-jun and JNK-signaling.
c-jun is after all an oncogene important to cell proliferation and tumorogenesis and one level of its regulation is via JNKs. In the case of laminin-DGC signaling, this is JNK1 (31, 47, 49). Phosphorylation of c-jun, ATF2, and of many other transcription factors by JNK accompanies stimulation by mitogenic growth factors; this kind of stimulation is usually transient activation. However, prolonged stimulation of JNK, by TNFα or ionizing irradiation, can also result in apoptosis (reviewed in Ref. 8). The c-jun and other members of the basic helix-loop-helix transcription factors form a variety of homo- and heterodimers that together are called AP-1. Thus AP-1-response element binding is not always constituted of the same protein subunits. Indeed, we found evidence for such changes in the 1E3 response. Low concentrations of 1E3 activate JNK and increase c-jun phosphorylation, and the AP-1 DNA-binding complex is nearly all bound by the phospho-c-jun antibody showing that AP-1 in this case contains predominantly phospho-S63-c-jun. At high 1E3 however, the stimulation of JNK is giving rise to a different AP-1 complex, most of which is not reactive with the phospho-c-jun antibody (Fig. 6), indicating that phospho-S63-c-jun is not a part of the AP-1 complex. Because of the large number of possible AP-1 complexes, the nature of these complexes was not characterized and awaits further investigation. Our experiments involve myotubes cultured with 1E3 for 48 h, and so this stimulation of JNK is prolonged and may be at least partially causal for the apoptosis occurring at high 1E3.
Additionally, JNK1 signaling is related to the various isoforms of JNK1. In skeletal muscle, these are JNK1-p46 and JNK-p54. In rabbit skeletal muscle, in laminin-depleted microsomes, phospho-JNK-p54 predominates, but the addition of laminin results primarily in phospho-JNK-p46 (31). The mdx mouse is a model for Duchenne muscular dystrophy. In two different mdx mouse muscle types, the predominant form is highly activated phospho-JNK-p54 and activation of JNK by transfected MKK7 aggravates the dystrophic phenotype while inhibition of this JNK by JNK inhibitory protein (JIP1) improves it (24). Thus there is a strong association between the degree of pathology with the level of phospho-JNK-p54. However, in C2C12 myotubes, only phospho-JNK-p46 is observed in the presence or absence of 1E3, regardless of dosage. Since this isoform is also found activated by hololaminin, it seems to be a normal signaling in myocytes that is not normally associated with pathology.
High levels of nuclear NF-κB result from atrophy and are found in mdx mouse where the DGC is severely disrupted (reviewed in Ref. 27). Nuclear NF-κB is highly regulated by the various inhibitors of NF-κB (IκB), which bind NF-κB and sequester it in the cytosol away from the nucleus. In turn, the levels of IκB are regulated by inhibitor of NF-κB kinase (IKK), which phosphorylates IκBs, causing their proteolysis, and results in higher levels of nuclear NF-κB (reviewed in Ref. 33). Muscle-specific deletion of IKKβ results in less nuclear NF-κB, less binding to DNA and increased myofiber number and muscle strength (1, 29). The NF-κB pathway and PI3K/Akt signaling are related by the increased IKKα phosphorylation by the latter. This phosphorylation results in IKKα activation and increased nuclear NF-κB (20, 21). NF-κB-induces expression of muRF1, a ubiquitin ligase responsible for the proteosome degradation of many key muscle proteins (3, 29). Nuclear NF-κB also increases expression of the YY1 transcription factor, which represses expression of muscle α-actin, creatinine kinase, and myosin heavy chain IIb (44). Thus increased levels of NF-κB, such that occur at low 1E3 (<1 μg/ml), would be detrimental to muscle and presumably also to the cultured myotubes used here. This increase in NF-κB may thus prevent the positive effects of c-jun phosphorylation and produce the biphasic effect on myotube survival.
However, the effects of NF-κB are not totally negative. In myoblast differentiation to myotubes, NF-κB is required for differentiation (33, 44). Our model of differentiated C2C12 did not allow us to investigate the effects of NF-κB on myocyte differentiation.
Previously, we (45) have shown that laminin binding to the DGC causes PI3K activity to increase in a process involving Gβγ-subunits of the Gs heterotrimer. We show here that PI3K and Akt activities were also biphasic (Figs. 9 and 10) following 1E3 binding to α-DG. When Akt activity is high (1E3 <1 μg/ml), nuclear NF-κB would also be high, and this will diminish as Akt decreases (1E3 >1 μg/ml). Akt is antiapoptotic in its action, and blocking laminin binding to the DGC with the IIH6 antibody increases apoptosis (26). This fall in Akt accompanying the fall of PI3K activity may thus be important to the apoptosis occurring at high 1E3.
The biphasic effect of 1E3 is thus likely due to the interplay of various signaling pathways. All cell signaling was not investigated, and there may well be other relevant signaling causal for this phenomenon. However, the signaling studied does provide some insight into how this may occur, as depicted in the model in Fig. 11. At low 1E3, PI3K and Akt (Figs. 9 and 10) (45) are high, as well as nuclear NF-κB, and active JNK1-p46. Akt resists apoptosis, JNK1 can also increase viability, but NF-κB is probably initiating cellular remodeling and ultimate apoptosis. As 1E3 is further increased, active JNK1-p46 remains high or higher, while PI3K/Akt are diminished and the combination will eventually lead to apoptosis even as NF-κB diminishes.
What is causing these changes in signaling is likely the reorganization of the hololaminin matrix observed long ago by Colognato et al. (6). The isolated E3 region disrupts the polymeric structure of the hololaminin matrix and this disrupts the organization of the DGC in the sarcolemma (6). Our much lower resolution images were not detailed enough to clearly observe this reorganization, though it does clearly show that the organization of the laminin matrix along the myotube is being altered (Fig. 4).
This work was supported by National Institute of Arthritis and Musculoskeletal and Skin Diseases Grant R01-AR-051440. UTSA RCMI is supported by National Center for Research Resources Grant 2G12-RR-013646-11.
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: Y.W.Z., J.M., and D.J. performed experiments; Y.W.Z., J.M., D.J., and H.W.J. analyzed data; Y.W.Z. and H.W.J. approved final version of manuscript; H.W.J. conception and design of research; H.W.J. prepared figures; H.W.J. drafted manuscript; H.W.J. edited and revised manuscript.
We thank Maria R. Galindo for excellent technical assistance. Also, we thank to the University of Texas, San Antonio, Research Centers in Minority Institutions (UTSA RCMI) for the use of equipment, software, and computers.
- Copyright © 2012 the American Physiological Society