Association of Phosphofructokinase-M with Caveolin-3 in Differentiated Skeletal Myotubes DYNAMIC REGULATION BY EXTRACELLULAR GLUCOSE AND INTRACELLULAR METABOLITES*

Caveolin-3 is a member of the caveolin family of pro- teins that is primarily expressed in striated muscle cell types (skeletal and cardiac). Here, we show that an (cid:59) 80-kDa protein specifically co-immunoprecipitates with caveolin-3 expressed in differentiated skeletal C2C12 myotubes. Microsequence analysis of this (cid:59) 80-kDa polypeptide revealed its identity as a key regulatory enzyme in the glycolytic pathway, namely phosphofruc-tokinase-M (PFK-M). Pulse-chase experiments demon- strate that PFK-M associates with caveolin-3 with a significant time lag after the biosynthesis of PFK-M. In addition, we show that this interaction is (i) highly regulated by the extracellular concentration of glucose and (ii) can be stabilized by a number of relevant intracellular metabolites, such as fructose 1,6-bisphosphate and fructose 2,6-bisphosphate, which are known allosteric activators of PFK. While the bulk of these experiments were performed in C2C12 cells, identical results were obtained using mouse skeletal muscle extracts. Taken together, our results suggest that glucose-dependent plasma membrane recruitment of activated PFK-M by caveolin-3 could have important implications for understanding the mechanisms that regulate energy metabo- lism in skeletal muscle fibers.

An extensive body of evidence suggests that certain glycolytic enzymes associate with other enzymes or with cytoskeletal components (1)(2)(3). Such interactions are thought to modulate the activities of these enzymes. Phosphofructokinase is the key enzyme in the control of glycolysis. It catalyzes the committed and rate-limiting step in glycolysis, the conversion of fructose 6-phosphate to fructose 1,6-bisphosphate. As such, phosphofructokinase is subject to a myriad of allosteric effectors, such as fructose 2,6-bisphosphate, fructose 6-phosphate, ATP, AMP, H ϩ , and citrate.
There are three different isoforms of phosphofructokinase (PFK) 1 : PFK-A, PFK-B, and PFK-C (4,5). These isoforms combine to give rise to homo-and hetero-tetrameric complexes (6 -8). PFK-A (also termed PFK-M for the rat and human isoform) is predominant in muscle tissues. PFK-B (PFK-L in rat and human) is most highly expressed in liver. Finally, PFK-C (PFK-C in rat, PFK-P in human) expression is relatively brain-specific.
A number of reports describe the association of PFK-M 2 with various cytoskeletal elements and signal transduction-related kinases. In heart muscle, a fraction of PFK-M is associated with phospholipase A 2 (9). In skeletal muscle, PFK-M associates with tubulin under certain conditions (10). Insulin plays a key role in this process, as it stimulates binding of both PFK and aldolase to the muscle cytoskeleton (11). PFK can undergo reversible phosphorylation by cAMP-dependent protein kinase (protein kinase A) and protein kinase C (12,13), and it is a substrate for regulation by a number of retroviral transforming protein kinases (14). Also, purified epidermal growth factor receptor and insulin receptor can phosphorylate PFK on tyrosine residues (15,16).
Epidermal growth factor receptor, as well as a large number of other molecules associated with signal transduction events, have been localized to caveolae, i.e. small, flask-shaped plasma membrane domains (17)(18)(19)(20)(21). The principal marker proteins for caveolae are a family of molecules called caveolins (reviewed in Ref. 22). This family consists of proteins that share a high degree of homology, have a molecular mass between 18 to 24 kDa, and are expressed in a tissue-specific manner (23)(24)(25)(26)(27). Caveolins may act as scaffolding proteins within caveolae membranes (28). Caveolins form high molecular mass homooligomers (ϳ14 -16 monomers per oligomer) (28 -30), and these caveolin homo-oligomers have the capacity to bind cholesterol (30,31) and self-associate into larger structures that resemble caveolae (28). Caveolin-1 mRNA and protein expression levels are highest in cell types that contain numerous caveolae, i.e. adipocytes, endothelial cells, smooth muscle cells, and fibroblasts (reviewed in Ref. 22). The tissue distribution of caveolin-2 mRNA greatly resembles the distribution of caveolin-1, and both appear to be co-expressed in the same cell types (23). In contrast, caveolin-3 expression is limited to muscle tissue types (skeletal muscle, diaphragm, and heart) where it is localized to the muscle cell's plasma membrane (sarcolemma) (24 -26). Similarly, caveolin-3 protein expression is dramatically induced during the differentiation of C2C12 skeletal myoblasts to myotubes in culture (24 -26). However, it remains unknown which molecules interact with caveolin-3 in vivo.
Here, we show that an ϳ80-kDa protein specifically co-immunoprecipitates with caveolin-3 expressed in differentiated skeletal C2C12 myotubes. Microsequence analysis of this 80-kDa polypeptide revealed its identity as PFK-M, a key regulatory enzyme in the glycolytic pathway. In addition, we demonstrate that association of PFK-M with caveolin-3 is highly dependent on extracellular glucose concentrations and can be stabilized by a number of relevant intracellular metabolites such as fructose 1,6-phosphate and fructose 2,6-phosphate.
Cell Culture-C2C12-3 cells (32) were derived from a single colony of C2C12 cells (33) cultured at clonal density and display a more stable phenotype than the parental cell line. C2C12-3 myoblasts were cultured as described previously (32). Briefly, proliferating C2C12-3 cells were cultured in high mitogen medium (DMEM containing 15% fetal bovine serum and 1% chicken embryo extract) and induced to differentiate at confluence in low mitogen medium (DMEM containing 3% horse serum). Overt differentiation was indicated by the assembly of multinucleated syncytia, which commenced 36 -48 h after the cells were switched to low mitogen medium. Rat aortic smooth muscle cells were the generous gift of Dr. Lee Graves (University of North Carolina, Chapel Hill, NC) and were isolated and characterized as described previously (25).
Pulse-Chase Labeling Experiments-Differentiated C2C12 cells were starved for 30 min in DMEM lacking cysteine and methionine and then labeled for the indicated amount of time in the same medium containing 0.3 mCi/ml Express Protein Labeling Reagent (1000 Ci/mmol). The cells were then washed twice with DMEM supplemented with unlabeled cysteine and methionine, and then fresh DMEM containing 300 M cycloheximide was added.
Immunoprecipitation-Immunoprecipitations were carried out using protein A-Sepharose CL-4B (Pharmacia Biotech Inc.) as described previously (34,35). Briefly, differentiated C2C12 cells were washed twice with cold phosphate-buffered saline and then scraped into lysis buffer (1% Triton X-100, 60 mM octyl-glucoside, 5 mM EDTA, 20 mM Tris, pH 8.0, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride). Insoluble debris was removed by centrifugation for 10 min at 15,000 ϫ g. The cleared tissue culture supernatant and the cell lysates were then incubated for 30 min at 4°C with protein A-Sepharose. The protein A-Sepharose was removed by centrifugation, and fresh protein A-Sepharose was added along with the corresponding antibody. Immunoprecipitations were performed for 3 h at 4°C; immunoprecipitates were then washed five times with lysis buffer (lacking octyl glucoside) and analyzed by SDS-PAGE.

Time-dependent Association of an ϳ80-kDa Protein with
Caveolin-3 in Differentiated Skeletal Myotubes-C2C12 cells offer a convenient model system to study caveolin-3, since both mRNA and protein levels of caveolin-3 are dramatically induced during the course of differentiation of C2C12 cells from myoblasts to myotubes (24 -26). To identify caveolin-3-associated proteins, differentiated C2C12 skeletal myoblasts that had formed myotubes were pulse-labeled with a mixture of 35 S-labeled methionine/cysteine and subjected to immunoprecipitation with anti-caveolin-3 IgG. Fig. 1 shows that, after a brief labeling period of 10 min, a faint band of ϳ80 kDa is immunoprecipitated with anti-caveolin-3 IgG (monoclonal antibody 26.2). During the chase period (15-120 min) with excess unlabeled amino acids, the signal for this ϳ80-kDa protein steadily increases (Fig. 1, lanes 5 through 8). As cycloheximide was present at all times of chase, no de novo protein biosynthesis could occur, suggesting that this 80-kDa protein associates with caveolin-3 in a time-dependent manner. However, on short exposures (such as the one shown here), there was no detectable signal around 20 kDa in the molecular mass range expected for caveolin-3.
Caveolins have the ability to form SDS-resistant oligomeric structures under certain conditions (23,24,28,29). These interactions are effectively disrupted in the presence of 100 mM NaOH. As we routinely subject all of our samples to NaOH treatment prior to gel loading, this excludes the possibility that the ϳ80-kDa band represents a trimeric form of caveolin-3. This and subsequent experiments show that the turnover rate of caveolin-3 in C2C12 myotubes is very slow, such that to visualize caveolin-3 by 35 S-methionine/cysteine metabolic labeling, very long autoradiographic exposures are required; similarly, the half-life of caveolin-1 has been shown to be ϳ24 -48 h. Also, it is important to note that the anti-caveolin-3 mono- clonal antibody used for these immunoprecipitation studies has been well characterized in our previous studies: (i) it recognizes an ϳ18 -20-kDa band by Western blotting that corresponds only to caveolin-3 and (ii) this antibody does not cross-react with other known members of the caveolin gene family, i.e. caveolins 1 and 2 (25).
As an internal control, half of the sample was immunoprecipitated with antibodies directed against a soluble protein, one of the GDP dissociation inhibitor proteins (GDI-3) (36). Lanes 1-4 show a constant signal for the 50-kDa GDI moiety, indicating that de novo protein synthesis was effectively blocked by cycloheximide.
Microsequence Analysis Identifies the ϳ80-kDa Protein as Phosphofructokinase-M-What is the identity of the ϳ80-kDa caveolin-3 associated protein? To address this issue, immunoprecipitations were scaled-up to obtain sufficient material for microsequence analysis of the 80-kDa region. In accordance with the data observed for 35 S-labeled extracts shown in Fig. 1, immunoprecipitation with anti-caveolin-3 IgG co-precipitated an 80-kDa protein from an unlabeled extract as visualized by silver staining (Fig. 2, lane 4). Note that this 80-kDa protein did not reflect one of the major proteins in the total extract (compare with lane 1), and its appearance was strictly dependent on the combination of three factors: (i) addition of C2C12 extracts; (ii) the presence of anti-caveolin-3 IgG, and (iii) the addition of protein A-Sepharose. These results exclude the possible fortuitous association of the ϳ80-kDa band with protein A-Sepharose.
Previously, we and others have demonstrated that caveolin-1 is difficult to visualize with conventional protein stains (Pon-ceau S, Coomassie Brilliant Blue, or silver staining) (39). Similarly, we note here that caveolin-3 also stains very poorly with silver staining (Fig. 2, lane 4).
Anti-caveolin-3 immunoprecipitates were next transferred to nitrocellulose, and the ϳ80-kDa region was excised after staining with Ponceau S and subjected to digestion with Lys-C. After digestion, a series of peptides were isolated by high performance liquid chromatography and subjected to microsequence

FIG. 3. Alignment of the rat muscle PFK-M isoform (top) with mouse liver PFK-B (middle) and rat brain PFK-C (bottom).
The four peptide sequences obtained by microsequence analysis are boxed (in bold). The full-length mouse PFK-M sequence is not available from any data base. However, mouse and rat sequences must be highly homologous, since the sequences obtained from the material isolated from C2C12 cells, a mouse cell line, are identical to the sequences found in the rat. Note that the protein sequence of PFK-M is clearly different from the protein sequences of PFK-B and PFK-C.  4) were pulselabeled for 20 min and chased for 3 h. The cells were then scraped into lysis buffer and either immunoprecipitated with anti-GDI antibodies (lanes 1 and 2) or anti-caveolin-3 IgG (lanes 3 and 4). Immunoprecipitates were analyzed by SDS-PAGE/fluorography. analysis. A total of 4 independent peptide sequences were obtained, and all of these peptides correspond to sequences within the muscle-specific isoform of phosphofructokinase (M-isoform; also called A-isoform), a known ϳ80-kDa protein (Fig. 3). These peptide sequences derive from regions clearly distinct from the B and C isoforms, unambiguously matching the sequence of the A isoform. This is in line with the notion that the M-or Aisoform is the most predominant form expressed in fully differentiated C2C12 cells.
Association of Phosphofructokinase-M with Caveolin-3 Occurs Only in Differentiated C2C12 Myotubes and Is Disrupted by High Ionic Strength-While all three PFK isoforms are expressed at the myoblast stage, PFK-M is strongly induced during myoblast differentiation to myotubes (40). Thus, the association of PFK-M with caveolin-3 should occur only in fully differentiated C2C12 myotubes, since there is no detectable caveolin-3 expression (both at the mRNA and protein level) in undifferentiated C2C12 myoblasts (24 -26). Fig. 4 (lanes 3 and 4) shows, as predicted, that PFK-M co-immunoprecipitates with anti-caveolin-3 IgG only in fully differentiated myotubes. To ensure that overall biosynthesis in myoblasts and myotubes was comparable, we immunoprecipitated these extracts with anti-GDI antibodies as a control for equal loading (Fig. 4, lanes 1 and 2). Note that the amount of GDI precipitated is constant regardless of the differentiation state.
We postulate that PFK-M and caveolin-3 form part of a stable hetero-oligomeric complex in vivo. However, one unlikely explanation for our current observations might be antibody cross-reactivity that would allow the anti-caveolin-3 IgG to directly recognize PFK-M. To test this possibility, we subjected anti-caveolin-3 immunoprecipitates to a high salt wash and observed its effect on the retention of PFK-M and caveolin-3 by anti-caveolin-3 IgG. Fig. 5 (upper panels) shows that retention of PFK-M was disrupted upon washing the immunoprecipitates with 500 mM NaCl. In contrast, the recovery of caveolin-3 in the immunoprecipitates remains unchanged (Fig. 5, lower panels; as indicated earlier, the gel was subjected to longer autoradiographic exposures to visualize the caveolin-3 signal). These results are consistent with the idea that PFK-M and caveolin-3 form a hetero-oligomeric complex, rather than simple cross-reactivity. This is further substantiated by the observation that anticaveolin-3 IgG recognizes an ϳ18 -20-kDa band that corresponds only to caveolin-3 by Western blot analysis; no crossreactivity with an ϳ80-kDa band has been observed (25).  at steady state A 10-cm plate of differentiated C2C12 cells was scraped into lysis buffer, divided into three equal aliquots, and immunoprecipitated with (i) protein A-Sepharose alone, (ii) protein A-Sepharose plus an irrelevant monoclonal antibody, or (iii) protein A-Sepharose plus an anticaveolin-3 monoclonal antibody. After immunoprecipitation, 100 l of each of these supernatants was analyzed by SDS-PAGE and Western blotting to determine the amount of nonprecipitated PFK-M. The signals were obtained by scanning densitometry. The signal obtained for protein A-Sepharose alone was taken as 100%. The PFK-M/Caveolin-3 Interaction Is Relevant in Vivo-While it is likely that complex formation between PFK-M and caveolin-3 has physiological relevance in vivo, this interaction might occur only after cell lysis. To evaluate this possibility, we chemically cross-linked the two proteins in intact cells prior to cell lysis and immunoprecipitation with anti-caveolin-3 IgG. We then took advantage of the salt sensitivity of the PFK-M/ caveolin-3 interaction demonstrated in Fig. 5. Under these conditions, only in vivo cross-linked material should be retained during the high salt wash, while noncovalently associated PFK-M would be dissociated. Fig. 6 shows the results of these cross-linking experiments. Intact differentiated C2C12 myotubes were subjected to in vivo cross-linking with a membrane-permeable homo-bifunctional cross-linker, dithiobis(succinimidyl propionate) (DSP). We and others have demonstrated that DSP is membrane-permeant (41,42). In addition, the two N-hydroxysuccinimide moieties in DSP are connected by a thiol-cleavable disulfide bond that readily allows dissolution of cross-linked material upon reducing SDS-PAGE. Note that an ϳ80-kDa band corresponding to PFK-M is observed only in samples pretreated with 500 M DSP prior to lysis (Fig. 6, right lane), while in the absence of DSP no signal in the 80-kDa range is detected.
These results clearly rule out post-lysis association between the two proteins. Furthermore, they suggest but do not prove that the interaction between PFK-M and caveolin-3 reflects direct physical contact between the two proteins as opposed to a third protein acting as a bridge between PFK-M and caveolin-3. As the efficiency of cross-linking two-member heterooligomeric complexes with N-hydroxysuccinimide-activated cross-linkers is generally quite low (Ͻ5%), it is unlikely that simultaneous cross-linking of three different components would be achieved with reasonable efficiency.
In further support of a direct interaction, rabbit PFK-M immobilized on Sepharose could bind and retain caveolin-3 from a mouse skeletal muscle cell lysate (Fig. 7). In contrast, a control Sepharose column containing immobilized protein A did not retain any detectable caveolin-3 as judged by Western blotting of SDS-treated column material.
The Majority of PFK-M Co-immunoprecipitates with Caveolin-3 at Steady State-We next attempted to quantitate the relative amount of PFK-M that co-immunoprecipitates with caveolin-3. Differentiated C2C12 cell lysates were divided into three equal parts, referred to as A, B, and C. Part A was loaded directly onto a SDS-PAGE gel to quantitate the total amount of PFK-M in the extract. Part B was immunoprecipitated with a nonimmune antiserum, and the remaining supernatant was loaded onto the gel. No significant decrease in the PFK signal was observed. Part C was immunoprecipitated with anti-caveo-  2 and 3 (lanes 2 and 3) were washed three times with DMEM lacking glucose and then incubated in the same medium for 60 min at 37°C. Plate 2 was then lysed. Plate 3 was incubated for an additional 60 min at 37°C standard DMEM containing 4.5 g/liter glucose. Plate 3 was then lysed as well, and all three lysates were immunoprecipitated with anti-caveolin-3 IgG and analyzed for the presence of PFK-M by SDS-PAGE/Western blotting (top panel). The same blot was also analyzed for the presence of caveolin-3 (bottom panel). As a control, ϳ5% of the cell lysate was analyzed to assess the total amount of PFK-M found in the lysate as a control for equal loading (center panel). B, the level of complex formation is proportional to the concentration of extracellular glucose. Four 10-cm plates of differentiated C2C12 cells were washed three times with glucose-free DMEM. Cells were then incubated for 60 min at 37°C in DMEM containing the indicated amounts of glucose, scraped into lysis buffer, and immunoprecipitated with anti-caveolin-3 IgG. These immunoprecipitates were analyzed for the presence of PFK by SDS-PAGE and Western blotting (lanes 1-4). Approximately 5% of the total lysate was analyzed for PFK to ensure that PFK is stable under these conditions (lower panel). C, dissociation of PFK-M from caveolin-3 occurs with rapid kinetics upon removal of extracellular glucose. Four 10-cm plates of differentiated C2C12 cells were pulse-labeled for 20 min and chased for 3 h. All four plates were then washed three times with glucose-free DMEM, incubated for the indicated amount of time at 37°C in the same medium, then scraped into lysis buffer and immunoprecipitated with anti-caveolin-3 IgG (lanes 1-4). The remaining supernatants were subsequently immunoprecipitated with anti-PFK antibodies to isolate PFK-M not associated with caveolin-3. lin-3 IgG. Immunoprecipitation of the lysate with anti-caveolin-3 IgG resulted in a reduction of the PFK-M signal by ϳ85-90%, indicating that under these conditions the bulk of PFK-M is associated with caveolin-3 ( Table I).
Association of PFK-M with Caveolin-3 Is Dependent on the Concentration of Extracellular Glucose-What is the functional significance of the interaction of PFK-M with caveolin-3? As PFK-M is a key regulator of the glycolytic pathway, its interaction with caveolin-3 might be dependent on the activation state of the enzyme and thus modulated by the concentration of extracellular glucose. High extracellular glucose is expected to activate both PFK-M and the glycolytic pathway; conversely, glucose starvation should have the opposite effect. Fig. 8 shows that interaction of PFK-M with caveolin-3 is strictly dependent on the concentration of extracellular glucose. All previously described experiments (Figs. 1-7) were performed with cells maintained in high glucose DMEM (4.5 g glucose/liter). To address the role of glucose in PFK-M-caveolin-3 complex formation, three identical plates of C2C12 myotubes were cultured and differentiated in high glucose medium. Cells from the first plate were immediately lysed without washing and immunoprecipitated with anti-caveolin-3 IgG, and the immuno-isolates were analyzed for the presence of PFK-M by SDS-PAGE and Western blotting (Fig. 8A). As expected, a strong signal for PFK-M was observed.
Cells from the second and third plates were incubated in glucose-free DMEM for 1 h. Plate 2 (Fig. 8) was then directly lysed, while plate 3 was incubated in high glucose medium for an additional 60 min after glucose starvation. Both samples were then analyzed for PFK-M associated with caveolin-3. While the levels of PFK-M and caveolin-3 remained constant in all three plates of cells (Fig. 8A, lower panels), this brief period of glucose starvation resulted in complete dissociation of the PFK-M/caveolin-3 complex (Fig. 8A, upper panel, lane 2). These findings cannot be attributed to cell death due to glucose deprivation, since complex formation was quantitatively restored after starvation upon transfer to high glucose DMEM and is therefore completely reversible (Fig. 8A, upper panel, lane 3).
Next, cells were incubated for 1 h in DMEM containing decreasing amounts of glucose; the amount of PFK-M associated with caveolin-3 was determined by SDS-PAGE and Western blot analysis of the immunoprecipitates. Fig. 8B shows that the levels of PFK-M associated with caveolin-3 are directly proportional to glucose levels in the DMEM, demonstrating that formation of the PFK-M-caveolin-3 complex is highly sensitive or responsive to the concentration of extracellular glucose.
Dissociation of PFK-M from caveolin-3 may occur very rapidly upon transfer to glucose-free DMEM. To observe the kinetics of this phenomenon, we pulse-labeled C2C12 myotubes for 20 min. The cells were then chased for 3 h in high glucose DMEM to allow newly synthesized PFK-M to undergo complex formation with caveolin-3 (see Fig. 1). Four separate plates of cells were then rapidly transferred to glucose-free medium and either lysed immediately (lane 1) or incubated for 2, 5, and 10 min in glucose free medium (lanes 2-4). All cells were then lysed and immunoprecipitated with anti-caveolin-3 IgG. Immunoprecipitates were analyzed by SDS-PAGE and fluorography (Fig. 8C). Supernatants were immunoprecipitated with anti-PFK antibodies to recover PFK not associated with caveolin- Lysates were then prepared, and various intracellular metabolites were added to a final concentration of 1 mM. After incubation for an additional 15 min at 37°C, insoluble material was removed by centrifugation, and samples were immunoprecipitated with either caveolin-3 IgG or with PFK antibodies to ensure the integrity of PFK under these conditions.
It is important to note that, under these conditions, we bias ourselves toward the newly synthesized pool of PFK-M that has not yet reached steady-state association with caveolin-3 (as judged from the time course presented in Fig. 1). Thus, focusing on the newly synthesized pool of PFK-M allows us to more clearly assess the positive effects of various intracellular metabolites, as a greater fraction of this PFK-M pool remains unassociated with caveolin-3. Note that this type of experiment does not reflect steady-state levels of total cellular PFK-M as seen by Western blot analysis in other experiments. Fig. 9 shows that complex formation between PFK-M and caveolin-3 is greatly promoted or stabilized by certain intracellular metabolites, such as fructose 1,6-bisphosphate and fructose 2,6-bisphosphate. A less dramatic positive effect was also observed with AMP and fructose 6-phosphate. Fructose 1,6bisphosphate, fructose 2,6-bisphosphate, fructose 6-phosphate, and AMP are all known allosteric activators of PFK. In contrast, allosteric inhibitors of PFK, citrate and ATP, tend to dissociate PFK-M from caveolin-3 as compared with the control sample processed in parallel. Note that two different exposures are shown for caveolin-3-associated PFK to illustrate linear exposures for all signals. These results are summarized in Table II. Thus, association of PFK-M with caveolin-3 is promoted by metabolites that are or resemble its physiological substrate that is derived from glucose, fructose 6-phosphate, and its product, fructose 1,6-bisphosphate. PFK-M is known to undergo feed-forward stimulation in the presence of fructose 1,6bisphosphate (product) which binds to a second allosteric site that is distinct from the active site of the enzyme.
Interaction of PFK-M with Caveolin-3, but Not Caveolin-1, in Bona Fide Skeletal Muscle Tissue-We have demonstrated that PFK-M and caveolin-3 form a hetero-oligomeric complex using skeletal C2C12 myotubes in culture. Does this interaction also occur in bona fide skeletal muscle tissue? Extracts were prepared from isolated mouse skeletal muscle fibers and subjected to immunoprecipitation with anti-caveolin-3 IgG or anti-caveolin-1 IgG. The corresponding immunoprecipitates were then analyzed for the presence of PFK-M by Western blotting (Fig.  10). As expected, anti-caveolin-3 IgG co-immunoprecipitate a band of ϳ80 kDa that is immunoreactive with anti-PFK antibodies. In contrast, no PFK-M co-precipitated with anti-caveolin-1 IgG, suggesting that caveolin-3 preferentially interacts with PFK-M.
Interestingly, when the same experiment was performed on a freshly lysed sample of mouse cardiac muscle, we were unable to co-immunoprecipitate PFK-M with either caveolin-1 or caveolin-3 IgGs (not shown). We do not yet know whether this reflects a difference in the metabolic states of the two striated muscle tissues or whether this phenomenon is truly specific only for skeletal muscle. A similar experiment in rat smooth muscle cells revealed that immunoprecipitation with anticaveolin-1 IgG does not co-immunoprecipitate PFK-M (not shown), confirming the notion that only caveolin-3 has the ability to stably interact with PFK-M under these conditions. DISCUSSION Here, we have demonstrated that under certain metabolic conditions PFK-M and caveolin-3 form a stable complex. Formation of this complex is relevant in vivo as judged by crosslinking experiments in intact C2C12 cells as well as co-immunoprecipitation of PFK-M with caveolin-3 in skeletal muscle tissue lysates. The level of complex formation can be modulated by the presence of various intracellular metabolites that are known allosteric activators or inhibitors of PFK activity. In line with these observations, PFK-M-caveolin-3 complex formation is exquisitely sensitive to extracellular glucose concentrations.
Since the interaction between PFK-M and caveolin-3 is observed under conditions of physiologic extracellular glucose, we believe it is an enzymatically active form of PFK that associates with caveolin-3. This type of interaction with caveolin-3 would recruit PFK-M to the muscle cell plasma membrane (sarcolemma), the site of glucose entry into the cell. In previous studies, we and others have shown that a fraction of the insulin-sensitive glucose transporter Glut4 partitions into caveolar membrane domains in adipocytes and that this fraction can be increased in response to insulin (35,43). While we have not studied this phenomenon in skeletal muscle cells that also express Glut4, it seems reasonable to hypothesize that glucose transporters and a key regulatory enzyme of glycolysis would be recruited to the plasma membrane and locally concentrated in caveolar domains. In support of this hypothesis, Glut4 has been previously localized to caveolae-like domains in skeletal muscle fibers by immunoelectron microscopy (44,45).
Adipocytes primarily express caveolins-1 and -2 (23, 35), have much less of an acute energy demand, and thus, would not require such a mechanism. Indeed, we were unable to find any evidence for an interaction of caveolin-1 with PFK-M despite fairly high sequence homology between caveolin-1 and caveolin-3 (85% similarity and 65% identity) (24). We and others have proposed the idea that caveolins-1, -2, -and -3 function as scaffolding proteins to locally concentrate molecules involved in signal transduction within caveolar microdomains of the plasma membrane (22,28,46). By analogy, caveolin-3 homooligomers could act as a plasma membrane-bound scaffold involved in meeting the high energy demand of skeletal muscle cells.
In line with this proposal, PFK-M is known to be regulated by a variety of serine and tyrosine kinases, which are signal- Inhibitor ϪϪ a A plus (ϩ) indicates that a given metabolite promotes complex formation, while a minus (Ϫ) denotes dissociation of the complex relative to the untreated control processed in parallel. transducing molecules (protein kinases A and C, and epidermal growth factor receptor; see Introduction).
PFK has been the subject of intense study for many years. Why has this interaction between PFK-M and caveolin-3 gone unreported? We believe that several reasons have allowed the PFK-M/caveolin-3 interaction to escape detection. First, the interaction between these two proteins is quite labile and may not survive a variety of harsh purification procedures. Second, conventional lysis protocols focus on the soluble fraction of PFK-M or lysis using the detergent Triton X-100. All caveolins described to date are not solubilized under these conditions and require the presence of additional detergents such as octyl glucoside (23,24). Third, and perhaps most important, all members of the caveolin gene family are very difficult to visualize. Caveolins stain poorly with conventional reagents such as Coomassie Blue, Ponceau S, or silver-staining protocols (39). Additionally, caveolin-3 has a very slow turnover rate, resulting in poor incorporation of 35 S-labeled amino acids during the course of pulse-chase experiments. The only way to effectively detect the presence of caveolin-3 is by Western blot analysis. We have only recently cloned the cDNA for caveolin-3 (24) and subsequently generated a caveolin-3-specific monoclonal antibody (25) that recognizes the unique N-terminal region of the protein that is not shared by other caveolin family members. Thus, molecular cloning of caveolin-3 and generation of a monospecific antibody has allowed us to identify PFK-M as a prominent caveolin-3-associated protein.
Importantly, we do not wish to imply that PFK-M represents the only protein interacting with caveolin-3. Caveolin-1 has been shown to interact directly with a number of signal transducing molecules, including H-Ras, heterotrimeric G proteins, Src-like kinases, and ecNOS (30,(47)(48)(49)(50)(51). We suspect that such interactions could also occur with caveolin-3, since peptides derived from caveolin-3 have similar inhibitory effects on the GTPase activity of G ␣ -subunits as caveolin-1 derived peptides (24). Thus, caveolar organization of signaling molecules and certain glycolytic enzymes could help to physically couple the generation of cellular energy with environmental cues provided by extracellular signals such as the concentration of glucose, and ligands that may activate growth factor receptors or G protein-coupled receptors.
While we and others have provided evidence that certain signaling molecules are concentrated in the inactive conformation within caveolae membranes (G proteins, Src-like kinases, protein kinase C) (30,47,52), activated Raf-1 is recruited to caveolae membranes in response to epidermal growth factorstimulation or Ras-mediated cell transformation (53). Thus, other molecules, such as PFK-M, may be recruited to plasmalemmal caveolae after activation by the appropriate stimuli.
As caveolin-3 is a component of the dystrophin complex (25), our current findings with PFK-M may also have implications for understanding the pathogenesis of Duchenne's and related muscular dystrophies. In support of this assertion, PFK-M demonstrates abnormal allosteric properties in mdx mice (54), suggesting that loss of dystrophin expression may adversely affect the regulated interaction of PFK-M with caveolin-3 in skeletal muscle fibers.