Glucose Activates Protein Kinase C-ζ/λ through Proline-rich Tyrosine Kinase-2, Extracellular Signal-regulated Kinase, and Phospholipase D

Insulin controls glucose uptake by translocating GLUT4 and other glucose transporters to the plasma membrane in muscle and adipose tissues by a mechanism that appears to require protein kinase C (PKC)-ζ/λ operating downstream of phosphatidylinositol 3-kinase. In diabetes mellitus, insulin-stimulated glucose uptake is diminished, but with hyperglycemia, uptake is maintained but by uncertain mechanisms. Presently, we found that glucose acutely activated PKC-ζ/λ in rat adipocytes and rat skeletal muscle preparations by a mechanism that was independent of phosphatidylinositol 3-kinase but, interestingly, dependent on the apparently sequential activation of the dantrolene-sensitive, nonreceptor proline-rich tyrosine kinase-2; components of the extracellular signal-regulated kinase (ERK) pathway, including, GRB2, SOS, RAS, RAF, MEK1 and ERK1/2; and, most interestingly, phospholipase D, thus yielding increases in phosphatidic acid, a known activator of PKC-ζ/λ. This activation of PKC-ζ/λ, moreover, appeared to be required for glucose-induced increases in GLUT4 translocation and glucose transport in adipocytes and muscle cells. Our findings suggest the operation of a novel pathway for activating PKC-ζ/λ and glucose transport.

In normal circumstances, insulin serves as the major regulator of glucose transport into skeletal muscles and adipocytes, and this regulation is effected by stimulating the translocation of GLUT4 and, to a lesser extent, GLUT1 and possibly other glucose transporters from internal membranes to the plasma membrane. This effect of insulin on glucose transporter translocation is mediated largely through the activation/action of phosphatidylinositol (PI) 1 3-kinase and 3-phosphoinositide-de-pendent protein kinase-1 (PDK-1) and their downstream effectors, viz. atypical protein kinase C (PKC) isoforms, and/or (1)(2)(3)(4), and protein kinase B (PKB; Akt) (5)(6)(7)(8). In diabetes mellitus, however, insulin is either deficient or ineffective, but, despite impaired insulin-stimulated glucose transport, whole body glucose disposal rates are surprisingly normal or increased (9 -11). The maintenance of glucose transport in diabetes mellitus apparently occurs at the expense of hyperglycemia, which concomitantly causes "glucotoxicity" (i.e. clinical insulin resistance and diabetic pathological complications). In promoting glucotoxicity, hyperglycemia most likely increases glucose uptake by simple mass action, but there is uncertainty as to whether hyperglycemia, like insulin, may also increases the translocation of glucose transporters to the plasma membrane (12)(13)(14)(15). This uncertainty is heightened by the fact that there is little or no insight into potential signaling mechanisms that glucose might use to stimulate glucose transporter translocation.
Presently, we evaluated the effects of glucose on signaling mechanisms that may be relevant to glucose transporter translocation/glucose transport in rat adipocytes and skeletal muscles. Our findings suggest that glucose does in fact increase GLUT4 translocation/glucose transport in these tissues, apparently by activating PKC-/ through a mechanism that is independent of PI 3-kinase but, interestingly, dependent upon both (a) the PYK2/GRB2/SOS/RAS/RAF/MEK1/ERK pathway, which we have recently reported to be activated by glucose (16), and (b) phospholipase D (PLD), which generates phosphatidic acid (PA), a known direct activator of atypical PKCs (17,18). Our findings further suggest that PLD functions downstream of the PYK2/ ERK pathway but upstream of PKC-/ during glucose action. This novel mechanism for activating PKC-/ and stimulating GLUT4 translocation sharply contrasts with that of insulin, which utilizes PI 3-kinase and PDK-1, rather than PYK2/ERK/ PLD, to activate PKC-/ and stimulate GLUT4 translocation/ glucose transport.

Studies in Rat Adipocytes
Rat Adipocyte Preparation and Incubation Conditions-Rat adipocytes were prepared by collagenase digestion of epididymal fat pads and either used directly or transfected and cultured overnight as described * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
For studies of epitope-tagged ERK activation, as described (16,19), 0.8 ml of 50% rat adipocyte suspension in DMEM was co-transfected by electroporation with 3 g of pCEP4 encoding HA-ERK2 (kindly supplied by Dr. Melanie Cobb; see Refs. 16 and 19), along with 7 g of pRK5 alone (vector) or pRK5 encoding wild-type or kinase-inactive PYK2 (see Refs. 16 and 24). After overnight incubation in DMEM/BSA medium to allow time for expression, cells were washed and incubated in glucosefree KRP medium containing 1% BSA, with or without added D-glucose or insulin, following which HA-ERK was precipitated with anti-HA antibodies and assayed as described (16,19).
For studies of epitope-tagged PKC-activation, as described (20,21,23), 0.8 ml of 50% rat adipocyte suspension in DMEM was co-transfected by electroporation with 1 g of pCDNA3 encoding HA-tagged PKC-or pCMV5 encoding FLAG-tagged PKC-, along with the above described plasmids encoding dominant-negative forms of PYK2, GRB2, SOS, RAS, or ERK1 or ERK2 (kinase-inactive forms of ERK were kindly supplied by Dr. Melanie Cobb). After overnight incubation in DMEM/ BSA medium to allow time for expression, cells were washed and incubated in glucose-free KRP medium containing 1% BSA, with or without added D-glucose or insulin, following which HA-PKC-or FLAG-PKC-was precipitated with anti-HA or anti-FLAG antibodies and assayed as described (20,21,23).
Endogenous GLUT4 and GLUT1 Translocation-Translocation of endogenous GLUT4 and GLUT1 was measured by immunoblotting plasma membranes and low density microsomes after purification by ultracentrifugation as described (25). Mouse monoclonal anti-Glut4 antibodies were obtained from Biogenesis, and rabbit polyclonal anti-GLUT1 antiserum was kindly provided by Dr. Ian Simpson.
Phospholipase D Assays-PLD was assayed as described (26,27). In brief, adipocytes were incubated for 16 -20 h with [ 3 H]oleic acid (10 Ci/ml DMEM medium) (PerkinElmer Life Sciences) to prelabel phosphatidylcholine (PC) and other phospholipid pools. The cells were then washed and equilibrated for 15 min in glucose-free KRP medium containing 1.5% ethanol (which, as a primary alcohol, competes with water during PLD-dependent displacement of choline from PC to yield phosphatidylethanol instead of PA) and then treated for 15 min with or without glucose. After incubation, authentic phosphatidylethanol was added to each sample, lipids were extracted, and phosphatidylethanol was isolated by TLC and counted for radioactivity as described (26,27).
Other Assays-Immunoprecipitable activities of total cellular or epitope-tagged forms of PKC- (1,2,20,21,23), PKB (21), and ERK (16,19) were measured as described. Tyrosine phosphorylation of specific peptide sequences in PYK2 was detected by Western analyses using antisera obtained from BIOSOURCE as described (16). Phosphorylation of threonine 410 in the activation loop of PKC-by Western analysis and autophosphorylation of PKC-in vitro were determined as described (21,23). Note that autophosphorylation was conducted in the presence of phosphatidylserine and phosphopedic acid has little additional effect on autophosphorylation (see Ref. 17).

Studies in Skeletal Muscles
Rat Skeletal Muscle Preparations and Incubation Conditions-As described previously (28), soleus and extensor digitorum longus muscles were ligated at both ends and stretched to maintain their resting lengths, slit lengthwise into strips to improve diffusion, and incubated at 37°for 30 min under 95% O 2 /5% CO 2 in Krebs-Ringer bicarbonate buffer containing 5 mM glucose and, where indicated, inhibitors. After this equilibration period, 25 mM mannitol (to provide an osmotically balanced "control") or 25 mM D-glucose was added along with tracer quantities of 2-[ 3 H]deoxyglucose (2 Ci/ml) to reflect glucose uptake and L-[ 14 C]glucose (0.2 Ci/ml) to correct for trapping of extracellular medium and associated nonspecific radioactivity (see Ref. 28). Incubation was continued for 30 min, after which, as described (29), muscle tissues were rinsed and homogenized (by Polytron) and centrifuged at 1000 rpm to remove debris and nuclei, and cell lysates were examined for immunoprecipitable PKC-/ activity (see Ref. 29) and uptake of glucose, calculated on the basis of the specific radioactivity of 2-[ 3 H]deoxyglucose per nmol of glucose, assuming that glucose and 2-deoxyglucose are similarly transported.

Studies in Rat Adipocytes
Effects of Glucose on Glucose Transporter Translocation-Over a concentration range of 0 -30 mM, glucose provoked in-cremental dose-dependent increases in Myc-GLUT4 translocation to the plasma membrane of rat adipocytes (Fig. 1A). Glucosedependent increases in Myc-GLUT4 translocation were maximal within 10 min (Fig. 1B) and largely sustained for 30 min (Fig. 1B) but tended to diminish at longer times of incubation, viz. at 60-240 min (Fig. 2B). Insulin-induced increases in GLUT4 translocation appeared to develop more slowly over 30 min (Fig. 1B), were surprisingly similar to or only modestly greater than those of 20 -30 mM glucose (Fig. 1, B and C; also see below), and were not additive to those of glucose (Fig. 1C). This nonadditivity may reflect the fact that relevant signaling pathways used by insulin and glucose, while initially different, appeared to converge distally, as described below.
Effects of glucose on Glut4 translocation were not due to hyperosmolality, as 20 -30 mM mannitol (see Fig. 1A) or Lglucose (data not shown) had little or no effect. Of further note, glucose-dependent Myc-GLUT4 translocation was dependent upon the continued presence of glucose and was lost within 5 min of glucose removal (Fig. 1D).
In addition to increasing the translocation of transfected epitope-tagged GLUT4, glucose, like insulin, increased the FIG. 2. Acute effects of glucose and insulin on translocation of endogenous GLUT4 and GLUT1 glucose transporters to the plasma membrane of rat adipocytes (A) and more prolonged effects of glucose on Myc-GLUT4 translocation in rat adipocytes (B). A, freshly isolated adipocytes were incubated in glucose-free KRP medium without treatment, incubated for 15 min with 25 mM glucose, or incubated for 30 min with 10 nM insulin. After incubation, plasma membranes and low density microsomes were isolated by ultracentrifugation and blotted for GLUT4 and GLUT1. Shown at the top of A are representative immunoblots, below which are shown bar graphs that depict mean Ϯ S.E. of percentage changes (relative to the control, set at 100%) of four determinations, as quantitated by measurement of chemiluminescence in a Bio-Rad Chemiluminescence/Phosphorescence Molecular Analyst Imaging System. All changes in levels of GLUT4 and GLUT1 in plasma membranes and microsomes were significant to at least p Ͻ 0.05 (paired t test). B, adipocytes were transfected with plasmids encoding Myc-GLUT4, and after overnight incubation in DMEM/BSA medium, cells were incubated in glucose-free KRP medium for 240 min without or with 20 mM glucose being present for the indicated times, i.e. prior to termination of the experiment. Values are mean Ϯ S.E. of four determinations. translocation of endogenous GLUT4 from microsomes to the plasma membranes of rat adipocytes (Fig. 2). Also, glucose, like insulin, increased endogenous GLUT1 (Fig. 2), as well as epitope-tagged GLUT1 (Fig. 3A), translocation to the plasma membrane.
Effects of Inhibitors on Glucose-stimulated GLUT4 Translocation-In contrast to insulin effects, glucose-dependent increases in Myc-GLUT4 translocation were not inhibited by the PI 3-kinase inhibitor, wortmannin (Fig. 1B). Further, as seen in Fig. 3B, glucose effects on Myc-GLUT4 translocation were inhibited by the tyrosine kinase inhibitor, genistein, the MEK1 inhibitor, PD98059, and dantrolene, which inhibits an internal Ca 2ϩ pool required for (among other things) PYK2 activation (24), but not by nifedipine, which inhibits Ca 2ϩ uptake through L-type channels. Also, although glucose increases de novo diacylglycerol synthesis and activates PKC in rat adipocytes (30,31), GF109203X, in concentrations that inhibit diacylglycerolsensitive PKCs (3-5 M), did not inhibit glucose-dependent Myc-GLUT4 translocation (Fig. 3B). The latter observation correlated well with the fact that glucose activates ERK independently of PKC in rat adipocytes (16). Note, however, that higher concentrations of GF109203X (20 -30 M) that effectively inhibit atypical PKCs markedly inhibited glucose effects on GLUT4 translocation (data not shown).
In comparison with glucose, insulin-stimulated Myc-GLUT4

FIG. 4. Effects of wild-type (WT) and dominant negative (DN) forms of PYK2 (kinase-inactive (KI)) (A), GRB2 (B), SOS (C), and RAS (D) on glucoseand insulin-stimulated translocation
of epitope-tagged GLUT4 to the plasma membrane of rat adipocytes. Adipocytes were co-transfected with plasmids encoding epitope-tagged GLUT4 and indicated signaling proteins. After overnight incubation in DMEM/BSA medium, cells were washed and incubated for 10 min in glucose-free KRP medium containing 1% BSA, with or without 20 mM glucose or 10 nM insulin, as indicated, following which the cell surface level of Myc-or HA-GLUT4 was measured. Values are mean Ϯ S.E. of n, the number of determinations, shown in parentheses.  Freshly isolated adipocytes were incubated in glucose-free KRP medium containing 1% BSA for the indicated times with 25 mM glucose, following which lysates were subjected to SDS-PAGE and blotted for specific tyrosine residues in PYK2 or examined for immunoprecipitable enzyme activity. Shown in A are mean values of three or four determinations. translocation was inhibited, as expected, by genistein (Fig. 3C) and wortmannin (Fig. 1B), but not by dantrolene or PD98059 (Fig. 3C).
In contrast to the inhibitory effects of the MEK1 inhibitor, PD98059, on glucose-induced increases in epitope-tagged GLUT4 translocation, the inhibitor of the p38 MAPK pathway, SB202190, was without effect on this translocation (data not shown). Thus, although glucose may activate p38, as well as ERK, MAPK, the ERK pathway appeared to be more important for glucose-dependent GLUT4 translocation.
Studies on the Role of the PYK2/ERK Pathway during Glucose-stimulated GLUT4 Translocation-The above described findings suggested that glucose stimulated GLUT4 translocation via a signaling pathway that is dependent on a tyrosine kinase, a dantrolene-sensitive Ca 2ϩ pool, and MEK1/ERK. Germane to these findings, it was previously shown that activation of the nonreceptor tyrosine kinase, PYK2, requires dantrolenesensitive Ca 2ϩ , and the ERK pathway can be activated by PYK2, which, via phosphotyrosine 402, activates SRC and phosphorylates SHC, and/or via phosphotyrosine 881, directly interacts with the SH2 domain of GRB2 (24). It was also previously shown that glucose rapidly activates (a) PYK2 by in-creasing phosphorylation of tyrosine 402, the autophosphorylation site, tyrosine 580, an activating residue in the catalytic domain, and tyrosine 881, the GRB2 interaction site, in PYK2, and (b) ERK by a mechanism that is inhibited by dantrolene and PD98059 and expression of dominant-negative forms of PYK2, GRB2, SOS, RAS, and MEK1 (16).
In keeping with the aforesaid requirements for PYK2, GRB2, In A, C, and D, cells were co-transfected with plasmids encoding epitope-tagged PKCand dominant-negative forms of indicated signaling proteins, incubated overnight in DMEM/BSA medium, and then washed and incubated in glucose-free KRP medium containing 1% BSA for 10 min with or without 20 mM glucose, following which epitope-tagged PKCwas immunoprecipitated and assayed. In B, freshly isolated adipocytes were incubated first for 15 min with indicated inhibitors and then for 10 min with or without 20 mM glucose, following which PKCwas immunoprecipitated and assayed. Values are mean Ϯ S.E. of four determinations. C, control; G, glucose.

TABLE I Effects of wortmannin, dantrolene, PD98059, and expression of kinase-inactive PDK-1 and kinase-inactive ERK-2 on insulin-induced activation of atypical PKCs in rat adipocytes
In the top and middle, adipocytes were incubated in glucose-free KRP medium for 15 min with or without the indicated inhibitors and then treated for 10 min with or without 10 nM insulin. At the bottom, adipocytes were co-transfected with FLAG-tagged PKC-and kinaseinactive forms of PDK-1 or ERK-2, and, after overnight incubation, cells were washed and incubated for 10 min in glucose-free KRP medium with or without 10 nM insulin. After incubation, total PKC-/ (top and middle) or FLAG-PKC-(bottom) was immunoprecipitated and assayed. Values are mean Ϯ S.E. of n determinations. C, control, G, glucose.  3B) but also by expression of (a) dominant-negative, but not wild-type, forms of PYK2 (Fig. 4A) and GRB2 (Fig. 4C) and (b) dominant-negative forms of SOS (Fig. 4B) and RAS (Fig. 4D).
In contrast, expression of dominant-negative forms of PYK2, GRB2, SOS, and RAS had little or no effect on insulin-dependent epitope-tagged GLUT4 translocation (Fig. 4, A-D), despite the fact that expression of these dominant-negative signaling proteins effectively inhibits insulin effects on epitope-tagged ERK (19). Thus, unlike the insulin-sensitive ERK pool, the glucosesensitive ERK pool is required for GLUT4 translocation.
In keeping with the findings suggesting that glucose-stimulated, but not insulin-stimulated, GLUT4 translocation was dependent on PYK2 and PYK2-mediated activation of the ERK pathway, glucose-stimulated, but not insulin-stimulated, 2-[ 3 H]deoxyglucose (and therefore glucose) uptake was inhibited by the PYK2 inhibitor, dantrolene, and the MEK1 inhibitor, PD98059 (Fig. 5). Thus, findings on epitope-tagged GLUT4 translocation appeared to be reflective of glucose uptake.
Studies on the Activation of PKC-/ by Glucose and the Role of PKC-/ during Glucose-stimulated GLUT4 Translocation-In addition to PYK2 and ERK, glucose rapidly activated total cellular PKC-/ but not PKB (Fig. 6). Moreover, PKC-/ appeared to be required for glucose-as well as insulin-stimulated translocation of epitope-tagged GLUT4 to the plasma membrane, as evidenced by inhibitory effects of both the cellpermeable myristoylated (see Ref. 2) PKC-pseudosubstrate and expression of kinase-inactive PKCon glucose-as well as insulin-stimulated GLUT4 translocation (Fig. 7).
The failure of glucose to activate PKB, and the fact that wortmannin did not inhibit glucose effects on glucose-stimulated GLUT4 translocation suggested that glucose activated PKC-/ and GLUT4 translocation independently of PI 3-kinase. Further, since both the PYK2/ERK pathway and PKC-/ appeared to be required for glucose-stimulated GLUT4 translocation and since PKC-/ is not required for glucose-induced activation of the PYK2/ERK pathway (16), we examined the possibility that the PYK2/ERK pathway may serve upstream of PKC-/ during glucose action. Indeed, as shown in Fig. 8, expression of dominant-negative forms of PYK2, GRB2, SOS, RAS, and ERK1 and -2, as well as dantrolene and PD98059, chemical inhibitors of PYK2 and MEK1, respectively, but not the PI 3-kinase inhibitor, wortmannin, inhibited glucose-induced activation of epitope-tagged PKC-. Expression of kinase-inactive PDK-1, like wortmannin, did not inhibit glucoseinduced activation of epitope-tagged PKC-(3.79-and 2.94-fold increases in presence and absence of PDK-1). In marked contrast, insulin-induced increases in PKC-/ were inhibited, as expected, by wortmannin and expression of PDK-1, but not by dantrolene or PD98059 or by expression of kinase-inactive ERK ( Table I).
The above described findings suggested that glucose activated PKC-/ by a mechanism that is independent of PI 3-kinase and decidedly different from that of insulin, which uses PI  A and B, freshly isolated adipocytes were used. In C, cells were transfected with plasmids encoding HA-GLUT4 and incubated overnight in DMEM/BSA medium. All cells were finally incubated in glucose-free KRP medium containing 1% BSA, first for 5 min with or without 1% n-butanol and then for 15 min with or without 25 mM glucose, as indicated. After incubation, lysates were assayed for immunoprecipitable PKC-/ and ERK activity and cell surface levels of HA-GLUT4. Values are mean Ϯ S.E. of n, the number of determinations, shown in parentheses.

FIG. 10. Effects of butanol on glucose-induced increases in PKC-/ activity (A), ERK activity (B), and HA-GLUT4 translocation (C) in rat adipocytes. In
3-kinase, PIP 3 , and PDK-1 to provoke sequential increases in phosphorylation of threonine 410 in the activation loop and threonine 560, the autophosphorylation site of PKC- (23). It was therefore of interest to find that, unlike insulin (see Refs. 21 and 23), glucose did not provoke increases in either the autophosphorylation of PKC-/ in vitro, which primarily reflects phosphorylation of threonine 560 of PKC- (21,23) or phosphorylation of the threonine 410 activation loop site of PKC-in intact adipocytes (Fig. 9).
Studies on the Activation of PLD by Glucose and the Role of PLD in Glucose-stimulated PKC-/ Activation and GLUT4 Translocation-With respect to alternative PI 3-kinase-independent mechanisms for activating PKC-, PA has been reported to directly activate PKC-by a mechanism that, like that used by PIP 3 , is presumably allosteric in nature, but, unlike that of PIP 3 , is not attended by increases in autophosphorylation (17), i.e. similar to the apparently phosphorylationindependent mechanism used by glucose, as described above. Of further interest, PA is generated by the action of a PCspecific PLD, which has also been shown to activate PKC- (18). It was therefore interesting to find that glucose-induced activation of PKC-was inhibited both by 1% n-butanol (Fig. 10A), which, as a primary alcohol, inhibits the production of PA via PC-PLD by competition with water during the PLD-dependent displacement of choline from PC, thus yielding phosphatidylbutanol instead of PA. Glucose-induced activation of PKC-/ was also inhibited by D609, an inhibitor of PLD and PLC (data not shown). This inhibitory effect of butanol on PKC-/ activation could not be considered to be nonspecific, since butanol did not inhibit insulin-induced PKC-/ activation (data not shown).
In addition to inhibiting glucose-stimulated PKC-/ activation, butanol inhibited glucose-stimulated HA-GLUT4 translocation (Fig. 10B). In contrast, butanol did not inhibit glucose-stimulated ERK activation (Fig. 10C). Thus, butanol apparently inhibited an event that was downstream of ERK but upstream of PKC-/ and GLUT4 translocation during glucose action.
In support of the possibility that PLD may function downstream of ERK and upstream of PKC-/ and glucose transport during glucose action, glucose activated PLD by a PD98059sensitive mechanism (Fig. 11A). Furthermore, the addition of exogenous PLD activated PKC-/ (Fig. 11B) and stimulated 2-deoxyglucose uptake (Fig. 11C), and these effects of PLD were not inhibited by MEK1 inhibitors, PD98059 (Fig. 11, B and C) or UO126 (data not shown). (Note that adenovirusmediated expression of kinase-inactive PKC-inhibited effects of exogenously added PLD on 2-deoxyglucose uptake in L6 myotubes (data not shown), but it was not possible to use adenoviruses in rat adipocytes to study transport, since these adipocytes became leaky, and stimulatory effects of agonists were no longer evident.) These findings were compatible with the possibility that PLD, which is activated by glucose, could function distal to ERK, but proximal to PKC-/, during glucose action.

Studies in Rat Skeletal Muscles
As might be expected, glucose uptake was substantially higher in slow twitch soleus and fast twitch extensor digitorum longus skeletal muscles incubated in medium containing higher glucose concentrations (Fig. 12). Of particular interest, however, as in rat adipocytes, glucose provoked ϳ2-fold increases in PKC-/ activity in both muscles (Fig. 12), and the MEK1 inhibitor, PD98059, inhibited glucose-induced increases in both glucose uptake and PKC-/ activity, as tested in the extensor digitorum longus muscle (Fig. 12). In this regard, note that dantrolene, an inhibitor of the internal pool of Ca 2ϩ required for PYK2 activation, has already been reported to in- hibit glucose-dependent increases in glucose uptake in rat skeletal muscles (32). Taken together, these results suggested that, as in adipocytes, glucose activates PKC-/ in skeletal muscle through the PYK2/ERK pathway, and, moreover, it seems reasonable to propose that this PYK2/ERK-dependent activation of PKC-/ contributes importantly to increases in glucose uptake observed during incubation of skeletal muscles in supraphysiological glucose concentrations. DISCUSSION It was particularly interesting to find that glucose rapidly activated PKC-/ in rat adipocytes and also apparently in skeletal muscles by a novel mechanism requiring PYK2, GRB2, SOS, RAS, MEK1, ERK1/2, and PLD. This mechanism is decidedly different from that used by insulin, which activates PKC-/ via PI 3-kinase-dependent increases in PIP 3 levels and PDK-1 action (2,21,23), independently of PYK2 and the ERK pathway.
The mechanism whereby glucose used the PYK2/ERK pathway to activate PLD and PKC-/ is not entirely clear, but there are several important precedents that should be mentioned. First, PLD has been found to be activated via ERK during formyl-Met-Leu-Phe action in human neutrophils (33); however, other than for the present study, to our knowledge, there are no other reports indicating that PLD operates downstream of ERK. Second, PA can directly activate PKC-without increasing the autophosphorylation of PKC- (17). Third, PLDinduced increases in PA have been reported to activate PKC- (17,18). Thus, it was not entirely surprising to find that (a) activation of the PYK2/ERK pathway could result in PLD activation and (b) PLD activation could result in PKC-/ activation. On the other hand, our findings provided the first evidence that (a) an agent can sequentially activate the nonreceptor tyrosine kinase, PYK2, the GRB2/ERK pathway, PLD, and PKC-/ and (b) glucose activates this PYK2/ERK/ PLD PKC-pathway in rat adipocytes, and, most likely, in skeletal muscle cells as well.
With respect to alternative non-PLD/PA mechanisms whereby ERK could conceivably activate PKC-/, it has been reported that ERK-activated MAPK-activated kinase-2 (34,35) and MSK-1 (36) can act like PDK-1 and/or PDK-2 and phosphorylate PKB on serine 308 and threonine 473 residues. However, glucose did not activate PKB in our studies, and, unlike our findings with insulin (21, 23), we did not observe increases in phosphorylation of PKC-following glucose treatment. Accordingly, it seemed unlikely that these ERK-activated kinasedependent mechanisms could explain glucose-induced increases in PKC-/ activation in our studies. The finding that GLUT4 translocation/glucose transport could be increased by a mechanism that appeared to be dependent on PKC-/ and upstream activators, PYK2/ERK/PLD, but independent of PI 3-kinase is noteworthy for several reasons. First, it suggested that, in addition to PI 3-kinase, PI 3-kinasedependent signaling factors, including PKB, are not essential for PKC-/-dependent effects on the glucose transport system. Second, agents that activate PYK2/ERK/PLD may potentially activate PKC-/ and the glucose transport system. However, it remains uncertain if agents, such as glucose, that use PYK2/ ERL/PLD to activate PKC-/ act on the same GLUT4 pool(s) or use similar distal translocating mechanisms as agents that use PI 3-kinase to activate PKC-/, such as insulin.
In comparing insulin effects with those of glucose, it may be noted that, although insulin provokes equally strong increases in ERK activity (16), unlike glucose, insulin does not activate PYK2, and it is considerably less potent than glucose for increasing PLD activity in rat adipocytes. 2 Moreover, inhibitors of PYK2, ERK, and PLD, despite largely inhibiting glucose effects, had little or no effect on insulin-induced increases in PKC-/ activation. On the opposite side of the coin, insulininduced increases in PKC-/ activity and GLUR4 translocation/glucose transport were, unlike those of glucose, completely or nearly completely dependent on PI 3-kinase. At present, we can only speculate as to why PKC-/ and GLUT4 translocation were activated as a consequence of glucose-induced, but not insulin-induced, ERK activation. Possibly relevant to this question is the fact that glucose uses PYK2 to activate the GRB2/ERK pathway (16), whereas insulin uses IRS-1 and/or SHC to activate the GRB2/ERK pathway. In this respect, it may be relevant to note that the nonreceptor tyrosine kinase PYK2 is dependent on a dantrolene-sensitive Ca 2ϩ pool, and both tyrosine kinases and Ca 2ϩ have been implicated in PLD activation. Perhaps it is the combination of both PYK2 and ERK that is important for assembling a signaling complex that results in strong PLD and subsequent PKC-/ activation during the action of glucose. Further work is needed to explore these possibilities.
The extent of inhibitory effects of dantrolene and PD98059 on glucose-stimulated 2-[ 3 H]deoxyglucose uptake and the failure of these agents to alter insulin-stimulated 2-[ 3 H]deoxyglucose uptake are also noteworthy. These findings suggested that (a) these agents did not act in a nonspecific manner and (b) increases in GLUT4/1 translocation, rather than a simple increase in mass action, contributed importantly to glucose-dependent increases in glucose uptake.
It should be noted that the presently observed acute effects of glucose on PKC-/ and GLUT4 translocation/glucose transport seemed to wane with time. In this regard, we have not observed PKC-/ activation or increases in glucose transport in adipocytes or skeletal muscles in a rat model of type 2 diabetes mellitus in which there is stable chronic elevation of serum glucose levels (29). Thus, as with persistent insulin action, the relatively acute effects of glucose on PKC-/ and the glucose transport system observed presently may be subject to downregulation in states of persistent hyperglycemia.
Finally, we have also observed activation of atypical PKCs by certain agonists that, like glucose, activate PLD through antecedent activation of the PYK2/ERK pathway. However, it is presently unclear if other mechanisms of PLD activation result in atypical PKC activation. Further studies are needed to answer this question.
In summary, our findings define a novel mechanism for activating atypical PKCs that requires the assembly and/or activation of a signaling complex that includes PYK2, components of the GRB2/ERK pathway, and PLD. Presumably, the PA generated by PLD action is directly responsible for activation of atypical PKCs presently seen during glucose action. Of further interest, this mechanism for atypical PKC activation is apparently used by glucose to stimulate the glucose transport system in adipocytes and skeletal muscles. Further work is needed to see if other agonists use this novel mechanism.