Glucagon and the Ca2+-linked Hormones Angiotensin 11, Norepinephrine, and Vasopressin Stimulate the Phosphorylation of Distinct Substrates in Intact Hepatocytes*

Recent studies have demonstrated that angiotensin 11, catecholamines, and vasopressin can stimulate the phosphorylation of hepatic cytosolic proteins via a Caz+-linked cyclic AMP-independent mechanism. The present study used high resolution, two-dimensional gel electrophoresis to determine if the proteins phosphorylated in response to the Ca2+-linked hormones were distinct from those affected by glucagon acting via the cyclic AMP-dependent pathway. Intact hepato- cytes labeled with [32P]P043- were stimulated with glucagon, angiotensin 11, 1-norepinephrine, and vasopres- sin and over 100 phosphorylated proteins resolved by two-dimensional electrophoresis and autoradiography. Six important enzymes known to be regulated through covalent modification were positively identified, in- cluding phosphorylase, phosphofructokinase, pyruvate kinase, fructose-6-phosphate 2-kinase, phenylalanine hydroxylase, and fructose-1,6-bisphosphatase. Computer analysis of the autoradiograms from control and hormone-treated cells demonstrated that glucagon increased the phosphorylation state of 12 phosphopro- teins and reduced the phosphorylation of one protein with a M,. = 21,000 and a PI = 5.9. The Ca2+-linked hormones stimulated the phosphorylation of 7 phosphoproteins and also reduced the phosphorylation state of the 21,000-dalton protein. Angiotensin 11, 1-norepinephrine, and vasopressin had equivalent effects on protein phosphorylation. There were six protein substrates uniquely affected by glucagon and one phos- phoprotein uniquely stimulated by the Ca2+-linked hormones. Seven substrates were affected by stimulation of the cell with either glucagon or the Ca2+-linked hormones. These results demonstrate that, while there is overlap in the substrates affected by glucagon and the Ca2+-linked hormones, each pathway is able to affect the phosphorylation of unique substrates. This finding suggests that the two types of hormones may have some distinct effects on hepatic function.

a Ca2+-linked pathway that is independent of cyclic AMP (see Refs. 1-4 for reviews). Although these two groups of hormones act via different mechanisms, their net effects on carbohydrate metabolism are similar, with both stimuli leading to increased glycogenolysis (1)(2)(3)(4) and gluconeogenesis (5)(6)(7)(8)(9)(10). Moreover, the two types of hormones appear to regulate metabolism at similar enzymatic sites and by the same molecular mechanism. For example, both glucagon and the Ca"-linked hormones increase glycogenolysis by stimulating phosphorylase activity (11)(12)(13)(14)(15) and inhibiting glycogen synthase activity (8,11,15,16) through an increase in the phosphorylation state of these proteins (14,15). The two pathways also appear to converge on several other phosphoproteins in the hepatocyte (14,15). These observations have been interpreted to mean that aadrenergic agonists, angiotensin 11, and vasopressin stimulate the activity of one or more Ca'+-sensitive protein kinases (14,15). Although a number of candidates exist (17)(18)(19)(20)(21), to date the relevant kinase(s) have not been identified or studied in detail.
While most studies have stressed the similarities between the glucagon and Ca2+-linked pathways in the hepatocyte (1,2), differences have been noted in the ability of the two pathways to phosphorylate and alter the activity of certain enzymes such as pyruvate kinase (15). More importantly, there is evidence that each of the pathways may have unique substrates (14,15). A clear demonstration of distinct substrates phosphorylated in response to glucagon and the Ca2+linked hormones would provide compelling evidence that these pathways activate different protein kinases. Moreover, knowledge of the differences between the efficacy and substrate specificity of the glucagon and Ca"-linked pathways will help us understand to what extent each pathway is involved in metabolic events within the cell.
In order to define the substrate specificity of the cyclic AMP and Ca2+-linked pathways, hepatocytes were labeled with [:"PplPO4"-, treated with glucagon or one of the Ca'+linked hormones and the cytosolic proteins resolved on twodimensional polyacrylamide gels. The "'P-labeled proteins were visualized by autoradiography and the autoradiographs were analyzed with the aid of a computer. The increased resolution provided by these techniques allowed positive identification of a number of enzymes in the gel pattern and demonstrated that glucagon and the Ca2+-linked hormones stimulate the phosphorylation of separate but overlapping groups of substrates.

Preparation of Hepatocytes, Incubation Procedures, and Sample
Preparation-Isolated liver cells were prepared from 200-to 300-g fasted, male rats by published methods (22). The hepatocytes were resuspended in a low phosphate (0.1 mM) Krebs

Identification of Proteins Phosphorylated in Hepatocytes
gassed with 95% Os, 5% CO'. The effect of the various hormones on protein phosphorylation in the intact cell was monitored according to published methods (14,15). Briefly, the exchangeable phosphate pools in the hepatocytes were labeled by incubating the cells at 37°C with 16 mM L-lactate, 4 mM pyruvate, and [3*P]P043for 45 min, the cells stimulated with hormones for 3-5 min, and the incubations terminated by one of two methods. The doses of hormone and the incubation times were chosen to give a maximal response (15). In early experiments, the incubations were stopped by diluting the cells 10-fold into an ice-cold buffer containing 10 mM Tes', pH 7.4, 100 mM NaF, 10 mM EDTA, and 80 m y sucrose, and the cytoplasmic proteins were prepared following homogenization as described (14). In later experiments, the cytosolic fraction of the cell was prepared by lysing the plasma membrane with digitonin according to the procedure of Janski and Cornell (23). In this procedure, the ["2P]P04"-labeled cells were diluted 5-fold into a slightly hypertonic fractionation medium containing 4 mg/ml of digitonin, 10 mM Tes, pH 7.4, 50 mM NaF, 10 mM EDTA, 5 mM EGTA, and 200 mM sucrose, and incubated for 6-8 s to release the cytosolic proteins. The remaining cell structures were removed by centrifugatioq through a hydrocarbon layer ( p = 1.05) a t 13,000 X g i n a n Eppendorf microfuge. The microfuge tubes containing the digitonin fractionation medium were maintained at 27 "C before use and the time of incubation was adjusted to insure 90-95% release of the cytoplasmic enzyme lactate dehydrogenase (23). The advantages of the digitonin fractionation procedure are that the cytosolic proteins are removed from the cell in less than 10 s and there is little contamination from mitochondrial, endoplasmic reticulum, or nuclear proteins (23). Moreover, since the proteins can be boiled immediately in SDS, the possibility of proteolytic activity is minimized. Although the time elapsed between breakage of the cell and boiling the cytosolic proteins in SDS ranged from 30-40 s for the digitonin fractionation procedure to 70 min for the homogenization procedure, no major differences were observed in either the two-dimensional protein or autoradiographic patterns. This result tends to validate the use of either method to obtain cytoplasmic proteins.
The concentration of the cytosolic proteins obtained by either the homogenizationor the digitonin fractionation procedure was adjusted to equality (5-7 mg/ml) and samples were prepared for two-dimensional electrophoresis by boiling in SDS according to the method of Anderson and Anderson (24). After the samples were cooled to 25 "C, urea, NP-40, and ampholines were added to final concentrations of 9.5 M (1 mg of urea/pl sample, 3% (v/v) and 2% (w/v), respectively. Because heating samples to 100 "C can potentially induce artifacts in the isoelectric focusing dimension (25), the low temperature sample preparation procedures of Garrels (26) or OFarrell (25) were employed in a few experiments. The protein patterns obtained with the latter two preparation methods were identical with those obtained with the Anderson method. However, boiling the sample in SDS enhanced resolution and sharpness so this method was used routinely.
Two-dimensional Gel Electrophoresis, Autoradiography, a n d Computer Analysis-Cytosolic proteins were resolved by the twodimensional gel electrophoresis system described by O'Farrell (25) with minor modifications. The acrylamide concentration of the isoelectric focusing gels was reduced to 3% (2.84% acrylamide and 0.16% bisacrylamide), the NP-40 concentration was raised to 3%, and the SDS dimension was run on 0.75-mm slab gels of 10% acrylamide. Garrels has described a technique for two-dimensional electrophoresis that enhances resolution and reproducibility and optimizes spot shapes for computer analysis (26). In some experiments, the protein and autoradiographic patterns obtained with the Garrels and O'Farrell procedures were compared. While the Garrels procedure did provide the anticipated increase in resolution, the quantitative changes presented under "Results" were not altered by use of this gel system (data not shown). The O'Farrell system was used routinely because the larger bore tubing used in the isoelectric focusing step allowed larger protein loads.
The two-dimensional gels were stained with Coomassie brilliant blue, destained, and dried on filter paper as described (14,15). In some experiments, the gels were stained with silver according to the method of Merril et aL (27). Autoradiography was performed by exposing Dupont Chronex-4 x-ray film to the dried gels in standard xray cassettes for 3-5 days. Chronex 4 was chosen for its low and consistent background and care was taken to keep the exposure durations within the linear response range of the fiim (28). The fiims were developed in a well regulated Kodak automatic processor and scanned on an Optronics P-lo00 high speed densitometer. The optical density of the spots displayed on the autoradiograms was analyzed and integrated by an interactive computer program (28).
Assays-Cyclic AMP was assayed by the automated radioimmunoassay described by Brooker et al. (29).
Cyclic AMP-dependent protein kinase was assayed using histone as substrate in the presence and absence of 5 p~ cyclic AMP by the method of Cherrington et al. (30). The results of this assay are presented as the activity ratio (-cAMP/+cAMP) in the text. Aliquots of the hepatocyte suspension were prepared for these two assays by published methods (14,15). The specificity and sensitivity of the protein kinase assay was increased by using the heat stable protein kinase inhibitor as suggested by van de Werve et al. (31). The inhibitor was purified as described (32). The [y-"PJATP used in these assays was prepared by the method of Johnson and Walseth (33). Protein was assayed by the method of Lowry et al. (34) using crystalline bovine albumin as a standard. Lactate dehydrogenase was measured spectrophotometrically as described by Janksi and Cornell (23). The M , of the proteins in the SDS dimension of the gel system was estimated by comparison with standard curves using the marker proteins described (14). The pH gradient of the isoelectric focusing dimension was estimated by slicing a focusing gel into 1-cm segments, incubating the segments in 1.0 ml of degassed HzO for 4-6 h, and measuring the pH of each sample with a pH meter (25).
Identification of Enzymes in the Two-dimensional Pattern-The proteins resolved on the two-dimensional gels were identified using immunoprecipitation or by running purified proteins through the gel system. Some proteins were identified by both methods. Immunoprecipitation was performed by adding antibody sufficient to inhibit 95% of the activity of the enzyme of interest to "'P-labeled cytosolic proteins from control and hormone-treated cells. The antibody precipitate was obtained by the method of Schimke (35) and the proteins in the antibody pellet were prepared for electrophoresis by the standard sample preparation procedure. Control experiments were run on one-dimensional gels to insure that the immunoprecipitation procedure did not lead to dephosphorylation or proteolysis of the proteins (15). Enzymes identified by immunoprecipitation include pyruvate kinase, phenylalanine hydroxylase, fructose-l,6-bisphosphatase, and glycogen synthase. Samples of purified proteins were prepared and subjected to electrophoresis according to the protocols used for cytosolic proteins. In order to insure positive identification, 1-2 pg of protein of the immunoprecipitates or purified proteins were supplemented with 10-15 pg of cytosolic proteins and run through the gel system. This method produces an overloaded spot that is easily recognized from its high protein concentration and its location relative to its neighboring proteins. Pure proteins run through the gel system (followed by references to the method of purification) include pyruvate kinase (36), phosphorylase (37), phosphofructokinase (38), glycogen synthase (39), fructose-6-phosphate 2 kinase (40). tyrosine amino transferase (41), acetyl-coA carboxylase (42), and fructose-1,6bisphosphatase (43).
Calculations a n d Expression of Results-Two-dimensional gel data presented under "Results" were chosen as representative from threee to ten experiments. Averaged data are presented as the mean & S.E. Significant differences between groups of data were determined by paired t test.

Identification of Known Enzymes in the Two-dimensional Pattern
Two-dimensional Analysis- Fig. 1 presents the cytosolic proteins of isolated hepatocytes resolved by the two-dimen- FIG. 1. A two-dimensional polyacrylamide gel of hepatic cytosolic proteins stained with Coomassie blue. Cytosolic proteins were prepared from homogenates of isolated liver cells by centrifugation at 100,OOO X g and then 60 pg of protein was subjected to gel electrophoresis as described under "Materials and Met'lods." Proteins marked with arrous have been positively identified as: 4, phosphorylase; 7, phosphofructokinase; 13, pyruvate kinase; 14, fructose-&phosphate 2 kinase; 17, phenylalanine hydroxylase; 25, fructose-1,6-bisphosphatase. See "Materials and Methods" for details. The numbering system corresponds to that used in Fig. 4 and Table I sional electrophoretic technique. More than 400 Coomassie blue-stained spots were visible on the original gels and many more appear if the gels are stained with silver (not shown). The high resolution obtained allows positive identification of a number of hepatic enzymes thought to be regulated via changes in protein phosphorylation. Identification of nine proteins has been attempted including acetyl-coA carboxylase, phosphorylase, glycogen synthase, phosphofructokinase, fructose-6-phosphate 2 kinase, tyrosine amino transferase, phenylalanine hydroxylase, pyruvate kinase, and fructose-1,6bisphosphatase. Positive identifications were obtained with phosphorylase, 4; phosphofructokinase, 7; pyruvate kinase, 13; fructose-6-phosphate 2 kinase, 14; phenylalanine hydroxylase, 17; fructose-1,6-bisphosphatase, 25. These proteins are marked by the black arrowheads in Fig. 1. The number adjacent to the arrowhead corresponds to the numbering system used on the autoradiographs as presented in Fig. 4. The molecular weights and isoelectric points of these proteins are presented in Table I along with other data (see below). A large body of literature demonstrates that each of these proteins can be phosphorylated in vitro with the cyclic AMP-dependent protein kinase or in the intact cell following stimulation with glucagon (6,38,43-53).
Two proteins, acetyl-coA carboxylase and glycogen synthase, entered the two-dimensional gel system when the samples were prepared from the purified proteins. Although these two proteins can be identified in extracts of cells on onedimensional SDS gels (15,42,54), they could not be observed in the two-dimensional patterns obtained from cytosolic extracts and their positions are not indicated in Fig. 1. The reason($ for the failure of the acetyl-coA carboxylase and glycogen synthase in cell extracts to enter the two-dimensional gel system are unknown.
of the substrates phosphorylated in intact ["ZP]P04"--labeled hepatocytes in response to glucagon or a Ca"-linked hormone such as vasopressin. Fig. 2 presents autoradiographs made from the two-dimensional gels of ""P-labeled cytosolic proteins obtained from control and glucagon-treated cells. About 100 phosphorylated proteins are evident in heavily exposed autoradiographs made from proteins of control cells. Approximately half of these spots are faint at these exposure levels and do not appear on the photographic reproductions. Fig. 2 shows that stimulation of hepatocytes with 100 nM glucagon for 4 min markedly changes the phosphorylation of 13 phosphoproteins. These proteins are indicated by the black arrowheads. Interestingly, note that the phosphorylation of one protein with a M , = 21,000 and a PI of about 5.9 is markedly decreased by glucagon treatment. The phosphorylation of all other substrates is stimulated. As will be seen in Fig. 3, glucagon and the Ca"-linked hormones stimulate the phosphorylation of different substrates in intact cells. The six substrates whose phosphorylation is uniquely stimulated by glucagon are indicated by arrowheads pointed downward. Proteins whose phosphorylation is changed by both groups of hormones are marked by arrowheads pointing upward.
ea"-dependent Phosphorylation- Fig. 3 shows that the stimulation of intact liver cells with 10 milliunits/ml of vasopressin for 3 min changes the phosphorylation state of eight substrates. As above, the proteins whose phosphorylation is changed by vasopressin are marked with black arrowheads. Note that vasopressin uniquely stimulates phosphorylation of one substrate with a M, = 35,000 and a PI of about 6.6. Following the convention established in Fig. 2, this protein is marked with a downward-pointing arrow. The upwardpointing arrows mark the seven proteins whose phosphorylation state is changed by glucagon or vasopressin. Interest-Effect of Glucagon and Ca2+-linked Hormones on Protein protein with a Mr = 21,000 and of 5.9. ingly, vasopressin also causes the dephosphorylation of the In experiments analogous to those presented in Fig. 3, r:"Pl Phosphorylation M W 9 3 K -118 K -  glucagon for 4 min, the cytosolic proteins prepared, and then separated by two-dimensional gel electrophoresis as described under "Materials and Methods." Autoradiographs were made by exposing Dupont Chronex 4 x-ray film to the dried gels for 5 days. The arroujs in the lower halfof the figure point to proteins whose phosphorylation state is changed by treatment of the intact cells with glucagon. The proteins identified by arr0uj.s pointing downurard are affected by glucagon but not by Ca"-linked hormones (compare Figs. 2 and 3). In each case, the arrouw point to the phosphorylated form of the protein (see Figs. 5 and 6).
labeled hepatocytes with either angiotensin I1 or a-adrenergic agonists caused phosphorylation changes very similar to those observed with vasopressin (see Table I below). Cells stimulated with norepinephrine were pretreated with 20 /.LM l-propranolol in order to insure an a-adrenergic response. This dose of propranolol has been demonstrated to block the rise in cellular cyclic AMP levels mediated via the hepatic /3-adrenergic receptor (9, 14, 15). The efficacy of propranolol in the present experiments was monitored by measuring cyclic AMP levels and the activity of the cyclic AMP-dependent protein kinase in two of the four norepinephrine experiments presented in Table I. The cyclic AMP levels and protein kinase activity ratios were 1.6 pmol/mg of protein and 0.18, respectively, before norepinephrine addition, and 1.7 pmol/mg of protein and 0.16 following 4 min of treatment (averages of n = 2 in each case). These results imply that the observed effects of norepinephrine occurred predominantly via the aadrenergic receptor.
Previous experiments using one-dimensional gels to resolve the '"P-labeled cytosolic proteins demonstrated that neither angiotensin nor vasopressin stimulated protein phosphorylation in intact cells when Ca2* ion was removed from the incubation medium (15). Two analogous experiments were performed using vasopressin as the agonist and the two-di-

Properties of the proteins whose phosphorytation state is increased by hormones
Cells were stimulated with 100 nM glucagon for 4 min, 10 p~ norepinephrine, 10 milliunits/ml of vasopressin or 100 nM angiotensin for 3 weights were rounded off to the nearest thousand. The magnitude of the hormonally induced change in protein phosphorylation was measured min. Subunit molecular weights and isoelectric points were determined as described under "Methods" from three experiments. Molecular as described under "Methods." These values are reported as -fold over control. (   * Blanks in this column indicate that the nonphosphorylated protein was not observed with Coomassie blue stain. Blanks in this column indicate that the identity of the protein is unknown. mensional technique to resolve the "P-labeled proteins. When Caz+ ion was removed from the incubation medium as described (15), the effects of vasopressin on protein phosphorylation were greatly reduced or abolished (data not shown). This result confirms that the response to these hormones is Ca2+-dependent.
Quantitation of Phosphorylation Changes-The density of the 37 darkest spots in the autoradiographs from control and hormone-treated cells was quantitated by scanning the x-ray films on an Optronics P-1000 densitometer and analyzing the output with an interactive computer program (28). The location of each of these spots is shown in Fig. 4. The 37 spots represent all of the observed hormonally induced changes and a number of other spots of varying intensity, M , and PI whose phosphorylation state did not change with hormonal treatment.' The spots whose phosphorylation does not change with hormone treatment represent important benchmarks to insure that equal amounts of protein were loaded on the gel, the exposure and development of the x-ray film is consistent, and that the densitometer performance is constant.
The quantitative effects of glucagon and the Ca2'-linked hormones on the phosphorylation of the 37 spots analyzed is presented in Table I. Three major points are evident from the data. First, the quantitative effects of all the Ca"-linked hormones, angiotensin 11, I-norepinephrine, and vasopressin ' This analysis is not exhaustive. However, current sensitivity is such that proteins containing 1-2 cpm of 12P are readily detectable on the autoradiographs. Therefore, it is likely that the great majority of the phosphorylated proteins entering the gel system have been detected.
were very similar (compare the two columns on the far right). For this reason, the data for angiotensin and vasopressin (comprising two experiments each) was combined to allow statistical evaluation. Second, as pointed out in Figs. 2 and 3, there are major differences in the substrate specificities of the two types of hormones. Note that six of the proteins whose phosphorylation state is markedly affected by glucagon (spots 7, 14, 20, 21, 22, 25) are not affected by angiotensin, I-norepinephrine, or vasopressin. Only one protein, spot 29, is a unique substrate for the Ca"-linked hormones. There are seven overlaps (spots 4, 13, 17, 23, 34-36). Third, as was suspected from previous experiments (14, 15), there are marked differences in the magnitude of the phosphorylation change induced by the two types of hormones on their common substrates. In some instances (spots 13, 17), the effects of glucagon are much larger than those of the Ca"-linked hormones. In one instance, (spot 34), the effects of the Ca2'-linked hormones are much greater than those of glucagon. In a few instances (spots 4,23, 35, 36), all hormones cause similar changes in protein phosphorylation. As noted above, acetyl-coA carboxylase and glycogen synthase do not enter the two-dimensional gel system. Thus, the list of proteins in Table I is not complete. How many other phosphoproteins exhibit this behavior or have molecular weights or isoelectric points that fall outside the limits used in this study is unknown. The lower half of Table  I

Effect of Phosphorylation on a Protein's Isoelectric Point
Charge Shifts in Pure Proteins-A question that has not been definitively answered in intact cell phosphorylation experiments is whether hormonally induced increases in radioactivity actually reflect a chemical increase in the phosphate content of the protein. Since increases in covalently bound phosphate are associated with a charge shift to a more acidic PI (25, 55), the two-dimensional gel system can answer this question. The effect of a net increase in a protein's phosphate content on its isoelectric point is shown in Fig. 5. In this experiment, pure liver pyruvate kinase was incubated with the catalytic subunit of the cyclic AMP-dependent protein kinase in the presence of [yJ2P]ATP for 1-5 min. The Coo-massie blue-stained gel in the top left part of Fig. 5 shows that the native form of pyruvate kinase consists of two isomeric forms, with most of the protein existing in a basic form. The lower three parts in the left half of Fig. 5 show that incubation of pyruvate kinase with the catalytic subunit for 1-5 min shifts all of the protein into the acidic form (denoted by the black arrowheads). The accompanying autoradiographs shown on the right side of Fig. 5 demonstrate that the acidic spot represents the phosphorylated form of the protein as judged by the gain in radioactivity during the incubation period. The thin arrows on the right side of Fig. 5 point to the position of the basic (dephospho) form of the protein on the gels from which the autoradiographs were made.
Charge Shifts in Intact Cells-Charge shifts are also observed in the proteins whose phosphorylation is increased in intact cells by hormone treatment. Fig. 6 presents two examples of such charge shifts following glucagon or vasopressin treatment of [J2P]PO~"-labeled hepatocytes. The top arrows in each part of Fig. 6 point to pyruvate kinase (spot 13) and the bottom arrows point to fructose-6-phosphate 2 kinase (spot 14). The heavy arrows point to the acidic (phosphorylated) form and the long thin arrows point to the basic, (nonphosphorylated) form of the proteins. The same convention holds for the stained protein (left side of Fig. 6) and the autoradiographs (right side of Fig. 6). The top parts show that in control cells, most of the protein in spots 13 and 14 exists in the basic, dephosphorylated form of the enzyme. This situation is analogous to that observed with native pyruvate kinase shown in Fig. 5. The middle parts shows that treatment of the cells with glucagon shifts most of the stained protein in spots 13 and 14 to the acidic, phosphorylated form (heavy arrows). The bottom parts show that treatment of the cells with vasopressin causes a partial shift of the stained protein of pyruvate kinase (spot 13) to the phosphorylated form. Consistent with the smaller amount of protein shifted to this form, the autoradiograph shows a smaller increase in the phosphorylation of pyruvate kinase. As noted in Table I, stimulation of cells with vasopressin does not increase the phosphorylation of fructose-6-phosphate 2 kinase (spot 14). This is consistent with the lack of a charge shift observed in the stained protein and no observed increase in density on the autoradiograph. All hormone-induced charge shifts that could be detected in gels stained with Coomassie blue are summarized in Table I.
Phosphorylated Proteins with Multiple Charges-The examples of charge shifts shown in Fig. 6 apparently represent the simple case of a protein having only two charged forms, phosphorylated and dephosphorylated. Some proteins can exhibit more complex behavior. An excellent example of a protein with a complex isoelectric focusing pattern is provided by spot 23 (M, = 49,000, PI = 4.6). While this protein appears as a streak on the left side of Figs. 2-4, it can be shown to be a series of six or seven spots by changing the ampholine content of the focusing gel to cover a more acidic range. Fig.  7 presents small sections of autoradiographs made from such an experiment with two benchmark proteins (spots 24 and 33) marked to provide reference points. Only the autoradiographs are shown because this protein is too low in concentration to be visible on a Coomassie blue-stained gel. In the top section representing the control, the two ends of the charge train are marked by arrows. When the cells are treated with glucagon, the entire train shifts toward the acidic region of the gel (the arrows are the same distance from spot 24 in each section). Note that the phosphorylation of some of the spots is markedly increased by hormone treatment. Treatment of the cells with any of the three Ca"-linked hormones produced an identical result (data not shown). The molecular reasons for the observed behavior of spot 23 are unknown. The train of spots is suggestive of a carbamylated protein that may have been artifactually produced during sample preparation (25,26,56). However, few other proteins in the sample exhibited multiple isoelectric forms (see Fig. 1) and two other methods of sample preparation did not alter this behavior of spot 23 (see "Materials and Methods" for details). Whatever the reason for spot 23 focusing as multiple spots, the charge shift observed following treatment of the intact cell with hormones is consistent with the addition of one phosphate to each molecular form of the protein. The phosphorylation of spot 23 in the intact cell has been the object of a great deal of study. Treatment of hepatocytes with insulin, glucagon, Ca2+-linked hormones, and the Ca2' ionophore A23187 increase the phosphorylation of this protein (14, 15,57). Unfortunately, neither the identity nor the function of this interesting protein is known.

DISCUSSION
It is now well established that a-adrenergic agonists, angiotensin 11, and vasopressin can stimulate hepatic glycogenolysis and gluconeogenesis through a Ca2+-sensitive, cyclic AMPindependent mechanism (1-5.7-9). Furthermore, recent studies indicate that the Ca2'-linked hormones can regulate the enzymes of glycogen metabolism via a Ca"-dependent increase in their phosphorylation state (14, 15). These experiments also demonstrated that treatment of hepatocytes with the Ca2'-linked hormones increased the phosphorylation of a number of other cytosolic proteins (15). Because one-dimensional SDS gels were used to separate the [J'P]PO~'--labeled of the gel ranging in M, from about 56,000 to 20,000 and includes spots 24 and 33 as reference points. The arrowheads mark the beginning and end of the charge train and are the same distance from spot 24 in each part of the figure. proteins, the resolution was not great enough to determine if the substrates affected by the Ca"-linked hormones were distinct from those stimulated by treatment of the cells with glucagon.
In order to provide a definitive answer regarding the substrates affected by glucagon and the Ca"-linked hormones, the present study used two-dimensional electrophoresis to separate the cytosolic proteins and computer techniques to analyze the autoradiographs. The data presented under "Results" clearly show that stimulation of intact cells with glucagon or the Ca2'-linked hormones leads to increases in the phosphorylation of distinct but overlapping groups of substrates. The overlaps in the substrate specificities of the two pathways can account in part for the similarity of the biochemical responses observed when liver is stimulated with these hormones. For example, both types of hormone stimulate the activity and phosphorylation state of phosphorylase to the same extent (Table I and Refs. 1-4, 14, 15).
While there are eight substrates whose phosphorylation is affected by all hormones, both the cyclic AMP and Ca2'dependent pathways do have unique substrates (Table   I).
This result implies that, although the functions of the two types of hormones in liver are usually considered to be similar (i.e. activation of glycogenolysis), there must be some distinct biochemical effects of the two pathways. An interesting example of this point is provided by the effects of the two pathways on the enzymes of gluconeogenesis. Table I shows that glucagon, acting through the cyclic AMP-dependent pathway, can stimulate the phosphorylation of four enzymes thought to control the futile cycling of carbon as it is converted from lactate to glucose. These enzymes are phosphofructokinase (spot 7), pyruvate kinase (spot 13), fructose-&phosphate 2 kinase (spot 14), and fructose-1,6-bisphosphatase (spot 25). The role of phosphorylation in controlling the activity of fructose-l,6-bisphosphatase and phosphofructokinase is uncertain (10, 38, 43, 45, 51).
However, it is now clear that glucagon can regulate the level of the important allosteric effector of these two enzymes, fructose-2,6-phosphate (45, 58-63), via changes in the phosphorylation state of the enzyme catalyzing its formation, fructose-&phosphate 2 kinase (47,48, 64). There is also general agreement that glucagon can regulate the flux of carbon at the level of pyruvate kinase via cyclic AMP-dependent phosphorylation of the enzyme (6,10, 46). In contrast to the effects of glucagon, the Ca2+-linked hormones do not increase the phosphorylation of phosphofructokinase, fructose-&phosphate 2 kinase, or fructose-1,6bisphosphatase and only moderately affect the phosphorylation of pyruvate kinase (Table I). Interestingly, while both glucagon and the Ca2+-linked hormones stimulate gluconeogenesis, the efficacy of the Ca"-linked hormones is often observed to be less than that of glucagon (7-10, 65). The possibility that these hormones do not cause a full stimulation of gluconeogenesis because they cannot regulate phosphofructokinase and fructose-1,6-bisphosphatase is an interesting question for future study.
Two-dimensional gels can be used to display the changes in a protein's isoelectric point that occur following covalent modification of the protein (25, 55). An important achievement of this study is the demonstration that the increases in protein phosphorylation observed in autoradiographs following hormonal stimulation of the intact cell are correlated with a shift of the protein to a more acidic isoelectric point (Figs.  5-7). Charge shifts were observed for all phosphoproteins able to be visualized in the Coomassie blue-stained gels and occurred following treatment of the cells with glucagon, a-adrenergic agonists, vasopressin, or angiotensin I1 (Table I).
These charge shifts clearly prove that the proteins gain chem-ical phosphate. Moreover, the ability of hormones to shift the isoelectric point of a protein following phosphorylation should be of general interest because it allows one to observe and quantitate hormonal effects by examination of Coomassie blue or silver-stained protein patterns. Thus, the use of large quantities of [32P]P043may not be necessary in all experiments.
The observation that stimulation of intact cells with glucagon or the Ca"-linked hormones leads to the phosphorylation of different sets of substrates strongly suggests that the pathways activated by these hormones lead to the activation of distinct protein kinases. The available evidence suggests that glucagon activates the cyclic AMP-dependent protein kinase through increases in the cellular level of cyclic AMP (2,12,15,30). Thus, 12 of the 13 substrates listed in Table I (excluding phosphorylase) whose phosphorylation is increased by glucagon potentially represent substrates for the cyclic AMPdependent protein kinase. Based on the literature (53) and the information in Table I, it appears that there are five or more as yet unidentified proteins whose phosphorylation is controlled by cyclic AMP.
An important question that remains to be answered is the mechanism by which the phosphorylation of proteins is increased by the Ca"-linked hormones. It is possible that the calcium signal stimulates a Ca"-sensitive kinase or inhibits a protein phosphatase. There is comparatively little known about the hormonal regulation of protein phosphatases in liver and evidence exists to suggest that the Ca'+-linked hormones do not activate phosphorylase by this mechanism (66). There are three Ca'+-stimulated protein kinases that may mediate the effects of the Ca2+-dependent hormones in liver. These enzymes include phosphorylase kinase (12, 17, 20, 66), a glycogen synthase kinase that is stimulated by Ca2+ and calmodulin (21), and the Ca2+ and phospholipid requiring kinase discovered by Nishizuka and co-workers (18, 19). Experiments performed with rats deficient in hepatic phosphorylase kinase show that the effects of the Ca'+-linked hormones on phosphorylase are mediated by phosphorylase kinase (66) but suggest that their effects on pyruvate kinase are not mediated by this enzyme (67). However, because of the difficulties in purifying phosphorylase kinase from liver, its true substrate specificity is unknown (20). To date, the calmodulin requiring glycogen synthase kinase has only been reported in rabbit liver (21). This enzyme has a very restricted substrate specificity in vitro and is unable to phosphorylate myosin light chain, histone, phosphorylase, or casein (21). The substrate specificity of the Ca2'-sensitive, phospholipid-requiring protein kinase in the liver is unknown. The role of each of these protein kinases in hepatic function bears further study.
A number of cell types appear to contain both Fyclic AMPdependent and Ca'+-dependent pathways capable of increasing the phosphorylation of important substrates. For example, Ca'+-sensitive kinases have been identified in brain synaptic membranes that can phosphorylate a number of membrane proteins that are similar to those phosphorylated by the cyclic AMP-dependent protein kinase (68-73). Treatment of intact platelets with the calcium ionophore A23187, thrombin, or collagen increases the phosphorylation of different proteins from those affected by stimulating the cell via cyclic AMP with prostaglandin E, (74). Similar results have been observed in isolated mast cells following stimulation of labeled cells with A23187 and the drug cromolyn (75, 76). It will be interesting to observe how each of these phosphorylation changes is linked to an appropriate physiological response. In the liver, most of the available data suggests that the effects of the cyclic AMP-dependent and the Ca2+-dependent pathways are directed toward similar biochemical responses (1-13). HOW-ever, the finding that the Ca"-linked hormones affect one unique substrate raises the possibility that these hormones may be able to elicit a unique response in the hepatocyte.