Hormonal effects on calcium homeostasis in isolated hepatocytes.

A new method is described to determine the cytosolic free Ca”-concentration in isolated hepatocytes. Calcium concentrations in the medium were monitored spectrophotometrically using arsenazo 111. Digitonin ( 5 pg/mg dry weight of cells) was added to make the plasma membrane permeable to Ca”, and the magnitude of the Ca“ change was measured for different initial external Ca2+ concentrations. The cytosolic free Ca2+ concentration was estimated by determining the Ca2+ null point when no net Ca2+ changes were observed upon addition of digitonin. Values for the Caz+ null point between 100 and 200 nm were obtained for normal hepatocytes from fed or fasted rats, depending upon the pH and Mg2+ concentration of the incubation medium. Rapid cell fractionation of calcium-loaded cells permitted determination of the intracellular calcium distribution under these conditions. In normal cells, the mitochondrial calcium content accounted for 60 to 70% of the total cell calcium content, with the remainder located mainly in the microsomes. This proportion was approximately constant over a 10-fold range of total cell calcium. Glucagon had no effect on the cytosolic free Ca2+. On the other hand, a-adrenergic stimulation increased the cytosolic free Ca2+ concentration 2to %fold; phenylephrine M) increased mean values from 0.19 ? 0.01 to 0.46 f 0.04 p~ within 2 min. The time course and dose-response  relationship of cytosolic free  Caz+ changes closely followed the increase of phosphorylase a. Rapid cell-fractionation studies showed that the calcium content of the mitochondrial fraction was decreased after norepinephrine addition to the intact cells, while that of the microsomal fraction was increased. The above effects were abolished by a-adrenergic antagonists but were little affected by /3-adrenergic antagonists. These data indicate that the mitochondrial calcium pool is highly labile and is influenced by an as yet unknown transducing signal which is regulated by interaction of a-adrenergic hormones with the plasma membrane receptor. Increased efflux of calcium from the mitochondria causes a rise of cytosolic free Ca2+ and regulates enzymes of carbohydrate metabolism possibly by enhanced binding of calcium to calmodulin.

normal hepatocytes from fed or fasted rats, depending upon the pH and Mg2+ concentration of the incubation medium. Rapid cell fractionation of calcium-loaded cells permitted determination of the intracellular calcium distribution under these conditions. In normal cells, the mitochondrial calcium content accounted for 60 to 70% of the total cell calcium content, with the remainder located mainly in the microsomes. This proportion was approximately constant over a 10-fold range of total cell calcium.
Glucagon had no effect on the cytosolic free Ca2+. On the other hand, a-adrenergic stimulation increased the cytosolic free Ca2+ concentration 2to %fold; phenylephrine M) increased mean values from 0.19 ? 0.01 to 0.46 f 0.04 p~ within 2 min. The time course and dose-response relationship of cytosolic free Caz+ changes closely followed the increase of phosphorylase a. Rapid cell-fractionation studies showed that the calcium content of the mitochondrial fraction was decreased after norepinephrine addition to the intact cells, while that of the microsomal fraction was increased. The above effects were abolished by a-adrenergic antagonists but were little affected by /3-adrenergic antagonists. These data indicate that the mitochondrial calcium pool is highly labile and is influenced by an as yet unknown transducing signal which is regulated by interaction of a-adrenergic hormones with the plasma membrane receptor. Increased efflux of calcium from the mitochondria causes a rise of cytosolic free Ca2+ and regulates enzymes of carbohydrate metabolism possibly by enhanced binding of calcium to calmodulin.
In recent years, it has become increasingly apparent that changes of intracellular free CaZ+ exert a vital role in the mediation of chemical events linking hormonal or other ex-* This work was supported by National Institutes of Health Grants AM 15120 and AM 19525. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be herebv marked "adoertisemenf" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. + T o whom correspondence and reprint requests should be addressed.
ternal stimuli to alterations of intracellular enzyme activities (1)(2)(3). In cardiac muscle and nerve, many studies have described the role of the action potential in the regulation of Ca2+ entry through the plasma membrane by a gating mechanism (4). With other excitable tissues such as skeletal muscle, as well as nonexcitable or secretory tissues having a lower resting membrane potential, flux of Ca2+ across the plasma membrane appears to be less important for the transduction of the biochemical signal. Alternatively, it has been suggested that membrane depolarization or hormone-receptor interactions might elicit a release of Ca'+ from intracellular storage sites and a flux of Ca" between intracellular organelles and Ca"-binding target proteins. The relative roles of the plasma membrane, microsomes, and mitochondria in this process are presently under active investigation (3,(5)(6)(7)(8). Of particular interest is the growing body of evidence suggesting that cyclic AMP and Ca" are parallel rather than sequential second messengers for the elicitation of biochemical events (9,10). Thus, a-adrenergic activity in liver is thought to be mediated by cyclic AMP-independent processes and to involve a direct increase of phosphorylase b kinase activity by a rise of intracellular Ca2+ concentration to between lo-' and M free Ca" (11-13).
Although there is a considerable body of data concerning the effects of catecholamines, glucagon, and other hormones on *%a'+ flux in liver (6, 7, 9, 10, 13-16), a more detailed understanding of the sequence and molecular mechanisms of the events involved has been severely hampered by a lack of knowledge of the intracellular Ca" concentrations as well as by a poor understanding of the detailed mechanisms of cellular calcium homeostasis. As a first step toward permitting a greater understanding of the mechanisms of hormone actions and stimulus-response activity of target cells, we have developed methods for the measurement of the concentrations of free Ca" and Mg" in the cytosol of isolated hepatocytes and in the mitochondrial matrix of isolated mitochondria (17). In this paper, we describe in detail a spectrophotometric method using arsenazo 111 as a calcium indicator for measurement of the free cytosolic Ca'+ concentration in isolated hepatocytes. In addition, a modification of the rapid cell-disruption technique of Tischler et al. (18) has been used to determine the calcium content of the mitochondria and microsomes after calcium loading of the hepatocytes and after hormonal stimulation. Data are presented describing the effects of the aadrenergic hormones norepinephrine and phenylephrine on the intracellular distribution of calcium and on the kinetics and dose-response relationship of changes of the free cytosolic Ca'+ concentration.

EXPERIMENTAL PROCEDURES'
Hepatocyte Preparation-Rat liver cells were obtained from fed I Portions of this paper (including the experimental procedures for the arsenazo method of Ca" determination and Figs. 1 to 3) are presented in miniprint at the end of this paper. Miniprint is easily or overnight-fasted rats by previously described (19) modifications of the collagenase liver perfusion method originally introduced by Berry and Friend (20). The hepatocytes were washed twice and subsequently incubated in modified Ca"-and Mi"-free Hanks' medium containing 137 mM NaCl, 5.4 mM KCl, 0.44 mM KHzPO~, 4.2 mM NaHCO:I, 0.33 mM Na2HPO:l, and 20 mM 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid (Hepes), pH 7.4, which was equilibrated with 100%;. 0,. Calcium loading of liver cells was achieved by incubation of the cells (approximately 5 mg dry weight/ml) in ice-cold modified Hanks' medium containing up to 10 mM CaCI? (8). followed by washing in Ca"-free medium.
Measurement of Total Calcium Contents in Cellular Subfractions-The total calcium content of isolated hepatocytes was deterof the cells through silicone oil into 14% (w/v) perchloric acid (28). mined by atomic absorption spectroscopy after rapid centrifugation Corrections for carrydown of medium CaZ+ were made from measurements of the sucrose space by inclusion of [U-"C]sucrose and ,'HLO in the incubation medium. The calcium content in the heavy particulate fraction (mainly mitochondria, but also containing peroxisomes, lysosomes, and nuclei) and the light particulate fraction (microsomes) were determined using a modification of the rapid cell-disruption technique of Tischler et al. (18). Cells were mixed with 3 mM EGTA,' 15 p~ ruthenium red, and 50 pg of digitonin/mg dry weight of cells, and 3 s later passed under controlled pressure (60 p s i ) through a 25gauge needle, followed by centrifugation through silicone oil into perchloric acid. The combination of the brief exposure time to digitonin and the shearing forces generated during passage of the cell suspension through the needle caused disruption of the plasma membrane with release of 85 to 95Yr of lactate dehydrogenase, a-glycerophosphate dehydrogenase, and glucose-6-phosphatase, and 5% release of glutamate dehydrogenase. After the above centrifugation to remove the heavy particulate fraction, the supernatant was recentrifuged at high speed (120,OOO X g) for 15 min in a Beckman bench-top ultracentrifuge to remove the microsomal fraction, which was subsequently analyzed for glucose-6-phosphatase activity and for calcium content after extraction with perchloric acid.
Assay Procedures-Protein was determined by a modification of the biuret procedure (29). ATP and glucose in neutralized perchloric acid extracts of mitochondria and cells were assayed fluorometrically by the hexokinase method (30). Chloride was assayed according to Van Rossum (31). Lactate dehydrogenase and glutamate dehydrogenase activities were assayed as described previously (18). glucose-6phosphatase was assayed according to De Duve et al. (321, and phosphorylase a was assayed as described by Hutson et ai. (33).
Materials-Digitonin (Sigma Chemical Co.) was recrystallized three times from hot ethanol. The ionophore A23187 and the uncoupler FCCP (carbonyl cyanide p-trifluoromethoxyphenylhydrazone) were gifts from Dr. Robert Hamill, Eli Lilly Co., Indianapolis, Ind., and Dr. Peter Heytler of the Research Division, DuPont Co., Wilmington, Del., respectively. Norepinephrine, phenylephrine, and propranolol were obtained from Sigma, and phentolamine was from Ciba Pharmaceutical Co. Phenoxybenzamine was a gift from Smith, Kline and French, Philadelphia, and glucagon was a gift from the Eli Lilly co.

Cytosolic Free Ca2+
Concentration in Hepatocytes-The effects of digitonin addition on changes of free Ca'+ in the medium of the hepatocyte suspensions are illustrated in Fig.  4. The Ca" changes were measured spectrophotometrically with arsenazo 111 using the wavelenth pair 675-685 nm. Either EGTA or Cat+ was added to adjust the free Ca" to desired levels prior to addition of digitonin. Calcium was removed from the medium by the digitonin-treated cells, but the final equilibrium free Ca2+ concentration was independent of the initial Ca" concentration. This equilibrium value provides a convenient rough estimate of the cytosolic free Cat+, which read with the aid of a standard magnifying glass.  compares favorably with the more accurate null point titration method described below. In this particular experiment, the medium contained 0.5 mM Mgl', which has the effect of increasing both the equilibrium Ca2+ value and the cytosolic Ca2+ null point as described subsequently. When cells were incubated in the absence of arsenazo, addition of digitonin caused no significant absorbance changes due to light scattering at the 675-685 nm wavelength pair. Likewise, negligible absorbance changes were observed after digitonin addition when hepatocytes were incubated with arsenazo and excess EGTA, or when digitonin was added to buffer containing arsenazo in the absence of cells (data not shown). A further point is that, although the rate of Ca" uptake by digitonized cells is affected by the digitonin concentration, when this is constant, the rate of Ca2+ removal is proportional to the initial Can+ concentration, since the intracellular calcium-sequestering reactions are well below saturation, The results of a series of experiments similar to those shown in Fig Fig. 5B. If there are no diffusion gradients of ions across the plasma membrane after digitonin addition, the null point should correspond to the cytosolic free Cas+ concentration, even though the calcium indicator is measuring the free Ca2+ concentration in the bulk phase of the cell incubation medium. Measurements of the chloride distribution between the digitonin-treated cells and the medium showed, in fact, that the plasma membrane potential calculated from the Donnan equilibrium (58 log [cl-],,,,/[Cl-]i,) collapsed essentially to 0 from the control value of -32 mV. Additional experiments showed that digitonin (5 pg/mg of protein) had no effect on calcium uptake or release by isolated microsomes or mitochondria, and also caused no further calcium changes when added to hepatocytes after addition of A23187 (IO nmol/mg dry weight of cells which, at this high concentration, results in equilibration of Cat+ across the plasma membrane (34). This latter experiment suggests that release of ions, ATP, proteins, and other cellular constituents by digitonin addition is not interfering with the spectrophotometric measurement of Ca".
It is readily apparent from the data in Figs. 4 and 5 that the size of the Ca2+ change after digitonin addition to the cells is much greater than that expected from a simple mixing of the medium and cytosol volumes. This is illustrated by the following calculation: assuming a cytosolic volume of 2 ml/g dry weight of liver cells (18), the total cytosolic volume in the cell incubation medium (2.5 ml) with 4 mg/ml of cells is only 20 pl. With an initial medium free Ca" concentration of 1 PM, equilibration of the spaces would only result in a change of the medium free Ca" of 0.01 p~. Since the measured change at 1.15 PM external free Ca" is about 0.09 p~, it is clear that calcium is being removed by intracellular organelles once the permeability restriction of the plasma membrane to Ca" movement is removed.
Both mitochondria and microsomes contribute to net uptake of calcium by the digitonin-treated cells, as illustrated below. Fig. 6 shows the effects of digitonin on cells incubated in the absence and presence of 15 IJM ruthenium red. Ruthenium red at this concentration completely inhibits Ca')' uptake by the mitochondria. Digitonin addition caused a 3-fold greater removal of calcium from the medium in the absence than in the presence of ruthenium red. The calcium taken up was released by the subsequent addition of FCCP. The calcium uptake or binding in the presence of ruthenium red can be attributed to nonmitochondrial intracellular components, while that released by FCCP can be attributed to mitochondria. Null point titrations with hepatocytes incubated in the absence and presence of 15 ,ILM ruthenium red prior to addition of digitonin are given in Fig. 7. The results show that although ruthenium red decreased the extent of the Ca2+ uptake more than 2-fold, it did not affect the null point. This indicates that the Ca" null point determination is independent of the activity of the mitochondrial Ca2+ uptake system.
Since the intracellular ionic composition is very different from that of the normal cell incubation medium, the effect of digitonin addition to the hepatocytes will be to cause a redistribution of ions across the plasma membrane. The effects of different Mg2+ and H' ion concentrations in the medium prior to digitonin treatment were investigated in detail, because it was felt that these ions may be particularly important in altering the binding of Ca" to intracellular sites, thereby changing the measured Ca2+ null point. Table I   influence of varying pH over the range from 6.8 to 7.8, and of Mg2+ ion concentration from 0 to 1 mM, on the measured Ca2' null points. The Ca'+-arsenazo dissociation constants shown in Fig. 2 for the different media were used for calculation of the free Ca'+ concentrations. As seen from Table I, the Ca2+ n d l point was not appreciably affected by pH, but tended to increase with increasing Mg2+ concentrations. The cytosolic free MgS+ concentration is thought to be between 0.5 and 1.0 mM (17,35), indicating that the cytosolic free Ca" concentration should be about 0.2 p~ (cc Fig. 4).
Distribution of Calcium between Intracellular Organelles in the Hepatocyte-It is well established that collapse of the Calcium Homeostasis in Liver Cells proton electrochemical gradient across the mitochondrial membrane by addition of uncoupling agents causes a release of Ca'+ from the mitochondria (for review, see Ref. 36). Likewise, induction of an electroneutral H+/Ca2+ antiport by addition of the divalent cation ionophore A23187 causes efflux of Ca'+ from the mitochondria at low external Ca'+ concentrations (37). Calcium release from intracellular storage sites of isolated hepatocytes has been observed following addition of these agents (7,38,39), which results in efflux of Ca" across the plasma membrane where it is detected by arsenazo 111 present in the medium. The maximal amount of Ca'+ released by A23187 (30 (IM) was greater than that released by uncoupler (15 (IM FCCP), suggesting that while uncoupler may be releasing Ca" from the mitochondrial pool alone, A23187 releases Ca2+ from both the mitochondria and other storage sites, such as microsomes (data not shown). Fig. 8 shows a comparison of the quantitative determination of Ca2+ released from hepatocytes by FCCP, measured by the absorbance change of arsenazo 111 and by the difference of the total   In order to investigate the distribution of calcium between the mitochondrial and microsomal fractions of the cell as a function of total cellular calcium, hepatocytes were loaded with calcium by incubation in the cold (8) in the presence of 3 p~ ruthenium red for different times up to 90 min with 5 or 10 mM extracellular c a z + . After incubation, the cells were centrifuged at 300 X g and washed twice with Ca2+-free medium. The calcium content of the different subcellular organelles was then estimated by a number of different methods, as shown in Table 11. The total calcium content of the cell is shown in Column A and ranged from 3 to 30 nmol/mg of cell protein (0.85 mg of cell protein is approximately equivalent to 1 mg dry weight of cells). The amount of Ca" released from the cells after addition of 5 (IM FCCP was measured directly in some experiments by arsenazo 111 (Column B) or by subtraction of the residual cell calcium from the initial cell calcium (Column C). On the assumption that uncoupler releases Ca2+ only from the mitochondrial pool, Columns B and C provide estimates of the mitochondrial cell calcium. An independent check of the mitochondrial cell calcium is provided by rapid disruption of the cells in the presence of EGTA and ruthenium red and separation of the mitochondria, as described under "Experimental Procedures." These data are shown in Column D of Table 11. The agreement between the numbers in Columns B, C, and D may be considered satisfactory, and the data indicate that the mitochondrial cell calcium increases roughly in proportion to the total cell calcium. Thus, 60 to 70% of the total cell calcium is located in the mitochondria for the different calcium-loaded cells. The microsomal calcium content was also estimated directly by further high speed centrifugation of the supernatant fraction from the cell disruption (Column E). These values may be compared with values obtained by subtracting the uncoupler-releasable Ca" from the total cell calcium (Column A -Column C) or by subtracting the mitochondrial calcium obtained by cell disruption from the total cell calcium (Column A -Column D).
Values obtained by direct measurement of the microsomal fraction were generally higher than those obtained by the subtraction methods, suggesting that some Ca'" uptake into the microsomes may have occurred during centrifugation. were investigated using phenylephrine (IO-' M) as the madrenergic agonist.

Hormonal Effects on Cytosolic Free Ca2' and Intracellular
In these experiments, phenylephrine increased the cytosolic free Ca"+ 2.4-fold from a control value of 0.19 p~ and increased phosphorylase a 2.2-fold. These effects were completely blocked by phentolamine, but were only slightly attenuated by propranolol, confirming the predominant a-agonist activity of phenylephrine.
The time courses of the effects of lo-" M phenylephrine on the changes of cytosolic free Ca" and phosphorylase a activity

Distribution of calcium between subcellular organelles in rat lit9er
cells Liver cells were calcium-loaded as described under "Experimental Procedures" and washed with Ca"-free medium before measurement of total calcium (A). The amount of Ca'+ released into the medium after addition of 5 p~ FCCP was measured directly using arsenazo I11 (B) or after centrifugation of the cells (C). In separate experiments, the cells were rapidly disrupted for measurement of the heavy particulate or mitochondrial calcium (E)). 'The supernatant from the cell disruption was centrifuged in a Beckman microfuge and the calcium content of the light pellet or microsomes was assayed (E). 3   Increased glucose production as a result of enhanced glycogenolysis, however, continued for at least 30 min at the higher hormone concentrations. The dose-response curve for the stimulation of phosphorylase a activity by phenylephrine was also remarkably similar to that for the increase of cytosolic free Ca" (Fig. 12). Half-maximal effects were obtained at about 5 x lo-" M phenylephrine.

Effects o f a and / 3 receptor antagonists on the response of cytosolic free CaZ' and phosphorylase to phenylephrine stimulation
The effect of norepinephrine on the intracellular distribution of calcium between the "mitochondrial" and "microsomal" fractions after rapid cell disruption is shown in Table  IV. The microsomal fraction was calculated from the difference between the total cell calcium and the heavy particulate (mitochondrial) fraction which was obtained by the rapid celldisruption technique. In these experiments, the mitochondrial calcium content represented 65 to 83% of the total cell calcium under control conditions. Of particular interest is the fact that Time course of changes of cytosolic free Caz+ and phosphorylase a following addition of M phenylephrine to isolated hepatocytes. Liver cells (4 mg dry weight/ml) from fed rats were incubated in modified Hanks' medium containing 0.5 mM Mg" for 15 min at 37°C prior to hormone additions. Phenylephrine dose-response curve. The incubation conditions were the same as those described in the legend to Fig. 11 except for variation of the phenylephrine concentration.

TABLE IV Effects of norepinephrine on intracellular calcium distribution
Liver cells from fed rats (8 mg of protein/ml) were incubated at 37°C in modified Can+-and Mg"-free Hanks' medium for 10 min prior to addition of norepinephrine. After removal of control samples for rapid cell disruption (see "Experimental Procedures"), 10 " M norepinephrine was added and a second sample was taken after 2.5 min. When present, a-adrenergic antagonists phenoxybenzamine or phentolamine at M were added 30 s prior to norepinephrine addition. Numbers of experiments are shown in parentheses. Experimental samples are compared with the paired control after no addition and after addition of norepinephrine or norepinephrine plus a-antagonist. norepinephrine decreased the calcium content of the mitochondrial fraction by 40% without appreciably affecting the total cell calcium, with the result that the calculated microsomal calcium content increased almost %fold. Glucose pro-duction was measured in parallel experiments 15 min after norepinephrine addition. The control rate of glucose output was 29 nmol/mg of protein/l5 min, and was increased by 768 after norepinephrine addition. These effects of norepinephrine were blocked by the a-adrenergic inhibitors phentolamine and phenoxybenzamine, which by themselves had little effect.

DISCUSSION
Validity of the Cytosolic Free Ca2+ Measurements-Very few other measurements of the cytosolic free Ca')' concentration in cells have been reported in the literature, and no others have been reported for hepatocytes. Three general methods have been used for estimation of intracellular free Ca" concentrations in different tissues. These involve use of ( a ) calcium-selective microelectrodes; ( b ) calcium-sensitive photoproteins such as aequorin, which emit a blue luminescence when exposed to Ca"; and (c) calcium-sensitive metallochromic indicators, which have different absorption properties when Ca2+ is bound. Calcium-sensitive microelectrodes with a tip diameter of about 1 pm and a detection limit of 6 X IO-' M in Ca"-buffered solutions have been described (40). However, they are less sensitive and reproducible in solutions approximating intracellular fluid, and few measurements of intracellular free Ca2+ using microelectrodes have as yet been reported (41). The useful range of Ca" measurements with aequorin is from 10" to M, and although it has a Ca'+ dissociation constant of 16 p~, it can be used at very low concentrations so that only about 2% of total calcium is bound (42). Both aequorin and arsenazo I11 have been injected into single cells such as barnacle muscle fibers (43) and giant axons of the squid (44-46), and have provided values of 100 nM or below for the intracellular free Ca" concentration. Both methods when applied to in situ measurements are subject to large potential absolute errors in this range as discussed by various authors (42, 43,46,47) and are more suitable for measuring relative changes of intracellular free Ca" after perturbing the cell. I n addition, this approach is strictly limited to cell types where microinjection is technically feasible.
Arsenazo I11 has a high selectivity for Ca2+, but is sensitive to ionic composition and pH (48). A major disadvantage with arsenazo 111 as an intracellular Ca" probe is that its dissociation constant for calcium is about 2 orders of magnitude greater than apparent cytosolic free Ca'+ concentrations. Thus, a t concentrations of arsenazo 111 which give a linear Ca2+ titration, the ratio of calcium bound to free Ca" is 3 or greater. This necessitates careful measurement of the Ca2+ arsenazo dissociation constants under the experimental conditions and correction for the amount of bound calcium. However, an advantage of arsenazo 111, especially when absorbance changes are measured in a highly sensitive dual wavelength spectrophotometer, is that its high extinction coefficient allows measurements to be made at two narrowly spaced wavelengths which minimize light-scattering artifacts. Since the above factors have been taken into account in the present study, it may reasonably be concluded that the free Ca" concentration at the extrapolated null point with digitonin-treated hepatocytes can be measured with precision.
The relationship between the Caa+ null point as determined by the extracellularly located arsenazo 111 and the true intracellular free Ca2+ concentration requires comment. A number offactors allow us to conclude that the values are very similar, particularly when the extracellular and intracellular Mg'+ concentrations are equilized prior to digitonin addition. These are ( a ) the rapid exchange of Ca" between cells and medium after digitonin addition; ( b ) the retention of basic structural integrity of the digitonized hepatocyte (49) and small loss of total protein, which suggests that there is an insignificant loss of Ca"-binding sites; and (c) the virtual collapse ofthe plasma membrane potential after digitonin addition, which indicates a negligible Donnan equilibrium distribution of Ca". The general validity and usefulness of the technique is also evidenced by the fact that FCCP, A23187, and a-adrenergic hormones as perturbing agents increase the Ca2+ null point 2to 4-fold. A distinct advantage of this method is that it is applicable in principle to any cell type in suspension and to cells in tissue culture attached to suitable cover slips that will fit inside a spectrophotometer cuvette. The data obtained, however, represent a statistical average for the cell population, and may be misleading if there is appreciable heterogeneity of the cell population.
Intracellular Calcium Distribution-Between 65 and 80% of the total cell calcium is mitochondrial (see also Refs. 6 and 7), with most of the remainder being associated with a small particulate (microsomal) fraction that can be sedimented by centrifugation at 120,000 X g for 15 min. This relative proportion is essentially independent of total cellular calcium content over a 10-fold range. Studies which will be reported in detail elsewhere (50) have shown that only about 0.6% of the mitochondrial calcium is free, and that the intramitochondrial free Ca9+ rises in proportion to the total calcium in the mitochondria. With liver mitochondria as normally isolated, which have about 10 nmol of calcium/mg of protein, the matrix free Ca2+ concentration is about 60 p~. A similar value can be predicted for nlitochondria in situ, since for normal cells with a total calcium content of 4.56 nmol/mg of cell protein, a mitochondrial calcium content of 2.74 nmol/mg of cell protein (Table  II), and a mitochondrial content of 28% of total protein, the calculated mitochondrial calcium content is 9.8 nmol/mg of mitochondrial protein. For a cytosolic free calcium concentration of 180 nM (0.004% of the total cell calcium), the gradient of free Ca2+ across the mitochondrial membrane in the hepatocyte is about 350 (see Ref. 51).
Hormonal Effects on Intracellular Calcium Distribution-The most important finding reported in this paper relates to the effect of the a-agonists norepinephrine and phenylephrine on calcium homeostasis in isolated hepatocytes. The present studies show that a-agonists cause a 2-to 3-fold stimulation of cytosolic free calcium that is closely related to the increase of phosphorylase a activity both kinetically and by a doseresponse relationship. Maximum changes were observed within 2 min, the peak effects were almost over within 5 min, and the half-maximal effective concentration of phenylephrine was about 5 X M. Increases of both the cytosolic free calcium and the phosphorylase a activity induced by phenylephrine were completely inhibited by the a-receptor antagonist phentolamine, but were little affected by the /?-receptor antagonist propranolol. It may be concluded, therefore, that there is a cause and effect relationship between binding of phenylephrine to the a-receptor (52) and an increase of cytosolic free calcium and activation of phosphorylase b to phosphorylase a (15,16). In contrast, glucagon failed to induce a change of cytosolic free calcium, which is in accordance with the generally accepted belief that the stimulatory effects of glucagon on carbohydrate metabolism in liver are mediated by CAMP-dependent rather than calcium-dependent mechanisms.
It has been ascertained that a-adrenergic agents inhibit glycogen synthase and activate glycogen phosphorylase and gluconeogenesis in hepatocytes (33, 53,54) and adipocytes (55) without changes in CAMP levels. Calcium is known to increase the activity of phosphorylase kinase and also to enhance the activity of CAMP-independent glycogen synthase kinases (56). Indirect evidence ( e g . Ref. 15) has indicated that a-adrenergic activation of phosphorylase may be caused by an increase of the intracellular calcium concentration. The present data confirm this hypothesis, and indicate that the maximum increase of the intracellular free calcium (from 0.19 to 0.46 PM) is the correct range to activate hepatic phosphorylase b kinase, which has a reported h, for calcium of 0.6 PM (11).
The present results also show that a-adrenergic stimulation of hepatocytes results in a net efflux of calcium from the mitochondria. After this work was completed, reports from two other laboratories have been published which reached the same conclusion (57,58). However, our results differ from the results in these and earlier studies from the same laboratories (7,16) in that we observed little or no net loss of calcium from the liver under the influence of a-adrenergic stimulation, in agreement with the data of Foden and Randle (6). We conclude, therefore, that flux of calcium across the liver plasma membrane cannot be an intrinsic factor in the mode of action of a-adrenergic agents. In our studies, the calcium released from the mitochondria appeared to be largely taken up by the endoplasmic reticulum. These effects on a redistribution of intracellular calcium can be attributed directly to the a-adrenergic hormone, since they were either abolished or greatly attenuated by the presence of a-adrenergic antagonists.
Calcium transport across the mitochondrial membrane is known to occur by two separate mechanisms (51, 59, 60). Energy-dependent calcium influx occurs by the electrophoretic transport of uncompensated calcium ions, and derives its energy from the outwardly directed proton pump of the electron transport chain (51, 59). The second mechanism of calcium transport in liver mitochondria is by a reversible electroneutral exchange of calcium with two H+ ions (51, [60][61][62]. This process has about 2 orders of magnitude lower capacity than the energy-dependent calcium uptake mechanism, while the apparent free Ca2+ K , for efflux is more than 1 order of magnitude greater than that for influx (50). Operation of these two independent influx and efflux pathways permits the mitochondria to buffer the intracellular free calcium concentration at a low value (51). An increase of cytosolic free calcium would thus be expected to result in an increased mitochondrial calcium content. Clearly, a depletion of mitochondrial calcium under the influence of a-adrenergic stimulation of hepatocytes cannot be secondary to the increased cytosolic free calcium concentration. A specific effect on mitochondria must be postulated; there is either an inhibition of the electrophoretic influx pathway or an increase of the electroneutral efflux pathway. An increase of the cytosolic free Mg'+ concentration, for instance, would have the effect of inhibiting the Ca2+ influx pathway (51,60). Alternatively, an oxidation of the mitochondrial nicotinamide nucleotides may stimulate the Ca2+ efflux pathway (61). At present, the nature and mechanism of the transducing signal from the a-adrenergic receptor on the plasma membrane to the mitochondria remains unknown. On the other hand, the increased calcium uptake by the endoplasmic reticulum after norepinephrine addition to hepatocytes is probably secondary to the rise of the intracellular free Ca'+. Rat liver microsomes are known to have an active MgATP-dependent Ca" sequestration system which has a high affinity for Ca2+ (62, 63).
In conclusion, it is evident that the mitochondrial calcium pool in the intact cell is labile. Furthermore, it may be suggested that hormone binding to the plasma membrane a receptor, by release of a chemical mediator, regulates the kinetics of Ca2+ transport across the mitochondrial membrane, which in turn directly influences the cytosolic free Ca" concentration. With a-adrenergic stimulation under the present experimental conditions, the endoplasmic reticulum Ca2+-ATPase must compete favorably with the plasma membrane