The Regulation of Gluconeogenesis in Isolated Rat Liver Cells by Glucagon, Insulin, Dibutyryl Cyclic Adenosine Monophosphate, and Fatty Acids*

SUMMARY Suspensions of rat liver cells from starved rats converted three carbon precursors to glucose and responded to hormones such as glucagon and insulin. Cells were isolated by perfusion of rat livers with calcium-free Hanks’ solution containing 0.01 g% of collagenase and 0.08 g% of hyaluronidase in the presence of 4% defatted albumin. Cells from 1% to 24-hour starved rats synthesized glucose de novo from L-alanine at rates comparable to those observed in the perfused liver. Glucagon increased cyclic adenosine 3’,5’-monophosphate accumulation in liver cells with the maximal effect being observed some 2 min after its addition. In cells from fed rats there was an immediate due to glucagon of glucose which be for by glycogenolysis. In cells from rats there almost and there a period before

glucose de novo from L-alanine at rates comparable to those observed in the perfused liver.
Glucagon increased cyclic adenosine 3',5'-monophosphate accumulation in liver cells with the maximal effect being observed some 2 min after its addition. In cells from fed rats there was an immediate increase due to glucagon of glucose release which could be accounted for by glycogenolysis.
In cells from starved rats there was almost no glycogen (0.5% for starved rats as compared to 7.5 % for fed rats), and there was a ZO-min lag period before glucose output was increased by glucagon. Glucagon stimulated gluconeogenesis in the presence of alanine, lactate, or pyruvate whereas added fatty acids inhibited gluconeogenesis from alanine and stimulated that from lactate or pyruvate.
Insulin increased glycogen deposition in cells from starved rats, and this accounted for the decrease in glucose output seen with insulin.
Dibutyryl cyclic AMP also stimulated gluconeogenesis in isolated liver cells. The effects of insulin and glucagon on liver cells were abolished by short term treatment of cell suspensions with trypsin. In contrast, the response to dibutyryl cyclic AMP was not altered in the trypsin-treated cells.
The perfused rat liver has been used estensively and successfully in studies of gluconeogenesis (l-10).
A major disadvantage of the system is that it is not esclusively a population of parenchymal cells. Secondly, the technique is rather cumbersome * This work was supported by Besearch Grants AM 10149 and AI\I 14648 from the National Institute of Arthritis and Metabolic Diseases.
$ Present address, Tufts University Medical School, Boston, Massachusetts 02111. and limits the number of variables that can be studied simultaneously.
Liver slices, at first glance, would appear to be a simple system which would offer an ideal method for studying cellular phenomena.
They provide many samples from the same animal, offering the possibility of studying many variables and simultaneously providing internal control samples. However, slices are actually nonuniform and complex systems in which oxygen diffusion to cells on the "inside" of the slice is limited, and the cells on the outer surface are mechanically damaged.
Homogeneous preparations of intact parenchymal cells isolated from one animal would allow the study of many variables at one time and would overcome many of the disadvantages mentioned in the other systems. There have been several procedures published for the isolation of rat liver cells (ll-15), but we have found only the procedure of Berry and Friend (15) to be satisfactory for studies on the hormonal regulation of gluconeogenesis.

MATERIALS
Ai'iD METHODS All experiments were done with 130-to 160-g female Sprague-Dawley rats (Charles River, CD strain) fed laboratory chow ad &turn. The rats were deprived of food for 18 to 24 hours prior to the start of each experiment except where indicated. Each rat was anesthetized by intraperitoneal injection of 10 mg of sodium pentobarbital, the abdomen opened, the portal vein cannulated, approximately 500 units of heparin injected to prevent clotting, and the liver placed in a perfusion-aeration apparatus at 37". The complete procedure should not take more than 4 min from the time the liver is cannulated until the liver is connected in the perfusion-aeration apparatus (Metaloglass Co.). The gas phase was 95% oxygen-5% carbon dioxide.
The liver was next perfused with 50 ml of calciumfree Hanks' solution containing 4% defatted albumin which was not recirculated but discarded. Following this, the blanched liver was perfused for 40 min with 50 ml of calcium-free Hanks' solution containing 4 T0 defatted albumin, 0.08 y0 hyaluronidase, and 0.01% collagenase which was recirculated.
The liver was removed at the end of the 40 min of perfusion with enzymes, placed in a plastic Petri dish, and gently pressed with a spatula to disperse the cells. The dispersed tissue was incubated with shaking for another 15 min in plastic flasks containing the (24 InM). The bovine fraction V albumin powder (pH 7.0) was defatted prior to use by the procedure of Guillory and Racker (16). The pH of the Hanks' solution containing 4% albumin was approximately 7.1 at the start of the experiment and approximately 7.0 at the end of the enzymatic perfusion of the liver. The procedure for isolation of liver cells was that of Berry and Friend (15) modified by the use of 0.01% eollagenase instead of 0.05% and 0.08% hyaluronidase instead of 0.10%. Viability of the cell preparations was assessed by the exclusion of vital dyes. Preparations used in t'hese experiments contained 90 to 98% viable cells as indicated by their exclusion of trypan blue.
Aliquots of the cell suspension containing from 5 to 8 X lo6 cells in 0.25 ml were added to plastic incubation tubes (17 X 100 mm) conta,ining 1.25 ml of the bicarbonate buffer, which contained 4oj0 albumin and the indicated substrate. The drugs and hormones were added prior to the addition of the cells unless otherwise indicated. Tubes were gassed with 95% oxygen-5% carbon dioxide, stoppered, and incubated with vigorous shaking at 37". At the end of the incubation, the samples were chilled in an ice bath, centrifuged, and 25-~1 aliquots of medium were taken fol glucose assay by t,he glucose osidase method (17).
The cell pellet remaiuillg iu each tube was analyzed for glycogen by the anthrone procedure after digestion in 30% (w/v) KOII (18). Cyclic Ah4Pl was assnyed by a.n adaptation of the binding assay described by Gilmau (19). The assay was conducted at pH 6.0 in a total volume of 35 ~1 using a binding protein 1 The abbreviation used is: cyclic AMP, cyclic adenosine 3',5'mcmophosphate.
which was purified up to the diethylaminoethyl cellulose step (19). Instead of collecting the bound cyclic AMP on Millipore filters, we separated the bound from the unbound cyclic AMP by adding 0.6 ml of Norit A charcoal suspended (60 mg per 100 ml) in 20 mM phosphate buffer (pH 6.0) containing 0.5% bovine serum albumin.
After a maximum of 10 and a minimum of 5 min, the tubes were centrifuged to pellet the charcoal. Aliquots (0.5 ml) of the supernatant were taken to determine the bound radioactive cyclic AMP. The use of charcoal is similar to the method of Brown et al. (20).
For the figure and table where cyclic AMP measurements are given, tubes were set up in parallel to hhe tubes used for glucose and glycogen measurements.
Trichloroacetic ac,id was added to the cells plus the medium to a final concentration of 5% in those tubes in which cyclic AMP was measured.

RESULTS
The procedure of Berry and Friend (15) for the isolation of liver cells resulted in parenchymal cells which responded to hormones, were morphologically intact, and could he isolated in high yield. Electron micrographs of cells isolated by the modified Berry and Friend procedure indicated that the structure of the cells, including nuclei, mitochondria, rough surfaced endoplasmic reticulum, and other cytoplasmic organelles were well preserved. 2 Berry and Friend (15) did not use buffers containing albumin, but we found that the yield of cells and hormonal sensitivity was greater for cells isolated and incubated in the presence of 4% defatted bovine albumin. 3 We also found that the additional step of perfusing the liver with calcium-and magnesium-free Hanks' solution containing 2 mM EDTA recommended by Berry and Friend (15) was unnecessary for a large yield of viable cells. Occasionally we have had problems with impurities in the albumin, collagenase, or hyaluronidase.
These problems were overcome by the use of defatted albumin and a reduction in the conoentratiou of hyaluronidase aud collagenase used in the perfusion medium.
A high yield of cells was obtained using our modification of the procedure of lserry and Friend (Table I). The leakage of lactate dehydrogenase was less than in cells prepared by the procedure of Howard et al. (13) in which liver slices were digested with collagenase and hyaluronidase ( Table I) (Table II). Upon further incubation (for up to 2 hours) the cyclic AMP accumulation in the presence of glucagon decreased but was still higher than in controls.
There was an appreciable amount of glycogen in the cells from fed rats, and glucagon increased glycogen breakdown and the appearance of glucose in the medium at all t'imes periods tested (Table II).
The increase in glucose production in cells from fed rats occurred immediately and could he accounted for solely by enhanced glycogen breakdown.
Since we were interested in examining effects of agents on gluconeogenesis, all further experiments were conducted Cth cells from rats starved for 24 hours whose glycogen content was very low (approximately 10 pg per lo6 cells or less). A test of metabolic integrity in liver preparations is the ability to synthesize glucose from 3-carbon precursors.
Krebs et al.' (9) have reported that 0.66 umole of glucose per min per g of tissue was formed, with 10 mM L-alanine as substrate, in perfused livers frotn 4%hour starved rats. The rate of glucose production obtained by Haynes (22) from liver slices of fasted rats using 10 mM nlanine as the substrate was 0.14 pniole per min per g of tissue. The rate of glucose production, 0.5 to 0.7 pmole per min per g of tissue (assuming that each liver cell weighs 2 x 1.0-g g) (23), which we observed in isolated liver cells preparations in the presence of alanine, compares favorably with that of the perfused liver and is higher than the values obtained in liver slices (Table III).
The addition of glucagon to cells from starved rats incubated with 10 mM pyruvate increased glucose accumulation in the medium after a lag period of more than 20 rnin (Fig. 1). The increase in glucose output was due to stimulation of gluconeogenesis rather than glycogenolynis, since the glycogen content in control cells was 0.5y0 (11 pg per 106 cells in glucose equivalents) and the decrease due to glucagon was less than 1 pg per lo6 cells for the experiment shown in Fig. 1.
Cyclic AMP accumulation was increased by glucagon with the maximal effect being seen at 2 min; by 20 min, the values had dropped ahnost to the level of controls (Fig. 1). At the later time period (60 to 120 min) when gluconeogenesis was stimulated, the cyclic AMP values were indistinguishable from those for controls.
Effects similar to those shown in Fig. 1 hare also been seen when liver cells from starved rats were incubated with 10 mM L-alanine.
Gluconeogenesis was stimulated by glucagon in cells from starved rats incubated in the presence of 10 mM pyruvate, lactate, or alanine, as well as 5 TIIM alanine plus either 5 mM pyruvate or lactate (Table III).
Oleate at a concentration of 0.5 or 1 mM decreased gluconeogenesis from alanine but increased gluconeogenesis from lactate or pyruvate (Table III). There was no inhibition of gluconeogenesis due to 1 mM oleate in the presence of glucngon (2.7 PM).
In an experiment conducted on aliquots of t.he same cells used for Experiment 1 in 'l':Ue III, it was fourld that epinephrine bitartrnte (50 PM) stimulated glucose release by 40% after 2 hours of incubation and 97% after 3 hours of incubation in the presence of 10 mu L-lactate.
Cyclic ,4lVIP levels in control cells were 3.  in liver cells from fed rats Liver cells (6.6 X lo6 cells per tube) obtained from a fed rat were incubated in buffer containing 47, albumin and 10 mM L-alanine for 5 min prior to addition of glucagon (2.7 PM).
The cells were then incubated for the indicated times. The values are the means of duplicate tubes. The values shown are for two different experiments and represent for each experiment the mean of duplicate t,ubes. The initial glycogen content in glucose equivalents was approximately 7 rg per lo6 cells, whereas that at the end of the incubation period was 6 ~g per lo6 for cells from the controls.
There was a drop of less than 1 rg in glycogen content for tubes incubated with glucagon and no change in cells incubated with oleate. The remainder of the studies were done with 10 mM L-alanine as the substrate.
The data in Fig. 2   to liver cells incubated with alanine (Fig. 3). The addition of sodium octanoate at concentrations ranging from 0.5 to 5 mhl depressed gluconeogenesis from alanine with the maximal inhibition being seen with 2.6 rnM octanoate (Fig. 3). The inhibitory effect of low concentrations of octanoate could be overcome by dibutyryl cyclic AMP, but at high concentrations the inhibition by octanoate predominated. The effects of octanoate on gluconeogenesis from alanine are similar to those of oleate shown in TabIe III.
Insulin has been shown to reduce the amount of glucose released by the fasting liver (7, 8), and this was also found to be the case with liver cells (Table IV and Fig. 4). Insulin increased glycogen deposition in liver cells as compared to controls (Fig. 4). The decrease in glucose release to the medium seen with insulin was probably not the result of inhibition of gluconeogenesis, since within the limits of experimental error the decrease in glucose release could be accounted for by diversion of glucose into glycogen.
In the studies shown in Fig. 4, cells were incubated with dibutyryl cyclic AMP in the presence and absence of insulin. Basal glucose production was reduced with insulin and elevated by all three concentrations of dibutyryl cyclic AMP.
In the presence of insulin, the stimulation of glucose production due to the higher concentrations of dibutyryl cyclic AMP was unaffected, whereas glucose production due to the lowest concentration of dibutyryl cyclic AMP was reduced by insulin. Dibutyryl cyclic AMP also markedly reduced the stimulation of glycogen deposition due to insulin.
In fact, glycogen con- tent was not significantly altered except in the presence of insulin alone (Fig. 4). Exposure of fat cells to trypsin has been found to selectively inactivate the receptors for insulin ,znd glucagon (24, 25), and similar results were seen in liver cells. The response of liver cells to insulin and glucagon was abolished by trypsin treatment, and basal glucose production was lowered (Table IV). However, the stimulation of glucose release by dibutyryl cyclic AMP was unaffected (Table IV). DISCUSSIOX We have tested several methods for the isolation of liver cells and found that of Berry and Friend (15) to be the most satisfactory.
The procedure of Jacob and Bhargava (II), which involved perfusion of the liver with calcium chelators followed by mechanical separation of the cells, proved unsatisfactory due to the very low yields of cells and the distorted structure seen in electron micrographs. Experience with this procedure indicated great variability in the preparations, low yields of cells, high rates of enzyme leakage (Table  I), and poor response to hormones4 all of which led to our abandonment of this procedure.
The procedure of Berry and Friend results in large numbers of parenchymal cells which can be divided among many tubes and are responsive to hormones.
It has been shown that glucose production from L-alanine is stimulated in the perfused livers from fasted rats by the  9) reported an enhancement of glucose formation from lactate in the presence of oleic acid and glucagon.
The reports by Krebs (9) and Struck (32) suggest that inhibition of gluconeogenesis by fatty acids or their metabolic products occurs at the conversion of alanine to pyrurnte.
Our data are in agreement with the observation that the inhibitory effect of fatty acids is unique for alanine since opposite effects are seen with other substrates. Also the greatest increase in glucose production was observed from la&ate in the presence of oleate plus glucagon (Table  III).
Eston et al. (4) observed a rapid increase in glucose output due to glycogenolysis with perfused livers from fed rats in the presence of glucagon (lactate as substrate).
We obtained similar results with isolated fed rat liver cells (Table II) (37) observed that glucagon in the presence of oleate did not stimulate ketogenesis when glycodiazine was present. However, they observed an additional effect of glucagon on gluconeogenesis which exceeded that observed in the absence of fatty acid. These findings indicate that fatty acid osidation may supply additional factors which are needed for optimal glucagon stimulation of gluconeogenesis and basal stimulation.
Eston et al. (3) found tha,t the ketogenic and gluconeogenic actions of glucagon in the liver were separable and that glucagon had a primary and direct effect on gluconeogenesis.
Therefore, it seems unlikely that fatty acid oxidation can be considered as the physiological regulator of gluconeogenesis (3,38).
The action of insulin in regulating glucose release is particularly interesting.
TTnder basal conditions in the normal fed rat, insulin has no effect on glycogenolysis or gluconeogenesis (10). According to Eston et al. (4,5), insulin can block the production of glucose in livers from starved rats by lowering the tissue concentration of cyclic AMP. Recently, Hepp (39) proposed that insulin inhibits glucagon-stimulated adenylate cyclase activity in particulate fractions of mouse liver. House (40) claims that the effect is not on cyclase but on phospho-die&erase.
Our results indicate that insulin opposes the stimulatory effect of glucagon on glucose release (Table TV). Whether this effect of insulin is on the accumulation of cyclic AMP, the action of cyclic -ihIP, or some other process remains to be demonstrated.
Glinsmann and llIortimore (10) were the first to report that insulin antagonizes the stimulation of glucose release by cyclic AMP in the perfused liver, a finding which was recently confirmed by Exton et al. (6). If the sole action of insulin was to inhibit adenylate cyclase, it should not affect the stimulation of glucose release by added cyclic AMP.
The major effect of insulin was not on gluconeogenesis in our experiments but rather to divert glucose into glycogen. This suggests that our effects may be similar to those of Bishop and Larner (41) a.nd Gerschenson and Casanello (42), who found a rapid increase in glycogen synthetase in liver after infusion of insulin (which could not be accounted for by changes in cyclic AMP content). Kono (25) and Fain and Loken (24) found that the response of both adipose tissue and isolated white fat cells to insulin or glucagon was abolished by preliminary treatment with trypsin.
Similar results were seen in the present studies with liver cells. The failure of trypsin treatment to affect the response to dibutyryl cyclic AMP suggests that trypsin impairs some early step in hormone action.
Possibly trypsin blocks the binding of insulin and glucagon to their receptors in hepatic cells since trypsin abolished the ability of fat cells to bind insulin (43).
The present results indicate that it is possible to obtain suspensions of liver cells by a modification of the procedure of Berry and Friend (15) which are responsive to those agents which affect gluconeogenesis in the intact liver.
The suitability of these cells for studies on the induction of enzymes, regulation of cyclic A?LIP accumulation, and other processes affected by hormones is currently under investigation.