Perspectives

The Glucose Paradox

Is Glucose a Substrate for Liver Metabolism?
Joseph Katz and J. Denis McGarry
Cedars-Sinai Medical Center, Los Angeles, California 90048;
Departments of Internal Medicine and Biochemistry, University
of Texas Health Science Center, Dallas, Texas 75235

iver contains the full complement of enzymes for the dog liver was perfused in situ with a large glucose load, it
synthesis as well as the catabolism of glucose, glycogen, and switched from glucose production to glucose uptake (see
fat, and in many species, including the pig, chicken, and man, reference 6 for review). These and numerous subsequent
it is the predominant if not the sole site of triglyceride studies formed the basis of the notion that liver regulates blood
synthesis. Because the ingestion of a carbohydrate diet after a glucose levels by releasing glucose (through glycogenolysis or
fast is followed promptly by replenishment of liver glycogen gluconeogenesis) in the postabsorptive and fasted states, and
and activation of hepatic lipogenesis, it has become commonly by removing glucose (through conversion into glycogen and
accepted that glucose is a major hepatic substrate and serves other products) after a carbohydrate meal. This concept is
as a direct' precursor for glycogen and fat. However, over the prevalent in textbooks and recent reviews (3-5).
last decade, evidence has mounted that under physiological The published experimental data on rates of glucose uptake
conditions glucose utilization by liver is rather limited and in vivo are ambiguous. In the studies of Soskin and other
that it is in fact a poor precursor for glycogen and fatty acids earlier investigators, significant glucose uptake was seen only
(much inferior to gluconeogenic substrates such as fructose,2 at very high glucose loads and at unphysiological blood glucose
glycerol, or lactate). This finding, at first sight at variance with levels; there was little if any response to insulin. A
common sense and established views, has been designated the major difficulty in such experiments is the sampling of portal
"glucose paradox" (Riesenfeld et al. [1]; Boyd et al. [2]). Our blood. Madison and co-workers (7) circumvented this obstacle
purpose here is to call attention to the problem and to present by using dogs with portacaval shunts. These animals survive
the case that longstanding concepts in this area are in need of only when kept on a high carbohydrate, low protein diet. In a
reevaluation. series of experiments, they demonstrated the cessation of
An exhaustive quotation of the voluminous literature in glucose production at a plasma glucose level of 6.5 mM and
the field will not be attempted since much of this has already an enhancement of glucose uptake with insulin. Landau et al.
been covered in the excellent reviews of Stalmans (3) and Hers (8) reexamined this problem with dogs kept on a high carbo-
(4, 5). Rather, we have adopted a selective but, we believe, hydrate (80%), low protein (16%) diet or on a "normal" diet
representative approach to the published work, both pro and (45% carbohydrate, 45% protein). Arterial, hepatic vein, and
con of our interpretations. portal vein blood was sampled. They found that in dogs kept
on "normal" diets, there was little or no uptake of glucose
Hepatic utilization of glucose even with massive glucose infusions. The production of glucose
The central role of the liver in glucose homeostasis has been ceased at arterial concentrations of 200 mg/100 ml and there
recognized since the time of Claude Bernard, over 100 years was no effect of insulin. On the other hand, with dogs kept on
ago. In the 1930's, Soskin and co-workers showed that when a high carbohydrate diet, the threshold for the hepatic glucose
Received for publication 21 May 1984. balance was 120 mg/ml, and this was lowered with insulin (9).
The only study in rats where portal blood was sampled was
1. As used here, the term "direct" implies the uptake of glucose by that by Remesy et al. (10), who conditioned the animals to a
the liver followed by a sequential series of reactions involving: glu- diet containing 42 or 79% carbohydrate and measured hepatic
cose - glucose--phosphate glucose-I-phosphate - UDP-glucose- substrate balances during food consumption after a 16-h fast.
glycogen, or glucose - glucose-6-phosphate - pyruvate - fat.
2. In liver, fructose is metabolized by phosphorylation to fructose-iP, At the time of sampling, portal vein glucose concentration was
which is split into dihydroxyacetone-P and glyceraldehyde. 8.8 mM in the rats eating the 42% carbohydrate diet but net
hepatic glucose output was still positive. Although animals
J. Clin. Invest. consuming the high carbohydrate diet exhibited hepatic glucose
© The American Society for Clinical Investigation, Inc. uptake, this was at the expense of a much higher portal vein
0021-9738/84/12/1901/09 $ 1.00 glucose concentration (13.6 mM). Absolute glucose balance
Volume 74, December 1984, 1901-1909 could not be quantified because blood flow was not measured.

1901 Glucose and Liver Metabolism

 In the great majority of studies with perfused livers, there the absorbed glucose passed through the liver to appear in the
was little or no glucose uptake at substrate concentrations systemic pool.
below 15 or 20 mM and no response to insulin. However, In summary, there is no question that in vivo a glucose
some reports indicated that when the perfusion media contained load will depress and perhaps even block glucose output. Also,
washed erythrocytes, glucose uptake could be demonstrated when blood glucose concentrations are experimentally raised
(11, 12). Recently, Storer et al. (13) showed that in a liver to a range of 12-15 mM, the liver will respond with substantial
preparation perfused with undiluted defibrinated rat blood, glucose uptake. It should be remembered that free glucose is
there was a threshold for hepatic glucose at 6.2 mM and a a rare dietary component, and even with high starch diets,
large uptake of the sugar at 12 mM. This was increased some portal glucose concentrations will seldom reach 12 mM. In
20% with insulin, up to a utilization of the order of 1.3 tmol/ rats and dogs, the threshold for glucose uptake is lowered to
min per g, and most of this was accounted for as glycogen. the physiological range when the animals are conditioned to a
Dilution of blood to a hematocrit of 19% markedly depressed high glucose diet. The available evidence shows clearly that in
glucose uptake and abolished the insulin effect (14). These humans on a Western-type diet, the uptake of glucose by the
interesting observations require confirmation. Perfusion with liver is quite low, and the administered glucose is metabolized
whole blood is difficult due to hemoconstriction factors which predominantly in the periphery.
have to be removed before perfusion (13, 14) and has rarely
been used. In addition, measurement of glucose removal from Glycogen synthesis
the perfusate will overestimate direct hepatic uptake to the
extent that erythrocyte glycolysis is operative. In most studies with isolated hepatocytes and perfused liver
In a variety of experiments with isolated hepatocytes, preparations, net glycogen synthesis was minimal when glucose
virtually no glucose utilization was observed at concentrations was the sole substrate at concentrations < 12-15 mM and
< 12 mM; significant rates were achieved only at levels of 30 became significant only when glucose levels were raised to 30
mM or above, and in no instance did insulin have a significant mM or above. Working with liver slices in 1942, Hastings and
effect (2, 15-18). This has also been the finding of the great Buchanan observed that media high in potassium stimulated
majority of studies with the perfused liver preparation.3 the incorporation of [14C]glucose into glycogen (25). More
Measurements of splanchnic glucose uptake after oral recently, Hue et al. ( 16) found with isolated hepatocytes, which
glucose ingestion in man are of limited value since they require were incubated in a medium 140 mM in sodium and 5 mM
assumptions on the rate of glucose entry into the blood and in potassium, virtually no glycogen synthesis from 15 mM
ignore intestinal glucose metabolism. Moreover, they have glucose and a rate of only 0.05-0.08 jmol/min per g from 50
produced conflicting results, even from workers at the same mM glucose. When the sodium was completely replaced with
institution. Thus, Felig et al. (21) concluded that in postab- potassium (140 mM), glycogen synthesis from glucose was
sorptive man the bulk of a 100 g oral glucose load is trapped increased fivefold. Even so, the maximal rate with 50 mM
in the splanchnic bed with only a minor fraction escaping for glucose (0.4 umol/min per g) was much less than rates obtained
use by nonhepatic tissues. Yet, from the same type of experi- in vivo. The mechanism for the stimulation by extreme
ment, Katz et al. (22) deduced that the initial metabolism of extracellular concentrations of potassium is obscure.
ingested glucose takes place primarily in skeletal muscle. The With rat hepatocytes, Hems et al. (26) and Seglen (15)
earlier claim that glucose, which is given intragastrically but found that efficient glycogen synthesis required the presence
not that delivered intravenously, is taken up efficiently by the of glucose plus gluconeogenic precursors (fructose, glycerol,
liver (21) could not be confirmed. lactate, etc.). Katz et al. (18, 27) showed that efficient glycogen
Maehlum et al. (23) measured arteriovenous (splanchnic)
synthesis also required the presence of glutamine or alanine.
These observations have been amply confirmed, as illustrated
glucose differences after an intravenous glucose infusion of 0.5 by the data from Boyd et al. (2) shown in Table I. In such a
g/kg per h to fasted, resting and exercising volunteers. Despite system glucose and glycogen are synthesized concurrently at
arterial glucose levels of 12 mM, arterio-venous differences the expense of the gluconeogenic precursor. The rate of
across the splanchnic bed were almost zero, indicating negligible glycogen deposition is similar to that attained in vivo. Exoge-
hepatic uptake of the sugar. The most extensive study in man, nous glucose is essential for glycogen synthesis but it serves as
using doubly labeled tracers, was that by Radziuk et al. (24). an activator rather than a substrate. The effect of the amino
They found that a 96-g glucose load depressed endogenous acids is to divert the gluconeogenic flux of glucose-6-phosphate
hepatic glucose production by '70%, but that at least 90% of (glucose-6-P)4 from glucose production into the formation of
UDP-glucose and glycogen. The mechanism of the amino acid
3. In many studies the incorporation of 14C from glucose into glycogen
effect is still unclear.
has been taken as a measure of glycogen synthesis. Such incorporation
frequently represents an exchange of 14C with unlabeled carbon and 4. Abbreviations used in this paper: F-2,6-P2, fructose-2,6-bisphosphate;
may occur in the absence of net glycogen deposition as a result of F-2,6-P2ase, fructose-2,6-bisphosphatase; glucose-6P, glucose-6-phos-
futile cycling between glucose, glucose-6P, and glycogen (19, 20). phate; glucose-6Pase, glucose-6-phosphatase.

1902 J. Katz and J. D. McGarry

Table I. Glycogen Synthesis in Hepatocytes from Fasted Rats circulating glucose. When 360 mg of glucose and 180 mg of
glycerol were administered together, these authors estimated
Substrate Net glycogen deposition that <20% of the carbon of glycogen was derived from glucose.
mg/g cells after 2 h That the gluconeogenic pathway itself is important for
glycogen synthesis in vivo was clearly shown by the studies of
Glucose (20 mM) 2.1 Shikama and Ui (30). NaH'4CO3 was injected intraperitoneally
Fructose (5 mM) 6.7 into fasted rats and an unlabeled glucose load was given
Lactate-Pyruvate (10-1 mM) 0 intravenously. CO2 is introduced into the carboxyl carbon of
Glutamine (10 mM) 0.4 phosphoenolpyruvate (through carboxylation of pyruvate and
Glucose; Fructose 10.8 subsequent reversible interconversion of oxaloacetate and fu-
Glucose; Fructose; Glutamine 15.6 marate) so that '4C-incorporation into glycogen becomes a
Glucose; Lactate-Pyruvate 7.5 measure of de novo synthesis from three carbon precursors.
Glucose; Lactate-Pyruvate; Glutamine 13.5 As shown in Table III, the glucose load increased greatly the
incorporation of '4CO2 into glycogen, lowered the specific
Hepatocytes from 20-h fasted rats were incubated with the indicated activity of circulating glucose, but increased markedly that of
substrates (after Boyd et al. [2]). glycogen, indicating that the latter was derived mainly from
small precursors. Similar observations were made by Newgard
et al. (31).
In vivo also, gluconeogenic compounds are better precursors To obtain a quantitative assessment of the directness versus
of glycogen than is glucose. It has been repeatedly observed indirectness of hepatic glycogen synthesis during glucose loading
that rats fed fructose (a gluconeogenic precursor2) deposit of rats, Newgard et al. (31) administered [3-3H, U-'4C]glucose
more glycogen in liver than when fed glucose. This was also to fasted animals at a rate sufficient to suppress endogenous
seen in humans in a rare study where liver and muscle biopsies glucose production. In the direct conversion of glucose into
were taken from volunteers (28). As shown in Table II, after glycogen, 3H and "'C would appear in glycogen glucose without
glucose infusion, plasma glucose rose to 310 mg/100 ml, and loss of specific activity relative to that in circulating glucose.
13 mg/g of glycogen was deposited in liver. The infusion of But to the extent that glucose first traversed the glycolytic
the same amount of fructose increased blood glucose only sequence (at whatever site) prior to glycogen synthesis, 3H
from 90 to 110 mg/ml, but hepatic glycogen deposition was would be lost at the triose phosphate level, whereas "'C could
now 45 mg/g. still enter glycogen from labeled three carbon precursors. As
More direct evidence that under most conditions in vivo shown in Table IV, when the doubly labeled glucose was given
the major sources of liver glycogen are three carbon compounds, intragastrically, the specific activity of 3H and "'C in newly
even in the presence of a glucose load, comes from studies synthesized glycogen was only 12 and 33%, respectively, of
with isotopic substrates. Moriwaki and Landau (29) found that that in the blood glucose. Thus, only a small fraction of
2 h after the intragastric administration of 360 mg/100 g body glycogen carbon could have been derived directly from glucose.
weight of [U-'4C]glucose to fasted rats, the specific activity of Similar results were obtained when glucose was infused intra-
liver glycogen-glucose was about half that of the circulating venously or consumed by the rats in their food.
glucose. When glycogen deposition was stimulated with cortisol, Radziuk (32), using a similar experimental approach in
the specific activity of glycogen was still only 60% that of man, obtained much the same results. After feeding 96 g of
labeled glucose to fasting volunteers, he calculated that at most
Table II. Glycogen Synthesis After Infusion
ofGlucose or Fructose into Humans Table III. Specific Activity of Glucose
and Glycogen after Injection of NaH'4C03
Maximum Liver glycogen Muscle glycogen
blood Blood glucose Liver glycogen
Infusion n glucose Before After A Before After
Injection Level Sp act Level Sp act
mg/100 ml mg/g mg/g
mg/100 ml cpm/mg mg/g cpm/mg
Glucose 6 310 46 59 13 15 18
Fructose 5 110 43 88 45 15 19 Saline 69 9,200 1.9 4,400
Glucose 155 3,000 4.2 7,100
Volunteers were fasted overnight and infused intravenously for 4 h
with I g/kg per h of either glucose or fructose. Liver and muscle
-
Rats were injected intraperitoneally with NaH'4CO3 and intrave-
(quadriceps) biopsies were taken at the start and the end of the ex- nously with either saline or 100 mg/100 g body weight of glucose.
periment. Initial blood glucose was 90 mg/100 ml (after Nilsson and The experiment was terminated 30 min after intravenous injection
Hultman [28]). (after Shikama and Ui [30]).

1903 Glucose and Liver Metabolism

Table IV. Glycogen Synthesis after Table V. Analyses of Liver Glycogen,
Intragastric Administration of Glucose Plasma Glucose, and Plasma Lactate after Infusion of
[1_4C]_ or [6-"C]Glucose into Fasted Rats
Portal plasma glucose 7.8 mM
Liver glycogen 16.5 mg/g "C on glucose C-l C-6
Liver glycogen (mg/g) 16.3±1.7 23.8±1.9
Relative specific activity
Arterial glucose-3H 100 Relative specific activity*:
Arterial glucose-'4C 100 Plasma glucose 0.75±0.03 0.83±0.04
Liver glycogen-3H 12 Plasma lactate 0.31±0.02 0.30±0.02
Liver glycogen-'4C 33 Liver glycogen 0.43±0.02 0.42±0.03
Randomization* in:
20-h starved rats were infused intragastrically with 167 mg/100 g Plasma glucose 0.095±0.01 0.11±0.02
body weight per h of [3-3H, U-'4C]glucose for 2 h. Relative specific Liver glycogen 0.27±0.01 0.27±0.02
activity is expressed as a percent of that in the infused glucose. Ini-
tially, portal glucose was 5 mM and liver glycogen 0.7 mg/g (after 20-h fasted rats were infused with glucose (labeled as indicated) at a
Newgard et al. [31]). rate of 167 mg/100 g body weight per h for 2 h. Values are
means±SEM for three to six animals. (Data from unpublished stud-
one third of the liver glycogen deposited could have come ies by Newgard, Golden, Foster, McGarry, and Katz).
directly from circulating glucose and that at least two thirds * Expressed as counts per minute per micromoles with the specific

had a gluconeogenic origin. activity of infused glucose set at 1.00.

The lack of randomization of "`C in glycogen after the t Expressed as CA4, C-5, C-6/(C-l-C-6), or C-1, C-2, C-3/(C-I-C-6)
injection of specifically labeled ["'C]glucose has been previously in experiments with [1-"4C] and [6-'4C]glucose, respectively.
taken as evidence that glucose is directly converted into
glycogen without prior cleavage. Hers (33) injected rats with 50% of that of circulating glucose. However, since the precursors
[1-"'C]glucose and [l-"'C]fructose and isolated liver glycogen. are also labeled, this is an underestimate. For a reliable
Glycogen formed from fructose had 50% of the "'C in C-1 calculation, determination of the specific activity of phospho-
and 30% in C-6. With glucose, 85% of the "'C in glycogen was enolpyruvate is required. Assuming that the specific activity
in C-1 and very little in C-6. These observations have been of phosphoenolpyruvate is one half that of plasma lactate, and
repeatedly confirmed. However, the conclusion that glucose is taking into account the randomization in glucose, calculations
predominantly converted to glycogen without cleavage is not to be presented elsewhere indicate that at least 70% of the
warranted. If the glycogen was formed in part directly from glycogen arises from three carbon precursors, in spite of the
glucose and in part after glucose cleavage to pyruvate, the 14C- infusion of a large glucose load. It should be stressed that the
yield in glycogen would depend on the relative specific activities "4C-data apply to the relative incorporation of carbon into
of these precursors. After a single injection, the specific activity glycogen rather than net conversion of glucose into glycogen.
of glucose is initially high but decays rapidly, and that of In the experiments of Table V, there was still hepatic output
lactate attains only a low value. Thus, a short period after of glucose at 20-25% of the basal rate. '4C-incorporation in
injection, the incorporation of 14C by direct conversion of the absence of net uptake of glucose or in the presence of
glucose will much exceed that from other labeled precursors, substantial glucose breakdown has been frequently observed
and most of the 14C will be incorporated in the early period in vivo and in vitro. Irrespective of the "'C-balance, this means
(minutes) after injection. Also, the administration of large that glucose, as far as hepatic balance is concerned, does not
boluses of glucose is attended by severe hyperglycemia which contribute net carbon to liver constituents. The "'C-incorpo-
will serve to force glucose through the sluggish glucokinase ration may be formally interpreted as an exchange of labeled
reaction in liver, and thus, favor the direct pathway of glycogen for unlabeled carbon.3 Alternatively, glucose uptake and pro-
synthesis (see below). A valid analysis requires conditions of duction proceed in different cell populations. We are concerned
physiological concentrations and constant specific activities of here with the physiological function of liver as a whole in
glucose and lactate, as are approximated with a continuous glucose balance.
infusion of [14Ciglucose. Indeed, we found that when a glucose The studies described above establish that, contrary to
load labeled either in C- 1 or C-6 was infused for 2 or 3 h, the widespread belief, continued carbon flow through the gluco-
isolated glycogen had 27% of its activity in the "opposite" end neogenic reactions of liver plays an important role in hepatic
of the glucose molecule compared with that of the infused glycogen synthesis during the fasted-to-fed transition. The fact
material (Table V). Randomization in circulating glucose at that this process is essential for efficient glycogen synthesis
that time was only 10%. A minimal value for the contribution
- postprandially has recently been convincingly demonstrated
of the nonglucose precursors to glycogen carbon may be by independent studies from two laboratories (Sugden et al.
obtained from the specific activity of glycogen which was 40- [34] and Newgard et al. [35]). In both cases, fasted rats were

1904 J. Katz and J. D. McGarry

treated with 3-mercaptopicolinic acid (an inhibitor of phos- [ 4C]glucose and 3HOH into liver and depot fat over a 24-h
phoenolpyruvate carboxykinase) prior to the administration of period. (In adipose tissue, glucose is the preferred substrate for
glucose. This resulted in a reduction in liver glycogen deposition lipogenesis.) They found that even after a meal most of the
of 80-90% despite similar elevation of the blood glucose liver fat was derived from nonglucose precursors. Baker et al.
concentration in control and experimental groups. Moreover, (39) came to the same conclusion. These authors fed mice
when the infused glucose was labeled with `4C in the 1 position, ['4C]glucose and injected them with 3HOH. Feeding promptly
the liver glycogen formed in the absence of 3-MPA showed stimulated hepatic lipogenesis but, as shown in Table VII,
extensive label randomization, whereas little randomization under most conditions the fatty acids were derived from
was seen in the small amount of glycogen formed in the nonglucose precursors. Only when an enormous load of glucose
presence of the inhibitor (indicating that when gluconeogenesis was administered to fed mice were most of the fatty acids
was blocked only the relatively inefficient, direct pathway for derived directly from this substrate.
hepatic glycogen synthesis was operative [35]). Rats refed a high carbohydrate diet after a fast exhibit a
In summary, there is overwhelming evidence that in rats very high liver glycogen content and very efficient hepatic
raised on "mixed" or sucrose diets, the capacity for hepatic lipogenesis. Clark et al. (36) have shown that in hepatocytes
glucose utilization is limited and glycogen deposition in liver from these rats, glycogen is the major source of lipid. It is
is predominantly a gluconeogenic process. The administration likely that refeeding after a fast leads first to replenishment of
of glucose stimulates glycogen deposition, but the effect is the glycogen stores which then serve as a source of pyruvate
mainly indirect with the products of glucose cleavage in the and acetyl-coenzyme A for lipogenesis. The results of Boyd et
periphery acting as the major proximate precursors of glycogen. al. (2) support this sequence in hepatocytes.
In rats and dogs conditioned on high glucose or starch diets Thus, the evidence in rodents is clearcut; glucose admin-
where the capacity for hepatic glucose utilization appears to istration acts as a trigger for lipid synthesis, but the hexose
be increased, glucose might become an important source of serves only indirectly (via lactate or glycogen) as a precursor
liver glycogen. Available evidence in humans, at least those for this process in liver. There are no relevant experiments
on a mixed Western diet, shows a limited capacity for hepatic with humans.
glucose utilization and that the major sources of liver glycogen
after a glucose meal are its cleavage products. There are no There is no "glucose paradox" with Japanese quail
data from human populations consuming predominantly starch The studies reviewed above were with rodents, dogs, and
diets. humans; very little relevant information is available for other
animal species. The exception is our study with a bird, the
Hepatic lipogenesis Japanese quail, which is strikingly different in that this species
Using rat hepatocytes, Clark et al. (36) noted that lactate was uses glucose as a major precursor for hepatic glycogen and
greatly superior to glucose as a precursor for fatty acids. As lipid synthesis. The Japanese quail (Coutournix coutournix
illustrated in Table VI, 2 mM lactate, alone or in the presence japonica) is a small (200 g) domestic bird. Plasma glucose is
of glucose, was a much better precursor than was 10 mM 15-20 mM and the in vivo rate of hepatic lipogenesis is very
glucose. A preferential use of lactate over glucose for lipogenesis high, especially during egglaying (Riesenfeld et al. [1] and
has also been shown in perfused rat liver by Brunengraber et references therein). Isolated quail hepatocytes, when incubated
al. (37) and Salmon et al. (38). The findings in hepatocytes with glucose as sole substrate, utilize the sugar at a moderate
have been amply confirmed (Boyd et al. [2]). rate, but compared with lactate its incorporation into fatty
Hems et al. (26) compared in rats the incorporation of

Table VI. Incorporation of Glucose and Table VII. Contribution of Glucose and Nonglucose
Lactate into Lipid by Rat Hepatocytes Carbon to Fatty Acid Synthesis in Mouse Liver In Vivo
'4C recovered in 3HOH in Micromoles of fatty
acid per minute per mouse from
Sub- Fatty Lipid Fatty Lipid
strate Glucose CO2 acids glycerol acids glycerol Dietary Glucose Plasma Glucose Nonglucose
state intubated glucose carbon carbon Total
mm patom C or H/g per h
mg/ml
Lactate 2 116 60 9.4 3.8 34 20
Glucose 10 - 40 1.4 4.6 24 15 Fasted 1.1 0.3 2.5 2.8
Lactate 2 42 821 60 9.61 Fed 2.2 10.0 73.0 83.0
and 124 61 l14 64 20 Fasted 250 mg* 5.5 18.0 53.0 71.0
glucose 10 - 42 J 1 4.2 Fed 250 mg* 3.2 220.0 110.0 330.0

Rats were meal fed a high carbohydrate diet. Lactate and glucose were labeled * Given by stomach tube as 50% glucose, equivalent to 12 g/kg body weight
uniformly with `4C (after Clark et al. [36]). (after Baker et al. [39]).

1905 Glucose and Liver Metabolism

Table VIII. Effect ofAlanine on Glucose and Lactate Metabolism in Hepatocytes from Japanese Quail
4C in
3H from
Labeled substrate Alanine Glucose Lactate CO2 Glucose Lipid 3HOH in lipid
mM umol/g/h patom C or H/g/h

[U-t4C]Glucose none -8 34 - 8 54
2.5 -54 104 64 188
10.0 -80 - 120 186 242
[U-'4C]Lactate none +84 -282 330 324 94 94
10.0 +16 -232 320 44 198 272

-, uptake; +, production. Cells were from mature egglaying birds. The concentration of glucose was 10 mM and of lactate 20 mM. Some 80-
90% of the added alanine was taken up by the cells (after Katz et al. [40]).

acids is limited (1, 40). However, when supplemented with Glucose uptake by hepatocytes and perfused liver becomes
several amino acids, glucose uptake, its incorporation into substantial at unphysiologically high substrate levels (2 ;9mol/
glycogen, and most noticeably, its incorporation into fatty min per g at 50 mM) and it shows no saturation even at 80
acids increases many fold. In the presence of alanine, glucose mM (43). The Km (or Ko.5, half maximal rate) appears to be
is equal or superior to lactate as a precursor of glycogen and - 30-40 mM (44) rather than the 5 mM calculated for the
fatty acids, as illustrated in Table VIII. In the preparation of purified enzyme. The results support the existence of another
these hepatocytes, there is an almost complete loss of glutamine form of glucose phosphorylating enzyme in liver with a very
and alanine from the cells. The latter is taken up very rapidly high Ko.5. Nordlie (45) suggested that this function might be
from the medium, with replenishment of near normal amino subserved by the microsomal glucose-6Pase with pyrophosphate
acid content of the tissue. Thus, to date the quail hepatocyte or carbamyl phosphate acting as donors of the high energy
is the only liver cell system found in which under physiological phosphate. So far, definite proof that such a mechanism
conditions glucose uptake is high and where this substrate is operates physiologically is lacking.
an efficient source of glycogen and fatty acids. To conclude, the limited capacity for hepatic glucose
utilization in the rat, and probably also in man,' is probably
Enzymes of glucose phosphorylation and uptake due to the low level of glucokinase combined with variable
The liver content of hexokinase is very low, and the major rates of futile cycling caused by the activity of glucose-6Pase.
enzyme for glucose phosphorylation is glucokinase (for review,
see reference 41). In the rat, the enzyme is unaffected by Regulation of hepatic glucose-6P metabolism
fasting periods up to 20 h (31), but is depressed by prolonged If during the fasted-to-fed transition the major fraction of liver
starvation (48-72 h) and in diabetes (41); it is elevated only glucose-6P is gluconeogenic in origin, how is this centrally
to a limited extent by a high carbohydrate diet. The Km for located metabolite diverted away from free glucose formation
glucose was reported to be -l0 mM (41), but a more recent and into the pathway of glycogen synthesis? A widely accepted
value, obtained with a highly purified rat liver enzyme, was theory is that proposed by Hers (4). Its essential features are
-5 mM (42). Maximal activity with cell extracts measured at that glucose loading causes activation of glycogen synthesis
100 mM glucose is 2-3 Amol/min per g, but at physiological secondary to inhibition of glycogen phosphorylase. This is
concentrations of substrate it is much less, 0.35 and 0.6 Atmol/ expected to "pull" UDP-glucose and glucose-6P into glycogen.
min per g for fed rats at 5 and 10 mM glucose, respectively, The predicted fall in glucose-6P concentration is considered
and also, considerably less than the rate of glycogen deposition sufficient to attenuate glucose-6P flux through the glucose-
seen in vivo (31). Moreover, rates of glucose phosphorylation 6Pase step since hepatic glucose-6P levels (generally in the
as measured by enzyme assays in liver homogenates do not region of 0.1 mM) are far below the Km of glucose-6Pase for
correspond to rates of glucose uptake in isolated hepatocytes. this substrate (2-3 mM). An opposing view, based primarily
In the latter, much of the glucose-6P is hydrolyzed back to on theoretical grounds, is that with glucose loading glucose-6P
glucose, due to the action of glucose-6-phosphatase (glucose- is "pushed" into glycogen as a result of inhibition of glucose-
6Pase). The phosphorylation of glucose in intact cells can be
measured by the loss of tritium from position 2 of glucose 5. In a survey of different mammalian species for hepatic glucokinase
(20). As illustrated in Table IX, the uptake of glucose at activity, Lauris and Cahill (46) stated that the activity of this enzyme
concentrations below 20 mM was only a small fraction of the "was from extremely low to absent in the toadfish, guinea pig, cat and
rate of phosphorylation. Futile cycling can also be extensive man." To our knowledge this interesting observation, made almost
in vivo (19). twenty years ago, has never been followed up.

1906 J. Katz and J. D. McGarry

Table IX. Glucose Phosphorylation and control of the glycolytic-gluconeogenic sequences in liver (for
Glucose-6-Pase in Rat Hepatocytes review, see reference 52). Its concentration is low in the fasted
Glucose state and increases dramatically with refeeding, consistent with
Change phos- Glucose its established role as an activator of phosphofructokinase-1
Diet Lactate in glucose* phorylation 6-Pase and inhibitor of fructose-1,6-bisphosphatase (F-1,6-P2ase).
pmol/h/g wet weight Moreover, its concentration in hepatocytes from fasted rats
- -24 106 82 rises many fold within minutes after exposure of the cells to
Meal fed + +22 102 124 elevated levels of glucose and insulin (53-57). In the context
- +6 32 38
of the present discussion, the latter observations presented
Fasted
+ +34 26 60 another curious paradox. If hepatic F-2,6-P2 levels increase
acutely in vivo with refeeding of fasted animals, how could
Fed - +4 7.8 12 glucose-6P continue to be generated so efficiently from three
diabetic + +46 4.0 50
carbon precursors (since F- 1 ,6-P2ase should now be shut
-, uptake; +, production. down)? Studies by Kuwajima et al. (58) promise to resolve
* Glucose uptake was measured by analysis of the medium; initially [glucose]
was 15 mM, [lactate] was 209 mM. Glucose phosphorylation was determined
this metabolic dilemma. Fasted rats were given a continuous
by the 3HOH yield from [2-3H]glucose corrected for incomplete detrition. Glu- intragastric infusion of glucose or were allowed to refeed on a
cose-6-Pase was obtained by difference (after Katz et al. [20]). regular chow diet ad lib. In both cases, despite acute elevation
of the blood glucose (and presumably insulin) concentration,
hepatic F-2,6-P2 levels remained low for the first 3-4 h. They
6Pase (47). More recent studies suggest that both "pull" and began to rise towards fed values only after liver glycogen stores
"push" mechanisms might in fact be operative. Niewoehner had been largely repleted (58). Similar observations were made
et al. (48) administered glucose by gavage to fasted rats in the by Claus et al. (59). Thus, a theoretical obstacle to active
amount of 4 g/kg body weight. This huge load increased portal gluconeogenic carbon flow into glycogen for several hours into
glucose concentration to 15 mM; hepatic glycogen synthesis the postprandial period was removed. The interesting possibility
was rapid and the tissue UDP-glucose level was halved. How- is raised that the late rise in hepatic F-2,6-P2 levels serves to
ever, the concentration of glucose-6P increased from 0.095 to prevent excessive glycogen deposition from gluconeogenic pre-
0.15 gmol/g liver at 20 min. Newgard et al. (49) infused cursors so that the latter, together with glycolytically derived
glucose intragastrically to fasted rats for 3 h at a rate of 167 pyruvate, are shunted into the pathway of lipogenesis (58).
mg/100 g body weight per h. Both UDP-glucose and glucose- The precise biochemical mechanisms underlying this temporal
6P levels fell at 1 h. Although the former remained low, the sequence of events with refeeding and the basis for the in
latter rebounded to basal levels by 2 h, at which time glycogen vitro/in vivo discrepancy remain to be elucidated.
synthesis was brisk and carbon flow through glucose-6Pase was
markedly suppressed. Role of insulin
These observations suggest that, in addition to causing As noted earlier, when isolated hepatocytes from fasted animals
activation of glycogen synthase, glucose loading results in were provided with appropriate substrate mixtures, insulin was
inhibition (or deactivation) of glucose-6P hydrolysis. A similar neither necessary for nor stimulatory to the process of glycogen
phenomenon might have been at work in the hepatocyte
studies of Katz et al. (20) and Okajima et al. (50) where amino (or fat) synthesis. Yet, ever since the classic studies of Madison
acids and mercaptopicolinate were shown to divert gluconeo- and co-workers, (60) it has been recognized that in the intact
genically derived glucose-6P away from glucose and into animal insulin is essential for these anabolic events to occur.
glycogen formation concomitant with rising levels of the How can these apparently conflicting observations be recon-
hexose phosphate. From the work of Arion and co-workers ciled? Although a simple answer has yet to emerge, it seems
(51), it appears that the phosphohydrolase itself is a nonspecific likely that the essentiality of insulin for the catabolic-to-
enzyme that resides within the lumen of the endoplasmic anabolic transition of liver metabolism in vivo stems, at least
reticulum. Specificity for glucose-6P is conferred by the presence in part, from its ability to suppress the secretion of glucagon
in the membrane of a translocase that transports glucose-6P and to antagonize the catabolic actions of the a-cell hormone
from the extra- to the intramicrosomal compartment. Our bias (and other counterregulatory hormones). Such a role for
is that the translocase rather than the phosphohydrolase itself insulin in liver is to be contrasted with its well-known ability
is subject to metabolic regulation. A search for potential to promote glucose uptake and metabolism in muscle and fat
regulators seems worthwhile. tissue when present as the sole hormone. It is also fully in
accord with the early studies of Exton and Park (61) on the
Role offructose-2,6-bisphosphate (F-2,6-P2) in thefasting- opposing roles of insulin and glucagon in the control of hepatic
to-fed transition of liver metabolism cyclic AMP levels and with the bihormonal hypothesis for the
The newly discovered regulatory molecule, F-2,6-P2, has added regulation of glucose homeostasis (62) and ketone body pro-
a new and important dimension to our understanding of the duction (63).

1907 Glucose and Liver Metabolism

Conclusion regimen differing from that of Western man would be of
Caution is required in extrapolating from observations made interest.
in vitro to metabolic and regulatory events operative in vivo. Finally, it should be emphasized that we do not challenge
In the former situation, normal metabolic effectors might be the fact that in the intact organism a major fraction of dietary
lost or subtle damage to the cell membrane might alter cellular glucose is ultimately converted into liver glycogen and fat.
behavior and response to hormones. However, most of the in What we question is whether, in quantitative terms, the
vitro studies cited above are consistent with the experimental ingested glucose is the immediate precursor of these storage
observations made with intact animals. materials and whether liver is the primary site of glucose
When taken together, the data presented support the disposal. As to these issues, which undoubtedly will be inter-
concept that in most conditions only with excessive glucose preted by many as a recent departure from accepted dogma,
loads, which are rarely encountered by rat or man, would we might do well to recall that the notion of an indirect
glucose be taken up efficiently in direct manner by liver. pathway from glucose to liver glycogen was expressed as long
Rather, it appears that in the immediate postprandial phase, ago as 1944 by Boxer and Stetton (68) but subsequently fell
dietary carbohydrate is converted into liver glycogen and fat into disfavor. Even more thought provoking are the words of
largely via an indirect mechanism involving the sequence: Claude Bernard written over a century ago, some 20 years
glucose -* C3 unit -- glycogen and lipids. Although the nature after his discovery of liver glycogen. The question under
of the three carbon intermediate has not been firmly identified, discussion was: What is the source of liver glycogen when
lactate (and possibly alanine) would seem to be a likely fasted animals are refed? His response was as follows: "I will
candidate since it is produced during glucose absorption from carefully refrain from making a definite decision on such a
the gut (Boyd et al. [2]; Nicholls et al. [64]) and by muscle fundamental question. It is not such a simple matter as it
and erythrocytes.6 Implicit in this formulation is that carbon appears. The indisputable fact is that the administration of
flow from the C3 level to glucose-6P must remain active for at cane sugar considerably increases the liver glycogen content;
least several hours after carbohydrate ingestion, but that the but how does the sugar act in this case-as a 'nutritive
glucose-6P formed is now diverted away from the glucose- stimulator' or as a substance which is directly converted to
6Pase reaction and into the pathway of glycogen synthesis. glycogen? I am inclined to believe, I must confess, that the
The mechanism of this crucial metabolic switch remains to be first suggestion is the more correct." (Claude Bernard, "Lecons
delineated. One factor is probably a glucose-induced activation sur le Diabite," Paris, 1877).
of glycogen synthase. An additional possibility, and one that
we find intuitively attractive, is that glucose loading somehow Acknowledgments
leads to inhibition (or deactivation) of the glucose-6Pase system, The authors' studies were supported by United States Public Health
which would direct glucose-6P carbon away from glucose and Service grants AM 18573, AM07307, AM 12604, AM 19576, and by the
into glycogen formation. In any event, clarification of the 30K Fund.
nature of this permissive glucose effect, predictable from the
early work of Soskin and colleagues, will fill a longstanding References
void in our understanding of glucose homeostasis.
The observations we have cited were made predominantly 1. Riesenfeld, G., P. A. Wals, S. Golden, and J. Katz. 1981. J.
with rodents, dogs, and man on standard diets. Care is required Biol. Chem. 256:9973-9980.
in generalizing to other animal species and to varied environ- 2. Boyd, M. E., E. B. Albright, D. W. Foster, and J. D. McGarry.
mental and dietary conditions. In dogs and rats, prolonged 1981. J. Clin. Invest. 68:142-152.
conditioning to a high carbohydrate diet alters the metabolic 3. Stalmans, W. 1976. Curr. Top. Cell Regul. 11:51-97.
patterns. Studies with human populations with a dietary 4. Hers, H. G. 1976. Annu. Rev. Biochem. 45:167-189.
5. Hers, H. G. 1981. In Short-Term Regulation of Liver Metabolism.
L. Hue, and G. Van de Werve, editors. Elsevier/North-Holland,
6. While studies in the rat would be consistent with this interpretation Amsterdam. 105-117.
(Table V), the situation in other species is less clear. For example, 6. Soskin, S., and R. Levine. 1946. Carbohydrate Metabolism.
Cherrington and co-workers (65) have recently concluded that dogs University of Chicago Press, Chicago, IL. 3-315.
consuming one mixed meal per day exhibit no net uptake of glucose 7. Madison, L. C., B. Combes, R. Adams, and W. Strickland.
by the liver, yet display a marked postprandial hyperlactatemia that is 1960. J. Clin. Invest. 39:507-522.
hepatic in origin. They postulated that in this model postprandial 8. Landau, B. R., J. R. Leonards, and F. M. Barry. 1961. Am. J.
hepatic glycogen synthesis must derive from sources other than circu- Physiol. 201:41-46.
lating lactate and glucose. Also to be noted are the studies by Jackson 9. Leonards, J. R., B. R. Landau, J. W. Craig, F. I. R. Martin, M.
et al. (66) and Radziuk and Inculet (67) in which substrate balance Miller, and F. M. Barry. 1961. Am. J. Physiol. 201:47-54.
across the human forearm showed a net uptake rather than a net 10. R6mesy, C., C. Demigne, and J. Aufrere. 1978. Biochem. J.
output of lactate in response to oral glucose loading. Clearly, further 170:321-329.
work is needed to elucidate the nature and sources of liver glycogen 11. Burton, S. D., and T. Ishida. 1965. Am. J. Physiol. 209:1145-
precursors in the postprandial period. 1161.

1908 J. Katz and J. D. McGarry

 12. Glinsmann, W. H., E. P. Herm, and A. Lynch. 1969. Am. J. 40. Katz, J., P. A. Wals, and S. Golden. 1983. In Isolation,
Physiol. 216:698-703. Characterization and Use of Hepatocytes. R. A. Harris and N. W.
13. Storer, G. B., D. L. Topping, and R. P. Trimble. 1981. Febs Cornell, editors. Elsevier Biomedical, New York. 505-516.
Lett. 136:135-137. 41. Weinhouse, S. 1976. Curr. Top. Cell. Regud. 11:1-50.
14. Topping, D. L., R. P. Trimble, and G. B. Storer. 1981. 42. Storer, A. C., and A. Cornish-Bowden. 1976. Biochem. J.
Biochem. Int. 3:101-106. 159:7-14.
15. Seglen, P. 0. 1974. Biochim. Biophys. Acta. 338:317-336. 43. Alvares, F. L., and R. C. Nordlie. 1977. J. Biol. Chem.
16. Hue, L., F. Bontemps, and H. G. Hers. 1975. Biochem. J. 252:8404-8414.
152:105-114. 44. Singh, J., and R. C. Nordlie. 1983. Febs Lett. 150:325-328.
17. Katz, J., P. A. Wals, S. Golden, and R. Rognstad. 1975. Eur. 45. Nordlie, R. C. 1974. Curr. Topics Cell. Regul. 8:33-117.
J. Biochem. 60:91-101. 46. Lauris, V., and G. F. Cahill, Jr. 1966. Diabetes. 15:475-479.
18. Katz, J., S. Golden, and P. A. Wals. 1979. Biochem. J. 180:389- 47. El-Refai, M., and R. N. Bergman. 1976. Am. J. Physiol.
402. 231:1608-1619.
19. Katz, J., and R. Rognstad. 1976. Curr. Top. Cell. Regul. 48. Niewoehner, C. B., D. P. Gilboe, and F. Q. Nuttall. 1984. Am.
10:237-289. J. Physiol. 246:E89-E94.
20. Katz, J., P. A. Wals, and R. Rognstad. 1978. J. Biol. Chem. 49. Newgard, C. B., D. W. Foster, and J. D. McGarry. 1984.
253:4530-4536. Diabetes. 33:192-195.
21. Felig, P., J. Wahren, and R. Hendler. 1975. Diabetes. 24:468- 50. Okajima, F., and J. Katz. 1979. Biochem. Biophys. Res.
475. Commun. 87:155-162.
22. Katz, L. D., M. G. Glickman, S. Rapoport, E. Ferrannini, and 51. Arion, W. J., A. J. Lange, H. E. Walls, and L. M. Ballas. 1980.
R. A. DeFronzo. 1983. Diabetes. 32:675-679. J. Biol. Chem. 255:10396-10406.
23. Maehlum, S., J. Jervell, and E. D. R. Pruett. 1976. Scand. J. 52. Hers, H. G., and E. Van Schaftingen. 1982. Biochem. J. 206:1-
Clin. Lab. Invest. 36:415-422. 12.
24. Radziuk, J., T. J. McDonald, D. Rubenstein, and J. Dupre. 53. Van Schaftingen, E., L. Hue, and H. G. Hers. 1980. Biochem.
1978. Metab. Clin. Exp. 27:657-669. J. 192:887-895.
25. Hastings, A. B., and J. M. Buchanan. 1942. Proc. Natl. Acad. 54. Richards, C. S., and K. Uyeda. 1980. Biochem. Biophys. Res.
Sci. USA. 28:478-482. Commun. 97:1535-1540.
26. Hems, D. A., P. D. Whitton, and E. A. Taylor. 1972. Biochem. 55. Hue, L., P. F. Blackmore, H. Shikama, A. Robinson-Steiner,
J. 129:529-538. and J. H. Exton. 1982. J. Biol. Chem. 257:4308-4313.
27. Katz, J., S. Golden, and P. A. Wals. 1976. Proc. Nat!. Acad. 56. Pilkis, S. J., T. D. Chrisman, M. R. El-Maghrabi, A. Colosia,
Sci. USA. 73:3433-3437. E. Fox, J. Pilkis, and T. H. Claus. 1983. J. Biol. Chem. 258:1495-
28. Nilsson, L. H., and E. Hultman. 1974. Scand. J. Lab. Invest. 1503.
33:5-10. 57. Chaekal, O., J. C. Boaz, T. Sugano, and R. A. Harris. 1983.
29. Moriwaki, T., and B. R. Landau. 1963. Endocrinology. 72:134- Arch. Biochem. Biophys. 225:771-778.
145. 58. Kuwajima, M., C. B. Newgard, D. W. Foster, and J. D.
30. Shikama, H., and M. Ui. 1978. Am. J. Physiol. 235:E354- McGarry. 1984. J. Clin. Invest. 74:1108-11 1.
E360. 59. Claus, T. H., F. Nyfeler, H. A. Muenkel, M. G. Burns, and
31. Newgard, C. B., L. J. Hirsch, D. W. Foster, and J. D. McGarry. S. J. Pilkis. 1984. Biochem. Biophys. Res. Commun. 122:529-534.
1983. J. Biol. Chem. 258:8046-8052. 60. Madison, L. L. 1969. Arch. Intern. Med. 123:284-292.
32. Radziuk, J. 1982. Fed. Proc. 41:110-116. 61. Exton, J. H., and C. R. Park. 1972. Handb. Physiol. 1:437-
33. Hers, H. G. 1955. J. Biol. Chem. 214:373-381. 455.
34. Sugden, M. C., D. I. Watts, T. N. Palmer, and D. D. Myles. 62. Unger, R. H. 1976. Diabetes. 25:136-151.
1983. Biochem. Int. 7:329-337. 63. McGarry, J. D., and D. W. Foster. 1980. Annu. Rev. Biochem.
35. Newgard, C. B., S. V. Moore, D. W. Foster, and J. D. McGarry. 49:395-420.
1984. J. Biol. Chem. 259:6958-6963. 64. Nicholls, T. J., H. J. Leese, and J. R. Bronk. 1983. Biochem.
36. Clark, D. G., R. Rognstad, and J. Katz. 1974. J. Biol. Chem. J. 212:183-187.
249:2028-2036. 65. Davis, M. A., P. E. Williams, and A. D. Cherrington. 1984.
37. Brunengraber, H., M. Boutry, and J. M. Lowenstein. 1973. J. Am. J. Physiol. In press.
Biol. Chem. 248:2656-2669. 66. Jackson, R. A., N. Peters, U. Advani, G. Perry, J. Rogers,
38. Salmon, D. M. W., N. L. Bowen, and D. A. Hems. 1974. W. H. Brough, and T. R. E. Pilkington. 1973. Diabetes. 22:442-458.
Biochem. J. 142:611-618. 67. Radziuk, J., and R. Inculet. 1983. Diabetes. 32:977-981.
39. Baker, N., D. B. Learn, and K. R. Bruckdorfer. 1978. J. Lipid 68. Boxer, G. E., and D. Stetten, Jr. 1944. J. Biol. Chem. 155:237-
Res. 19:879-893. 242.

1909 Glucose and Liver Metabolism