Effect of physiological concentrations of insulin and glucagon on the relationship between nonesterified fatty acids availability and ketone body production in humans

Effect of Physiological Concentrations of Insulin and Glucagon on the Relationship Between Nonesterified Fatty Acids Availability and Ketone Body Production in Humans M. Beylot,

S. Picard,

C. Chambrier,

H. Vidal,

M. Laville,

R. Cohen,

A. Cotisson,

and R. Mornex

To determine the effect of insulin and glucagon on the transformation of nonesterified fatty acids (NEFA) into ketone bodies (KB), we measured simultaneously in normal subjects NEFA and KB kinetics at different NEFA levels in the presence of basal (control test) or increasing insulin concentrations with glucagopenia (somatostatin + insulin infusion, insulin test) and without glucagopenia (somatostatin + lnsulin + glucagon infusion, glucagon test). NEFA levels were controlled during these tests by an intravenous (IV) infusion of a triglyceride emulsion. During the control test, a moderate increase of NEFA (464 ? 30 to 715 -c 56 pmol/L) increased the percentage of NEFA converted into KB (13.3% -c 1.4% to 26.4% + 2.1%. P < .05), and there was a linear relationship between this percentage and NEFA levels (I = .788,P < .Ol). During the insulin and glucagon tests,the progressive increase in NEFA induced by the triglyceride emulsion infusion was associated, despite the increase of insulinemia, with an increase in KB production rate (P < .05) and in the proportion of NEFA used for ketogenesis in the presence (8.1% -t 1.2% to 14.2% 2 6.3%, P < .05) and absence (15.7% ? 2.8% to 25.2% 2 3.99%. P < 0.05) of glucagopenia. In both tests, this percentage was always linearly related with NEFA levels (P < .05) and the slopes of these relationships were comparable to that observed in the control test. However, the fraction of NEFA used for ketogenesis was always higher (P < .05) during glucagon substitution than in its absence. Last, the study of type 1 diabetic patients with various degrees of metabolic control showed that this percentage of NEFA used for ketogenesis was rather related to increasing NEFA level (P < .05) than to decreasing insulin concentration (P > .lO). In conclusion, within the range of NEFA and insulin attained in this study, we did not obtain evidence for a direct hepatic antiketogenic action of insulin, whereas glucagon appears to stimulate ketogenesis at physiological concentrations. Copyright 6 1991 by W. B. Saunders Company

T

HE PRODUCTION of ketone bodies (KB) is dependent on the amount of the precursor-nonesterified fatty acids (NEFA)-supplied to the liver and on the intrahepatic metabolic fate of fatty acids.’ Thus, KR production can be regulated at the prehepatic and/or at the hepatic level and it is important during studies of regulation of ketogenesis to delineate the respective roles of these two processes in KB production. Among many metabolic and hormonal factors, insulin and glucagon are considered to play a central role in the regulation of KB production, insulin acting at the prehepatic and hepatic levels and glucagon only on the liver.’ However, the demonstration of a direct hepatic effect of glucagon to stimulate ketogenesis is mainly based on in vitro experiments2-4 and on in vivo studies in dogs.5 The existence of this ketogenic action in humans is debated, since this action was found to be either moderate6 or absent7.’ With respect to insulin, there is no doubt that insulin restrains KE3 production by decreasing lipolysis and NEFA availability.’ However, whether insulin has or does not have an additional action to decrease the hepatic transformation of fatty acids into KB-ketogenesis in the strict sense-remains controversial. In vitro experiments showed’-” or failed to show2,‘2a direct antiketogenic

From IN&ERM U. 197, FacuNC de Medecine Alexis Carrel, Lyon; the Department of Endocrinology, Hopital Edouard Hem’ot, Lyon; the Service de Radioanalyse, Hopital Nemo-Cardiologique, Lyon; and the Institut Pasteur, Lyon, France. Supported by the INSERM and the Hospices Civils de Lyon (Reseau de Recherche Clinique no. 488 004). Address reprint requests to M. Beylot, MD, INSERM U. 197, Faculte de Medecine Alexis Carrel, rue G. Paradin, 69372 Lyon Cidex 08, France. Copyright 0 1991 by U?B. Saunders Company 00260495/91/4011-0005$03.00/0

1138

action. Conflicting results have also been reported during in vivo studies in animals’3,‘4 and humans.‘,’ Miles et al found that insulin did not exert an antiketogenic action at the hepatic level,6 while Keller et al’ described a fatty acidindependent inhibition of KE3production by insulin. The aim of the present study was to determine whether insulin and/or glucagon modify the hepatic transformation of fatty acids into KEI in humans. To address this question, we measured simultaneously NEFA and KEl kinetics in the presence of various concentrations of NEFA, insulin, and glucagon. This allowed us to examine the effect of physiological concentrations of insulin and glucagon on the relationship between NEFA availability and KEl production. METHODS Subjects

Informed consent was obtained from 19 normal subjects and six insulin-dependent diabetic patients after full explanation of the nature, purpose, and possible risks of the study. The normal subjects were 19 men, aged 22 to 38 years (mean, 26) weighing 57 to 76 kg (mean, 66). All were within 10% of their ideal body weight. None took any medication or had a familial history of diabetes melhtus. All had a stable body weight and ate their usual diet before the study. The diabetic patients were five men and one woman, aged 22 to 55 years, weighing 61 to 77 kg, treated for insulin-dependent diabetes for more than 6 years. None had detectable C peptide after intravenous (IV) injection of glucagon; all had normal hepatic, renal, and cardiac functions. All studies were performed in accordance with the principles of the Declaration of Helsinki. Procedures Normal subjects. Three series of tests were performed. In the first, we determined the relationship between NEFA availability and KB production in the presence of basal concentrations of insulin and glucagon. In the second and third studies, we examined

Metabolism, Vol40, No 11 (November), 1991: pp 1138-l 146

1139

INSULIN AND GLUCAGN CONTROL OF KETOGENESIS

the influence of increasing insulin concentrations on the relationship between NEFA and KB flux in the presence of glucagon levels that were decreased (second study, low-glucagon study) or maintained at normal peripheral postabsorptive values (third study, normal-glucagon study). All tests were initiated between 8:OOand 900 AM after a lZ-hour overnight fast. IV catheters were inserted into veins of one forearm for the different infusions (all studies) and for continuous blood glucose analysis by an artificial pancreas (Biostator GCIIS II, Miles Laboratory, Frankfurt, Germany) (second and third studies) and into a vein of the controlateral hand for sampling of arterialized bloodI (all studies). At least 30 minutes of rest was allowed before initiating the different infusions. In the first study (10 subjects), tests were initiated by a continuous infusion of albumin-bound [1-“Clpalmitate (0.015 pmol . kg-’ . min-‘) and of [3,4-“CJacetoacetate (0.05 to 0.06 kmol kg-’ . min-’ after a priming dose of 0.5 to 0.6 pmol kg-‘). After a 90-minute period for isotopic equilibration, blood was sampled for the various determinations. The test was thereafter interrupted in five subjects. It was continued in the other five for an additional period of 90 minutes, along with an IV infusion of a triglyceride emulsion (Intralipid 20% [KabiVitrum, Limoges, France], 0.025 mL .kg-’ . min-‘) in order to raise NEFA levels to moderately increased concentrations. In the second and third studies, tests were initiated by simultaneous infusions of somatostatin (250 Kg. h-l), insulin, [3,4-“C,] acetoacetate, and albumin-bound [1-“Clpalmitate, which were continued during the 500 minutes of the tests. Insulin was infused at increasing rates (see below for the exact infusion rates), with a priming dose for each rate calculated according to Rizza et al.” Insulin infusion periods lasted 120 minutes, except the first one, which lasted 140 minutes. Triglyceride emulsion (Intralipid 20%) was also infused IV (0.016 mL- kg-‘. min-‘) to prevent the insulin-induced decrease of NEFA. Since preliminary studies showed that the first insulin infusion rate resulted in the maintenance of insulinemia close to normal postabsorptive values and did not decrease NEFA levels, Intralipid was infused only during the last three insulin infusion periods of each test. Exogenous glucose was infused when necessary to prevent any decrease in blood glucose. KH,PO, (40 mmol K’/L glucose) was added to prevent hypokalemia and hypophosphatemia. Blood was sampled before initiating the tests and during each insulin infusion period for the various determinations. The remainder of the procedure was dependent on the study performed. During the second protocol (low-glucagon study, five subjects), insulin was infused at progressing rates of 0.05,0.09,0.15, and 0.30 mU . kg-’ . min-‘. During the third study (normal-glucagon study, four subjects), insulin was infused at higher rates (0.09,0.15,0.30, and 0.40 mU . kg-’ min-‘) and glucagon was replaced by a continuous 500-minute infusion (0.8 ng . kg-’ min-I). Diabetic patienrs. Diabetic patients were also studied in the postabsorptive state. Subcutaneous insulin was withdrawn and IV insulin infusion was adjusted to obtain a metabolic status ranging from almost perfect to poor control and stable during the 24 hours preceding the study. The morning of the study, NEFA and KB kinetics were determined by a primed continuous 250-minute infusion of [3,4-“CJacetoacetate (0.06 to 0.133 kmol . kg-’ . min-‘) and a continuous infusion of albumin-bound [l”C]palmitate (0.02 to 0.036 kmol . kg-’ min-I).

Materials [l-‘3C]palmitic acid (99% APE) and [3,4-“CJethylacetoacetate (98% APE) were from CEA (Gif-suryvette, France). Isotopic purity was determined by gas chromatography-mass spectrometry

(GC-MS). [3,4?Jacetoacetate was prepared by hydrolysis of labeled ethylacetoacetate as previously described.16 [1-“Clpalmitate was bound for infusion to human serum albumin (Institut Merieux, Lyon, France). Labeled acetoacetate was diluted with sterile isotonic saline. All tracers were tested for pyrogen before administration and infused through a 0.22~km filter. Aliquots of the tracers were collected at the end of each test to check the concentration and calculate’the actual infusion rate. Somatostatin (Clin-Midy, Montpellier, France) was dissolved with sterile water. Insulin (Actrapid HM Novo, Copenhagen, Denmark) and glucagon (Novo, Copenhagen, Denmark) were diluted with sterile isotonic sodium chloride containing human serum albumin (2%). Analytical Procedures Plasma glucose, glycerol, acetoacetate, and D-P-hydroxybutyrate were determined in neutralized perchloric extracts of plasma by enzymatic methods as previously described.” KB refers to the sum of acetoacetate and D-P-hydroxybutyrate. Plasma triglycerides” and NEFA19 were measured by enzymatic methods, and plasma total (normal subjects) or free (diabetic patients) immunoreactive insulin (IRI),N.Z’glucagon (IRG),” and C peptide= by radioimmunoassay. The isotopic enrichment (MPE) of acetoacetate, D-phydroxybutyrate, and palmitate was determined by capillary GC-MS as previously described. 16~24z The individual concentrations of each plasma NEFA, from lauric acid to linolenic acid, were also determined by GC-MSz4 using heptadecanoic acid as standard. Since in the present study we infused simultaneously “C-labeled palmitate and acetoacetate, the validity of the determinations of KB and fatty acids kinetics requires that there is no significant apparition in KB of “C coming from palmitate during the infusion of tracer amount of [1-‘-‘C]palmitate and no significant incorporation in plasma fatty acids of 13C coming from KB during the infusion of labeled acetoacetate. Therefore, control experiments were performed in humans (n = 3) and rats (n = 4), with KB and fatty acids flux ranging from moderate to high values and infused either with [1-“C] palmitate or with [2-3?Jacetoacetate alone. These experiments showed that, using conventional GC-MS techniques, there was no detectable enrichment in ‘C of KB when plasma palmitate enrichment was 2% to 3%, and no detectable enrichment of palmitate when acetoacetate enrichment was increased to 5% to 8%. Thus, the simultaneous infusion of labeled palmitate and acetoacetate introduces no error in the determination of KB and palmitate enrichment. Calculations For the determination of KB kinetics, we used the traditional approach of first calculating the mean total KB enrichment, since this approach has been validated for steady-state and non-steadystate conditionsI It has been suggestedz6 that the accuracy of the determination of KB kinetics could be dependent, in part, on the ratio of enrichment obtained in acetoacetate and B-hydroxybutyrate (MPE AcAc/MPE DPOHB). During all the studies performed, this ratio was between 1.25 and 3.13 (mean, 2.43), without any difference between the different studies. Therefore, there were no systematic differences between the different tests in the isotopic equilibrium of the two KB that could lead to artifactual modifications of KB flux. In diabetic patients, KB and palmitate kinetics were calculated using steady-state equations. In normal subjects, appearance (Ra) and disappearance (Rd) rates were calculated using either steady-state equations, when neither concentration nor enrichment variations exceeded lo%, or the non-steady-state equations of Steele” adapted to stable isotope.‘6.Z’The effective volume of distribution was assumed to be 0.20 L. kg-’ for KB**and the plasma volume 0.05 L kg-’ for palmitate.29sMTotal NEFA

1140

BEYLOT

kinetics were calculated from palmitate kinetics and the total NEFA concentration to palmitate concentration ratio, assuming that palmitate is representative of all NEFA.37 For the calculation of the conversion of NEFA into KB, we assumed that all KB arc derived from plasma NEFA. Knowing the concentration of the different plasma fatty acids, one can calculate for each sample the mean carbon chain length of fatty acids. It is next possible to calculate from this mean chain length and from the Rd of NEFA the production rate of KB that would have been obtained if all plasma NEFA had been converted into KB (maximal KB Ra = [NEFA Rd] x [mean carbon chain length]/4). The proportion of NEFA converted into KB is then the actual KB Ra to maximal KB Ra ratio.24 Results are shown as individual values or mean ? SEM. Within-test comparisons were performed by two-way ANOVA, and next the Newman-Keuls procedure to locate the specific differences. Between-test comparisons were made by one-way ANOVA and by ANCOVA. Correlations were established by linear regression analysis (least-square method). RESULTS

Normal Subjects First study. Table 1 shows the individual values for palmitate, total NEFA and KB concentrations, and NEFA and KB kinetics determined in the postabsorptive state (10 subjects) and during the last 30 minutes of the Intralipid infusion (five subjects). In these five subjects, Intralipid infusion induced no modifications of plasma insulin (7.8 c 0.8 v 7.2 2 0.6 mu/L), C peptide (1.71 ? 0.22 v 1.65 ? 0.13 kg/L), or glucagon levels (98 2 23 v 83 +- 21 ng/L). As expected, NEFA concentrations, as well as Ra and Rd, increased (P < .05). KE3 concentrations and KB Ra also increased (P < .05). However, these increases were more important than the increases in NEFA levels and NEFA Ra. In addition, the percentage of plasma NEFA used for KB production (calculated from NEFA Rd and KB Ra) was increased (P < .0.5). Moreover, there was a linear relationship between the percentage of plasma NEFA used

Table 1. Concentrations

ET AL

for ketogenesis and NEFA concentrations considering either only the values obtained in the five subjects investigated before and during Intralipid infusion (r = .658, P < .05, y = 0.0318x + 1.192), or the results obtained in all subjects (r = ,788, P < .OOl. y = 0.033x + 0.215). Thus, in the absence of any detectable variation of endocrine pancreas secretion. a moderate increase in NEFA concentration appeared to increase KB production not only by increasing fatty acids availability, but also by diverting, in part, fatty acid metabolism toward ketogenesis. Second study (low glucagon levels). Endocrine pancreas secretion was suppressed by somatostatin infusion, since C peptide and glucagon decreased to a level near or under the limit of detection (C peptide 0.5 kg. I-‘, glucagon 15 ng . 1-l) (Fig 1). Insulin level in the postabsorptive state was 11.5 t 1.7 mu. L-‘. Exogenous insulin infusion first resulted in the maintenance of peripheral insulinemia near to this postabsorptive value (9.5 + 0.4 mu. 1-l). Thereafter. insulin level was increased progressively by the exogenous infusion to 13.1 ? 0.7, 16.1 2 2.5, and 28.0 f 3.8 mU . L-’ (Fig 1). Postabsorptive glycemia was 4.88 c 0.23 mmol/L. Plasma glucose was maintained throughout the test between 4.45 and 5.10 mmol . L-’ (Fig 2) by the infusion of small amounts of exogenous glucose. NEFA (Fig 2) and glycerol concentrations (data not shown) during the first insulin infusion period were comparable to the values observed in the postabsorptive state. Intralipid infusion resulted in a sharp increase of glycerol from 42 + 5 kmol . 1-l to approximately 200 kmol L-‘. NEFA increased progressively throughout the test from a value of 312 t 87 kmol . 1-l (end of first insulin infusion period) to 419 * 68 (NS), 466 2 29 (P < .05), and 502 2 64 kmol ‘1.’ (P < .05) (Fig 2). Thus, although there was a significant increase above baseline values, Intralipid infusion was successful in maintaining NEFA to physiological postabsorptive values. Plasma palmitate concentrations at the end of the first

and Kinetics of NEFA and KS in Normal Subjects Studied in the Postabsorptive State’-‘O and During a go-Minute lntralipid lnfusior?”

Subject NO.

Palmitate (km01 L-‘)

NEFA (pm01L-‘)

KB

(~moi. L ‘)

(pmol

NEFA Rd kg-‘. min-‘)

(pmol

KB Ra kg ’ min ‘)

NEFA - Ka I”4

213 + 12

a3 2 5

3.40 + 0.10

1.27 t- 0.03

9.4 -c 0.8

2

150 + 4

432 2 16

172 + 14

6.02 + 0.36

3.56 2 0.10

15.4 k 1.6

3

101 26

330 k 10

94+

4.50 rt 0.20

1.49 * 0.09

a.8 *

4

91 +5

265 + la

116?6

3.80 + 0.15

1.17 2 0.02

7.8 -+ 0.2

1

71

k3

10

0.2

5

170 + 6

428 + 17

145 + 6

5.13 + 0.17

2.60 f. 0.09

13.2 + 0.2

6

143 t 3

481 + lo

la0

4.44 2 0.13

2.80 -c 0.05

14.9 + 0.7 22.8 * 1.8

+ IL

558 r 22

228 + 26

4.67 + 0.08

4.10 2 0.45

96 2 4

395 * 15

110t4

3.55 * 0.07

2.51 t 0.10

17.3 t 1.4

+ IL

128 2 3

654 * 31

379 2 28

4.96 t 0.35

6.37 + 0.35

30.8 + 0.8

110*3

415 If: 17

74 + 5

+ IL

168 + 11

901 2 35

493 + 29

185 + a

565 + 26

125 + 2

4.14 f 0.22

2.33 2 0.08

13.3 _t 1.5

+ IL

208 2 7

739 + 7

446222

5.83 5 0.21

6.70 + 0.49

24.9 2 0.4

137 2 5

465 r 26

149 + a

5.08 * 0.18

2.75 f 0.08

12.2 + 1.0

+ IL

197 -c 5

723 2 59

707 -t 74

6.19 + 0.36

a.70 k 0.98

32.1 2 1.5

7 a 9 10

NOTE.

154k-8

+ 6

Each value is the mean

of the lntralipid Abbreviations:

+ SEM of four values

determined

5.30 + 0.23 10.03

over a 30-minute

+ 0.30

period

in the postabsorptive

I .a0 it 0.04

9.0 z? 0.5

9.04 -e 0.20

21.6 r 0.6

state and during

the last 30 minutes

infusion. Rd, disappearance

rate; Ra, appearance

rate; NEFA

-

KB, proportion

of plasma

NEFA used for KB production;

IL, Intralipid.

INSULIN AND GLUCAGN CONTROL OF KETOGENESIS

1141

6.53 f 0.24 pmol *kg-’ . min-’ during the second, third, and fourth insulin infusion periods) and Rd (5.55 2 0.71, 7.31 f 0.60, and 6.50 f 0.52 umol . kg-’ . min-‘) increased progressively and were linearly related to NEFA concentrations (r = .861, P < .OOl, for NEFA Rd and levels). RR Ra also increased (2.56 + 0.68, 3.50 ? 0.71, and 4.38 + 2.44 pmol . kg-’ . min-‘) and were more than doubled at the end of the test. The calculated percentage of NEFA disappearing from plasma used for ketogenesis at the end of the first insulin infusion period was 8.1 + 1.2, a value slightly lower than that found in postabsorptive normal subjects (Table 1). Thereafter, this percentage increased (Fig 4) at the end of second, third, and fourth insulin infusion periods, to 10.9 f 2.4, 11.9 2 3.1, and 14.2 f 6.3, respectively (P < .05 v the initial value). As in the first study, this percentage appeared to be linearly related to NEFA levels (P < .05) (Fig 5) and the slope of the relationship (y = 0.0319x + 2.580) was not modified by the increasing insulin concentrations. Thus, we found no evidence for a restraining effect of a mild increase of insulin level on hepatic ketogenesis when plasma NEFA were maintained close to normal postabsorptive values. Third study (normal glucagon levels). C peptide levels were decreased by somatostatin infusion as in the first study, while peripheral glucagonemia was maintained to postabsorptive values by the infusion of exogenous gluca-

25w ;

20

E ;

15_ lo-

t m x e rr 5 :

1,5_ lQ5-

60 c; p

40

is 20 i

6

l4b TIME

A0

3180

500

minutes

Fig 1. Evolution of plasma immunoreactive insulin (BE), C peptide, and glucrgon (IRG) during the low-glucagon study (O-O) and the normal-glucagon study (0-O). Somatostatin and insulin, with or without glucagon, were infused from 0 to 500 minutes. Results are shown as mean 2 SEM. l/J < .05, l*P < .Ol, l*+P < ,001, means significantly different from the initial (-5 and 0 minutes) values of the low-glucagon study; $ P < 0.05, $$ P < 0.01, $$$ P < .OOl, means significantly different from the initial values of the normal-glucagon study; l P < .05,V < .Ol, significantly different of the corresponding values of the normal glucagon study.

insulin infusion period were 75 f 10 pmol . L-’ (26% of total NEFA) and remained stable (79 f 14, 75 -C 6, and 83 f 11 pmol . I-‘), whereas its contribution to total NEFA decreased slightly (20%, NS; 16.6%, P < .05; and 16.5%, P < .05). KB were also first maintained close to postabsorptive values (107 + 26 v 109 ? 26 umol *L-l). Thereafter, KB increased progressively (174 f 57, ns; 274 ? 110, P < .05; and 270 2 111 pmol . L-‘, P < .05, at the end of the second, third, and fourth insulin infusion periods, respectively) and were more than doubled at the end of the test (Fig 2). Figure 3 shows the Ra of NEFA and KB during the test. At the end of the first insulin infusion step, both Ra were within normal postabsorptive values (RR, 1.44 2 0.15 t.r.rnol. kg . min-’ ; NEFA, 4.70 ? 0.71 u.mol * kg-’ . min-‘). Thereafter, NEFA Ra (5.72 f 0.62, 6.87 + 0.50, and

600, z 400X iz w z

200o400_ 300-

5

200..

: lOO_ r

I

0

140 TIME

260

380

500

minutes

Fig 2. Glucose, NEFA, and KB evolution during the low-glucagon (O-O) and the normal-glucagon (O+) studies. lP < .05, different from the initial values (-5 and 0 minutes) of the low-glucagon study; S P < .05, different from the initial value of the normal-glucagon study; lP < .05, different from the corresponding value of the low-glucagon study.

BEYLOT ET AL

140

TIME

fir

500

380

260

140

minutes

TIME

360

minutes

Fig 3. NEFAand KB Rs during (A) the low-glucagon study and (B) the normal-glucagon study. lP < .05, different from the values of the I 10 to MO-minute period (low-glucagon study); S P < .05, $$ P < .oOl, different from the values of the llO- to MO-minute period (normalglucagon study).

(Fig 1). Postabsorptive insulinemia was 7.3 k 0.4 mU . L-‘. During the test, insulinemia was first maintained by the exogenous infusion to postabsorptive value (7.7 f 0.3 mu. L-l) and next increased progressively to 12.2 ? 0.5, 19.1 f 0.2, and 29.6 + 0.5 mU . L-’ (Fig 1). The replacement of glucagon secretion by exogenous glucagon infusion resulted in an increase of glycemia at the end of the first insulin infusion period and during the second one (Fig 2). Glucose returned to basal value during the third period and was maintained close to this level until the end of the gon

1

a

25

20

a

!

\

a

I*

il

I

32 28 24

*

z

0

15

ta G? 9) s

IO

5

0

l I

test. Exogenous glucose was infused only during these last two periods. The changes of NEFA and glycerol concentrations were comparable to those observed during the first study. During the first insulin infusion period, both concentrations were maintained to the initial postabsorptive values. Thereafter, Intralipid infusion resulted in a sharp increase of glycerol (data not shown) and in a progressive and moderate increase of NEFA between 370 and 470 kmol . L-’ (Fig 2). Plasma palmitate concentrations were not significantly modified (62 + 17, 62 k 14, 76 +- 13, and 77 + 14 pmol . L-‘) and its contribution to total NEFA decreased slightly (21%, 17.6%, 17.9%, and 18%) as during the low-glucagon study. KB were slightly increased

20 16 12 a 4

km

;

P

m

400

NEFA

600 ~mol.l-’

’ 800

ldoo

IE

Fig 4. Percentage of NEFA flux used for KB production during the four periods (I to IV) of the low-ghrcagon (m) and the normalglucagon (0) studies. lP < 55, different from the first value (I) of the corresponding study; ‘f < .05, different from the correeponding value of the low-glucagon study.

Fig 5. Relationship between plasma NEFA concentrations and the calculated percentage of free fatty acids flux used for ketogenesls in the presence of unchanging (C), decreasing (Db), or increasfng (0, low-gfuugon study; H, normal-glurgon study) insulin Ieveh. The C and Db mlationshlp were calculated from the lndlvldwl date of Tables 1 and 2, respectively.

INSULIN AND GLUCAGN CONTROL OF KETOGENESIS

1143

above postabsorptive values (130 + 25 *mol. L-’ v 118 f 21 pmol . L-’ , P > .lO) during the first period of the test, and increased progressively (156 + 38, 259 f 60, and 285 -C 90 pmol . L-‘) at the end of the second, third, and fourth periods (Fig 2). The kinetics of NEFA and KB are shown in Fig 3. Again KB and NEFA Ra were within normal postabsorptive values at the end of the first period. However, KB Ra (2.60 + 0.35 pmol . kg-’ . min-‘) was slightly higher (P > .lO) than during the corresponding period of the low-glucagon study, whereas NEFA Ra (4.26 c 0.12 km01 . kg-’ . mm’) was comparable. The percentage of NEFA flux used for KB production was higher in the presence of glucagon substitution that in its absence (15.70 -C 2.83 v 8.05 + 1.19, P < .05). During Intralipid infusion, NEFA Ra (5.35 +- 0.50, 5.76 ? 0.92, and 6.01 + 0.47 pmol . kg-’ . min-‘) and Rd (5.31 5 0.52, 5.65 2 0.88, and 6.01 & 0.47 kmol . kg-’ . min-‘) increased progressively and were always linearly related to NEFA levels (r = .866, P < .OOl for NEFA Rd and concentrations). KB Ra also increased progressively (3.49 f 0.71, 4.90 f 0.77, and 6.01 t 0.71 pmol . kg-’ * min-‘). Again, the increase in KB production was more important than the increase in NEFA flux. Despite the progressive increase of insulinemia, the percentage of NEFA flux used for ketogenesis increased to 20.4 f 4.9, 22.0 2 4.2, and 25.2 + 3.9 (P < 0.05 v initial value) during the second, third, and fourth periods of the test (Fig 4) and was always linearly related (P < .05, y = 0.0518x + 0.296) to NEFA concentration (Fig 5). Thus, we again found in the presence of glucagon substitution no evidence for a restraining effect of a mild hyperinsulinemia on hepatic ketogenesis. In addition, we found that, within the range of NEFA and insulin concentrations obtained in this study, the percentage of NEFAtlux used for ketogenesis was always higher (P < .05) in the presence of glucagon substitution than in its absence (Fig 5). Diabetic Patients

The individual values obtained in the six diabetic patients are shown in Table 2. As expected, low insulin levels were associated with higher glucose concentrations and with increased NEFA and RR concentrations and Ra. The proportion of NEFA used for ketogenesis also increased. However, this proportion was only weakly related to insulin level (r = -.663, P > .lO) and was again linearly related to NEFA concentrations (r = .886, y = 0.0226x + 7.300, P < .05). Moreover, it is noteworthy that the regression

line between this percentage and NEFA concentrations found in diabetic subjects was close to those observed in normal subjects (Fig 5), despite the fact that in diabetic patients the increase in NEFA was associated with a decrease in insulin, whereas in normal subjects this increase was associated with no variation (first study) or an increase (second and third studies) of insulinemia.

DISCUSSION The present studies examined the effects of physiological concentrations of insulin and glucagon on hepatic ketogenesis. When NEFA availability was maintained by Intralipid infusion, we found no restraining action of a mild hyperinsulinemia on the transformation of fatty acids into RR, whereas glucagon appeared to have a direct hepatic ketogenit action. Since liver is exposed mainly to portal venous blood, the validity of our results requires that we were effectively able to control and manipulate portal insulin and glucagon concentrations. Actually, somatostatin infusion in normal subjects was effective to decrease C peptide andfor the low-glucagon study-glucagon levels close to the limit of detection, indicating that endogenous insulin and glucagon secretions were almost, if not totally, suppressed. In these conditions, portal and peripheral insulin and glucagon concentrations should be close. Last, we have previously showr? that the glucagon infusion rate used in this study was effective in increasing endogenous glucose production above the values observed without glucagon substitution, as evidenced in the present report by the moderate increase of glycemia. We have also shown that the progressive increase of insulinemia obtained by the stepwise insulin infusion was able to induce a progressive decrease of endogenous glucose production in the presence and in the absence of glucagon substitution.3* These actions on glucose metabolism provide good evidence that the hormonal environment of liver was modified as expected throughout the different periods of the studies. During the experiments with somatostatin infusion, we did not replace growth hormone (GH), and the possible impact of this suppression of basal GH secretion has to be considered, since GH can exert in vivo lipolytic and ketogenie actions.33.34However, in those studies, the ketogenic effect of GH was, with the exception of insulin-deficient subjects, explained by an enhancement of lipolysis and an increase in NEFA supply. There was no argument for a direct ketogenic action. Moreover, those studies examined the actions of GH raised several-fold over basal values and

Table 2. Individual Hormonal and Metabolic Values of the Diabetic Patients Patient

IRG

NO.

(ng. L-9

1

FIRI (mu

L-1)

54

36.0

PG (mmol/L)

KB (pmd/L)

KB Rt (pmol

kg-’

NEFA Rt

NEFA min-‘1

bmol/L)

(kmol

kg-’

min-‘)

NEFA + KB 1%)

3.20

106

2.10

322

3.69

13.6 13.2

2

4.8

18.5

5.50

108

2.15

292

3.95

3

61

14.0

4.34

73

2.95

361

3.87

18.6

4

96

11.0

10.8

390

7.10

719

9.03

21.3

5

115

4.5

16.27

1,416

13.74

990

9.23

35.0

6

4.8

3.5

9.10

1,118

7.40

909

7.73

22.9

Abbreviations: FIRI, free immunoreactive insulin: PG. plasma glucose: Rt, turnover rate.

1144

the role of basal GH secretion remains to be established. In any case, this would not hamper the comparison between the low-glucagon and the normal-glucagon studies. The interpretation of our results is also dependent on the validity of the tracer method used for determining NEFA and KB kinetics. We infused [l-‘3C]palmitate, since it is considered to be representative of all plasma NEFA.” During Intralipid infusion, the contribution of palmitate to total NEFA decreases, since the proportion of palmitate relative to that of unsaturated fatty acids (oleic, linoleic, and linolenic acids) is lower in this triglyceride emulsion than in normal plasma. However, this decrease was quite moderate (see Results), and palmitate appeared to always represent an important percentage of total NEFA. For the determination of KB kinetics, we used the traditional approach of infusing one labeled KB (acetoacetate) and calculating from the measured individual enrichment of acetoacetate and D-B-hydroxybutyrate, the “mean total KB enrichment”” (or “mean total KB-specific activity”” for radioactive tracers). This last value is then used for the calculation of KB kinetics.‘“.28 This approach has been criticized,26.35and it has been proposed that the simultaneous infusion of differently labeled acetoacetate and D-B-hydroxybutyrate and the use of a two-pool model could provide more accurate results. However, contrary to Bailey et alZ6in their study in dogs, Bougneres and Ferre36 found a good agreement between single-isotope calculations and the two-pool model in human studies. Moreover, the single-isotope method has been validated in dogs against net hepatic production3’ Last, the two-pool model is limited to steady-state conditions, whereas the singleisotope method has been validated for both steady-state and non-steady-state conditions, using radioactive or stable isotope-labeled tracers.‘6,28*37.38 Our results provide no evidence for a direct hepatic antiketogenic action of insulin, in the presence or in the absence of glucagon deficiency. In both studies, despite the increase of insulin levels, both total KB Ra and the calculated percentage of NEFA used for KB production increased instead of decreasing when any decrease of NEFA was prevented. Actually, Intralipid infusion resulted in a moderate increase of NEFA levels during both experiments. There is a great amount of in vitro evidence that increasing the concentration of fatty acids promotes the orientation of hepatic fatty acid metabolism toward oxidation and ketogenesis: the increase of intracellular concentration of long-chain acyl COASTdecreases acetyl CoA carboxylase activitya and malonyl CoA concentration,J and inhibits malonyl CoA binding to carnitine palmitoyl transferase I.“’ These two processes, reduced malonyl CoA content and binding to carnitine palmitoyl transferase I, promote fatty acid oxidation. Thus, one could argue that since NEFA concentrations increased in these experiments, the increase in the fraction of NEFA used for ketogenesis could have been greater if insulinemia had been kept to the same level instead of increasing. However, when we examined the relationship between NEFA levels and the percentage of NEFA used for KB production obtained in the presence of

BEYLOT ET AL

increasing insulin concentration (low-glucagon and normalglucagon studies) with that observed in the first study (no variations of insulin), the slopes of the regression lines appeared nearly identical (Fig 5). Last, when we examined this relationship in the presence of decreasing insulin level, as obtained in the diabetic patients study (Table l), this slope was, if anything, rather decreased than increased (Fig 5). Thus, since the slopes of the relationships between NEFA level and the proportion of NEFA flux directed toward ketogenesis appear nearly identical in the presence of increasing, constant, or decreasing insulin levels, we have no evidence for a direct hepatic antiketogenic action of insulin. These results in humans agree with those of Miles et al,” who found that KB production following elevation of plasma NEFA was identical in the presence or absence of insulin deficiency. They contrast with the recent report of Keller et al,’ showing a restraining effect of high physiological insulin concentration on hepatic ketogenesis. Differences in the insulin levels obtained could explain these discrepancies. We and Miles et al6 have examined the effects of low physiological insulin levels (10 to 30 mU L-‘), whereas Keller et al’ investigated the action of high insulin concentration (100 to 120 mU L -I). Taken together, these studies suggest that, under most physiological conditions, insulin inhibits ketogenesis mainly by decreasing lipolysis and NEFA availability, but that, in pathological situations associated with large hyperinsulinemia-such as severe sepsis”’ or massive obesity43-it may have an additional direct hepatic effect. It should also be stressed that we studied the effect of insulin during a relatively short period (6 hours) and it remains possible that a direct hepatic action would have appeared if we had infused insulin over a longer period. In vitro studies have shown that insulin is able to acutely stimulate acetyl CoA carboxylase activity and increase intracellular malonyl CoA concentration.J4.45 However, insulin acts also to modify the activity of carnitine-palmitoyl transferase I and its affinity for malonyl CoA: starvation and onset of diabetes in rats are associated with an increase in the activity of this enzyme and a decrease in its sensitivity to inhibition by malonyl COA.~-~~These modifications are reversed slowly; for example, full restoration of the sensitivity of the enzyme to malonyl CoA by treatment of streptozotocin-induced diabetes with insulin was not observed before 24 hours of treatment and this period was not sufficient to decrease the activity of the enzyme.J7 Thus, it remains possible that our study was too short to allow insulin to exert all its actions. Our results strongly support a role for physiological concentrations of glucagon in stimulating hepatic ketogenesis since, within the range of insulin and NEFA attained, the percentage of NEFA used for KB production was always lower in presence of glucagon deficiency than during glucagon substitution. These observations are consistent with in vitro studies showing a direct enhancement of hepatic ketogenesis by glucagor? and with in vivo studies in dogs5 They are at variance with the report of Miles et al6

1145

INSULIN AND GLUCAGN CONTROL OF KETOGENESIS

who found a small ketogenic action of glucagon in man only during insulin deficiency, and they disagree with those of Sonnenberg et al’ and Keller et al,’ who found no in vivo ketogenic action of glucagon in the presence’ or absence of insulin deficiency.’ However, both Miles et al6 and Keller et al’ studied the effect of glucagon in the presence of NEFA levels increased between 1.5 and 2.3 mmol . L -‘. It is quite possible that these high NEFA levels resulted already in a near maximal stimulation of ketogenesis and left no place for an additional effect of glucagon. This possibility is supported by the in vitro demonstration4 that the ketogenic action of glucagon is less apparent when fatty acids levels are high. Actually, high fatty acids levels per se result in modifications of hepatic fatty acids metabolism similar to those induced by glucagon in the presence of low concentrations of fatty acids.4 Sonnenberg et al* found no ketogenic action of glucagon in the presence of basal NEFA levels.

However, NEFA were approximately 50% lower during their tests with glucagon replacement than during glucagon deficiency (380 v 685 kmol . L-‘). The decrease of ketogenesis induced by such a difference in NEFA levels may have been sufficient to mask a ketogenic action of glucagon. In conclusion, the present results confirm that NEFA level per se is an important regulating factor of hepatic ketogenesis. Physiological glucagon levels appear to have a stimulating action on hepatic ketogenesis in the presence of basal NEFA concentrations, whereas low to physiological insulin levels have no direct hepatic antiketogenic action at least within a period of a few hours. ACKNOWLEDGMENT We wish to thank N. Bureau for her help in the preparation of i&sates, M. Odeon and F. Vieilly for their technical assistance, and S. Terfous for her secretarial work.

REFERENCES

1. Johnston DG, Alberti KGMM: Hormonal control of ketone body metabolism in the normal and diabetic state. Clin Endocrinol Metab 11:329-361,1982 2. Witters LA, Trasko CS: Regulation of hepatic free fatty acid metabolism by glucagon and insulin. Am J Physiol 237:E23-E29, 1979 3. McGarry JD, Wright P, Foster DW: Hormonal control of ketogenesis. Rapid activation of hepatic ketogenic capacity in fed rats by anti-insulin serum and glucagon. J Clin Invest 55:1202-1209, 1975 4. McGarty JD, Foster DW: Effect of exogenous fattyconcentration on glucagon induced changes in hepatic fatty acid metabolism. Diabetes 29:236-240,198O 5. Keller U, Chiasson J, Liljenquist A, et al: The role of insulin, glucagon and free fatty acids in the regulation of ketogenesis in conscious dogs. Diabetes 26:1040-1051,1977 6. Miles J, Haymond MW, Nissen SL, et al: Effect of free fatty acid availability, glucagon excess and insulin deficiency on ketone body production in post-absorptive man. J Clin Invest 71:15541561,1983 7. Keller U, Gerber PPG, Stauffacher W: Fatty-acid independent inhibition of hepatic ketone body production by insulin in humans. Am J Physiol254:E694-E699,1988 8. Sonnenberg GE, Stauffacher W, Keller U: Failure of glucagon to stimulate ketone body production during acute insulin deficiency or insulin replacement in man. Diabetologia 23:94-R& 1982 9. Agius L, Chowdhuty H, Davis SN, et al: Regulation of ketogenesis, gluconeogenesis and glycogen synthesis by insulin and proinsulin by rat monolayer cultures. Diabetes 35:1286-1293, 1986 10. Bates EJ, Topping DL, Sooranna SP, et al: Acute effects of insulin on glycerolphosphate acyl transferase activity ketogenesis and serum free fatty acids concentration in perfused rat liver. FEBS Lett 84:225-228,1987 11. Laker ME, Hayes PA: Investigation onto the direct effects of insulin on hepatic ketogenesis, lipoprotein secretion and pyruvate deshydrogenase activity. Biochim Biophys Acta 795:427-430, 1984 12. Boyd ME, Albright EB, Foster DW, et al: In vitro reversal of the fasting state of liver metabolism in the rat. Reevaluation of the role of insulin and glucagon. J Clin Invest 68:142-152,198l 13. Bates MN, Linn LC, Huen AHJ: Effects of oleic acid infusion on plasma free fatty acids and blood ketone bodies in the fasting rat. Metabolism 25:361-373,1976

14. Biederdorf FA, Charnick SS, Score RG: Effect of insulin and acute diabetes on plasma FFA and ketone bodies in the fasting rat. J Clin Invest 49:1685-1693, 1970 15. Rizza RA, Mandarin0 LJ, Gerich JE: Dose-response characteristics for effects of insulin on production and utilization of glucose in man. Am J Physiol240:E630-E635,1981 16. Beylot M, Beaufrere B, Normand S, et al: Determination of human ketone body kinetics using stable isotope labelled tracers. Diabetologia 29:90-96,1986 17. Beylot M, Riou JP, Bienvenu F, et al: Increased ketonaemia in hyperthyroidism. Evidence for a beta-adrenergic mechanism. Diabetologia 19:505-510,198O 18. Wahlefed AW: Triglycerides: Determination after enzymatic hydrolysis, in Bergmeyer HU (ed): Methods of Enzymatic Analysis. San Diego, CA, Academic, 1974, pp 1831-1836 19. Okabe H, Osi Y, Nagashima K, et al: Enzymic determination of free fatty acids in serum. Clin Chem 13:476-480, 1973 20. Hales CM, Randle PJ: Immunoassay of insulin with insulin antibody precipitate. Biochem J 88:137-148,1963 21. Nakagawa S, Nakayama H, Sasaki T, et al: Simple method for the determination of serum free insulin levels in insulin-treated patients. Diabetes 22:590-600,1973 22. Harris J, Faloona GR, Unger RH: Glucagon, in Jaffe BM, Behrman HR (eds): Methods of Hormone Radioimmunoassay. San Diego, CA, Academic, 1979, pp 643-671 23. Horwitz D, Starr J, Mako M, et al: Proinsulin, insulin and C peptide concentrations in human portal and peripheral blood. J Clin Invest 55:1278-1283,1975 24. Beylot M, Beaufrere B, Riou JP, et al: Effect of epinephrine on the relationship between non esterified fatty acid availability and ketone body production in post-absorptive man: Evidence for an hepatic anti-ketogenic effect of epinephrine. J Clin Endocrinol Metab 65:914-921, 1987 25. Beylot M, Guiraud M, Grau G, et al: Regulation of ketone body flux in septic patients. Am J Physiol257:E665-E674,1989 26. Bailey JM, Haymond MW, Miles JM: Validation of two pool model for in vivo ketone body kinetics. Am J Physiol Endocrinol Metab 258:E850-E855,1990 27. Steele R: Influence of glucose loading and of injected insulin on hepatic glucose output. Ann NY Acad Sci 82:420-426,1959 28. Keller U, Sonnenberg GF, Stauffacher W: Validation of a tracer technique to determine non steady-state ketone body turnover rates in man. Am J Physiol240:E253,1981

1146

29. Armstrong DT, Steele R, Altzuler N, et al: Regulation of plasma free fatty acid turnover. Am J Physiol201:9-14,196l 30. Galster AO, Clutter WE, Cryer PE, et al: Epinephrine plasma threshold for lipolytic effects in man. Measurement of fatty acids turnover with [1-“C]palmitic acid. J Clin Invest 67:1729,1981 31. Hagenfeldt L, Wahren J, Pernon B, et al: Uptake of individual free fatty acids by skeletal muscle and liver in man. J Clin Invest 51:2324-2330,1972 32. Chambrier C, Picard S, Vidal H, et al: Interactions of glucagon and free fatty acids with insulin in control of glucose metabolism. Metabolism 39:976-984,199O 33. Keller U, Schnell H, Girard J, et al: Effect of physiological elevation of plasma growth hormone levels on ketone body kinetics and lipolysis in normal and acutely insulin-deficient men. Diabetologia 26:103-108,1984 34. Moller N, Butler PC, Antsiferov MA, et al: Effects of growth hormone on insulin sensitivity and forearm metabolism in man. Diabetologia 32: 105-l lo,1989 35. Cobelli C, Nosadini R, Toffolo G, et al: Model of the kinetics of ketone bodies in humans. Am J Physiol243:R7-R17,1982 36. Bougneres PF, Ferre P: Study of ketone body kinetics in children by a combined infusion of “C and ‘H, tracers. Am J Physiol243:E496-E502,1987 37. Keller U, Cherrington AO, Liljenquist JE: Ketone body turnover and rat hepatic ketone production in fasted and diabetic dogs. Am J Physiol235:E238-E247,1978 38. Miles JM, Haymond MW, Rizza RA, et al: Determination of 14Cradioactivity in ketone bodies: A new, simplified method and its validation. J Lipid Res 21:646-650,198O

BEYLOT ET AL

39. Bortz WM, Lynen F: Elevation of long chain acyl CoA derivatives in livers of fasted rats. Biochem Z 329:77-82, 1963 40. Bortz WM, Lynen F: The inhibition of acetyl CoA carboxylase by long chain acylCoA derivatives. Biochem Z 337:505-509, 1963 41. Mills SE, Foster DW, McGarry JD: Interaction of malonyl CoA and related compounds with mitochondria from different rat tissues. Biochem J 214:83-90,1983 42. Stoner HB: Metabolism after trauma and sepsis. Circ Shock 19:75-87,1986 43. Nosadini R, Avogaro A, Trevisan R, et al: Acetoacetate and 3-hydroxybutyrate kinetics in obese and insulin-dependent diabetic humans. Am J Physiol248:R611-R620,1985 44. Beynen DC, Vaartjes J, Geelen MJH: Opposite effects of insulin and glucagon in acute hormonal control of hepatic lipogenesis. Diabetes 28:828-835,1979 45. Witters LA, Watts TD, Daniuls DL. et al: Insulin stimulates the dephosphorylation and activation of acetyl CoA carboxylase. Proc Nat1 Acad Sci USA 85:5473-5477,1988 46. Cook GA, Gamble MS: Regulation of carnitine palmitoyltransferase by insulin results in decreased activity and decreased apparent Ki for malonyl CoA. J Bio Chem 262:2050-2055,1987 47. Gamble MS, Cook GA: Alteration of the apparent Ki of carnitine palmitoyl transferase for malonyl CoA by the diabetic state and reversal by insulin. J Biol Chem 260:9516-9519, 1985 48. Grantham BD, Zammit VA: Role of carnitine palmitoyl transferase I in the regulation of hepatic ketogenesis during the onset and reversal of chronic diabetes. Biochem J 249:409-414, 1988