In Vivo Changes in Central and Peripheral InsulinSensitivity in a Large Animal Model of Obesity
Clare L. Adam, Patricia A. Findlay, Raymond P. Aitken, John S. Milne,and Jacqueline M. Wallace
Obesity and Metabolic Health Theme, Rowett Institute of Nutrition and Health, University of Aberdeen,Bucksburn, Aberdeen AB21 9SB, Scotland, United Kingdom
Obesity disrupts homeostatic energy balance circuits leading to insulin resistance. Here we exam-ined in vivo peripheral and central insulin sensitivity, and whether central insensitivity in terms ofthe voluntary food intake (VFI) response occurs within the hypothalamus or at blood-brain transferlevel, during obesity and after subsequent weight loss. Sheep with intracerebroventricular (icv)cannulae were fed complete diet for 40 wk ad libitum (obese group) or at control level (controls).Thereafter, obese sheep were food restricted (slimmers) and controls fed ad libitum (fatteners) for16 wk. Dual-energy x-ray absorptiometry (DEXA) measured total body fat, insulin analyses in bloodand cerebrospinal fluid (CSF) assessed blood-brain transfer, iv glucose tolerance test (GTT) andinsulin tolerance test (ITT) measured peripheral insulin sensitivity, and VFI responses to icv insulinassessed intrahypothalamic sensitivity. Insulinemia was higher in obese than controls; plasma in-sulin correlated with DEXA body fat and CSF insulin. Insulinemia was higher in fatteners thanslimmers but ratio of CSF to plasma insulin correlated only in fatteners. Plasma glucose baseline andarea under the curve were higher during GTT and ITT in obese than controls and during ITT infatteners than slimmers. GTT and ITT glucose area under the curve correlated with DEXA body fat.VFI decreased after icv insulin, with response magnitude correlating negatively with DEXA bodyfat. Overall, insulin resistance developed first in the periphery and then within the brain, thereaftercorrelating with adiposity; central resistance in terms of VFI response resulted from intrahypotha-lamic insensitivity rather than impaired blood-brain transfer; modest weight loss improved pe-ripheral but not central insulin sensitivity and induced central hypoinsulinemia. (Endocrinology153: 3147–3157, 2012)
Circulating insulin and leptin act within the brain toregulate body weight and energy balance in mamma-
lian species, including humans. In view of the obesity pan-demic, it is important to elucidate how these homeostaticcircuits are disrupted and how central insulin and leptinresistance develops (1). Normally these hormones providepositive nutritional feedback to the hypothalamus wherethey activate anorexigenic and inhibit orexigenic path-ways, thereby maintaining energy balance. However, de-spite high circulating concentrations in obesity, they fail toenact appropriate responses, and there is no spontaneousloss of appetite or body weight, reflecting central insulinand leptin resistance (1). The site of resistance may occur
at or downstream of hypothalamic receptors, but it mayalso result from decreased transport into the brain. Usingour large-animal model of obesity, we have recently dem-onstrated in vivo that central resistance to raised circulat-ing leptin concentrations is attributable to impaired leptintransport into the brain rather than insensitivity within thehypothalamus (2). In this study, we use the same approachto investigate the basis for central insulin resistance.
Systemic insulin crosses the blood-brain barrier via asaturable transport mechanism to exert its effects withinthe central nervous system, and most, if not all, of theinsulin in the brain is of peripheral origin (3, 4). The rel-ative concentrations of insulin in peripheral circulation
Abbreviations: AUC, Area under the response curve above baseline; CSF, cerebrospinalfluid; DEXA, dual-energy x-ray absorptiometry; GTT, glucose tolerance test; icv, intrace-rebroventricular; ITT, insulin tolerance test; 3V, third ventricle; VFI, voluntary food intake.
E N E R G Y B A L A N C E - O B E S I T Y
Endocrinology, July 2012, 153(7):3147–3157 endo.endojournals.org 3147
and brain cerebrospinal fluid (CSF) therefore provide anindication of blood-brain insulin transfer. The hypotha-lamic arcuate nucleus, rich in insulin receptors, is the ma-jor site of insulin action on appetite and energy balanceaxes (5). Sheep provide an animal model, of similar bodyweight and adiposity to humans, amenable to long-termintracerebroventricular (icv) cannulation (2). Thus, wecan explore the dynamics of endogenous blood-braintransport in vivo from concurrent peripheral blood and icvCSF samples while also repeatedly testing intrahypotha-lamic insulin sensitivity by direct icv administration. Inthis way, the relative importance of these two sites of cen-tral insulin resistance may be elucidated.
A previous study found linear correlation betweenplasma and CSF insulin concentrations in sheep on anincreasing nutritional plane, although not obese, consis-tent with no change in insulin blood-brain transport effi-ciency with increasing insulinemia (6). In that same study,icv insulin infusion decreased voluntary food intake (VFI),in agreement with earlier literature (7), and the magnitudeof response (i.e. sensitivity) correlated negatively with thelevel of body fat (as measured by subjective adiposityscore) (6). Here we extend these findings by using obeseanimals (i.e. fatter than before) studied over a longer du-ration, including a period of postobesity calorie-restrictedweight loss, and objective measurement of total body fatby dual-energy x-ray absorptiometry (DEXA). Increasingadiposity in sheep has been shown to be associated withchanges in steady-state gene expression for appetite-reg-ulating peptide genes in the hypothalamus (8). The extentto which this might be due to altered insulin levels is un-known, but these earlier findings suggest that increasedadiposity may alter the brain’s set point for appetite reg-ulation. Here we aim to examine functional changes inhypothalamic insulin sensitivity in terms of the appetiteresponse to intrahypothalamic (third ventricular) insulinadministration, or in other words, central insulin sensi-tivity in terms of the food intake response. Although in-sulin is also known to activate neurons in other brain re-gions such as the brainstem (9), the hypothalamus is themain effector of changes in food intake induced by insulin(10).
Insulin sensitivity and insulin resistance are terms usu-ally associated with altered insulin-stimulated glucose up-take by peripheral tissues, mainly muscle and fat. Glucoserequirements are similar between nonruminants and ru-minants, although differences in digestive physiologymean that circulating glucose in ruminants comes mainlyfrom hepatic gluconeogenesis rather than from dietarycarbohydrate (11). Nonetheless, insulin is equally criticalin ruminants and nonruminants for maintaining normalblood glucose concentrations, acting primarily at periph-
eral sites (11, 12). Insulin resistance occurs when a givenamount of insulin fails to stimulate the same peripheraltissue uptake of glucose from the circulation or when moreinsulin is required for a given level of glucose uptake, typ-ically detected by insulin tolerance tests (ITT) and glucosetolerance tests (GTT). Compared with nonruminants, ru-minants tend to be relatively insensitive to insulin in termsof tissue glucose uptake (12), and chronically increasedadiposity can lead to hyperinsulinemia with no statisti-cally significant change in glycemia (8). However, in com-mon with other mammalian species, obesity in sheep isclearly associated with increased insulin resistance (13). Inour sheep model of obesity and subsequent weight loss, wetherefore investigated changes in peripheral insulin sensi-tivity, using peripherally administered ITT and GTT.
Therefore, the objectives of this study were to deter-mine in vivo concurrent changes in peripheral and centralinsulin sensitivity during obesity and whether central in-sensitivity in terms of the food intake response occurswithin the hypothalamus or at the level of insulin blood-brain transfer. The investigation went on to test the hy-pothesis that the presumed adverse effects of obesity oninsulin sensitivity would be ameliorated by subsequentweight loss brought about by global calorie restriction.
Materials and Methods
All procedures involving animals were conducted underthe United Kingdom Animals (Scientific Procedures) Act1986 and received approval from the local Ethical ReviewCommittee.
Animals and dietary treatmentsEighteen young adult female sheep (Dorset Horn � Greyface
crossbreeds, �10 months old) were surgically prepared with in-dwelling icv cannulae into the lateral cerebral ventricle and thirdventricle (3V) (14). They were housed in individual pens andgiven complete diet (12 MJ metabolizable energy and 140 g crudeprotein/kg; comprising 30% chopped hay, 42% rolled barley,and 17% soyabean meal) twice daily at 0800 and 1600 h, eitherad libitum (approximately three times the amount required forbody weight maintenance; obese group) or restricted to 1.25 �maintenance (control group) for 40 wk (n � 9 per group). There-after, for 16 wk, food was restricted to 0.75 � maintenance forthe obese group (now slimmers), and controls were fedad libitum(now fatteners). VFI of ad libitum food was monitored daily at0800 h by removing and weighing uneaten food (refusal margin�15%).
Body composition measurementsBody weight was recorded every 2 wk. At 4, 16, 28, 40, and
56 wk, body composition (total fat and lean content) was de-termined by DEXA (Norland XR-26 Mark II; Norland Corp.,Fort Atkinson, WI). DEXA scans were conducted in the morning
3148 Adam et al. Insulin Resistance in Obesity Endocrinology, July 2012, 153(7):3147–3157
(0800–1200 h); food and water were not withheld beforehand,but provision of fresh food was delayed until after the scan.General inhalation anesthesia (�2–5% halothane; HalothaneBP; Concord Pharmaceuticals Ltd., Essex, UK) was used by maskto immobilize the animal for the duration of the scan (15–20min). The coefficients of variation for DEXA measurements ofwhole-body fat and lean mass were less than 5%.
Blood sampling and endocrine challengesBlood and CSF samples were immediately chilled on ice, and
then plasma and CSF were stored at �20 C until analyzed forinsulin. At approximately 3- to 4-wk intervals, midmorningblood samples (2 h after feeding) were taken by jugular veni-puncture into heparinized vacutainer tubes, and CSF was sam-pled from the lateral cerebral ventricle. Central (icv) insulin chal-lenges were conducted at 16, 28, 40, and 56 wk, and peripheralGTT and ITT were conducted at 4, 16, 28, 40, and 56 wk asfollows.
Insulin (icv) challengesInjections were given via 3V cannula at 0800 h, immediately
before the morning feed, and again at 1000 h, with control in-jections (0.1 ml 0.9% saline) on d 1 and insulin (5 IU in 0.1 ml0.9% saline; Hypurin porcine insulin; CP Pharmaceuticals, Wr-exham, UK) on d 2. Blood samples were taken via temporaryjugular catheters at 30-min intervals from �30 min to �6 hrelative to the first icv injection. Plasma was assayed for insulinand glucose. For ad libitum-fed sheep, VFI was measured athourly intervals (replacing uneaten food) for 6 h after first in-jection. For restricted-fed sheep, VFI was measured of an addi-tional test meal of the same diet supplied ad libitum for 30 minat 1030 h.
GTT and ITTGTT and ITT were performed according to published pro-
tocols (15–17). One week after icv challenge, two temporaryjugular catheters (one for injection, one for sampling) were in-serted for GTT and ITT conducted with a 2-d interval in be-tween. For each test, sheep were fasted overnight (food removedat 1700 h; water freely available) to ensure steady-state rumenfermentation in the absence of recent feeding sessions and pro-vide baseline metabolic status at the start. Three baseline bloodsamples were taken (�30, �15, and 0 min) and, for ITT only, abaseline CSF sample. Glucose (0.5 g/kg 50% glucose solution;TPS Medical, Glasgow, UK) or insulin (0.5 IU/kg insulin made upin 0.9% saline to 5 ml; Hypurin) was administered iv at time zero(approximately 0830 h), and blood samples were collected at�5, 10, 15, 20, 30, 45, 60, 90, 120, and 180 min into chilledheparinized tubes; during ITT only, CSF was sampled at 60 min.After 180 min, animals were immediately re-fed with their nor-mal morning meal. GTT plasma samples were assayed for glu-cose and insulin, all ITT plasma samples for glucose, and a subset(�30, �15, 0 and 60 min) for insulin, and CSF was analyzed forinsulin.
Terminal proceduresAnimals were euthanized by lethal iv dose of pentobarbital
sodium (Euthatal; Merial Animal Health Ltd., Harlow, Essex,UK). Perirenal and visceral (omental and mesenteric) body fatdepots were dissected and weighed immediately postmortem.
Plasma and CSF analysesInsulin concentrations in plasma and CSF were measured by
RIA (18), with a limit of detection 0.2 �IU/ml and inter- andintraassay coefficients of variation less than 10%. Plasma glu-cose concentrations were measured by Yellow Springs Instru-ments (YSI, Yellow Springs, OH) dual biochemistry analyzer(model 2700), calibrated with known standards after everyfourth determination, and variation between duplicates was lessthan 5%.
Statistical analysesFor GTT and ITT, mean baseline (�30–0 min) plasma con-
centration and the area under the response curve above baseline(AUC; 5–180 min) were calculated for glucose (GTT and ITT)and insulin (GTT only). All statistical analyses used Minitab(Minitab Inc., State College, PA). Effects of diet on plasma andCSF concentrations, GTT and ITT parameters, and body com-position data were examined by ANOVA using the general linearmodel with time, group, and their interaction as specified terms,followed by Tukey’s post hoc tests. Effects of icv insulin on VFIwere examined by ANOVA using the general linear model withtime, group, day (icv insulin or saline), and their interactions asspecified terms, followed by Tukey’s post hoc tests. Correlation(Pearson’s product moment) and regression analyses were usedto explore relationships between CSF and plasma insulin con-centrations and between various response measurements andtotal body fat (DEXA values). Results are presented as groupmeans � SEM.
Results
Body weight and body compositionMean body weight of ad libitum-fed obese sheep in-
creased by 49 kg over the 40 wk of the first phase ofexperiment, whereas controls gained just 17 kg (P �
0.001; Fig. 1A). During the second 16-wk phase of exper-iment, mean body weight decreased by 4 kg in calorie-restricted slimmers (4% weight loss) but increased by 24kg in ad libitum-fed fatteners (P � 0.001; Fig. 1A). Therewas no difference in mean body weight between groups atthe end.
DEXA measurements showed that the obese group hadgreater total lean mass from 4 wk (P � 0.05–0.01) andgreater total body fat mass (P � 0.001) and percent bodyfat from 16 wk (P � 0.001) compared with controls in thefirst phase of experiment (Fig. 1, B–D). After the 16-wksecond phase, total body fat mass and percent body fatwere not different between groups, but total lean mass waslower in slimmers than fatteners (P � 0.01) (Fig. 1, B–D).Postmortem mean adipose tissue weights at 56 wk werenot different between slimmers and fatteners for perirenal(2.76 � 0.239 vs. 2.59 � 0.109 kg) or visceral fat (3.19 �
0.219 vs. 3.29 � 0.236 kg), but perirenal fat weight cor-related with DEXA total fat mass (P � 0.01; Fig. 1E).
Endocrinology, July 2012, 153(7):3147–3157 endo.endojournals.org 3149
Plasma and CSF insulinemiaInsulin concentrations were consistently higher in
obese than control plasma (P � 0.001) and CSF (P � 0.01)in midmorning samples taken from fed animals in the first
phase of experiment (Fig. 2, A and B);throughout this period, CSF concentra-tions correlated with plasma concen-trations for all animals (r2 � 0.45; P �0.001; Fig. 2C) and separately for obese(r2 � 0.22; P � 0.001) and controlgroups (r2 � 0.18; P � 0.001). In thesecond phase, plasma insulin washigher in fatteners than slimmers (P �0.001), with group � time interaction(P � 0.01) reflecting increasing valuesin fatteners but decreasing values inslimmers (Fig. 2D). CSF concentrationswere similarly higher in fatteners thanslimmers (P � 0.001; Fig. 2E), and theycorrelated with plasma concentrationsfor fatteners (P � 0.001) but not forslimmers (Fig. 2F).
Samples taken at the start of ITTwere used to obtain baseline insulinconcentrations after overnight fast.During the first phase of experiment,baseline plasma values increased inobese sheep to be higher than those incontrols from 16 wk (Fig. 2G, P �0.001), and baseline CSF insulin con-centrations correlated with plasma val-ues across both groups (Fig. 2H, P �0.001). The slope and intercept of thisbaseline regression equation (0.10 �0.028 and 2.44 � 0.508, respectively;Fig. 2H) did not differ significantlyfrom those of the equivalent relation-ship in the fed state (0.14 � 0.017 and2.44 � 0.608, respectively; Fig. 2C).At the end of the second phase of ex-periment, ITT baseline plasma insulinwas higher in fatteners (previouslycontrol) than slimmers (previouslyobese; 25.9 � 3.60 vs. 11.5 � 0.98�IU/ml, P � 0.01); however, therewere insufficient CSF samples for cor-relation analysis.
Plasma and CSF insulin concentra-tions were both increased more than10-fold in samples taken 1 h after ex-ogenous insulin iv injection during ITT,and they were significantly correlated
in the first phase of experiment (Fig. 2I; P � 0.001), al-though the slope of this relationship was lower (0.05 �0.013, P � 0.001) than the slope of the equivalent endog-enous relationships above; there were insufficient CSF
30
40
50
60
70
80
90
100
0 7 13 20 26 32 38 44 50 56
Bod
y w
eigh
t (kg
)
Time (weeks)
***A
0
10
20
30
40
50
60
4 16 28 40 56
Tota
l lea
n m
ass
(kg)
Time (weeks)
***
****
B
0
10
20
30
40
50
60
4 16 28 40 56
Tota
l fat
mas
s (k
g)
Time (weeks)
C
0
10
20
30
40
50
4 16 28 40 56
Tota
l bod
y fa
t (%
)
Time (weeks)
***
D***
***
r² = 0.43**1.5
2.0
2.5
3.0
3.5
4.0
4.5
20 25 30 35 40 45 50
Perir
enal
fat w
eigh
t (kg
)
DEXA total fat mass (kg)
E
***
***
***
***
FIG. 1. A–D, Body weight (A), whole body lean mass (B), whole body fat mass (C), and totalbody fat percentage (D) determined by DEXA in obese sheep (solid symbols and black bars)given ad libitum food for 40 wk followed by restricted food (now slimmers) for 16 wk andcontrol sheep (open symbols and white bars) given restricted food for 40 wk followed by adlibitum food (now fatteners) for 16 wk (n � 9 per group); E, relationship betweenpostmortem weight of perirenal fat and DEXA total body fat mass at 56 wk for all sheep. InA, the vertical line indicates when dietary intakes were changed, and the arrows indicatewhen endocrine challenges were conducted. *, P � 0.05; **, P � 0.01; ***, P � 0.001.
3150 Adam et al. Insulin Resistance in Obesity Endocrinology, July 2012, 153(7):3147–3157
samples available for correlation in the second phase ofexperiment.
Baseline plasma insulin during ITT correlated with DEXAtotal percent body fat in the first phase of experiment (y � 0.62x �5.36; r2 � 0.41; P � 0.001) and overall in both phases ofexperiment (y � 0.60 x �4.82; r2 � 0.34; P � 0.001).
Peripheral GTT and ITTSimilar patterns of response were seen at all times for
plasma glucose and insulin during GTT and plasma glu-cose during ITT, and these are illustrated in Fig. 3 for obeseand control sheep at 40 wk. Mean group response param-eters at all time points are presented in Fig. 4. In the first
0
20
40
60
80
4 6 9 12 15 20 22 24 26 32 38Plas
ma
insu
lin (µ
IU/m
l)
Time (weeks)
***
02468
101214
4 6 9 12 15 20 22 24 26 32 38
CSF
insu
lin (µ
IU/m
l)
Time (weeks)
***
0
5
10
15
42 45 48 51 54 56C
SF in
sulin
(µIU
/ml)
Time (weeks)
***
*
0
20
40
60
80
100
42 45 48 51 54 56
Plas
ma
insu
lin (µ
IU/m
l)
Time (weeks)
y = 0.14x + 2.44r² = 0.45**
0
5
10
15
20
0 20 40 60 80
CSF
insu
lin (µ
IU/m
l)
Plasma insulin (µIU/ml)
y = 0.15x - 0.87r² = 0.66***
y = 0.03x + 5.55r² = 0.03
0
5
10
15
20
20 40 60 80 100 120
CSF
insu
lin (µ
IU/m
l)
Plasma insulin (µIU/ml)
***
y = 0.10x + 2.44r² = 0.22***
0
2
4
6
8
10
12
0 10 20 30 40 50
CSF
insu
lin (µ
IU/m
l)
Plasma insulin (µIU/ml)
0
5
10
15
20
25
30
4 16 28 40
Plas
ma
insu
lin (µ
IU/m
l)
Time (weeks)
****** ***
y = 0.05x + 16.7r² = 0.28***
0
20
40
60
80
0 200 400 600 800
CSF
insu
lin (µ
IU/m
l)
Plasma insulin (µIU/ml)
A B C
D E F
G H I
FIG. 2. Insulin concentrations in midmorning fed samples of plasma (A) and CSF (B) of obese sheep (solid symbols) given ad libitum food andcontrol sheep (open symbols) given restricted food for 40 wk (n � 9 per group); C, relationship between concomitant CSF and plasma fed insulinconcentrations (using data from A and B); D and E, fed insulin concentrations in plasma (D) and CSF (E) of slimmers (solid symbols, previouslyobese) given restricted food and fatteners (open symbols, previously control) given ad libitum food for 16 wk; F, relationship between concomitantCSF and plasma fed insulin concentrations (using data from D and E); G, baseline plasma insulin concentrations after overnight fast; H, relationshipbetween concomitant baseline CSF and plasma insulin concentrations; I, relationship between concomitant CSF and plasma insulin concentrations1 h after peripheral insulin injection during ITT in obese (black bars and closed symbols) and control sheep (white bars and open symbols). *, P �0.05; **, P � 0.01; ***, P � 0.001.
0
5
10
15
20
25
30
-30 0 30 60 90 120 150 180
Plas
ma
gluc
ose
(mm
ol/l)
Time (min) relative to glucose injection
0
0.5
1
1.5
2
2.5
3
3.5
4
-30 0 30 60 90 120 150 180
Plas
ma
gluc
ose
(mm
ol/l)
Time (min) relative to insulin injection
0
20
40
60
80
100
120
140
160
-30 0 30 60 90 120 150 180
Plas
ma
insu
lin (u
IU/m
l)
Time (min) relative to glucose injection
A B C
FIG. 3. A and B, Plasma glucose response (A) and plasma insulin response (B) during GTT; C, plasma glucose response during ITT in obese (solidsymbols) and control (open symbols) sheep at 40 wk.
Endocrinology, July 2012, 153(7):3147–3157 endo.endojournals.org 3151
phase of experiment, mean baseline plasma glucose con-centration was higher in obese than control sheep duringGTT at 4–28 wk and ITT at 4 wk (Fig. 4, A and D; P �0.01–0.001), and glucose AUC was higher in obese thancontrol sheep during GTT at 16–40 wk (Fig. 4B; P �0.01–0.001) and during ITT at 4–28 wk (Fig. 4E; P �0.001). At 56 wk (end of the second phase), GTT baselineglucose and glucose AUC were not different (Fig. 4, A andB), but ITT baseline glucose and glucose AUC were higherin fatteners than slimmers (P � 0.01 and P � 0.001, re-spectively; Fig. 4, D and E). Thus, for all animals through-out both phases of experiment, DEXA total percent bodyfat correlated with GTT glucose AUC (P � 0.001; Fig. 4C)and ITT glucose AUC (P � 0.001; Fig. 4F).
Baseline plasma insulin concentrations during GTTmatched those measured during ITT, with values in obesesheep higher than controls at 16–40 wk in the first phase(P � 0.01–0.001) and values similar between groups at 56wk at the end of the second phase of experiment (Fig. 4G).Plasma insulin AUC during GTT was also greater in obesethan control sheep at 16–40 wk (P � 0.001) but similarbetween slimmers and fatteners at 56 wk (Fig. 4H). Forboth parameters, there was a significant increase over time
for the obese group (P � 0.001), and GTT insulin AUCcorrelated with DEXA total percent body fat (P � 0.001;Fig. 4I).
Insulin (icv) challengeFirst it was established that VFI by ad libitum-fed sheep
on d 1 (icv saline) of each challenge was representative ofthat individual’s daily intake over the preceding week. VFIwas decreased after icv insulin (d 2) on all occasions inboth groups (P � 0.001). Cumulative intake remainedlower in the obese group and fatteners throughout the 6 hafter icv insulin (Fig. 5A), but 24-h VFI was not affected,and VFI of the test meal at 3 h after icv insulin was de-creased in controls and slimmers (Fig. 5A). Because mea-surements for the controls (first phase of experiment) andslimmers (second phase) were confined to the test mealperiod 2.5–3 h after the first icv injection, proportionaldecreases inVFI for3hafter injectionby the ad libitum-fedobese group and fatteners were used for group compari-sons (Fig. 5B). The magnitude of response in the first phaseof experiment was similar between groups at 16 wk butwas greater in control than obese sheep at 28 wk (P �0.001) and 40 wk (P � 0.05) and in the second phase of
2.0
2.5
3.0
3.5
4.0
4 16 28 40 56
ITT
base
line
gluc
ose
(mm
ol/l)
Time (weeks)
**
D***
0.00
0.04
0.08
0.12
4 16 28 40 56
ITT
gluc
ose
AUC
(m
mol
/l/m
in)
Time (weeks)
*** *** *** ***
E
2.0
2.5
3.0
3.5
4.0
4 16 28 40 56GTT
bas
elin
e gl
ucos
e (m
mol
/l)
Time (weeks)
** **
0.2
0.4
0.6
0.8
1.0
4 16 28 40 56
GTT
glu
cose
AU
C
(mm
ol/l/
min
)
Time (weeks)
******
***
B***
r² = 0.34***
50
100
150
200
10 20 30 40 50 60 70
GTT
glu
cose
AU
C
(mm
ol/l/
min
)
DEXA total body fat (%)
C
r² = 0.10***
5
10
15
20
25
10 20 30 40 50 60 70
ITT
gluc
ose
AUC
(m
mol
/l/m
in)
DEXA total body fat (%)
F
0
10
20
30
4 16 28 40 56
GTT
bas
elin
e in
sulin
(u
IU/m
l)
Time (weeks)
***
G
0
1
2
3
4
5
4 16 28 40 56
GTT
insu
lin A
UC
(u
IU/m
l/min
)
Time (weeks)
H
**
***
******
***r² = 0.44***
0123456
10 20 30 40 50 60 70
GTT
insu
lin A
UC
(u
IU/m
l/min
)
DEXA total body fat (%)
I
A
FIG. 4. A–C, Baseline plasma glucose concentration (A), glucose AUC (B), and relationship between glucose AUC and DEXA percent total body(C) fat during peripheral GTT; D–F, baseline plasma glucose concentration (D), glucose AUC (E), and relationship between glucose AUC and DEXApercent total body fat (F) during peripheral ITT; G–I, baseline plasma insulin concentration (G), insulin AUC (H), and relationship between insulinAUC and DEXA percent total body fat (I) during GTT in obese sheep (black bars and closed symbols) given ad libitum food for 40 wk followed byrestricted food (now slimmers) for 16 wk and control sheep (white bars and open symbols) given restricted food for 40 wk followed by ad libitumfood (now fatteners) for 16 wk (n � 9 per group). **, P � 0.01; ***, P � 0.001.
3152 Adam et al. Insulin Resistance in Obesity Endocrinology, July 2012, 153(7):3147–3157
experiment was similar between slimmers and fatteners at56 wk (Fig. 5B). Within the obese group, there was a de-crease in magnitude of response between 16 and 28 wk(P � 0.01), no change between 28 and 40 wk, and nochange in response between 40 and 56 wk during the sec-ond phase of experiment when they had become slimmers(Fig. 5B). Overall, the proportional VFI response corre-lated negatively with DEXA percent total body fat for allanimals over the whole experiment (P � 0.01, Fig. 5C).
Plasma insulin concentration during the 6 h after icvinsulin as opposed to icv saline was increased in eachgroup on each occasion (P � 0.01–0.001), but there wasno difference between groups in the magnitude of increase,and there was a corresponding slight decrease in plasmaglucose during this period (only significant for controls at28 wk, P � 0.05; Table 1).
Discussion
These data demonstrated in vivo insulin resistance devel-oping first in the periphery and then centrally as obesity
developed, with resistance intensifying as adiposity levelscontinued to increase. The central resistance (in terms ofthe VFI response) appeared to result from intrahypotha-lamic insensitivity to insulin rather than impaired blood-brain insulin transfer. Subsequent modest weight loss im-proved peripheral insulin sensitivity and decreasedproportional blood-brain insulin transport, but there wasno evidence for improvement in central insulin sensitivity.
In the first phase of experiment, the increased weightgain in ad libitum-fed obese animals was attributablelargely to greater adipose tissue gain. Obesity (�33%body fat) was attained by 16 wk, but adiposity levels none-theless continued to increase up to 40 wk (Fig. 1). Corre-spondingly, weight loss was mainly due to fat loss whenformerly obese animals were food restricted (slimmers)during the second phase of experiment. DEXA measure-ments of body fat in these female sheep were substantiatedby the close correlation between postmortem perirenal fatweight and antemortem DEXA total body fat mass, sim-ilar to the correlation recorded previously between fat de-pot weights and DEXA fat measurements in male sheep
300
500
700
900
1100
1 2 3 4 5 6
VFI (
g)
Time (h)
16 weeks
200
400
600
800
1000
1 2 3 4 5 6Time (h)
28 weeks
200
400
600
800
1000
1 2 3 4 5 6Time (h)
40 weeks
200
400
600
800
1000
1 2 3 4 5 6Time (h)
56 weeks
y = -0.009x + 0.63r² = 0.10**
-0.20
0.00
0.20
0.40
0.60
0.80
10 20 30 40 50 60 70
Prop
otio
nal d
ecre
ase
in V
FI
DEXA total body fat (%)
C
0
0.1
0.2
0.3
0.4
0.5
16 28 40 56
Prop
or�
onal
dec
reas
e in
VFI
Time (weeks)
SED = 0.033
***
B
A
*
FIG. 5. A, VFI during 6 h after saline control (triangles, dotted line) or insulin icv injection (circles, solid line) in ad libitum-fed obese sheep (solidsymbols) at 16, 28, and 40 wk and fatteners (open symbols) at 56 wk and VFI of ad libitum test meal at 3 h for control sheep (open symbols) at16–40 wk and slimmers (solid symbols) at 56 wk. Obese sheep were given ad libitum food for 40 wk followed by restricted food (now slimmers)for 16 wk and control sheep were given restricted food for 40 wk followed by ad libitum food (now fatteners) for 16 wk (n � 9 per group). B,Proportional decrease in VFI during 3 h after icv injection of insulin compared with saline control injection at 16–40 wk in obese (black bars) andcontrol sheep (white bars) and at 56 wk in slimmers (formerly obese, black bars) and fatteners (formerly control, white bars). C, Relationshipthroughout the experiment between proportional decrease in VFI after icv insulin and percent total body fat (by DEXA) in obese/slimmers (solidsymbols) and control/fatteners (open symbols). **, P � 0.01; ***, P � 0.001.
Endocrinology, July 2012, 153(7):3147–3157 endo.endojournals.org 3153
(2). The changes in body fat content were associated withunderlying changes in peripheral and central insulinsensitivity.
Peripheral and central (CSF) insulinemia were in-creased in obese sheep. Indeed, baseline plasma insulinconcentrations correlated closely with total body fat con-tent for all animals throughout the experiment, and takenwith the additional correlation between CSF and plasmaconcentrations, these data support the purported role ofcirculating insulin as an adiposity signal within the brain(19). The constant CSF to plasma concentration ratio seenin both fed and overnight fasted obese and control animals(Fig. 2, C and H) was strongly suggestive of unimpededproportional blood-brain insulin transport during physi-ological insulinemia (20). These results apparently con-trast with reports of decreased blood-brain insulin trans-port in genetically obese rats (21, 22) and decreased brainuptake of iv-injected radiolabeled insulin in obese mice(23). This apparent discrepancy could reflect bona fidespecies differences or could be due to differences in insulintransport into the CSF vs. brain tissue uptake of insulinand/or differences in the transport of the radiolabeled vs.native insulin molecule, but this is the first report of nativeendogenous insulin blood-brain transfer in a nongeneti-cally altered obese animal model. Furthermore, the CSF toplasma insulin ratio in our model was decreased whensupraphysiological plasma concentrations were presentafter iv insulin administration (Fig. 2I), indicative ofblood-brain transport saturation during extreme hyper-insulinemia (20). This observation applied to both controland obese groups, providing no evidence for obesity-im-paired insulin blood-brain transport, in contrast to leptin(2). Although it has previously been reported that iv in-sulin infusion rapidly increases lumbar CSF insulin con-centrations in humans (24), our model detected the in-
crease in central CSF closer to insulin’s central nervoussystem sites of action (5). Clearly, the evidence indicatesthat acute increases in systemic insulin are rapidly re-ported to the brain. Interestingly, however, the data alsoindicated that decreases in systemic insulin may not be soaccurately reported to the brain, because CSF concentra-tions were disproportionately lower than those in plasmaduring calorie-restricted weight loss. This is similar to thedisproportionate decrease in CSF leptin seen in calorie-restricted obese sheep and would further contribute to themisleading information received by the brain, underre-porting levels of adiposity and resisting additional weightloss (2). Recent recognition of the role of central insulinresistance in the pathogenesis of human obesity and type2 diabetes assumes that insufficient insulin is entering thebrain, and treatments are being developed to specificallyincrease brain insulin concentrations using intranasal ad-ministration or commercial insulins with altered pharma-cokinetics (25, 26); however, the present findings of per-sistent obesity-induced hypothalamic insulin insensitivitywould indicate that such approaches may not be effica-cious in reducing appetite and body weight.
The GTT and ITT provided information on glucosetolerance and peripheral insulin sensitivity in our model.After just 4 wk of dietary treatments, preobese animalshad elevated baseline glycemia and higher glucose AUCduring ITT (Fig. 4), providing early signs of impaired glu-cose handling and peripheral insulin insensitivity. From16 wk, when obesity was established, glucose baseline andAUC remained higher during GTT and ITT, and insulinbaseline and AUC during GTT were also higher in obesethan control sheep, demonstrating their ongoing periph-eral insulin insensitivity (Fig. 4). Furthermore, these mea-sures worsened over time with further increasing adipositylevels after obesity was established, supported by the sig-
TABLE 1. Average plasma insulin and glucose concentrations during 6 h after icv saline or insulin injection
Obese sheep were given ad libitum food for 40 wk followed by restricted food for 16 wk (now slimmers), and control sheep were given restrictedfood for 40 wk followed by ad libitum food for 16 wk (now fatteners) (n � 9 per group). At 16, 28, 40, and 56 wk, icv injections were given ofsaline (d 1) and insulin (d 2). P values are shown for icv insulin vs. saline, within groups.a P � 0.05; b P � 0.01; c P � 0.001.
3154 Adam et al. Insulin Resistance in Obesity Endocrinology, July 2012, 153(7):3147–3157
nificant correlations between DEXA total body fat mea-surements and glucose AUC during GTT, glucose AUCduring ITT, and insulin AUC during GTT (Fig. 4). Con-versely, modest weight loss (4%, largely fat) decreasedbaseline insulin, insulin AUC during GTT, and glucoseAUC during ITT, indicating an improvement in peripheralinsulin sensitivity, in agreement with findings in formerlyobese humans on hypocaloric diets (27, 28). This improve-ment occurred in the present slimmers despite their re-maining obese levels of body fat (40%, Fig. 1D), indicatingthe metabolic benefits of calorie restriction and weight loss(however small) even when the obesity is not cured.
A recent commentary highlights the current workinghypothesis for insulin action in the brain, and althoughmany brain sites are insulin responsive, the hypothalamusis the main effector of metabolic changes induced by in-sulin (10). Insulin acts on its receptor in the hypothalamusto lower food intake (29). The neural insulin receptorknockout mouse is hyperphagic (30), and central insulinadministration decreases food intake in rats (31), mice(32), and baboons (33) as well as sheep (6, 7). Here, weused our in vivo sheep model to focus on functionalchanges in the food intake response to insulin. The foodintake responses to icv insulin provided information onintrahypothalamic sensitivity because the injection wasplaced into the 3V at the middle of the hypothalamus, andsignificant decreases in intake were indeed recorded (Fig.5A). Food intake data typically attract large variation;nonetheless, differences in magnitude of response betweengroups and over time were detected, and examination ofindividual data revealed negative correlation between themagnitudeof responseand total body fat content across allanimals throughout the experiment (Fig. 5C). This rela-tionship was also seen in our earlier study in overfed males(6) and now reveals a graded response in obese females.Together with the finding that icv insulin decreases foodintake in lean but not obese Zucker rats (34), the datashow that central (intrahypothalamic) insulin resistance interms of the appetite response increases with obesity andincreasing adiposity. Interestingly, after modest weightloss, the apparent central insulin sensitivity remained com-mensurate with the level of adiposity as described by thenegative regression relationship between the magnitude ofthe decrease in food intake and total body fat. It wouldappear that more substantial weight loss in the obese isnecessary for a measurable improvement in central insulinsensitivity.
Blood samples taken over the 6-h period after icv in-jections provided additional information on insulinemiaandglycemia (Table1).After icv saline, plasma insulinandglucose concentrations did not show major fluctuations,as is normal for ruminants on roughage diets (35), and the
6-h average values endorsed the periodic midmorningsample data by showing sustained increases in insulinemiaand glycemia in obese vs. control animals and fatteners vs.slimmers. The increased peripheral insulinemia after icvinsulin was most likely due to leakage of exogenous insulinfrom the brain rather than a specific action of insulinwithin the brain, because the magnitude of increase wassimilar between groups and the direct effect of increasedbrain insulin on insulin secretion is purportedly inhibi-tory, not stimulatory (36). Consequently, glucose concen-trations decreased slightly, but they remained in the nor-mal glycemic range, and no difference was detected inglucose handling between groups.
The second phase of this experiment revealed not onlyeffects of weight loss in the obese animal but also under-lying differences between animals of equal adiposity andyet in opposing energy balance trajectories. Thus, meta-bolic endocrine differences existed between sheep withsimilar total body fat content in that those in negativeenergy balance (slimmers) had significantly lower periph-eral and brain insulinemia than those in positive energybalance (fatteners). Furthermore, peripheral insulin sen-sitivity appeared to remain lower in (formerly obese) slim-mers because their glucose AUC during ITT was higherthan that of fatteners. Previously, we have found differ-ences in hypothalamic gene expression in sheep of similaradiposity but in opposing energy balance trajectories, withorexigenic genes up-regulated in negative energy balanceand anorexigenic genes up-regulated in positive energybalance (37). There was no evidence from the present ex-periment for differences in central insulin sensitivity (interms of VFI response to icv insulin) due to opposing nu-tritional trajectories at the same level of adiposity. How-ever, whereas the CSF to plasma concentration ratio infatteners fitted the relationship predicted in the first phaseof the experiment, CSF concentrations in slimmers weredisproportionately low, revealing the brain in negative en-ergy balance to be experiencing exaggerated hypoinsu-linemia. As indicated earlier, and matching the scenariowith central hypoleptinemia (2), this would increase thebrain’s drive to reverse the negative energy balance tra-jectory even when the individual is still obese, makingcompliance to restricted food intake more difficult.
This experiment has also revealed novel findings on therelative timing of insulin resistance in the brain and pe-riphery. Whereas decreased peripheral insulin sensitivitywas evident (4 wk) before obesity developed (16 wk), therewas no evidence for a change in central insulin sensitivityuntil after obesity was established (28 wk). Thereafter, theintensity of insulin resistance continued to increase in boththe periphery and the brain as obese individuals continuedto gain body fat. Subsequently, very modest weight loss
Endocrinology, July 2012, 153(7):3147–3157 endo.endojournals.org 3155
(4%) markedly improved peripheral insulin sensitivity af-ter 16 wk, whereas the amount of body fat lost was in-sufficient to significantly improve central insulin sensitiv-ity. It appears that obesity may have longer-lasting adverseconsequences for central than for peripheral insulin sen-sitivity. Previously, changes in steady-state hypothalamicgene expression for appetite-regulating peptide genes thatoccur with chronically increased adiposity in sheep led tothe suggestion that increased adiposity may alter thebrain’s set point for appetite regulation (8). The presentfunctional in vivo data add support to this hypothesis inthat the brain’s set point for insulin-responsive changes infood intake was indeed altered by increased adiposity, andfurthermore, it did not appear to be restored by subse-quent modest loss of adiposity.
In conclusion, in our in vivo model, insulin resistancedeveloped first in the periphery and then centrally as obe-sity developed, and thereafter, resistance at both sites con-tinued to correlate with adiposity. Modest weight loss im-proved peripheral but not central insulin sensitivity andwas associated with relative hypoinsulinemia within thebrain. The central insulin resistance, in terms of the foodintake response, resulted from intrahypothalamic insen-sitivity rather than impaired blood-brain transfer, butweight loss in obese individuals was associated with re-duced proportional blood-brain insulin transport.
Acknowledgments
We thank Rowett Farm staff for routine daily animal care andRowett Bioresources Group for assistance with surgeries,DEXA, and veterinary care.
Address all correspondence and requests for reprints to:Dr. Clare Adam, Obesity and Metabolic Health Theme,Rowett Institute of Nutrition and Health, University of Ab-erdeen, Bucksburn, Aberdeen AB21 9SB, United Kingdom.E-mail:[email protected].
The research was funded by the Scottish Government.Disclosure Summary: The authors have nothing to disclose.
References
1. Belgardt BF, Brüning JC 2010 CNS leptin and insulin action in thecontrol of energy homeostasis. Ann NY Acad Sci 1212:97–113
2. Adam, CL, Findlay PA 2010 Decreased blood-brain leptin transferin an ovine model of obesity and weight loss: resolving the cause ofleptin resistance. Int J Obes (Lond) 34:980–988
3. Gerozissis K, Kyriaki G 2003 Brain insulin: regulation, mechanismsof action and functions. Cell Mol Neurobiol [Erratum (2003) 23:873] 23:1–25
4. Laron Z 2009 Insulin and the brain. Arch Physiol Biochem 115:112–116
5. Schwartz GJ 2009 Your brain on fat: dietary-induced obesity im-pairs central nutrient sensing. Am J Physiol Endocrinol Metab 296:E967–E968
6. Miller DW, Bennett EJ, Harrison JL, Findlay PA, Adam CL 2011Adiposity and plane of nutrition influence reproductive neuroendo-crine and appetite responses to intracerebroventricular insulin andneuropeptide-Y in sheep. Reprod Fertil Dev 23:329–338
7. Foster LA, Ames NK, Emery RS 1991 Food intake and serum insulinresponses to intraventricular infusions of insulin and IGF-I. PhysiolBehav 50:745–749
8. Henry BA, Tilbrook AJ, Dunshea FR, Rao A, Blache D, Martin GB,Clarke IJ 2000 Long-term alterations in adiposity affect the expres-sion of melanin-concentrating hormone and enkephalin but not pro-opiomelanocortin in the hypothalamus of ovariectomized ewes. En-docrinology 141:1506–1514
9. Ghuman SP, Morris R, Scherzer J, Routly JE, Mobasheri A, SmithRF, Dobson H 2011 Neuronal responses in the brainstem and hy-pothalamic nuclei following insulin treatment during the late follic-ular phase in the ewe. Reprod Domest Anim 46:121–129
10. Filippi BM, Mighiu PI, Lam TK 2012 Is insulin action in the brainclinically relevant? Diabetes 61:773–775
11. Brockman RP 1978 Roles of glucagon and insulin in the regulationof metabolism in ruminants. A review. Can Vet J 19:55–62
12. Sasaki S 2002 Mechanism of insulin action on glucose metabolismin ruminants. Anim Sci J 73:423–433
13. Bergman EN, Reulein SS, Corlett RE 1989 Effects of obesity oninsulin sensitivity and responsiveness in sheep. Am J Physiol 257:E772–E781
14. Miller DW, Findlay PA, Morrison MA, Raver N, Adam CL 2002Seasonal and dose-dependent effects of intracerebroventricular lep-tin on luteinising hormone secretion and appetite in sheep. J Endo-crinol 175:395–404
15. Wallace JM, Milne JS, Aitken RP 2010 Effect of weight and adi-posity at conception and wide variations in gestational dietary in-take on pregnancy outcome and early postnatal performance inyoung adolescent sheep. Biol Reprod 82:320–330
16. Wallace JM, Milne JS, Adam CL, Aitken RP 2012 Adverse meta-bolic phenotype in low-birth-weight lambs and its modification bypostnatal nutrition. Br J Nutr 107:510–522
17. Poore KR, Cleal JK, Newman JP, Boullin JP, Noakes DE, HansonMA, Green LR 2007 Nutritional challenges during developmentinduce sex-specific changes in glucose homeostasis in the adultsheep. Am J Physiol Endocrinol Metab 292:E32–E39
18. MacRae JC, Bruce LA, Hovell FD, Hart IC, Inkster J, Walker A,Atkinson T 1991 Influence of protein nutrition on the response ofgrowing lambs to exogenous bovine growth hormone. J Endocrinol130:53–61
19. Woods SC, D’Alessio DA 2008 Central control of body weight andappetite. J Clin Endocrinol Metab 93(Suppl 1):S37–S50
20. Banks WA, Jaspan JB, Kastin AJ 1997 Selective, physiological trans-port of insulin across the blood-brain barrier: novel demonstrationby species-specific radioimmunoassay. Peptides 18:1257–1262
21. Baskin DG, Stein LJ, Ikeda H, Woods SC, Figlewicz DP, Porte D Jr,Greenwood MR, Dorsa DM 1985 Genetically obese Zucker ratshave abnormally low brain insulin content. Life Sci 36:627–633
22. Gerozissis K, Orosco M, Rouch C, Nicolaidis S 1993 Basal andhyperinsulinemia-induced immunoreactive hypothalamic insulinchanges in lean and genetically obese Zucker rats revealed by mi-crodialysis. Brain Res 611:258–263
23. Urayama A, Banks WA 2008 Starvation and triglycerides reverse theobesity-induced impairment of insulin transport at the blood-brainbarrier. Endocrinology 149:3592–3597
24. Wallum BJ, Taborsky Jr GJ, Porte Jr D, Figlewicz DP, Jacobson L,Beard JC, Ward WK, Dorsa D 1987 Cerebrospinal fluid insulinlevels increase during intravenous insulin infusions in man. J ClinEndocrinol Metab 64:190–194
25. Ketterer C, Tschritter O, Preissl H, Heni M, Häring HU, Fritsche A
3156 Adam et al. Insulin Resistance in Obesity Endocrinology, July 2012, 153(7):3147–3157
2011 Insulin sensitivity of the human brain. Diabetes Res Clin Pract93(Suppl 1):S47–S51
26. Ott V, Benedict C, Schultes B, Born J, Hallschmid M 2012 Intra-nasal administration of insulin to the brain impacts cognitive func-tion and peripheral metabolism. Diabetes Obes Metab 14:214–221
27. De Luis DA, Gonzalez Sagrado M, Conde R, Aller R, Izaola O,Castro MJ, Romero E 2010 Lack of effect of a moderate hypocaloricdiet on visfatin levels in morbid obese patients: relationship withinsulin resistance. Eur Rev Med Pharmacol Sci 14:1031–1036
28. Mitterberger MC, Mattesich M, Klaver E, Lechner S, Engelhardt T,Larcher L, Pierer G, Piza-Katzer H, Zwerschke W 2010 Adipokineprofile and insulin sensitivity in formerly obese women subjected tobariatric surgery or diet-induced long-term caloric restriction. JGerontol A Biol Sci Med Sci 65:915–923
29. Schwartz MW, Woods SC, Porte Jr D, Seeley RJ, Baskin DG 2000Central nervous system control of food intake. Nature 404:661–671
30. Brüning JC, Gautam D, Burks DJ, Gillette J, Schubert M, Orban PC,Klein R, Krone W, Müller-Wieland D, Kahn CR 2000 Role of braininsulin receptor in control of body weight and reproduction. Science289:2122–2125
31. Clegg DJ, Riedy CA, Smith KA, Benoit SC, Woods SC 2003 Dif-
ferential sensitivity to central leptin and insulin in male and femalerats. Diabetes 52:682–687
32. Brown LM, Clegg DJ, Benoit SC, Woods SC 2006 Intraventricularinsulin and leptin reduce food intake and body weight in C57BL/6Jmice. Physiol Behav 89:687–691
33. Woods SC, Lotter EC, McKay LD, Porte D Jr 1979 Chronic intra-cerebroventricular infusion of insulin reduces food intake and bodyweight of baboons. Nature 282:503–505
34. Ikeda H, West DB, Pustek JJ, Figlewicz DP, Greenwood MR, PorteJr D, Woods SC 1986 Intraventricular insulin reduces food intakeand body weight of lean but not obese Zucker rats. Appetite 7:381–386
35. Sasaki Y, Takahashi H, Aso H, Hikosaka K, Hagino A, Oda S 1984Insulin response to glucose and glucose tolerance following feedingin sheep. Br J Nutr 52:351–358
36. Benedict C, Brede S, Schiöth HB, Lehnert H, Schultes B, Born J,Hallschmid M 2011 Intranasal insulin enhances postprandial ther-mogenesis and lowers postprandial serum insulin levels in healthymen. Diabetes 60:114–118
37. Miller DW, Harrison JL, Bennett EJ, Findlay PA, Adam CL 2007Nutritional influences on reproductive neuroendocrine output: in-sulin, leptin and orexigenic neuropeptide signaling in the ovine hy-pothalamus. Endocrinology 148:5313–5322
Members can search for endocrinology conferences, meetings, and webinars on the Worldwide Events Calendar.
www.endo-society.org/calendar
Endocrinology, July 2012, 153(7):3147–3157 endo.endojournals.org 3157
