Content uploaded by Katherine H Ingram
Author content
All content in this area was uploaded by Katherine H Ingram on Jun 30, 2016
Content may be subject to copyright.
Available via license: CC BY-NC-ND 3.0
Content may be subject to copyright.
Anna E. Thalacker-Mercer,
1,2,3,4
Katherine H. Ingram,
2,5
Fangjian Guo,
2
Olga Ilkayeva,
6
Christopher B. Newgard,
6,7
and W. Timothy Garvey
2,3
BMI, RQ, Diabetes, and Sex
Affect the Relationships
Between Amino Acids and
Clamp Measures of Insulin
Action in Humans
Previous studies have used indirect measures of
insulin sensitivity to link circulating amino acids with
insulin resistance and identify potential biomarkers
of diabetes risk. Using direct measures (i.e.,
hyperinsulinemic-euglycemic clamps), we examined
the relationships between the metabolomic amino
acid profile and insulin action (i.e., glucose disposal
rate [GDR]). Relationships between GDR and serum
amino acids were determined among insulin-
sensitive, insulin-resistant, and type 2 diabetic
(T2DM) individuals. In all subjects, glycine (Gly) had
the strongest correlation with GDR (positive
association), followed by leucine/isoleucine (Leu/Ile)
(negative association). These relationships were
dramatically influenced by BMI, the resting
respiratory quotient (RQ), T2DM, and sex. Gly had
a strong positive correlation with GDR regardless of
BMI, RQ, or sex but became nonsignificant in T2DM.
In contrast, Leu/Ile was negatively associated with
GDR in nonobese and T2DM subjects. Increased
resting fat metabolism (i.e., low RQ) and obesity
were observed to independently promote and
negate the association between Leu/Ile and insulin
resistance, respectively. Additionally, the
relationship between Leu/Ile and GDR was
magnified in T2DM males. Future studies are
needed to determine whether Gly has a mechanistic
role in glucose homeostasis and whether dietary
Gly enrichment may be an effective intervention in
diseases characterized by insulin resistance.
Diabetes 2014;63:791–800 | DOI: 10.2337/db13-0396
Prevalence rates for type 2 diabetes (T2DM), pre-
diabetes, metabolic syndrome, and cardiovascular
disease have been increasing globally (1) and are re-
sponsible for an increased burden of patient suffering
and social costs. Insulin resistance is integral in the
pathogenesis of these disorders and involves defects in
glucose production by the liver and insulin-stimulated
glucose uptake and utilization by peripheral tissues.
While obesity is associated with insulin resistance, gen-
eral adiposity explains only a minor portion of variability
in insulin resistance among nondiabetic individuals (2,3).
1
Department of Cell, Developmental and Integrative Biology, University of
Alabama, Birmingham, AL
2
Nutrition Sciences, University of Alabama, Birmingham, AL
3
Birmingham Veterans Affairs Medical Center, Birmingham, AL
4
Division of Nutritional Sciences, Cornell University, Ithaca, NY
5
Department of Exercise Science and Sport Management, Kennesaw State
University, Kennesaw, GA
6
Department of Medicine, Duke University, Durham, NC
7
Department of Pharmacology and Cancer Biology, Duke University, Durham, NC
Corresponding author: W. Timothy Garvey, garveyt@uab.edu.
Received 11 March 2013 and accepted 4 October 2013.
This article contains Supplementary Data online at http://diabetes
.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-0396/-/DC1.
A.E.T.-M. and K.H.I. contributed equally to this work.
The opinions expressed are those of the authors and not necessarily those of the
NIH or any other organization with which the authors are affiliated.
© 2014 by the American Diabetes Association. See http://creativecommons
.org/licenses/by-nc-nd/3.0/ for details.
Diabetes Volume 63, February 2014 791
GENETICS/GENOMES/PROTEOMICS/METABOLOMICS
Numerous studies have found that insulin resistance and
diabetes are associated with fat accumulation in the
visceral compartment, skeletal muscle, and liver tissue
(4–7). Recent research suggests that amino acids may
also be important in the development of insulin re-
sistance as alterations in circulating levels of several
amino acids, including branched-chain amino acids
(BCAA) and aromatic amino acids (AAA), are associated
with obesity (8–11) and insulin resistance (8,10) and
identified as the best early predictor for the future de-
velopment of diabetes (12). Moreover, baseline levels of
BCAA, AAA, and related metabolites are prognostic for
improvement in insulin sensitivity in response to a
dietary/behavioral intervention (13) and are tightly
correlated with improvement in glucose homeostasis and
insulin sensitivity after bariatric surgery (14). While
these studies have focused primarily on BCAA and AAA,
other amino acids may also be relevant in the de-
velopment of insulin resistance and T2DM (11,15).
Furthermore, multiple intrinsic factors (e.g., amino acid
metabolism, protein metabolism [16]), hormonal
changes, and extrinsic factors (e.g., dietary intake,
physical activity [11]) can contribute to changes in amino
acid concentrations (17).
Reported relationships between amino acid concen-
trations and insulin resistance in humans have primarily
involved surrogate measures of insulin sensitivity (e.g.,
homeostasis model assessment of insulin resistance
[HOMA-IR]) (10,18,19), which may limit the accuracy of
their predictive value (20). The objective of the current
study was to examine, for the first time, the relationships
between amino acid levels and the gold standard measure
of insulin sensitivity, the hyperinsulinemic-euglycemic
clamp, in human subjects. This technique quantifies
whole-body insulin action under conditions where the
bulk of the insulin-stimulated glucose uptake is into
skeletal muscle (21), which is responsible for the vast
proportion of in vivo glucose uptake in response to in-
sulin. We have confirmed a relationship between BCAA
and insulin resistance and, importantly, have demon-
strated a major signal for glycine (Gly), as well as the
influence of BMI, race, and respiratory quotient (RQ) on
these relationships.
RESEARCH DESIGN AND METHODS
Subjects were recruited from advertisements and word-of-
mouth referrals and sequentially enrolled. An effort was
made to have equal enrollment of European and African
Americans such that only African Americans were entered
into the study after the full complement of European
Americans had been recruited. The final study group
comprised 124 volunteers (63 European American and 60
African American) with ages between 21 and 59 years.
None of the volunteers had cardiovascular, renal, or
hepatic disease, and all were chemically euthyroid. No
subjects were pregnant or taking pharmacological agents
known to affect carbohydrate or lipid metabolism.
Weight was stable (63%) for $3 months before study,
BMI was between 21 and 46 kg/m
2
, and none of the
study subjects engaged in regular exercise. Race was de-
termined by self-report. Premenopausal females were
studied between days 3 and 10 of the menstrual cycle.
Studies were performed in the morning after a 12-h fast.
Subjects were equilibrated on an isocaloric diet with
macronutrient composition of 30% fat, 55% carbohy-
drate, and 15% protein for 3 days prior to studies.
Protocols were approved by the University of Alabama
at Birmingham Institutional Review Board. Written in-
formed consent was obtained from every subject.
Insulin Action
In vivo insulin action was assessed as maximal insulin re-
sponsiveness via hyperinsulinemic-euglycemic glucose
clamp technique at a maximally effective steady-state
serum insulin concentration as previously described
(6,22,23). Briefly, glucose and KPO4 were administered
through a catheter inserted into the brachial vein. A dorsal
hand vein was cannulated in a retrograde manner and kept
in a warming device (65°C) to provide arterialized venous
blood for sampling. Regular insulin (Humulin; Eli Lilly,
Indianapolis, IN) was administered at 200 mU $m
22
$min
21
to produce a mean steady-state insulin concentration of
501 620 mIU/mL. This level is maximally effective for
suppressing hepatic glucose production and has been
shown to predominantly reflect maximally stimulated
skeletal muscle glucose uptake under these experimental
conditions (21). Serum glucose was clamped at 90 mg/dL
for a minimum of 3 h within a ,5% coefficient of variation.
Maximal glucose uptake was determined as the mean glu-
cose infusion rate over the final three 20-min intervals.
Whole-body glucose uptake was calculated as the glucose
infusion rate corrected for changes in the glucose pool size,
assuming a distribution volume of 19% body weight and
a pool fraction of 0.65. Glucose uptake was normalized per
kilogram lean body mass to yield the glucose disposal rate
(GDR). Lower GDR values indicate insulin resistance.
HOMA-IR was calculated from fasting plasma insulin and
glucose levels with the following formula: HOMA-IR =
plasma insulin (mU/mL) 3plasma glucose (mmol/L)/22.5
(24). Higher HOMA-IR values indicate insulin resistance.
Amino Acids Measured by Mass Spectrometry
Fasting serum samples, collected prior to initiating the
clamp procedures, were analyzed for amino acid con-
centrations by flow-injection tandem mass spectrometry
as described (25). Sixteen amino acids were measured
using stable isotope dilution techniques: alanine (Ala),
glycine (Gly), valine (Val), leucine/isoleucine (Leu/Ile),
phenylalanine (Phe), tyrosine (Tyr), glutamate/glutamine
(Glx), aspartate/asparagines (Asx), arginine (Arg), citrul-
line (Cit), histidine (His), methionine (Met), ornithine
(Orn), proline (Pro), and serine (Ser). Sample preparation
methods were performed as previously described (10,25).
Briefly, samples were equilibrated with a cocktail of
792 Amino Acids and Insulin Resistance Diabetes Volume 63, February 2014
internal standards and deproteinated by precipitation
with methanol, and then aliquoted supernatants were
dried and then esterified with hot, acidic n-butanol. The
data were acquired using a Micromass Quattro micro-TM
system equipped with a model 2777 autosampler,
a model 1525 mHPLC solvent delivery system, and
a data system controlled by MassLynx 4.0 operating
system (Waters, Milford, MA).
Anthropometric and Body Composition Measurements
BMI was calculated as weight in kilograms divided by the
square of height in meters. Waist and hip circumferences
were measured using a tension-controlled tape measure.
Dual-energy X-ray absorptiometry (DEXA), using Prodigy
(GE Medical Systems LUNAR, Madison, WI) with software
version 6.10.029 (enCORE 2002), provided total body fat
and lean body mass independent of bone mass (26).
Statistical Analyses
Differences in variables of interest were compared using
univariate ANOVA and reported as mean 6SD. Principle
components analysis was performed to identify
mechanistic-related groupings among the 16 amino
acids. Partial correlations controlled for age, sex, race,
and BMI were used to examine the relationships among
amino acids (including identified components) and insulin
action in the overall population and also stratified by di-
abetes status and BMI. Sensitivity analyses were per-
formed to detect race or sex influence in the correlations.
In analyses stratified by race or sex, these stratification
variables were not used as controlling variables.
Stepwise multiple regression analyses were used to
determine which, if any, amino acids were most pre-
dictive of GDR in the overall cohort, as well as in both
BMI groups and T2DM patients. The most predictive
amino acids revealed in the regression analysis were then
used in additional stepwise multiple regression analyses
to assess the predictability of GDR, along with RQ, BMI,
sex, and race. Missing data were handled by pairwise
deletion. Analyses were performed using SPSS 20.0 for
Windows (SPSS, Chicago, IL), and differences were ac-
cepted as significant at P,0.05.
RESULTS
Descriptive characteristics of study subjects, stratified by
diabetes status and insulin sensitivity, are delineated in
Table 1. As expected, T2DM subjects were most insulin
resistant (IR); however, nondiabetic subjects displayed
a wide variability in insulin responsiveness and were
categorized into insulin-sensitive (IS) and -resistant
subgroups based on values above and below the median
value of GDR. Differences in insulin responsiveness be-
tween T2DM and IR, compared with IS, were further
observed from differences in HOMA-IR. T2DM (vs. IS
and IR) also displayed elevated fasting glucose and re-
duced HDL compared with the other two groups, while
the IS group displayed the lowest fasting glucose and
Table 1—Descriptive characteristics of study subjects
All subjects
Non-T2DM
T2DMIS IR
N124 61 32 31
Race 51% EA, 48% AA 46% EA, 53% AA 47% EA, 53% AA 64% EA, 36% AA
Sex (% male) 41 34 47 48
Age (years) 42 610 41 6940611 45 69
GDR (mg/kg LBM/min) 12.4 64.7 16.1 63.1 9.6 62.0†7.4 62.4†‡
HOMA-IR 4.72 63.71 3.18 61.57 5.51 62.95†6.97 65.68†
Waist (cm) 99 614 94 612 105 614†103 614†
BMI (kg/m
2
) 31.0 65.0 30.2 65.0 33.3 66.0§ 30.4 66.0
Fat (%) 37.6 610.0 37.5 611.0 42.3 68.0§ 32.8 68.0‡
Lean body mass (kg) 52.3 612.0 49.0 610.0 54.0 612.0 57.1 614.0§
REE (kcal/day) 1608 6298 1542 6255 1661 6314 1684 6335
Resting RQ 0.85 60.06 0.86 60.06 0.84 60.07 0.82 60.04†
FFA (mmol) 0.53 60.24 0.51 60.22 0.53 60.16 0.66 60.47
HDL (mg/dL) 46.0 618.7 53 621 43 614§ 35 610†‖
LDL (mg/dL) 119.4 638.2 117 641 122 637 120 636
Fasting insulin (mIU/ml) 17 611 14 6723611†15 615
Fasting glucose (mg/dL) 122 665 90 6997611†214 677†‡
Data are means 6SD unless otherwise indicated. AA, African American; EA, European American; FFA, free fatty acids; REE, resting
energy expenditure. §P,0.05 compared with IS. †P,0.01 compared with IS. ‡P,0.01 compared with IS. ‖P,0.05 compared with IS.
diabetes.diabetesjournals.org Thalacker-Mercer and Associates 793
highest HDL. The mean BMI was similar among IS, IR,
and T2DM subgroups. Waist circumference was lowest in
the IS group.
Compared with IS, IR had reduced levels of Gly, Ser,
and Cit but elevated Glx (Table 2). T2DM had reduced
Gly and His, but elevated Leu/Ile, Val, Asx, and Glx
compared with IS, and elevated Leu/Ile and Asx and re-
duced His levels compared with IR.
Principle components analysis of the 16 amino acids
yielded two extracted components. Component 1 in-
cluded the BCAAs, Leu/Ile and Val (46.4% variance), and
component 2 included Gly and Ser (41.2% variance).
Together, these two components explained 87.6% of the
variance in the data set.
BMI correlated with GDR in the overall population
(r=20.18, P,0.05). Therefore, statistical analyses were
controlled for BMI where appropriate. For the entire
cohort (i.e., IS, IR, and T2DM combined), Gly had the
strongest (positive) correlation with GDR when age, BMI,
sex, and race and were controlled for, while Leu/Ile had
the strongest negative correlation (Fig. 1Aand B). These
relationships with GDR were closely followed by com-
ponents 2 (positive relationship) and 1 (negative re-
lationship [Fig. 1A]). Sensitivity analyses were performed
to explore whether sex, race, or diabetes impacted the
correlations between GDR and individual amino acids
(Gly, Ser, Leu/Ile, and Val) and the two principle com-
ponents (Table 3): while only slight racial differences
were detected, stronger sex differences were revealed.
The relationships between GDR, the amino acids, and
their respective components were strong and significant
in the female population but attenuated in the male
population, with the exception of Gly, which remained
strong in both sexes. The relationship between Leu/Ile
and GDR was intensified in the T2DM males (r=20.726,
P= 0.017) when the data were stratified by sex. It is
important to note, however, that the sample size in this
group is only 13, so this result may not be generalizable
to other populations. The positive relationship between
GDR and Gly was strong in normoglycemic (i.e., IS and
IR) subjects but attenuated in T2DM, while the opposite
phenomenon occurred with Leu/Ile: the relationship was
strong in T2DM but attenuated in normoglycemic sub-
jects. Due to these differences, subsequent analyses
consider T2DM separately, and all analyses are controlled
for sex and race.
For examination of the influence of obesity on the
relationships between GDR and amino acids, subjects
were stratified by BMI and diabetes status and partial
correlation analyses controlled for age, sex, and race were
performed (Fig. 2). Descriptive characteristics and serum
amino acid contents for these groups are provided in
Supplementary Tables 1 and 2, respectively. In nonobese
subjects (BMI ,30 kg/m
2
), Gly and component 2 were
positively correlated to GDR, while Leu/Ile and compo-
nent 1 were negatively related. In obese individuals (BMI
$30 kg/m
2
), Gly, component 1, and six other amino
acids were positively related to GDR, while Leu/Ile and
component 2 were not related. Thus, BMI was an im-
portant determinant as to whether BCAAs were associ-
ated with insulin resistance, while Gly and component 2
remained correlated with insulin action across the BMI
spectrum. Finally, in T2DM subjects, only Leu/Ile was
significantly correlated with GDR (Fig. 2).
For determination of whether relationships between
amino acid levels and GDR were affected by differences
Table 2—Circulating amino acid levels in IS, IR, and T2DM subgroups
Non-T2DM
T2DMIS IR
N61 32 31
Amino acids (mmol/L)
Gly 306.8 693.9 257.0 658.3†246.8 661.8†
Leu/Ile 163.5 636.9 180.8 637.0 204.4 636.0†‡
Ala 350.4 6106.3 350.5 6130.3 357.1 6118.0
Ser 111.8 626.6 98.7 619.7§ 107.3 620.4
Pro 195.1 665.9 189.2 658.3 185.8 657.6
Val 254.7 655.5 271.9 640.9 286.8 652.2§
Met 18.6 64.6 18.2 63.5 16.7 63.8
His 67.4 613.5 66.5 612.1 59.2 610.2†‡
Phe 73.0 614.8 75.2 613.2 75.6 612.4
Tyr 74.5 622.0 83.0 625.3 77.9 628.1
Asx 82.9 645.6 85.7 642.7 121.2 662.1†‡
Glx 83.8 621.7 96.5 621.3§ 103.5 622.0†
Orn 49.9 613.6 50.6 615.6 53.7 612.7
Cit 33.5 69.4 29.0 67.7§ 31.7 68.8
Arg 83.9 624.0 75.9 620.7 79.7 621.9
Data are means 6SD. §P,0.05 compared with IS. †P,0.01 compared with IS. ‡P,0.05 compared with IS.
794 Amino Acids and Insulin Resistance Diabetes Volume 63, February 2014
in fuel preference (i.e., fat vs. carbohydrate oxidation),
correlation analyses were performed after stratification
into subgroups with low and high resting RQ values
(below and above the median RQ value) and with con-
trolling for age, race, sex, and BMI (Fig. 3). A strong,
negative relationship between GDR and Leu/Ile was ob-
served in subjects with an RQ below the median RQ,
while no relationship was detected in those with an RQ
higher than the median RQ. In contrast, positive and
statistically significant associations were observed be-
tween GDR and Gly in both RQ groups.
Our studies have indicated that BMI, resting RQ,
and sex can affect the correlations between amino acid
levels and insulin responsiveness. Therefore, stepwise
multiple regression analyses were performed to determine
the extent to which these factors acted independently to
determine insulin action measured by GDR (Table 4).
In the overall cohort and in the BMI ,30 kg/m
2
subgroup,
only Leu/Ile and Gly entered the regression equation
with statistical significance and exerted independent
effects that predicted the GDR. In the BMI $30 kg/m
2
group, Gly, but not Leu/Ile, in combination with RQ,
sex, and BMI, were independently predictive of GDR.
In the T2DM group, only Leu/Ile was predictive of GDR
(Table 4).
The correlations between HOMA-IR and both Gly (r=
20.211, P= 0.021) and component 2 (r=20.204, P=
0.026) were weaker than the observed correlations with
GDR (above). However, the relationships between
HOMA-IR and Leu/Ile (r= 0.341, P,0.0001) and
component 1 (r= 0.366, P,0.0001) were similar to the
results from the hyperinsulinemic clamp (reported
above).
DISCUSSION
This is the first study to examine the relationship be-
tween circulating amino acids and insulin resistance in
humans using the gold standard measure of whole-body
Figure 1—A: Relationships between insulin sensitivity, measured by hyperinsulinemic-euglycemic clamp, and circulating amino acid
concentrations (N= 120). Correlations are controlled for age, BMI, sex, and race. ■, significant correlation; □, nonsignificant correlations.
B: Scatter plot showing the correlation between insulin sensitivity and amino acids Gly (left) and Leu/Ile (right). ○, normoglycemic subjects;
▼, those with T2DM. During the clamp studies, plasma glucose was clamped at 90 mg/dL in all subjects within a coefficient of variation
<5%. The mean steady-state serum insulin level achieved during the clamps at the indicated infusion rate was 501 620 mU/mL. LBM,
lean body mass.
diabetes.diabetesjournals.org Thalacker-Mercer and Associates 795
insulin action, the hyperinsulinemic-euglycemic clamp
technique. At maximally effective steady-state serum
insulin concentrations, the clamp technique quantifies
whole-body insulin action on glucose uptake with the
bulk of insulin-stimulated glucose uptake occurring into
skeletal muscle (21). Therefore, the current study is the
first to allow an assessment of circulating amino acid
concentrations in relation to insulin action largely in
skeletal muscle, the critical tissue for insulin action
defects that mediate the clinical manifestations of in-
sulin resistance. To test this hypothesis, previous
studies have used obesity (i.e., BMI) or surrogate in-
dices of insulin sensitivity involving mathematical
derivations of fasting glucose and insulin; however,
neither BMI nor these indices of insulin sensitivity
display robust correlations with clamp measures of
insulin responsiveness (20). For example, measures of
general adiposity, such as BMI, only explain ;8% of
Table 3—Impact of diabetes status, sex, and race on the correlations between GDR and amino acids GLY, Leu/Ile, Ser, and Val
n
Gly Leu/Ile Ser Val Component 1‖Component 2¶
rPrPrPrP r P r P
All subjects†120 0.422 0.000 20.344 0.000 0.173 0.064 20.207 0.026 20.324 0.000 0.358 0.000
Normoglycemic†93 0.418 0.000 20.120 0.262 0.296 0.005 20.045 0.678 20.133 0.213 0.403 0.000
T2DM†27 0.106 0.629 20.469 0.024 20.030 0.893 20.259 0.252 20.362 0.090 0.084 0.702
All women‡71 0.484 0.000 20.454 0.000 0.360 0.003 20.330 0.006 20.454 0.000 0.489 0.000
All men‡49 0.308 0.037 20.154 0.308 20.147 0.330 20.020 0.893 20.107 0.478 0.110 0.467
Non-DB women‡57 0.468 0.000 20.251 0.068 0.354 0.009 20.143 0.303 20.267 0.051 0.457 0.001
Non-DB men‡36 0.295 0.095 0.079 0.661 0.135 0.452 0.119 0.511 0.082 0.652 0.250 0.160
T2DM women‡14 0.534 0.091 20.280 0.405 0.474 0.141 0.049 0.886 20.143 0.675 0.654 0.029
T2DM men‡13 20.235 0.513 20.726 0.017 20.237 0.509 20.678 0.031 20.688 0.028 20.182 0.615
All EA§ 59 0.409 0.002 20.430 0.001 0.042 0.759 20.290 0.029 20.392 0.003 0.299 0.024
All AA§ 59 0.482 0.000 20.231 0.086 0.338 0.011 20.115 0.400 20.241 0.073 0.463 0.000
AA, African American; EA, European American; Non-DB, normoglycemic. †Controlled for age, BMI, race, and sex. ‡Controlled for age,
BMI, and race. §Controlled for age, BMI, and sex. ‖Component 1 includes Leu/Ile and Val. ¶Component 2 includes Gly and Ser.
Figure 2—Impact of BMI and T2DM on the relationships between amino acid levels and insulin sensitivity assessed by clamp. x-Axis
values represent correlation coefficients. ■, significant correlation; □, nonsignificant correlations. Component (Comp) 1, Leu/Ile and Val;
Comp 2, Gly and Ser.
796 Amino Acids and Insulin Resistance Diabetes Volume 63, February 2014
individual differences in insulin sensitivity when
assessed by hyperinsulinemic-euglycemic clamp, while
measures of central fat distribution, such as trunk–to–
leg fat ratio, can explain 20% of the variance (27).
Because of the current study design, we have been able
to advance our understanding of amino acid meta-
bolomic profiles and insulin resistance in humans.
Specifically,whilewehaveconfirmed the previous as-
sociation between insulin resistance and elevated
BCAAs, we have now shown that these relationships
canbedramaticallyaffectedbyBMI,RQ,andT2DM.
Importantly, a new observation is the uniquely strong
and persistent correlation between Gly and insulin
action among nondiabetic individuals regardless of
BMI and RQ.
In our cohort, multiple amino acids had strong asso-
ciations with GDR over a broad range of insulin sensi-
tivity. In particular, Gly emerged as the amino acid with
the strongest positive correlation with insulin action and
Leu/Ile with the strongest negative correlation. Fur-
thermore, principle components analysis empirically
identified only two components significantly correlated
with GDR; component 1 with Gly and Ser was positively
associated with GDR and component 2 with the BCAAs,
which was negatively associated. However, the strength
of these associations was again significantly influenced by
BMI, RQ, diagnosed T2DM, and sex.
Gly was found to have a strong positive correlation
with GDR in both lean and obese subgroups and in
subgroups with low and high resting RQ, raising the
Figure 3—Impact of resting RQ on the relationships between amino acid levels and insulin sensitivity assessed by clamp. x-Axis values
represent correlation coefficients. ■, significant correlation; □, nonsignificant correlations. Component (Comp) 1, Leu/Ile and Val; Comp
2, Gly and Ser.
Table 4—Stepwise multiple regression analyses assessing the independent effects of amino acids, RQ, and BMI as predictors
of insulin sensitivity
Group
Stepwise multiple
regression models R
2
SE of the
estimate R
2
change Fchange
Significance
of Fchange
All subjects (N= 120) 1 Leu/Ile 0.177 4.23 0.178 25.4 0.000
2 Leu/Ile, Gly 0.39 3.68 0.208 39.4 0.000
BMI ,30 kg/m
2
(N= 43) 1 Leu/Ile 0.094 4.04 0.094 4.271 0.045
2 Leu/Ile, Gly 0.322 3.54 0.228 13.46 0.001
BMI $30 kg/m
2
(N= 49) 1 Gly 0.276 3.22 0.276 17.902 0.000
2 Gly, RQ 0.365 3.05 0.089 6.456 0.014
3 Gly, RQ, sex 0.423 2.94 0.058 4.542 0.039
4 Gly, RQ, sex, BMI 0.474 2.83 0.051 4.291 0.044
T2DM (N= 27) 1 Leu/Ile 0.234 2.17 0.234 7.626 0.011
diabetes.diabetesjournals.org Thalacker-Mercer and Associates 797
question as to whether Gly was a passive marker or
exerted a causal effect to enhance insulin action. How-
ever, the mechanisms by which Gly interacts with GDR
have yet to be elucidated. Data from studies involving
rodents and cultured cells are consistent with a causal
role. C57BL/6J mice, with diet-induced obesity and
depressed glucose infusion rates, have reduced levels of
Gly (28). Gly administration has been shown to sup-
press proinflammatory adipokines (e.g., tumor necrosis
factor-a, interleukin [IL]-6) and increase adiponectin in
3T3-L1 adipocytes (29,30) and in lean mice (30,31). Ad-
ditionally, in obese mice, Gly suppressed TNF-aand IL-6
gene expression in fat tissue and reduced IL-6, resistin,
and leptin protein levels (29,30). Gly was further found to
improve glucose tolerance in lean, but not obese, mice
(29,30). Gly is also a substrate for glutathione bio-
synthesis, raising the possibility that high Gly could en-
hance antioxidant defense. While speculative, these
findings indicate that Gly could enhance glucose homeo-
stasis and perhaps insulin action by influencing adipose
tissue biology and inflammatory cytokine production, al-
though, again, favorable changes in glucose metabolism in
vivo were only observed in lean and not obese mice. Fur-
ther research is warranted to determine the role of Gly on
insulin sensitivity and inflammation in metabolic
dysfunction.
In humans, the current data are consistent with pre-
vious observations, including a clear decrease in Gly
levels in obese IR subjects compared with lean control
subjects (10), reduced levels in Japanese patients with
metabolic syndrome that were then increased after life-
style modification (11), an increase in Gly levels in re-
sponse to bariatric surgery (14), and decreased Gly levels
in IR offspring of two T2DM parents (15). In addition,
exercise training leading to an increase in insulin sensi-
tivity measured by the frequently sampled intravenous
glucose tolerance test was associated with increments in
Gly and Pro levels (18). While we have demonstrated
a quantitative relationship with GDR values, it remains
unclear in humans based on the current study and
existing literature whether Gly is causally related to
insulin action; nevertheless, the data support a trial
assessing effects of dietary Gly enrichment in IR
patients.
In patients with T2DM, Gly levels were significantly
reduced compared with nondiabetic subjects, and the
positive correlation was weakened and no longer statis-
tically significant. Since insulin resistance in T2DM is
exacerbated by hyperglycemia, with a consequent in-
crease in glucose metabolism via the hexosamine bio-
synthetic pathway (32), it is tempting to speculate that
Gly does not actively participate in, or protect against,
glucose-induced insulin resistance. In cultured adipo-
cytes, the presence of amino acids such as Gly, Thr, and
L-glutamine are permissive for the full expression of in-
sulin resistance induced by high glucose (33). Even so, in
this scenario, the severity of the component of insulin
resistance due to hyperglycemia would not be quantita-
tively related to the plasma Gly level in vivo.
In agreement with previous studies, we were able to
confirm a negative relationship between BCAAs, in-
cluding Leu/Ile and the principal component comprising
both Leu/Ile and Val, and insulin action. In the current
study, this relationship was established using clamp
measures that we assume reflect insulin action in skeletal
muscle. Furthermore, this relationship was modulated by
BMI, resting RQ, and sex. Leu/Ile had a strong negative
correlation with GDR in the nonobese subgroup and in
T2DM patients but not in the nondiabetic obese sub-
group. Thus, the presence of obesity obviated the re-
lationship between BCAAs and insulin action, even
though the obese individuals had higher levels of BCAA
compared with the nonobese subgroup. Interestingly,
observed relationships between BCAA and GDR were
influenced by sex, such that the relationship between
Leu/Ile and GDR was strengthened in the males.
Newgard et al. (10) demonstrated in rodents that
BCAA supplementation of a high-fat diet contributes to
insulin resistance; however, this was not observed when
BCAAs were supplemented into normal chow. This sug-
gests that the availability or preference of fat as a fuel
source may be a driver of the relationship between BCAA
and insulin resistance. To explore this possibility in
humans, we performed analyses in nondiabetic subjects
stratified by low and high resting RQ values. Leu/Ile
correlated with GDR in the low-RQ group, who prefer
oxidation of fat to maintain resting energy expenditure,
but not in subjects with high-RQ, who prefer carbohy-
drates as a fuel source. In T2DM patients, mean RQ was
lower than in the nondiabetic subgroups, and only Leu/
Ile exhibited a significant and negative correlation with
GDR. The data indicate that BCAAs are related to insulin
action only under conditions of high lipid metabolism
whether induced by high-fat feeding in rodents or low
RQ in humans. In the current study, no difference in the
mean RQ value could be detected in the obese versus
nonobese subgroup; therefore, differences in resting fuel
preference could not explain the loss of association be-
tween Leu/Ile and GDR in the obese subjects. In fact, in
multiple regression models, obesity and RQ exerted in-
dependent effects to modulate this relationship.
In previous studies, subjects with metabolic syndrome
(6,8) and obesity (8) have been reported to have elevated
BCAA concentrations. Additionally, infusion of amino
acids during a hyperinsulinemic-euglycemic clamp in-
duced insulin resistance in healthy young males (34).
Overnutrition involving a high-protein diet is also asso-
ciated with insulin resistance (35). Together, these
studies suggest that BCAAs could be causally related to
insulin resistance. Elevated BCAA may result from the
following: decreased BCAA metabolism in adipose tissue
or skeletal muscle; reduced insulin-stimulated, anti-
proteolytic mechanisms within the skeletal muscle; in-
creased dietary intake; decreased physical activity; or
798 Amino Acids and Insulin Resistance Diabetes Volume 63, February 2014
increased autophagy. A potential mechanism for the
observed negative relationship between Leu/Ile and ac-
tion in IR and T2DM subgroups could involve an im-
paired ability of insulin to inhibit skeletal muscle
proteolysis, leading to an increase in BCAA in the skeletal
muscle pool (36). It is unclear, however, why higher Leu/
Ile levels, in obesity, are related to GDR in nonobese but
not in obese humans. One possibility is an adipose tissue
cut point after which the impact of BCAAs on insulin
action is diminished. In any case, the higher levels of
Leu/Ile in the obese appear to exist independent of
changes in insulin action.
While the relationship between Gly and GDR
remained strong when the data were stratified by sex, the
general lack of correlations between GDR and amino
acids in the males is potentially due to fewer males in the
analyses. Support for this inference comes from the ad-
ditional analyses after stratification by sex in the BMI
and T2DM subgroups; however, the relationship between
GDR and Leu/Ile was intensified in the T2DM males.
Future research is warranted to determine whether sex
influences the relationships between amino acids and
GDR and the associated mechanisms.
When HOMA-IR was used as the measure of insulin
sensitivity, the relationship with Gly was attenuated,
while the relationship with Leu/Ile remained strong.
Differences in the magnitude of the identified relation-
ships between amino acids, especially Gly, and markers of
insulin action from the clamp versus surrogate HOMA-IR
measurement may be due to the fact that the maximally
stimulated clamp effectively shuts down hepatic glucose
production and largely reflects insulin action in skeletal
muscle (21), while HOMA reflects both hepatic and
muscle glucose metabolism. The correlation between
GDR and HOMA in our data was 20.461 (P,0.0001),
which is in agreement with a previous study where we
demonstrated that caution is warranted in the in-
terpretation of data using insulin sensitivity indices such
as homeostasis model assessment (20).
In summary, metabolomic amino acid profiles and
hyperinsulinemic clamps performed in nondiabetic and
T2DM individuals over a broad range of GDR and BMI
have demonstrated that 1) the amino acid with the most
robust positive correlation with insulin action is Gly and
strongest negative correlation is Leu/Ile; 2) the associa-
tion between Gly and insulin action remains strong re-
gardless of BMI, RQ, or sex but is weakened and
nonsignificant in T2DM; 3) the relationship between
Leu/Ile and insulin resistance is profoundly influenced by
BMI, fuel metabolism, and sex: Leu/Ile is associated with
insulin resistance in the nonobese and T2DM subjects
only and intensified in T2DM males; and 4) increased
resting fat metabolism (i.e., low RQ) and obesity in-
dependently promote and negate the association be-
tween Leu/Ile and insulin resistance, respectively. While
it is unlikely that amino acid levels are the sole contrib-
utors to the observed differences in GDR, future research
identifying the metabolic disturbances that link Gly and
Leu/Ile with GDR is necessary to fully understand the
pathogenesis of insulin resistance and diabetes. Addi-
tionally, future studies are needed to determine whether
Gly has a mechanistic role in glucose homeostasis and
whether dietary Gly enrichment may be an effective in-
tervention in diseases characterized by insulin resistance.
Acknowledgments. The authors sincerely appreciate the time and
effort put forth by the participants to complete this research.
Funding. This work was supported by grants from the National Institutes of
Health (DK-038765, DK-083562, PO1 HL-55782, and P01 DK58398 to W.T.G.)
and the Merit Review program (to W.T.G.) of the Department of Veterans
Affairs. The authors also acknowledge support from the University of Alabama
at Birmingham (UAB) Center for Clinical and Translational Science (UL1
RR025777), the Nutrition and Obesity Research Center (P30-DK-56336), and
the UAB Diabetes Research and Training Center (P60 DK-079626).
Duality of Interest. No potential conflicts of interest relevant to this
article were reported.
Author Contributions. A.E.T.-M. and K.H.I. wrote the manuscript,
were responsible for statistical analysis, and contributed to the interpretation of
data. F.G. was responsible for statistical analysis and contributed to the
interpretation of data. O.I. was responsible for analyzing the amino acids and
contributed to the interpretation of data. C.B.N. designed the study and
contributed to the interpretation of data. W.T.G. initiated the concept of the
study, designed the study, contributed to the interpretation of data, and wrote
the manuscript. W.T.G. is the guarantor of this work and, as such, had full
access to all the data in the study and takes responsibility for the integrity of
the data and the accuracy of the data analysis.
Prior Presentation. Parts of this study were presented in abstract form
at the Experimental Biology Conference, Boston, MA, 20–24 April 2013.
References
1. Wild S, Roglic G, Green A, Sicree R, King H. Global prevalence of diabetes:
estimates for the year 2000 and projections for 2030. Diabetes Care 2004;
27:1047–1053
2. Lara-Castro C, Newcomer BR, Rowell J, et al. Effects of short-term very
low-calorie diet on intramyocellular lipid and insulin sensitivity in nondia-
betic and type 2 diabetic subjects. Metabolism 2008;57:1–8
3. Reaven GM. The insulin resistance syndrome: definition and dietary ap-
proaches to treatment. Annu Rev Nutr 2005;25:391–406
4. Frayn KN. Adipose tissue and the insulin resistance syndrome. Proc Nutr
Soc 2001;60:375–380
5. Wagenknecht LE, Langefeld CD, Scherzinger AL, et al. Insulin sensitivity,
insulin secretion, and abdominal fat: the Insulin Resistance Atherosclerosis
Study (IRAS) Family Study. Diabetes 2003;52:2490–2496
6. Ingram KH, Lara-Castro C, Gower BA, et al. Intramyocellular lipid and
insulin resistance: differential relationships in European and African
Americans. Obesity (Silver Spring) 2011;19:1469–1475
7. Lara-Castro C, Garvey WT. Intracellular lipid accumulation in liver and
muscle and the insulin resistance syndrome. Endocrinol Metab Clin North
Am 2008;37:841–856
8. Felig P, Marliss E, Cahill GF Jr. Plasma amino acid levels and insulin
secretion in obesity. N Engl J Med 1969;281:811–816
9. Holm G, Björntorp P, Jagenburg R. Carbohydrate, lipid and amino acid
metabolism following physical exercise in man. J Appl Physiol 1978;45:
128–131
diabetes.diabetesjournals.org Thalacker-Mercer and Associates 799
10. Newgard CB, An J, Bain JR, et al. A branched-chain amino acid-related
metabolic signature that differentiates obese and lean humans and con-
tributes to insulin resistance. Cell Metab 2009;9:311–326
11. Kamaura M, Nishijima K, Takahashi M, Ando T, Mizushima S, Tochikubo O.
Lifestyle modification in metabolic syndrome and associated changes in
plasma amino acid profiles. Circ J 2010;74:2434–2440
12. Wang TJ, Larson MG, Vasan RS, et al. Metabolite profiles and the risk of
developing diabetes. Nat Med 2011;17:448–453
13. Shah SH, Crosslin DR, Haynes CS, et al. Branched-chain amino acid levels
are associated with improvement in insulin resistance with weight loss.
Diabetologia 2012;55:321–330
14. Laferrère B, Reilly D, Arias S, et al. Differential metabolic impact of gastric
bypass surgery versus dietary intervention in obese diabetic subjects de-
spite identical weight loss. Sci Transl Med 2011;3:re2
15. Perseghin G, Ghosh S, Gerow K, Shulman GI. Metabolic defects in lean
nondiabetic offspring of NIDDM parents: a cross-sectional study. Diabetes
1997;46:1001–1009
16. Biolo G, Ciocchi B, Lebenstedt M, et al. Short-term bed rest impairs amino
acid-induced protein anabolism in humans. J Physiol 2004;558:381–388
17. Newgard CB. Interplay between lipids and branched-chain amino acids in
development of insulin resistance. Cell Metab 2012;15:606–614
18. Huffman KM, Slentz CA, Bateman LA, et al. Exercise-induced changes in
metabolic intermediates, hormones, and inflammatory markers associated
with improvements in insulin sensitivity. Diabetes Care 2011;34:174–176
19. Tai ES, Tan ML, Stevens RD, et al. Insulin resistance is associated with
a metabolic profile of altered protein metabolism in Chinese and Asian-
Indian men. Diabetologia 2010;53:757–767
20. Pisprasert V, Ingram KH, Lopez-Davila MF, Munoz AJ, Garvey WT. Limi-
tations in the use of indices using glucose and insulin levels to predict
insulin sensitivity: impact of race and gender and superiority of the indices
derived from oral glucose tolerance test in African Americans. Diabetes
Care 2013;36:845–853
21. DeFronzo RA, Jacot E, Jequier E, Maeder E, Wahren J, Felber JP. The
effect of insulin on the disposal of intravenous glucose. Results from in-
direct calorimetry and hepatic and femoral venous catheterization. Dia-
betes 1981;30:1000–1007
22. Garvey WT, Olefsky JM, Griffin J, Hamman RF, Kolterman OG. The effect of
insulin treatment on insulin secretion and insulin action in type II diabetes
mellitus. Diabetes 1985;34:222–234
23. Wu X, Wang J, Cui X, et al. The effect of insulin on expression of genes and
biochemical pathways in human skeletal muscle. Endocrine 2007;31:5–17
24. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner
RC. Homeostasis model assessment: insulin resistance and beta-cell
function from fasting plasma glucose and insulin concentrations in man.
Diabetologia 1985;28:412–419
25. Lien LF, Haqq AM, Arlotto M, et al. The STEDMAN project: biophysical,
biochemical and metabolic effects of a behavioral weight loss inter-
vention during weight loss, maintenance, and regain. OMICS 2009;13:
21–35
26. Paradisi G, Smith L, Burtner C, et al. Dual energy X-ray absorptiometry
assessment of fat mass distribution and its association with the insulin
resistance syndrome. Diabetes Care 1999;22:1310–1317
27. Lara-Castro C, Garvey WT. Diet, insulin resistance, and obesity: zoning in
on data for Atkins dieters living in South Beach. J Clin Endocrinol Metab
2004;89:4197–4205
28. Shearer J, Duggan G, Weljie A, Hittel DS, Wasserman DH, Vogel HJ.
Metabolomic profiling of dietary-induced insulin resistance in the high fat-
fed C57BL/6J mouse. Diabetes Obes Metab 2008;10:950–958
29. Garcia-Macedo R, Sanchez-Muñoz F, Almanza-Perez JC, Duran-Reyes G,
Alarcon-Aguilar F, Cruz M. Glycine increases mRNA adiponectin and di-
minishes pro-inflammatory adipokines expression in 3T3-L1 cells. Eur J
Pharmacol 2008;587:317–321
30. Alarcon-Aguilar FJ, Almanza-Perez J, Blancas G, et al. Glycine regulates
the production of pro-inflammatory cytokines in lean and monosodium
glutamate-obese mice. Eur J Pharmacol 2008;599:152–158
31. Almanza-Perez JC, Alarcon-Aguilar FJ, Blancas-Flores G, et al. Glycine
regulates inflammatory markers modifying the energetic balance through
PPAR and UCP-2. Biomed Pharmacother 2010;64:534–540
32. Marshall S, Garvey WT, Traxinger RR. New insights into the metabolic
regulation of insulin action and insulin resistance: role of glucose and
amino acids. FASEB J 1991;5:3031–3036
33. Traxinger RR, Marshall S. Role of amino acids in modulating glucose-
induced desensitization of the glucose transport system. J Biol Chem
1989;264:20910–20916
34. Krebs M, Krssak M, Bernroider E, et al. Mechanism of amino acid-
induced skeletal muscle insulin resistance in humans. Diabetes 2002;
51:599–605
35. Um SH, D’Alessio D, Thomas G. Nutrient overload, insulin resistance, and
ribosomal protein S6 kinase 1, S6K1. Cell Metab 2006;3:393–402
36. Luzi L, Castellino P, DeFronzo RA. Insulin and hyperaminoacidemia regu-
late by a different mechanism leucine turnover and oxidation in obesity.
Am J Physiol 1996;270:E273–E281
800 Amino Acids and Insulin Resistance Diabetes Volume 63, February 2014