Original Article

Natural Antibiotics and Insulin Sensitivity

The Role of Bactericidal/Permeability-Increasing Protein
Carme Gubern,1 Abel Lopez-Bermejo,1 Josefina Biarnes,1 Joan Vendrell,2 Wifredo Ricart,1
and Jose Manuel Fernandez-Real1

The innate immune system can immediately respond to

microorganism intrusion by helping to prevent further
invasion. Bactericidal/permeability-increasing protein
(BPI) is a major constituent of neutrophils that possesses
anti-inflammatory properties. Inflammation is increasingly
recognized as a component of the metabolic syndrome. We
hypothesized that the production of BPI could be linked to
insulin sensitivity and glucose tolerance. We studied circulating BPI across categories of glucose tolerance. We also
studied whether these cross-sectional associations were of
functional importance. For this reason, we investigated
circulating bioactive lipopolysaccharide and the effects of
changing insulin actionafter treatment with an insulin
sensitizer (metformin) on circulating BPI in subjects
with glucose intolerance. Finally, we tested whether a
3-untranslated region (UTR) BPI polymorphism led to
differences in BPI and insulin action among nondiabetic
subjects. Age- and BMI-adjusted circulating BPI was significantly lower among patients with type 2 diabetes. Circulating BPI correlated negatively with fasting and postload
glucose and insulin concentrations. In subjects with glucose intolerance, BPI was also linked to BMI, waist-to-hip
ratio, and age- and BMI-adjusted insulin sensitivity. Bioactive lipopolysaccharide was negatively correlated with circulating BPI (r 0.57, P < 0.0001) and positively with
plasma lipopolysaccharide-binding protein (r 0.54, P
0.002). In parallel to improved insulin sensitivity, plasma
BPI significantly increased in the metformin group but not
in the placebo group. A 3-UTR BPI polymorphism was
simultaneously associated with plasma BPI concentration,
waist-to-hip ratio, fasting and postload insulin concentration, fasting plasma triglycerides, and insulin sensitivity.
These findings suggest that this component of the innate
immune system is associated with metabolic pathways.
Diabetes 55:216 224, 2006

From the 1Section of Diabetes, Endocrinology and Nutrition, Institut

dInvestigacio Biomedica de Girona, Girona, Spain; and the 2Research Unit,
University Hospital of Tarragona Joan XXIII, Institut Pere Virgili, Tarragona,
Spain.
Address correspondence and reprint requests to J.M. Fernandez-Real, MD,
PhD, Unit of Diabetes, Endocrinology and Nutrition, Hospital de Girona Dr.
Josep Trueta, Ctra. Franca s/n, 17007 Girona, Spain. E-mail: uden.
jmfernandezreal@htrueta.scs.es.
Received for publication 25 August 2005 and accepted in revised form 6
October 2005.
BPI, bactericidal/permeability-increasing protein; CIGMA, continuous infusion of glucose with model assessment; ELISA, enzyme-linked immunosorbent assay; HOMA, homeostasis model assessment; LBP, lipopolysaccharidebinding protein; sTNFR, soluble fraction of tumor necrosis factor- receptor;
UTR, untranslated region.
2006 by the American Diabetes Association.
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance
with 18 U.S.C. Section 1734 solely to indicate this fact.

216

he first lines of defense against invasion by

microbial agents are the physical barriers represented by the skin and mucosas. These physical
barriers are susceptible to injuries that allow the
entry of opportunistic microbial agents. The innate immune system can immediately respond to this intrusion by
helping to prevent further invasion. This immune response
includes phagocytosis by neutrophils and macrophages
and their production of reactive oxygen intermediates that
kill microbial agents (1).
A number of endogenous antimicrobial proteins produced by the neutrophils have been shown to play an
integral part in innate immunity. Bactericidal/permeability-increasing protein (BPI) is a major constituent of neutrophils (from 0.5 to 1% of total protein) (2). This singlechain cationic protein (molecular weight 55 kDa) is
located mainly in the primary granules but is also found on
the neutrophil surface (3). Additionally, BPI has been
found on the surface of monocytes, in eosinophils, and as
a product of mucosal and skin epithelial cells (3,4). In
addition to its well-known antimicrobial function against
gram-negative bacteria, BPI also possesses anti-inflammatory properties (3).
Impressive evidence has accumulated over the past
decade that the metabolic syndrome is linked to inflammatory pathways (59). The molecular mechanisms leading to development of the metabolic syndrome (the
clustering of central obesity, alterations of glucose and
lipid metabolism, and arterial hypertension) are not fully
understood. However, it has become clear that chronic
subclinical inflammation, increased levels of inflammation-sensitive plasma proteins, and alterations in the function of the innate immune system are intrinsic to the
metabolic syndrome (59). Insulin resistance is central to
the pathophysiology of all these alterations, which run
together with the accumulation of fat and the presence of
specific genetic components of the inflammatory cascade
(9).
By virtue of its anti-infective and neutralizing properties,
BPI is a proximal anti-inflammatory effector. BPIs crystal
structure reveals a boomerang-shaped bipartite molecule
that includes a cationic lysine-rich NH2-terminal half containing the antibacterial and lipopolysaccharide (endotoxin)-neutralizing activities of the molecule and a COOHterminal half that contributes to the opsonic activity of BPI
(10). Plasma lipopolysaccharide-binding protein (LBP) is a
product of the liver that has a net anionic charge and
serves to greatly amplify responses to lipopolysaccharide
DIABETES, VOL. 55, JANUARY 2006

C. GUBERN AND ASSOCIATES

by delivering lipopolysaccharide monomers to a monocyte

receptor complex containing CD14, MD2, and Toll-like
receptor 4 (11,12).
BPI competes with LBP for the binding of endotoxin,
but BPI-lipopolysaccharide complexes (in contrast to LBP
and lipopolysaccharide) do not activate macrophages and
other lipopolysaccharide-responsive host cells. Thus, BPI
and LBP are functionally antagonistic (3). In fact, the
affinity of LBP for lipopolysaccharide is 70-fold lower
than that of BPI and may explain why LBP does not exhibit
any antibacterial activity (3).
We and others have previously reported that subtle
deficiencies in proteins of the innate immune system
(soluble CD14 and mannose-binding lectin) were associated with alterations of glucose metabolism and atherosclerosis. These deficiencies lead to enhancement of the
inflammatory cascade, resulting in a worsening of insulin
resistance (13,14). We thus hypothesized that the production of BPI could be linked to insulin sensitivity and
glucose tolerance. To this aim we studied circulating BPI
in healthy subjects, in patients with glucose intolerance,
and in patients with type 2 diabetes. We also studied
whether these cross-sectional associations were of functional importance. For this reason, we investigated whether
circulating bioactive lipopolysaccharide was linked to BPI
in healthy volunteers. We also aimed to evaluate the
effects of changing insulin actionafter treatment with an
insulin sensitizer (metformin) on circulating BPI in subjects with glucose intolerance. Finally, we hypothesized
that there were differences in insulin sensitivity according
to the capacity of BPI production. To this end, we tested
whether the rs1131847 BPI polymorphism at the 3untranslated region (UTR) led to differences in BPI and
insulin action among nondiabetic subjects.
RESEARCH DESIGN AND METHODS
We recruited and studied 174 Caucasian subjects, including analysis of
glucose tolerance and insulin sensitivity, within an ongoing study on nonclas
sical cardiovascular risk factors. All normotolerant subjects (n 114) had
fasting plasma glucose 7.0 mmol/l and 2-h postload plasma glucose 7.8
mmol/l after a 75-g oral glucose tolerance test. Glucose intolerance was
diagnosed in 60 subjects according to American Diabetes Association (ADA)
criteria (postload glucose between 7.8 and 11.1 mmol/l). Inclusion criteria
were 1) BMI 40 kg/m2, 2) absence of systemic disease, and 3) absence of
infection within the previous month. None of the control subjects were under
medication or had evidence of metabolic disease other than obesity. Alcohol
and caffeine were withheld within 12 h of performing the insulin sensitivity
test. Smokers were defined as any person consuming at least one cigarette a
day in the previous 6 months. Resting blood pressure was measured as
previously reported (14). Liver disease and thyroid dysfunction were specifically excluded by biochemical workup.
Patients with type 2 diabetes, according to American Diabetes Association
criteria, were prospectively recruited from diabetes outpatient clinics based
on stable metabolic control in the previous 6 months, as defined by stable
HbA1c (A1C) values. Exclusion criteria included the following: 1) clinically
significant hepatic, neurological, endocrinologic, or other major systemic
disease, including malignancy; 2) history or current clinical evidence of
hemochromatosis; 3) history of drug or alcohol abuse, defined as 80 g/day in
men and 40 g/day in women, or serum transaminase activity more than twice
the upper limit of normal; 4) elevated serum creatinine concentration; 5)
acute major cardiovascular event in the previous 6 months; 6) acute illness
and current evidence of acute or chronic inflammatory or infective disease;
and 7) mental illness rendering the subject unable to understand the nature,
scope, and possible consequences of the study. Pharmacological treatment for
these patients was as follows: insulin: 44.8%; oral hypoglycemic agents: 72.9%;
statins: 38.0%; fibrates: 10.6%; blood pressurelowering agents: 61.5%; aspirin:
42.7%; and allopurinol: 4.2%. All subjects gave written informed consent after
the purpose of the study was explained to them. The institutional review
board of the institution approved the protocol.
DIABETES, VOL. 55, JANUARY 2006

Anthropometric and clinical measurements. BMI was calculated as the

weight (in kilograms) divided by the square of height (in meters). The
subjects waists were measured with a soft tape midway between the lowest
rib and the iliac crest. The hip circumference was measured at the widest part
of the gluteal region. The waist-to-hip ratio was then calculated. Fat mass and
percent fat mass were calculated using bioelectric impedance (body composition analyzer; Holtain, Crosswell, Crymych, Dyfed, Wales, U.K.).
Study of insulin sensitivity. Insulin sensitivity and glucose effectiveness
were measured using the frequently sampled intravenous glucose tolerance
test. In brief, the experimental protocol started between 8:00 and 8:30 A.M.
after an overnight fast. A butterfly needle was inserted into an antecubital
vein, and patency was maintained with a slow saline drip. Basal blood samples
were drawn at 30, 10, and 5 min, after which glucose (300 mg/kg body
wt) was injected over 1 min starting at time 0, and insulin (0.03 units/kg;
Actrapid, Novo Nordisk, Denmark) was administered at time 20. Additional
samples were obtained from a contralateral antecubital vein up to 180 min, as
previously described (14).
Functional studies. BPI in association with bioactive lipopolysaccharide. Lipopolysaccharide was measured by a limulus amebocyte lysate test
(Pyrochrome; Cape Cod, Falmouth, MA), as recommended by the manufacturer. In brief, plasma was collected in nonpyrogenic EDTA tubes and frozen
at 20C until assay. All procedures were performed under nonpyrogenic
conditions. Plasma was diluted 1:20 or 1:40 and heat inactivated at 75C for 10
min. The reaction was read using kinetics, i.e., measuring the time to reach a
given absorbance at 405 nm. Recovery of spiked lipopolysaccharide was
between 50 and 200%. Sensitivity of the assay was 0.005 Ehrlich units/ml (0.5
pg/ml).
Study of the effects of an insulin sensitizer (metformin) on BPI. Patients
were recruited from the Diabetes and Endocrinology Unit at Hospital de
Girona Dr. Josep Trueta. They were offered participation in the study if they
were between 30 and 65 years old and fulfilled all the inclusion criteria. Men
and women were both included, but women were included only when surgical
sterility was documented, when they were postmenopausal, or when a reliable
method of contraception was used. The inclusion criteria were as follows: 1)
BMI between 22 and 35 kg/m2, 2) impaired glucose tolerance by an oral
glucose tolerance test performed 2 months before the beginning of the study
or a fasting glucose level between 6.11 and 7.8 mmol/l, and 3) stability of diet
and physical exercise within the past 2 months. All subjects signed a written
informed consent. Exclusion criteria were 1) type 2 diabetes by American
Diabetes Association criteria; 2) pregnant or nursing women; 3) patients with
renal impairment defined as plasma creatinine values 1.5 mg/dl for men and
1.4 mg/dl for women; 4) patients affected by cardiac or respiratory insufficiency likely to cause central hypoxia or reduced peripheral perfusion; 5) past
history of lactic acidosis; 6) noncontrolled hypertension; 7) acute or chronic
infection; 8) liver disease, including alcoholic liver disease as demonstrated by
abnormal liver-function tests or alcohol abuse; 9) patients taking drugs that
could modify glucose tolerance; 10) participants in another clinical trial
within the last 30 days; and 11) legal incapacity as a study patient.
The clinical study was performed in accordance with the Declaration of
Helsinki (revised version, Hong Kong, 1989) as well as with the European
Community Note for Guidance on Good Clinical Practice for Studies on
Medicinal Products in the European Community. It was approved by the local
ethics committee and the Spanish health department (clinical assay no.
97/337).
We randomized 31 patients either to placebo or metformin, using randomization tablets (Lipha), according to the computer program Rancode 3.1. A
dietitian with the aim of assuring a stable weight gave general diabetic dietary
advice at the beginning of the study. At 2 months after the first visit (visit 2),
subjects started taking metformin or placebo, one tablet per day (850 mg) for
the 1st week and then two tablets per day (one after breakfast and one after
dinner) for the next 11 weeks. Drug compliance was checked by tablet counts,
and any side effect was recorded at visit 3 (6 weeks after randomization) and
at visit 4 (at the end of the study).
Insulin sensitivity was evaluated by homeostasis model assessment
(HOMA), using basal (mean of three samples obtained at 5-min intervals)
glucose and insulin (15,16). -Cell function was calculated by continuous
infusion of glucose with model assessment (CIGMA) from achieved C-peptide
and glucose values (17). The CIGMA test consists of a continuous intravenous
infusion of 5 mg glucose kg ideal body wt1 min1, using a 10 g/dl glucose
solution with model assessment of glucose, insulin, and C-peptide (radioimmunoassay; Byk-Sangtec Diagnostica, Dietzenbach, Germany) values obtained before (basal value, mean value of three samples obtained at 5-min
intervals) and at the end (achieved value, mean value of samples obtained at
50, 55, and 60 min) of the test. C-peptide detection level was 0.1 ng/ml and had
intra- and interassay coefficients of variation (CVs) of 2.6 and 4.4%, respectively. It shows 25% cross-reactivity with proinsulin but not with insulin.
217

BPI AND INSULIN ACTION

TABLE 1
Anthropometrical and biochemical variables of study subjects

n
Age (years)
Weight (kg)
BMI (kg/m2)
Systolic blood pressure (mmHg)
Diastolic blood pressure (mmHg)
Waist-to-hip ratio
Fat-free mass (kg)
Fat mass (kg)
Fasting insulin (mU/l)
Fasting glucose (mmol/l)
A1C (%)
HDL cholesterol (mg/dl)
Triglycerides (mg/dl)
Insulin sensitivity (104 mU/l)
Glucose effectiveness
BPI (ng/ml)
LBP (g/ml)

Normotolerant men

Glucose-intolerant
men

Type 2 diabetic
men

P*

114
46.2 11.9
78 12.1
27.01 3.6
124 14.5
78.3 10.7
0.92 0.06
72.1 9.9
6.1 (2.613.2)
8.23 4.1
5.15 0.4
4.78 0.34
52.8 12.1
97 54.6
2.72 (1.954.39)
0.020 (0.0160.024)
14.22 (6.628.5)
27.82 (9.850.1)

60
53.2 11.2
84.4 12
29.5 3.9
132.8 15.9
83.4 9.8
0.95 0.06
71.8 9.2
10.22 (5.2418.27)
11.8 5.7
5.9 1.0
5.00 0.58
51.1 10.7
121.3 63.2
1.32 (0.772.24)
0.018 (0.0140.020)
17.66 (8.631)
17.26 (9.5944.93)

170
57.2 11.8
85.5 19
32.3 7.0
140.0 21.5
81.2 10.8

10.0 4.7
7.3 1.7
50.5 16.4
223.6 142.0

10.64 (6.019.1)
65.03 (57.972.08)

0.0001
0.002
0.0001
0.004
0.8
0.009
0.0001
0.002
0.0001
0.0001
0.003
0.3
0.015
0.0001
0.01
0.51
0.47

Data are means SD or median (interquartile range). *P values for the comparison between normotolerant and glucose-intolerant men; P
0.0001, compared with normotolerant men.
Analytical methods. Serum glucose concentrations were measured in duplicate by the glucose oxidase method, using a Glucose Analyzer II (Beckman
Instruments, Brea, CA). Total serum cholesterol was measured through the
reaction of cholesterol esterase with cholesterol oxidase and peroxidase.
Total serum triglycerides were measured through the reaction of glycerolphosphate-oxidase and peroxidase. Uric acid was measured by routine
laboratory tests. A1C was measured by high-performance liquid chromatography (Bio-Rad, Muenchen, Germany) and a Jokoh HS-10 autoanalyzer. Intraand interassay CVs were 4% for all of these tests.
Serum insulin levels were measured in duplicate by monoclonal immunoradiometric assay or enzyme-amplified sensitivity immunoassay (Medgenix
Diagnostics, Fleunes, Belgium). Intra- and interassay CVs were similar to
those previously reported (14).
Enzyme-linked immunosorbent assay of BPI and LBP. Plasma EDTA BPI
concentrations were measured by a sandwich enzyme-linked immunosorbent
assay (ELISA; human BPI ELISA kit; HyCult Biotechnology, Uden, the
Netherlands) according to the manufacturers instructions. The assay has a
sensitivity of 250 pg/ml. Intra- and interassay CVs were 5%. Serum LBP levels
were determined with a commercially available human LBP ELISA kit (HyCult
Biotechnology, Uden, the Netherlands). Serum samples were diluted at least
1,000 times and assayed according to the manufacturers instructions. The
assay has a sensitivity of 1 ng/ml and a measurable concentration range of
0.8 50 ng/ml. Intra- and interassay CVs were between 5 and 10%.
Study of the effects of a 3-UTR BPI gene polymorphism on circulating
BPI and insulin action among nondiabetic subjects. Genomic DNA was
extracted from peripheral blood leukocytes according to standard procedures
(QIAamp DNA Blood Mini Kit; Qiagen, Hilden, Germany).
For the detection of the polymorphism rs1131847 (NCBI [National Center
for Biotechnology Information]), G-to-A transition at 3-UTR, we used a
method based on TaqMan technology suitable for allelic discrimination (ABI
Prism 7000 sequence detection system; Applied Biosystems, Darmstadt,
Germany). The samples were genotyped with an Applied Biosystems Taqman
assay (assay-on-demand C 308491 1 ), using minor-groove binding reporter
probes (VIC-labeled for the A allele and FAM-labeled for the G) and an
end-read protocol. The PCR conditions were as recommended by the manufacturer, and a sample containing water instead of DNA, as a negative control,
was used for each PCR run.
Statistical methods. Descriptive results of continuous variables are the
means SD. Before statistical analysis, normal distribution and homogeneity
of the variances were evaluated using Levenes test, and then variables were
given a log transformation if necessary. These parameters (insulin sensitivity
index [Si], glucose effectiveness index [SG], triglycerides, LBP, and BPI) were
analyzed on a log scale and tested for significance on that scale. The
antilog-transformed values of the means (geometric mean) are reported in
the Tables. Relationships between variables were tested using Pearsons test
and stepwise multiple linear regression analysis. We used 2 test for comparisons of proportions and unpaired t tests for comparisons of quantitative
218

variables. A general linear model for repeated measures with Bonferronis

correction was used to compare BPI levels pre- and posttreatment (metformin
study). A general linear model was also used to calculate circulating BPI
values after adjusting for age and BMI. The statistical analyses were performed using the SPSS program (version 11.0).

RESULTS

Comparison of circulating BPI across categories of

glucose tolerance. BPI was significantly lower among
patients with type 2 diabetes (Table 1). Median (95% CI)
age- and BMI-adjusted BPI values were 11.42 (9.24 14.12)
in type 2 diabetic patients, 16.98 (12.722.6) in subjects
with glucose intolerance, and 17.53 (14.3521.47) in control subjects (P 0.006 type 2 diabetes versus control, P
0.028 type 2 diabetes versus glucose intolerance). For any
given medication, BPI concentrations did not differ between treated and untreated patients.
Metabolic studies in nondiabetic subjects. Characteristics of the subjects and the comparisons with type 2
diabetic subjects are shown in Table 1. Subjects with
glucose intolerance were significantly older, heavier, and
showed lower insulin sensitivity and glucose effectiveness
than normotolerant subjects. Circulating BPI correlated
negatively with LBP (r 0.31, P 0.0001). This relationship was especially significant among normotolerant
subjects (r 0.35, P 0.0001) but not in subjects with
glucose intolerance (r 0.18, P 0.16).
When all nondiabetic subjects were considered as a
whole, circulating BPI was significantly associated with
age (Table 2). This was caused by the older age and higher
plasma BPI levels of glucose-intolerant subjects. BPI correlated negatively with fasting and postload glucose, fasting and postload insulin, and A1C. After stratifying by
glucose tolerance, all of these associations were observed
only among subjects with glucose intolerance (Fig. 1), in
whom BPI was also linked to BMI, waist-to-hip ratio, and
to age- and BMI-adjusted insulin sensitivity (Tables 2 and
3 and Fig. 2A). Circulating BPI was also positively associated with HDL cholesterol and soluble fraction of tumor
necrosis factor- receptor 2 (sTNFR2), and these associaDIABETES, VOL. 55, JANUARY 2006

C. GUBERN AND ASSOCIATES

TABLE 2
Correlationships between circulating BPI, LBP, and clinical variables in nondiabetic subjects

n
Age
BPI
LBP
BMI
BPI
LBP
Waist-to-hip ratio
BPI
LBP
Fat mass
BPI
LBP
Fat-free mass
BPI
LBP
Systolic blood pressure (mmHg)
BPI
LBP
Diastolic blood pressure (mmHg)
BPI
LBP

All subjects

Normotolerant
subjects

Glucose-intolerant
subjects

174

114

60

0.19
0.22

0.02
0.007

0.14
0.29

NS
0.005

0.30
0.12

0.02
NS

0.13
0.21

0.07
0.005

0.07
0.13

NS
0.1

0.26
0.40

0.04
0.002

0.12
0.12

0.1
0.1

0.04
0.04

NS
NS

0.39
0.20

0.002
0.1

0.12
0.10

0.1
NS

0.04
0.19

NS
NS

0.24
0.04

0.08
NS

0.03
0.19

NS
0.04

0.006
0.17

NS
0.1

0.04
0.17

NS
NS

0.04
0.02

NS
NS

0.05
0.008

NS
NS

0.03
0.15

NS
NS

0.08
0.06

NS
NS

tions were attributable to the findings in normotolerant

subjects (Table 3). BPI was also positively associated with
white blood cell and neutrophil counts (Table 3).
LBP was negatively associated with age but only among

FIG. 1. Relationship between fasting insulin (A) and A1C (B) with
plasma BPI concentration among subjects with glucose intolerance.
DIABETES, VOL. 55, JANUARY 2006

0.10
0.10

NS
NS

0.005
0.30

NS
0.02

normotolerant subjects (Table 2). As a possible reflection

of its source, LBP was positively associated with fat-free
mass and C-reactive protein (Tables 2 and 3). The remaining findings regarding plasma LBP were specular to those
with BPI. Plasma LBP was positively associated with
several components of the metabolic syndrome, such as
BMI and diastolic blood pressure (Table 2), fasting and
postload glucose concentrations, fasting insulin, A1C, and
fasting triglycerides (Table 3) and negatively with insulin
sensitivity (Table 3 and Fig. 2B) in subjects with glucose
intolerance. In addition, LBP was negatively associated
with glucose effectiveness when all subjects were considered as a whole (Table 3).
We performed a multiple linear regression analysis to
predict circulating BPI. We considered as independent
variables those with significant association on univariant
analysis. When all subjects were considered as a whole,
sTNFR2 (P 0.003), neutrophil count (P 0.003), and
HDL cholesterol (P 0.01) contributed to 19% of BPI
variance (7, 6, and 6%, respectively) after controlling for
the effects of age, BMI, waist-to-hip ratio, fasting glucose,
and insulin sensitivity. Regarding LBP, fat-free mass alone
contributed to 11% of LBP variance (P 0.02).
Functional studies. BPI in association with bioactive
lipopolysaccharide. Bioactive lipopolysaccharide was
negatively correlated with circulating BPI (r 0.57, P
0.0001) (Fig. 3) and positively with LBP (r 0.54, P
0.002).
Effects of insulin sensitizer (metformin) on BPI. Of
118 preselected subjects, 31 were randomized for the
study (16 to metformin and 15 to placebo). Six patients
(three on metformin and three on placebo) withdrew
early, and two noncompliant patients were excluded (both
in metformin group). Three patients in the metformin
group were excluded because no plasma EDTA was
available for BPI measurement. We finally analyzed 20
patients (8 on metformin and 12 on placebo).
Table 4 shows the baseline characteristics of the ana219

BPI AND INSULIN ACTION

TABLE 3
Correlationships between circulating BPI, LBP, and biochemical variables in nondiabetic subjects

n
Fasting glucose
BPI
LBP
30 OGTT
BPI
LBP
60 OGTT
BPI
LBP
90 OGTT
BPI
LBP
120 OGTT
BPI
LBP
A1C
BPI
LBP
Fasting insulin
BPI
LBP
120 OGTT
insulin
BPI
LBP
Triglycerides
BPI
LBP
HDL cholesterol
BPI
LBP
C-reactive protein
BPI
LBP
sTNFR1
BPI
LBP
sTNFR2
BPI
LBP
Si
BPI
LBP
SG
BPI
LBP
White blood cell count
BPI
LBP
Neutrophil count
BPI
LBP
Eosinophil count
BPI
LBP

All subjects

Normotolerant
subjects

Glucose-intolerant
subjects

174

114

60

0.28
0.14

0.001
0.09

0.05
0.01

NS
NS

0.47
0.37

0.0001
0.006

0.05
0.02

NS
NS

0.06
0.03

NS
NS

0.10
0.30

NS
0.03

0.10
0.02

NS
NS

0.05
0.07

NS
NS

0.23
0.43

0.07
0.001

0.10
0.09

NS
NS

0.09
0.11

NS
NS

0.25
0.47

0.05
0.0001

0.20
0.21

0.02
0.018

0.16
0.08

0.1
NS

0.29
0.47

0.02
0.0001

0.19
0.17

0.03
0.05

0.11
0.12

NS
NS

0.39
0.35

0.009
0.002

0.24
0.10

0.003
NS

0.04
0.03

NS
NS

0.41
0.30

0.001
0.02

0.17
0.11

0.04
0.1

0.13
0.03

NS
NS

0.15
0.24

NS
0.08

0.03
0.11

NS
0.1

0.08
0.04

NS
NS

0.32
0.31

0.01
0.02

0.16
0.05

0.03
NS

0.19
0.007

0.03
NS

0.10
0.21

NS
0.1

NS
0.001

0.03
0.27

NS
0.004

0.05
0.33

NS
0.01

NS
NS

0.14
0.002

NS
NS

0.01
0.32

NS
0.02

0.36
0.23

0.0001
0.02

0.20
0.03

NS
NS

0.01
0.26
0.09
0.10
0.34
0.16

0.001
0.07

0.07
0.02

NS
NS

0.05
0.19

NS
0.015

0.02
0.05

NS
NS

0.31
0.40

0.015
0.002

0.06
0.16

NS
0.07

0.06
0.23

NS
0.07

0.21
0.12

0.01
NS

0.17
0.07

0.1
NS

0.28
0.20

0.02
NS

0.25
0.06

0.002
NS

0.24
0.02

0.02
NS

0.30
0.11

0.02
NS

0.05
0.07

NS
NS

0.13
0.34

NS
0.01

0.08
0.17

NS
0.04

OGTT, oral glucose tolerance test (30, 60, 90, and 120 indicate the minutes after glucose intake).

lyzed patients. They were of similar age, BMI, and sex.

Fasting glucose, insulin, and C-peptide levels were similar.
Insulin sensitivity and -cell function were also not significantly different. After a 12-week treatment, fasting and 2-h
glucose decreased in the metformin group (P 0.003 and
220

P 0.045, respectively). Fasting insulin (P 0.005),

fasting C-peptide (P 0.01), achieved insulin (P 0.013),
and achieved C-peptide (P 0.02) levels were also statistically lower after metformin treatment (Table 4). Insulin
sensitivity by HOMA (P 0.0001) improved after metDIABETES, VOL. 55, JANUARY 2006

C. GUBERN AND ASSOCIATES

BPI polymorphism and insulin sensitivity. Given the

associations between BPI and BMI, we screened 373
control subjects. GG homozygotes and A/ carriers
showed no significant difference in BMI (27.5 3.4 vs.
28.12 4.1). In a sample of 174 consecutive subjects, we
performed a more detailed study concerning insulin sensitivity. Interestingly, GG homozygotes showed significantly lower waist-to-hip ratio, lower fasting and postload
insulin concentration, lower plasma triglycerides, and
higher Si and plasma BPI concentration than A/ subjects
(Table 5, Fig. 4).
DISCUSSION

FIG. 2. Relationship between plasma BPI concentration (A) and LBP

(B) with insulin sensitivity among subjects with glucose intolerance.

formin treatment and was not modified by placebo. -Cell

function was not modified in either group (Table 4). In
parallel to improved insulin sensitivity, plasma BPI significantly increased only in the metformin group in a general
linear model for repeated measures with Bonferronis
correction (P 0.023) (Table 4).
Given that we have used different methods to assess the
degree of insulin resistance in the subjects studied, we
performed a Bland-Altman plot between log-HOMA and
log-Si derived insulin sensitivity and between log-HOMA
and log-CIGMA. The comparisons showed that the methods did not differ significantly.

According to previous reports, an underlying activation of

the immune system leads to a higher white blood cell
count, mainly neutrophil count, and to insulin resistance in
apparently healthy subjects (18). In this study we substantiated this finding at the molecular level, providing evidence that this particular defense against infection (BPI)
runs in parallel to insulin sensitivity among healthy subjects. First, we found that circulating BPI concentration
was significantly different across categories of glucose
tolerance: BPI was significantly lower in patients with type
2 diabetes. In subjects with glucose intolerance, we found
the strongest associations between plasma BPI and central
obesity, glucose metabolism, insulin sensitivity, and components of the metabolic syndrome. Interestingly, the
associations between these components and LBP were
specular to those of BPI.
Second, we tested the functional significance of these
findings. Bioactive lipopolysaccharide was significantly
associated with both BPI and LBP. In fact, the latter two
proteins were also negatively associated. These findings
suggest that, with decreasing BPI, the ability to buffer
lipopolysaccharide is impaired, leading to increased LBP
synthesis by the liver. Again, the specular relationships of
BPI and LBP with several anthropometrical and metabolic
variables support their antagonistic function.
Third, we further substantiated our hypothesis by studying the effects of an insulin sensitizer, metformin, on
circulating BPI. In patients receiving metformin, but not in
those receiving placebo, we observed improved insulin
sensitivity and raised circulating BPI concomitantly.
How can all of these associations be explained? Insulin
action may lead to increased plasma BPI concentration. A
recent study has demonstrated that insulin is a strong

FIG. 3. Relationship between bioactive lipopolysaccharide and plasma BPI concentration in healthy subjects.
DIABETES, VOL. 55, JANUARY 2006

221

BPI AND INSULIN ACTION

TABLE 4
Study of the effects of an insulin sensitizer (metformin) on circulating BPI in subjects with glucose intolerance
Metformin
Sex (M/F)
Age (years)
BMI (kg/m2)
Waist circumference
Waist-to-hip ratio
Fasting glucose (mmol/l)
120 OGTT glucose (mmol/l)
Fasting insulin (mU/l)
Fasting C-peptide (ng/ml)
Achieved glucose (mmol/l)
Achieved insulin (mU/l)
Achieved C-peptide (ng/ml)
HOMA sensitivity (%)
CIGMA -cell function (%)
BPI (ng/ml)

Placebo

Baseline

12 weeks

Baseline

12 weeks

4/4
46.6 5.2
26.7 4.1
92.8 12.3
0.96 0.1
6.13 0.63
9.06 1.48
11.5 5.5
2.4 0.7
10.71 1.09
30.9 14.8
5.3 1.3
37.8 16.2
83.5 24.6
2.75 (1.267.73)

26.2 3.8
93.2 10.7
0.92 0.07
5.41 1*
7.27 2.66
7.4 2.9*
1.7 0.5
9.73 2.09
22.2 7.4
4.1 1.1
57.7 23.0*
87.9 33.6
7.49 (2.5638.8)

5/7
46.5 6.7
28.6 3.9
95.5 10.4
0.91 0.07
5.93 0.49
7.23 1.72
11.2 3.3
2.1 0.7
10.82 0.95
27.7 11.3
4.4 1.4
35.4 10.1
71.7 27.2
3.24 (1.585.3)

28.4 3.8
93.9 10.6
0.92 0.07
5.90 0.36
7.75 1.6
11.5 3.3
2.0 0.8
10.05 0.77
31.3 14.2
4.5 1.5
34.6 9.9
83.1 30.0
2.34 (1.068.2)

Data are n, means SD, or median (interquartile range). Achieved glucose, insulin, and C-peptide refers to these parameters after
intravenous glucose (see RESEARCH DESIGN AND METHODS). *P 0.01; P 0.05; P 0.023 (within group). OGTT, oral glucose tolerance test
(120 indicate the minutes after glucose intake).

regulator of the main neutrophil functions in nondiabetic

healthy subjects (19). The cellular functions of human
neutrophils, including bactericidal activity, require energy derived from glucose. Although insulin does not
stimulate hexose transport in this immune cell, previous
reports have clearly shown that this hormone is able to
regulate glucose metabolism in neutrophils (20,21). Impaired neutrophil function would lead to decreased BPI
exocytosis and impaired ability to buffer circulating
lipopolysaccharide. The consequence would be increased liver C-reactive protein and LPB synthesis and,
through enhancement of the inflammatory cascade, decreased insulin action in a vicious cycle. Metformin would
help to reverse this situation.

Finally, we found that the 3-UTR BPI gene polymorphism studied was associated with both increased BPI and
raised insulin sensitivity concomitantly. This polymorphism may encompass mRNA destabilizing signals, thus
affecting mRNA stability and leading to a reduction of BPI
abundance in polymorphonuclear leukocytes. In this
sense, the sequence of events would be precipitated in
these individuals with poor ability to produce BPI. Interestingly, in animal models, the insulin signaling pathway
modulates both inherent longetivity and pathogen resistance, increasing resistance to infection, to affect overall
survival (22). Nevertheless, these results should be interpreted with caution, given the relatively small sample size.
Furthermore, the BPI gene polymorphism may possibly

TABLE 5
BPI 3-UTR polymorphism and insulin sensitivity in nondiabetic subjects

Men
Normotolerant/glucose intolerant
Age (years)
Weight (kg)
BMI (kg/m2)
Waist-to-hip ratio
Fat-free mass (kg)
Fat mass (kg)
Systolic blood pressure (mmHg)
Diastolic blood pressure (mmHg)
Fasting glucose (mmol/l)
Postload glucose 120 (mmol/l)
Fasting insulin (mU/l)
Postload insulin 30 (mU/l)
Postload insulin 120 (mU/l)
A1C (%)
HDL cholesterol (mg/dl)
Triglycerides (mg/dl)
Insulin sensitivity (104 mU/l)
Glucose effectiveness
BPI (ng/ml)
LBP (g/ml)

A/ subjects

GG homozygotes

115
73/42
49.9 12.3
80.8 13.1
28.12 4.1
0.94 0.07
71.6 9.8
7.0 (2.615.1)
127 15.6
79.8 10.7
5.43 0.76
7.52 2.60
10.73 5.7
79.9 50.2
57 (30.492.7)
4.86 0.47
51.6 10.9
116.8 75.2
1.98 (1.13.05)
0.019 (0.0150.024)
17.4 (7.830.7)
20.23 (8.543.3)

59
41/18
47.0 12.2
79.1 10.4
27.5 3.4
0.91 0.07
71.7 8.9
8.5 (3.9714.66)
125.5 14.2
79.8 10.1
5.46 0.63
6.88 2.20
8.26 3.6
58.7 32.7
41.2 (24.673.3)
4.73 0.44
53.0 13.1
94.1 54.2
2.56 (1.744.29)
0.019 (0.0160.023)
23.85 (10.1 44.4)
14.04 (9.65 41.3)

0.45
0.14
0.42
0.33
0.03
0.98
0.69
0.44
0.99
0.82
0.12
0.002
0.002
0.04
0.07
0.47
0.026
0.008
0.79
0.01
0.88

Data are n, means SD, or median (interquartile range).

222

DIABETES, VOL. 55, JANUARY 2006

C. GUBERN AND ASSOCIATES

FIG. 4. Insulin sensitivity according to 3-UTR BPI gene

polymorphism in nondiabetic subjects.

represent a marker for another susceptibility gene in this

region that could be the basis for the observed associations. In this sense, MODY1 (23) and agouti (24) genes are
close to this region.
We cannot exclude that BPI could lead to increased
insulin action. The selective and potent action of BPI
against gram-negative bacteria and their lipopolysaccharides is fully manifest in biological fluids, including
plasma, serum, and whole blood (25). In multiple animal
models of gram-negative sepsis and/or endotoxemia, administration of BPI congeners is associated with improved
outcome (26). Recombinant NH2-terminal proteins derived
from BPI have demonstrated potent antiendotoxic activity
in phase II/III trials (2729). BPI has been demonstrated to
improve hyperglycemia and other metabolic events after
lipopolysaccharide administration in animal models (30).
In addition, the administration of erythromycin, and of
other members of this family of antibiotics, has been
reported to increase both insulin action (31) and BPI
exocytosis (32). Insulin secretion and BPI exocytosis
might be interrelated events reflecting common cellular
pathways. In addition, BPI is capable of inhibiting all of the
many proinflammatory activities of lipopolysaccharide,
including induction of cytokine release (27,33). As a
reflection of its source, plasma BPI was significantly
associated with neutrophil count. However, BPI was not
significantly associated with eosinophil count, possibly
reflecting the minor relative importance of this cellular
type.
In summary, our findings suggest that components of
innate immunity, such as BPI and LBP, are not only linked
to inflammatory pathways but also seem to be associated
with insulin action. These molecules seem particularly
important among patients with glucose intolerance and
type 2 diabetes.
ACKNOWLEDGMENTS

This work was supported by research grants from the

Ministerio de Educacion y Ciencia (BFU2004-03654) and
the Instituto de Salud Carlos III (RCMN C03/08, RGDM
G03/212, and RGTO G03/028).
A.L.-B. is an investigator of the Fund for Scientific
Research Ramon y Cajal (Ministerio of Education and
Science, Spain). We thank Maria Garcia for her help in
statistical analyses.
DIABETES, VOL. 55, JANUARY 2006

REFERENCES
1. Fearon DT, Locksley RM: The instructive role of innate immunity in the
acquired immune response. Science 272:50 53, 1996
2. Weiss J, Olsson I: Cellular and subcellular localization of the bactericidal/
permeability-increasing protein of neutrophils. Blood 69:652 659, 1987
3. Elsbach P: The bactericidal/permeability-increasing protein (BPI) in antibacterial host defense. J Leukoc Biol 64:14 18, 1998
4. Levy O, Canny G, Serhan CN, Colgan SP: Expression of BPI (bactericidal/
permeability-increasing protein) in human mucosal epithelia. Biochem Soc
Trans 31:795 800, 2003
5. Ridker PM, Buring JE, Cook NR, Rifai N: C-reactive protein, the metabolic
syndrome, and risk of incident cardiovascular events: an 8-year follow-up
of 14,719 initially healthy American women. Circulation 107:391397, 2003
6. Festa A, DAgostino R Jr, Howard G, Mykkanen L, Tracy RP, Haffner SM:
Chronic subclinical inflammation as part of the insulin resistance syndrome: the Insulin Resistance Atherosclerosis Study (IRAS). Circulation
102:42 47, 2000
7. Engstrom G, Hedblad B, Stavenow L, Lind P, Janzon L, Lindgarde F:
Inflammation-sensitive plasma proteins are associated with future weight
gain. Diabetes 52:20972101, 2003
8. Schmidt MI, Duncan BB, Sharrett AR, Lindberg G, Savage PJ, Offenbacher
S, Azambuja MI, Tracy RP, Heiss G: Markers of inflammation and prediction of diabetes mellitus in adults (Atherosclerosis Risk in Communities
study): a cohort study. Lancet 353:1649 1652, 1999
9. Fernandez-Real JM, Ricart W: Insulin resistance and chronic cardiovascular inflammatory syndrome. Endocr Rev 24:278 301, 2003
10. Beamer LJ, Carroll SF, Eisenberg D: Crystal structure of human BPI and
two bound phospholipids at 2.4 angstrom resolution. Science 276:1861
1864, 1997
11. Fenton MJ, Golenbock DT: LPS-binding proteins and receptors. J Leukoc
Biol 64:2532, 1998
12. Ulevitch RJ, Tobias PS: Recognition of gram-negative bacteria and endotoxin by the innate immune system. Curr Opin Immunol 11:19 22, 1999
13. Megia A, Gallart L, Fernandez-Real JM, Vendrell J, Simon I, Gutierrez C,
Richart C: Mannose-binding lectin gene polymorphism is associated with
gestational diabetes mellitus. J Clin Endocrinol Metab 89:50815087, 2004
14. Fernandez-Real JM, Broch M, Richart C, Vendrell J, Lopez-Bermejo A,
Ricart W: CD14 monocyte receptor, involved in the inflammatory cascade,
and insulin sensitivity. J Clin Endocrinol Metab 88:1780 1784, 2003
15. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner
RC: Homeostasis model assessment: insulin resistance and B-cell function
from fasting plasma glucose and insulin concentrations in man. Diabetologia 28:412 419, 1985
16. Bonora E, Targher G, Alberiche M, Bonadonna RC, Saggiani F, Zenere MB,
Monauni T, Muggeo M: Homeostasis model assessment closely mirrors the
glucose clamp technique in the assessment of insulin sensitivity: studies in
subjects with various degrees of glucose tolerance and insulin sensitivity.
Diabetes Care 23:57 63, 2000
17. Hosker JP, Matthews DR, Rudenski AS, Burnett MA, Darling P, Bown EG,
Turner RC: Continuous infusion of glucose with model assessment:
measurement of insulin resistance and B-cell function in man. Diabetologia 28:401 411, 1985
223

BPI AND INSULIN ACTION

18. Vozarova B, Weyer C, Lindsay RS, Pratley RE, Bogardus C, Tataranni PA:
High white blood cell count is associated with a worsening of insulin
sensitivity and predicts the development of type 2 diabetes. Diabetes
51:455 461, 2002
19. Walrand S, Guillet C, Boirie Y, Vasson MP: In vivo evidences that insulin
regulates human polymorphonuclear neutrophil functions. J Leukoc Biol
76:1104 1110, 2004
20. Furukawa S, Saito H, Matsuda T, Inoue T, Fukatsu K, Han I, Ikeda S,
Hidemura A, Muto T: Relative effects of glucose and glutamine on reactive
oxygen intermediate production by neutrophils. Shock 13:274 278, 2000
21. Munroe JF, Shipp JC: Glucose metabolism in leucocytes from patients with
diabetes mellitus, with and without hypercholesteremia. Diabetes 14:584
590, 1965
22. Garsin DA, Villanueva JM, Begun J, Kim DH, Sifri CD, Calderwood SB,
Ruvkun G, Ausubel FM: Long-lived C. elegans daf-2 mutants are resistant
to bacterial pathogens. Science 300:1921, 2003
23. Stoffel M, Le Beau MM, Espinosa R 3rd, Bohlander SF, Le Paslier D, Cohen
D, Xiang KS, Cox NJ, Fajans SS, Bell CI: A yeast artificial chromosomebased map of the region of chromosome 20 containing the diabetessusceptibility gene, MODY1, and a myeloid leukaemia related gene. Proc
Natl Acad Sci U S A 93:39373941, 1996
24. Kwon HY, Bultman SJ, Loffler C, Chen WJ, Furdon PJ, Powell JG, Usala AL,
Wilkison W, Hansmann I, Woychik RP: Molecular structure and chromosomal mapping of the human homolog of the agouti gene. Proc Natl Acad
Sci U S A 91:9760 9764, 1994
25. Weiss J, Elsbach P, Shu C, Castillo J, Grinna L, Horwitz A, Theofan G:
Human bactericidal/permeability-increasing protein and a recombinant
NH2-terminal fragment cause killing of serum-resistant gram-negative
bacteria in whole blood and inhibit tumor necrosis factor release induced
by the bacteria. J Clin Invest 90:11221130, 1992
26. Evans TJ, Carpenter A, Moyes D, Martin R, Cohen J: Protective effects of
a recombinant amino-terminal fragment of human bactericidal/permeabil-

224

ity-increasing protein in an animal model of gram-negative sepsis. J Infect

Dis 171:153160, 1995
27. von der Mohlen MA, van Deventer SJ, Levi M, van den Ende B, Wedel NI,
Nelson BJ, Friedmann N, ten Cate JW: Inhibition of endotoxin-induced
activation of the coagulation and fibrinolytic pathways using a recombinant endotoxin-binding protein (rBPI23). Blood 85:34373443, 1995
28. Giroir BP, Quint PA, Barton P, Kirsch EA, Kitchen L, Goldstein B, Nelson
BJ, Wedel NJ, Carroll SF, Scannon PJ: Preliminary evaluation of recombinant amino-terminal fragment of human bactericidal/permeability-increasing protein in children with severe meningococcal sepsis. Lancet 350:
1439 1443, 1997
29. Levin M, Quint PA, Goldstein B, Barton P, Bradley JS, Shemie SD, Yeh T,
Kim SS, Cafaro DP, Scannon PJ, Giroir BP: Recombinant bactericidal/
permeability-increasing protein (rBPI21) as adjunctive treatment for children with severe meningococcal sepsis: a randomised trial: rBPI21
Meningococcal Sepsis Study Group. Lancet 356:961967, 2000
30. Lin Y, Kohn FR, Kung AH, Ammons WS: Protective effect of a recombinant
fragment of bactericidal/permeability increasing protein against carbohydrate dyshomeostasis and tumor necrosis factor-alpha elevation in rat
endotoxemia. Biochem Pharmacol 47:15531559, 1994
31. Ueno N, Inui A, Asakawa A, Takao F, Tani S, Komatsu Y, Itoh Z, Kasuga M:
Erythromycin improves glycaemic control in patients with type II diabetes
mellitus. Diabetologia 43:411 415, 2000
32. Schultz MJ, Speelman P, Hack CE, Buurman WA, Van Deventer SJ, Van Der
Poll T: Intravenous infusion of erythromycin inhibits CXC chemokine
production, but augments neutrophil degranulation in whole blood stimulated with Streptococcus pneumoniae. J Antimicrob Chemother 46:235
240, 2000
33. Von der Mohlen MA, Kimmings AN, Wedel NI, Mevissen ML, Jansen J,
Friedmann N, Lorenz TJ, Nelson BJ, White ML, Bauer R: Inhibition of
endotoxin-induced cytokine release and neutrophil activation in humans
by use of recombinant bactericidal/permeability-increasing protein. J Infect Dis 172:144 151, 1995

DIABETES, VOL. 55, JANUARY 2006