www.elsevier.com/locate/jim

Journal of Immunological Methods 284 (2004) 99–106

Direct comparison of traditional ELISAs and membrane protein

arrays for detection and quantification of human cytokines

S. Copeland, J. Siddiqui, D. Remick*

Department of Pathology, University of Michigan Medical School, M2210 Med. Sci. I, 1301 Catherine Road, Ann Arbor, MI 48109-0602, USA

Received 22 July 2003; received in revised form 14 October 2003; accepted 14 October 2003

Abstract

Many labs wish to measure cytokines in an accurate, reproducible, and rapid manner. An antibody-based membrane array

for measuring cytokines has been developed based on the same technology as the traditional ELISA. The aim of this study was

to compare results obtained with the traditional ELISA method with those from the membrane array technology, a form of low-

cost proteomics. Diluted human whole blood was stimulated with live bacteria (Escherichia coli, or Staphylococous aureus), or

LPS and cytokines were measured both by ELISA and the membrane protein array. Of the 16 cytokines measured via ELISA,

only IFN-g was below detection level. The other 15 cytokines were present in concentrations up to several thousand picograms/

ml. Of the 20 cytokines measured via membrane protein array, only 3 could be detected (IL-6, IL-8 and MIP-1h). Additionally,the membrane protein array did not detect TNF-a from the LPS-stimulated blood. These results indicate that the low-cost

membrane protein array may lack sufficient sensitivity to adequately detect cytokines levels in complex biological fluids such as

human plasma.

D 2003 Elsevier B.V. All rights reserved.

Keywords: Blood; Proteomics; Inflammation; Multiplex assay

1. Introduction bound proteins are subsequently detected and quanti-

With the advent of genomic profiling, the improve-

ments in protein detection have been driven to decode

unknown relationships between disease patterns and

protein profiles. One of the frontiers for profiling lies

in the development of low-cost proteomics, including

protein-detecting micorarrays. Typically, a protein-

detecting microarray consists of numerous affinity

reagents arrayed on a solid surface. A complex mixture

of proteins binds to its specific target protein, and those

0022-1759/$ - see front matter D 2003 Elsevier B.V. All rights reserved.

doi:10.1016/j.jim.2003.10.011

* Corresponding author. Tel.: +1-734-763-6454; fax: +1-734-

763-6476.

E-mail address: remickd@umich.edu (D. Remick).

fied (MacBeath, 2002). These micorarrays may be

used to detect multiple low-abundance proteins from

one sample, such as cytokines. However, the measure-

ment of low abundance proteins is often fraught with

problems such as the high cost of dedicated equipment

and the specificity of the affinity reagents (MacBeath,

2002). Furthermore, there has not been the develop-

ment of a high-throughput technology for isolation of

highly specific protein ligands, lending to difficulty in

the ability to distinguish between different forms of the

same protein (Cahill, 2001). Additional problems exist

with analysis of conventional proteomic approaches,

which employ two-dimensional polyacrylamide gel

electrophoresis coupled with mass spectrometry;

S. Copeland et al. / Journal of Immunological Methods 284 (2004) 99–106100

therefore, absolute quantitation is difficult (Kodadek,

2001). In this technology-driven atmosphere for

change, new technologies have attempted to overcome

these limitations.

Recently, Schweitzer et al. (2002) described cou-

pling of isothermal rolling-circle amplification (RCA)

to antibodies for the simultaneous detection of 75

cytokines. RCA has allowed the development of highly

sensitive and multiplexed microarrays. Through the

RCA-amplified signals located on the microarrayed

spot, detection of protein can be seen at the zeptomolar

concentration (Schweitzer et al., 2002). In this appli-

cation, supernatants from human dendritic cells ex-

posed to LPS and/or TNFa were analyzed via RCA.

Results, confirmed through standard ELISA, revealed

LPS and TNF-a-induced 16 and 12 cytokines, respec-

tively (Schweitzer et al., 2002). However, the sophis-

ticated technology required to develop the RCA limits

its broad application, and this restriction has led to the

development of alternative, less expensive methods for

quantitation. Onemethod for human cytokine detection

was reported by Huang et al. (2001). This technology

employs a microspotted nitrocellulose membrane-

based array developed using enhanced chemilumines-

cence (Huang et al., 2001). This type of antibody-based

membrane array is based on the same type of technol-

ogy as the conventional ELISA. The antibody-spotted

membranes are incubated with biological samples, and

the bound proteins are recognized by biotin-conjugated

antibodies (Huang et al., 2001). RayBiok Human

Cytokine Array, the first membrane-based human cy-

tokine protein array commercially available, proposed

to have the ability to identify protein expression pro-

files on spotted arrays. However, there are no definitive

results correlating this form of low-cost proteomics

with the conventional method for measuring human

cytokines. Therefore, the aim of this study was to

compare the traditional ELISA method with the

RayBiok Human Cytokine Array for analysis of

cytokine production in stimulated human whole blood.

2. Materials and methods

2.1. Basic experimental design

Whole blood (10 ml) was collected from five

normal human volunteers. Results from donor four

were excluded from all statistical analyses due to

abnormally high baseline cytokine concentrations

(i.e. before any stimulation). Blood was drawn and

anticoagulated with 10 Al/ml heparin, and diluted

1:3 with Dulbecco’s modified Eagle’s medium

(DMEM; Bio-Whittaker, Walkersville, MD), for a

total volume of 12 ml. An initial sample (3 ml) of

unstimulated blood was collected for the Time 0

(T0) cytokine measurements. The experimental

groups consisted of the following: 1�107 colony-

forming units (CFUs) of Methicillin-resistant Staph-

ylococcus aureus (MRSA), 2� 107 CFUs of Escher-

ichia coli, 3� 10 ng/ml lipopolysaccharide (LPS), 4.

100 Al DMEM. Based on extensive and successful

therapeutic use of the antibiotic imipenem in a sepsis

mouse model, it was included in the preparation of

all whole blood stimulations. Imipenem is bacteri-

cidal resulting in the inhibition of cell wall synthesis

(Balfour et al., 1996). Imipenem has in vitro activity

against a wide range of Gram-positive and Gram-

negative organisms (Balfour et al., 1996).

2.2. Bacteria and antibiotics

2.2.1. Methicillin-resistant S. aureus (MRSA)

The initial isolate was obtained from a patient at

the University of Michigan. Organisms were stored

at � 20 jC and prior to use were subcultured on

blood agar plates. One colony was suspended in 3

ml of LB broth base (Gibco, Life Technologies,

Rockville, MD) and the suspension was incubated

for 16 h at 37 jC. Bacteria were quantitated via

spectrophotometry using the 0.5 McFarland Stan-

dard (an absorbance of 0.075 at 670 nm= 1.5� 108

CFU/ml).

2.2.2. E. coli (ATCC strain 25922)

The organisms were stored at � 20 jC and prior

to use were subcultured on blood agar plates. One

colony was suspended in 3 ml of LB broth base

(Gibco, Life Technologies) and the suspension was

incubated for 16 h at 37 jC. Bacteria were quanti-

tated via spectrophotometry using the 0.5 McFarland

Standard.

2.2.3. Imipenem

(PrimaxinR I.V., Merck) was diluted in sterile

saline for stock concentration of 10 Ag/ml. The final

Table 1

Cytokines position on custom membrane protein array

a b c d e F g h

1 Pos Pos Neg Neg Eotaxin GCSF GM-CSF GRO

2 Pos Pos Neg Neg Eotaxin GCSF GM-CSF GRO

3 IFN-y IL-10 IL-12 IL-1b IL-1ra IL-2 IL-4 IL-6

4 IFN-y IL-10 IL-12 IL-1b IL-1ra IL-2 IL-4 IL-6

5 IL-8 IP-10 MCP-1 MIG MIP-1a MIP-1b RANTES TNF-a

6 IL-8 IP-10 MCP-1 MIG MIP-1a MIP-1b RANTES TNF-a

7 Blank Blank Blank Blank Blank Blank Blank Pos

8 Blank Blank Blank Blank Blank Blank Blank Pos

Custom RayBio Human Cytokines Array. The name and location of

each cytokine/chemokine custom spotted for this set of experiments

are listed. A total of 20 cytokines were placed on the array. Note the

set of six positive controls, four in the upper left-hand corner and

two in the lower right-hand corner.

S. Copeland et al. / Journal of Immunological Methods 284 (2004) 99–106 101

concentration of imipenem in each whole blood

experimental group was 833 ng/ml.

2.3. Blood collection and preparation

Blood was collected by venipuncture by using a

pyrogen-free collecting system and anti-coagulated

with heparin (Elkins-Sinn, Cherry Hill, NJ; final

concentration, 10 IU/ml of blood). Four milliliters of

whole blood was diluted 1:3 with DMEM for a final

volume of 12 ml. The diluted blood was then ali-

quoted 3� 4 ml. To this diluted blood, 107 CFU of

MRSA or E. coli was added, followed immediately by

the addition of imipenem. For the negative control,

DMEM was added without the bacteria and/or anti-

biotics. For a positive control, LPS (10 ng/ml, E. coli

O111:B4) was added without any antibiotics or bac-

teria. The tubes were incubated at 37 jC for 24 h in a

CO2 incubator with continuous mixing. The blood

was centrifuged at 2000� g for 5 min, plasma har-

vested, and stored at � 70 jC until cytokine measure-

ments were performed.

2.4. Enzyme linked immunosorbent assay (ELISA, the

gold standard)

All cytokine ELISAs were measured using

matched antibody pairs (R&D Systems, Minneapolis,

MN) using 96-well plates (Nunc Immunoplate Max-

isorb, Neptune, NJ). As previously described (Nem-

zek et al., 2001), plates were blocked and incubated

at room temperature for 1 h on a rotator. Samples

were added (50 Al/well) in duplicate and incubated

for 1 or 2 h at room temperature on a rotator. The

samples were diluted at least 1:10 prior to assay, and

in some situations up to 1:500 in order to keep the

results on the scale of the standard curve. Biotiny-

lated antibodies were added (50 Al/well) and plates

were incubated for 1 or 2 h at room temperature on a

rotator. A half-hour incubation with streptavidin–

horseradish peroxidase (HRP) at a dilution of

1:20,000 was followed by detection with TMB

(3,3V,5,5V-tetramethylbenzidine (Sigma T-2885) in

DMSO (Sigma D-2650) dissolved to 1% in 0.1 M

citric acid and 0.1 M sodium acetate pH= 6 and

0.016% H2O2 for 30 min. The reaction was stopped

by addition of 1.5 M H2SO4. Plates were read using

dual wavelengths (465 and 590 nm) on a Biotek

microplate reader (Biotek Instruments, Winooski,

VT) and total cytokine concentrations were calculat-

ed using the standard curve prepared from recombi-

nant cytokines. The lower limit of detection for the

cytokines based on the standard curves ranged from

1 to 3 pg/ml.

2.5. Antibody-based protein array system

A Custom RayBio Human Cytokine Array kit

was purchased from RayBiotech (catalog no. H010;

Norcross, GA). RayBiotech custom spotted the cus-

tom membrane per our request of 20 specific cyto-

kines. The layout of the membrane is listed in Table

1. The directions were followed precisely as stated in

the directions from the manufacturer. Briefly, mem-

branes were placed in an eight well tissue culture

plate and washed twice with Tris-buffered saline

(TBS), and incubated with 2 ml 1� Blocking Buffer

at room temperature for 30 min. 1 ml of sample (10-

fold diluted with Blocking Buffer) was added and

incubated for 1 h. After decanting samples, all

membranes were washed 3 times with 2 ml of 1�Wash Buffer I at room temperature, 5 min per wash,

followed by washing two times with 1� Wash

Buffer II at room temperature, 5 min per wash.

Eight milliliters of 1:500 diluted biotin-conjugated

antibodies were prepared, and incubated at room

temperature for 1 h, and washing steps were repeated

as before. Sixteen milliliters of 1:20,000 diluted

HRP-conjugated streptavidin was added and mem-

Table 2

Coefficient of variation for the six positive spots on membrane

arrays

S. Copeland et al. / Journal of Immunological Methods 284 (2004) 99–106102

branes incubated at room temperature for 30 min,

followed by additional washing. Spots were visual-

ized using enhanced chemiluminescence (ECL,

Fig. 1. Custom RayBio Human Cytokines Array. A total of 20

cytokines were placed on the array. Representative membrane

protein arrays incubated with E. coli-stimulated blood (panel A),

MRSA-stimulated blood (panel B), and LPS-stimulated blood

(panel C). Four spots in the upper left and two lower right corners of

the membranes indicate the positive controls.

Individual membranes Average

1 2 3 4

E. coli 42 14 31 3.4 23

MRSA 14 33 45 34 32

LPS 6.4 8.2 6.0 16 9.2

Time 0 35 37 42 5.1 30

Coefficient of variation of positive controls for membrane arrays.

Expressed as a percentage, the coefficient of variation (CV) was

calculated from the mean and standard deviation of the actual

pixel densities of the 6 positive controls. The table represents

the CV for the four individual membranes and also the average

CV.

Western Blotting Analysis System, Amersham Phar-

macia Biotech; RPN2109). The ECL solution was

loaded onto the membranes to cover the entire

surface (approximately 250 Al) and incubated for 1

min. Membranes were wrapped in Saran Wrap and

exposed to Kodak X-Omat radiographic film for 1

min, then processed. Each film was scanned (Hewlett

Packard ScanJet 5300C) into Scion Image for Win-

dows (Scion, National Institute of Health, USA) and

spots were digitized into pixel densities. The densi-

ties were exported into Excel and the background

intensity was subtracted prior to analysis.

Fig. 2. Cytokines detected via membrane protein array. Repre-

sented values are actual pixel densities as measured via Scion

Image. The data represents the meansF S.E.M. for four donors.

Specific values and statistical results are listed in Tables 2, 3b,

and 4b.

Table 3a

Pro-inflammatory cytokines detected via ELISA

E. coli (pg/mlF S.E.M.) MRSA (pg/mlF S.E.M.) LPS (pg/mlF S.E.M.) T0 (pg/mlF S.E.M.)

IL-6 140000F 20000* 50000F 17000 110000F 17000* 200F 30

IL-8 40000F 4000* 30000F 5000* 6000F 2000 50F 3

IL-1h 8000F 2000* 5000F 1000* 2000F 500 BDL

TNF-a 6000F 1000* 4000F 1000* 2000F 200 BDL

IL-12 100F 40* BDL BDL BDL

IFN-g BDL BDL BDL BDL

Pro-inflammatory cytokines detected via ELISA. With the exception of IFN-g, ELISA results indicated that experimental treatments produced

detectable amounts of pro-inflammatory cytokines. The lower limit of detection for IL-6 after considering the dilution of the plasma was 150 pg/

ml, IL-8, IL-1h, and TNF-a was 50 pg/ml, and IL-12 was 5 pg/ml. The data represents the meansF S.E.M. for four donors. BDL= below

detection limit of the ELISA.

S. Copeland et al. / Journal of Immunological Methods 284 (2004) 99–106 103

2.6. Statistical analysis

Values were expressed as meansF S.E.M. of the

experiment. Data were analyzed using an ANOVA

with Tukey’s post test and P < 0.05 were considered

statistically significant.

*P< 0.05 compared to Time 0.

3. Results

3.1. Traditional ELISA (the gold standard)

Literally thousands of peer-reviewed publications

have used traditional ELISAs to measure human

cytokines and at this time they should be considered

the gold standard for detection and quantification of

Table 3b

Pro-inflammatory cytokines detected via membrane protein array

E. coli

(pixelsFS.E.M.)

MRSA

(pixelsFS.E.M.)

LPS

(pixelsFS.E.M.)

T0

(pixelsFS.E.M.)

IL-6 49F 19 28F 11 125F 36 0

IL-8 132F 11* 85F 20* 163F 5* 0

IL-1h BDL BDL BDL BDL

TNF-a BDL BDL BDL BDL

IL-12 BDL BDL BDL BDL

IFN-g BDL BDL BDL BDL

Pro-inflammatory cytokines detected via the membrane array. The

membrane array detected IL-6 and IL-8 for all experimental groups.

IL-8 densities were significantly higher in all experimental groups

when compared to Time 0. However, only IL-6 and IL-8 were

detected using the membrane array and no other pro-inflammatory

cytokines were detected. The data represents the meansF S.E.M.

for four donors. BDL= below detection limit of the membrane array.

*P< 0.05 compared to Time 0.

cytokines. A total of 16 cytokines were measured

using our traditional, previously published protocols

(Nemzek et al., 2001).

3.2. Custom RayBio Human Cytokines Array

The map of the cytokines specially spotted on our

membranes is listed in Table 1. Representative exam-

ples of the membrane protein arrays are shown in Fig.

1 (A, E. coli; B, MRSA; C, LPS). Positive controls are

located in the upper left-hand corner (four spots) and

lower right-hand corner (two spots) of each mem-

brane. The coefficient of variation (CV) was calculat-

ed for the pixel densities of the six-spot control for

each membrane. The individual and average %CV for

the membranes is depicted in Table 2. Awide range of

CV% was a result of the partial development of the

six-spot control on many of the membranes and

extended from 6% to 42%. The membranes were only

Table 4a

Anti-inflammatory cytokines detected via ELISA

E. coli

(pg/mlFS.E.M.)

MRSA

(pg/mlFS.E.M.)

LPS

(pg/mlFS.E.M.)

Time 0

(pg/mlFS.E.M.)

IL-10 500F 200 600F 200 100F 70 100F 100

IL-1ra 400F 200 1000F 500* 200F 100 200F 0

IL-4 100F 40 70F 40 100F 60 40F 8

TNF-SRII 900F 200 1000F 300 600F 60* 2000F 200

TNF-SRI 110F 4* 103F 3* 108F 2* 144F 16

Anti-inflammatory cytokines detected via ELISA. ELISA results

indicated that several anti-inflammatory cytokines were present in

both the stimulated and unstimulated samples. The data represents

the meansF S.E.M. for four donors.

*P < 0.05 compared to Time 0.

Table 4b

Anti-inflammatory cytokines detected via membrane protein array

E. coli

(pixelsFS.E.M.)

MRSA

(pixelsFS.E.M.)

LPS

(pixelsFS.E.M.)

T0

(pixelsFS.E.M.)

IL-10 BDL BDL BDL BDL

IL-1ra BDL BDL BDL BDL

IL-4 BDL BDL BDL BDL

TNF-SRII NPOM NPOM NPOM NPOM

TNF-SRI NPOM NPOM NPOM NPOM

Anti-inflammatory cytokines detected via the membrane array. The

membrane arrays failed to detect any anti-inflammatory cytokines.

It should be noted that some anti-inflammatory cytokines were not

available for the membrane array such as the soluble receptors for

TNF. The data represents the meansF S.E.M. for four donors.

BDL= below detection limit of the membrane array; NPOM=not

present on membrane array.

S. Copeland et al. / Journal of Immunological Methods 284 (2004) 99–106104

capable of detecting levels of IL-6, IL-8, and MIP-1hfrom the stimulated whole blood (Fig. 2). Actual pixel

densities are depicted in Tables 3b, 4b, and 5b.

3.2.1. Pro-inflammatory cytokines

With the exception of IFN-g, ELISA results indi-

cated that experimental treatments produced detect-

able amounts of pro-inflammatory cytokines.

Significant differences (P < 0.05) were seen between

IL-6, IL-8, IL-1h, and TNF-a concentrations when

comparing E. coli vs. Time 0. IL-8, IL-1h, and TNF-aconcentrations were significantly higher in MRSA

treatment groups as compared to Time 0. As expected,

and previously reported, LPS induced significant

amounts of all pro-inflammatory cytokines (DeForge

and Remick, 1991; DeForge et al., 1992a,b,c). The

membrane array detected IL-6 and IL-8 for all exper-

imental groups. IL-8 densities were significantly

Table 5a

Chemokines detected via ELISA

E. coli (pg/mlF S.E.M.) MRSA (pg/mlF S.E.M

MIP-1h 800000F 300000* 600000F 140000

MIP-1a 69000F 8000* 30000F 8000

MCP-1 100F 40 600F 30*

h NGF 100F 30* 50F 0

RANTES 28000F 19000 20000F 8000

Chemokines detected via ELISA. All treatment groups produced a con

however, MCP-1 and hNGF were at or below the lowest level of detectio

1:500 dilution of the sample for MIP-1h was 5000 pg/ml, and based on a

limit of detection for hNGF was 50 pg/ml. The data represents the meansF*P < 0.05 compared to Time 0.

higher in all experimental groups when compared to

Time 0. However, no other pro-inflammatory cyto-

kines were detected. It is noteworthy that neither TNF

nor IL-1h were detected in the membrane array after

stimulation of whole blood since several authors have

detected these cytokines after whole blood stimulation

(Waage and Bakke, 1988; Strieter et al., 1990).

Additionally, the original, non-custom membrane ar-

ray also failed to detect TNF-a or IL-1-h in stimulated

human whole blood (data not shown) (Tables 3a and

3b).

3.2.2. Anti-inflammatory cytokines

ELISA results indicated that several anti-inflam-

matory cytokines were present in both the stimulated

and unstimulated samples. These results were not

surprising since it is well recognized that normal

volunteers have endogenous, circulating levels of

these mediators. In contrast, the membrane arrays

failed to detect any anti-inflammatory cytokines. It

should be noted that some anti-inflammatory cyto-

kines were not available for the membrane array such

as the soluble receptors for TNF (Tables 4a and 4b).

3.2.3. Chemokines

All treatment groups produced a considerable

amount of RANTES, MIP-1a, and MIP-1h when

measured via ELISA; however, MCP-1 and hNGFwere at or below the lowest level of detection for

some of the stimuli. MIP-1a and MIP-1h and hNGFconcentrations were significantly different for E. coli

vs. Time 0. The membrane arrays only detected levels

of MIP-1h, with significant differences seen with E.

coli and LPS vs. Time 0. The membrane arrays did not

.) LPS (pg/mlF S.E.M.) Time 0 (pg/mlF S.E.M.)

1400000F 300000* BDL

50000F 5000* 18000F 9000

200F 80 BDL

70F 20 BDL

13000F 5000 12000F 1700

siderable amount of RANTES, MIP-1a and MIP-1h via ELISA;

n for some of the stimuli. The lower limit of detection based on the

1:10 dilution the limit of detection for MCP-1 was 30 pg/ml and the

S.E.M. for four donors. BDL= below detection limit of the ELISA.

Table 5b

Chemokines detected via Membrane protein array

E. coli

(pixelsFS.E.M.)

MRSA

(pixelsFS.E.M.)

LPS

(pixelsFS.E.M.)

T0

(pixelsFS.E.M.)

MIP-1h 78F 6* 50F 19 198F 5* 0

MIP-1a BDL BDL BDL BDL

MCP-1 BDL BDL BDL BDL

h NGF NPOM NPOM NPOM NPOM

RANTES BDL BDL BDL BDL

Chemokines detected via the membrane array. The membrane arrays

only detected levels of MIP-1h, with significant differences seen

with E. coli and LPS vs. Time 0. The data represents the

meansF S.E.M. for four donors. BDL= below detection limit of

the membrane array; NPOM=not present on membrane array.

*P< 0.05 compared to Time 0.

S. Copeland et al. / Journal of Immunological Methods 284 (2004) 99–106 105

detect any MIP-1a, even though the concentrations

determined by ELISA were up to 69,000 pg/ml.

Additionally, the membrane failed to detect any

RANTES in any of the samples (Tables 5a and 5b).

4. Discussion

Our study found these membranes lacked the

ability to accurately profile cytokine expression in

plasma from bacterial and LPS-stimulated human

whole blood. Addressing the reason for the mem-

brane arrays not detecting additional cytokines is not

the focus of this paper; rather the attention is

directed to correlating conventional ELISA results

with this novel technology. There is some correlation

between ELISA and the membrane arrays noted in

the IL-6, IL-8, and MIP-1h results; however, the

lack of detection of TNF-a from the positive control

(LPS-stimulated) leaves the membrane array results

in question. LPS has been well established as a

potent inducer of cytokines in human blood.

DeForge and Remick (1991) found normal human

blood stimulated with 10 Ag/ml LPS produced a

rapid TNF-a mRNA induction, peaking at 1 h post

LPS stimulation. Additionally, IL-6 mRNA levels

showed similar induction post LPS stimulation

(DeForge and Remick, 1991). Frieling (Frieling et

al., 1997) demonstrated a dose-dependent increase of

IL-1h and IL-6 concentrations after the addition of

LPS to normal human whole blood. Mitov et al.

(1997) reported TNF-a, and IL-6 concentrations

from LPS doses ranging from 0.00001 to 0.1 Ag/ml in normal human whole blood. Kleinschmidt et

al. (1998) demonstrated mRNA induction of TNF-a,

IL-1h, and IL-6 from peripheral blood mononuclear

cells isolated from LPS-stimulated (1 Ag/ml) normal

human whole blood. Heesen et al. (2002) stimulated

normal human whole blood with 100 ng LPS for 4

h and found significant levels of TNF-a, IL-1h, andIL-6. Additional studies have shown LPS stimulation

of TNF-a, IL-1h, and IL-6 in normal human whole

blood (Ahne and Jarre, 2002; Boneberg and Hartung,

2002; Weigand et al., 2000). Therefore, multiple

authors have demonstrated production of TNF-a

and IL-1h after stimulating human whole blood.

The failure of the membrane array to detect these

cytokines while the ELISA did both detect and

quantitate is not due to false positive results with

the ELISA.

At first, it might appear to be an issue of the

membrane’s sensitivity since the cytokines that were

detected via the membranes were generally in the

highest tier of concentrations. It should be noted that

MIP-1a and RANTES concentrations in some sam-

ples were 69,000 and 28,000 pg/ml and even with

these extremely high levels the membrane array failed

to detect their presence. However, the MIP-1a con-

centrations, via ELISA, were far above the IL-8 con-

centrations for all stimulants and the membrane array

did detect the presence of IL-8. Thus the failure to

detect the cytokines is not merely due to lack of

sensitivity. As further evidence that it is not just a

sensitivity issue, TNF-a concentrations for the E. coli-

stimulated whole blood were 6000F 1000 pg/ml,

while IL-8 concentrations for LPS-stimulated whole

blood was 6000F 2000 pg/ml, indicating that some

cytokines may be detected (IL-8) but other cytokines

present at the same concentration may not be measur-

able (TNF-a). While the use of novel protein micro-

arrays offers potential for low cost screening, the

present technology appears to lack sufficient rigor to

work in a protein complex environment such as

human plasma.

Acknowledgements

This work was supported in part by NIH grants

GM 50401 and GM 44918.

S. Copeland et al. / Journal of Immunological Methods 284 (2004) 99–106106

References

Ahne, W., Jarre, T., 2002. Environmental toxicology: polychlori-

nated biphenyls impair TNF-alpha release in vitro. J. Vet. Med.,

A. Physiol. Pathol. Clin. Med. 49, 105–106.

Balfour, J.A., Bryson, H.M., Brogden, R.N., 1996. Imipenem/cilas-

tatin: an update of its antibacterial activity, pharmacokinetics

and therapeutic efficacy in the treatment of serious infections.

Drugs 51, 99–136.

Boneberg, E.M., Hartung, T., 2002. Granulocyte colony-stimulating

factor attenuates LPS-stimulated IL-1beta release via suppressed

processing of proIL-1beta, whereas TNF-alpha release is in-

hibited on the level of proTNF-alpha formation. Eur. J. Immu-

nol. 32, 1717–1725.

Cahill, D.J., 2001. Protein and antibody arrays and their medical

applications. J. Immunol. Methods 250, 81–91.

DeForge, L.E., Remick, D.G., 1991. Kinetics of TNF, IL-6, and IL-

8 gene expression in LPS-stimulated human whole blood. Bio-

chem. Biophys. Res. Commun. 174, 18–24.

DeForge, L.E., Fantone, J.C., Kenney, J.S., Remick, D.G.,

1992a. Oxygen radical scavengers selectively inhibit interleu-

kin 8 production in human whole blood. J. Clin. Invest. 90,

2123–2129.

DeForge, L.E., Kenney, J.S., Jones, M.L., Warren, J.S., Remick,

D.G., 1992b. Biphasic production of IL-8 in lipopolysaccharide

(LPS)-stimulated human whole blood. Separation of LPS- and

cytokine-stimulated components using anti-tumor necrosis fac-

tor and anti-IL-1 antibodies. J. Immunol. 148, 2133–2141.

DeForge, L.E., Tracey, D.E., Kenney, J.S., Remick, D.G., 1992c.

Interleukin-1 receptor antagonist protein inhibits interleukin-8

expression in lipopolysaccharide-stimulated human whole

blood. Am. J. Pathol. 140, 1045–1054.

Frieling, J.T., Mulder, J.A., Hendriks, T., Curfs, J.H., van der Lin-

den, C.J., Sauerwein, R.W., 1997. Differential induction of pro-

and anti-inflammatory cytokines in whole blood by bacteria:

effects of antibiotic treatment. Antimicrob. Agents Chemother.

41, 1439–1443.

Heesen, M., Bloemeke, B., Heussen, N., Kunz, D., 2002. Can the

interleukin-6 response to endotoxin be predicted? Studies of the

influence of a promoter polymorphism of the interleukin-6 gene,

gender, the density of the endotoxin receptor CD14, and inflam-

matory cytokines. Crit. Care Med. 30, 664–669.

Huang, R.P., Huang, R., Fan, Y., Lin, Y., 2001. Simultaneous de-

tection of multiple cytokines from conditioned media and pa-

tient’s sera by an antibody-based protein array system. Anal.

Biochem. 294, 55–62.

Kleinschmidt, S., Wanner, G.A., Bussmann, D., Kremer, J.P., Zie-

genfuss, T., Menger, M.D., Bauer, M., 1998. Proinflammatory

cytokine gene expression in whole blood from patients under-

going coronary artery bypass surgery and its modulation by

pentoxifylline. Shock 9, 12–20.

Kodadek, T., 2001. Protein microarrays: prospects and problems.

Chem. Biol. 8, 105–115.

MacBeath, G., 2002. Protein microarrays and proteomics. Nat.

Genet. 32, 526–532.

Mitov, I.G., Kropec, A., Benzing, A., Just, H., Garotta, G., Galanos,

C., Freudenberg, M., 1997. Differential cytokine production in

stimulated blood cultures from intensive care patients with bac-

terial infections. Infection 25, 206–212.

Nemzek, J.A., Siddiqui, J., Remick, D.G., 2001. Development and

optimization of cytokine ELISAs using commercial antibody

pairs. J. Immunol. Methods 255, 149–157.

Schweitzer, B., Roberts, S., Grimwade, B., Shao, W., Wang, M., Fu,

Q., Shu, Q., Laroche, I., Zhou, Z., Tchernev, V.T., Christiansen,

J., Velleca, M., Kingsmore, S.F., 2002. Multiplexed protein

profiling on microarrays by rolling-circle amplification. Nat.

Biotechnol. 20, 359–365.

Strieter, R.M., Remick, D.G., Ham, J.M., Colletti, L.M., Lynch,

J.P.d., Kunkel, S.L., 1990. Tumor necrosis factor-alpha gene

expression in human whole blood. J. Leukoc. Biol. 47,

366–370.

Waage, A., Bakke, O., 1988. Glucocorticoids suppress the produc-

tion of tumour necrosis factor by lipopolysaccharide-stimulated

human monocytes. Immunology 63, 299–302.

Weigand, M.A., Schmidt, H., Zhao, Q., Plaschke, K., Martin, E.,

Bardenheuer, H.J., 2000. Ketamine modulates the stimulated

adhesion molecule expression on human neutrophils in vitro.

Anesth. Analg. 90, 206–212.