A selective protein sensor for heparin detection

ANALYTICAL

Analytical Biochemistry 326 (2004) 33–41

BIOCHEMISTRY

www.elsevier.com/locate/yabio

A selective protein sensor for heparin detection

Shenshen Cai,a,1 Jodi L. Dufner-Beattie,b,1,2 and Glenn D. Prestwicha,*

a Department of Medicinal Chemistry and Center for Cell Signaling, The University of Utah,

419 Wakara Way, Suite 205, Salt Lake City, UT 84108-1257, USAb Echelon Biosciences Inc., 675 South Arapeen Drive, Suite 302, Salt Lake City, UT 84158-0537, USA

Received 29 July 2003

Abstract

No clinical assays for the direct detection of heparin in blood exist. To create a heparin sensor, the hyaluronan (HA)-binding

domain (HABD) of a protein that binds heparin and HA was engineered. GST fusion proteins containing one to three HABD

modules were cloned, expressed, and purified. The affinities of each construct for heparin and for HA were determined by a com-

petitive enzyme-linked immunosorbent assay using immobilized HA or heparin. Each of the constructs showed modest affinity for

immobilized HA. However, heparin was 100-fold more potent than HA as a competing ligand. With immobilized heparin, affinity

increased as the HABD copy number increased. The three-copy construct, GST-HB3, detected unfractionated free heparin (UFH) as

low as 39 ng/ml (equivalent to approximately 0.1U/ml) with a signal-to-noise ratio of 5.6. GST-HB3 also showed 100-fold selectivity

for heparin in preference to other glycosaminoglycans. The plot of logKd vs log [Naþ] showed 2.5 ionic interactions per heparin–HB3

interaction. GST-HB3 showed a linear detection of both UFH (15 kDa) and low-molecular-weight heparin (LMWH; 6 kDa) added to

human plasma. For UFH, the range examined was 78 to over 2000 ng/ml (equivalent to 0.2 to 5.0U/ml). For LMWH, the useful range

was 312 to over 2000 ng/ml. The coefficient of variance for the assay was <9% for six serial heparin dilutions and <12% for three

plasma samples. In clinical use, GST-HB3 could accurately measure therapeutic heparin levels in plasma (0.2 to 2U/ml).

� 2003 Elsevier Inc. All rights reserved.

Keywords: Hyaluronan; Receptor for HA-mediated motility; Glycosaminoglycan; Helical binding domain; Competitive ELISA; Plasma levels;

Biotinylated heparin

The receptor for hyaluronan (HA)3-mediated motility

(RHAMM) [1] features a 62-amino acid HA-binding

domain (HABD) that contains two base-rich BX7B

* Corresponding author. Fax: 1-801-585-9053.

E-mail address: [email protected] (G.D. Prestwich).1 These authors contributed equally to this paper.2 Present address: Department of Biochemistry and Molecular

Biology, University of Kansas, Kansas City, KS 66160-7421, USA.3 Abbreviations used: ACT, activated coagulation time; AT-III,

antithrombin III; BSA, bovine serum albumin; CS-A, chondroitin

4-sulfate; CS-C, chondroitin 6-sulfate; CV, coefficient of variance;

DTT, dithiothreitol; ELISA, enzyme-linked immunosorbent assay;

GAG glycosaminoglycan; HA, hyaluronan; HABD, HA-binding

domain; HB, helical binding; HMT, heparin management test; HRP,

horse radish peroxidase; HS, heparan sulfate; IPTG, isopropyl b-DD-thiogalactoside; KS, keratan sulfate; LMWH, low-molecular-weight

heparin; ORF, open reading frame; PBS, phosphate-buffered saline;

PET, polyelectrolyte theory; PMSF, phenylmethylsulfonyl fluoride;

PTT, partial thromboplastin time; RHAMM, receptor for HA-

mediated motility; rt, room temperature; SA, streptavidin; TBS; tris-

buffered saline; UFH, unfractionated free heparin.

0003-2697/$ - see front matter � 2003 Elsevier Inc. All rights reserved.

doi:10.1016/j.ab.2003.11.017

motifs and possesses an overall helix–turn–helix struc-

ture. This HABD also has significant affinity for hepa-

rin, a polysulfated glycosaminoglycan (GAG) (Fig. 1)

[2]. RHAMM mediates cell migration and proliferation[3], and isoforms can be found in cytoplasm and on the

surfaces of activated leukocytes, subconfluent fibro-

blasts [4,5], and endothelial cells [6]. RHAMM expres-

sion in cell-surface variants promotes tumor progression

in selected types of cancer cells [7]. Recently, intracel-

lular RHAMM has been shown to bind to cytoskeletal

proteins, to associate with erk kinase, and to mediate the

cell cycle through its interaction with pp60v-src [8]. Themajority of studies of RHAMM, including the discovery

of HA-mimetic peptides that block HA–RHAMM

HABD interaction [9], have focused on its binding to

HA. Thus, we envisaged the use of multiple repeats of

this HABD for HA detection, as previously accom-

plished using both native and recombinant HABDs

from aggrecan [10–12]. We describe herein the surprising

mail to: [email protected]

Fig. 1. Partial tetrasaccharide structures of hyaluronan and heparin.

34 S. Cai et al. / Analytical Biochemistry 326 (2004) 33–41

result that these constructs instead show high affinityand selectivity for heparin, thereby producing a com-

pletely novel high-affinity, heparin-specific detection

reagent.

The structures of HA and heparin GAGs differ sub-

stantially, although both are GAGs with alternating

uronic acid and aminoglycoside residues. HA is unsulf-

ated and homogeneous, with a regular repeating disac-

charide consisting of alternating glucuronic acid andN -acetylglucosamine residues in alternating b-1,4- and

b-1,3-glycosidic linkages. Heparin has only 1,4-glycoside

linkages but no regular repeat unit; it is highly hetero-

geneous, having two epimeric uronic acids and both Nand O sulfation. In blood, heparin interacts with anti-

thrombin III (AT-III), which blocks activation of factor

Xa and thereby prevents blood coagulation [13]. Two

kinds of heparin, unfractionated free heparin (UFH)and low-molecular-weight heparin (LMWH), are em-

ployed as therapeutic agents to reduce blood clot for-

mation and thrombosis [14–18]. Although LMWH has

better bioavailability, the less expensive UFH is still

widely used in the United States [13].

Plasma heparin levels can be detected by several

clinically approved methods: (i) determination of acti-

vated coagulation time (ACT), (ii) partial thrombo-plastin time (PTT) [19], (iii) the heparin management

test (HMT) [20,21], or (iv) the anti-factor Xa assay [22].

Recently, one chemical method measured heparin by

monitoring inhibition of thrombin activity on a fluoro-

genic substrate [23]; however, this method lacked the

sensitivity required for clinical use. For over 30 years,

the measurement of PTT has remained the most widely

used protocol for prescribing and monitoring the use ofanticoagulants in patients. As a biochemical assay, PTT

continues to be problematic because it correlates poorly

with heparin concentration in certain clinical situations

[24,25]. While the anti-Xa assay is an improved method

with increased dependability, its considerably higher

expense deters widespread acceptance for clinical use

[13]. Hence, a convenient, selective, and economical

heparin quantification method would be an importantclinical tool for management of millions of patients.

Initial studies indicated that RHAMM appeared to

have a lower affinity for heparin than for HA [26], an

observation confirmed during preparation of the

62-amino acid minimal HABD of RHAMM for struc-

tural studies (RHAMM-P1) [9]. It is also possible thatother regions of RHAMM contribute to the specificity

for HA by reducing its reactivity with other GAGs, in-

cluding heparin. The polycationic binding motifs in

RHAMM-P1 might be expected to show increased af-

finity with heparin simply as a result of increased elec-

trostatic interactions with this polysulfated GAG [2,27].

To test this hypothesis, we generated GST-tagged re-

combinant proteins containing one, two, or three copiesof RHAMM-P1. Based on the helix–turn–helix struc-

ture predicted and experimentally determined for

RHAMM-P1, we refer to these as helical binding (HB)

domain fusion proteins GST-HB1, GST-HB2, and

GST-HB3. Detailed examination of the affinity and se-

lectivity of GST-HB3 for GAGs led to the development

of a rapid ELISA for clinical determination of heparin

levels in human plasma.

Materials and methods

Plasmid construction

RHAMM(518–580) cDNA was provided by Dr.

M.R. Ziebell (The University of Utah) [9]. The modifiedvector pGEX-ERL was developed from pGEX by

changing endonuclease sites in the multicloning site. A

forward primer, 50-CGGGATCCGGTGCTAGCCGT

GACTCCTATGCACAGCTCCTTGG-30, with BamHI

and NheI cleavage sites at 50 and a reverse primer, 50-GGAGCGGTCGACACGGATGCCCAGAGCTTTA

TCTAATTC-30, with a SalI site at 50 were synthesized to

amplify RHAMM(518–580). The PCR product wasdigested with BamHI and SalI and ligated into the

modified pGEX vector, which had also been digested

with BamHI and XhoI to obtain the HB1 construct. This

subcloning step eliminates the downstream restriction

sites so that the insert cannot be excised during sub-

sequent manipulations. To connect the consecutive

multiple copies of the P1 open reading frame (ORF), a

(GlySer)9Gly linker was introduced using the forwardprimer 50-GATCCGGTCTCGAGGGAAGTGGTTCT

GGAAGTGGTTCAGGTTCGGGTAGCGGATCTG

GTTCAGGAAGTGGTT-30 containing a XhoI site

and the reverse primer 50-CTAGAACCACTTCCTG

AACCAGATCCGCTACCCGAACCTGAACCACTT

S. Cai et al. / Analytical Biochemistry 326 (2004) 33–41 35

CCAGAACCACTTCCCTCGAGACCG-30 containinga BamHI site. The vector with single P1 ORF was lin-

earized with BamHI and NheI and ligated with the an-

nealed linker primers. This intermediate product was

again digested with BamHI and XhoI and then ligated

with another PCR-amplified P1 ORF cDNA, which had

been digested with BamHI and SalI to give the HB2

recombinant construct. The HB3 construct was synthe-

sized by repeating the above steps with another linkerand amplified P1 cDNA. All recombinant constructs

were sequenced to confirm the presence of in-frame

fusions with GST and the absence of mutations that

may have been introduced during PCR amplification of

RHAMM cDNA.

Protein synthesis

Each of the GST-HB plasmids, and the empty

pGEX-ERL vector were transformed into Escherichia

coli strain BL21 (DE3) (Novagen). Bacteria were grown

in 20ml LB culture at 37 �C overnight, transferred to 1

liter of fresh LB, and incubated at 37 �C for 3 h. Ex-

pression was induced by addition of 0.1mM IPTG

(Pierce, Rockford, IL) (for GST alone and GST-HB1)

or 0.5mM IPTG (for GST-HB2 and GST-HB3) andincubated at 22 �C for 4 h. The bacterial pellet was col-

lected by centrifugation (4000g, 15min), resuspended

with 100ml of STE buffer (10mM Tris, pH 8.0, 150mM

NaCl, and 1mM EDTA), and incubated for 15min on

ice. Next, a mixture of 1mM each of four protease in-

hibitors (PMSF, aprotinin, pepstatin A, and leupeptin,

Sigma, St. Louis, MO) and 5mM dithiothreitol (DTT,

Sigma) was added. The expressed proteins were releasedinto solution by sonication, and the 13,000g (10min)

supernatant was loaded onto a 10ml total volume of

glutathione–Sepharose 4B bead slurry (equal to 5ml of

beads, Amersham Pharmacia, Piscataway, NJ) to bind

GST-tagged proteins. After six washes with PBS (pH

7.4, 0.1M), the desired proteins (GST, GST-HB1, GST-

HB2, and GST-HB3) were eluted with 10ml of 20mM

GSH (Sigma) in Tris–HCl (100mM, pH 8.0, 120mMNaCl, 0.1% Triton X-100). The elution was repeated

two additional times to give three samples for each

protein. Protein concentrations were determined by

Bradford reagent (Sigma) with bovine serum albumin

(BSA; Pierce) as standard control. Purified proteins were

stored at )80 �C in small portions. For each use, an

aliquot was thawed and discarded after use in a given

experimental set.

Enzyme-linked immunosorbant assay

In a 96-well plate precoated with streptavidin (SA)

(Roche, Indianapolis, IN), 50 ll of 10 lg/ml biotinylated

heparin (average 15 kDa, Celsus, Cincinnati, OH) was

loaded into each well and incubated at 4 �C overnight.

Following three washes with TBS (20mM Tris, 150mMNaCl, pH 7.5), 100 ll StabilGuard solution (Surmodics,

Eden Prairie, MN) was applied to each well to block the

unbound SA sites. After 1-h incubation at room tem-

perature (rt), followed by three washes with TBS, trip-

licate 100-ll aliquots of GST, GST-HB1, GST-HB2,

and GST-HB3 were added at increasing concentrations.

After 1 h incubation at rt, followed by four washes with

TBS, 50 ll of mouse anti-GST antibody (Sigma) (1:1000diluted in TBS) was added. After incubation (1 h, rt), the

plate was washed four times with TBS. Then, 50 llhorseradish peroxidase (HRP)-conjugated anti-mouse

IgG (Sigma) (1:3000 diluted in TBS) was added. After

incubation (1 h, rt), the plate was washed four times with

TBS, and then 100 ll of 3,30,5,50-tetramethyl benzidene

(Sigma) was added. The wells gradually developed a

dark blue color during 15min incubation. Finally, 100 llof 1M H2SO4 was added and the resulting yellow color

was read by measuring absorbance at 430 nm.

For the competitive ELISA with different GAGs, an

aliquot of 100 ll/well of unlabeled GAG was added to

the GST or GST-HB proteins (50 lg/ml) after the Sta-

bilGuard blocking step but before the antibody loading

step. GAGs employed included chondroitin 4-sulfate

(CS-A), chondroitin 6-sulfate (CS-C), keratan sulfate(KS), heparan sulfate (HS) (all from Sigma), HA

(190 kDa, produced by acid degradation of 1200 kDa

HA; Clear Solutions Biotech, Inc., Stony Brook, NY),

and UFH (average 15 kDa, Sigma).

Heparin quantification using GST-HB3 protein

The GST-HB3 protein was selected for further hep-arin measurements using the competitive ELISA [28].

Thus, serial twofold dilutions of UFH were prepared

from 10 lg/ml to 20 ng/ml, and duplicate aliquots of

100 ll/well were used as competitors as described

above, with 100 ll� 50 lg/ml aliquot per well of GST-

HB3. In addition, 100 ll/well human plasma sample

(Sigma) was premixed with 100 ll/well serially diluted

heparin and added to the plate. In this simulatedplasma assay, both UFH and LMWH (6 kDa; Sigma)

were employed as competitors. Gradient concentrations

were also used in this assay to study the feasibility of a

role for the GST-HB3 protein in heparin detection in

plasma samples.

Characterization of HB3 binding with heparin

The heparin-binding ELISA was performed using

different NaCl concentrations in TBS to observe the salt

effect. Thus, the HB3 concentration was varied from 0 to

300 lg/ml and the NaCl concentration was varied from

150 to 1000mM. After the GST-HB3 was loaded into

the wells and incubated with the plate for 1 h, an aliquot

from each plate well was transferred into another


96-well plate in spatially corresponding wells. The HB3contained in those aliquots was considered to be free

and the concentration was measured using the Bradford

reagent (Sigma). Next, the amount of bound HB3–

heparin was calculated using Scatchard analysis from

the proportional ELISA signal (Amax ¼ 2:00 in our ex-

periment at 150mM NaCl). All added heparin was im-

mobilized, as verified in previous titrations with different

heparin amounts (data not shown). Thus, the amount offree heparin equaled the total heparin (corresponding to

the maximum signal) minus the bound heparin (corre-

sponding to the measured absorbances). Therefore, the

binding Kd value is given as Kd ¼ [free HB3][free hepa-

rin]/[HB3–heparin complex]. Absorbance signals at

300 lg/ml were selected for Kd calculation because sig-

nals at lower concentrations were too weak and vari-

able. Next, log Kd at different NaCl concentrations wasplotted against log [NaCl] to give the number of ionic

interactions between HB3 and heparin based on poly-

electrolyte theory (PET) [29].

Fig. 2. Preparation of GST-HB1, GST-HB2, and GST-HB3 con-

structs. (A) Protein sequence of HB3 protein, starting at the thrombin

cleavage site. For the GST-HB3 construct, the GST protein is N ter-

minal of this site. The three RHAMM(518–580) P1 repeats are shown

in boldface. (B) Cloning strategy for construction of the GST-HB

expression plasmids.

Results and discussion

To obtain a high-affinity HA-binding protein, weused tandem repeats of the region of the RHAMM(518–

580) cDNA (Fig. 2A) [9] separated by a linker that

encoded alternating glycine and serine residues. We

reasoned that generating a protein with more than one

HA-binding domain would increase avidity due to co-

operative or synergistic binding, as previously observed

for phosphatidylinositol 3-phosphate binding to the 2X

FYVE protein [30]. The subcloning scheme is summa-rized in Fig. 2B and was accomplished in five steps: (i)

preparation of an engineered GST expression vector

with appropriate restriction sites, (ii) insertion of

RHAMM(518–580) ‘‘P1’’ domain to obtain the GST-

HB1 construct, (iii) insertion of an oligonucleotide en-

coding a 19-residue Gly-Ser linker region (GSGSGSGS

GSGSGSGSGSG) to separate P1 domains, (iv) addition

of a second P1 domain to obtain the GST-HB2 con-struct, and (v) attachment of the linker plus a third P1

domain to complete the GST-HB3 construct.

Thus, the cDNA corresponding to the P1 region,

RHAMM(518–580), was subcloned into the modified

pGEX vector to give GST-HB1, GST-HB2, and GST-

HB3 with 1, 2, and 3 repeats of the P1 region, respec-

tively (Fig. 2). The sequences of these recombinant

constructs were confirmed by DNA sequencing. Theseconstructs were first expressed at 37 �C. However, the

large proportion of protein was present in insoluble

form; by reducing the expression temperature to 22 �C,the percentage of soluble protein was dramatically in-

creased (Fig. 3A). Subsequently, GST protein alone and

GST-HB1, GST-HB2, and GST-HB3 were purified by

affinity chromatography on immobilized GSH and

electrophoresed on SDS–PAGE to show the expected

sizes of 25, 30, 38, and 46 kDa (calculated masses¼ 28,

35, 44, and 53 kDa), respectively (Fig. 3B). Protein

concentrations decreased as the inserted fragment sizeincreased. Thus, GST and GST-HB1 were obtained at

yields of 30mg per liter bacterial culture, while we

initially obtained yields of 10mg/L for GST-HB2 and

Fig. 4. Protein titration for three GST-HB proteins using ELISA with

immobilized heparin. }, GST alone; �, GST-HB1; n, GST-HB2; X,

GST-HB3.

Fig. 5. Competition ELISAs for three GST-HB proteins using im-

mobilized heparin. Competitors, (A) HA, CS-A, CS-C, and UFH at

200lg/ml; (B) HS at 5 and 200lg/ml; KS at 5 and 200lg/ml. Control,

no competitor added.

Fig. 3. Expression and purification of GST-HB proteins. (A) SDS–

PAGE of postsonication supernatant protein expression; boxes show

the GST alone and GST-HB fusion proteins. (B) Protein purification

on GSH–Sepharose beads, following elution of GST and GST-HB

proteins with GSH. Lanes: 1, GST; 2, GST-HB1; 3, GST-HB2; 4,

GST-HB3. Note: Overloading of lane 2 resulted in appearance of

GST-HB2 protein in lanes 1 and 3.


5mg/L for GST-HB3. The yield of GST-HB3 was in-

creased to 14mg/L by adding 120mM NaCl and 0.1%

Triton X-100 to the elution buffer. All proteins were

relatively stable when maintained at or below )20 �C;nonetheless, binding activity gradually diminished at

4 �C over several months.

GST-HB proteins selectively bind heparin

The affinity and selectivity of GST and the GST-HB

proteins for HA were examined first, using an ELISA

system similar to that described herein but with bioti-nylated HA as the immobilized ligand [31]. The GST-

HB3 protein bound with highest affinity to immobilized

HA and was selective for HA as compared to CS-A and

CS-C (data not shown). However, to our considerable

surprise, 190 kDa HA was also a poor competitor for

displacement of this binding, while 1000-fold lower

concentrations of heparin effectively competed for the

interaction of GST-HB3 with immobilized HA. Ap-parently, the tandem repeats of P1 selectively amplified

the heparin affinity while reducing the HA affinity. Thus,

we repeated the ELISA protocols using biotinylated

heparin instead of the biotinylated HA. Each of the

GST-HB proteins was readily displaced using UFH as

the competitor, with a protein concentration of 50 lg/ml

(100 ll/well) of GST-HB3 (Fig. 4).

To evaluate the specificity of GST-HB proteins, acompetitive ELISA was performed with CS-A, CS-C,

HA, KS, HS, and UFH as the competitors at 200 lg/ml

(Fig. 5). The results indicated that the GST-HB proteins

bound to heparin with higher affinity and selectivity

relative to other GAGs. Moreover, both affinity and

selectivity appeared to increase with the number of

tandem P1 domains. This can be attributed in part to

increased electrostatic interactions between the highly

sulfated heparin and the HS with the polybasic nature of

the binding site. The differences between heparin andHS, which differ little in net charge, can be attributed to

stereospecific ligand recognition. Nonetheless, the high

selectivity suggested potential applications of the GST-

HB3 protein for detection of heparin at low levels in

biological fluids. To explore this possibility, serial dilu-

tions of HA, CS-C, CS-A, and UFH were used with

Table 1

Estimated IC50 values (lg/ml) for GAGs as competitors in ELISA with

immobilized heparin and GST-HB detection

GAG GST-HB1 GST-HB2 GST-HB3

HA 20–50 >200 >200

KS >1000 >1000 >1000

CS-A 10–20 20–50 20–50

CS-C 100–200 20–50 20–50

Heparin 0.1–0.2 <0.1 0.1–0.2

HS <1 <5 <5

Fig. 7. Measurement of UFH by ELISA with immobilized heparin and

GST-HB3 detection using serial 1:2 dilutions. Inset shows a log–log

plot demonstrating linear response over greater than three orders of

magnitude of UFH concentration (39 ng/ml to 10 lg/ml). Relative A430

is the absorbance at 430 nm compared to the control (no UFH added).

Based on parallel experiments with a commercial kit, the conversion

factor was approximately 400 ng/U for UFH.


GST and each GST-HB protein. Table 1 presents the

estimated IC50 values for competitive displacement for

each protein, illustrating a 100- to 2000-fold selectivity

for heparin relative to the less-sulfated GAGs. Fig. 6

depicts the raw data for GST-HB3.

Quantification of free heparin in solution

GST-HB3 was selected for further study as a detec-

tion protein for determination of heparin concentra-

tions. First, serial twofold dilutions of UFH were

prepared in the range 10 lg/ml to 20 ng/ml. The UFH

sodium salt used was from porcine mucosa. The ELISA

data for these dilutions yielded a logarithmic plot of

absorbance vs UFH concentration, and a log–log plot of

relative absorbance (corrected for no heparin blank) vsconcentration gave the expected linear relationship

(Fig. 7). This calibration curve demonstrates that GST-

Fig. 6. Quantitative competitive ELISAs using immobilized heparin and dete

UFH.

HB3 binding to immobilized biotinylated heparin pro-vides a linear range for detection of free UFH of at least

three orders of magnitude, suggesting that this ELISA

has significant potential for measurement of heparin

concentrations with high sensitivity and high selectivity.

We explored the effect of ionic strength by varying the

salt concentration from 50 to 1000mM NaCl. The op-

timal sensitivity was observed at 150mM NaCl, the

physiological concentration employed for this assay

ction with GST-HB3. (A) HA (Mw 190 kDa); (B) CS-A; (C) CS-C; (D)


(data not shown). An inverse ELISA, in which immo-bilized GST-HB3 was coupled to detection by biotiny-

lated heparin and HRP-SA, gave essentially identical

results for sensitivity of heparin detection (data not

shown).

Quantification of heparin in human plasma

To determine the suitability of GST-HB3 for deter-mining therapeutic heparin levels in plasma, we spiked

human plasma with heparin calibration standards.

Aliquots of human plasma were mixed with equal vol-

umes of serial dilutions prepared from both UFH (av-

erage size 15 kDa) and LMWH (average size 6 kDa).

The log–log plot of relative absorbance vs heparin

concentration was linear and showed the same slope as

that for the calibration standards in buffer alone (Fig. 8).Moreover, both UFH and LMWH showed the same

slopes. Essentially, no loss of sensitivity was observed

for detection of UFH in serum vs buffer (dashed line)

but, as expected, the LMWH was detected with lower

sensitivity. The optimal ranges for heparin measurement

appeared to be from 78 to 2000 ng/ml for UFH and

from 312 to 2000 ng/ml for LMWH. We correlated these

concentrations in a parallel experiment using the Ac-cucolor Heparin Kit (Sigma) to obtain conversion fac-

tors. For UFH, 1U/ml was equivalent to ca. 400 ng/ml,

while for LMWH, 1U/ml was equivalent to ca. 1000 ng/

ml. Thus, the useful range for heparin detection is 0.2 to

5U/ml for UFH and 0.3 to 2U/ml for LMWH. Thera-

peutic levels in plasma are generally between 0.2 and

2.0U/ml, indicating that the assay is sufficiently sensitive

to monitor therapeutically relevant changes in heparinlevels. Our experiments showed that the intraassay co-

efficient of variance (CV) was <9% for six serial UFH

dilutions from 78 ng/ml to 2.5 lg/ml, while the interassay

CV was <12% for three different commercial plasma

Fig. 8. ELISA quantification of heparin standards in human plasma.

The dashed line (overlaps solid line for UFH) shows the UFH quan-

tification in the absence of human plasma, as illustrated in Fig. 7 inset.

The conversion factors are approximately 400 ng/U for UFH and

1000 ng/U for LMWH.

products. At the lower limit of 78 ng/ml, the standarddeviation was 14 ng/ml, and thus S/N was 5.6. More-

over, throughout this detection range, no interference

was caused by the presence of 5 lg/ml HA in the diluted

plasma samples (data not shown).

The addition of fresh human plasma did not reduce

the absorbance in this ELISA (Fig. 9), indicating that a

human plasma sample itself would not interfere with the

competition observed with heparin. That is, no netchange in the slopes or intercepts for the linear log–log

plots was observed when plasma was added in the assay.

However, plasma samples stored at 4 �C for 4 months

did affect ELISA absorbance somewhat, suggesting that

interfering materials can accumulate in outdated plasma

(Fig. 9). Ideally, therefore, fresh plasma samples should

be used in the assay.

Our preliminary data suggest that, even given patientvariability, the clinical concentration of UFH (or

LMWH) could be read following performance of a ge-

neric calibration. This new detection method could offer

a substantial improvement to the current heparin mea-

surement protocols. For clinical use, additional factors

including heparin degradation in samples and interfer-

ence by polysaccharides or other polyelectrolytes will

need to be assessed. Nonetheless, this new direct,sensitive, and quantitative heparin measurement could

be readily integrated into a hospital clinical chemistry

service.

Heparin and GST-HB3 binding characterization

To further understand the interactions between GST-

HB3 and heparin and the ionic contributions involved,we tested the binding affinity changes as the ionic

strength was varied. By increasing NaCl concentrations

from 15–1000mM in TBS, the binding between HB3

and heparin decreased (Fig. 10). By obtaining the

Fig. 9. Effect of adding human plasma on heparin ELISA. See Ma-

terials and methods for experimental details.

Table 2

Kd values at different NaCl concentrations in ELISA with immobilized

heparin and GST-HB detection

[NaCl] (M) Kd (nM)

0.15 2.7� 102

0.30 2.2� 103

0.50 2.6� 103

0.75 6.1� 103

1.0 1.8� 104

Fig. 11. Analysis of PET data for heparin–HB3 binding using a log Kd

vs log [NaCl] plot.

Fig. 10. Effect of [NaCl] on heparin ELISA.


concentrations of free HB3, free heparin, and bound

HB3–heparin complex, we calculated the Kd value at

different NaCl concentrations (Table 2) to quantify the

decreased binding with increased ionic strength. It is

expected that for most heparin-binding proteins, a

substantial contribution to binding would arise from theelectrostatic interactions between the highly anionic

heparin and a corresponding cationic protein. Increased

ionic strength would lessen these ionic interactions be-

tween the negatively charged sulfate and carboxylate

groups on heparin and the positively charged Arg and

Lys residues of the protein. For a given heparin-binding

interaction, an equation based on PET is used to de-

scribe such ionic interactions:

logKd ¼ logK 0d þ ZW log½Naþ�:

Here K 0d is the dissociation constant at 1M [Naþ], Z

refers to the number of ionic interactions involved in the

binding, and W is defined as the fraction of Naþ bound

per heparin charge and released upon binding to HB3

(estimated to be �0.8 [32]). Thus, by plotting log Kd vslog [Naþ], we obtained ZW from the slope. The intercept

gave the log K 0d value, permitting calculation of the

nonionic interactions (Fig. 11). From Fig. 11, Z ¼ 2:50,showing between two and three ionic interactions per

binding heparin–HB3 interaction. Based on the Gilbert

equation,

DG ¼ �RT ðlnKdÞ;where R ¼ 8:314 J/(molK) and T ¼ 298K, when

Kd ¼ K 0d at 1M [NaCl], a nonionic interaction with

DG ¼ 27:1 kJ would be considered to be nonionic.

Compared with Kd at [NaCl]¼ 150mM, when DG ¼37:4 kJ, we calculate that the binding contribution fromnonionic interactions equals the ratio of these two free

energy values (27.1/37.4¼ 72%) and thus the ionic

interactions contribute only 28% of the total binding

energy. This binding character is in the middle range of

known heparin–protein interactions and is acceptable

for development of HB3 as a heparin sensor.

Acknowledgments

We thank the NIH (R43 CA81820 to J.L.B.), Echelon

Biosciences Inc., and the Center for Cell Signaling, aUtah Center of Excellence, for financial support. The

method described herein is patent pending. We thank

Dr. M.R. Ziebell (The University of Utah) for providing

the plasmid encoding the P1 HABD region of RHAMM

and Mr. M.J. Mostert (Echelon Biosciences, Inc. and

Lifespan Technologies, Inc.) for advice on coagulation

assays. Information about assay availability can be

found at www.echelon-inc.com.

References

[1] E. Turley, P. Noble, L. Bourguignon, J. Biol. Chem. 277 (2002)

4589–4592.

[2] B. Yang, B. Yang, R. Savani, E. Turley, EMBO J 13 (1994)

286–296.

[3] Y. Akiyama, S. Jung, B. Salida, S. Lee, S. Hubbard, M. Taylor, T.

Mainprize, K. Akaishi, W. Van Furth, J. Rutka, J. Neurooncol.

53 (2001) 115–127.

[4] V. Assmann, J. Marshall, C. Fieber, M. Hofmann, I. Hart, J. Cell

Sci. 111 (1998) 1685–1694.

[5] A. Day, G. Prestwich, J. Biol. Chem. 277 (2001) 4585–4588.

[6] R. Savani, G. Cao, P. Pooler, A. Zaman, Z. Zhou, H. DeLisser, J.

Biol. Chem. 276 (2001) 36770–36778.

http://www.echelon-inc.com


[7] H. Li, L. Guo, J. Li, N. Liu, R. Qi, J. Liu, Int. J. Oncol. 17 (2000)

927–932.

[8] G. Perdew, H. Wiegand, J. Heuvel, C. Mitchell, S. Singh,

Biochemistry 36 (1997) 3600–3607.

[9] M.R. Ziebell, Z.-G. Zhao, B. Luo, Y. Luo, E.A. Turley, G.D.

Prestwich, Chem. Biol. 8 (2001) 1081–1084.

[10] T. Shibutani, K. Imai, A. Kanazawa, Y. Iwayama, J. Periodont.

Res. 33 (1998) 265–273.

[11] M. Anttila, R. Tammi, M. Tammi, K. Syrjanen, S. Saarikoski, V.

Kosma, Cancer Res. 60 (2001) 150–155.

[12] R. Pirinen, R. Tammi, M. Tammi, P. Hirviroski, J. Parkkinen, R.

Johansson, J. Bohm, S. Hollmen, V. Kosma, Int. J. Cancer 95

(2001) 12–17.

[13] D. Nelson, Clin. Lab. Sci. 12 (1999) 359–364.

[14] F. Schmidt, C. Faul, J. Dichgans, M. Weller, J. Neurol. 249 (2002)

1409–1412.

[15] D. Kereiakes, G. Montalescot, E. Antman, M. Cohen, H. Darius,

J. Ferguson, C. Grines, K. Karsch, N. Kleiman, D. Moliterno, P.

Steg, P. Teirstein, F. Van de Werf, I. Wallentin, Am. Heart J. 144

(2002) 615–624.

[16] H. Kock, A. Handschin, Clin. Appl. Thromb. Hemost. 8 (2002)

251–255.

[17] V. Rodie, A. Thomson, F. Stewart, A. Quinn, I. Walker, I. Greer,

Br. J. Obstet. Gynecol. 109 (2002) 1020–1024.

[18] R. Hull, G. Pineo, P. Stein, Int. Angiol. 17 (1998) 213–224.

[19] J. Hirsch, T. Wendt, P. Kuhly, W. Schaffartzik, Anaesthesia 56

(2001) 760–763.

[20] D. Giavarina, M. Carta, A. Fabbri, J. Manfredi, E. Gasparotto,

G. Soffiati, Perfusion 17 (2002) 23–26.

[21] M. Wallock, W. Jeske, M. Bakhos, J. Walenga, Perfusion 16

(2001) 147–153.

[22] C. Edstrom, J. McBride, D. Theriaque, K. Kao, R. Christensen, J.

Perinatol. 22 (2002) 475–477.

[23] Z. Zhong, E. Anslyn, J. Am. Chem. Soc. 124 (2002) 9014–

9015.

[24] D. Murray, W. Brosnahan, B. Pennell, D. Kapalanski, J.

Weiler, J. Olson, J. Cardiothorac. Vasc. Anesth. Online 11

(1997) 24–28.

[25] M. Furubashi, N. Ura, K. Hasegawa, H. Yoshida, K. Tsuchib-

ashi, T. Miura, K. Shimamoto, Nephrol. Dial. Transplant. 17

(2002) 1457–1462.

[26] B. Yang, C. Hall, B. Yang, R. Savani, E. Turley, J. Cell Biochem.

56 (1994) 455–468.

[27] B. Yang, L. Zhang, E. Turley, J. Biol. Chem. 268 (1993) 8617–

8623.

[28] H. Maeda, H. Fujita, Y. Sakura, K. Miyazaki, M. Goto, Biosci.

Biotechnol. Biochem. 63 (1999) 892–895.

[29] M. Jairajpuri, A. Lu, U. Desai, S. Olson, I. Bjork, S. Bock, J. Biol.

Chem. 278 (2003) 15941–15950.

[30] D.J. Gillooly, A. Simonsen, H. Stenmark, Biochem. J. 355 (2001)

249–258.

[31] T. Pouyani, G.D. Prestwich, Bioconjugate Chem. 5 (1994) 370–

372.

[32] S. Olson, I. Bjork, J. Biol. Chem. 266 (1991) 6353–6364.

A selective protein sensor for heparin detection

Documents

Transcript of A selective protein sensor for heparin detection