ANALYTICAL BIOCHEMISTRY 121, 244-250 (1982)

Affinity Chromatography of Heme-Binding Proteins: An Improved Method

for the Synthesis of Hemin-Agarose’

KENTSUTSUI’ AND GERALDC. MUELLER~

McArdle Laboratory for Cancer Research, The University of Wisconsin, Madison, Wisconsin 53706

Received August 11, 198 I

Hemin was conjugated to aminoethyl-agarose with high efficiency using l,l’-carbonyldi-

imidazole as a condensing agent. The hemin-agarose was an efficient affinity resin for certain
heme-binding proteins. It retained 4-5 mg of globin per milliliter of packed gel, whereas
hemoglobin did not bind to the resin. The specificity of the adsorbent was further demonstrated
by the selective binding of hemopexin from the whole sera of several species of a new heme-
binding protein (HBP.93) from rabbit serum. Serum albumin, a major protein in serum which
can also bind heme, did not bind to the adsorbent.

A high-capacity hemin-agarose has been lular heme-binding proteins. The ease of

prepared for the selective adsorption and preparation, high capacity, and selectivity
identification of intracellular heme-binding for heme-binding proteins of this affinity
proteins that might play a role in the ter- resin are compared with those of other prep-
minal differentiation of erythroid cells (1). arations that have been described in earlier
This was accomplished using water-free di- publications (2-6).
methylformamide to solubilize hemin during
the coupling to aminoethyl-agarose with EXPERIMENTAL PROCEDURES
1, I’-carbonyldiimidazole. The resulting af-
Materials. Aminoethyl-agarose and hemin
finity resin contains 2.2 pm01 of covalently
chloride (type I) were obtained from Sigma
bound hemin, bound through a single car-
Chemical Company (St. Louis, MO.). l,l’-
boxy1 group of the hemin, per milliliter of
Carbonyldiimidazole ( CDI)4 and N&V-di-
agarose. The resin exhibits a high and spe-
methylformamide (DMF, spectrophotomet-
cific capacity for binding globin, hemopexin,
ric grade) were purchased from Aldrich
and a new heme-binding protein from rabbit
Chemical Company (Milwaukee, Wise.) and
serum (15) while excluding serum albumin.
used without purification.
The hemin-agarose produced in this proto-
Mouse hemoglobin was purified from lysed
col promises also to be a useful reagent for
red blood cells by CM-Sephadex chroma-
the concentration and isolation of intracel-
tography (7) and globin was prepared by
treating a hemoglobin solution with cold
I This work was supported by Grants CA-07 175 and HCl-acetone (8). Tetrameric globin was re-
CA-23076 and Training Grants CA-09020 and CA- constituted by dialyzing a dissociated globin
09230 from the National Cancer Institute. solution against 0.01 M sodium phosphate
’ Present address: Department of Biochemistry, Can-
cer Institute, Okayama, University Medical School, 2- buffer, pH 6.8 (7).
5-1 Shikata-Cho, Okayama 700, Japan.
3 Recipient of a Research Career Award, CA-00685, 4 Abbreviations used: CDI, I, I’carbonyldiimidazole;
from the National Cancer Institute. To whom requests DMF, N,N-dimethylformamide; EDC, I-ethyl-3-(3-di-
for reprints should be sent. methylaminopropyl)carbodiimide.

244
0003-2697/82/060244-07$02.00/O
Copyright 0 1982 by Academic Press, Inc.
All rights of reproduction in any form reserved.
 IMPROVED METHOD FOR
Synthesis of hemin-agarose. Hemin (200
mg) was dissolved in DMF ( 15 ml) and CD1
(200 mg) was added immediately. The mix-
ture was heated at 80°C for 15 min and then
cooled at room temperature for 30 min. The
derivatized hemin in DMF was added to 10
ml of aminoethyl-agarose (4.5 pmol diami-
noethane/ml packed gel), which had been
washed successively with distilled water (200
ml), 33% DMF (50 ml), 66% DMF (50 ml),
and finally 100% DMF ( 100 ml). The cou-
pling reaction was allowed to proceed at
room temperature for 18 h while the tube
was shaken gently. The brownish colored
beads were washed on a sintered glass funnel
with decreasing concentrations of DMF:
100% DMF (100 ml), 66% DMF (50 ml),
and 33% DMF (50 ml). After washing with
distilled water (200 ml), the resin was re-
washed with 25% pyridine until the wash had
no color (about 200 ml) and finally with
distilled water (200 ml). The hemin-agarose
can be stored at 4°C with appropriate pre-
servatives. While controlled studies have not
been conducted, it is recommended as a pre-
caution against degradation that light and
oxygen be excluded during storage.
Determination of hemin immobilized on
agarose. The conjugated hemin was released
from the agarose by heating the hemin-
agarose (50 ~1) in 0.1 M NaOH (3 ml) at
75°C for 30 min. After cooling to room tem-
perature, 1 ml of pyridine was added and the
residual agarose beads (colorless) were re-
moved by centrifugation. The hemin deriv-
ative released from the matrix was converted
to pyridine hemochrome by reducing the fer-
ric iron with sodium dithionite and the ab-
sorbance at 420 nm was recorded immedi-
ately. For standards, 1 mM heme in pyridine
(freshly prepared) was mixed with nonder-
ivatized aminoethyl-agarose and the mixture
treated in the same way as the hemin-aga-
rose.
Detection of heme-binding proteins in
serum. Serum (50 ~1) was shaken with 40
~1 of hemin-agarose (added as a 50% sus-
pension in 0.5 M NaCl- 10 mrvi sodium phos-
HEMIN-AGAROSE SYNTHESIS 245
phate, pH 7.5), either directly or after prein-
cubation with 2.5 nmol of hemin at 37°C for
20 min. The hemin-agarose beads were
washed three times with 0.5 M NaCl-10 mM
sodium phosphate, pH 7.5 (0.5 ml) and the
adsorbed protein was released and solubi-
lized with 100 ~1 of electrophoresis sample
buffer containing 1% SDS, 2.5% 2-mercap-
toethanol, 0.0625 M Tris-HCl (pH 6.8),
and 0.25 M sucrose. Sample eluates were
boiled for 2 min before loading on the gel.
SDS-polyacrylamide gel electrophoresis was
performed according to the method of
Laemmli (9).
RESULTS
Synthesis and Properties of Hemin-
Agarose
Preliminary attempts to couple hemin to
aminoethyl-agarose were carried out using
1 - ethyl - 3 - (3 - dimethylaminopropyl)car-
bodiimide (EDC) as the coupling agent, as
is usually recommended for the conden-
sation of carboxyl groups (10); however, this
reagent requires a pH of 4-6 for optimal
coupling. At this pH range, solubility prob-
lems with hemin were encountered even
when dimethyl sulfoxide, dioxane, or di-
methylformamide-water mixtures were used
as the solvent phase. As a result the coupling
efficiency was rather low.
These problems were largely eliminated
by using 1, I’-carbonyldiimidazole (CDI) as
the coupling agent in pure DMF, as is prac-
ticed in the esterification of fatty acids ( 11).
The carboxyl groups of hemin are deriva-
tized initially to imidazolides with CD1 in
DMF. The progression of the reaction is vi-
sualized by the conversion of a brownish
hemin solution to one with a red tinge. This
color change is probably caused by the co-
ordination of the heme iron with imidazole
that is released from CDL For the coupling
reaction, a 13-fold excess of the hemin acyl-
imidazolide was used with respect to the
number of aminoethyl groups on agarose.
This increases the likelihood of having one
246 TSUTSUI AND MUELLER

could not be eluted with hemin solution due

to an extensive adsorption of free hemin to
the resin. Figure 2 shows that there is a sig-
nificant level of heme-binding sites on the
hemin-agarose. The binding very likely oc-
curs through the interaction of free hemin
with matrix-bound hemin since aminoethyl-
agarose did not bind significant amounts of
hemin. At saturation, about l-2 mol of
hemin was bound per mole of immobilized
hemin. This hemin was readily eluted by
washing the column with 8 M urea. It re-
400 450 MO 550 600
WNE LENGTH (“In)
quires further investigation to clarify the
mechanism for the failure of free hemin to
FIG. 1. Comparison of the absorption spectrum of
displace proteins adsorbed on the resin. We
pyridine hemochromes from hemin, before and after
coupling with agarose. Hemin was released from hemin- have not pursued this phenomenon since our
agarose by hydrolysis with NaOH as described under goal was to obtain heme-free apoproteins for
Experimental Procedures. The adsorption spectrum of studies on the effects of heme on their sur-
the hemin was measured in pyridine after reduction with vival and subcellular associations.
sodium dithionite. The spectrum was plotted on a com-
mon abscissa in which the absorption at 600 nm is the
baseline. (a) Hemin derivative released from hemin- Interaction of Globin with Hemin-Agarose
agarose by alkali treatment, (b) hemin.
The retention of a heme-binding protein
to hemin-agarose was examined using re-
free carboxyl group on each immobilized
hemin molecule. After the coupling reaction,
the hemin which was not covalently linked
was removed by washing the derivatized 50- A
agarose with 25% (v:v) pyridine:H,O, fol-
lowed by distilled water. The bead structure 25
of the agarose was retained throughout these 20
preparative steps.
To measure the level of covalently at-
tached hemin, the hemin-agarose was treated
with alkali and the released hemin derivative
was converted to pyridine hemochromes for
spectrophotometric determination (Fig. 1).
The almost identical absorption pattern with
the starting hemin indicated that the reso- 0 I
HEMIN
2
(ml)
3 4 0 I
“REAh:,

nance structure of the porphyrin ring, in-

FIG. 2. Binding of hemin to hemin-agarose. Columns
ciuding the coordination with the central (0.25 ml) of aminoethyl-agarose and hemin-agarose
iron, remained intact throughout the syn- were saturated by treatment with a solution of 0.5 mM
thetic process. Quantitation of the released hemin in 0.5 M NaCI-IO mM sodium phosphate (pH
hemin showed that the hemin-agarose con- 7.5). Hemin concentration in the eluate was estimated
from the absorbance at 390 nm (A3w). After the A390
tains about 2.2 pmol hemin covalently at-
reached a constant level the columns were washed with
tached to 1 ml of agarose. 10 ml of the buffer and the adsorbed hemin was eluted
During the course of our study, we noticed with 8 M urea. Eluate from aminoethyl-agarose (0),
that proteins bound on the hemin-agarose from hemin-agarose (0).
 IMPROVED METHOD FOR HEMIN-AGAROSE SYNTHESIS 247

bin and matrix-bound hemin is mainly hy-

-n-
GLOBIN (ml) UREA bnr,
FIG. 3. Binding of globin to hemin-agarose. Columns
(0.25 ml) of aminoethyl-agarose and hemin-agarose
were saturated by treatment with a solution of I mg/
ml globin in NP buffer. The globin concentration in the
eluate was monitored by measuring the absorbance
280 nm (&a). After the &,, reached a plateau,
columns were washed with IO ml of NP buffer, and the
globin was eluted with 8 M urea. Eluate from amino-
ethyl-agarose (0) from hemin-agarose (0).
natured globin as a model protein. To pre-
vent nonspecific ionic interaction between
protein and the affinity resin, the binding was
carried out in a high-ionic-strength buffer
(0.5 M NaCl and 10 mM sodium phosphate,
pH 7.5) (NP buffer). Passage of a globin
solution through a column of hemin-agarose
resulted in a delayed appearance of globin
in the eluate due to a highly effective binding
of globin to the column (Fig. 3). Bound glo-
bin could be eluted, in turn, from the column
with 8 M urea. No binding was observed on
the control column containing aminoethyl-
agarose. The capacity of the hemin-aga-
rose for globin was 4-5 mg protein/ml
packed gel.
A variety of conditions were tested for the
elution of bound globin. Nearly complete
elution was attained with 8 M urea or 1%
SDS; however, 20-25s of globin was re-
tained on the resin when 0.1 M acetic acid
was used as the eluting agent. On the other
hand, practically no protein was eluted with
a highly chaotropic salt such as 3 M NaSCN,
suggesting that the interaction between glo-
drophobic in nature.
Specificity of hemin-agarose for the apo
form of a heme-binding protein has been
demonstrated by comparing the binding of
globin and hemoglobin to the resin (Fig. 4).
The observed binding was clearly dependent
on the absence of protein-bound heme. The
small amount of protein which was retained
on the column when a hemoglobin solution
was applied may have been due to the pres-
ence of some partially heme-depleted mol-
ecules of hemoglobin since it varied with the
preparation.
Identification of Heme-Binding Proteins in
Serum
at
the Human serum contains two major heme-
binding proteins, serum albumin and he-
mopexin (12); it also contains at least one
minor protein called histidine-rich glycopro-
tein, which has been shown recently to bind
more than 10 hemes per mole of protein
(13). To determine whether these proteins
could be retained specifically on hemin-aga-
rose, serum samples from man and several
other species were incubated with a small
amount of hemin-agarose and the bound
proteins analyzed by SDS-polyacrylamide
gel electrophoresis (Fig. 5). In the absence
of added hemin, the affinity column bound
FIG. 4. Selective binding of globin to hemin-agarose.
Solutions (0.5 ml) of globin and hemoglobin (1 mg/ml
in NP buffer) were applied to columns (0.25 ml) of
hemin-agarose and washed with NP buffer until the
A2800f the eluate returned to the baseline. Bound protein
was then eluted with 8 M urea after washing the columns
with IO ml of NP buffer. Globin (0) hemoglobin
(0).
248 TSUTSUI AND MUELLER

SERUM CALF F!AT HUMAN RABBIT similar to that of hemopexin in the native
-II
HEWN gel system (Fig. 6~). We have recently pu-
TREATMENT ; - - + - +
rified this protein (HBP.93) by affinity chro-
matography on hemin-agarose ( 15). Several
lines of evidence have suggested that HBP.93
**p #@‘-HBP(93K)
has about 30 heme-binding sites per mole-
cule. This observation is consistent with the
* -~ --Hpx(GSK)
fact that the binding of HBP.93 was not in-
hibited by the presence of hemin in a low
concentration (Fig. 5, lane 6) although the
I*---
binding was completely abolished with a
higher hemin concentration (data not shown).
FIG. 5. Selective binding of serum heme-binding pro- The other minor bands seen on the gel are
teins to hemin-agarose. The indicated sera were incu-
bated with hemin-agarose with (+) or without (-)
probably the proteins nonspecifically bound
preincubation with hemin as described in the text. After to the resin since the binding of these pro-
removing the nonadsorbed proteins by washing the teins was not precluded by preincubating the
beads with NP buffer, the adsorbed proteins were eluted serum with a large dose of hemin.
with a buffer containing 1% SDS and analyzed by elec-
trophoresis in 7.5% polyacrylamide gels. Protein bands DISCUSSION
corresponding to the hemopexin (Hpx) in calf, rat, hu-
man, and rabbit sera and the heme-binding protein of The preparation of hemin-agarose using
rabbit serum (HBP) are labeled. CD1 as a condensing agent has several ad-
vantages over the procedures using EDC
a protein with an apparent molecular weight
RABBIT HUMAN
of 65,000-70,000 from all sera tested (lanes I
1-3, and 5). Since serum albumin and he- ‘S a b c IS
mopexin migrate to similar positions in the
SDS-gel system, the bound proteins from
human and rabbit sera were further electro-
phoresed in a native gel system to separate
these proteins (Fig. 6). Most of the stained
material was detected at the position of he-
mopexin and no band corresponding to al-
bumin was demonstrable. The identity of the
protein as hemopexin was further confirmed -
by amino acid analysis (Table 1). The cor- -*
responding coefficient between the corre- FIG. 6. Absence of binding of albumin to hemin-
sponding amino acid values was 0.94. Thus, agarose. Rabbit serum and human serum (5 ml each)
the hemin-agarose binds hemopexin but not were diluted with NP buffer to 20-fold and loaded on
serum albumin. When serum was preincu- small columns (2.0-ml bed) of hemin-agarose. After
washing the columns with 500 ml of NP buffer, the
bated with just enough hemin to saturate the
adsorbed proteins were eluted and fractionated by ap-
binding site of hemopexin (one site per mol- plying a decreasing pH gradient (pH 7.5-3.5). Fractions
ecule), the binding of this protein to the containing proteins were pooled, dialyzed against dis-
hemin-agarose was abolished (Fig. 5, lanes tilled water, and then subjected to electrophoresis in
4 and 6). a 7.5% polyacrylamide gel prepared according to
In rabbit serum, another protein was re- Laemmli (9). omitting SDS. Whole serum (S); rabbit
serum proteins eluted at pH 6.3-5.3 (a), pH 5.2-4.8
tained by the resin (Fig. 5, lanes 5 and 6); (b), pH 4.7-4.1 (c), and human serum proteins eluted
this protein had a molecular weight of ap- at pH 6.3-5.3 (d), and pH 5.1-4.1 (e). Serum albumin
proximately 93,000 and showed a mobility is marked by the arrows.
 IMPROVED METHOD FOR HEMIN-AGAROSE SYNTHESIS 249

TABLE 1 fraction when rabbit serum was fractionated

COMPARISON OF THE AMINO ACID COMPOSITION OF on the column. This hemin appears to have
THE~~,OOOM,PROTEIN FROM RAEIIX~SERUMWITH had its origin in the adsorbed HBP.93, which
THAT OF RABBIT HEMOPEXIN
binds extra hemin on the sites and is released
Mole percentage from this protein by the elution procedure.
During the preparation of this manuscript,
65,000 M, the synthesis of a similar hemin-agarose
Amino acid protein” Hemopexinb
adsorbent was reported by Olsen (5). In con-
Il.2 11.2 trast to our observation, this hemin-agarose
Asp
Thr 5.6 5.8 preparation was reported to bind serum al-
Ser 10.6 1.9 bumin. In this case, a matrix with a longer
Glu 11.1 9.4 spacer (diaminohexane) was used to conju-
Pro 6.3 7.1
gate the hemin; in addition, a lower-ionic-
GUY 12.5 10.4
Ala 1.2 6.4 strength buffer was employed in their bind-
Val 5.8 6.4 ing study. While the differences in the per-
Ile 2.8 2.8 formance of the two resins are not resolved,
Lcll 8.7 9.1 the difference in affinity of the resins may
TY~ 1.8 3.5
4.6
be preliminary evidence that albumin has a
Phe 2.6
His 2.1 4.3 deeper heme-binding pocket than either
LYS 5.1 5.6 hemopexin or HBP.93.
Arg 6.7 5.6 For the elution of proteins bound on
hemin-agarose, it appears that an acidic
a The 65,000 M, protein was separated from the
93,000 M, polypeptide by a pH-gradient elution as de- buffer is superior to other protein denatur-
scribed in the legend to Fig. 6 (the 65K component ants, although 8 M urea was exclusively used
eluted at pH 6.3-5.3). Lyophilized sample (about 500 in this study to elute globin. For the proteins
rg protein) was hydrolyzed in 6 N HCI at 110°C for 24 studied, a relatively mild acidity was re-
h and the hydrolysate analyzed on a Beckman amino
quired for the elution (e.g., hemopexin and
acid analyzer. Contents of cysteine, methionine, and
tryptophan were not determined. HBP.93 were released at about pH 6.0 and
b Based on the data of Hrkal and Muller-Eberhard 5.0, respectively). In this case a simple di-
(14). Values for the above three amino acids were alysis of the eluted proteins against neutral
omitted from the calculation to compare with the com- buffer fully restored their heme-binding ac-
position of the 65K protein.
tivity (15). Accordingly, the batch adsorp-
tion method, as described in Fig. 5, provides
(5,lO). In the final instance, there is no need a useful method for demonstrating the pres-
to adjust the pH during the acylimidazolide ence of heme-binding proteins in complex
formation since it is not a proton-consuming protein mixtures. Finally, the preequilibra-
reaction like the one with EDC. In addition, tion of a protein with varying levels of hemin
the coupling efficiency with CD1 was about can provide information on the number of
10 times higher than that obtained with heme-binding sites and their relative affinity
EDC. The time required to remove the un- for heme. This technique can also be used
reacted hemin from the hemin-agarose was to estimate the ratio of apo and heme-bound
significantly reduced by introducing a 25% forms of hemoproteins in a crude prepara-
pyridine wash. It is thus possible to synthe- tion when combined with radioimmunoas-
size the resin in 2 days, provided that ami- says.
noethyl-agarose is available.
Although no hemin was released from the ACKNOWLEDGMENTS
hemin-agarose column with any of the elu- We thank Dr. Charles B. Kasper for the amino acid
tion media used here, a small amount of analysis and Mary LeMahieu for her assistance in the
hemin was always eluted with the hemopexin preparation of the manuscript.
250 TSUTSUI AND MUELLER

REFERENCES L., and Moldave, K., eds.), Vol. 12B, pp. 747-
769, Academic Press, New York.
1. Lo, S., Aft, R., and Mueller, G. C. (1981) Cuncer 9. Laemmli, U. K. (1970) Nature (London) 227,680-
Rex 41, 864-870. 685.
2. Suttnar, J., Hrkal, Z., Vodrazka, Z., and Rejnkova, 10. Cuatrecasas, P., and Anfinsen, C. B. ( 1971) in
J. (1979) .I Chromatogr. 169, 500-504. Methods in Enzymology (Jackoby, W. B., ed.),
Vol. 22, pp. 345-378, Academic Press, New
3. Suttnar, J., Hrkal, Z., Vodrazka, Z. (1977) J.
York.
Chromarogr. 131,453-457.
II. Ko, H., and Royer, M. (1974) J. Chromatogr. 88,
4. Conway, T. P., and Muller-Eberhard, U. (1973)
253-263.
Fed. Proc. 32, 469.
12. Muller-Eberhard, U., and Morgan, W. T. (1975)
5. Olsen, K. W. (1980) Anal. Biochem. 109,250-254.
Ann. N.Y. Acad. Sci. 244, 624-650.
6. Olsen, K. W., Carey, D., Thompson, N., and Hutch- 13. Morgan, W. T. ( 1978) Biochim. Biophys. Acta 535,
inson, M. E. (1978) Fed. Proc. 37, 1514. 319-333.
I. Winterhalter, K. H., and Huehns, E. R. (1964) J. 14. Hrkal, Z., and Muller-Eberhard, U. (1971) Bio-
Biol. Chem. 239, 3699-3705. chemistry 10, 1746- 1750.
8. Schapira, G., Rosa, J., Malenknia, N., and Podieu, 15. Tsutsui, K., and Mueller, G. C. (1982) J. Biol.
P. ( 1968) in Methods in Enzymology (Grossman, Chem., in press.