ANALYTICAL
196,
BIOCHEMISTRY
104-110
(1991)
Quantitative Determination of Albumin in Urine by On-Line lmmunoadsorptive Cleanup and Reversed-Phase Chromatography E. Reh,l
M. Kratzer,
Boehringer-Mannheim
Received
December
D. Zdunek,
and F. Lang
GmbH, Department
DB-VA, D-8132 Tutzing, Bahnhofstrasse
9-15, Germany
13,199O
Albumin in urine is selectively adsorbed on an immunoadsorber (human serum albumin-specific antibodies coupled covalently with a silica stationary phase) and after elution with 0.1% HCl is quantitatively determined by reversed-phase chromatography with detection of native fluorescence. The optimization of sample preparation and characteristics of the method such as recovery, linearity, reproducibility, detection limit, and selectivity are discussed. 0 1991 Academic Press, Inc.
derivatized with a purified analyte-specific immunoglobulin (2-7). The use of this selective sample preparation in combination with the separating performance of modern protein-RP phases should permit a reliable quantification of proteins in complex matrices. This paper shows that the presented method can serve as a candidate reference method for clinical hSA assays. MATERIALS
The importance of human serum albumin (hSA)2 quantification lies in its high diagnostic value for the differential diagnosis of kidney diseases (1). Exact quantification of the albumin content of urine serves, among other things, for the diagnosis of incipient renal disease by an elevation of albumin excretion (microalbuminuria) . The following results describe the development and the characterization of a quantitative fully automatic HPLC method with fluorescence detection for the determination of hSA in urine after on-line immunosorptive cleanup. Determination of low-molecular analytes by a combination of automated column valve switching and reversed-phase chromatography (RPC) is an often-described method for the analysis of complex materials. The range of application of this method can be considerably extended by replacing the RP cleanup column by an immunoadsorber, i.e., a silica carrier material
’ To whom correspondence should be addressed. r Abbreviations used: hSA, human serum albumin; RPC, reversedphase chromatography; TFA, trifluoroacetic acid; AN, acetonitrile; RSD, relative standard deviations; SDS, sodium dodecyl sulfate; IS, immunosorbent; HIA, homogeneous immunoassay.
AND METHODS
Chromatographic System
The chromatographic system was as follows: Chromatograph 1: a HP 109OL with cooled autosampler (HewlettPackard, Waldbronn, Germany); Chromatograph 2: a Gynkotek 600/200 (Gynkotek, Munich, Germany). The on-off valves were Rheodyne 7010 and 7067 tandem valves (Rheodyne, Cotati, CA). The Detector was a Hitachi 650-10 with an 18-~1 flow-through cell, X,,, = 280 nm, x = 340 nm (Hitachi Ltd., Tokyo, Japan). The data s;tem was a Maxima 820 (Waters, Milford, MA). The immunoadsorber was PAB-(anti-hSA)-S-IgG(IS), immunosorptively purified polyclonal antibodies on Spherosil (lOO-200 pm, mean pore diameter 300 nm); 10 X 4.6 mm i.d. (Boehringer-Mannheim, Mannheim, Germany). The HPLC column was a Synchropak RP 18 WP, 6.5 pm, 60 X 4.0 mm i.d. The precolumn was 5 X 4.6 mm i.d., dito (Synchrom, Linden, IN). Eluents
The eluents used were (A) 10 mM HCl in H,O; (B) 0.1% trifluoroacetic acid (TFA) in H,O; (C) 0.1% TFA in acetonitrile; (D) 50 mM KH,PO,/K,HPO, buffer, 50 mM NaCl, pH 6.8; and (E) 250 mM KH2P0,/K2HP0, buffer, pH 6.8.
104 All
Copyright 0 1991 rights of reproduction
ooo3-2697/91 $3.00 by Academic Press, Inc. in any form reserved.
CHROMATOGRAPHIC
DETERMINATION
OF
ALBUMIN
IN
105
URINE
Time Programming Time (min)
Flow rate (ml/min)
5 5
0
95
3 6
95
8
15
9
15
85 85
72 72 0
28 28 0
0 72 66 66 25 25 95
0 28 34 34 75 75 5
10 11
12 13 13.5 14 15 20 21 25 27 34 34.5 35 36
0.4
Valve
1
Valve
0
0
Notes
2 Binding
to adsorber
with
buffer
C
1
0
0.4 1.0 1.0 0.5 1 0.5
1 0
0
1.0
1.0 0.4
Washing adsorber with buffer D and bypass Washing adsorber with buffer D and bypass Washing adsorber with buffer D and bypass Bypass equilibration of RPC Bypass equilibration of RPC Supplying HCl eluent Supplying HCl eluent Supplying HCl eluent Precolumn switch into HCl eluent Desorption analyte onto RPC Start of RPC elution RPC gradient RPC gradient RPC gradient RPC gradient Reequilibration of equipment
cleaning cleaning cleaning
of RPC of RPC of RPC
STOP
Figure 1 shows the arrangement of the precolumn units. For the sake of simplicity valve 2 can be omitted and eluent E can be used for sample preparation and for washing.
Gerau, Germany). Human albumin (hSA) was from Behringwerke AG (Marburg, Germany). RESULTS
Reagents Hydrochloric acid Suprapur (HCl, 30%), potassium dihydrogen phosphate Suprapur, dipotassium hydrogen phosphate Suprapur, and sodium chloride were all from E. Merck (Darmstadt, Germany). Acetonitrile (AN) and trifluoroacetic acid (TFA) were from Baker (Gross-
t LI waste
Ive I
waste FIG.
1.
Arrangement
of the HPLC
system.
Optimization of Immunoadsorptive Sample Preparation Sample preparation by means of a precolumn can in principle be divided into three steps: (a) adsorption of the analyte, (b) rinsing of the matrix components, and (c) desorption for subsequent separation. Optimization of the conditions for adsorption, washing, and desorption of hSA was investigated. The immunoadsorber was regenerated simply by a changeover to adsorption buffer. Adsorption. The kinetics of the adsorption process constitute a characteristic feature of immunoadsorptive binding of proteins. Whereas the binding of a low-molecular analyte (e.g., digoxin) to the immobilized antibodies occurs quickly and without problems (2), the tertiary structure of proteins causes a change in the kinetic phenomena on interaction with the antibody. Not only do rate-determining transport processes take place, but the protein must also be present in the correct conformation for binding with the Fab part of the antibody. This results in a native milieu for adsorption (pH, temperature, etc.). A buffer of 50 mmol phosphate/50 mmol NaCl, pH 6.8, is therefore fairly uncritical as a mobile phase which transfers the analyte onto the immunoadsorber. Furthermore, the slowed-down kinetics must be matched by a slow flow rate in sample application. There is a remarkable loss due to unbound analyte
106
REH
ET
breaking through the immunoadsorber at a flow rate of 0.5 ml/min for three successive injections (6% at 1.3 pg hSA/injection) in comparison with the conditions at a flow rate of 0.2 mUmin. It is therefore necessary to reduce the flow rate for adsorption, whereas it can be increased again for desorption and subsequent separation. Rinsing. In addition to the specific interaction of the analyte with the antibody-binding site, nonspecific binding of substances in other antibody regions is often observed. The foreign substances adsorbed in this way can influence the separation after desorption and thus lower the selectivity of the method. To reduce such nonspecific binding rinsing stages, which do not break the specific antibody-hapten binding, but which wash out loosely bound nonspecific material, are included. Dilute detergents (Tween, Brij) and/or small proportions of organic solvents are often used for this purpose. An additional difficulty in sample preparation of protein analytes is that the analyte (in this case hSA) may not get denatured, because this may lead to incomplete dissolution in the desorption step. Figure 2 shows the reduction in recovery of hSA by rinsing with various contents of aeetonitrile in the rinsing solution. This is presumably the result of increased denaturation of hSA on the immunoadsorber, which is not reversible during the desorption process in transfer onto the separation column. The addition of Triton X-100 (I ml/liter) to the washing solution leads to large signals in blank gradients which overlap the hSA peak in RPC separation. Whereas this rinsing of the immunoadsorber during cleanup of low-molecular analytes (digoxin (2)) produced much better blank gradients, rinsing with addi-
of
recovery
4
6
0
10
acetonitrile FIG.
2.
Influence
of acetonitrile
AL.
tives is not advisable in the immunoadsorptive cleanup of proteins. It can only be advisable to prolong the adsorption mode (in this case with 250 mmol phosphate buffer). Desorption. An optimal desorption milieu should be selected for the desorption of proteins from an immunoadsorptive phase. In the first place, the binding of the antibody to the antigen (here hSA) should be broken without irreversible damage to the antibody (the cartridge is used several times). Second, the conditions must not be such that the analyte is too strongly denatured and thus binds irreversibly and nonspecifically to the immunoadsorber. Acidic desorption conditions have proven their worth in this respect, and in some circumstances a small proportion of organic solvent (e.g., 2propanol) can be added. A subtle balance of this compromise is illustrated in Fig. 3, where desorption occurs with increasing proportions of HCl in the desorbent mobile phase. Recovery is optimal with between 0.1 and 1~01% of hydrochloric acid. Higher HCI concentrations tend to cause increased irreversible denaturation, which is marked by increased carryover to the next analytical run. Figure 3b shows a continuous increase of the memory effect in blank injection, following sample injections. The following parameters have emerged for quantitative routine operation: Immunoadsorber-column dimensions: 10 X 4.6 mm i.d. Adsorption: 50 mmol KH,PO, + 50 mmol NaCl, pH 6.8, flow rate 0.2 ml/min, 6 min. Rinsing: 250 mmol KH,PO,, pH 6.8, flow rate 0.2 ml/ min, 4 min. Desorption: 0.1% HCl, flow rate 0.5 ml/min, 5 min.
hSA
from
the
immunadsorber
i2
content in the rinsing
14
[%
16
V/VI solution
on recovery.
10
20
CHROMATOGRAPHIC
a
1600
recovery
of
hSA
DETERMINATION
from
the
OF
ALBUMIN
0.10
0.20
0.30
0.40 HCl
FIG. 3.
Dependence on the HCI concentration
of hSA desorption and the magnitude of the desorption reagent.
0.50
content
of the Chromatography of hSA
Reversed-phase chromatography of larger proteins is often accompanied by poor peak symmetry and minor recovery. Comparing different suppliers proved the chosen stationary phase to be a good compromise. Indeed there is a small decrease in retention time (-0.5 min/lOO injections) and a small carry-over effect from run to run. This carryover has been quantitated to 2.3, 0.8, and 0% of the sample application peak area (three buffer injections without analyte after hSA chromatography). In order to minimize the influence of the memory effect, the first run of a triplicate determination was discarded. CHARACTERIZATION
Quantitation
OF THE
0.60
METHOD
of Aqueous hSA Standard
Linearity. Calibration chromatograms (Fig. 4) show linearity throughout the working range of O-18 pg of injected hSA (360 mg/liter). The regression equation is y = 1033.3 + 220808.4*x and the correlation coefficient r2 = 0.9976. Reproducibility. For the injection of aqueous hSA standard the following relative standard deviations in
of
hSA
0.70
to
the
0.80
next
run
0.90
1.00
[% V/V]
of the relative
The conditions represent an optimum for the immunoadsorber used. For every other immunoadsorber the sample preparation stage must be checked. This is only necessary once, as the clean-up phase is very often regenerated and used again, and after fresh packing with the same material does not change. Optimization
107
URINE
immunadsorber
,, carryover
0.00
IN
memory
effect
(first
buffer
injection
after
hSA chromatography)
series (RSD) were measured for the peak area: RSD = 2.3% (1 pg/40 ~1; n = 12). Stability. Immobilized antibodies are very stable and permit repeated use of a clean-up cartridge for many injections after regeneration in the LC. As a result of the strong antigen/antibody binding, denaturating conditions such as 0.1% HCl are necessary for rapid and complete desorption. However, it has been demonstrated that the immunoadsorber is stable over a large number of injections under these conditions. For example, repeated injection of 18 pg hSA from the same sample shows no decrease in recovery over a period of 67 h (111 injections)! The RSD for all measurements is 1.8%. A clean-up cartridge could thus be used for more than 300 injections of aqueous standard and urines. Only then was the immunoadsorber changed as a precaution. The stationary phase was also stable in its selectivity over all this time, and showed only the already mentioned shift in retention time. Investigation of urine samples. All urines were stored at 4°C and within 1 week they were analyzed after centrifuging, either undiluted or, for concentrations ~1000 mg/liter, suitably diluted. Linearity. Pooled normal urine was spiked with various concentrations of hSA. The resulting regression line has the following equation, when the natural hSA content of the urines is taken into account: y = 122815.7 + 41728.8 * c, (y = area in pV*s,
r2 = 0.9957, n = 32
c = concentration
in mg/liter).
108
REH
calibration
run
ET
for
AL.
hSA
1.00
0.60
0.60 Y ;
16000000
0.40
14000000 ?
lW
4000000--
4
I , 6
6
calibration
4.
Calibration
run
for
12
[pg/Inj of hSA
hSA
for
I 10
concentration
FIG.
graph
calibration
in
in aqueous
14
hSA
I
I
16
10
I 20
] solution.
urine
.bration
graph
for
hSA
in
urine
I
0
20
40
60 concentration
FIG.
5.
Calibration
100
[mg/l] of hSA
in urine.
120
CHROMATOGRAPHIC
DETERMINATION
OF
ALBUMIN
i
, 2.80
IN
109
URINE
4.00
g 3.00 : -b - 2.00 x
_. ____
1.00
O.OOi 2.60
2.80
3.00 x 10’
FIG.
6.
Chromatogram
of a urine
3.20
with
3.6 mg/liter
hSA.
I,
1,.
2.60
I,.
/I.
2.80 x 10’
3.00
III.
I
3.20
3.40
minutes
FIG. 7. Comparison of RP chromatograms immunosorptive sample preparation.
(a) with
3.00
and (b) without
3.20
3.40
minutes
FIG. 8. Comparison of RP chromatograms immunosorptive sample preparation.
(a) with
and (b) without
30 m X 0.5 mm, steel capillary; flow: Fl:F2 = 1:l; reagent: 0.1% TFA/40 mmol SDS/GO ~01% 2-propanol, 60°C). The peak area under reaction conditions can thus be compared during chromatography with the same sample injection into a capillary without a column. The recovery rate for a new immunoadsorber and new RP column is the same as after 18 h of operating time: 80%, n = 9, RSD = 0.9%. Using this denaturation detection for the analysis of a broad range of serum samples also demonstrated that in general application for serum albumin it was possible to neglect the structural influence of the protein on the quantitation and omit denaturation detection for routine analysis.
0
1
1 x 10’
The resulting calibration function is presented and three of the chromatograms are reproduced in Fig. 5. Detection limit. A detection limit of 0.4 mg/liter is calculated from the determination of the content of 3.6 mg/liter hSA in a pathological urine sample (Fig. 6) after the injection of 200 ~1. Reproducibility. After the injection of 40 ~1 of a pathological urine sample with an hSA content of 320 mg/liter a RSD of 1.3% (n = 12) was calculated for the peak area. Recovery. No reference method for checking the method was available. The recovery was determined by comparison of a capillary injection without column with peak area determination after chromatographic separation and adaptation of the fluorescence quantum yield using a denaturation-reaction detector. The fluorescence quantum yield of proteins depends very much on the degree of denaturation of the protein and on the physical properties of the medium (e.g., pH, viscosity, polarity). Comparable fluorescence quantum yields are obtained for identical proteins by complete on-line denaturation after separation before the detector (reactor:
2.40
2.80,
3.40
minutes
sample
1
100
50
correlation number
coefficient: of
data:
FIG. 9. Correlation tion method (IS-RPC) Boehringer-Mannheim)
150 200 HIA [w/l1
250
300
0.998
21
between the chromatographic determinaand homogeneous immunoassay (HIA, for urines.
REH ET AL.
110
Spiking of normal urine with hSA permits checking the recovery with respect to the urine matrix: Normal urine
Spiking
6.4 mg/liter
289.7
Relative recovery
Recovery
mg/liter
307.3
of
mg/liter
103.8%
Selectivity. Figures 7 and 8 (traces a) show that affinity chromatographic sample preparation leads to chromatograms which even on urine injection have virtually no interfering components besides the hSA peak. Figures 7 and 8 (traces b) show chromatograms of urine samples without immunoadsorptive cleanup. Only substances bound nonspecifically or cross-reactively to the immunoadsorber should produce interference in chromatography. Among the approximately 110 urine samples examined further peaks in the chromatogram occurred only in a few cases without influence on the quantitation. The cross-reactivity of the immobilized antibody was checked by the following spiking tests with normal urine: (a) h-transferrin added
Cross-reactivity
with
human
transferrin hSA concentration determined (mg/ml)
concentration (mg/ml) 20 40 100
2.0 1.8 1.6 (b)
Human
Cross-reactivity
with
human
IgG hSA concentration determined (mg/ml)
IgG concentration added (ma/ml)
6.3 6.1 6.1
50 100 150 (c) Cross-reactivity Human
IgA concentration added (mg/ml) 50 100 150
with
human
IgA hSA concentration determined (mg/ml) 6.4 6.7 6.9
The invariance following the addition of h-IgG at 6.2 mg/liter blank level can be evaluated positively; the increase with increasing quantities of IgA (above 6.2 mgl liter blank level) could be explained by an hSA impurity in the IgA (0.5%). The reduction in the recovery of hSA (beyond 2.0 mg/liter blank level) after the addition of
transferrin has not yet been explained. However, as the physiological range is about 5 mg/liter, an influence on the method due to different transferrin contents is unlikely. The analytical reliability of the quantitative determination of hSA is therefore considerably increased by cleanup with the immunoadsorber. Method comparison. The presented method (ISRPC) was compared with a turbidity test (HIA) for an automatic hospital instrument (Hitachi 717). Figure 9 shows the good correlations between these methods: IS-RPC = + 1.77 + 0.995 *HIA, r2 = 0.998 (standardized main component analysis). Conclusions. The presented results demonstrate the principal procedure in the development and characterization of selective chromatographic methods of protein analysis by means of affinity chromatographic sample preparation and analytical chromatography on RP columns. The virtually interference-free evaluation of the hSA peak substantially improves analytical accuracy, and reproducibility. The method was successfully applied for the determination of hSA in urines and standards over a period of 10 months, which is an indication of its robustness in routine use. The immunoadsorber was stored in a NaN; solution at 4°C and showed no significant change in properties over a period of more than 1 year. In addition, the use of small 10 X 4.6-mm cartridges for the immunoadsorber requires only small quantities of immobilized antibodies. It should be possible to adapt this simple and automatizable sample preparation method for proteins in complex matrices to a large number of other proteins in biological fluids. Reliable quantification of a large number of relevant proteins in biological systems would then become possible by combining a selective and reproducible cleanup procedure with a standard RP separation and a well-known, stable, direct fluorescence detection. REFERENCES 1. Weber, M. H. (1987) GZT Labo Me&in 2. Johansson, B. J. (1986) Chromatography
7/S, 327-335. 381, 107-113.
3. Reh, E. (1988) J. Chromatogr. 433, 119-1130. 4. Farjam, A., De Jong, G. J., Frei, R. W., and Brinkman, (1988) J. Chromatogr. 452,419-433. 5. Janis, L. J., and Regnier, F. R. (1988) J. Chromatogr. 6. Hayashi, T., and Sakamoto, S. (1989) Chromatographia 580. 7. Janis, L. J., Grott, A., Regnier, F. E., and Smith-Gill, J. Chromatogr. 476, 235-244.
U. A. Th. 444, l-11. 27,574S. J. (1989)
196,
BIOCHEMISTRY
104-110
(1991)
Quantitative Determination of Albumin in Urine by On-Line lmmunoadsorptive Cleanup and Reversed-Phase Chromatography E. Reh,l
M. Kratzer,
Boehringer-Mannheim
Received
December
D. Zdunek,
and F. Lang
GmbH, Department
DB-VA, D-8132 Tutzing, Bahnhofstrasse
9-15, Germany
13,199O
Albumin in urine is selectively adsorbed on an immunoadsorber (human serum albumin-specific antibodies coupled covalently with a silica stationary phase) and after elution with 0.1% HCl is quantitatively determined by reversed-phase chromatography with detection of native fluorescence. The optimization of sample preparation and characteristics of the method such as recovery, linearity, reproducibility, detection limit, and selectivity are discussed. 0 1991 Academic Press, Inc.
derivatized with a purified analyte-specific immunoglobulin (2-7). The use of this selective sample preparation in combination with the separating performance of modern protein-RP phases should permit a reliable quantification of proteins in complex matrices. This paper shows that the presented method can serve as a candidate reference method for clinical hSA assays. MATERIALS
The importance of human serum albumin (hSA)2 quantification lies in its high diagnostic value for the differential diagnosis of kidney diseases (1). Exact quantification of the albumin content of urine serves, among other things, for the diagnosis of incipient renal disease by an elevation of albumin excretion (microalbuminuria) . The following results describe the development and the characterization of a quantitative fully automatic HPLC method with fluorescence detection for the determination of hSA in urine after on-line immunosorptive cleanup. Determination of low-molecular analytes by a combination of automated column valve switching and reversed-phase chromatography (RPC) is an often-described method for the analysis of complex materials. The range of application of this method can be considerably extended by replacing the RP cleanup column by an immunoadsorber, i.e., a silica carrier material
’ To whom correspondence should be addressed. r Abbreviations used: hSA, human serum albumin; RPC, reversedphase chromatography; TFA, trifluoroacetic acid; AN, acetonitrile; RSD, relative standard deviations; SDS, sodium dodecyl sulfate; IS, immunosorbent; HIA, homogeneous immunoassay.
AND METHODS
Chromatographic System
The chromatographic system was as follows: Chromatograph 1: a HP 109OL with cooled autosampler (HewlettPackard, Waldbronn, Germany); Chromatograph 2: a Gynkotek 600/200 (Gynkotek, Munich, Germany). The on-off valves were Rheodyne 7010 and 7067 tandem valves (Rheodyne, Cotati, CA). The Detector was a Hitachi 650-10 with an 18-~1 flow-through cell, X,,, = 280 nm, x = 340 nm (Hitachi Ltd., Tokyo, Japan). The data s;tem was a Maxima 820 (Waters, Milford, MA). The immunoadsorber was PAB-(anti-hSA)-S-IgG(IS), immunosorptively purified polyclonal antibodies on Spherosil (lOO-200 pm, mean pore diameter 300 nm); 10 X 4.6 mm i.d. (Boehringer-Mannheim, Mannheim, Germany). The HPLC column was a Synchropak RP 18 WP, 6.5 pm, 60 X 4.0 mm i.d. The precolumn was 5 X 4.6 mm i.d., dito (Synchrom, Linden, IN). Eluents
The eluents used were (A) 10 mM HCl in H,O; (B) 0.1% trifluoroacetic acid (TFA) in H,O; (C) 0.1% TFA in acetonitrile; (D) 50 mM KH,PO,/K,HPO, buffer, 50 mM NaCl, pH 6.8; and (E) 250 mM KH2P0,/K2HP0, buffer, pH 6.8.
104 All
Copyright 0 1991 rights of reproduction
ooo3-2697/91 $3.00 by Academic Press, Inc. in any form reserved.
CHROMATOGRAPHIC
DETERMINATION
OF
ALBUMIN
IN
105
URINE
Time Programming Time (min)
Flow rate (ml/min)
5 5
0
95
3 6
95
8
15
9
15
85 85
72 72 0
28 28 0
0 72 66 66 25 25 95
0 28 34 34 75 75 5
10 11
12 13 13.5 14 15 20 21 25 27 34 34.5 35 36
0.4
Valve
1
Valve
0
0
Notes
2 Binding
to adsorber
with
buffer
C
1
0
0.4 1.0 1.0 0.5 1 0.5
1 0
0
1.0
1.0 0.4
Washing adsorber with buffer D and bypass Washing adsorber with buffer D and bypass Washing adsorber with buffer D and bypass Bypass equilibration of RPC Bypass equilibration of RPC Supplying HCl eluent Supplying HCl eluent Supplying HCl eluent Precolumn switch into HCl eluent Desorption analyte onto RPC Start of RPC elution RPC gradient RPC gradient RPC gradient RPC gradient Reequilibration of equipment
cleaning cleaning cleaning
of RPC of RPC of RPC
STOP
Figure 1 shows the arrangement of the precolumn units. For the sake of simplicity valve 2 can be omitted and eluent E can be used for sample preparation and for washing.
Gerau, Germany). Human albumin (hSA) was from Behringwerke AG (Marburg, Germany). RESULTS
Reagents Hydrochloric acid Suprapur (HCl, 30%), potassium dihydrogen phosphate Suprapur, dipotassium hydrogen phosphate Suprapur, and sodium chloride were all from E. Merck (Darmstadt, Germany). Acetonitrile (AN) and trifluoroacetic acid (TFA) were from Baker (Gross-
t LI waste
Ive I
waste FIG.
1.
Arrangement
of the HPLC
system.
Optimization of Immunoadsorptive Sample Preparation Sample preparation by means of a precolumn can in principle be divided into three steps: (a) adsorption of the analyte, (b) rinsing of the matrix components, and (c) desorption for subsequent separation. Optimization of the conditions for adsorption, washing, and desorption of hSA was investigated. The immunoadsorber was regenerated simply by a changeover to adsorption buffer. Adsorption. The kinetics of the adsorption process constitute a characteristic feature of immunoadsorptive binding of proteins. Whereas the binding of a low-molecular analyte (e.g., digoxin) to the immobilized antibodies occurs quickly and without problems (2), the tertiary structure of proteins causes a change in the kinetic phenomena on interaction with the antibody. Not only do rate-determining transport processes take place, but the protein must also be present in the correct conformation for binding with the Fab part of the antibody. This results in a native milieu for adsorption (pH, temperature, etc.). A buffer of 50 mmol phosphate/50 mmol NaCl, pH 6.8, is therefore fairly uncritical as a mobile phase which transfers the analyte onto the immunoadsorber. Furthermore, the slowed-down kinetics must be matched by a slow flow rate in sample application. There is a remarkable loss due to unbound analyte
106
REH
ET
breaking through the immunoadsorber at a flow rate of 0.5 ml/min for three successive injections (6% at 1.3 pg hSA/injection) in comparison with the conditions at a flow rate of 0.2 mUmin. It is therefore necessary to reduce the flow rate for adsorption, whereas it can be increased again for desorption and subsequent separation. Rinsing. In addition to the specific interaction of the analyte with the antibody-binding site, nonspecific binding of substances in other antibody regions is often observed. The foreign substances adsorbed in this way can influence the separation after desorption and thus lower the selectivity of the method. To reduce such nonspecific binding rinsing stages, which do not break the specific antibody-hapten binding, but which wash out loosely bound nonspecific material, are included. Dilute detergents (Tween, Brij) and/or small proportions of organic solvents are often used for this purpose. An additional difficulty in sample preparation of protein analytes is that the analyte (in this case hSA) may not get denatured, because this may lead to incomplete dissolution in the desorption step. Figure 2 shows the reduction in recovery of hSA by rinsing with various contents of aeetonitrile in the rinsing solution. This is presumably the result of increased denaturation of hSA on the immunoadsorber, which is not reversible during the desorption process in transfer onto the separation column. The addition of Triton X-100 (I ml/liter) to the washing solution leads to large signals in blank gradients which overlap the hSA peak in RPC separation. Whereas this rinsing of the immunoadsorber during cleanup of low-molecular analytes (digoxin (2)) produced much better blank gradients, rinsing with addi-
of
recovery
4
6
0
10
acetonitrile FIG.
2.
Influence
of acetonitrile
AL.
tives is not advisable in the immunoadsorptive cleanup of proteins. It can only be advisable to prolong the adsorption mode (in this case with 250 mmol phosphate buffer). Desorption. An optimal desorption milieu should be selected for the desorption of proteins from an immunoadsorptive phase. In the first place, the binding of the antibody to the antigen (here hSA) should be broken without irreversible damage to the antibody (the cartridge is used several times). Second, the conditions must not be such that the analyte is too strongly denatured and thus binds irreversibly and nonspecifically to the immunoadsorber. Acidic desorption conditions have proven their worth in this respect, and in some circumstances a small proportion of organic solvent (e.g., 2propanol) can be added. A subtle balance of this compromise is illustrated in Fig. 3, where desorption occurs with increasing proportions of HCl in the desorbent mobile phase. Recovery is optimal with between 0.1 and 1~01% of hydrochloric acid. Higher HCI concentrations tend to cause increased irreversible denaturation, which is marked by increased carryover to the next analytical run. Figure 3b shows a continuous increase of the memory effect in blank injection, following sample injections. The following parameters have emerged for quantitative routine operation: Immunoadsorber-column dimensions: 10 X 4.6 mm i.d. Adsorption: 50 mmol KH,PO, + 50 mmol NaCl, pH 6.8, flow rate 0.2 ml/min, 6 min. Rinsing: 250 mmol KH,PO,, pH 6.8, flow rate 0.2 ml/ min, 4 min. Desorption: 0.1% HCl, flow rate 0.5 ml/min, 5 min.
hSA
from
the
immunadsorber
i2
content in the rinsing
14
[%
16
V/VI solution
on recovery.
10
20
CHROMATOGRAPHIC
a
1600
recovery
of
hSA
DETERMINATION
from
the
OF
ALBUMIN
0.10
0.20
0.30
0.40 HCl
FIG. 3.
Dependence on the HCI concentration
of hSA desorption and the magnitude of the desorption reagent.
0.50
content
of the Chromatography of hSA
Reversed-phase chromatography of larger proteins is often accompanied by poor peak symmetry and minor recovery. Comparing different suppliers proved the chosen stationary phase to be a good compromise. Indeed there is a small decrease in retention time (-0.5 min/lOO injections) and a small carry-over effect from run to run. This carryover has been quantitated to 2.3, 0.8, and 0% of the sample application peak area (three buffer injections without analyte after hSA chromatography). In order to minimize the influence of the memory effect, the first run of a triplicate determination was discarded. CHARACTERIZATION
Quantitation
OF THE
0.60
METHOD
of Aqueous hSA Standard
Linearity. Calibration chromatograms (Fig. 4) show linearity throughout the working range of O-18 pg of injected hSA (360 mg/liter). The regression equation is y = 1033.3 + 220808.4*x and the correlation coefficient r2 = 0.9976. Reproducibility. For the injection of aqueous hSA standard the following relative standard deviations in
of
hSA
0.70
to
the
0.80
next
run
0.90
1.00
[% V/V]
of the relative
The conditions represent an optimum for the immunoadsorber used. For every other immunoadsorber the sample preparation stage must be checked. This is only necessary once, as the clean-up phase is very often regenerated and used again, and after fresh packing with the same material does not change. Optimization
107
URINE
immunadsorber
,, carryover
0.00
IN
memory
effect
(first
buffer
injection
after
hSA chromatography)
series (RSD) were measured for the peak area: RSD = 2.3% (1 pg/40 ~1; n = 12). Stability. Immobilized antibodies are very stable and permit repeated use of a clean-up cartridge for many injections after regeneration in the LC. As a result of the strong antigen/antibody binding, denaturating conditions such as 0.1% HCl are necessary for rapid and complete desorption. However, it has been demonstrated that the immunoadsorber is stable over a large number of injections under these conditions. For example, repeated injection of 18 pg hSA from the same sample shows no decrease in recovery over a period of 67 h (111 injections)! The RSD for all measurements is 1.8%. A clean-up cartridge could thus be used for more than 300 injections of aqueous standard and urines. Only then was the immunoadsorber changed as a precaution. The stationary phase was also stable in its selectivity over all this time, and showed only the already mentioned shift in retention time. Investigation of urine samples. All urines were stored at 4°C and within 1 week they were analyzed after centrifuging, either undiluted or, for concentrations ~1000 mg/liter, suitably diluted. Linearity. Pooled normal urine was spiked with various concentrations of hSA. The resulting regression line has the following equation, when the natural hSA content of the urines is taken into account: y = 122815.7 + 41728.8 * c, (y = area in pV*s,
r2 = 0.9957, n = 32
c = concentration
in mg/liter).
108
REH
calibration
run
ET
for
AL.
hSA
1.00
0.60
0.60 Y ;
16000000
0.40
14000000 ?
lW
4000000--
4
I , 6
6
calibration
4.
Calibration
run
for
12
[pg/Inj of hSA
hSA
for
I 10
concentration
FIG.
graph
calibration
in
in aqueous
14
hSA
I
I
16
10
I 20
] solution.
urine
.bration
graph
for
hSA
in
urine
I
0
20
40
60 concentration
FIG.
5.
Calibration
100
[mg/l] of hSA
in urine.
120
CHROMATOGRAPHIC
DETERMINATION
OF
ALBUMIN
i
, 2.80
IN
109
URINE
4.00
g 3.00 : -b - 2.00 x
_. ____
1.00
O.OOi 2.60
2.80
3.00 x 10’
FIG.
6.
Chromatogram
of a urine
3.20
with
3.6 mg/liter
hSA.
I,
1,.
2.60
I,.
/I.
2.80 x 10’
3.00
III.
I
3.20
3.40
minutes
FIG. 7. Comparison of RP chromatograms immunosorptive sample preparation.
(a) with
3.00
and (b) without
3.20
3.40
minutes
FIG. 8. Comparison of RP chromatograms immunosorptive sample preparation.
(a) with
and (b) without
30 m X 0.5 mm, steel capillary; flow: Fl:F2 = 1:l; reagent: 0.1% TFA/40 mmol SDS/GO ~01% 2-propanol, 60°C). The peak area under reaction conditions can thus be compared during chromatography with the same sample injection into a capillary without a column. The recovery rate for a new immunoadsorber and new RP column is the same as after 18 h of operating time: 80%, n = 9, RSD = 0.9%. Using this denaturation detection for the analysis of a broad range of serum samples also demonstrated that in general application for serum albumin it was possible to neglect the structural influence of the protein on the quantitation and omit denaturation detection for routine analysis.
0
1
1 x 10’
The resulting calibration function is presented and three of the chromatograms are reproduced in Fig. 5. Detection limit. A detection limit of 0.4 mg/liter is calculated from the determination of the content of 3.6 mg/liter hSA in a pathological urine sample (Fig. 6) after the injection of 200 ~1. Reproducibility. After the injection of 40 ~1 of a pathological urine sample with an hSA content of 320 mg/liter a RSD of 1.3% (n = 12) was calculated for the peak area. Recovery. No reference method for checking the method was available. The recovery was determined by comparison of a capillary injection without column with peak area determination after chromatographic separation and adaptation of the fluorescence quantum yield using a denaturation-reaction detector. The fluorescence quantum yield of proteins depends very much on the degree of denaturation of the protein and on the physical properties of the medium (e.g., pH, viscosity, polarity). Comparable fluorescence quantum yields are obtained for identical proteins by complete on-line denaturation after separation before the detector (reactor:
2.40
2.80,
3.40
minutes
sample
1
100
50
correlation number
coefficient: of
data:
FIG. 9. Correlation tion method (IS-RPC) Boehringer-Mannheim)
150 200 HIA [w/l1
250
300
0.998
21
between the chromatographic determinaand homogeneous immunoassay (HIA, for urines.
REH ET AL.
110
Spiking of normal urine with hSA permits checking the recovery with respect to the urine matrix: Normal urine
Spiking
6.4 mg/liter
289.7
Relative recovery
Recovery
mg/liter
307.3
of
mg/liter
103.8%
Selectivity. Figures 7 and 8 (traces a) show that affinity chromatographic sample preparation leads to chromatograms which even on urine injection have virtually no interfering components besides the hSA peak. Figures 7 and 8 (traces b) show chromatograms of urine samples without immunoadsorptive cleanup. Only substances bound nonspecifically or cross-reactively to the immunoadsorber should produce interference in chromatography. Among the approximately 110 urine samples examined further peaks in the chromatogram occurred only in a few cases without influence on the quantitation. The cross-reactivity of the immobilized antibody was checked by the following spiking tests with normal urine: (a) h-transferrin added
Cross-reactivity
with
human
transferrin hSA concentration determined (mg/ml)
concentration (mg/ml) 20 40 100
2.0 1.8 1.6 (b)
Human
Cross-reactivity
with
human
IgG hSA concentration determined (mg/ml)
IgG concentration added (ma/ml)
6.3 6.1 6.1
50 100 150 (c) Cross-reactivity Human
IgA concentration added (mg/ml) 50 100 150
with
human
IgA hSA concentration determined (mg/ml) 6.4 6.7 6.9
The invariance following the addition of h-IgG at 6.2 mg/liter blank level can be evaluated positively; the increase with increasing quantities of IgA (above 6.2 mgl liter blank level) could be explained by an hSA impurity in the IgA (0.5%). The reduction in the recovery of hSA (beyond 2.0 mg/liter blank level) after the addition of
transferrin has not yet been explained. However, as the physiological range is about 5 mg/liter, an influence on the method due to different transferrin contents is unlikely. The analytical reliability of the quantitative determination of hSA is therefore considerably increased by cleanup with the immunoadsorber. Method comparison. The presented method (ISRPC) was compared with a turbidity test (HIA) for an automatic hospital instrument (Hitachi 717). Figure 9 shows the good correlations between these methods: IS-RPC = + 1.77 + 0.995 *HIA, r2 = 0.998 (standardized main component analysis). Conclusions. The presented results demonstrate the principal procedure in the development and characterization of selective chromatographic methods of protein analysis by means of affinity chromatographic sample preparation and analytical chromatography on RP columns. The virtually interference-free evaluation of the hSA peak substantially improves analytical accuracy, and reproducibility. The method was successfully applied for the determination of hSA in urines and standards over a period of 10 months, which is an indication of its robustness in routine use. The immunoadsorber was stored in a NaN; solution at 4°C and showed no significant change in properties over a period of more than 1 year. In addition, the use of small 10 X 4.6-mm cartridges for the immunoadsorber requires only small quantities of immobilized antibodies. It should be possible to adapt this simple and automatizable sample preparation method for proteins in complex matrices to a large number of other proteins in biological fluids. Reliable quantification of a large number of relevant proteins in biological systems would then become possible by combining a selective and reproducible cleanup procedure with a standard RP separation and a well-known, stable, direct fluorescence detection. REFERENCES 1. Weber, M. H. (1987) GZT Labo Me&in 2. Johansson, B. J. (1986) Chromatography
7/S, 327-335. 381, 107-113.
3. Reh, E. (1988) J. Chromatogr. 433, 119-1130. 4. Farjam, A., De Jong, G. J., Frei, R. W., and Brinkman, (1988) J. Chromatogr. 452,419-433. 5. Janis, L. J., and Regnier, F. R. (1988) J. Chromatogr. 6. Hayashi, T., and Sakamoto, S. (1989) Chromatographia 580. 7. Janis, L. J., Grott, A., Regnier, F. E., and Smith-Gill, J. Chromatogr. 476, 235-244.
U. A. Th. 444, l-11. 27,574S. J. (1989)
