Anal. Chem. 1991, 63, 627-629
627
Guanidinium- Induced Differential Kinetic Denaturation of Alkaline Phosphatase Isozymes Willie H. Lewis, Jr. and Sarah C. Rutan* Department of Chemistry, Box 2006, Virginia Commonwealth University, Richmond, Virginia 23284
This paper examines the solution kinetics of bovine intestinal and liver alkaline phosphatase (ALP) isozymes. 4-Methyiumbelliferyi phosphate is used as the substrate to study the differential kinetic behavior of ALP isozymes in the presence of guanidlnlum hydrochloride, a denaturant. The recursive Kalman filter algorithm for parameter estimation is used for analysis of the resulting kinetic data. A two-component first-order kinetic model with a zero-order component is used to successfully quantify intestinal and liver isozymes in synthetic mixtures. This work serves as a basis for the development of an electrophoresis separation method for ALP isozyme quantification with differential kinetic detection.
INTRODUCTION Alkaline phosphatase (ALP), orthophosphoric monoester phosphohydrolase (EC 3.1.3.1), is a zinc containing dimeric glycoprotein that catalyzes the cleavage of monophosphate esters ( I , 2). This enzyme is found in a wide range of mammalian species and has optimum activity a t an alkaline pH. Its actual physiological function is not known, but ALP is believed to be involved in the transport of phosphate across cell membranes. ALP levels in serum and other biological matrices were shown well over 50 years ago to be of utmost clinical significance in diagnostic enzymology. T h e greatest concentration of ALP isozymes in serum is derived from the liver, bone, intestinal, and placental tissues. T h e goal of the work described here is to achieve a n understanding of bovine ALP catalyzed kinetics in solution. Guanidinium hydrochloride is used as a denaturant to discriminate between bovine intestinal and liver ALP isozymes. Ultimately, the kinetic behavior observed in these exploratory solution studies will be used as a model for solid surface kinetics on low-resolution electrophoretic media, which will be used to effect the separation of ALP isozymes in human serum. This approach should allow for the quantification of ALP isozymes, even in the event of partially resolved bands and in the presence of rarer subforms such as cancer-associated isozymes. While bone ALP is normally severely overlapped with the liver subform, quantities of the bone form are not commercially available. Therefore, our initial studies have focused on the kinetic characterization of mixtures of liver and intestinal bovine isozymes. Other methods reported in the literature have exploited the kinetic and/or electrophoretic properties of the ALP isozymes in an attempt to quantify the different forms. Investigators have developed methods for quantifying ALP isozymes based on the discovery that certain chemicals differentially inhibit the specific isozymes (3-6). The inhibitors or denaturants that are commonly used include guanidinium hydrochloride (GuHCl), L-homoarginine, L-phenylalanine, L-tryptophan, and urea. Miggiano and co-workers studied the kinetics of concurrent instantaneous and time-dependent inhibition of bone isozyme by urea and related compounds. The resulting data indicated that GuHCl and thiourea are significantly more efficient for isozyme differentiation than urea. Samples with up to three components could be determined by monitoring
the time course of the reaction (5,6). Shephard and Peake have also studied the use of GuHCl as a protein denaturant and its potential to selectively discern ALP isozymes. They concluded that GuHCl was more potent and more stable and that better discrimination between intestinal and placental isozymes can be achieved with its use (6, 7). These investigators used a single-point activity determination for each isozyme present; however, this approach may yield results with low precision due to the propagation of error in solving the simultaneous equations (8). The mathematical rationale applied to the kinetic profiles obtained in this study has been well documented in the literature (4-12). Assuming Michaelis-Menten kinetics, with excess substrate present and first-order deactivation of the isozymes a t different rates, yields the following kinetic model
where AP is the difference between P,, the signal observed if the reaction goes to completion, and Po,the initial signal, and kd is the first-order deactivation constant. I and L denote the intestinal and liver isozymes, respectively, and (k&))I,,d denotes the activity of the undeactivated intestinal isozyme. The activity of each isozyme can be calculated from the product of AP and kd. The feasibility of using linear and nonlinear least-squares curve-fitting methods to process kinetic data that obey this type of model has been demonstrated in recent publications (12, 13-15).
EXPERIMENTAL SECTION Bovine intestinal and liver ALP isozymes and guanidine hydrochloride were obtained from Sigma Chemical Company (St. Louis, MO) and used as received. The substrate, 4-methylumbelliferyl phosphate (4-MUP),was also obtained from Sigma. This substrate is hydrolyzed by ALP to a highly fluorescent product, 4-methylumbelliferone,which exhibits strong fluorescence at 460 nm when excited at 370 nm. The rate of fluorescent product formation is used to determine ALP activity present in the original sample. 4-MUP is one of the most sensitive substrates known for ALP determinations and allows for detection of product formation early in the reaction. This makes 4-MUP extremely suitable for subsequent electrophoresis experiments. The buffer, 2-amino-2-methyl-1-propanol (2A2MlP), was supplied by Eastman Kodak Company (Rochester, NY). 2A2MlP was chosen as the buffer since this amino alcohol binds to the enzyme substrate complex, which can result in an acceleration of the reaction rate (16). A 0.90 M 2A2MlP solution was prepared in deionized water and the pH adjusted to 8.50 with concentrated HCl. ALP has an optimum activity at pH 10, but our best results have been obtained at pH 8.5. No reproducible results for K , values could be obtained at pH 10. All data discussed here are collected at M pH 8.5. The 2A2MlP buffer solution also contains 2.5 X Mg2+,since Mg2+plays a role in maintaining the structure required for ALP catalytic activity (2, 17). The concentrations of buffer and Mg2+are taken from the Gelman Sciences’ electrophoresis method for the qualitative determination of ALP isozymes (18). Activity is expressed in international units (U), which is defined as that amount of enzyme activity that will convert 1 pmol of substrate per minute under the conditions specified for the assay. The factor to convert the raw intensity values to U/L is 8.48 X IO3. Specific activities for the intestinal and liver isozymes were
0003-2700/91/0363-0627$02.50/0C 199 1 American Chemical Society
628
ANALYTICAL CHEMISTRY, VOL. 63, NO. 6, MARCH 15, 1991
24'0 20.0
16. 0
t
Table I. Regression Analysis slope
1
I
intercept
intestinal liver
Day 1 1.02 f 0.02 0.6 f 0.4 1.25 f 0.07 -1.5 f 0.7
intestinal liver
1.11 f 0.07 1.42 f 0.09
34
o J
o
0.48 0.84
Day 2 -1.4 f 1.2 -1.6 f 1.3
+/
l
eo.
160.
240.
Time
320.
400.
480.
560.
1.71 1.09
1
k
0.
std error
2-
(6)
Figure 1. Kinetic profiles for mixtures of intestinaMiver ALP isozymes in the presence of 2.30 M GuHCl where a is the pure intestinal isozyme, b-j are intestinal/liver mixtures, and k is the pure liver isozyme.
84-
0-
8.5 and 10 U/mg of solid, respectively (as determined by Sigma for p-nitrophenyl phosphate substrate). Isozyme solutions were prepared with 1-2 mg of protein/25 mL of 2A2MlP buffer. The resulting intestinal isozyme solution had an activity of 35 U/L, and the liver isozyme solution had an activity of 19 U/L for the 4-MUP substrate a t pH 8.5. Since we are operating under different assay conditions (pH and substrate), the U/L activity values for our approach are approximately 30-fold less than for the standard assay conditions for p-nitrophenyl phosphate at pH 10.5. A Farrand MK-2 spectrofluorometer (Valhalla, NY) interfaced to a Compupro 816-D computer (Viaysn, Hayward, CA) was used for data collection and was equipped with a thermostated cell compartment set at 30 "C. Experiments were conducted to obtain Michaelis-Menten constants, K,, for the various isozymes. For these studies, substrate concentrations ranged from 1.18 X to 1.57 X mM for the intestinal K , determination. Four concentrations within this range were selected, and kinetic profiles were run in duplicate at each concentration. For the liver K,, substrate concentrations ranged from 3.90 X to 1.56 X mM, and four concentrations within this range were also selected, but kinetic profiles were run in triplicate at each concentration. The Michaelis-Menten constants (K,) were obtained from Lineweaver-Burk plots. During the determination of the K , values, it was decided to carry out all experiments at pH 8.5. A possible reason for the irreproducibility of the results a t higher pH may be enhanced sensitivity to variations in other conditions (Le., Mg2+ concentration and temperature) at the optimal pH. The results at pH 8.5 were 0.028 f 0.003 mM and 0.012 f 0.002 mM for the intestinal and liver isozymes, respectively, where the K , value obtained for the intestinal isozyme is about twice that found for the liver isozyme. For all subsequent experiments, the substrate concentration used was 2.10 mM, which was approximately 100-foldgreater than the K , value found for the intestinal isozyme. To denature the isozymes, other kinetic experiments were conducted with 2.30 M GuHCl. Mixtures of intestinal and liver isozymes were prepared to contain the following volume percent composition (v/v): 90:10, 8020,7030,6040,50:50,4060,3070,2080, and 1090, respectively. Experiments were conducted on single-component isozyme solutions and isozyme mixtures with and without GuHCl present in the reaction cuvette. All reactions were monitored for 500 s, which is about 2 half-lives of the deactivation response for the intestinal isozyme. In all cases, the last reaction component added to the cuvette was the enzyme. Pascal Kalman filtering programs have been developed in our laboratory that are capable of calculating rate constants and AP values. Data were obtained by using 1 X 10" as the measurement variance and 1 X 10-'-1 X lo4 as variances on the initial guesses for the parameters. The quality of the fits was evaluated by using the variance of fit, defined as the mean square error of the residuals. For the nonlinear models, an iterative method was used where the best fit was defined as that which preceded the first increase in the variance of fit.
6-
2
6
10 14 18 22 THEORETICAL ACTIVITY (U/L)
26
30
Figure 2. Intestinal activity estimated in the presence of liver isozyme using the two-component plus zero-order model, where day 1 (+) and day 2 ( 0 )are plotted against theoretical activity (-).
RESULTS AND DISCUSSION Figure 1 is a graphical representation of the kinetic profiles for mixtures of intestinal and liver isozymes with GuHCl present. Similar experiments were done with no denaturant present. Those kinetic profiles were fit to a zero-order kinetic model, where the variances of fits ranged from 2 x lo4 t o 2 x 10-5. In the presence of 2.30 M GuHC1, all data were fit t o (a) first-order, (b) zero-/first-order, (c) two-component first-order, and (d) two-component first-order/zero-order kinetic models (eq 1). Models a and b require a nonlinear fitting procedure, while models c and d are linear functions and are based on presumed values for the deactivation constants. T h e most accurate results were obtained with model d. We fit the kinetic profiles for the pure intestinal and liver isozymes to a zero- plus first-order kinetic model. We obtained kd values of 2.70 X and 1.24 X lo-* s-l for the intestinal and liver isozyme, respectively. The intestinal isozyme had a substantial zero-order contribution, while the liver zero-order contribution was less than 5% of the total liver isozyme activity and was considered negligible. These deactivation constants were used to estimate the respective isozyme activities in the mixtures from fits to model d. The results showed an average difference for the intestinal isozyme of only 1.0 U / L between the true activity and the activity estimated by the model. T h e liver activities obtained resulted in an average difference of 1.1U/L. The variances of fits ranged from 2.3 X lo+ to 4.5 X lo+. We repeated these experiments under identical conditions several days later, using freshly prepared solutions. Values of 2.68 X and 1.22 X lo-* were obtained as deactivation constants for the pure intestinal and liver isozymes, respectively, which are in close agreement with the previous results. Once again, kinetic profiles for the mixtures were fit to model d. Average differences between the true activity and estimated activity were 1.0 and 2.9 U/L for intestinal and liver isozymes, respectively. The variances of fits ranged from 2.4 X lo4 to These results from the first and second data sets 5.6 X are graphically depicted in Figures 2 and 3. Table I includes the regression parameters and standard error of regression obtained for both data sets. The slopes of the curves were
ANALYTICAL CHEMISTRY, VOL.
23
I
t
I
0
+I
5 31
3
I
I
5
I
1
7
I
I
9
I
I
11
I
1
13 THEORETICAL ACTIVITY (U/L)
I
I
15
I
t
17
Figure 3. Liver activity estimated in the presence of intestinal isozyme using the two-component plus zeroorder model, where day 1 (+) and day 2 ( 0 )are plotted against theoretical activity (-).
greater than 1 in three of the four cases. Possible reasons for this behavior are discussed below. Studies were conducted to determine whether GuHCl a t 2.30 M caused instantaneous deactivation. While the pure component kinetic reactions were in progress, GuHCl was added to the reaction mixture and the resulting kinetic profiles were measured. When the data were fit to a zero-/first-order kinetic model, an enhancement of enzyme activity was observed for both isozymes. I t is our belief that this enhancement may be due to ionic strength effects, which result in an increase in activity during the early part of the reaction (19). This may be the reason for the slopes having values greater than one in Table I. As GuHCl continues to interact with the isozymes, the protein continues to unfold and the changes in secondary structure result in lower enzymatic activity.
CONCLUSIONS The ability to distinguish between ALP intestinal and liver isozymes has been demonstrated in the results presented above. The concentration of GuHCl used provided us with a deactivation rate consant ratio of 4.6 to 1,which we believe is sufficient to be able to differentiate between the intestinal and liver isozymes. This ratio allows us to detect intestinal isozyme in the presence of as much as 5-fold excess liver
63,NO. 6, MARCH 15. 1991
629
isozyme and provides us with a useful activity range of 5 (12). The kinetic profiles were studied for 2 half-lives of the intestinal isozyme, and good fits to our linear models were obtained. If we had used a nonlinear model, where the deactivation rate constants are not known a priori, the fitting range studied should be increased to a t least 3 half-lives to obtain reasonable quality fits (12). While the activities measured in these studies are significantly greater than the normal ranges for the isozymes, the S I N characteristics of fluorescence spectroscopy are such that isozyme activities at least 30-fold more dilute can be easily measured. Our improved understanding of ALP deactivation kinetics should allow us to develop improved approaches for quantitative analysis in conjunction with electrophoretic methods.
LITERATURE CITED (1) Anderson, R. A.; Bosron, W. F.; Kennedy, F. S.; Vallee, B. L. Proc. Natl. Acad. Sci. U.S.A. 1975, 72, 2989. (2) Bosron, W. F.; Anderson, R. A.; Falk, M. C.; Kennedy, F. S.; Vallee, B. L. Biochemistry 1977, 76, 610. (3) Miggiano, G. A. D.; Mordente, A,; Martorana, G. E.; Meucci, E.;Castelli, A. I t . J. Siochem. 1983, 32, 223-230. (4) Miggiano, G. A. D.; Mordente, A,; Pileri, M.; Martorana, G. E.: Meucci, E.: Castelli, A. Enzyme 1984, 32, 162-169. (5) Farley, J. R.; Chesnut, C. H.; Baylink, D. J. Clin. Chem. 1981, 27, 2002-2007. (6) SheDhard, M. D.: Peake, M. J.; Walmsley, R. N. J. Clin. Pathol. 1988, 39, '1025-1030. ( 7 ) Shephard, M. D. ; Peake, M. J.; Walmsley, R. N. J. Clin. Pathol. 1988, 39. 1031-1038. (8) TillyerlC. R. Cin. Chem. 1988, 34, 2490-2493. (9) Fernley, H. N.; Walker, P. G. Biochem. J. 1985, 9 7 , 95. (10) Weiser, W.E.; Pardue, H. L. Anal. Chem. 1986, 58, 2523-2527. (11) Harner, R. S.:Pardue, H. L. Anal. Chim. Acta 1981, 727, 23-38. (12) Fitzpatrick, C. P.; Pardue, H. L. Anal. Chem. 1989, 67,2551-2556. (13) Corcoran, C. A.; Rutan, S. C. Anal. Chem. 1988, 60, 1146-1153. (14) Rutan. S. C.; Fitzpatrick, C. P.; Skoug, J. W.; Weiser, W. E.; Pardue, H. L. Anal. Chim. Acta 1989, 224, 243-261. (15) Corcoran, C. A.; Rutan, S. C. Anal. Chim. Acta 1989, 224, 315-328. (16) Lowry, 0. H.; Roberts, N. R.; Wu, M.; Nixon, W. S.; Crawford. E. J. J. Biol. Chem. 1954, 207, 19. (17) McComb, R. B.; Bowers, G. N., Jr.; Posen, S. Alkaline Phosphatase: Plenum Press: New York, 1979; Chapter 7. (18) Alk-Pbos Isozyme Reagent Set: Gelman Sciences: Ann Arbor, MI, 1985. (19) McComb, R. B.; Bowers, G. N., Jr.; Posen, S. Alkaline Phosphatase; Plenum Press: New York, 1979; Chapter 8.
RECEIVED for review May 17,1990. Accepted December
21,
1990. This work was supported in part by Grant No. DEFG05-88ER13833 from the U S . Department of Energy.
