(Received for publication, October 8, 1996, and in revised form, December 3, 1996)
From the Department of Biochemistry, Medical College of Wisconsin, Milwaukee, Wisconsin 53226
We report the construction of an expression plasmid for rat mevalonate kinase and the overexpression of recombinant enzyme in Escherichia coli. The homogeneous enzyme had a specific activity of 30 units/mg and an observed subunit molecular mass of 42 kDa. The Michaelis constants (Km) for DL-potassium mevalonate (288 µM) and for ATP (1.24 mM) were in agreement with values reported for enzymes isolated from rat liver (Tanaka, R. D., Schafer, B. L., Lee, L. Y., Freudenberger, J. S., and Mosley, S. T. (1990) J. Biol. Chem. 265, 2391-2398). Recombinant rat mevalonate kinase was inactivated by the lysine-specific reagent, pyridoxal phosphate (PLP). ATP (5 mM) afforded protection against inactivation, suggesting reaction of PLP with an active-site lysine. Mapping, isolation, and Edman degradation of the ATP-protectable peptide from [3H]PLP-inactivated borohydride-reduced mevalonate kinase allow assignment of lysine 13, a residue invariant in known mevalonate kinase sequences, as the modification site. These results represent the first identification of an active-site residue in mevalonate kinase. The function of lysine 13 was evaluated by replacing this residue with methionine. Vm of the mutant protein is diminished by 56-fold, suggesting that lysine 13 facilitates catalysis. Kd values of wild-type and mutant proteins for ATP were determined in electron spin resonance competition experiments. The observed 56-fold diminution in affinity for the mutant enzyme supports an additional role for lysine 13 in stabilization of ATP binding.
Mevalonate kinase (EC 2.7.1.36) catalyzes the transfer of the
-phosphoryl group of ATP to the C-5 hydroxyl oxygen of mevalonic acid (1) to produce a key intermediate in the pathways for biosynthesis
of polyisoprenoids and sterols. Recently, the significance of this
enzyme has been elevated by the documentation of defects that arise
when production of 5-phosphomevalonate is either diminished or elevated
from the usual physiological range. Mevalonic aciduria is an inherited
metabolic disease that results from diminished mevalonate kinase
activity in infants and children (2). At least one missense mutation
(N301T) has been documented as the cause of the disease in the index
patient (3). Recently, diminished mevalonate kinase activity has been
implicated in a second disorder of sterol metabolism, Zellweger
syndrome (4). However, the molecular basis for the decreased mevalonate
kinase activity associated with this disease remains to be elucidated.
In contrast, human hepatocytes that overexpress mevalonate kinase as a
result of insertional activation by hepatitis B virus have an increased rate of cellular transformation. Presumably, this results from changes
in prenylation of hepatocyte signal transduction effectors (5). While
the above observations suggest that a detailed understanding of the
factors that modulate production of 5-phosphomevalonate would be
desirable, mevalonate kinase has been subjected to far less scrutiny
than other enzymes in the pathway for polyisoprenoid and sterol
biosynthesis.
Kinetic studies (6) suggest that mevalonate kinase catalyzes an ordered sequential reaction; mevalonic acid is the first substrate bound, and phosphomevalonate is the first product released. The downstream products of the polyisoprenoid pathway, farnesyl pyrophosphate and geranyl pyrophosphate are potent competitive inhibitors of mevalonate kinase with respect to ATP (7). The enzyme has been reported to contain reactive cysteine (6) and lysine (8) residues. However, despite the recent elucidation of cDNA sequences for mevalonate kinase from a variety of species, the identity and functional significance of those reactive amino acids remain to be determined.
Availability of a recombinant form of the enzyme would facilitate studies aimed at elucidation of the structure/function correlations that account for the reaction catalyzed by mevalonate kinase as well as for the features responsible for feedback inhibition. Using cDNA that encodes rat mevalonate kinase (9), we have developed a bacterial expression system that allows overexpression and facile isolation of the enzyme. The utility of this recombinant model system is illustrated by a series of experiments that account for the first identification and functional assignment of a mevalonate kinase active-site amino acid. A summary of a portion of this work has recently appeared (10).
Materials
The pGR50 plasmid harboring the full-length (1286 base pairs)
cDNA that encodes rat mevalonate kinase was a gift from the Bristol-Myers Squibb Co. Restriction endonucleases and T4 DNA ligase
were obtained from New England Biolabs Inc. Deoxyoligonucleotides were
synthesized by the Protein and Nucleic Acid Shared Facility at the
Medical College of Wisconsin. The pET-3d expression plasmid along with
Escherichia coli strain BL21(DE3) was provided by Novagen. Isopropyl--D-thiogalactopyranoside was purchased from
Research Products International Corp. Kits for DNA purification were
products of QIAGEN Inc. DNA was sequenced using an AutoRead sequencing kit and a Pharmacia-LKB-ALF DNA sequencer, both from Pharmacia Biotech
Inc. Fast Q anion exchange resin was also a product of Pharmacia.
Phenyl-agarose, DL-mevalonic acid lactone, pyridoxal phosphate (PLP),1 trypsin, immobilized
chymotrypsin, and all other biochemicals and reagents were purchased
from Sigma unless otherwise specified.
Methods
Construction of the Rat Mevalonate Kinase pET-11a Expression PlasmidA pET-11a expression plasmid was selected because of its
utility in overexpression in E. coli of cDNA-encoded
proteins. Insertion of a cDNA into pET-11a must occur at an
NdeI site that is positioned just downstream from a T7
promoter. Since the cDNA for rat mevalonate kinase does not contain
an NdeI site, it was necessary to create one by modifying a
single base upstream from the start codon. The cDNA was cut at an
NcoI site located 68 base pairs downstream from the
translation start site. This 68-base pair fragment was replaced with a
69-base pair oligonucleotide cassette composed of 4 deoxyoligonucleotides, the sequences of which were identical to the
amino-terminal coding sequence of mevalonate kinase except for two
changes: 1) a silent mutation at position 51 of the coding sequence
(CAT CA
) that eliminated an internal NcoI
site and 2) an extra T at the
1 position. The extra T created an
NdeI restriction site that overlapped the start codon,
allowing ligation into the pET-11a multiple cloning site immediately
downstream from the Shine-Dalgarno sequence. The 3
end of the cassette
contained an NcoI overhang, allowing ligation to the
downstream coding sequence of the NcoI-digested cDNA.
The sequences of the deoxyoligonucleotides follow, with non-wild-type
bases underlined; 1) 5
-
AT GTT GTC AGA AGT CCT GCT GGT GTC
TGC TCC-3
, 2) 5
-AGG GAA AGT CAT TCT CCA
GG AGA ACA TGC
TGT GGT C-3
, 3) 5
-CAT GGA CCA CAG CAT GTT CTC C
T GGA-3
,
4) 5
-GAA TGA CTT TCC CTG GAG CAG ACA CCA GCA GGA CTT CTG ACA ACA-3
.
An identical procedure was used to construct the K13M mutant with the
exception that the oligonucleotide cassette contained an additional
mutation that changed the lysine codon (AAA) to a methionine codon
(ATG). To confirm the integrity of the plasmid, DNA was sequenced in
both directions using a Pharmacia AutoRead sequencing kit and a
Pharmacia-LKB-ALF automated DNA sequencer with fluorescein-labeled
primers.
E. coli
transformants containing the pET-11a construct were grown in 0.5 liter
of LB medium at 37 °C, induced with
isopropyl-1-thio--D-galactopyranoside when the
A600 reached 0.8, and harvested after 6 h
of induction. The cells were suspended and lysed by passage through a
French pressure cell at 16,000 p.s.i. The lysate was centrifuged at
100,000 × g for 1 h, and the supernatant was
dialyzed against 2 liters of buffer containing 25 mM
Tris-Cl, pH 8.0, and 1 mM dithiothreitol. The dialyzed
supernatant was loaded onto a column (1 × 20 cm) of Fast Q anion
exchange resin equilibrated with the same buffer. The column was washed
until the A280 was <0.1, and mevalonate kinase
was eluted with a 0-100 mM linear gradient of potassium phosphate at pH 8.0. The fractions containing mevalonate kinase were
pooled, and solid (NH4)2SO4 was
added to bring the final salt concentration to 1.0 M. The
protein was then loaded onto a phenyl-agarose column (1 × 10 cm)
equilibrated with 1.0 M
(NH4)2SO4 in 25 mM
HEPES, pH 7.5, and mevalonate kinase was eluted with a decreasing
linear salt gradient (1.0-0 M
(NH4)2SO4). Protein purity and
subunit molecular mass were determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (11), and protein concentration was determined using the method of Bradford (12) with
bovine serum albumin as a standard. Measurement of enzyme activity was
carried out in the presence of 5 mM MgCl2 at
30 °C, pH 7.0, using a modification of a spectrophotometric assay
described elsewhere (13). The initial velocity was determined on the
basis of the rate of decrease in absorbance at 340 nm following
addition of mevalonate to the reaction mixture. The reaction velocities at various substrate concentrations were fitted to the Michaelis-Menten and Lineweaver-Burk equations using the program EnzFitter (14).
[3H]PLP was
prepared as described previously (15). The inactivation reaction was
carried out on ice using mevalonate kinase (1 nmol) in 100 µl of
HEPES buffer (50 mM, pH 7.5). PLP (or
[3H]PLP) was added to the enzyme solution at zero time,
and at the indicated time points samples were removed from the reaction
and reduced with ice-cold NaBH4 (2 mM in 10 mM Na2CO3). After 5 min of
reduction the enzyme activity was determined as described above. As a
control, a sample of mevalonate kinase was incubated with borohydride
only; control activity was equivalent to untreated enzyme. The
stoichiometry of PLP incorporation was determined on a sample of enzyme
modified with nonradioactive PLP. The reduced PLP-mevalonate kinase
adduct was separated from unbound PLP on a Sephadex G-50 column, and
the quantity of PLP that co-eluted with rat mevalonate kinase was
determined by measuring the pyridoxamine absorbance peak at 327 nm
(327 = 9400 M
1).
Borohydride-reduced [3H]PLP-labeled mevalonate kinase (13 nmol) was denatured, reduced with dithiothreitol, and carboxymethylated with iodoacetic acid. The protein was precipitated with trichloroacetic acid (6%) and pelleted, and the pellet was dissolved in NH4HCO3 (0.4 M) plus urea (4.8 M). Following dilution with 5 volumes of water, trypsin (20 µg or 1%) was added once every hour and stirred at 37 °C for a total of 4 h of digestion. The digestion was quenched by addition of trifluoroacetic acid (0.1%).
The mevalonate kinase peptides that resulted from exhaustive trypsin digestion were separated by HPLC using a Vydac C-18 (218TP54) reversed phase column equilibrated in aqueous trifluoroacetic acid (0.1%). The peptides were eluted with a linear gradient of acetonitrile (0-60%) over a period of 90 min at a flow rate of 0.7 ml/min. The column eluent was monitored at 215 nm using an absorbance detector with 3H radioactivity measured in tandem using a radioactivity detector.
The radioactive peptide that eluted at 58 min was recovered by using a fraction collector and bypassing the radioactivity detector. Because the 58-min fraction contained contaminating nonradioactive peptides visible on the UV trace, further purification of the radiolabeled peptide was required. The 58-min fraction was subjected to further digestion with 100 milliunits of immobilized chymotrypsin. After a 2-h digestion at 37 °C, the immobilized chymotrypsin was removed, and trifluoroacetic acid (0.1%) was added to the peptide mixture. The mixture was injected onto the Vydac column and eluted with a linear gradient of acetonitrile (20-40%) over a period of 50 min at 0.7 ml/min. A single radioactive peak was detected by counting samples of the eluted fractions. The radioactive peak correlated with a single symmetric UV peak, suggesting the recovery of a highly purified peptide, an expectation validated upon subsequent Edman degradation analysis.
ATPThe spin-labeled ATP analog,
ATPSAP, was synthesized as described previously (16). All spectra
were acquired at 22 °C using a Varian E-Line Century Series 9 GHz
EPR spectrometer. Rotational correlation times were calculated from
bound spectra using the program of Freed and co-workers (17, 18).
In a typical binding study, the concentration of ATPSAP was kept
constant at 30 µM, and the concentration of protein
(either wild type or K13M) was varied. Under the conditions used to
obtain the spectra, only the signal produced by unbound ATP
SAP is
observed. Therefore, the fraction of ATP
SAP free in each sample was
calculated by dividing the amplitude of the high field spectral line
produced in the presence of various concentrations of protein by the
amplitude of the signal produced in the absence of protein;
[ATP
SAP]bound = ([ATP
SAP]total
[ATP
SAP]free). A Scatchard plot of the data was used
to determine the binding constants and binding stoichiometry of
ATP
SAP to both wild-type and mutant proteins.
Competitive displacement experiments were performed in order to measure
ATP binding constants for both wild-type and K13M proteins. Varying
amounts of MgATP were added to a mixture containing ATPSAP (30 µM) and wild-type or mutant enzyme (75 and 150 µM, respectively). The signal produced by the displaced
ATP
SAP was used in a calculation (19) that allows determination of
the binding constant for the competing ligand
(Kd(ATP)) in terms of the known binding
constant of the displaced ligand
(Kd(ATP
SAP)).
Overexpression of rat mevalonate kinase using a pET-11a vector facilitates isolation of homogeneous enzyme (Table I; Fig. 1) in amounts adequate to support detailed study by chemical and physical approaches. The recombinant enzyme has a subunit molecular mass (42 kDa, Fig. 1) and kinetic characteristics (Table II) that are comparable to those reported for the tissue-isolated protein (20), suggesting that the recombinant protein represents a useful experimental model system.
|
|
An early study on the hog liver enzyme (8) suggests the presence of a
reactive lysine, but no identification or functional assignment to this
or any other active-site amino acid in mevalonate kinase has been
reported. To address this issue, the recombinant rat enzyme was
subjected to incubation with [3H]pyridoxal phosphate.
Borohydride reduction of the mixture resulted in the
time-dependent inactivation of the enzyme (Fig.
2) with incorporation of 2 mol of pyridoxal phosphate
per mol of enzyme subunit. When a protective agent such as ATP was
included in the incubation, the enzyme was completely protected from
inactivation and only 1 mol of pyridoxal phosphate was incorporated per
mol of subunit. Conversely, high levels of mevalonic acid (50 mM) provided only modest protection from inactivation. The
high level of efficacy of ATP in blocking inactivation implicated the
protected amino acid as an active-site lysine. The marked protection
afforded by ATP was also reflected in the striking differences apparent upon comparison of tryptic peptide maps produced when
[3H]pyridoxal phosphate was used to modify the enzyme in
the presence or absence of ATP (Fig. 3). The
distribution of radioactivity in this chromatogram is compatible with
the equivalent labeling of two distinct sites in the sample (Fig.
3A), one of which is protected against modification in the
presence of ATP (Fig. 3B). The "ATP-protectable"
radioactive peak apparent at 58 min was recovered for further HPLC
analysis. The presence of unlabeled contaminating tryptic peptides that
flanked the radioactive peptide (detectable in the
A215 nm HPLC map; data not shown) prompted a
subsequent chymotryptic cleavage of the partially purified peptide. This digestion produced a shorter radiolabeled peptide obtainable in
highly purified form under modified HPLC conditions. Edman degradation
analysis2 allowed assignment of the 10 amino acids that comprise the chymotryptic peptide. The sequence
obtained is in agreement with the deduced amino acid sequence near the
amino-terminal end of mevalonate kinase (9). The release of a
3H-labeled derivative in cycle 7, together with the absence
of the phenylthiohydantoin-lysine that would appear at easily
detectable levels in this step if no modification of mevalonate kinase
had been performed, indicated that lysine 13 is the active-site target of pyridoxal phosphate. Identification of this residue, an invariant amino acid in all mevalonate kinase sequences reported to date, coupled
with the observation of protection against pyridoxal phosphate modification by ATP, allows the first assignment of a mevalonate kinase
active-site residue.
Assignment of Function to Lysine 13; Kinetic Characterization
The basic properties of the side chain of lysine
13, together with the observation that ATP protected this residue
against modification by pyridoxal phosphate, not only argues for an
active-site location for this lysine but also prompts speculation that
lysine 13 could stabilize ATP binding because of interaction between the positively charged -amino group and the negatively charged phosphoryl groups of ATP. In a test of this hypothesis, site-directed mutagenesis was employed to substitute methionine for lysine at position 13. The mutant mevalonate kinase was expressed and isolated (Fig. 1) in a manner identical to that observed for wild-type enzyme.
Kinetic characterization (Table II) indicated that elimination of the
basicity at residue 13 depresses catalytic turnover by 56-fold. An
effect of this magnitude suggests that lysine 13 facilitates catalysis
but is not crucial to reaction chemistry.
Substrate Km values for the mutant enzyme change by approximately 1 order of magnitude, becoming numerically larger for mevalonate and smaller for ATP. Since for an ordered sequential reaction Km is a complex parameter that includes elements of both rate and binding constants, extrapolation from empirically determined apparent Km values to binding affinities may be misleading. For this reason, a more reliable evaluation of the possible function of the residue 13 side chain in stabilizing ATP binding was afforded by the determination and comparison of physical binding constants.
Biophysical Characterization of Wild-type and Mutant Mevalonate KinasesThe spin-labeled ATP analog, ATPSAP (16), contains a
paramagnetic PROXYL group tethered to the
-phosphoryl of ATP
S. In buffered solution, this analog exhibits the sharp three-line ESR signal
typical of nitroxide spin-labels free in solution (Fig. 4A). Upon binding to mevalonate kinase, the
amplitude of signals attributable to free ATP
SAP diminishes and the
broad spectral features assignable to an immobilized spin-label become
apparent (Fig. 4B). The spectrum of the bound species
exhibits a maximum hyperfine separation of 64.4 G. Spectral simulation
of the bound component (17, 18) indicates a correlation time,
c = 26 ns. Such a value corresponds to immobilization of
the nitroxide on a spherical particle with a Stokes radius of 29 Å.
Assuming a partial specific volume of 0.75 ml/g, a protein with such a
Stokes radius would have a molecular mass of 83 kDa, in good agreement with a dimeric form for a mevalonate kinase composed of 42-kDa subunits.
Calculation of free and bound ATPSAP is accomplished by monitoring
the change in signal amplitude due to the free analog. Performing such
an analysis with samples that vary in the ATP
SAP/mevalonate kinase
ratio allowed construction of a Scatchard plot, which indicated that
mevalonate kinase binds this analog both stoichiometrically (n = 0.95) with respect to enzyme subunits and with
high affinity (Kd = 3.1 µM; Table II).
The bound analog was displaced by addition of ATP (Fig.
5A), indicating that this probe binds at the
substrate site. By titrating samples prepared using a fixed ratio of
ATP
SAP/enzyme with variable amounts of ATP, the competitive displacement of analog was monitored (Fig. 5A) and a
Kd(ATP) (1.9 µM; Table II)
was calculated (19).
When the mevalonate kinase mutant that contains a methionine at
position 13 was used for similar ESR binding experiments, a binding
stoichiometry (n = 1.05) and c (26 ns)
comparable to that observed with wild-type enzyme were observed. These
data represent a rigorous test of the integrity of the tertiary
structure of the mutant, indicating that it contains a full complement
of functional substrate binding sites, although the observed
affinity for ATP
SAP was diminished
(Kd(ATP
SAP) = 49 µM; Table II). Moreover, when ATP was used to
competitively displace the analog from the K13M mutant (Fig.
5B), data analysis generated a
Kd(ATP) (110 µM; Table II)
that indicated >50-fold weakening of ATP binding to the mutant
mevalonate kinase in comparison with that observed for wild-type
enzyme. Thus, the collected data support the hypothesis that the lysine
13
-amino stabilizes binding of the substrate ATP.
The development of an expression system that produces a recombinant form of eukaryotic mevalonate kinase in the amounts documented in this report will undoubtedly expedite progress toward elucidation of the molecular basis for catalysis of 5-phosphomevalonate production. The high level of mevalonate kinase expression simplifies isolation of a stable homogeneous protein that exactly matches the amino acid sequence of the tissue-derived protein. The physical and kinetic properties of the enzyme expressed in E. coli argue strongly for the fidelity of this recombinant model and validate its expanded use in future studies.
In initial protein chemistry work on mevalonate kinase, we were guided by earlier reports on the sensitivity of the enzyme to various modification reagents (6, 8). The strategy of pursuing a reactive lysine was based on an expectation that pyridoxal phosphate, which may be expected to function as a group-specific modification reagent, could preferentially bind to the active site of a phosphotransferase due to the phosphoryl moiety of the reagent. The high degree of protection afforded by ATP against inactivation of mevalonate kinase by pyridoxal phosphate supported this expectation. This observation, coupled with the subsequent mapping of the protectable lysine as a residue invariant in the mevalonate kinase sequences, supported a provisional active-site assignment to lysine 13.
Another potential function of lysine 13 involves peroxisomal targeting of mevalonate kinase. Rachubinski and Subramani (23) propose that peroxisomal targeting may be mediated by peroxisomal targeting sequence 2, which is characterized by the consensus sequence (R/K)(L/V/I)(X5)(HQ)(LA). Lysine 13 corresponds to the first residue of this peroxisomal targeting sequence 2 motif. However, Purdue and Lazarow (24) do not suggest an alternative to arginine as the first residue of this consensus sequence. Furthermore, Tsukamoto et al. (25) report that lysine cannot replace arginine in this targeting sequence. Additionally, the work of Glover et al. (26) suggests that ketoacyl-CoA thiolase is poorly targeted to the peroxisome when lysine replaces arginine in the targeting sequence. Nonetheless, Krisans and co-workers (27, 28) report the peroxisomal localization of mevalonate kinase, suggesting that lysine 13 may well function as part of a targeting sequence.
Regardless of whether lysine 13 is crucial to peroxisomal targeting,
our data clearly implicate lysine 13 as an active-site residue. The
magnitude in diminution of Vm observed upon
conservative substitution to eliminate basicity at residue 13 suggests
that lysine 13 significantly influences but is not crucial to
catalysis. Any contribution to substrate binding would be difficult to
evaluate on the basis of the contrasting changes in substrate
Km values, which are not necessarily reliable
indicators of substrate affinity. Thus, the availability of sufficient
amounts of homogeneous protein for binding experiments was important to
any evaluation of the lysine 13 function. The other important tool for
such studies was a novel spin-labeled ATP analog that we recently
reported in studies on another phosphotransferase (16, 22). The results of binding studies for ATPSAP and of competition experiments with
ATP clearly demonstrate that substitution to eliminate the basicity of
lysine 13 impairs the binding of adenine nucleotides to mevalonate
kinase. The >50-fold effect on ATP Kd values is
compatible with hypotheses that lysine 13 either stabilizes the
negative charge of the ATP by an electrostatic interaction with the
phosphoryl groups or hydrogen bonds to another region in the
nucleotide.
With respect to ATP binding, three domains that have highly conserved amino acid sequences exist in mevalonate kinase. One of these domains found in a central region of the protein has been proposed by Tsay and Robinson (29) to be part of an ATP binding consensus sequence. The consensus they propose, GXGXXAX(13-26)K, is found in Saccharomyces cerevisiae mevalonate kinase, with the lysine 24 residues downstream from the consensus alanine. However, when comparing sequences of mevalonate kinases from a variety of species, the lysine contained in this consensus sequence is not invariantly conserved. In the rat and human enzymes it is 22 residues downstream from the consensus alanine, in the A. thaliana sequence it is 18 residues downstream, and in the sequences from Schizosaccharomyces pombe and Methanobacterium thermoautotrophicus no lysine is found within 26 residues of the consensus alanine.
A second domain that has been proposed as an ATP binding motif resides in the carboxyl-terminal region of both the human (3) and the rat proteins (9). This sequence conforms to the GXGXXGX(15-21)AXK consensus found in protein kinases. Human and rat mevalonate kinases contain the AXK motif 17 residues downstream from the last consensus glycine. However, the sequences from Arabidopsis thaliana, S. pombe, S. cerevisiae, and M. thermoautotrophicus do not contain such a motif.
The nucleotide binding sequences found in numerous proteins including protein kinases (30) as well as a variety of other nucleotide-binding proteins and phosphotransferases have been reported (21). A striking feature common to all these consensus sequences is the presence of either a lysine or arginine residue that interacts with a phosphoryl group on the nucleotide. This report on the function of the mevalonate kinase lysine 13 provides the only evidence for participation of an invariant basic residue in ATP binding. Without any experimental evidence to directly implicate the proposed central or carboxyl-terminal consensus sequences in nucleotide binding, it is difficult at present to evaluate their significance. However, our recombinant model may facilitate investigation of this issue.
Dr. Jennifer Runquist synthesized the [3H]pyridoxal phosphate used in these studies. Kelly Chun assisted in construction of the expression plasmid that encodes the K13M mutant. Edman degradation was performed under the direction of Dr. Liane Mende-Mueller at the Protein and Nucleic Acid Shared Research Facility, Medical College of Wisconsin. We thank Dr. G.W. Robinson and Bristol-Myers Squibb for supplying plasmid pGR50.