Characterization of the nucleotide binding properties and ATPase activity of recombinant hamster BiP purified from bacteria.

HSP70 family proteins bind ATP and hydrolyze it, but the precise role of these activities in their in vivo chaperoning function has not been determined. In this report, we characterized wild-type hamster BiP isolated from bacteria in terms of its ATP binding and ATPase activities. Recombinant BiP behaved essentially the same as endogenous BiP in terms of oligomeric status, protease digestion patterns, and ATPase properties. By engineering a Factor Xa cleavable site following the His tag which was used for affinity purification, we demonstrated that the six histidines had no effect on either the structural or ATPase properties of recombinant BiP. We also found that bacteria-synthesized BiP had a tightly bound ADP that was resistant to dialysis. Removal of the bound nucleotide allowed us to directly measure the binding affinity of ATP and ADP to BiP (K of 0.2 μM for ATP and 0.29 μM for ADP) by equilibrium dialysis. Careful characterization of wild-type BiP will allow us to use this system to characterize BiP ATP binding site mutants that can be used to probe the role of ATP binding and ATPase activity in BiP functions.

The heat shock protein 70 (HSP70) 1 family is comprised of a highly conserved class of molecular chaperones. Members of this family exist in all species from Escherichia coli to human and in all the organelles of eukaryotic cells (1,2). Most of these proteins are constitutively expressed and are further induced under conditions of stress like heat shock or other physiological insults to the cells (1). Immunoglobulin heavy chain binding protein (BiP) is the eukaryotic endoplasmic reticulum (ER) member of the HSP70 family, and is one of the major resident ER proteins. BiP is proposed to play a role in protein folding, subunit assembly, and subsequent transport of proteins from the ER (2). BiP associates transiently with a variety of nascent secretory proteins (3)(4)(5)(6)(7) and more stably with malfolded or unassembled proteins whose transport from the ER is blocked (3-5, 8, 9). However, the extent of BiP's involvement in these processes is not clearly understood.
All HSP70 family members possess a highly conserved NH 2terminal ATP binding domain and a more variable COOHterminal protein binding domain (10,11). The three-dimensional structure of the NH 2 -terminal 44-kDa proteolytic fragment of bovine brain hsc70, a cytoplasmic HSP70, has been determined. This fragment contains the ATP binding site, and its structure is very similar to that of two other adenine nucleotide binding proteins: actin and hexokinase, even though their amino acid sequence homology is not high (12,13). Studies on both recombinant and purified bovine hsc70 revealed that hsc70 binds ATP and ADP with K d values in the order of 10 Ϫ5 -10 Ϫ8 M (14 -18). Although all family members presumably have a similar nucleotide binding structure, direct binding studies have not been done on BiP or other HSP70 members. The HSP70s also have a rather weak intrinsic ATPase activity. For BiP, the reported turnover numbers for ATPase activity range from 0.02 to 0.35 min Ϫ1 (19,20). Studies with normal endogenous BiP purified from dog pancreas demonstrated that the ATPase activity is optimal at acidic pH and low salt concentrations, requires magnesium, and is strongly inhibited by calcium (20). There has been no thorough characterization of the enzymatic properties of bacterially expressed wild-type recombinant BiP (rBiP).
In this report, we describe the nucleotide binding properties and ATPase activity of rBiP purified from bacteria. Hamster BiP was tagged with six histidines at the NH 2 terminus (6X-His), expressed, and affinity purified on Ni 2ϩ -agarose to near homogeneity. By engineering a Factor Xa cleavage site immediately following the histidines and comparing the 6X-Histagged BiP with the tag-cleaved BiP, we found that the sixhistidine tag had no effect on either the structural or ATPase properties. The ATPase activity of rBiP was characterized in terms of pH optimum, salt, and divalent cation requirements, and these properties were essentially the same as those reported for BiP purified from tissue (20). Like hsc70, the rBiP purified from bacteria contained tightly bound ADP, and removal of this nucleotide allowed us to measure the nucleotide binding directly by equilibrium dialysis. These characterizations are not only important for establishing the enzymatic parameters of BiP, but will also enable us to identify BiP ATPase and ATP binding mutants that can be used to determine the role of nucleotide in BiP's function.

MATERIALS AND METHODS
Expression of 6X-His Tagged BiP in Bacteria-A cDNA clone for hamster BiP (21) was ligated into the QE10 vector (Qiagen) in such a way that the leader sequence of BiP was removed and replaced with six histidines and seven non-native amino acids (22). Recombinant hamster BiP protein was expressed in M15 E. coli and purified on Ni 2ϩagarose. Protein quantitation was done by Bradford's dye binding assay using bovine serum albumin as a standard (Bio-Rad).
Factor Xa Site Engineering and Subsequent Cleavage of the Histidine Tag-The Factor Xa tetrapeptide recognition site, IEGR, was engineered immediately NH 2 -terminal of the first amino acid of the mature hamster BiP protein (signal sequence-cleaved BiP) and just COOHterminal of the six histidines by using a primer extension method (23). A 39-mer oligonucleotide was synthesized that contained a BamHI site, followed by an IEGR-encoding DNA sequence, and ending with the nucleotides encoding the first six amino acids of mature BiP (CGG GAT CCG ATC GAA GGT AGA GAG GAG GAG GAC AAG AAG). This oligo was used as the 5Ј primer, and an antisense 18-base pair oligo flanking Gly 226 of BiP was used as the 3Ј primer for amplification by polymerase chain reaction. The polymerase chain reaction-amplified fragment was digested with BamHI and BstEII and cloned into the pTZBiP vector (22) in place of the original BamHI-BstEII fragment. The sequence of the inserted fragment was confirmed by dideoxy sequencing, and the entire BiP construct was then cloned into the pQE10 vector for production of rBiP protein. The resultant protein (His-Xa-BiP) had an NH 2 -terminal 6X-His tag followed by the IEGR motif, that is recognized and cleaved by Factor Xa. The Ni 2ϩ -agarose purified protein was digested with Factor Xa (U. S. Biochemical Corp.) for 18 h at 30°C with 1:100 (w/w) of Factor Xa to His-Xa-BiP in our modified Xa buffer, containing 20 mM HEPES (pH 7.5), 100 mM NaCl, 2.0 mM CaCl 2 , and 1 mM sodium azide. We were unable to use the classical Tris containing buffer for Xa cleavage (24), because the purified rBiP protein precipitated in this buffer. Any uncleaved His-Xa-BiP remaining after 18 h of digestion was removed by rebinding the sample to Ni 2ϩ -agarose beads.
Nondenaturing Gel Electrophoresis-2-15% gradient nondenaturing polyacrylamide gels were prepared as described previously (25). 10 g of rBiP was loaded and electrophoresed at 4°C for 16 h at 150 V. After electrophoresis, the gel was stained with Coomassie Brilliant Blue.
ATPase Assay-The ATPase activity of BiP was analyzed by quantitating the free P i liberated during ATP hydrolysis by the method of Seals et al. (26) with some modifications. The standard assays were carried out at 37°C in a volume of 100 l. At appropriate times, 20-l aliquots were removed from assay mixtures and pipetted into tubes containing 20 l of 2% SDS. After mixing to stop the reaction, 50 l of fresh phosphate reagent (4 N H 2 SO 4 , 4% ammonium molybdate, and 0.02 M silicotungstic acid) and 200 l of 65:35 xylene:isobutyl alcohol (v/v) were added, and the samples were vortexed twice for 10 s. A 100-l aliquot was removed from the upper organic phase and counted in a liquid scintillation counter to determine [ 32 P]P i released. For the studies to determine the pH dependence of ATPase activity, citrate buffer was used for pH 4.5 and the rest of the buffers (acetate for pH 5.0, MES for pH 5.5, 6.0, and 6.5, HEPES for pH 7.0 and 7.5, and Tris for pH 8.0 and 9.0) were the same as described (20).
Proteolysis of Recombinant BiP in the Presence of Nucleotide-Proteinase K digestions were performed on rBiP as described by Kassenbrock and Kelly (20) with some modifications. 65-l reactions were assembled that contained 10 g of rBiP, 2 g of proteinase K (or similar volume of buffer for control), and 100 M ATP or ADP in the standard ATPase buffer (20 mM HEPES, 25 mM KCl, 2 mM MgCl 2 , 0.1 mM EDTA, and 0.5 mM dithiothreitol at pH 7.0). After incubation at 37°C for 25 min, the reaction was stopped by adding 10 l of 1 mg/ml phenylmethylsulfonyl fluoride and incubating it on ice for 30 min. The digested rBiP was then analyzed by SDS-PAGE.
HPLC Analysis of BiP Nucleotide Content and Preparing Nucleotidefree BiP-The protocol of Gao et al. (18,27) was used to determine the nucleotide content of rBiP and to prepare nucleotide-free protein. For nucleotide determinations, 10 M rBiP was acid denatured and then precipitated to separate protein from released nucleotide. The sample was neutralized with KOH and clarified by centrifugation before applying it to the HPLC column. An HPLC system consisting of a Pharmacia LKB gradient pump 2249, a Pharmacia LKB VWM 2141 monitor, and a LKB 2221 integrator was used. The nucleotide-free BiP was prepared by incubating purified rBiP with a 600-fold molar excess of AMPPNP to replace the bound nucleotide and then dialyzing it in the standard ATPase buffer. The extent of nucleotide removal was determined by HPLC.
Nucleotide Binding-Nucleotide binding was performed by equilibrium dialysis as described by Gao et al. (27). In these experiments, cylindrical 200-l volume dialysis chambers fitted with dialysis membranes (MWCO 25 kDa) were used. 0.3 mg/ml bovine serum albumin was tested and shown not to bias the equilibrium of ATP or ADP, and was therefore included in both sides of the membrane to stabilize the rBiP protein. The reaction was carried out in the standard ATPase buffer with 2.0 M rBiP on one side and 0.25-4.0 M [ 14 C]ATP or ADP (with constant specific activity) on both sides of the dialysis membrane. The dialysis chamber was rotated at 4°C with a speed of 10 rpm and equilibrium was reached within 24 h as monitored by counting an aliquot from both sides of the membrane at various times during dialysis.

Engineering of a Factor Xa Site following the Six Histidine
Tag and Subsequent Purification of Tag-cleaved rBiP-In a previous report from this laboratory (22), a 6X-His tag was inserted in place of the BiP signal sequence, which is normally cleaved when BiP is translocated into the ER of mammalian cells. This 6X-His tag allows a simple one-step purification of rBiP. However, it was not determined whether the presence of six positively charged amino acids at the NH 2 terminus of the ATP binding domain influences the ATP binding or hydrolysis properties of the tagged-BiP. This concern is particularly valid for kinetic assays.
In order to evaluate any effects of these histidines, a specific protease cleavage site was inserted between the 6X-His tag and the sequence corresponding to the NH 2 terminus of the mature BiP protein (Fig. 1A). The Factor Xa recognition tetrapeptide, IEGR, was chosen for the following reasons: 1) this sequence does not exist in the BiP molecule, and 2) cleavage with Xa occurs immediately after the arginine residue in the tetrapeptide, producing a BiP molecule identical in amino acid sequence to mature BiP. The bacterially expressed BiP with 6X-His-IEGR on the NH 2 terminus was first purified on a Ni 2ϩ -agarose column and subsequently digested with Factor Xa for varying amounts of time (Fig. 1B). The cleavage was nearly complete at 18 h. After rebinding the sample to Ni 2ϩ -agarose to remove any remaining uncleaved BiP, the Xa-cleaved protein is quite pure (Fig. 1B, purified). Isolation of pure rBiP protein containing the 6X-His tag, as well as protein devoid of the His tag, allowed us to determine whether the His tag affected either structural or ATPase properties of rBiP.
Oligomerization and Conformation of Tag Cleaved as Well as Uncleaved rBiP-In mammalian cells, BiP exists as both a dimer and monomer (25). Nondenaturing gel electrophoresis was used to analyze the oligomerization status of the tagcleaved and uncleaved forms of rBiP. Both of the 6X-His-tagged proteins (His-Xa-BiP and His-BiP) and the Xa-cleaved BiP (mature) existed as a combination of monomers, dimers, and some higher order oligomers ( Fig. 2A), suggesting that the histidine tag did not grossly alter the structure of the rBiP proteins. The Xa-cleaved BiP migrates slightly faster on nondenaturing gels, presumably due to the removal of the positivecharged peptide. The migration patterns of the three purified rBiP proteins are nearly identical to each other and to that observed when purified mammalian BiP and hsc70 were similarly analyzed (28,29).
Partial protease digestion is often used as a measure of the structural integrity of a protein. The HSP70 proteins produce very distinctive proteolytic patterns when digested in the presence of nucleotides. ATP protects 60-and 44-kDa fragments, whereas, ADP protects only the 44-kDa fragment (11,20). In order to check the structural integrity of our three forms of rBiP, we digested the various samples with proteinase K, a nonspecific serine protease, in the presence of ATP or ADP. For all three of the BiP preparations, both a 60-and a 44-kDa fragment were protected in the presence of ATP, and a 44-kDa fragment was protected when ADP or no nucleotide was present (Fig. 2B). These protected fragments were all derived from the NH 2 -terminal domain, because they could still bind to Ni 2ϩ -agarose (His-Xa-BiP and His-BiP) but could not be immunoprecipitated with a COOH-terminal specific antiserum (data not shown). The observation that samples with no nucleotide added produced the same digestion pattern as samples to which ADP was added implies that either the rBiP already contained bound ADP or that the NH 2 -terminal domain folds very compactly even without nucleotide. Together, these data suggest that bacterially produced rBiP proteins retained the structural integrity of the native BiP protein, and that the presence of a histidine tag did not grossly alter this structure.
Characterization of the ATPase Activity of rBiP and Comparison of Tag Cleaved versus Uncleaved rBiP in Terms of ATPase Activity-Knowing that our recombinant proteins maintained the structural properties of endogenous BiP, we next wished to examine the ATPase activity of these rBiP proteins. In this study, an ATPase assay measuring the release of inorganic phosphate was chosen (26). This method is sensitive, easy to perform, and suitable for measuring ATP hydrolysis across a wide range of ATP concentrations. The ATPase activities of the three different forms of rBiP were compared under identical conditions (Fig. 3A). For this experiment, His-Xa-BiP was mock-digested to control for any effects of the digestion conditions. Interestingly, we found that mock-digested BiP had identical ATPase activity when compared to His-Xa-BiP that was kept at 4°C, indicating that BiP is fairly stable at 30°C for at least 18 h (not shown). The ATPase activity of the cleaved rBiP (mature BiP) was essentially identical to both of the tagged forms of rBiP (Fig. 3A).
The pH effect on BiP's ATPase activity was examined by using the same buffers described for assaying endogenous BiP (20). Peak ATPase activity was obtained at pH 5.0 while either increasing or decreasing the pH from pH 5.0 significantly lowered the ATPase activity (Fig. 3B). For all further ATPase determinations, pH 7.0 was used because: 1) the ER pH is estimated to be approximately pH 7.0; 2) pH 7.0 is in a region of the pH curve with a shallow slope, thus minimizing the potential for large measurement errors; and 3) the ATPase and ATP binding activities of hsc70 were analyzed at pH 7.0 (14,18,30). Thus, by performing our ATPase assays at pH 7.0, it will be possible to compare findings between these two related proteins.
We next assayed the effects of salt and divalent cations on BiP's ATPase activity. The data for cleaved and His-tagged BiP were indistinguishable, and thus, for simplicity, only that for cleaved BiP (mature) is presented. The optimal salt concentration was about 25 mM for both NaCl and KCl, but at all concentrations measured, KCl produced a higher ATPase activity than NaCl (Fig. 3C). Therefore, all remaining assays were performed with 25 mM KCl. Calcium was strongly inhibitory to BiP's ATPase activity, and magnesium was required for the optimal enzyme activity (Fig. 3D). This is consistent with the requirements for both native BiP and hsc70 (20,30), but unlike data on hsc70, Mg(CH 3 CO) 2 was not preferred over MgCl 2 . The optimal concentration of MgCl 2 required for rBiP's ATPase activity ranged from between 1 and 10 mM. The ATPase activity of our rBiP could be stimulated 2-3-fold by peptides (see Ref. 43, accompanying article), which is similar to data for native mammalian BiP and hsc70 (28,31). Taken together, these data demonstrate that the ATPase properties of our rBiP purified from bacteria are extremely similar to those of BiP purified from dog pancreas (20). In no case was there a significant difference between 6X-His-tagged and tag-cleaved BiP, demonstrating that the 6X-His tag does not affect any of the various aspects of BiP's ATPase activity. In the following study (43), 6X-His BiP was used in order to avoid additional manipulations to the recombinant proteins during purification.
Nucleotide Content of rBiP and Preparation of Nucleotidefree rBiP-The inability of our rBiP to bind nucleotide in equilibrium dialysis experiments alerted us to the possibility that the rBiP purified on Ni 2ϩ -agarose might contain bound nucleotide, even though our rBiP was not purified on ATP-agarose columns. If this was the case, the nucleotide must have bound in the bacteria and must be resistant to dialysis, since the rBiP was dialyzed before the binding study. To determine if our rBiP contained bound nucleotide, the recombinant protein was perchloric acid-denatured and precipitated. The supernatant, which should contain any released nucleotides, was analyzed by HPLC. The data showed that 0.82 mol of ADP was bound per mol of rBiP (Fig. 4A). The absence of ATP in this BiP preparation was not due to hydrolysis during the extraction procedure, since no hydrolysis occurred when an ATP standard was treated in parallel (data not shown).
At this point it was necessary to prepare nucleotide-free BiP in order to measure the nucleotide binding constants for ATP and ADP. We used the method developed for preparing nucleotide-free hsc70 (18). A 600-fold molar excess of AMPPNP was added to rBiP, the sample was incubated at room temperature for 1 h to replace the bound ADP, and then the bound AMP-PNP, as well as free nucleotides, were removed by dialysis. Over 95% of the bound ADP was removed after this treatment, as shown by HPLC analysis (Fig. 4). The nucleotide-free BiP was stable for about 2 weeks when kept at 4°C as determined by measuring ATPase activity (not shown). The oligomerization state and protease digestion patterns were not altered by removing the nucleotide, suggesting that ADP is not required for BiP oligomerization and that nucleotide-free BiP is structurally similar to BiP containing bound ADP (not shown).
Kinetic Properties of rBiP-The kinetic properties of the nucleotide-free rBiP's ATPase activity were examined using the optimized conditions derived from the initial ATPase characterization. In this experiment, several enzymatic criteria were strictly observed. First, the enzyme concentration used was in the linear region in the initial velocity versus [rBiP] plot (Fig.  5A). Second, the lowest substrate (ATP) concentration was at least 10-fold greater than the enzyme concentration. And third, at the lowest substrate concentration for a given amount of enzyme, the incubation time was chosen such that it was in the linear portion of the time course curve (Fig. 5B). Under standard assay conditions and using a constant amount of enzyme (0.05 M), ATP concentrations from 0.5 to 6.0 M were used for assaying the ATPase activity (Fig. 5C). The initial velocities obtained and the corresponding ATP concentrations were entered into the Clelend's kinetic program that determines the V max and K m values based on a Lineweaver-Burk's plot (Fig.  5D). A V max of 5.2 pmol/min/g and a K m of 1.5 M were obtained. The V max corresponds to a k cat of 0.40 min Ϫ1 and is consistent with the slow rate of hydrolysis reported for native BiP and other HSP70 family members.
Equilibrium Dialysis Measuring Nucleotide Binding of rBiP-Equilibrium dialysis was used to measure the dissociation constants of nucleotide-free rBiP for ADP and ATP. With increasing ADP concentrations, the portion of bound nucleotide increased until it approached saturation at 4 M (Fig. 6A). Further increases in the ADP concentration up to 200 M did not increase the amount of nucleotide bound to BiP (not shown). For the ATP binding measurement, creatine kinase and creatine phosphate were included in both chambers since a significant amount of the ATP was hydrolyzed if the ATP regenerating system was not included (not shown). The efficiency of creatine kinase and creatine phosphate to convert ADP to ATP was almost 100%, as demonstrated by thin layer chromatography performed at the end of dialysis (not shown). The ADP and ATP binding data were replotted in a Scatchard plot (Fig. 6, A and B, insets), and the resultant number of binding sites ranged from 0.7 to 1.1 for both ATP and ADP depending on different preparations. The K d values of 0.2 Ϯ 0.1 and 0.29 Ϯ 0.12 M were obtained for ATP and ADP, respectively. However, within a single protein preparation, the variations were small (Fig. 6). The stability of rBiP under these binding conditions was examined by measuring the amount of bound nucleotide at both 24 h (time required to reach equilibrium) and at 48 h. There was no loss of bound nucleotide even after 48 h of dialysis, suggesting that rBiP was stable under the binding conditions used. DISCUSSION The ability to identify and fully characterize BiP ATP binding and ATP hydrolysis mutants requires both a knowledge of the normal kinetic values for wild-type BiP, and a method to isolate mutant proteins that does not require ATP binding or that does not result in co-contamination with wild-type BiP protein. Tagging hamster BiP with six histidines at its amino terminus allowed easy purification of the rBiP from bacteria, but unlike native BiP, resulted in the addition of non-native amino acids to the ATP binding domain that could affect kinetic measurements. To determine if the 6X-His tag affected BiP, we engineered a factor Xa cleavage site immediately downstream of the 6X-His tag so it could be removed, producing a protein that was identical in sequence to mature hamster BiP isolated from cells.
The 6X-His tagged and the tag-cleaved rBiP seemed to be structurally identical to native mammalian BiP as judged by protease protection assays (20) and by their ability to form dimers and monomers (25,28). Similar to other studies performed on native BiP (29,31), we found that the dimers were converted to monomers by peptide binding (not shown), and that the ATPase activity was stimulated approximately 2-3fold by peptide (Ref. 43, accompanying paper). This is in contrast to another report, in which rBiP was isolated, in part, on ATP-agarose columns (29). Their rBiP was primarily monomeric and was not stimulated by peptide. We found that all three forms of our rBiP had the same rates of ATP hydrolysis, and that their pH and salt optima, as well as the divalent cation requirements for this activity were the same as those reported for native BiP (20). Thus, our purification method produced rBiP proteins that appeared to be structurally very similar to native BiP, and the 6X-His tag did not apparently affect structural or enzymatic parameters.
Our recombinant BiP possessed tightly bound ADP, just like native hsc70 isolated from bovine brain (14,27). Unlike the purification scheme for bovine hsc70 (14,27), we did not use nucleotide columns and ATP elution in our isolation. Therefore, the bound ADP must come from the bacteria and is most likely a normal component of HSP70 proteins. After producing nucleotide-free BiP, we performed ATPase assays to determine the K m (1.48 Ϯ 0.1 M) and V max (5.12 pmol/min/g) of wild-type BiP. The variations between reported values for V max of HSP70 proteins are surprisingly large. The V max values for BiP range from 0.2 to 4.7 (19,20), for hsc70 from 1.2 to 12 (32,33), and for dnaK from 2.5 to 600 pmol/min/g (34,35). This variation could stem from the fact that, in most studies, the percent of the preparation that was active is not known, no attempt was made to remove bound nucleotide before performing the assays, and finally, the purity of the various preparations used in these studies is not known. Because the HSP70s are very poor ATPases, even a small amount of contamination with a much stronger ATPase could have profound effects on the V max value. We feel that our determination of V max and K m values are reliable, because we ensured that our protein possessed nearly full ATP binding activity and was free of nucleotide before performing kinetic studies. We determined that our preparation was free of contamination by immunoblotting with anti-dnaK antisera and by performing in vitro kinase assays (data not shown). The purity of our rBiP preparation was further corroborated by our isolation of an ATPase Ϫ mutant, T229G BiP (43) that was purified in the same way. Finally, care was taken to ensure that our studies followed strict enzymatic criteria.
There is currently no reported data on direct nucleotide binding for BiP, however, there are reports for hsc70 prepared either from bovine brain or bacteria with K d values for ADP or ATP ranging from 10 Ϫ5 to 10 Ϫ8 M. Using initial rate binding, Schmid et al. (14) 4. HPLC analysis of nucleotide bound to rBiP. a, buffer alone was extracted as a control and the background profile was determined. b, nucleotide was extracted from 10 M rBiP as described under "Materials and Methods." c, rBiP was preincubated with 600 molar excess of AMPPNP to replace bound nucleotide and then extensively dialyzed to remove both bound and free nucleotides. Nucleotide was extracted as in b. A control sample containing a mixture of ADP, AMPPNP, and ATP was injected to determine where each of them eluted and each is denoted with a downward arrow. K d of 0.012 M for Mg-ATP and 0.018 M for Mg-ADP for nucleotide-free hsc70 purified from bovine brain. These values are somewhat lower than those for recombinant hsc70 or BiP, and the discrepancy may result either from differences in the experimental methods used, or from an intrinsic difference between native and recombinant proteins.
The weak ATPase activity and high ATP binding affinity of BiP measured in vitro suggests one of two things. First, other ATPase activating proteins may exist. In bacteria, two heat shock proteins, dnaJ and grpE, stimulate the ATPase activity of dnaK up to 50-fold (36). Cytosolic homologues of dnaJ have been identified in human and yeast cells (37)(38)(39)(40), and an ER protein with homology to dnaJ has been isolated in yeast (41). However, no data are available to demonstrate that these proteins modulated the ATPase activity of eukaryotic HSP70 proteins. Alternatively, ATP binding may be more important to the function of the HSP70 proteins than ATP hydrolysis. This may be particularly true for BiP since the assumed ER conditions, pH 7.4 and millimolar calcium, strongly inhibit BiP's ATPase activity. An in vitro study using purified proteins demonstrated that ATP binding, but not ATP hydrolysis, was required for the release of bound polypeptides from hsp70 and dnaK (42,43).  6. Binding of ATP and ADP to nucleotide-free rBiP. The binding of ADP (A) and ATP (B) to nucleotide-free rBiP was determined by equilibrium dialysis. Varying amounts of [ 14 C]ATP or ADP (abscissa) with constant specific activity were added to both sides of the dialysis membrane. The amount of labeled nucleotide associated with the protein at equilbrium is plotted on the ordinant. The data was analyzed in a Scatchard plot (inset) using the calculated amount of bound and free nucleotide.
In summary, we have isolated recombinant BiP protein from bacteria and demonstrated that enzymatically it behaves very much like native BiP isolated from mammalian cells. Using rBiP purified by a single step, we determined the K m and V max values for wild-type BiP and provided the first determination of nucleotide binding affinities for BiP. The characterization of wild-type BiP activity permits us to analyze the various BiP ATPase mutants (43) to determine whether they represent nucleotide binding or hydrolysis mutants. This should then allow us to distinguish the role of ATP binding from that of ATP hydrolysis in BiP function.