Binding of the viral immunogenic octapeptide VSV8 to native glucose-regulated protein Grp94 (gp96) and its inhibition by the physiological ligands ATP and Ca 2+ Ming Ying and Torgeir Flatmark Section of Biochemistry and Molecular Biology, Department of Biomedicine, University of Bergen, Norway Grp94 (gp96), a major chaperone of the ER lumen and a paralogue of the cytoplasmic chaperone Hsp90, plays an essential role in the structural maturation and ⁄ or secretion of a subset of client (cargo) proteins destined for transport to the cell surface [1]. Related studies have indicated that Grp94 also associates with a wide array of immunogenic peptides generated in the cytosol or in the early secretory pathway. Such Grp– peptide complexes have been proposed to interact with antigen-presenting cells (APC) in a specific manner that eventually leads to the presentation of peptides by the major histocompatability complex (MHC) class I molecules of the APCs [2]. The immunodominant viral octapeptide RGYVYQGL (VSV8), derived from vesicu- lar stomatitis virus nucleoprotein (residues 52–59), has been identified as a MHC class I H2-Kb ⁄ H2-Kd epi- tope (http://immunax.dfci.harvard.edu/bioinformatics/ epimhc/) and a Grp94 ligand in VSV-infected cells [3], and has been shown to bind to Grp94 in vitro [4–7]. The generally reported low affinity and low stoichio- metry of peptide binding to Grp94 in vitro, and the finding that this binding seems to be almost irrevers- ible, have questioned the proposed in vivo peptide acceptor–donor function of the chaperone [8,9]. Thus, in previous in vitro studies peptide binding to Grp94 was artificially enhanced by experimental conditions that elicit a more open conformational state of the chaperone, i.e. by heat shock denaturation (50–60 °C) [5–7,10], by a chemically induced conformational change [6] or by high salt concentrations [5]. More- over, the binding assays were based on incubations of Grp94 and radiolabelled ligand for 1–20 h [7]. Finally, no physiological ligands favouring such conformation- al states of Grp94 have been defined, and a better understanding of the interaction of the native chaper- one with ‘client’ peptides and the possible regulation Keywords ATP; cations; Grp94; SPR; VSV8 Correspondence T. Flatmark, Section of Biochemistry and Molecular Biology, Department of Biomedicine, University of Bergen, Jonas Lies vei 91, N-5009 Bergen, Norway Fax: +47 55 586360 Tel: +47 55 586428 E-mail: torgeir.flatmark@biomed.uib.no (Received 30 September 2005, revised 1 December 2005, accepted 2 December 2005) doi:10.1111/j.1742-4658.2005.05084.x The molecular chaperone Grp94 (gp96) of the endoplasmic reticulum (ER) lumen plays an essential role in the structural maturation and ⁄ or secretion of proteins destined for transport to the cell surface. Its proposed role in binding and transferring peptides for immune recognition is, however, controversial. Using SPR spectroscopy, we studied the interaction of native glycosylated Grp94 at neutral pH and 25 and 37 °C with the viral immunogenic octa- peptide RGYVYQGL (VSV8), derived from vesicular stomatitis virus nucleoprotein (52–59). The peptide binds reversibly with low affinity ([A] 0.5  640 lm) and a hyperbolic binding isotherm, and the binding is par- tially inhibited by ATP and Ca 2+ at concentrations that are present in the ER lumen, and the effects are explained by conformational changes in the native chaperone induced by these ligands. Our data present experimental support for the recent proposal that, under native conditions, VSV8 binds to Grp94 by an adsorptive, rather than a bioselective, mechanism, and thus further challenge the proposed in vivo peptide acceptor–donor function of the chaperone in the context of antigen-presenting cell activation. Abbreviations APC, antigen-presenting cell; dn ⁄ dc, refractive index increment; ERGIC, ER ⁄ Golgi intermediate compartment; Grp78 (BiP), ER glucose- regulated protein of 78 kDa; Grp94 (gp96), ER glucose-regulated protein of 94 kDa; MHC, major histocompatability complex; NECA, 5¢-N-ethylcarboxamidoadenosine; RU, resonance unit; TAP, transporter associated with antigen presentation. FEBS Journal 273 (2006) 513–522 ª 2006 The Authors Journal compilation ª 2006 FEBS 513 of this function by ligand interactions is considered imperative [9]. In this study, SPR spectroscopy was used to study multiple ligand-binding events in real time to the immobilized native glycosylated Grp94 (rGrp94) and nonglycosylated recombinant Grp94 at 25 or 37 °C. Special attention was paid to the reversi- ble binding of the viral immunogenic octpeptide RGY- VYQGL (VSV8) and physiological low-molecular mass ligands of the ER lumenal compartment (Ca 2+ and ATP), known to regulate the function of Grp94 as a protein chaperone [4,11,12]. Thus, the SPR binding studies provide new insights into native Grp94–ligand interactions related to its proposed in vivo peptide binding and antigen presentation activity. Results Previous in vitro studies on the peptide-binding activity of Grp94 [6,7] have not given any equilibrium-binding parameters for peptide binding to the native chaperone or identified any physiological factors in the ER lumenal environment which may have a regulatory effect on the Grp94–peptide interactions. In this study we addressed both questions by choosing SPR as the experimental approach to study the interactions ofGrp94 in real time with the viral immunogenic octapeptide VSV8 and some known physiological low- molecular mass ligands of the chaperone in the ER lumenal compartment which may perturb this inter- action. Binding of the viral immunogenic octapeptide VSV8 to native rGrp94 Native glycosylated rat Grp94 (rGrp94) and recombin- ant nonglycosylated Grp94 were immobilized using carbodiimide-activated amine coupling so that the chip surface might present random orientations of Grp94 molecules. Figure 1A shows a series of sensorgrams for the binding of VSV8 to rGrp94 at 25 °C and pH 7.4, corrected for the change in refractive index observed on the adjacent mock-immobilized surface of the same chip. The kinetics of both association and dissociation are near the limit of being too fast to be analysed quantitatively with confidence. Thus, the response values at the steady-state plateau were taken as a measure of the amount of ligand bound to the sensor surface. The sensorgrams show increasing response to concentrations of VSV8 in the range 10 lm to 1 mm, and from the hyperbolic equilibrium binding isotherm (Fig. 1B) a [A] 0.5 of  640 lm and a global R max (obs) of 27 RUÆ(ng proteinÆmm )2 ) was estimated. The binding affinity was decreased at 37 °C, Fig. 1. Binding of VSV8 to immobilized native glycosylated rGrp94 as measured by SPR. rGrp94 was immobilized on the sensor chip at 15 856 RU, and the binding of VSV8 was studied at 25 °C and pH 7.4 in HBS-P buffer. (A) Representative sensorgrams at 0, 100, 500 and 1000 l M VSV8. (B) The equilibrium-binding isotherm with a half-maximal response ([A] 0.5 )at 640 lM of VSV8 and a global R max (obs) of 27 RUÆ(ng proteinÆmm )2 ). (C) D ouble-reciprocal plot of the equilibrium-binding isotherms obtained in the absence of ATP ⁄ Ca 2+ (d), in the presence of 500 lM ATP (.) and in the pres- ence of 500 l M Ca 2+ (s) in the flow buffer. Grp94–ligand interactions M. Ying and T. Flatmark 514 FEBS Journal 273 (2006) 513–522 ª 2006 The Authors Journal compilation ª 2006 FEBS but slightly increased by lowering the pH from 7.4 to 6.5 [approximate pH of the ER ⁄ Golgi intermediate (ERGIC) compartment] [13] (not shown). Essentially identical results and binding parameters were obtained for recombinant nonglycosylated Grp94 (not shown). Moreover, the low-molecular mass analytes ATP and Ca 2+ at concentrations near the physiological levels in the ER were both found to inhibit rGrp94 binding of VSV8 (Fig. 1C), most pronounced for ATP at low concentrations of the peptide (see below). However, at this ATP concentration we observed no significant effect on the binding of VSV8 to the recombinant ER chaperone Grp78 ⁄ BiP (not shown). Although SPR typically offers no direct information about stoichiometry, an estimate was obtained from Eqn 2 (Experimental Procedures) by assuming a single binding site (n ¼ 1) per rGrp94 monomer (96 kDa) for VSV8 and the same refractive index increment [dn ⁄ dc (cm 3 Æg )1 )] as for the immobilized protein [14]. Thus, an R max (theoretical) of 4.9 RUÆ(ng proteinÆmm -2 ) was cal- culated for the rGrp94 monomer, which represents  18% of the R max (obs) of 27 RUÆ(ng proteinÆmm -2 )as calculated from the binding isotherm (Fig. 1B). VSV8 is a water soluble peptide and is shown structurally to be in an extended and flexible monomeric state [15], and it was most likely a monomer at the experimental conditions used. This indicates that rGrp94 contains multiple low-affinity binding sites for VSV8. Binding of ATP, MgATP and NECA to rGrp94 The function of Grp94 as a protein chaperone is repor- ted to be sensitive to ER lumenal ATP levels [11]. The chaperone has been reported to bind ATP and ADP with estimated K d -values in the millimolar concentration range [16], and the recently obtained three-dimensional structure of its N-terminal domain in complex with ATP and ADP has revealed a nucleotide-induced conforma- tional switch in the chaperone [17]. SPR analyses of the full-length glycosylated protein are in agreement with both observations. ATP binds to the full-length rGrp94 and the binding results in a negative SPR signal (Fig. 2A). This unusual response can only be explained by a ligand-induced conformational change, because a positive signal was expected due to the increased surface mass contribution of bound ATP (507 Da) as observed in a control experiment with the ER chaperone Grp78 ⁄ BiP (Fig. 2C). From the negative response iso- therm obtained within the concentration range of 0–500 lm ATP an apparent [A] 0.5 of  90 lm was esti- mated (Fig. 2B). Owing to the mixed contribution of ATP binding to the overall SPR signal response, i.e. a positive DRU due to an increased surface mass concentration, Fig. 2. Binding of ATP to immobilized native glycosylated rGrp94 and recombinant Grp78 ⁄ BiP as measured by SPR. rGrp94 and Grp78 ⁄ BiP were immobilized on the sensor chip at 15 945 and 2323 RU, respectively, and the binding of ATP was studied at 25 °C and pH 7.4 in HBS-P buffer. (A) Representative sensorgrams for rGrp94 at 10, 75, 150 and 300 l M of ATP. (B) The equilibrium- response isotherm for rGrp94 within the range 0–500 l M ATP with a half-maximal response ([A] 0.5 )at 90 lM ATP. (C) The equilib- rium-response isotherm for Grp78 ⁄ BiP with a half-maximal response ([A] 0.5 )at 170 lM ATP. M. Ying and T. Flatmark Grp94–ligand interactions FEBS Journal 273 (2006) 513–522 ª 2006 The Authors Journal compilation ª 2006 FEBS 515 and a negative DRU due to the ligand-induced conform- ational change, no exact value for the binding affinity and the stoichiometry of ATP binding could be deter- mined. By contrast, in the same concentration range MgATP (at a molar ratio of Mg 2+ : ATP ¼ 2 : 1) gave a positive SPR signal with an apparent [A] 0.5 of 5.8 mm (Fig. 4B). Almost the same response isotherm was obtained for nonglycosylated recombinant Grp94 (not shown), and from Fig. 2C it is seen that in a control experiment with Grp78 ⁄ BiP ATP binds with a posit- ive SPR response isotherm, and [A] 0.5 was  170 lm. DRU max [1.75 RUÆ(ng proteinÆmm )2 )] calculated from the response isotherm by nonlinear regression analysis is similar to the R max (theoretical) ¼ 1.63 calculated from Eqn 2 using the estimated dn ⁄ dc value for ATP (i.e. 63% of that for VSV8), indicating an apparent 1 : 1 stoichiometry of ATP binding to Grp78 ⁄ BiP monomer (78 kDa). Grp94 was found to bind the substituted adenosine analogue 5¢-(N-ethylcarboxamido) adenosine (NECA) with submicromolar affinity [16]. The crystal structure of a monomeric double-truncated form of canine Grp94 (residues 69–337D40) has more recently revealed that NECA binds to a conserved adenine-binding cav- ity and a second partially hydrophobic pocket, and the presence of an adenosine nucleotide-induced conforma- tional switch was predicted [18] and recently confirmed [17]. Our SPR analyses (Fig. 3) also support the con- clusion that NECA (308 Da) binds to native rGrp94 with high affinity, relative to ATP. In contrast to ATP the SPR response to NECA was positive, and the equilibrium binding isotherm was hyperbolic with a well-defined saturating SPR response and an apparent [A] 0.5 of  550 nm. Binding of divalent cations to rGrp94 The function of Grp94 as a protein chaperone is also sensitive to ER lumenal calcium levels [12,19]. Like the other abundant lumenal proteins of the ER, Grp94 is a low-affinity, high-capacity calcium-binding protein, making it one of the important calcium-storage and -buffer proteins of the ER [20]. Our SPR binding stud- ies (Fig. 4A) revealed that Ca 2+ (40 Da) binds to rGrp94 at pH 7.4 with a positive SPR signal and the Fig. 3. Binding of NECA to immobilized native glycosylated rGrp94 as measured by SPR. rGrp94 was immobilized on the sensor chip at 16 239 RU, and the binding of NECA was measured within the concentration range of 0–90 l M at 25 °C and pH 7.4 in HBS-P buf- fer containing 1% (v ⁄ v) dimethylsulfoxide. The hyperbolic equilib- rium binding isotherm revealed a well-defined saturating SPR response with a half-maximal response ([A] 0.5 )at 550 nM. Fig. 4. Binding of divalent cations to immobilized native glycosylat- ed rGrp94 as measured by SPR. rGrp94 was immobilized on the sensor chip at 18 230 RU, and the binding was studied at 25 °C and pH 7.4 in HBS-P buffer. (A) Representative sensorgrams obtained at 100, 400, 1000 and 3000 l M Ca 2+ . (B) The equilibrium- binding isotherms for Ca 2+ (d), Mg 2+ (s), MgATP (.) and Na + (n) with a half-maximal response ([A] 0.5 ) at 2.8 mM for Ca 2+ and 9.6 m M for Mg 2+ . Grp94–ligand interactions M. Ying and T. Flatmark 516 FEBS Journal 273 (2006) 513–522 ª 2006 The Authors Journal compilation ª 2006 FEBS ‘square-wave’ sensorgram is typical for a low-affinity binding analyte [21,22]. From the slightly sigmoidal equilibrium response isotherm (n H ¼ 1.2) an apparent [A] 0.5 of 2.8 mm and an R max of 39 RUÆ(ng pro- teinÆmm )2 ) was estimated by nonlinear regression ana- lysis (Fig. 4B). By contrast, at pH 6.5 a hyperbolic response isotherm (n H ¼ 1.0) was obtained with an apparent [A] 0.5 of 8.6 mm, demonstrating a reduced affinity and noncooperative binding at this acidic pH (not shown). From Fig. 4B it is seen that rGrp94 also binds Mg 2+ (24 Da) at pH 7.4 in the same concentra- tion range as Ca 2+ , but with a lower apparent affinity ([A] 0.5 ¼ 9.6 mm), indicating that at least some of the cation-binding sites are of a mixed Ca 2+ ⁄ Mg 2+ type. By contrast, Na + (NaCl) gave only a very small SPR response indicating very little contribution of an ionic strength effect to the responses observed for Ca 2+ (CaCl 2 ) and Mg 2+ (MgCl 2 ). Interestingly, the SPR response to Mg 2+ was higher than for MgATP at equal concentrations (Fig. 4B) which may be explained by the negative SPR signal resulting from the binding of ATP alone (Fig. 2). By assuming the same dn ⁄ dc (cm 3 Æg )1 ) for Ca 2+ as for rGrp94 (which is likely an overestimation), an approximate theoretical R max - value due to the contribution of the surface concentra- tion (ngÆmm )2 ) of the analyte alone was estimated from Eqn 2 to R max (theoretical) ¼ 43 RU at 17 000 RU immobilized, i.e.  2.5 RUÆ(ng proteinÆmm )2 ), for the n ¼ 15 binding sites previously estimated [20]. From the response isotherm (Fig. 4B) a R max of 39 RUÆ (ng proteinÆmm )2 ) was obtained by nonlinear regres- sion analysis indicating that the SPR response to Ca 2+ largely may reflect a concentration dependent global conformational change of the protein in addition to the increased surface mass concentration. Interaction between Grp94 and other molecular chaperones of the ER Several ER resident proteins are calcium-binding with low affinity and high capacity [20], and they have been proposed to weakly interact through a ‘calcium matrix’ [19]. In order to test whether Grp94 interacts with two of these proteins, the molecular chaperones Grp78 ⁄ BiP and calreticulin, rGrp94 and recombinant Grp78 ⁄ BiP were immobilized on CM5 sensor chips by the stand- ard procedure at a surface concentration of 3.4–5.6 ng proteinÆmm )2 , respectively. In the experiments with immobilized rGrp94 the injection of the recombinant forms of Grp78 ⁄ BiP and calreticulin, dissolved at vari- able concentrations in the flow buffer (HBS-P at pH 7.4 or 6.5), revealed no binding in the absence or presence of up to 3 mm Ca 2+ or 500 lm ATP (not shown). Similar negative binding results were obtained when Grp78 ⁄ BiP was immobilized on the sensor chip and rGrp94 was injected over the surface (not shown). Discussion The ER provides a tightly regulated environment for the folding and maturation of proteins destined to enter the secretory pathway [23], and the most abundant resident proteins (Grp94, Grp78 ⁄ BiP, protein disulfide isomerase and calreticulin) all function in protein fold- ing. A characteristic feature of these molecular chaper- ones is the wide diversity of ligand-binding properties, including specificity and affinity [23]. In this study we addressed the proposal that Grp94 also has a peptide acceptor–donor function related to antigen pres- entation [2], using SPR spectroscopy as a biophysical method to study in vitro its multiple binding properties. ATP- and Ca 2+ -induced conformational changes of native rGrp94 Although there is general agreement that Grp94 binds ATP and has a low ATPase activity, their functional significance has been controversial [4,16,17,24]. In this study, ATP was found to bind to native glycosylated and nonglycosylated recombinant Grp94 with relat- ively low apparent affinity, and with a complex SPR response isotherm. In contrast to the titration of the ER chaperone Grp78 ⁄ BiP with ATP (Fig. 2C), the SPR signal obtained with Grp94 was net negative at all concentrations of the nucleotide (Fig. 2B), showing that the response is dominated by an ATP-induced conformational change. This conclusion is in agree- ment with the change in conformation observed as a loss of an epitope-specific antibody binding in the pres- ence of ATP [7] and more recently crystallographically as a large conformational change in the N-terminal nucleotide binding domain of a C-terminal truncated form of Grp94 upon ATP binding [17]. These struc- tural studies also revealed a different binding mode for the adenosine analog NECA which in this study binds to rGrp94 with relatively high affinity and with a net positive SPR response. Interestingly, a similar negative conformation-dependent SPR response as observed for ATP binding to rGrp94 has been reported for the binding of maltose to the maltose-binding protein of Escherichia coli [25]. In that case it was shown crystal- lographically that the ligand binding induced a con- formational change which caused a net decrease in the hydrodynamic radius of the protein [26], and the observed negative SPR response was considered to be a function of the net change in hydrodynamic radius M. Ying and T. Flatmark Grp94–ligand interactions FEBS Journal 273 (2006) 513–522 ª 2006 The Authors Journal compilation ª 2006 FEBS 517 that occurs upon maltose binding [25]. By contrast, in our control experiment with Grp78 ⁄ BiP ATP was found to bind with a positive SPR signal (Fig. 2C), and the response isotherm gave a [A] 0.5 of 168 lm. Although Grp78 ⁄ BiP and Grp94 are the two major recipients of the pool of ATP translocated into the lumen of ER [24] the effects of ATP binding are quite different for the two chaperones. Thus, Grp78 ⁄ BiP has a weak ATPase activity which is stimulated by its binding of exogenous polypeptides [27] and is further regulated by ER cochaperones [28]. Grp94, by con- trast, has an unusually weak ATPase activity [4,16] which is inhibited or not stimulated by exogenous pep- tides [4], and the precise role of ATP binding ⁄ hydro- lysis in the in vivo chaperoning function of Grp94 remains unclear [1,29]. Four ER lumenal calcium-binding glycoproteins were originally reported in isolated rat liver microsomes [20]. A major component was Grp94, suggesting that one of its roles might be in the calcium-storage and -buffer function of the ER. Moreover, calcium has been shown to be required for the retention of Grp94 in the ER [12], as well as to be involved in the transport of secretory proteins out of the ER [19], and Grp94 may thus also be considered as a Ca 2+ sensor protein. In this study, it is confirmed that native rGrp94 possess multiple low-affin- ity binding sites for Ca 2+ [20]. As reported above (see Results) the theoretical maximum SPR signal [R max (the- oretical) ¼ 2.5 RUÆ(ng proteinÆmm )2 )] assuming 15 binding sites for Ca 2+ [20] was markedly lower than the observed SPR value [R max ¼ 39 RUÆ(ng proteinÆ mm )2 )], indicating a major contribution of a Ca 2+ - induced conformational transition to the overall SPR response. This finding is in agreement with a previous far-UV CD spectroscopic study that calcium causes a conformational change in Grp94 with a decrease in the a-helical content from 40 to 34% [20]. Interestingly, the large conformation-dependent enhancement of the SPR response observed in our study is similar to that previ- ously observed for Ca 2+ binding to tissue transglutami- nase [25], an allosteric enzyme which undergo significant conformational changes, including an increase in its hydrodynamic radius, upon binding of Ca 2+ [30]. Peptide-binding properties of native rGrp94 and its inhibition by physiological ER ligands Since the original studies by Srivastava et al. [31] it has been considered that Grp94 has the ability to bind a subset of immunogenic peptides and is actively involved in loading transporter associated with antigen presentation (TAP)-translocated peptides onto MHC class I molecules [2]. However, a direct role of Grp94 in antigen presentation is highly controversial and has still to be formally proven [8,9]. In particular, the reported low affinity and low stoichiometry of peptide binding to Grp94 in vitro, and the finding that the pep- tide binding seems to be almost irreversible at the selected experimental in vitro conditions (usually invol- ving heat shock denaturation and aggregation of the chaperone), are observations which have questioned its proposed peptide acceptor–donor function in vivo [8]. In this study, a low affinity was observed for the fully reversible binding of VSV8 to native glycosylated and nonglycosylated recombinant Grp94 (pH 7.4 or 6.5 and 25 or 37 °C), with a hyperbolic equilibrium bind- ing isotherm and half-maximal binding at  640 lm (pH 7.4 and 25 °C). A similar low affinity was observed for the binding of VSV8 to Grp78 ⁄ BiP, i.e. [A] 0.5  560 lm (not shown), in agreement with the previously published low-affinity binding of a library of 7-mer peptides to this chaperone [27]. Moreover, at saturation (extrapolated value) an apparent maximal stoichiometry  6 mol peptide per mol of rGrp94 monomer was estimated (see Results) which presents experimental support for the recent hypothetical pro- posal [9] of a binding of peptides to this chaperone by an adsorptive, rather than a bioselective, mechanism. A further argument [9] against the function of Grp94 in binding and transferring peptides for antigen presen- tation has been that peptide binding in vitro is consid- ered to be very stable, almost irreversible [5,7,8,10]. Thus, previous in vitro studies on Grp94 binding of radiolabelled synthetic peptides were performed under nonequilibrium binding conditions using a chaperone exposed to either heat shock denaturation and aggre- gation (50–60 °C) or to a chemically induced conform- ational change in connection with long incubation times (1–20 h) [5–7,10]. In this study, the SPR analyses revealed an equilibrium binding of VSV8 to native rGrp94 at pH 7.4 and 25 or 37 °C which was com- pletely reversible. Moreover, the rate of dissociation was very rapid (t ½ ¼ 7.2 ± 1.2 s; n ¼ 10) and incom- patible with the proposed peptide acceptor–donor function of the chaperone [31]. Thus, the VSV8-bind- ing sites ⁄ interactions in native Grp94 (this study) appear to be different from that observed for non- native forms of the protein [7,10]. A further argument is our finding that the binding of VSV8 was partly inhibited by ATP at concentrations (500 lm) consid- ered to be physiological in the ER and ERGIC lume- nal compartments [24], and most pronounced at the low concentrations of VSV8. The inhibitory effect of ATP is most likely related to the ATP-induced con- formational change discussed above, and a certain con- formation-related inhibition was also observed at the Grp94–ligand interactions M. Ying and T. Flatmark 518 FEBS Journal 273 (2006) 513–522 ª 2006 The Authors Journal compilation ª 2006 FEBS reported high (mm) total concentration of Ca 2+ in the ER lumen [32]. This may explain a previous observa- tion that Grp94 and six other ER proteins were selec- tively bound to an affinity column of denatured histone and specifically eluted by ATP, and that the release of Grp94 was further stimulated by Ca 2+ ⁄ Mg 2+ [33]. Additional, as yet unidentified, cofactors present in the lumenal ER compartment in vivo may also affect the peptide-binding properties of native Grp94, inclu- ding its binding parameters. However, interactions of Grp94 with its two related ER lumenal chaperones Grp78 ⁄ BiP and calreticulin were excluded from this study. By analogy to Grp78 ⁄ BiP [28] and the cytoplas- mic Hsp90, it is possible that the function of Grp94 may be regulated by physical interactions with cochap- erones, but to date there is no direct evidence for the existence of ER homologues of Hsp90-associated pro- teins [1]. Moreover, it should be noted that a low affin- ity of peptide binding to native Grp94 has been observed also in experiments at the cellular level. Thus, Lammert et al. [34], in their studies on streptolysine- permeabilized cells, found that radiolabelled peptides are translocated into the ER lumen by the TAP trans- porter and thus bind to Grp94. The fact that the affinity-purified Grp94 could only be labelled with TAP-translocated peptides containing a photo-cross- linker (a photoreactive phenylalanine label) was inter- preted as reflecting a weak interaction of the peptides with the chaperone. Finally, recent quantitative and structural analyses of peptides extracted from a Grp94-enriched glycoprotein fraction, isolated by con- cavalin A affinity chromatography from a T-cell lym- phoma tumour cell line, revealed a peptide occupancy of only 0.1–0.4% compared with  100% occupancy for MHC class I molecules [35]. The specific nature of Grp94-associated peptides, compared with those pre- sented by MHC class I molecules, together with the far substoichiometric occupancy of Grp94, strongly suggested that Grp94 is not a peptide chaperone involved in antigen processing [35]. Concluding remarks Our data present experimental support for the recent proposal by Nicchitta et al. [9] that VSV8 under native conditions binds to Grp94 by an adsorptive, rather than a bioselective, mechanism, and thus further chal- lenge the proposed in vivo peptide acceptor–donor function of the chaperone in the context of APC acti- vation [2,31]. Moreover, Nicchitta et al. [8,9] also pro- posed an alternative mechanism, in which Grp94 may function by inducing an activation of the immune system independent of bound peptides. Such a mech- anism is supported by a recent study [36] demonstra- ting that Grp94 and its N-terminal fragment induce an enhancement of the humoral immune responses to a protein antigen (HbsAg), related to the chaperone as an adjuvant. Experimental procedures The octapeptide VSV8 (RGYVYQGL), derived from VSV nucleoprotein (residues 52–59), was synthesized by Euro- gentec (Seraing, Belgium). NECA and ATP were purchased from Sigma (St. Louis, MO). All solutions used in the SPR analyses were purchased from Biacore AB (Uppsala, Sweden). Recombinant Grp94 and Grp78⁄ BiP were obtained from Stressgen Bioreagents (Victoria, Canada), and recombinant calreticulin from Abcam (Cambridge, UK). Polyclonal anti-Grp94 sera were obtained from StressGen Biotechnologies Corp. (Victoria, Canada). Purification of rGrp94 rGrp94 was purified from rough microsomes of rat pan- creas using a published procedure [37] with the following modifications. The lumenal protein fraction was applied to a concanavalin A–Sepharose column equilibrated with 50 mm phosphate buffer, pH 6.5 containing 0.2 m KCl at a slow flow rate at 4 °C for 12 h, and washed with the same buffer until A 280nm returned to  0. Glycoproteins were eluted using 0.5 m of methyl-a-d-mannopyranoride, and the eluate was concentrated and excess methyl-a-d-mannopyra- onoside removed by ultrafiltration (Centricon 30 from Amicon) before size-exclusion chromatography on a Super- dex 26 ⁄ 60 column (Pharmacia, Uppsala, Sweden). Purified rGrp94 was concentrated by ultrafiltration to a final concentration of 0.24 mgÆmL )1 . Small aliquots were flash- frozen in liquid N 2 and stored at )80 °C. The purity of the rGrp94 preparations was ‡ 95% as judged by SDS ⁄ PAGE and staining with Comassie Brilliant Blue and stains-all [38], and immunoblotting revealed only a single band of the expected subunit molecular mass ( 96 kDa) (not shown). The yield was about 32 lg rGrp94 per g of rat pancreas. SPR measurements SPR analyses were performed using the Biacore 3000 bio- sensor system (Biacore AB). Dimeric Grp94 (0.24 lgÆmL )1 ) was desalted, diluted in 10 mm sodium acetate, pH 4.0 to a final concentration of 0.03 lgÆmL )1 (exposure time  10 min), and immobilized covalently to the hydrophilic carboxymethylated dextran matrix of a CM5 sensor chip by the standard primary amine coupling reaction as described by the manufacturer. The amount of immobilized protein was estimated by assuming that 1000 RU M. Ying and T. Flatmark Grp94–ligand interactions FEBS Journal 273 (2006) 513–522 ª 2006 The Authors Journal compilation ª 2006 FEBS 519 correspond to  1 ng of immobilized proteinÆmm )2 [39], and a surface concentration of 10–19 ngÆmm )2 was used in the analyses. The responses to the different analytes were found to be proportional to the amount of immobilized proteins (range 10–19 ngÆ mm )2 ). A reference surface was subjected to the same procedure, but with no protein. A stable baseline was obtained in the cell with immobilized protein by a continuous flow (50 lLÆmin )1 ) of HBS-P run- ning buffer (10 mm Hepes, 150 mm NaCl, pH 7.4) for  1 h. This equilibration also removed any low-affinity lig- ands bound to the protein as isolated. All measurements were normally performed at 25 or 37 ° C with running buffer (pH 7.4 or 6.5) at a constant flow of 5–30 lLÆmin )1 . Each compound was dissolved in the running buffer and ana- lysed (in triplicate) using a two-to-five-fold dilution series. All sensorgrams were processed by first subtracting the SPR response observed for the reference surface. Because all the analytes, except NECA, were found to bind with low affinity we were as expected [21,22] unable to measure kin- etic rate constants for association and dissociation. More- over, for at least three of the analytes (ATP, Ca 2+ and Mg 2+ ) a conformational change of Grp94 represented a major contribution to the overall SPR response. The sensor- grams were therefore analysed by simple Langmuir bind- ing ⁄ response isotherms, and the equilibrium responses (R eq ¼ DRU at t ¼ 3 min) as a function of the free analyte (A) concentration was used to determine the concentration at half-maximal response ([A] 0.5 ) and the global R max (obs) by nonlinear regression analysis using the Jandel scientific Sigma Plot technical graphing software. The experimental error for the SPR response in replicate injections of the ana- lyte was found to be < 4%. The theoretical maximum R eq , R max (theoretical), was estimated using the equation [40], R max ðtheoreticalÞ¼ðM r;analyte =M r;ligand ÞnðAÞR immobilized ð1Þ where M r,analyte and M r,ligand are the relative molecular mass of the analyte and ligand, respectively; n is the number of analyte-binding sites on the protein; (A) is the fraction of active sites on the immobilized protein; and R immobilized is the RU of protein on the surface. The activity (A) of the immobilized ligands was assumed to be  40% at the high immobilization levels used (10–19 ng Æmm )2 ) [40]. Because the refractive index increment [dn ⁄ dc (cm 3 Æg )1 )] of the analytes may be different from that of the protein [14] this parameter was also considered in the calculations, giving the modified equation, R max ðtheoreticalÞ¼ðM r;analyte =M r;ligand Þnð0:4ÞR immobilized ððdn=dc analyte Þ=ðdn=dc ligand ÞÞ ð2Þ where (dn ⁄ dc analyte ) and (dn ⁄ dc ligand ) are the refractive index increment for the analyte and ligand, respectively. Similar (dn ⁄ dc) values have been determined for bovine serum albu- min (0.190 cm 3 Æg )1 ) and alanine (0.192 cm 3 Æg )1 ) [14], and the same value was therefore used for Grp94 and VSV8. Note that Eqns 1 and 2 do not include any parameter for the global conformational changes which are often observed on low-molecular-mass ligand binding to immobi- lized proteins [25,41–43]. To account for different immobilization levels of protein, the SPR responses to analyte binding in each sensorgram were divided by the calculated amount of immobilized pro- tein and expressed as DRUÆ(ng proteinÆmm )2 ). Acknowledgements This work was supported by the Novo Nordic Foun- dation, Rebergs Legat, the Blix Family Fund and the University of Bergen. References 1 Argon Y & Simen BB (1999) GRP94, an ER chaperone with protein and peptide binding properties. Semin Cell Dev Biol 10, 495–505. 2 Srivastava PK (2005) Immunotherapy for human cancer using heat shock protein–peptide complexes. Curr Oncol Rep 7, 104–108. 3 Nieland TJ, Tan MC, Monne-van Muijen M, Koning F, Kruisbeek AM & van Bleek GM (1996) Isolation of an immunodominant viral peptide that is endogenously bound to the stress protein GP96 ⁄ GRP94. Proc Natl Acad Sci USA 93, 6135–6139. 4 Li Z & Srivastava PK (1993) Tumor rejection antigen gp96 ⁄ grp94 is an ATPase: implications for protein fold- ing and antigen presentation. EMBO J 12, 3143–3151. 5 Blachere NE, Li Z, Chandawarkar RY, Suto R, Jaik- aria NS, Basu S, Udono H & Srivastava PK (1997) Heat shock protein–peptide complexes, reconstituted in vitro, elicit peptide-specific cytotoxic T lymphocyte response and tumor immunity. J Exp Med 186, 1315– 1322. 6 Wassenberg JJ, Reed RC & Nicchitta CV (2000) Ligand interactions in the adenosine nucleotide- binding domain of the Hsp90 chaperone, GRP94. II. Ligand-mediated activation of GRP94 molecular chaperone and peptide binding activity. J Biol Chem 275, 22806–22814. 7 Vogen S, Gidalevitz T, Biswas C, Simen BB, Stein E, Gulmen F & Argon Y (2002) Radicicol-sensitive peptide binding to the N-terminal portion of GRP94. J Biol Chem 277, 40742–40750. 8 Nicchitta CV (2003) Re-evaluating the role of heat- shock protein–peptide interactions in tumour immunity. Nat Rev Immunol 3, 427–432. 9 Nicchitta CV, Carrick DM & Baker-Lepain JC (2004) The messenger and the message: gp96 (GRP94)–peptide interactions in cellular immunity. Cell Stress Chaperones 9, 325–331. Grp94–ligand interactions M. Ying and T. Flatmark 520 FEBS Journal 273 (2006) 513–522 ª 2006 The Authors Journal compilation ª 2006 FEBS 10 Gidalevitz T, Biswas C, Ding H, Schneidman-Duhovny D, Wolfson HJ, Stevens F, Radford S & Argon Y (2004) Identification of the N-terminal peptide binding site of glucose-regulated protein 94. J Biol Chem 279, 16543–16552. 11 Pasqualini E, Caillol N, Mas E, Bruneau N, Lexa D & Lombardo D (1997) Association of bile-salt-dependent lipase with membranes of human pancreatic microsomes is under the control of ATP and phosphorylation. Biochem J 327, 527–535. 12 Ying M, Sannerud R, Flatmark T & Saraste J (2002) Colocalization of Ca 2+ -ATPase and GRP94 with p58 and the effects of thapsigargin on protein recycling sug- gest the participation of the pre-Golgi intermediate compartment in intracellular Ca 2+ storage. Eur J Cell Biol 81, 469–483. 13 Appenzeller-Herzog C, Roche A-C, Nufer O & Hauri H-P (2004) pH-induced conversion of the transport lec- tin ERGIC-53 triggers glycoprotein release. J Biol Chem 279, 12943–12950. 14 Tumolo T, Angnes L & Baptista MS (2004) Determina- tion of the refractive index increment (dn ⁄ dc) of mole- cule and macromolecule solutions by surface plasmon resonance. Anal Biochem 333, 273–279. 15 Ota N & Agard DA (2001) Binding mode prediction for a flexible ligand in a flexible pocket using multi-confor- mation simulated annealing pseudo crystallographic refinement. J Mol Biol 314, 607–617. 16 Rosser MF & Nicchitta CV (2000) Ligand interactions in the adenosine nucleotide-binding domain of the Hsp90 chaperone, GRP94. I. Evidence for allosteric reg- ulation of ligand binding. J Biol Chem 275, 22798– 22805. 17 Immormino RM, Dollins DE, Shaffer PL, Soldano KL, Walker MA & Gewirth DT (2004) Ligand-induced con- formational shift in the N-terminal domain of GRP94, an Hsp90 chaperone. J Biol Chem 279, 46162–46171. 18 Soldano KL, Jivan A, Nicchitta CV & Gewirth DT (2003) Structure of the N-terminal domain of GRP94. Basis for ligand specificity and regulation. J Biol Chem 279, 48330–48338. 19 Sambrook JF (1990) The involvement of calcium in transport of secretory proteins from the endoplasmic reticulum. Cell 61, 197–199. 20 Van PN, Peter F & Soling H-D (1989) Four intracister- nal calcium-binding glycoproteins from rat liver micro- somes with high affinity for calcium. No indication for calsequestrin-like proteins in inositol 1,4,5-trispho- sphate-sensitive calcium sequestering rat liver vesicles. J Biol Chem 264, 17494–17501. 21 Karlsson R & Sta ˚ hlberg R (1995) Surface plasmon reso- nance detection and multispot sensing for direct moni- toring of interactions involving low-molecular-weight analytes and for determination of low affinities. Anal Biochem 228, 274–280. 22 Myszka DG (2004) Analysis of small-molecule interac- tions using Biacore S51 technology. Anal Biochem 329, 316–323. 23 Ellgaard L & Helenius A (2003) Quality control in the endoplasmic reticulum. Nat Rev Mol Cell Biol 4, 181– 191. 24 Clairmont CA, De Maio A & Hirschberg CB (1992) Translocation of ATP into the lumen of rough endo- plasmic reticulum-derived vesicles and its binding to luminal proteins including BiP (GRP 78) and GRP 94. J Biol Chem 267, 3983–3990. 25 Gestwicki JE, Hsieh HV & Pitner JB (2001) Using receptor conformational change to detect low molecular weight analytes by surface plasmon resonance. Anal Chem 73, 5732–5737. 26 Sharff AJ, Rodseth LE, Spurlino JC & Quiocho FA (1992) Crystallographic evidence of a large ligand- induced hinge-twist motion between the two domains of the maltodextrin binding protein involved in active trans- port and chemotaxis. Biochemistry 31, 10657–10663. 27 Flynn GC, Pohl J, Flocco MT & Rothman JE (1991) Peptide-binding specificity of the molecular chaperone BiP. Nature 353, 726–730. 28 YuM, Haslam RH & Haslam DB (2000) HEDJ, an Hsp40 co-chaperone localized to the endoplasmic reticulum of human cells. J Biol Chem 275, 24984– 24992. 29 Rosser MF, Trotta BM, Marshall MR, Berwin B & Nicchitta CV (2004) Adenosine nucleotides and the reg- ulation of GRP94–client protein interactions. Biochemis- try 43, 8835–8845. 30 Di Venere A, Rossi A, De Matteis F, Rosato N, Agro AF & Mei G (2000) Opposite effects of Ca 2+ and GTP binding on tissue transglutaminase tertiary structure. J Biol Chem 275, 3915–3921. 31 Srivastava PK & Udono H (1994) Heat shock protein– peptide complexes in cancer immunotherapy. Curr Opin Immunol 6, 728–732. 32 Bastianutto C, Clementi E, Codazzi F, Podini P, De Giorgi F, Rizzuto R, Meldolesi J & Pozzan T (1995) Overexpression of calreticulin increases the Ca 2+ capa- city of rapidly exchanging Ca 2+ stores and reveals aspects of their lumenal microenvironment and func- tion. J Cell Biol 130, 847–855. 33 Nigam SK, Goldberg AL, Ho S, Rohde MF, Bush KT & Sherman MY (1994) A set of endoplasmic reticulum proteins possessing properties of molecular chaperones includes Ca 2+ -binding proteins and members of the thioredoxin superfamily. J Biol Chem 269, 1744–1749. 34 Lammert E, Arnold D, Nijenhuis M, Momburg F, Hammerling GJ, Brunner J, Stevanovic S, Rammensee H-G & Schild H (1997) The endoplasmic reticulum-resi- dent stress protein gp96 binds peptides translocated by TAP. Eur J Immunol 27, 923–927. M. Ying and T. Flatmark Grp94–ligand interactions FEBS Journal 273 (2006) 513–522 ª 2006 The Authors Journal compilation ª 2006 FEBS 521 35 Demine R & Walden P (2005) Testing the role of gp96 as peptide chaperone in antigen processing. J Biol Chem 280, 17573–17578. 36 Li HT, Yan JB, Li J, Zhou M-H, Zhu X-D, Zhang Y-X & Tien P (2005) Enhancement of humoral immune responses to HBsAg by heat shock protein gp96 and its N-terminal fragment in mice. World J Gastroenterol 11, 2858–2863. 37 Wearsch PA & Nicchitta CV (1996) Purification and partial molecular characterization of GRP94, an ER resident chaperone. Protein Expr Purif 7, 114–121. 38 Campbell KP, MacLennan DH & Jorgensen AO (1983) Staining of the Ca 2+ -binding proteins, calsequestrin, calmodulin, troponin C, and S-100, with the cationic carbocyanine dye ‘Stains-all’. J Biol Chem 258 , 11267– 11273. 39 Jo ¨ nsson U, Fa ¨ gerstam L, Ivarsson B, Johnsson B, Karlsson R, Lundh K, La ¨ fas S& et al. (1991) Real-time biospecific interaction analysis using surface plasmon resonance and a sensor chip technology. Biotechniques 11, 620–627. 40 . Karlsson R, Roos H, Bruno J & Stolz 1 (1997) Practi- cal aspects concerning direct detection of low molecular weight analytes using BIACORE 2000. BIA J 18–21. 41 Salamon Z, Wang Y, Brown MF, Macleod HA & Tollin G (1994) Conformational changes in rhodopsin probed by surface plasmon resonance spectroscopy. Biochemistry 33, 13706–13711. 42 Sota H & Hasegawa Y (1998) Detection of conforma- tional changes in an immobilized protein using surface plasmon resonance. Anal Chem 70, 2019–2024. 43 Flatmark T, Stokka AJ & Berge SV (2001) Use of sur- face plasmon resonance for real-time measurements of the global conformational transition in human phenyl- alanine hydroxylase in response to substrate binding and catalytic activation. Anal Biochem 294, 95–101. Grp94–ligand interactions M. Ying and T. Flatmark 522 FEBS Journal 273 (2006) 513–522 ª 2006 The Authors Journal compilation ª 2006 FEBS . Binding of the viral immunogenic octapeptide VSV8 to native glucose-regulated protein Grp94 (gp96) and its inhibition by the physiological ligands ATP and Ca 2+ Ming Ying and Torgeir Flatmark Section. 2C), and the response isotherm gave a [A] 0.5 of 168 lm. Although Grp78 ⁄ BiP and Grp94 are the two major recipients of the pool of ATP translocated into the lumen of ER [24] the effects of ATP binding. rGrp94 contains multiple low-affinity binding sites for VSV8. Binding of ATP, MgATP and NECA to rGrp94 The function of Grp94 as a protein chaperone is repor- ted to be sensitive to ER lumenal ATP