Phosphoenzyme Conversion of the Sarcoplasmic Reticulum Ca 2 1-ATPase MOLECULAR INTERPRETATION OF INFRARED DIFFERENCE SPECTRA

Time-resolved Fourier transform infrared difference spectra of the phosphoenzyme conversion and Ca release reaction (Ca2E1-P 3 E2-P) of the sarcoplasmic reticulum Ca-ATPase were recorded at pH 7 and 1 °C in H2O and H2O. In the amide I spectral region, the spectra indicate backbone conformational changes preserving conformational changes of the preceding phosphorylation step. b-sheet or turn structures (band at 1685 cm) and a-helical structures (band at 1653 cm) seem to be involved. Spectra of the model compound EDTA for Ca chelation indicate the assignment of bands at 1570, 1554, 1411 and 1399 cm to Ca chelating Asp and Glu carboxylate groups partially shielded from the aqueous environment. In addition, an E2-P band at 1638 cm 21 has been tentatively assigned to a carboxylate group in a special environment. A Tyr residue seems to be involved in the reaction (band at 1517 cm in H2O and 1515 cm 21 in H2O). A band at 1192 cm 21 was shown by isotopic replacement in the g-phosphate of ATP to originate from the E2-P phosphate group. This is a clear indication that the immediate environment of the phosphoenzyme phosphate group changes in the conversion reaction, altering phosphate geometry and/or electron distribution.

Time-resolved Fourier transform infrared difference spectra of the phosphoenzyme conversion and Ca 2؉ release reaction (Ca 2 E 1 -P 3 E 2 -P) of the sarcoplasmic reticulum Ca 2؉ -ATPase were recorded at pH 7 and 1°C in H 2 O and 2 H 2 O. In the amide I spectral region, the spectra indicate backbone conformational changes preserving conformational changes of the preceding phosphorylation step. ␤-sheet or turn structures (band at 1685 cm ؊1 ) and ␣-helical structures (band at 1653 cm ؊1 ) seem to be involved. Spectra of the model compound EDTA for Ca 2؉ chelation indicate the assignment of bands at 1570, 1554, 1411 and 1399 cm ؊1 to Ca 2؉ chelating Asp and Glu carboxylate groups partially shielded from the aqueous environment. In addition, an E 2 -P band at 1638 cm ؊1 has been tentatively assigned to a carboxylate group in a special environment. A Tyr residue seems to be involved in the reaction (band at 1517 cm ؊1 in H 2 O and 1515 cm ؊1 in 2 H 2 O). A band at 1192 cm ؊1 was shown by isotopic replacement in the ␥-phosphate of ATP to originate from the E 2 -P phosphate group. This is a clear indication that the immediate environment of the phosphoenzyme phosphate group changes in the conversion reaction, altering phosphate geometry and/or electron distribution.
Muscle relaxation is mediated by the removal of cytosolic Ca 2ϩ by the Ca 2ϩ -ATPase of the sarcoplasmic reticulum membrane. The ATPase couples active Ca 2ϩ transport to the hydrolysis of ATP (1)(2)(3)(4) in a reaction cycle that is shown in Scheme 1. In an essential reaction step, ATP reacts with Asp-351 to form a phosphoenzyme intermediate (Ca 2 E 1 3 Ca 2 E 1 -P), which then converts from an ADP-sensitive form (Ca 2 E 1 -P) to an ADP-insensitive form (E 2 -P) that is more rapidly hydrolyzed. This phosphoenzyme conversion is associated with Ca 2ϩ release toward the sarcoplasmic reticulum lumen against the concentration gradient (1,3,5). The structural origin of the change of accessibility of the Ca 2ϩ sites and of the essential reduction of Ca 2ϩ affinity upon phosphoenzyme conversion Ca 2 E 1 -P 3 E 2 -P have yet to be elucidated.
The potentially large body of structural information regarding the ATPase reaction cycle provided by infrared spectroscopy has only recently begun to be exploited using the approach of effector molecule-induced infrared difference spectroscopy (6 -13). The method uses the release of effector molecules from biologically "silent" photolabile derivatives, termed caged compounds (14 -17), to generate high quality infrared difference spectra. (The reaction products and infrared difference spectra of caged ATP photolysis and of side reactions have been characterized (18 -20).) The absorbance changes seen in these difference spectra give evidence for conformational changes of the polypeptide backbone and for alterations to the environment of amino acid side chains that take place in the reaction investigated.
Spectra of the phosphoenzyme conversion reaction in the 1800 to 1000 cm Ϫ1 spectral range have previously been described (8) and in that work were calculated by subtracting two difference spectra obtained with two different types of samples; a normalized difference spectrum of the Ca 2 E 1 3 Ca 2 E 1 -P reaction was subtracted from a normalized difference spectrum of the Ca 2 E 1 3 E 2 -P reaction. The use of two different samples in the subtraction and the normalization to identical protein content limit the reliability of these spectra and make it desirable to obtain the phosphoenzyme conversion spectrum more directly. This is possible using time-resolved rapid scan Fourier transform infrared (FTIR) 1 spectroscopy (10). Using this approach, we present here the spectra of the phosphoenzyme conversion reaction obtained in H 2 O and 2 H 2 O after the release of unlabeled ATP or [␥- 18 O 3 ]ATP. In particular, bands of the putative Ca 2ϩ chelating carboxylate groups and of the phosphoenzyme phosphate group are discussed.

MATERIALS AND METHODS
Sample Preparation-Samples for time-resolved infrared spectroscopy of the Ca 2 E 1 P 3 E 2 -P reaction were prepared as described previously (8, 10) by removal of free water from a sarcoplasmic reticulum suspension equilibrated in H 2 O or 2 H 2 O buffer. Samples were immediately rehydrated with H 2 O or 2 H 2 O with or without 20% Me 2 SO. This method resulted in active ATPase samples (6). Approximate concentrations were 0.7 mM ATPase, 300 mM imidazole, pH 7.0, 1 mM CaCl 2 , 20 mM glutathione, 20 mM caged ATP, 0.5 mg/ml A23187, 2 mg/ml adenylate kinase, and 20% Me 2 SO in approximately 1 l of sample volume. FTIR Measurements-Time-resolved FTIR measurements of the Ca 2 E 1 -P 3 E 2 -P reaction were performed at 1°C with a modified Bruker IFS 66 spectrometer as described previously (10). Difference spectra for the reaction were obtained by subtracting a spectrum representing predominantly Ca 2 E 1 -P, and to a small extent E 2 -P (recorded 3.3-11 s after photolysis of caged ATP in H 2 O or after 11-19 s in 2 H 2 O), from a spectrum representing E 2 -P (recorded after 88 and 146 s in H 2 O and 2 H 2 O, respectively). The resulting spectrum was normalized as described (10) to the full amplitude of the absorbance changes associated with the phosphoenzyme conversion reaction.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
As the difference spectra were obtained directly from the time-resolved measurements, a normalization of spectra to an identical protein concentration is in principle not necessary. However, for a better comparison of samples in H 2 O and 2 H 2 O and to prevent the possible predominance of individual samples with high protein content in the averaged spectra, spectra were normalized to an identical protein concentration before averaging, as described (21).
Absorbance spectra in H 2 O and 2 H 2 O of the model compound for Ca 2ϩ release EDTA were recorded with and without Ca 2ϩ at 20°C and pH 12.8. Small variations in the path length between different samples were corrected by normalizing the water absorption of every sample to a single water spectrum that served as a standard for the path length. Sample concentrations were 84 mM.
Band-narrowing Procedures-Apart from second-derivative spectra, the following procedure was applied. The original spectrum was smoothed over 24 cm Ϫ1 and multiplied by 0.95. This spectrum was then subtracted from the original spectrum, which eliminates broad features of the original spectrum. The resulting spectrum is dominated by the fine-structure in the original spectrum, and thus we term the method "fine-structure enhancement" in the following text. When tested with protein absorbance spectra, this method gives results similar to Fourier self-deconvolution (data not shown). For a clearer presentation, finestructure enhanced spectra were multiplied by 10 and second-derivative spectra by Ϫ15.
To assess whether peaks in the fine-structure enhanced spectra correspond to "true" component bands or are the result of artifacts in the band-narrowing procedure (see Fig. 3), the original difference spectra were fitted, and the resulting fit was fine-structure enhanced and compared with the fine-structure enhanced original spectra (data not shown). Interestingly, it was found that the spectral region of 1700 -1670 cm Ϫ1 can accurately be fitted by only three bands at 1693 (Ϫ), 1686 (Ϫ), and 1670 cm Ϫ1 (ϩ) despite the plateau at 1678/1676 cm Ϫ1 in the fine-structure enhanced H 2 O spectrum (solid line in Fig. 3B). It is not necessary to introduce an additional band in the fitting model to reproduce that plateau.

RESULTS AND DISCUSSION
Model Spectra of Ca 2ϩ Release-From site-directed mutagenesis studies it is thought that Glu-309, Glu-771, and Asp-800 form part of the high affinity Ca 2ϩ binding sites of Ca 2 E 1 and Ca 2 E 1 -P (4). Thus, we studied the effect of Ca 2ϩ binding on the infrared spectrum of carboxylate groups using the Ca 2ϩ chelator EDTA. Fig. 1A shows absorbance spectra in 2 H 2 O of free EDTA (solid line) and of the EDTA complex with Ca 2ϩ (dotted line). The bands near 1590 and 1410 cm Ϫ1 can be assigned to the COO Ϫ antisymmetric stretching vibration ( as ) and the symmetric stretching vibration ( s ), respectively. The shoulder at 1612 cm Ϫ1 indicates some heterogeneity of the COO Ϫ groups, resulting from a small proportion of EDTA with protonated nitrogen atoms (22,23). Upon Ca 2ϩ release there is a downshift of 4 -8 cm Ϫ1 for both of the main bands (at 1590 and 1414 cm Ϫ1 ), which translates in the difference spectrum of Ca 2ϩ release (absorbance of free EDTA minus absorbance of the Ca 2ϩ complex) into the two minimum/maximum features at 1590/1568 and 1418/1400 cm Ϫ1 .
When 2 H 2 O is replaced by H 2 O, the bands of the antisymmetric stretching vibration are downshifted by 6 -8 cm Ϫ1 for free EDTA and for the complex (data not shown). The band of the symmetric stretching vibration is less sensitive and shifts SCHEME 1. downwards only for free EDTA (by 2 cm Ϫ1 ). A similar behavior has been observed for sodium acetate (24) and for Asp and Glu (25,26). Thus, the model spectra have identified marker band pairs for Ca 2ϩ release from carboxylate groups near 1575 and 1410 cm Ϫ1 . The former is expected to be sensitive to H 2 O 3 2 H 2 O replacement with a higher frequency in 2 H 2 O.
The Phosphoenzyme Conversion Spectrum- Fig. 2A shows the difference spectrum of the phosphoenzyme conversion and Ca 2ϩ release reaction, Ca 2 E 1 -P 3 E 2 -P (solid line), and the respective spectrum in 2 H 2 O buffer (dashed line). Positive bands are characteristic for E 2 -P, negative bands for Ca 2 E 1 -P.  (25). However, a strong salt bridge or hydrogen bond to one of the oxygens of a carboxylate group could shift the band well above 1600 cm Ϫ1 (24, 32), and a reasonable scenario for the band at 1638 cm Ϫ1 could be that the parting Ca 2ϩ is replaced by a strong hydrogen bond donor or a positive charge.
Bands at 1758 and 1710 cm Ϫ1 have tentatively been assigned to the protonation of at least two chelating carboxylate groups (33), which is in line with the small downshift observed upon Protein Backbone: the Amide I Region-The largest absorbance changes (up to 0.5% of the total protein absorbance) are observed in the amide I region of the spectrum (1700 -1610 cm Ϫ1 ) as are the strongest effects of protein deuteration. In this region, the amide I mode of the polypeptide backbone as well as the side chains of Asn, Gln, Arg, HisH ϩ , and Lys absorb strongly (25,26,30). In the lower wavenumber part of that region, Asp and Glu COO Ϫ groups with strong interactions may contribute as discussed above.
Large downshifts (Ն30 cm Ϫ1 ) of bands upon H 2 O 3 2 H 2 O replacement are characteristic for the side chain absorptions of Asn, Gln, Lys, and Arg, as are small upshifts as discussed above for COO Ϫ bands and downshifts of up to 10 cm Ϫ1 for amide I and HisH ϩ bands. No band shifts are expected for groups that are located in parts of the protein that are not accessible to deuteration.
The large number of alterations in this spectral region, when H 2 O is replaced by 2 H 2 O, makes it difficult to arrive at a unique explanation for the band shifts, and additional experiments as well as data processing were necessary, as described below.
The recording of spectra as soon as possible after H 2 O 3 2 H 2 O replacement (30 min to 1 h at 1°C) shows that most of the observed effects take place in protein regions readily accessible to deuteration. The spectrum shortly after deuteration (data not shown) is very similar to the one after prolonged incuba- tion. The only clear exception is the minimum at ϳ1630 cm Ϫ1 that develops over a few hours of incubation. This position is characteristic of amide I modes of ␤-sheet structures and of the CϭO group of deuterated Asn or Gln residues (26).
Band-narrowing procedures (see "Materials and Methods") were applied to identify band shifts upon 1 H 3 2 H exchange. Fig. 3A shows the original difference spectra and Fig. 3, B and C, the resulting spectra after band narrowing. These reveal essentially the same peak positions in H 2 O and 2 H 2 O (see Fig.  3, B and C), with one exception. The negative band at 1689 cm Ϫ1 in the unprocessed original H 2 O spectrum (solid line in Fig. 3A) is composed of at least two bands, giving minima in the processed spectrum at 1693 and 1685 cm Ϫ1 (solid line in Fig. 3, B and C). The highest wavenumber component of the negative band seems to be nearly unaffected by protein deuteration and is observed in 2 H 2 O at 1692 cm Ϫ1 . It could be caused by a conformational change of ␤-sheet or turn structures or an Asn, Gln, or Arg side chain that is located in the core of the protein and is inaccessible to deuteration.
The other component of the 1689 cm Ϫ1 band seems to shift from its position in the processed spectrum (Fig. 3B) (Fig. 3A). This shift is characteristic of amide I modes of the polypeptide backbone, and the position then indicates a conformational change of ␤-sheet or turn structures. These structures are predicted to occur only in the extramembraneous domains of the protein (34,35), and thus the observed conformational change is likely to take place in these protein domains.
As the positions of the other bands in the amide I region are hardly affected by deuteration, the isotope effects are the result either of intensity changes or of bands that are not evident after applying the band-narrowing procedures because they are broader than the ones detected. Narrow bands tend to dominate the processed spectra.
The bands hardly affected by 1  residue in a helical or unordered conformation. The imide band is found approximately 20 cm Ϫ1 lower than an amide band (36). Thus, the band at 1607 cm Ϫ1 , the position of which seems to be too low for an amide I band, could also be caused by a Pro imide group. Interestingly, there are three Pro residues in the putative Ca 2ϩ binding transmembrane helices M4 and M6, mutation of which affects Ca 2ϩ affinity and phosphoenzyme conversion (4). Alternatively, both bands may be caused by COO Ϫ side chain groups not interacting with bulk water. This assumption relies on the fact that the bands do not show the upshift upon H 2 O 3 2 H 2 O replacement characteristic for carboxylate groups in water.
Protein Backbone: the Amide II Region-None of the three bands in the amide II region (1570 -1530 cm Ϫ1 ) shows the strong sensitivity toward 1 H 3 2 H exchange expected for the amide II mode of the protein backbone. The amide II mode of backbone elements accessible to deuteration therefore does not seem to be affected by phosphoenzyme conversion and Ca 2ϩ release. Two of the bands (at 1570 and 1554 cm Ϫ1 ) have been tentatively attributed above to the as vibration of COO Ϫ groups.
Protein Backbone: the Amide III Region-The amide III mode absorbs between 1400 and 1200 cm Ϫ1 and is sensitive to deuteration (37). This property is observed in the spectra for the bands at 1337 and 1318 cm Ϫ1 (see Fig. 2A), which therefore might be attributed to amide III modes. The position of these bands is characteristic for turn structures (37), and they are probably related to the amide I band at 1687 cm Ϫ1 , which has tentatively been assigned to turn or ␤-sheet structures. Alternatively, the bands at 1337 and 1318 cm Ϫ1 may be caused by the ␦(COH) mode of Ser, Asp, or Glu with a weakly bonded OH group.
Side Chain Modes Other than Carboxyl Modes-As mentioned above, bands of the side chains of Asn, Gln, Arg, and Lys in the amide I region show relatively large shifts upon deuteration. The extinction coefficient of the former three residues is relatively high (25,26), whereas that of Lys is smaller. Thus, Lys bands may be masked by stronger bands. The HisH ϩ mode near 1631 cm Ϫ1 (30) shows a 10 cm Ϫ1 downshift in 2 H 2 O (26) and absorbs relatively strongly in H 2 O (30). These characteristic shifts have not been observed in the spectra, and thus there is no clear evidence for the participation of Asn, Gln, Arg, and HisH ϩ in the phosphoenzyme conversion and Ca 2ϩ release reaction. This statement holds only for those residues that are accessible to 1 H 3 2 H exchange, which should include residues in the ATP binding site, the catalytic site, and at least part of the Ca 2ϩ binding sites, because several bands that were tentatively assigned to the Ca 2ϩ binding sites show an effect upon The position of the band at 1517 cm Ϫ1 and its slight down-shift of 2 cm Ϫ1 in 2 H 2 O is characteristic for a ring mode of protonated Tyr (25,26,30). Also, the band pairs at 1283 and 1264 cm Ϫ1 may be attributed to the (C-O) mode of Tyr but also to a Trp mode, which is observed for indole at 1276 cm Ϫ1 (38,39). It has been suggested that Tyr-763 is involved in the cytoplasmic gate to the Ca 2ϩ binding sites (4). The very small band at 1365 cm Ϫ1 might be caused by a ␦ s (CH 3 ) mode of aliphatic side chains.
Absorption of the Phosphate Group- Fig. 2B shows a comparison between the conversion spectrum obtained with unlabeled ATP and that obtained with [␥-18 O 3 ]ATP. With the heavier isotope, phosphate bands are expected to be downshifted, thereby enabling the identification in the spectrum of alterations to the phosphate group. As the ␥-phosphate is transferred to Asp-351 before phosphoenzyme conversion, differences between the spectra with the two isotopes will identify the absorbance of the phosphoenzyme phosphate group. As expected, the two spectra superimpose very well above 1250 cm Ϫ1 , where phosphate groups do not absorb. However, the band at 1192 cm Ϫ1 is reduced upon isotopic substitution. Instead, the intensity is higher for [␥- 18 O 3 ]ATP between 1180 and 1150 cm Ϫ1 . A difference spectrum between labeled and unlabeled phosphate group (data not shown) shows that the band at 1192 cm Ϫ1 for the unlabeled phosphate seems to shift to 1157 cm Ϫ1 for the labeled phosphate, in agreement with the expected isotopic shift of 20 -30 cm Ϫ1 for phosphate groups (18,40). This identification of a phosphate band in the difference spectrum clearly shows that the conversion reaction considerably changes the environment of the phosphoenzyme phosphate group, thus affecting its electron density distribution and/or binding geometry. The band position at 1192 cm Ϫ1 is rather unusual for a phosphate group and could be explained in two ways: (i) a widening of the P-O angles leading to a stronger coupling between the P-O vibrations and thus to a stronger splitting between the as and s modes; or (ii) an increase in electron density of some or all of the P-O bonds, either by breaking of the hydrogen bonds to all phosphate oxygens or by strong hydrogen binding to one or two of the phosphate oxygen atoms, thus increasing the electron density in the other P-O bond(s). The band is also affected by H 2 O 3 2 H 2 O replacement, which seems to indicate that the phosphate oxygen interacts with either a water molecule or deuterated protein residues.
The active site of E 2 -P was shown to be in a closed conformation (41) shielded from bulk water with a hydrophobic environment detected close to the ribose OH groups of a fluorescent ATP analogue (42,43). This finding is in contrast to the properties of Ca 2 E 1 -P, and it is thought that the decrease in the active site water activity upon the Ca 2 E 1 -P 3 E 2 -P conversion is responsible for the higher hydrolysis rate of E 2 -P as compared with Ca 2 E 1 -P (44). The infrared spectra presented here show that this conformational change has a direct effect on phosphate conformation and/or electron density distribution. Creation of a hydrophobic environment alone however, may not be sufficient to explain the phosphate band at 1192 cm Ϫ1 , because dehydration of the model compound acetyl phosphate shifts the as PO 3 2Ϫ band only from 1132 to 1177 cm Ϫ1 at most (data not shown).
Comparison with Spectra of ATPase Phosphorylation- Fig. 4 shows spectra in 2 H 2 O of ATPase phosphorylation Ca 2 E 1 ⅐ATP 3 Ca 2 E 1 -P (solid line) (21) and of the overall reaction of phosphorylation and phosphoenzyme conversion Ca 2 E 1 ⅐ATP 3 E 2 -P (dotted line). These spectra were obtained from timeresolved measurements of the same set of samples. Negative bands are characteristic for Ca 2 E 1 ⅐ATP, positive bands either for Ca 2 E 1 -P (solid line) or E 2 -P (dotted line). (The spectrum of Ca 2 E 1 ⅐ATP 3 Ca 2 E 1 -P also shows a small contribution because of the E 2 -P that is already formed, as indicated by the E 2 -P marker band near 1750 cm Ϫ1 .) The spectral region shown includes the amide I region (1700 -1610 cm Ϫ1 ) that is sensitive to conformational changes of the protein backbone. The comparison indicates that bands characteristic for Ca 2 E 1 -P are still present in E 2 -P, which is especially obvious for the bands at 1682, 1630, and 1593 cm Ϫ1 . This finding indicates that at least some of the alterations to the protein conformation induced by ATPase phosphorylation seem to be preserved or even "enhanced" in the subsequent transition to E 2 -P. It seems as if the enzyme conformation on the way from Ca 2 E 1 -P to E 2 -P goes further away from the pre-phosphorylation conformation of Ca 2 E 1 ⅐ATP instead of returning to it. CONCLUSIONS Infrared difference spectra show that the protein conformation changes in the phosphoenzyme conversion reaction, preserving conformational changes of the preceding step of enzyme phosphorylation. ␤-sheet or turn structures of the extramembraneous domains and most likely ␣-helical structures seem to be affected. The net change of secondary structures, however, is small (10).
The release of Ca 2ϩ proceeds from carboxylate groups that seem to be partly shielded from the aqueous environment. It is associated with the protonation of at least two carboxylate groups presumably involved in Ca 2ϩ chelation. A change of environment of a Tyr residue was detected as well as a direct effect of phosphoenzyme conversion on the geometry and/or electron density distribution of the phosphoenzyme phosphate group. The latter is likely to be a prerequisite for the higher susceptibility toward hydrolytic attack of E 2 -P as compared with Ca 2 E 1 -P. Interestingly, the currently assigned bands of the phosphate group and of Ca 2ϩ release from carboxylate groups appear with the same reaction rate (10), showing that Ca 2ϩ release and the change of phosphate environment proceed at the same time. This finding rules out a significant population of Ca 2 E 2 -P, a postulated state (45) with phosphate properties of E 2 -P that still binds Ca 2ϩ .