Hydrogen/deuterium exchange studies of native rabbit MM-CK dynamics

Abstract

Creatine kinase (CK) isoenzymes catalyse the reversible transfer of a phosphoryl group from ATP onto creatine. This reaction plays a very important role in the regulation of intracellular ATP concentrations in excitable tissues. CK isoenzymes are highly resistant to proteases in native conditions. To appreciate localized backbone dynamics, kinetics of amide hydrogen exchange with deuterium was measured by pulse-labeling the dimeric cytosolic muscle CK isoenzyme. Upon exchange, the protein was digested with pepsin, and the deuterium content of the resulting peptides was determined by liquid chromatography coupled to mass spectrometry (MS). The deuteration kinetics of 47 peptides identified by MS/MS and covering 96% of the CK backbone were analyzed. Four deuteration patterns have been recognized: The less deuterated peptides are located in the saddle-shaped core of CK, whereas most of the highly deuterated peptides are close to the surface and located around the entrance to the active site. Their exchange kinetics are discussed by comparison with the known secondary and tertiary structures of CK with the goal to reveal the conformational dynamics of the protein. Some of the observed dynamic motions may be linked to the conformational changes associated with substrate binding and catalytic mechanism.

Creatine kinase (CK) isoenzymes play a crucial role in the bioenergetics of the cells of excitable tissues (Wyss et al. 1992). The muscle cytosolic form (MM-CK), as well as the other cytosolic isoenzymes, is a dimeric molecule in contrast with the mitochondrial isoenzymes, which have the unique property of forming interconvertible dimers as well as octamers. Sequence alignments of the CK isoenzymes reveal six highly conserved regions that form a compact core involved in substrate binding and catalysis (Fritz-Wolf et al. 1996). The native enzyme contains four thiol groups but no disulfide bond. X-ray–resolved structures of octameric mitochondrial CK isoform (Fritz-Wolf et al. 1996); dimeric cytosolic CKs from rabbit, ox, and man (Rao et al. 1998; Shen et al. 2001; Tisi et al. 2001); and monomeric arginine kinase (Yousef et al. 2003) were recently obtained. They all share the same subunit topology: Each monomer has a small α-helical N-terminal domain and a large C-terminal domain containing an eight-stranded antiparallel β-sheet flanked by seven α-helices. The active site is located in a cleft between the domains. Surface loops with a high flexibility have been observed in all CKs of known structure around residues 59–69, 182–204, and 320–330. They are believed to be involved in rearrangements of the structure upon substrate binding, to shield the active site from water during catalysis (Kabsch and Fritz-Wolf 1997). In all CKs, several monomer–monomer interfaces allow for the formation of a very stable dimer (Webb and Morris 2001; Lahiri et al. 2002).

Native CK isoenzymes proved to be highly resistant to specific proteases such as trypsin or chymotrypsin despite the existence of numerous potential digestion sites (Price and Stevens 1982; Morris 1989; Wyss et al. 1993). However, they are cleaved by nonspecific proteases, such as proteinase K, at a single site that is hypersensitive to protease attack (Williamson et al. 1977; Price et al. 1981; Lough et al. 1985; Morris et al. 1985; Lebherz et al. 1986; Wyss et al. 1993). This site was localized in a C-terminal region of the protein (Wyss et al. 1993; Leydier et al. 1997), which was later shown to be an exposed mobile surface loop (Fritz-Wolf et al. 1996; Rao et al. 1998). The main consequence of limited proteolysis is the complete inactivation of the enzyme. The two fragments remain associated with each other, and the global physical and chemical properties of the enzyme are not affected as long as the dimer is not dissociated by addition of a denaturant (Clottes et al. 1997; Raimbault et al. 1997).

This high resistance to proteolysis indicates that CK has a compact structure. However, an enhanced susceptibility to proteases has been described as a consequence of exposure to a low denaturant concentration (Webb et al. 1997). This destabilization results from partial unfolding within the hydrophobic shell of the protein and from a disruption of monomer–monomer contacts (Fontana et al. 1997; Webb and Morris 2001), which also allows some monoclonal antibodies to gain access to their otherwise hidden epitopes (Morris and Cartwright 1990).

Measurements of protein dynamics are key to understanding their behavior; the static protein structure by itself is insufficient to describe the complex functions they can perform. Isotopic exchange rates of amide hydrogens (NHs) have long been used to provide information about the structural dynamics of proteins. When a protein is placed in a solution of D₂O, labile hydrogens in the protein are replaced with deuterium at varying rates, depending on intramolecular hydrogen bonding and solvent accessibility. Interest in this technique has increased substantially with the introduction of NMR and mass spectrometry (MS) to measure deuterium incorporation (Engen and Smith 2000) and by the development of deuterium labeling followed by pepsin digestion to localize exchange to short fragments of large proteins (Zhang and Smith 1993; Zhang et al. 1996). MS offers some advantages such as a high sensitivity, which allows working with low concentrations, and the ability to analyze large proteins. Hydrogen exchange detected by MS has been particularly effective for detecting different conformational states.

Hydrogen exchange may involve a two-step process in which a structural change occurs in the native state of a protein to facilitate isotopic exchange (Englander and Kallenbach 1984). The structural change may be localized to a small region, facilitating the exchange of one or several NHs within this segment, or it may involve global unfolding of a domain making way for the exchange of all of its NHs. Exchange through local and global unfolding may operate in parallel in different regions of a protein.

The two steps required for NH exchange are illustrated in equation 1, where F is the folded form and U is the unfolded form (Hvidt and Nielsen 1966).

(1)

If k_cl ≫ k_c, the structural change facilitating hydrogen exchange occurs many times before isotopic exchange takes place, and the exchange is described by EX2 kinetics (Hvidt and Nielsen 1966; Bai et al. 1994). EX2 kinetics lead to a random distribution of deuterium among the peptide units, which gives a single envelope of isotope peaks in the mass spectra of partially labelled peptides. In contrast, if k_cl ≪ k_c, all sites exposed to the deuterated solvent will undergo isotopic exchange in the first opening. This form of exchange, which is called EX1 kinetics, often gives bimodal isotope patterns.

EX1 and EX2 kinetics can be clearly distinguished from the distribution of isotope peaks in mass spectra. A random distribution of deuterium among the sample molecules, detected as a single binomial distribution of isotope peaks in mass spectra, indicates hydrogen exchange by EX2 kinetics. A bimodal distribution of isotope peaks is evidence for EX1 kinetics (Miranker et al. 1993). The two envelopes of mass are a low-mass envelope, which corresponds to the less exchanged population and a high-mass envelope, which represents the fully exchanged population. Therefore NH exchange has proven to be an excellent and sensitive probe of protein structure and dynamics. This technique has been successfully used to analyze the native conformational dynamics of several proteins: aldolase (Zhang et al. 1996), the hematopoietic cell kinase SH3 and SH2 domains (Engen et al. 1997, 1999), and MAP kinase kinase-1 (Resing and Ahn 1998). The aim of the present work was, by using hydrogen/deuterium (H/D) exchange and MS, to document the fluctuations and solvent accessibility of MM-CK in order to map its solution dynamics.

Results

Pepsin digestion of native CK

The main steps in the protein fragmentation/MS methods are as follows: (1) H/D exchange for a specified time, (2) quenching of isotopic exchange by lowering pH and temperature, (3) fragmenting the deuterated protein by pepsin, and (4) determining the molecular weights of the deuterated peptides by coupled HPLC/ESIMS. The total time required for proteolytic fragmentation, desalting, and HPLC separation was ~30 min. Under the quenching conditions, the exchange half life at most peptide linkages is >1 h (Bai et al. 1993). The extent of back exchange for the various proteolytic peptides ranged from 20% to 48% during quench and liquid chromatography (LC)-MS analysis. These values are consistent with the 25% to 55% or 18% to 54% reported by Wang et al. (1998a^,b).

The sequence (numbered from Pro 1) and the secondary structure of rabbit muscle MM-CK (Rao et al. 1998) are presented in Figure 1 ▶, with lines indicating the 47 peptides identified by MS/MS and analyzed in this study. The shortest and longest segments analyzed in this study contained three and 36 peptide linkages, respectively, and 96% of the entire CK backbone was covered.

Figure 1.

Location with respect to the rabbit MM-CK amino acid sequence of pepsin-cleaved mass-identified fragments for which deuterium incorporation was measured. Secondary structural elements determined by X-ray crystallography (Rao et al. 1998) are shown above the sequence of the native protein numbered from proline 1.

Throughout this section, reference will be made to digestion sites and antibody epitopes identified by the Morris group (Webb and Morris 2001) in the chick muscle CK, which shows >90% identity with rabbit muscle CK (Mühlebach et al. 1994).

Rates of amide isotopic exchange

Deuterium levels in the 47 peptides derived from CK that has been incubated in D₂O for 3 sec to 90 min were determined from the molecular weights of the corresponding peptic fragments. Isotope distributions of all the fragments were monomodal, the average molecular weight shifting more or less rapidly to higher values with increasing deuteration time.

Figure 2 ▶ shows the kinetics of deuterium incorporation monitored by ESIMS, for two doubly charged peptic peptides. Figure 2A ▶ shows the 202–224 peptide, which is slowly and lowly deuterated, whereas Figure 2B ▶ displays an example of a more quickly and fully deuterated peptide (312–330).

Figure 2.

Incorporation of deuterium by two peptic fragments of CK. ESI mass spectra are shown for the +2 charge state of the 202–224 peptide (A) and the 312–330 peptide (B). The deuteration times are indicated. Results for CK that contained no deuterium are presented in the top panel. “0% Ref” corresponds to CK exposed to deuterium only during quenching and proteolysis steps, and “100% Ref” corresponds to CK that was completely deuterated.

Results from 14 segments representative of most of the CK backbone are presented in Figure 3 ▶. The corrected percentage of deuteration of each peptide is plotted versus incubation time. The curves showing the kinetics of deuteration of these peptides have been arranged arbitrarily according to the extent and rates of hydrogen exchange and have been modeled by using first-order kinetics according to equation 3 with two or three exponentials. For the analysis of these kinetics, the NHs in each fragment were grouped according to their exchange rates. Some of the most labile hydrogens were exchanged within the dead-time of the experiment. The fraction of very slow exchanging NHs was estimated by subtracting the number of NHs exchanged after 90 min from the total number of exchangeable NHs in each peptide. The number of exchangeable NHs and their rate constants in the various probes analyzed in this report are summarized in Table 1. Those peptides with exchange rate constants >~10 min⁻¹ correspond to the protons that are not shielded or engaged in H bonds. They are freely accessible to the D₂O solvent because exchange rate constants >10 min⁻¹ are typical of unfolded peptides (Bai et al. 1993). The constants determined for the rapidly (0.1 to 4.5 min⁻¹) and slowly (0.02 to 0.07 min⁻¹) exchanging NHs are related to protons that are shielded or engaged in intramolecular H bonds. Their values are in the same range as those found for the breathing of other proteins (Zhang et al. 1997, 2001; Resing and Ahn 1998; Engen et al. 1999; Wang et al. 2001). The last column concerns NHs with exchange that is too slow to be measured within the time range of our experiments.

Figure 3.

Plots of deuterium incorporation into representative fragments produced by pepsin digestion of CK as a function of incubation time after adjustment for deuterium gain/loss during digestion and analysis. The number of backbone exchangeable amide hydrogens is given in parentheses. The data are fitted to equation 3 with two or three exponential terms. The slowest exchanging protons are not detectable within the time scale of the experiments. Segments displaying similar exchange kinetics are grouped in panels A–D.

Table 1.

Exchange rate constants for 14 probes of native rabbit muscle MM-CK at 20°C and pH 7

			Medium		Slow
Peptide fragments	Na	Very fastb nd	nd	k (min−1)	nd	k (min−1)	Very slowc nd
1–22	20	5	3	2.7	11	0.07	1
162–186	23	8	6	1.3	6	0.05	3
362–380	17	6	5	0.4	2	0.02	4
51–72	19	10			3	0.02	6
193–201	6	4	2	0.8			0
312–330	18	13	1	1.5			4
23–37	13	2	3	0.3			8
38–50	11		2	4.5	2	0.03	7
90–126	34	8	11	0.2			15
202–225	22	4			6	0.05	12
235–249	14	2	4	0.1			8
262–280	17	2	4	0.2			11
287–311	23	5	4	0.3			14
334–348	14	4			4	0.03	6

^a The total number of exchangeable amide hydrogens in the peptide.

^b Very fast refers to hydrogens exchanging with a non accurately determinable rate constant >10 min⁻¹.

^c Number of hydrogens which are very slowly exchanged (k < 0.02 min⁻¹).

^d The fitted number of protons exchanged with the rate constant k.

Figure 3A ▶ shows the exchange-in pattern of three peptides that are nearly completely deuterated and that display similar exchange kinetics. They can be satisfactorily fitted to equation 3 with three exponential terms.

Among the 20 exchangeable NHs of the 1–22 peptide, five are rapidly exchanged, three have exchange rate constants of ~2.7 min⁻¹, and 11 have rate constants of ~0.07 min⁻¹.

This N-terminal peptide is almost devoid of secondary structure because it contains only the short α1-helix (residues 15–18). It thus can be extensively exchanged. Furthermore, because the crystallographic position of the six first amino acids cannot be resolved, the NHs of this very mobile segment may correspond to the five that are very rapidly exchanged.

However, according to Webb and Morris (2001), 11 residues (7–10, 13, 16–21) belong to a monomer–monomer interface residue segment (IRS 1). The interaction between the two monomers may decrease the rate of exchange because of a reduction of the accessibility to the solvent of part of this probe. The NHs of this IRS could be among the most slowly exchangeable ones of the peptide. Interestingly, this peptide contains the epitope for the monoclonal antibody CK-STAR, which reacts only with mildly denatured CK and not with native CK (Morris and Man 1992).

The deuteration of peptide 162–186 follows a similar pattern. This peptide is totally deuterated for the longest incubation times. Among the 23 exchangeable NHs, eight are rapidly exchanged, six are deuterated with a rate constant of 1.3 min⁻¹, and six are deuterated with the rate constant of 0.05 min⁻¹. The exchange of the last three NHs is too slow to be observed in our experimental conditions. Part of this peptide belongs to a conserved surface segment 176–204 (Shen et al. 2001) that includes the α8-helix (180–187). The fragment also contains a 3₁₀-helix (166–168) and the β2 strand (170–174), which is the more external β strand of the saddle lining the active site of the enzyme. All of these elements are external in the three-dimensional (3D) structure and thus accessible to D₂O. The backbone hydrogens of the nonstructured 162–165 and 177–179 segments could be included in the eight most rapidly exchanged protons. According to Webb and Morris (2001), a residue of this peptide, Lys 176, belongs to the monomer–monomer interface IRS 4. Therefore, several vicinal residues could be protected and deuterated more slowly. This shielding may be the reason why proteolytic attack by chymotrypsin and endoproteinase Glu-C at Leu 175, Glu 180, and Glu 182 residues requires a mildly denatured state (Webb et al. 1997).

A third peptide (362–380) follows the same pattern. The six NHs that are rapidly exchanged could be the last of this C-terminal peptide because they belong to a nonstructured and very accessible region. The rest of the peptide (a minor part of the α12-helix and a nonstructured stretch that could interact with the α12-helix) exchanged more slowly: five NHs with a rate constant of 0.4 min⁻¹, two with the rate constant of 0.02 min⁻¹, and four with a rate too slow to be measured.

Figure 3B ▶ shows the deuteration pattern of three peptides that exchange a larger percentage of their NHs during the fast phase. The data were fitted to an equation including two exponential terms. In the 3D structure of CK, these regions are relatively accessible while being in the vicinity of the active site. The 193–201 probe totally exchanged its NHs within 5 min; four NHs are rapidly deuterated, and two are deuterated at a rate constant of 0.8 min⁻¹. This largely unstructured peptide (it contains only two residues out of the three of the α9-helix) belongs to a surface segment (176–204) but included Lys 195, which is a monomer–monomer interface residue (IRS 5). Two other peptides 192–201 and 192–202 behave similarly, and the last one includes Leu 202, which is a part of the IRS 6. This interfacial position may explain why two NHs are less rapidly exchanged than the others and why partial denaturation is required for the binding of monoclonal antibodies CK-JAC and CK-JIL, the epitopes of which partially overlap this peptide (Morris and Cartwright 1990).

The 312–330 peptide contains a mobile loop in which the position of the 322–330 residues has not been resolved by crystallography and in which resides the proteinase K cleavage site (Wyss et al. 1993; Leydier et al. 1997). We have analyzed two overlapping peptides that behave similarly (312–333 and 315–333). The NHs of this loop certainly number among the 13 very rapidly exchanged NHs.

This peptide also includes the end of the α11-helix, the β9 strand, and the beginning of β10, which is the other external strand of the saddle. Among the hydrogens of the amide linkages of these elements, one exchanges with an average rate constant of 1.5 min⁻¹ and four exchange very slowly. These slowly exchanging NHs may be part of α11 and β9, which is an inner strand of the saddle.

The 51–72 peptide contains another mobile surface loop (59–69), which could explain in part the rapid exchange of 10 hydrogens. The presence of the α4-helix (52–61) and IRS 2 (Asp 53, Ile 56, Gln 57, Val 60, Asp 61, Asn 62) may explain the slow exchange of three hydrogens (0.02 min⁻¹) and the very low exchange of six others. This peptide also contains a major chymotrypsin digestion site, Phe 67, in mildly denatured CK (Webb et al. 1997).

Figure 3C ▶ shows the peptides of the N-terminal domain that have a low percentage of deuteration. The kinetics were fitted to a two exponential equation. The 23–37 peptide is composed of a minor nonstructured part, of the α2-helix, and of a small part of α3. Only two NHs exchange with a very rapid rate constant. The others have slower rate constants: 0.3 min⁻¹ for three of them and a low nonmeasurable rate for the last eight. Some of them may be located in secondary structure elements. In this region, two proteolytic sites (Lys 31 and Glu 36) are unveiled in the presence of a low concentration of GdmHCl.

The deuteration of the next peptide (38–50) is also low. In contrast to the 23–37 peptide, it contains a large nonstructured part and only half of the α3-helix. However, it also comprises a part of the interface IRS 2 (Pro 47, Ser 48, and Gly 49). This peptide contributes to monomer–monomer interaction and is therefore expected to be partially protected from hydrogen exchange. Two NHs are exchanged with a rate constant of 4.5 min⁻¹, two with a rate constant of 0.03 min⁻¹, and seven at a very low rate. By comparison with the deuteration data of three other probes of the same region (38–49, 38–53, 38–54), it can be concluded that the less exchangeable NHs of these peptides probably include their C-terminal ends and the IRS residues. A low concentration of GdmHCl (0.625 M) unmasks a chymotrypsin cleavage site (Tyr 38) in this peptide (Webb et al. 1997).

The third peptide (90–126) comprises the second half of the α6-helix, a 3₁₀-helix, and a surface area (114–121). Eight NHs are rapidly exchanged, 11 are exchanged more slowly (k = 0.2 min⁻¹), and 15 are exchanged only very slowly. The hydrogen exchange behavior of this peptide is unexpected because it constitutes a surface segment (Fig. 4 ▶). Yet, the 90–114 segment has similar deuteration kinetics, and the 115–126 probe is also slowly exchanged (not shown).

Figure 4.

Three-dimensional structure of dimeric rabbit muscle creatine kinase indicating the locations and H/D exchange of peptic peptides. The peptides of Figure 3 ▶, A–D, are represented in red, yellow, green, and dark blue, respectively. The peptides exhibiting minimal exchange are presented in light blue, and the regions not identified are in cyan. Coordinates (Protein Data Bank code: 2crk) were displayed with the visualization program Protein Explorer.

Figure 3D ▶ shows the deuterium exchange pattern of five peptides belonging to the C-terminal domain and having a low degree of deuteration. Except for the 262–280 peptide, they are located in the hydrophobic core of the protein and participate in the formation of the saddle-shaped β-sheet.

The 202–225 segment includes the β3 strand and the beginning of the β4 one. Four hydrogen amides are rapidly exchanged, possibly into the turn between β3 and β4. Six hydrogens exchanged with a rate constant of 0.05 min⁻¹, and 12 are exchanged at a very slow rate. This area contains the IRS 6 segment, which comprises Leu 202, Ala 207, Arg 208, Asp 209, Trp 210, and Asp 212 residues. Mild denaturation of CK exposes two trypsin digestion sites in this area (Arg 208 and 214). The presence of this large IRS and of the β strands may explain the low level of deuteration in this peptide. It is noteworthy that Trp 210 and 217 are located in a hydrophobic environment and that the mutation of Trp 210 facilitates the dissociation of the dimer (Gross et al. 1994; Perraut et al. 1998).

Peptide 235–249 includes the β5 strand, which lays in the middle of the saddle, and the N-terminal end of the α10-helix. Two NHs are rapidly exchanged, four are more slowly exchanged (k = 0.1 min⁻¹), and the exchange of the last eight hydrogens is not measurable in our conditions. It should be noted that the 235–239 peptide, which is H-bonded to both β4 and β1, is not deuterated (data not shown). Its four NHs are certainly included in the eight nonexchangeable hydrogens.

Peptide 262–280 comprises the end of the α10-helix, β6, and β7, which form a small β-sheet distinct from the main eight-stranded β-sheet. Two hydrogens exchange rapidly, four NHs exchange at 0.2 min⁻¹, and 11 have a very low exchange rate. Two sites (Glu 261 and Glu 274) susceptible to cleavage by endoprotease Glu-C following mild denaturation are located in this area (Webb et al. 1997).

Peptide 287–311 includes the β8 strand of the saddle and the first half of the α11-helix. Five NHs are rapidly exchanged. They may be located in the 298–306 unstructured stretch, which is accessible in the 3D structure. Four additional hydrogens have a slower rate constant (k = 0.3 min⁻¹), and 14 others exchanged very slowly.

The last peptide (334–348) also contains a large part of β10, which is a saddle strand, and the beginning of α12. The four of its NHs that exchanged rapidly may be located in the 338–344 loop. Four additional hydrogens exchanged with a rate constant of 0.03 min⁻¹, and six exchanged at a very slow rate.

In addition, we have found some peptides that undergo very little exchange (data not shown). All except the 225–234 peptide, which also contains the turn between β4 and β5, are included in secondary structure elements. This peptide contains the β4 strand in which Trp 227 is located. This residue, which is crucial for catalytic activity, is poorly accessible to the solvent and is the strongest contributor to intrinsic fluorescence (Gross et al. 1994; S. Darmochod, O. Marcillat, and C. Vial, unpubl.).

The 250–261 and 349–358 peptides form the central part of the two largest helices, α10 and α12. As already mentioned, the 235–239 segment forming the β5 strand located in the middle of the saddle was not deuterated. Finally, the large peptide 127–158 contains the β1 strand and the major part of the α7-helix. Because β1 is located in the middle of the saddle between β5 and β8, and because the turn between β1 and α7 and the N-terminal part of α7 constitute an interface segment (IRS 3), this may explain the low deuteration level of this fragment.

The peptide 81–84, which constitutes the small α5-helix, was very slowly deuterated to a final level of 50%, whereas the following segment, 85–89, which comprises the N-terminal end of the large α6-helix, was not deuterated at all.

Discussion

The kinetics of localized deuteration of native CK were investigated by labeling the protein for various times ranging from 3 sec to 90 min, decreasing the pH and the temperature to quench isotope exchange, fragmenting with pepsin, and analyzing the resulting mixture by LC-MS. The locations of the segments of the native MM-CK having different patterns of deuteration with respect to the primary and secondary structure are illustrated in Figure 1 ▶, and their positions in the tertiary structure of the CK dimer are shown in Figure 4 ▶.

Figure 2 ▶ shows two ESIMS isotope patterns representative of all identified peptides. Both have the form of a single binomial distribution. This pattern is normal for hydrogen exchange within the EX2 kinetic limit leading to a random distribution of deuterium. Protein NH exchange under native conditions almost always proceeds via this mechanism (Perrett et al. 1995; Kaltashov and Eyles 2002). A few exceptions have however been reported: For instance, nine of 64 peptic peptides of aldolase and two of six of the kinase SH3 domain have been found to exhibit EX1 kinetics (Zhang et al. 1996; Engen et al. 1997).

Two types of regions are expected to be significantly deuterated: nonstructured and solvent-accessible regions of the protein, and those that experience localized unfolding during the deuteration pulse. Conversely, folded and/or inaccessible regions are expected to be little or not exchanged. Indeed, in a study of the conformational dynamics of a tetrameric aldolase, a high level of correlation between the slowing of hydrogen exchange and intramolecular hydrogen bonding was found. One exception occurs at the subunit interface, where the NHs in one nonstructured sensor peptide have slower exchange rates than expected, indicating that they were effectively shielded from the solvent (Zhang et al. 1996).

Accordingly, several patterns of increase of the deuteration levels as a function of time can be recognized in the peptic peptides of MM-CK (Figs. 2 ▶, 3 ▶). For some, represented by those in Figures 2A ▶ and 3 ▶, C and D, the deuterium level was initially low but increased with time. Indeed, as indicated in the results section, most of these peptides are localized in structured and/or nonaccessible areas. This is also the case for other peptides not presented in Figure 3 ▶ because of their very low levels of deuteration. These peptides include the strands constituting the hydrophobic core of the protein (127–158, 202–225, 225–234, 235–239, 287–311, 334–348), the α-helical peptides (81–84, 85–89, 250–261, 349–358), and the peptides located at monomer–monomer interfaces (38–50, part of 127–158, and part of 202–225). Among the peptides with a low deuteration, the 90–126 segment is an exception. Indeed, although it has a low content of secondary structure, and despite its surface localization, it is slowly exchanged. This segment is a linker that connects the N- and C-terminal structural domains (Rao et al. 1998). When the 3D structure is considered, this segment runs at the surface of the protein between the C1 and C2 subdomains defined by Webb and Morris (2001) as two hydrophobic folding units. It contains a relatively high number of hydrogen bonds, which explains the observed protection against hydrogen exchange. In the first part of the peptide (90–114), these hydrogen bonds are mainly intrasegment bonds, but the second part (115–126) also H-bonds to both subdomains C1 and C2. This would indicate that upon folding of the two subdomains, this peptide fastens them together and locks them in their final position. Furthermore, there are no protease susceptible regions in this segment of the mildly denatured enzyme in spite of the presence of several potential digestion sites (Webb and Morris 2001). Although the 90–126 peptide is solvent-exposed and although its extremities belong to the two dynamic domains 1 and 3, which move upon substrates binding (Yousef et al. 2003), it should be noted that the deuteration of this peptide is not altered by the formation of the transition state analog complex (Mazon et al. 2003). These observations confirm that the number of hydrogen bonds is high enough to prevent its deuteration and indicate a prominent role for the 90–126 region in the cohesion of protein domains. In all these segments, a low percentage of fast deuterium exchange is observed together with a high percentage of nonexchanging or very slowly exchanging NHs. In these areas of the protein, the degree of localized unfolding, which could lead to a medium or small rate of exchange, is relatively low. Thus, these regions seem to have very slow dynamics. Such a low degree of deuterium incorporation was found for peptic peptides derived from cytochrome c (Maier et al. 1997). In one case, it was attributed to resistance to exchange-in due to strong hydrogen bonding in the C-helical domain, in the other one to shielding effects of the heme group.

In contrast, for the peptides represented in Figures 2B ▶ and 3 ▶, A and B, a high final level of deuteration was found. These peptides include unstructured and/or solvent accessible regions. Indeed, the three peptides presented in Figure 3B ▶ (51–72, 193–201, 312–330) contain a high fraction of surface loops that are rapidly exchanged during the first phase of deuteration. The second phase leads to the slower deuterium incorporation into more structured parts of these segments due to localized unfolding (“breathing”) of the protein (Wagner and Wüthrich 1979). It should be noticed that the fragments 51–72 and 312–330 contain the two mobile loops that close the active site upon the binding of the two substrates (Rao et al. 1998; Lahiri et al. 2002).

Three other peptides, for which deuteration kinetics are illustrated in Figure 3A ▶, are relatively accessible to the solvent. The pattern of deuteration of the 1–22 peptide may be explained by the high dynamics of its very N-terminal moiety, which could progressively induce a total exchange in the remaining part of the peptide despite its inclusion in a monomer–monomer interface with a restricted solvent accessibility.

The deuteration pattern of the C-terminal end (362–380) is similar to that of the N-terminal end. As mentioned in the results section, the most rapidly exchanged hydrogens are likely to be the last nonstructured residues of the sequence, whereas the others are exchanged more slowly because of their interactions with the α12-helix.

The third peptide (162–186) having a similar pattern is quite accessible but more structured than the two previous ones, which may explain its progressive deuteration. Furthermore, a low concentration of guanidinium chloride is sufficient to expose digestion sites in this segment (Webb and Morris 2001).

Peptide 127–158 also contains several cleavage sites that become exposed only after mild denaturation. However, in contrast to the 162–186 peptide, it contains a large interface segment between monomers, which is likely to restrict the dynamics of this area and to prevent D₂O access into the native protein, as observed at the aldolase interface (Zhang et al. 1996).

As seen in the 3D picture of the enzyme (Fig. 4 ▶), the less-deuterated peptides (dark blue and light blue) are located in the saddle-shaped core of the CK. Most of the easily exchanged peptides (red and yellow) are located around the entrance to the active site and at the N- and C-terminal ends. It can be noted that the 90–126 (green) and 162–186 (red) peptides are both solvent-exposed but do not have the same deuteration behavior. Surprisingly, the 90–126 peptide, which has less secondary structure than the other, is more protected against hydrogen exchange. As previously stated, the 90–126 peptide could be stabilized by interactions with other parts of the protein. This is not the case for the 162–186 peptide. Many of the peptides presented in Figure 3 ▶, A through C, belong to the dynamic domains and hinge regions identified by Yousef et al. (2003) in arginine kinase. The three dynamic domains move relative to a fixed domain and to each other as a consequence of a nucleotide-induced conformational change, leading to the closing of the active site. In contrast, the “dark blue” and “light blue” peptides are more frequently located in the fixed domain.

Thus, although native CK is highly resistant to most proteolytic enzymes, several areas of the protein are sufficiently mobile to have access to the solvent or to exchange their amide protons with deuterons through localized unfolding. However, these motions, because they are either too small or too fast, do not allow docking with the proteases active site and subsequent cleavage. On the other hand, we identified a protein surface–localized, secondary structure–free but highly protected peptide, which is thought to play an active role in the cohesion of domains.

Materials and methods

Protein preparation

CK from rabbit muscle (MM-CK) was purchased from Roche. The enzyme was desalted by using a PD10 Sephadex G25 column (Pharmacia) equilibrated in 50 mM Tris-HCl buffer (pH 7). The MM-CK concentration was estimated by using a molar extinction coefficient of 76,000 M⁻¹ cm⁻¹ at 280 nm.

Materials

D₂O (99.9%) and chloroacetic acid (99%) were purchased from Aldrich; trifluoroacetic acid (TFA), dithiothreitol (DTT), and pepsin were from Sigma. Tris was purchased from Interchim; acetonitrile, from SDS.

Hydrogen exchange

MM-CK (80 μM) in 50 mM Tris-HCl/H₂O buffer (pH 7) was diluted (1 : 1, v/v) into 50 mM Tris-HCl/H₂O buffer (pH 7) containing 5 mM DTT as a reducing agent to reduce or prevent the formation of adventitious disulfide bond. Deuterium exchange was initiated by diluting the solution 20-fold with 50 mM Tris-DCl/D₂O buffer (pD 7) at 20°C. All pD measurements are given as read from the pH meter with no adjustment for isotope effects (Connelly et al. 1993). Isotope exchange was quenched after different times by adding 0.2 M chloroacetic acid (H₂O at pH 2) to decrease the pH to 2.5 and by decreasing the temperature to −3°C in a ice/H₂O/ammonium sulfate mixture.

Pepsin digestion

The pepsin digestion allowed determination of the extent of deuterium incorporation into MM-CK segments during the incubation time. The sample containing ~1.5 μM MM-CK was immediately proteolyzed by pepsin (in 10 mM NaH₂PO₄-AcOH/H₂O buffer at pH 2) for 3.5 min at 0°C (CK/pepsin ratio of 1 : 1, w/w) and analyzed by directly coupled LC-MS.

The resulting peptides (186 pmoles of MM-CK) were desalted and concentrated (Peptide Trap C8, 3 × 8 mm, Michrom Bioresources) and separated on a C18 microbore column (50 × 1 mm, Interchim). The HPLC column and the entire injector assembly were packed in ice to minimize H/D exchange at peptide amide linkages during analysis. A 5-min desalting step with 5% mobile phase B (90% acetonitrile/0.03% TFA/H₂O) in phase A (0.03% TFA/H₂O) was used before connecting the HPLC to the mass spectrometer (flow rate, 300 μL/min). Peptides were then eluted at 50 μL/min directly into the mass spectrometer within 21 min by using a 5% to 50% gradient of mobile phase B. The column was washed with 100% phase B for 5 min after the peptides were eluted. Peptic cleavage sites are numerous and difficult to predict on the basis of the sequence alone but are reproducible under identical digestion conditions. Therefore, all peptides were identified by using MS/MS (Biemann 1992). Data were processed by centroiding an isotopic distribution corresponding to the +1, +2, or +3 charge state of each peptide.

Back-exchange controls

The undeuterated and totally deuterated controls were analyzed to adjust for deuterium back-exchange during analysis (Zhang and Smith 1993). The 0% reference sample was prepared by mixing a stock solution of MM-CK with buffer containing no deuterium. The 100% deuterated reference sample was prepared by incubating the protein in 50 mM Tris-DCl/D₂O (pH 2.5) for 3 h at 35°C. These controls were also quenched and digested with the same D₂O/H₂O ratio as the samples to correct for gain or loss of deuterium.

Correction was done according to equation 2, where D is the deuterium content of the peptide; m, m_0%, and m_100% are the average molecular weights of the same peptide in the sample, the undeuterated form, and the totally deuterated form, respectively; and N is the number of exchangeable NHs in the peptide (number of peptide bonds not involving Pro residues):

(2)

The HPLC step was performed with protiated solvents, thereby removing deuterium from side-chains and N/C termini that exchange much faster than amide linkages (Bai et al. 1993). Therefore, an increase in mass is a direct measure of deuteration at peptide amide linkages.

Data analysis

Deuterium levels were plotted versus the exchange time and fitted to a sum of first-order rate expressions according to equation 3:

(3)

Where a_i is the number of deuterium exchanged with a similar rate constant k_i, and N is the sum of a_i (number of peptide amide linkages in which hydrogen exchange rate is measurable). Data were fitted to an expression containing the highest number of exponential terms giving the best fit of the data. The a_i values were rounded to the nearest integer.

Mass spectrometry

Online LC-MS and LC-MS/MS were performed on a quadripole ion trap mass spectrometer (esquire 3000+, Bruker Daltonics) operating under the following conditions: capillary voltage, 4 keV; nebulizer, 10 psi; dry gas, 8 L/min; and dry temperature, 250°C. By using electrospray ionization in positive ion mode, mass spectra were acquired from 200 to 2000 m/z, averaging 25 spectra, and the ICC target was 50,000. For the MS/MS analysis, mass spectra were acquired from 50 to 2000 m/z, averaging 15 spectra, and the ICC target was 30,000. The threshold was 50,000, the amplitude fragmentation was 2 V between 50% and 150%, and the number of precursor ions was three.

Electronic supplemental material

It is a complement to Table 1, which shows the distribution of peptide NH exchange rate constants for the 47 identified peptides in the native rabbit muscle MM-CK at 20°C and pH 7. The data were fitted to a one, a two, or a three exponential equation as permitted by the data.

Abbreviations

• H/D, hydrogen/deuterium

• DTT, dithiothreitol

• LC/MS, liquid chromatography/mass spectrometry

• MM-CK, cytosolic dimeric creatine kinase MM

• MS/MS, tandem mass spectrometry

• NH, amide hydrogen

• TFA, trifluoroacetic acid