Conformational transitions of the three recombinant domains of human serum albumin depending on pH.

Human serum albumin (HSA) is a protein of 66.5 kDa that is composed of three homologous domains, each of which displays specific structural and functional characteristics. HSA is known to undergo different pH-dependent structural transitions, the N-F and F-E transitions in the acid pH region and the N-B transition at slightly alkaline pH. In order to elucidate the structural behavior of the recombinant HSA domains as stand-alone proteins and to investigate the molecular and structural origins of the pH-induced conformational changes of the intact molecule, we have employed fluorescence and circular dichroic methods. Here we provide evidence that the loosening of the HSA structure in the N-F transition takes place primarily in HSA-DOM III and that HSA-DOM I undergoes a structural rearrangement with only minor changes in secondary structure, whereas HSA-DOM II transforms to a molten globule-like state as the pH is reduced. In the pH region of the N-B transition of HSA, HSA-DOM I and HSA-DOM II experience a tertiary structural isomerization, whereas with HSA-DOM III no alterations in tertiary structure are observed, as judged from near-UV CD and fluorescence measurements.

Human serum albumin (HSA) is a protein of 66.5 kDa that is composed of three homologous domains, each of which displays specific structural and functional characteristics. HSA is known to undergo different pH-dependent structural transitions, the N-F and F-E transitions in the acid pH region and the N-B transition at slightly alkaline pH. In order to elucidate the structural behavior of the recombinant HSA domains as standalone proteins and to investigate the molecular and structural origins of the pH-induced conformational changes of the intact molecule, we have employed fluorescence and circular dichroic methods. Here we provide evidence that the loosening of the HSA structure in the N-F transition takes place primarily in HSA-DOM III and that HSA-DOM I undergoes a structural rearrangement with only minor changes in secondary structure, whereas HSA-DOM II transforms to a molten globulelike state as the pH is reduced. In the pH region of the N-B transition of HSA, HSA-DOM I and HSA-DOM II experience a tertiary structural isomerization, whereas with HSA-DOM III no alterations in tertiary structure are observed, as judged from near-UV CD and fluorescence measurements.
The albumin molecule is composed of three homologous, predominantly helical evolutionarily related domains. Each domain is made up of two subdomains, which share a common helical motif. Specific characteristics, for example the location of the various ligand-binding sites, were assigned to some of the domains or subdomains in the past (1,2). We recently described the production of the three domains of HSA 1 as recombinant proteins, and we were able to show that they have structural features as well as ligand binding properties correlating to their specific functionality in the context of the native protein (3). Clear evidence could be provided that the primary binding site for hemin is located on HSA-DOM I and that for diazepam on HSA-DOM III.
Human serum albumin undergoes several transitions in dependence of pH, the N-F transition between pH 5.0 and 3.5, the F-E transition or acid expansion below pH 3.5, and the N-B transition between pH 7.0 and 9.0 (1,2). The acid-induced structural changes of bovine and human serum albumin have been studied by a wide range of methods (1,2) and are characterized by changes in secondary as well as tertiary structure. An indication for the former is a significant decrease in helical content observed by far-UV CD measurements which was interpreted by Era et al. (4,5) to represent a helix 3 ␤ and a helix 3 coil transition in the case of the N-F transition. As the pH of an albumin solution is reduced below 3.5, further unfolding occurs until about pH 2.5, when the molecule appears to be expanded to the full extent that is allowed by its disulfide bonding structure. Under slightly alkaline conditions, between pH 7.0 and 9.0, HSA and BSA undergo another conformational change, known as the N-B transition. It is more subtle and more gradual in onset and has been proposed to have physiological importance, similar to the acid-induced transformations (1,2). The B isomerization is supposed to be a structural fluctuation, a loosening of the molecule with loss of rigidity, particularly affecting the N-terminal region and thereby impacting ligand binding (6 -9).
One way to explore the molecular origins of these transitions is by using albumin fragments. Attempts have been made in the past to assign specific roles and contributions of the albumin domains to the characteristic structural transitions of both HSA and, more frequently, its closely related bovine homolog, BSA. For many of these studies, fragments have been used, which had been produced by enzymatic or chemical cleavage (8 -17).
An F-like transformation was observed with a tryptic fragment of bovine serum albumin (BSA-T 377-582 ), shedding some light on the nucleation points for the intramolecular changes that occur during this transition (13). It was shown that this fragment unfolds and abruptly loses structure at pH 4, whereas fragment BSA-P 1-385 does not do so until pH 3.5 (12), as determined by difference spectroscopic methods. Domain III of BSA was consequently proposed to have a less constrained conformation than the rest of the molecule and to expand through separation of its subunits in the course of the N-F transition. Bos et al. (17) concluded from their work with two large fragments of HSA that the conformational change between pH 6.0 and 9.0 of the P-46 fragment (residues 1-387 of HSA) clearly resembles the N-B transition of albumin, whereas the T-45 fragment (residues 198 -585 of HSA) does not display such a conformational change. However, the choice of fragments used in these studies was dependent on the availability of suitable cleavage sites and did not necessarily take into account the natural borders of the domains as they can be defined based on amino acid sequence and atomic structure.
Based on our previous results on the structure and ligand binding properties of the three recombinant HSA domains, we report here in a systematic way on the conformational behavior of these new proteins in the acidic and slightly alkaline pH region and on their contributions to the origins of the structural transitions of the native molecule. For this purpose, fluorescence and CD spectroscopic measurements were performed depending on pH, and the results are discussed in the context of the atomic coordinates of HSA.

EXPERIMENTAL PROCEDURES
Materials-HSA (essentially free of fatty acids and globulin) was from Sigma (catalog number A3782). This material was further purified by size exclusion chromatography over Superdex 200 prep grad (Amersham Pharmacia Biotech) and delipidated by passing over a Lipidx-1000 column (Canberra-Packard).
The three domains of HSA were cloned, expressed, and purified as has been described before in detail (3). The domains encompassed the following amino acid residues of HSA: HSA-DOM I, 1-197; HSA-DOM II, 189 -385; HSA-DOM III, 381-585. Due to the restriction site used for cloning, all domains had Glu-Phe as the N-terminal amino acids. All other reagents were of analytical grade.
Protein Solutions-Protein stock solutions were prepared by dissolving the delipidated proteins in aqua bidest. Protein concentrations were measured by their absorbance at 278 nm on a Hewlett-Packard HP8453 spectrophotometer. For HSA, an absorption coefficient (A 278 nm,1 cm 0.1% ) of 0.58 was used (8). For HSA-DOM I, HSA-DOM II, and HSA-DOM III, the absorption coefficients were found to be 0.50, 0.79, and 0.30, respectively (3).
pH-dependent CD Measurements-Measurements were made using a Jasco J-600 spectropolarimeter using a 1-mm cell at 25°C in a thermostated cell holder at a concentration of 4.5 M in the far-UV region. Scans were made from 250 to 190 nm. The slit width was programmed for a half-bandwidth of 1 nm, and the dynode voltage never exceeded 0.6 kV. 10 l of a 100 times concentrated protein stock solution (in aqua bidest.) were mixed with 990 l of the corresponding buffer (pH 9.0 to 8.5, 10 mM boric acid/Borax; pH 8.0 to 5.0, 10 mM sodium phosphate; pH 4.5 to 3.5, 10 mM acetic acid/sodium acetate; pH 3.0 and less was adjusted with HCl). All spectra were recorded at least twice. After each measurement, the pH was confirmed with a standard pH meter. The data were expressed as mean residue ellipticity ([] MRW , degree cm 2 dmol Ϫ1 ), using the mean residue weights of 114.0 g mol Ϫ1 for the intact molecule, 114.9 g mol Ϫ1 for HSA-DOM I, 113.2 g mol Ϫ1 for HSA-DOM II, and 113.0 g mol Ϫ1 for HSA-DOM III.
The fractional content of the secondary structure elements of the proteins was calculated from the far-UV CD spectra using the procedure of Provencher and co-workers (18,19) (CONTIN) with a set of 16 reference proteins.
In the near-UV region, measurements were made using a 1-cm cell at 25°C in a thermostated cell-holder at a protein concentration of 20 M. Scans were made from 340 to 250 nm. The slit width was programmed for a half-bandwidth of 1 nm, and the dynode voltage never exceeded 0.4 kV. As above, a 100 times concentrated protein stock solution was diluted with the desired buffer (pH 9.0, 0.1 M boric acid/Borax; pH 7.4, 0.1 M sodium phosphate; pH 4.0, 0.1 M acetic acid/sodium acetate; pH 2.0 was adjusted with HCl) to reach the final protein concentration of 20 M. The results are expressed as molar ellipticity ([], degree cm 2 dmol Ϫ1 ).
pH-dependent Fluorescence Spectroscopy-Fluorescence measurements were made with a Hitachi F-4500 fluorescence spectrometer equipped with a thermostated cell holder at 25°C. All spectra were recorded in a 0.5 ϫ 1-cm stirred cell (0.5 cm at emission and 1 cm at excitation side) with the excitation and the emission slit width set to 5 nm. For HSA and HSA-DOM II, which contain only a single tryptophanyl residue (Trp-214), an excitation wavelength of 295 nm was routinely used, to ensure that the light was absorbed almost entirely by the lone tryptophanyl residue. HSA-DOM I and HSA-DOM III were excited at 280 nm to record the tyrosyl emission. pH-dependent experiments were carried out as described above with a protein concentration of 4.5 M, and all spectra were recorded at least twice. Each spectrum was corrected for buffer base-line fluorescence. After each measurement the pH was controlled with a standard pH meter.

RESULTS
Far-UV CD-In order to get information about the secondary structure of the proteins used in this study, far-UV CD measurements were performed between 250 and 190 nm depending on the pH. A spectrum for each of the proteins in the native (pH 7.4) and denatured state (9.5 M urea) is shown in Fig. 1A, demonstrating full disruption of the secondary structure of all proteins at 9.5 M urea.  Table I for pH 9.0, 7.4, 4.0 and 2.5.
In the acid pH region from pH 5.0 to 3.5 and further down to pH 2.5, the ellipticity observed with HSA shows a two-step sigmoidal increase, and calculated helix content (f ␣ ) decreases, whereas sheet structures (f ␤ ) increase. These changes in secondary structure are in accordance with the concept of the N-F transition followed by the acid expansion (1,2,4,5). A similar observation is made with HSA-DOM III, which also displays a characteristic two-step transition correlated with a decrease in ␣-helical content and a gain in ␤-sheet structure. However, the two transitions are clearly split. The onset of the first sigmoidal transition is shifted to higher pH as compared with HSA, followed by a second change in the pH region between 3.0 and 2.0. These findings clearly prove that the N-F transition originates in a structural loosening at the C-terminal end of the albumin molecule, probably followed by a separation of domains and subdomains in the course of the acid expansion, as has been suggested before by several authors (1,2,13,15,16). For HSA-DOM I, a slight increase in [] 222 is observed in the pH region from 5.5 to 3.5, which indicates only a minor change in secondary structure, whereas below pH 3.5 this domain shows an acid expansion similar to HSA and HSA-DOM III, which is further evidence for the disruption of intradomain structure of HSA in this pH region. By diminishing the pH of a HSA-DOM II solution from 6.0 to 4.0, unexpectedly a sigmoidal decrease in [] 222 , is observed, which is correlated with a slight increase in ␣-helix content accompanied by a reduction of ␤-sheet structure. HSA-DOM II shows, in contrast to HSA-DOM I and HSA-DOM III, only a slight increment in ellipticity in the acid expansion region of HSA. Below pH 2.0, the ellipticity of all four proteins investigated is decreased. This can be interpreted as a gain in secondary structure, as was observed previously for many proteins in the absence of salt by Goto et al. (20) and Fink et al. (21).
HSA shows a small increase in the ellipticity at 222 nm in the pH range from 7.4 to 9.0, indicating a slight reduction of ␣-helical content and an increase in ␤-structural elements, in accordance with the concept of the well known N-B transition in the absence of added salt (2,22). Neither of the recombinant domains exhibits significant fluctuation in ellipticity between pH 6.0 and 9.0. In this pH region, only minor changes in secondary structure can therefore be expected, which is further supported by the data from the secondary structure resolved analysis.
Tryptophanyl Fluorescence-HSA has a single tryptophanyl residue, Trp-214, located in domain II. To examine the conformational variations around this residue in HSA and HSA-DOM II, fluorescence was excited at 295 nm, which provides no excitation of tyrosine residues and therefore neither emission nor energy transfer to the lone indole side chain.
The pH profiles of the fluorescence intensity and the wavelength at the respective emission maximum are displayed in Fig. 2A for HSA and HSA-DOM II. The fluorescence of HSA in the acid pH region exhibited a two-step decrease, the first from pH 6.0 to 4.5 and the second, more pronounced, from pH 3.5 to 2.5, representing the N-F and F-E transitions (16,23,24). The wavelength maximum of the tryptophanyl fluorescence shifted from 340 nm at the high pH end of the N-F transition (pH 5.0) to approximately 332 nm at pH 3.5. By raising the pH of an HSA solution from 7.4 to 9.0, we observed a slight decrease of max and fluorescence intensity ( Fig. 2A), which was reported previously for BSA in the absence of salt (22). At pH 7.4, the fluorescence quantum yield of the tryptophanyl residue in HSA-DOM II is 2.6-fold lower compared with Trp-214 in the native molecule, and the emission maximum is shifted from 340 (HSA) to 332 nm (HSA-DOM II). As the pH is reduced, HSA-DOM II shows a sigmoidal rise in fluorescence intensity, reaching a maximum increase of 2.4-fold at pH 4.0. At the onset of this intensity enhancement no change of the emission maximum is observable, whereas from pH 5.5 to pH 4.0 a slight decrease of 1 nm can be detected. The pronounced rise in fluorescence intensity of HSA-DOM II between pH 7.4 and 4.0 coincides with the secondary structural rearrangement  observed by far-UV CD in the same pH region. At pH 4.0, the tryptophanyl residues of both HSA-DOM II and HSA have superimposable fluorescence spectra (data not shown), suggesting that at this pH the tryptophanyl residue in both proteins has a similar environment.
By further reduction of the pH, the fluorescence intensity and emission maximum of HSA-DOM II and HSA run nearly parallel, but the diminution in fluorescence intensity below pH 3.0 followed by the increase below pH 2.0 is not as pronounced in HSA-DOM II as it is in HSA. Even at pH 2.0, the tryptophanyl residues are not in a solvent-exposed environment, as indicated by denaturing both proteins with 9.5 M urea, which leads to a red shift of the fluorescence maximum (349 nm for both proteins) and a reduction of the fluorescence intensity (data not shown). In the slightly alkaline pH region HSA-DOM II shows a gain in tryptophanyl fluorescence intensity, which is accompanied by an increase of the emission maximum.
Tyrosyl Fluorescence-Since HSA harbors only a single tryptophanyl residue, which is located in domain II, we investigated the tyrosyl fluorescence of HSA-DOM I and HSA-DOM III in order to obtain additional insight into the structural characteristics of these two proteins. The pH profiles of the fluorescence intensity and the wavelength at the emission maximum are visualized in Fig. 2B.
In the acidic pH region, HSA-DOM I shows a two-step alteration in fluorescence intensity. At first, by lowering the pH to 3.5, the fluorescence increases, which is aligned with a slight, but significant, decrease in the emission maximum. By further reducing the pH, the fluorescence intensity rises to a maximum value at pH 2.0; however, no change in the position of the emission maximum is observed. The spectra of HSA-DOM I in 9.5 M urea and at pH 2.0 (data not shown) are nearly superimposable, indicating that the seven tyrosyl residues in this protein have the same environment at pH 2.0 as in the fully denaturated state at 9.5 M urea. In the neutral and slightly alkaline pH range the tyrosyl fluorescence and the position of the emission maximum of HSA-DOM I show no significant alterations, which indicates no significant changes in the microenvironments of the seven tyrosyl residues.
As the pH of the HSA-DOM III solution is reduced to pH 4.0, there is a sharp sigmoidal decrease in the tyrosyl fluorescence intensity and a slight increase in the emission maximum, however, by further lessening of the pH of the solution there is a smooth and continuous increase in fluorescence intensity. The tyrosyl fluorescence intensity of HSA-DOM III in the neutral to slightly alkaline pH region decreases, whereas the emission maximum increases slightly but reproducibly from ϳ304 nm at pH 6.0 to ϳ305 nm at pH 9.0. By raising the pH, a shoulder near 350 nm becomes obvious, which indicates ionization of some tyrosyl side chains (data not shown) (25).
Near-UV CD-To study the structural alterations in more detail, CD measurements in the near-UV region, which infer to tertiary structure, were performed at specific pH values, and the results are shown in Fig. 3, A-D, for HSA and each of the three recombinant domains.
The near-UV CD spectra of HSA are shown in Fig. 3A. At pH 7.4, minima at 262 and 268 nm are observed as well as two shoulders near 276 and 283 nm, which is in accordance with previous findings (26 -28). By reducing the pH to 4.0 and further to 2.0 there is an increase in the ellipticity at 262 and 268 nm and a decrease between 290 and 300 nm, denoting loss of tertiary structure in both the N-F transition and the acid expansion, in agreement with the alterations in secondary structure and tryptophanyl fluorescence. These findings agree with results published previously (29,30). However at pH 2.0 there are still significant CD signals left, compared with the spectrum at 9.5 M urea, suggesting remaining tertiary structure even at this low pH. By changing the pH of the HSA solution from 7.4 to 9.0, we observe a slight gain in the CD signal at 262 and 268 nm, which might reflect perturbations around the numerous disulfide bridges (22,30). In addition, a small reduction in ellipticity in the region between 280 and 300 nm is seen, which was ascribed to changes in the tryptophanyl environment (22,30) and which is corroborated here by the changes in tryptophanyl fluorescence that we observed in this pH region. Fig. 3B presents the near-UV CD spectra of HSA-DOM I. Like HSA at pH 7.4, two minima at 262 and 268 nm and a shoulder at 276 nm are seen, and an additional shoulder near 255 nm is evident. Upon reduction of the pH of the solution to 4.0 the spectrum of HSA-DOM I exhibits an increase of [] 262 and [] 268 , a flattening of the shoulder at 255 nm, and a decrease in the region above 280 nm with loss in fine structure. This is a clear indication for a structural rearrangement of HSA-DOM I in the pH region of the N-F transition of HSA, which is further emphasized by the pronounced increase in tyrosyl fluorescence intensity and the shifting emission maximum. By a further decrease of the pH to 2.0, an overall increase in ellipticity compared with the spectrum at pH 4.0 is perceivable, indicating further loss of tertiary structure elements of this domain in the pH region of the F-E transition of HSA. At 9.5 M urea, a dramatic increase in ellipticity was observed. The difference between the spectrum at pH 2.0 and 9.5 M urea denotes that even at pH 2.0 some tertiary structural elements are left. By raising the pH of the HSA-DOM I solution to 9.0, the CD signal at 262 and 268 nm increases slightly and the shoulder near 255 nm flattens, whereas there is a decrease in the region above 285 nm, highlighting the contribution of HSA-DOM I to the N-B transition of HSA.
The near-UV CD spectra of HSA-DOM II at different pH values and at 9.5 M urea are presented in Fig. 3C. Similar to HSA the spectrum at pH 7.4 has minima at 262 and 268 nm, shoulders at 276 and 283, and an additional shoulder at 287 nm. As the pH is reduced to 4.0, there is an overall loss in ellipticity below 300 nm, and additional minima at 282 and 291 nm arise. These alterations in the near-UV CD spectra of HSA-DOM II provide evidence for major fluctuations in the tertiary structure and may reflect the changes in the environment of the tryptophanyl residue, which are documented by the 2.4-fold increase in fluorescence intensity as well as the secondary structural changes observed by far-UV CD in the same pH region. At pH 2.0 the ellipticities at 283 and 295 nm are only slightly increased, whereas the changes at 262 and 268 nm are more pronounced, indicating further loss in asymmetry around disulfide bridges and/or aromatic residues. The spectrum of HSA-DOM II in 9.5 M urea is different in shape compared with that at pH 2.0; however, the signal strength is similar, denoting that only minor tertiary structure elements are left at pH 2.0, in contrast to HSA and the other two domains. By increasing the pH of the solution to 9.0, a slight enhancement of the ellipticity at 262 and 268 nm is evident. The shoulder at 287 nm is transformed to a minimum and an additional shoulder at 295 nm arises. These findings and the changes in the tryptophanyl fluorescence intensity and emission maximum suggest that the tertiary structure is changing in the alkaline pH region, whereas there are only small fluctuations in secondary structure (Fig. 1B).
In Fig. 3D the near-UV CD spectra of HSA-DOM III are shown. Similar to HSA, minima at 262 and 268 nm and shoulders near 276 and and near 283 nm are observed at pH 7.4. By reduction of the pH from 7.4 to 4.0, the ellipticity below 295 nm increases, and a loss in fine structure is detectable in the wavelength region between 295 and 270 nm. As the pH is diminished to 2.0, ellipticities below 310 nm are further enhanced, however less pronounced than at 9.5 M urea, providing evidence that even at pH 2.0 some tertiary structure elements are left. Increasing the pH from 7.4 to 9.0 leads only to modest changes, which indicates only minor perturbations of tertiary structure in the pH region of the N-B transition of HSA. This clearly demonstrates that the major structural rearrangement in the N-B transition of HSA is not likely to happen in domain III of the native albumin molecule. DISCUSSION We described previously the production and characterization of the three recombinant domains of HAS, and we discussed in depth their potential as models for the albumin molecule with an emphasis on studying their basic structural features and ligand binding activities (3). The present study highlights the structural integrity and the behavior under varying pH conditions of these new proteins in comparison with intact albumin. Measurements with HSA were used throughout our experiments as a reference in order to compare all observations made with the three recombinant domains directly to the data obtained with the native protein. In general, good agreement was achieved with published data for HSA.
Far-UV CD-In the pH region of the N-F transition (pH 5.0 to 3.5), a reduction of helical structure occurs in HSA, whereas the content of sheet structure increases ( Fig. 1B and Table I). These findings are in accordance with previously published observations for both HSA and BSA (4, 5, 29 -31), where it has been proposed that the N-F transition might represent helix 3 ␤ and helix 3 coil transitions. In the acid-expansion region (pH 3.5 to 2.5), we found a further marked reduction in the helix content and an increase in sheet structure which is in agreement with Era et al. (4,5) and Sogami et al. (30). However, in contrast to these publications we did not find a significant increase in random coil content, and the changes that we observed were not as pronounced as those found by these authors. Furthermore, the onset of the N-F transition occurred about 0.5 pH units earlier in our experiments, which could be ascribed to the different buffer systems that were used in the pH range of the N-F transition. We used acetic acid/sodium acetate in the pH region of 4.5 to 3.5, whereas Era et al. (4,5) and Sogami et al. (30) adjusted the pH with HCl in 0.1 M NaCl or KCl. Interaction of undissociated acetic acid by hydrogen bonding to carboxylic groups of proteins was demonstrated for BSA and other proteins as revealed by electrophoresis (32)(33)(34) and CD (35,36). Furthermore, Leonard and Foster (37) showed that the pH profile of the N-F transition of BSA in acetic acid/acetate differs from that in chloride. In addition, the earlier onset of the N-F transition of the HSA used in our experiments could be due to differences in the preparation method, which is known to contribute to the structural behavior of albumin (38 -40) and/or to the absence of added salt and bound fatty acids, shifting the midpoint of the N-F transition of BSA to higher pH (41).
A number of interesting observations was made with the HSA domains in the acid pH range. A high degree of similarity in behavior to the native protein was observed with HSA-DOM III, which displayed a two-step transition with a decrease in helical and an increase in sheet content ( Fig. 1B and Table I). However, the two transitions are more distinctly separated from each other than the N-F, and the F-E transitions are in HSA under our conditions; furthermore, the onset of the first structural change occurs earlier and that of the second later than in our control experiments with the native protein. The crystal structure of HSA strongly suggests at least five interdomain salt bridges and six hydrogen bondings for domain III at neutral pH. The fact that HSA-DOM III in its recombinant form is missing these stabilizing interactions may be an explanation for the earlier onset of the first and the delay of the second transition in HSA-DOM III. This notion is further supported by the findings of Sogami et al. (30,42), who reported for BSA that the initial part of the N-F transition was shifted to higher pH and the acid expansion was moved to lower pH by increasing the salt concentration, especially by using strongly binding anions such as ClO 4 Ϫ , which dissociate hydrophilic and strengthen hydrophobic interfaces. Based on these findings we conclude that the major structural changes in the N-F transition of HSA arise from domain III, as has previously been shown for a C-terminal fragment of BSA (13) underlining that HSA-DOM III as a stand-alone protein clearly reflects the structural behavior of the C terminus of the native molecule.
With HSA-DOM I only a minor decrease in helix content is observed in the region of the N-F transition of HSA, whereas a marked structural change occurs in the pH region of the acid expansion of HSA ( Fig. 1B and Table I). This is in good agreement with the concept that the albumin molecule is fully extended in the E-form, resulting from a loss of inter-domain and inter-subdomain contacts and a disruption of the structure in the hinge and link regions (1,2).
HSA-DOM II undergoes a rearrangement of secondary structure in the pH region between 6.5 and 4.0 (Fig. 1B), which surprisingly leads to a slight increase in helical content and a reduction in sheet structure (Table I). As we have previously shown (3) the isoelectric point of HSA-DOM II is 5.4, which is almost the midpoint of this transition. It is conceivable that near the isoelectric point, where ionic repulsion between side chains of the same charge is minimized, hydrophobic interactions result in a rearrangement of secondary structure. This suggests that HSA-DOM II has a very flexible structure, which is not surprising since this domain has lost contacts to the two neighboring domains, leading to two previously unexposed surface patches, one at its N terminus and, more importantly, the second in the central part of HSA-DOM II as judged from inspection of the crystal structure. By further lowering the pH, an acid expansion-like transition is observed, which is, however, not as pronounced as compared with HSA-DOM I, HSA-DOM III, and HSA.
The partial structural restoration as it is observed with all four proteins at strong acidic pH can be explained by a minimization of the hydrophobic surface area, which is promoted by neutralization of positively charged repulsive forces by chloride anions of HCl, as has been reported by Goto et al. (20) and Fink et al. (21) for many proteins.
In the alkaline pH range between pH 7.4 and 9.0, HSA displays a slight reduction in helical content and a small increase in sheet structure, whereas the changes in coil structure are not significant (Fig. 1B and Table I). These findings are in good agreement with the data for BSA reported by Era et al. (22) in the absence of salt. The recombinant domains did not show significant alterations in their secondary structure content in the alkaline pH region (Fig. 1B and Table I). The N-B isomerization is interpreted to be "a structural fluctuation, a loosening of the molecule with higher configurational adaptability" (2). The B-form in the absence of added salt might be a state where mutual movement of subdomains, connected by flexible hinge regions, takes place, accompanied by fluctuations of the structures of the subdomains themselves relative to each other (22). It is therefore possible that the main losses in secondary structure are affecting the two interdomain helices (h10 DOM I -h1 DOM II , h10 DOM II -h1 DOM III ) of HSA and that the secondary structural integrity of a domain per se is not impaired in the N-B transition as supported by our data presented here.
Tryptophanyl Fluorescence-To explore further the unexpected structural changes of HSA-DOM II, tryptophanyl fluorescence measurements were performed, since the lone tryptophanyl residue of HSA is located almost centrally in helix 2 of domain II. The position of the energy maximum ( max ) from the arising emission spectrum depends on the properties of the environment of the tryptophanyl residue (43). The fluorescence intensity depends upon the degree of exposure of the tryptophanyl side chain to the polar, aqueous solvent and upon its proximity to specific quenching groups, such as protonated carboxyl, protonated imidazole, deprotonated ⑀-amino groups, and tyrosinate anions (44,45).
With HSA, we obtained pH profiles of Fl em max (where Fl is fluorescence intensity) in the acid pH region showing a twostep change ( Fig. 2A), one corresponding to the N-F transition and the other to the acid expansion, as reported previously for HSA (16,23,24) and for BSA (46,47). The major changes in Fl em max were complete at pH 4.5, which is approximately at the onset of the N-F transition under conditions where salt is added. It was reported (41) that delipidated BSA and albumin without added salt show a shift of the N-F transition to higher pH regions. The observed shift of the energy maximum to shorter wavelength of HSA in the N-F transition ( Fig. 2A) was attributed to sandwiching of the indole side chain in a rigid portion of the protein matrix (24).
The pH-dependent decrease of the tryptophanyl fluorescence of HSA at the onset of the N-F transition may be attributed to quenching by protonated imidazole and carboxyl groups, resulting from protonation of such residues in the vicinity of Trp-214. Three amino acids with carboxylate side chains (Glu-292 of domain II and Glu-450 and Asp-451 of domain III) and one histidine residue (His-242 of domain II) are found within 10 Å distance from Trp-214, based on the atomic coordinates of HSA (Protein Data Bank entries 1UOR (48) and 1AO6 (49)). In addition, Cowgill (50) reported that loss of helical conformation is correlated with a decrease in fluorescence intensity, which is the case in the N-F transition, as discussed in the far-UV CD section. The reduction in fluorescence intensity below pH 3.5 may arise from quenching by protonation of carboxylate groups, whereas the increase in fluorescence at pH values below 2.5 can be attributed to restoration of secondary structure at strong acidic pH without added salt (20,21), which was reported to trigger fluorescence intensity (50). Consequently, a more hydrophobic environment of Trp-214 may be created, leading to increased tryptophanyl fluorescence intensity as described by Halfman and Nishida (44,45).
The observed decreases of max and fluorescence intensity of HSA in the pH region from 7.4 to 9.0 ( Fig. 2A) might either be due to changes in the secondary structure as discussed in the far-UV CD section and/or to quenching of the tryptophanyl fluorophor by deprotonation of groups that are in close proximity to the indole chromophore, such as ⑀-amino groups of lysyl residues. Two such residues are found in the crystal structures of HSA, viz. Lys-199 at 3.7 Å and Lys-195 at 7.4 Å distance to Trp-214.
Profound differences between the fluorescence behavior of the lone tryptophanyl residue in HSA-DOM II and in HSA at pH 7.4 are highlighted by the 2.6-fold lower quantum yield of HSA-DOM II and the pronounced shift of the emission maximum to shorter wavelength ( Fig. 2A), denoting a dramatically changed microenvironment of this residue in HSA-DOM II. As can be deduced from the crystal structures of human serum albumin, Trp-214 acts as an important stabilizer (by hydrophobic packing force) for the interface between subdomain IIA and IIIA (1). This hydrophobic area, which is centered in the crystal structure of domain II, loses its natural counterpart when expressed in the form of a single domain, leading to modified water accessibility of the tryptophanyl residue in the recombinant protein, thereby affecting the alterations in fluorescence behavior of the tryptophanyl residue in HSA-DOM II. The reduced fluorescence intensity may indicate an impairment of the helix harboring the lone tryptophanyl in the native molecule, which is supported by the reduced helical content of HSA-DOM II compared with HSA at pH 7.4 (Table I). By using light meromyosin, a helical muscle protein, Cowgill (50) demonstrated a 3-fold reduction of the tryptophanyl fluorescence intensity by disaggregation of the helical conformation and attributed this observation to the quenching by peptide carbonyl groups in disordered polypeptides. The shifted wavelength maximum of HSA-DOM II compared with HSA may be due to a more hydrophobic environment of the tryptophanyl residue and/or sandwiching of the side chain in the rigid portion of the protein matrix, as has been described Eftink and Ghiron (24) for HSA.
In the weak acid pH region, there is a sigmoidal 2.4-fold increase in fluorescence intensity of HSA-DOM II ending near pH 4.0 ( Fig. 2A). Interestingly, at pH 4.0 HSA-DOM II and HSA have almost superimposable tryptophanyl fluorescence spectra suggesting similar microenvironments encircling this residue in both proteins. The pronounced increase in fluorescence intensity of HSA-DOM II between pH 7.4 and 4.0 is correlated with the secondary structural rearrangement observed by far-UV CD in the same pH region. It has been reported that loss of helical conformation resulted in decreased fluorescence (see above) (50). Conversely, we detect an increased ␣-helical content with decreasing pH (Fig. 1B and Table I), which therefore may trigger the fluorescence of the tryptophanyl residue in HSA-DOM II. In addition, Steiner and Edelhoch (51) reported that an increase in intensity in the acid pH region is very likely to be connected with a definite conformational change in the protein structure, because all side chains that are known to quench the tryptophanyl fluorescence intensity become protonated in this pH region and should therefore quench the fluorescence (tryptophanyl fluorescence quenching groups: -COOH, protonated imidazole, deprotonated ⑀-NH 2 , -SH, -S-S-, tyrosinate anion).
The spectra of the tryptophanyl residues of HSA and HSA-DOM II in 9.5 M urea (data not shown) are superimposable and show reduced intensity caused by collisional quenching with the surrounding solvent molecules with an emission maximum red-shifted to approximately 349 nm, which is characteristic for a fully exposed tryptophanyl residue (24,52).
In the weak alkaline pH range, the fluorescence intensity of HSA-DOM II increases, which is accompanied by a shift of the emission maximum to higher wavelength ( Fig. 2A). The decrease in fluorescence quenching in this pH region may be due to the deprotonation of imidazole side chains with unusually high pK, which are known to be located in domain II of HSA (17); however, the reduction of the emission maximum suggests a structural transition in the zone surrounding the tryptophanyl residue in HSA-DOM II.
Tyrosyl Fluorescence-The fluorescence of proteins originates almost entirely from the tyrosyl and tryptophanyl residues. In proteins that lack tryptophanyl residues (class A, tryptophanyl-free proteins), tyrosyl fluorescence is used as a probe for conformational changes (53).
HSA-DOM I harbors seven tyrosyl residues. As indicated by the crystal structures (Protein Data Bank entries: 1UOR (48) and 1AO6 (49)) four of them (residues 30, 138, 140, and 161) are located in helical regions, and all of them are at least minimally exposed to the solvent. We observed a significant increase in tyrosyl fluorescence intensity of HSA-DOM I in the pH region from 6.0 to 3.5 (Fig. 2B). Inspection of the atomic coordinates of HSA shows that two carboxylate side chains, Asp-38 and Asp-108, are within 5 Å of Tyr-84 and Tyr-148, respectively. As has been demonstrated by Cowgill (54), carboxylate groups serve as acceptors for the phenolic proton in the excited state and thus are among the known quenchers for tyrosyl fluorescence. In addition, it was reported that tyrosyl fluorescence is strongly quenched by hydrogen bonding of the phenolic hydroxyl to carbonyl or amide groups that could be supplied by the protein backbone and/or by side chains of asparagine and glutamine next to tyrosyl residues (53,55,56). As a matter of fact, hydrogen bonds between the OH group of Tyr-30 and the amide side chain of Asn-99, as well as between the OH group of Tyr-84 and backbone carbonyl of Gln-33 are strongly indicated by the atomic coordinates of HSA. Therefore, protonation of the aspartate side chains mentioned above accompanied by structural changes which lead to disruption of these hydrogen bondings may explain the observed rise in fluorescence intensity. All this suggests a rearrangement of tertiary rather than of secondary structure of HSA-DOM I in the region of the N-F transition of HSA, which is corroborated by the results of near-UV CD measurements discussed below.
Three out of the four tyrosyl residues of HSA-DOM III are known to be located in helical regions (residues 401, 411, and 452 of HSA) (Protein Data Bank entries 1UOR (48) and 1AO6 (49)). As the pH of a HSA-DOM III solution is reduced to pH 4.0 there is a sharp sigmoidal decrease in the tyrosyl fluorescence intensity and a slight increase in the emission maximum (Fig.  2B). This decrease in fluorescence intensity may be due to a loss of helical structure around the three tyrosyl residues located in ␣-helical regions, in conformity with the pronounced reduction of helical content that we observed in the identical pH range, as described above in the far-UV CD section. Similar observations have been made by Cowgill (54) in the case of helical muscle proteins. In addition, no carboxylate side chains that could potentially trigger the fluorescence by their protonation can be found in 5 Å surrounding the four tyrosyl residues. By further reduction of the pH of the solution, the fluorescence rises continuously, which may be ascribed to the disruption of hydrogen bondings between the hydroxyl groups of tyrosyl residues in HSA-DOM III and carbonyl groups of the peptide backbone (Tyr-452 to Asn-429 and Tyr-497 to Lys-534) (55) emphasized by the loss of secondary structure of HSA-DOM III in the F-E transition.
By increasing the pH of a HSA-DOM III solution, the intrinsic tyrosyl fluorescence decreases, whereas a shoulder in the emission spectrum near 350 nm occurs (data not shown). These findings strongly suggest the deprotonation of the phenolic hydroxyl group of some tyrosyl residues in HSA-DOM III, as was suggested previously (25,54,57). In addition, Edelhoch et al. (57) reported that significant inhibition of phenol emission occurs by uncharged amino groups. In this pH region some lysyl residues may be deprotonated and lead to quenching of tyrosyl fluorescence. In fact, in the crystal structure of HSA, amino groups of lysyl residues are found in close proximity (5 Å surrounding) to all 4 tyrosyl residues of HSA-DOM III: Tyr-401 to Lys-525, Tyr-411 to Lys-414, Tyr-452 to Lys-432 and Tyr-497 to Lys-534. All these findings and the fact that HSA-DOM III undergoes only minor changes in secondary structure in the slightly alkaline pH region provide evidence that the observed decrease in fluorescence intensity of HSA-DOM III is not correlated with a structural change, which in addition is strongly supported by the near-UV CD results discussed below.
Near-UV CD-CD spectra in the near-UV region provide information about the asymmetry of the structure around the aromatic amino acid side chains and disulfide bridges and therefore on fluctuations in the tertiary structure of a protein.
The main contributions to the ellipticity from tryptophanyl and tyrosyl groups are obvious above 265 nm, with the largest maxima around 279, 284, and 291 nm for tryptophanyl, around 277 nm for tyrosyl, and around 255, 261, and 268 nm for phenylalanine residues. The ellipticity of S-S bridges in proteins can be significant from 320 to 250 nm (58 -61).
The observed alterations in ellipticity of HSA in the acid pH range (Fig. 3A) reflect the tertiary structural alterations in the N-F and F-E transition, in correlation with the loss of secondary structure described above. These findings are in close agreement with previous results obtained with BSA (30). The observed decrease of [] 290 -300 in the spectrum at pH 4.0 was ascribed to the immobilization of the tryptophanyl residues of BSA (30). Leonard and Foster (37) reported that approximately four buried tyrosyl residues of BSA appear to become solventexposed during the acid expansion, whereas none do so in the N-F transition. This may, in addition to the environmental perturbations of the disulfide bridges, be an explanation for the overall enhancement of the ellipticity and the loss in fine structure below 280 nm between pH 4.0 and 2.0. The residual near-UV CD signals at pH 2.0, which are significant compared with the spectrum at 9.5 M urea, strongly suggest the existence of tertiary packing, at least around aromatic residues and disulfide bridges even at this low pH.
The alterations in the spectra between pH 7.4 and 9.0 demonstrate tertiary structural changes in the N-B transition of HSA (Fig. 3A). This was ascribed to environmental perturbations of the 17 disulfides in BSA, such as variations in the dihedral angles, with some contribution of phenylalanine and tyrosyl residues (29,30), which were shown by UV difference spectroscopy to gain increased accessibility in this pH range (62). In addition, a slight reduction in ellipticity in the region between 280 and 300 nm emerged, which might be due to an immobilization of the tryptophanyl side chain, as was described in the case of the N-F transition of BSA by Sogami et al. (30) and is supported by the observed changes in the tryptophanyl fluorescence discussed above.
The spectra of the three recombinant domains are dominated by the large number of disulfide bridges analogous to HSA, whereas some difference in fine structure is obvious, which may be owing to different relative amounts of aromatic amino acid residues (Fig. 3, B-D). The general similarity of the spectra of HSA, HSA-DOM I, HSA-DOM II, and HSA-DOM III at pH 7.4 reflects their structural, evolutionary-based relationship and emphasizes that the recombinant domains as standalone proteins have the ability to adopt folds that are similar to their respective structure in the context of the whole protein.
In the pH region of the N-F transition of HSA, all three domains display increases in ellipticity of the minima at 262 and 268 nm (Fig. 3, B-D), which may be due to loss of asymmetry around disulfide bridges and/or aromatic residues, indicating disruption of tertiary structure. In addition, major global changes with additional minima in the spectra of HSA-DOM II occur between pH 7.4 and 4.0. This may reflect dramatic changes in the tryptophanyl environment, supported by the observed 2.4-fold increase in fluorescence intensity and the secondary structural changes ( Fig. 2A and Fig. 1B). Whereas the pH-dependent changes of the spectra of HSA-DOM II and HSA-DOM III coincide with the observed alterations in secondary structure and tryptophanyl and tyrosyl fluorescence respectively, the most striking result of our CD measurements in the near ultraviolet region is the fact that HSA-DOM I undergoes fluctuations in tertiary structure that are accompanied by only a small loss in secondary structure, which could so far only be deduced from experiments with larger fragments (12). HSA-DOM I shows flattening of the shoulder near 255 nm as the pH is diminished to 4.0 and also as it is increased to 9.0, indicating perturbations of phenylalanine residues. An opening of the crevice harboring Cys-34 has been reported to occur during both the N-F and the N-B transition of BSA (63,64). The crystal structure of HSA shows that Phe-26 and Phe-37 are located in the same crevice, and it is therefore possible that the observed flattening of the shoulder at 255 nm may partly be attributed to perturbations of these particular residues.
By reducing the pH further, all three recombinant proteins lose tertiary structure as indicated by the rise in ellipticity at 262 and 268 nm and the loss of fine structure (Fig. 3, B-D), in agreement with the observed decrease of the secondary structure content in the region of the acid expansion of HSA. By inspection of the spectra of HSA-DOM I, HSA-DOM II, and HSA-DOM III at pH 2.0 and 9.5 M urea, which represents the fully denaturated state as indicated by the spectra in the far-UV region (Fig. 1A), it becomes evident that HSA-DOM I and HSA-DOM III, similar to HSA, display significant tertiary structural elements even at pH 2.0. It has been reported for some proteins (20,21), especially for those that are heavily cross-linked by disulfide bridges, as is the case for albumin and its domains, that the intramolecular charge-charge repulsion is insufficient to cause complete unfolding in the range of pH 2 in the absence of salt. This may serve as an explanation why we could observe residual secondary and tertiary structure with HSA, HSA-DOM I, and HSA-DOM III, even at this low pH.
In contrast, the ellipticities of HSA-DOM II at pH 2.0 and 9.5 M urea are comparable (Fig. 3C). The fact that the spectra are different in shape despite the supposed complete denaturation at these conditions may be explained by residual tertiary packing around the lone tryptophanyl residue, supported by the difference in its fluorescence intensity and emission maximum at pH 2.0 and 9.5 M urea, respectively. Similar observations with a number of proteins including bovine serum fetuin have been described (Ref. 65 and references therein). Therefore we speculate that HSA-DOM II transforms to a molten globulelike state in the acidic pH region with pronounced loss of tertiary but preserved secondary structure (66).
At pH 9.0, HSA-DOM I and HSA-DOM II display a slight enhancement of the CD signal at 262 and 268 nm, some fluctuations in fine structure most evident in the spectrum of HSA-DOM I in the wavelength region above 285 nm, as well as a flattening of the shoulder near 255 nm, indicating a perturbation of Phe-26 and Phe-37 in the pH region of the N-B transition as discussed above. Since there are no significant changes in the secondary structural content of these two proteins, these findings clearly denote a tertiary structural rearrangement, which has been speculated to happen in the Nterminal region of the native HSA and BSA molecules during the N-B transition (2,11,17,64,(67)(68)(69). No changes in the spectra of HSA-DOM III are observable in this pH region which, in the context with the observations discussed above, leads us to the conclusion that HSA-DOM III is not involved in the N-B transition of the intact molecule.
Conclusion-It is clear that complex chemical intra-and inter-domain forces stabilize the albumin structure and that an isolated domain cannot be expected to fully reflect its behavior in the context of the whole protein because of the absence of all inter-domain contacts. This is especially true for HSA-DOM II which, due to its central position in the albumin molecule, experiences the greatest proportional loss of framework structure when expressed as a stand-alone protein, underlined by the proposed formation of a molten globule-like state of HSA-DOM II at acidic pH in contrast to the other two recombinant domains. However, it is evident from the work presented here that certain features, e.g. the origin of the molecular rearrangements during the N-F and N-B transition, can be attributed to specific albumin domains unequivocally, in particular to HSA-DOM III and HSA-DOM I, respectively. In conclusion we were able to demonstrate that the recombinant domains of HSA represent powerful tools for the dissection and study of important structural and functional characteristics of the native molecule.