Pure ovalbumin was purchased from Sigma Aldrich (UK). Recombinant E. coli CyaY and S. cerevisiae Yfh1 were produced as previously described in detail.11,15 In short, CyaY (Swiss-Prot:P27838) was cloned into a pET-derived plasmid vector as a His-tagged glutathione-S-transferase (GST) fusion protein comprising a tobacco etch virus protease cleavage site, which leaves a short N-terminal GlyAla tag. For Yfh1, we used a pET11A expression vector in which residues 52–174, which correspond to the mitochondrial mature form of the protein, had been subcloned. This vector was a generous gift of Stemmler (Wayne State University). ¹⁵N labeled and ¹⁵N labeled samples were produced by growing the bacteria in minimal medium using ammonium sulfate as the sole source of nitrogen. The proteins were expressed in Escherichia coli BL21-(DE3) cells grown at 37°C, induced with in 1 mM IPTG for 6 h, lysed with a French press and sonicated. CyaY was first subjected to affinity chromatography (glutathione-S-Sepharose) and then further purified by gel filtration chromatography on a Superdex G75 16/60 column (Pharmacia). Yfh1 was purified by two ammonium sulfate precipitation steps with a 40% cut to precipitate contaminating proteins and a 65% cut to precipitate Yfh1. After dialysis, the protein was subjected to anion exchange chromatography using a Pharmacia Q-Sepharose column with a gradient to 1M NaCl, followed by a Pharmacia phenyl-Sepharose column with a decreasing 1M ammonium sulfate gradient. The EDTA containing protease inhibitor cocktail (Roche) was added before the cells were lysed and at each purification step. EDTA and salts were removed by dialysis before concentration of the protein. Protein purity and identity were checked using SDS-PAGE and mass spectrometry.

Several types of chicken egg whites and one from quail egg were compared and evaluated in relationship to their features, notably pH, in relation to ageing. Samples of egg white, to be used as medium in NMR experiments, were taken either from commercial chicken eggs or from free-range chicken eggs collected in two villages of Mallorca (Son Sardina and Calonge). Mechanical separation from yolk and subsequent filtering with cheesecloth allowed elimination of chalazae and membranes, to retain only the thin portion of HEW. Samples of HEW were prepared and used immediately before NMR measurements. Lyophilized proteins were dissolved in about 0.5 mL of neat HEW together with 40 μL of D₂O.

NMR spectra were recorded on a Varian INOVA spectrometer and on a Bruker AVANCE, both operating at 600 MHz ¹H frequency. Typically, measurements were carried out using protein (unlabeled or ¹⁵N uniformly labeled) concentrations of 0.3–0.5 mM. Water suppression was achieved by the WATERGATE pulse-sequence,16 HSQC experiments were used as described by Bax et al.17 The spectra were processed, zero-filled to the next power of two, or just analyzed using the NMRPipe program18 or SPARKY19 software. Baseline correction was applied when necessary.

¹⁵N T₁, T₂, and NOE NMR relaxation measurements were performed at 600 MHz and 25°C on approximately 0.25 mM samples. Both T₁ data and T₂ data were acquired with eight relaxation delays (10, 100, 200, 400, 700, 1000, 1500, and 2000 ms and 8.7, 17.4, 26.1, 34.8, 52.2, 69.6, 87.0, and 104.4 ms, respectively). Experimental steady-state NOE values were determined from the peak intensity ratios of amide signals obtained by recording interleaved 2D Watergate ¹H-¹⁵N HSQC spectra with and without a proton saturation delay of 4 s and a repetition delay of 4.2 s. T₁ and T₂ relaxation times were obtained by fitting the data with a two-parameter single exponential decay function. The T₁ and T₂ values of residues that differ by more than two standard deviations from the mean value were not considered in the correlation time calculations.

T₁ and T₂ values for each residue were used to determine R₁, R₂ that in turn were fed to the r2r1_tm program (Palmer's group, http://www.hhmi.umbc.edu/toolkit/analysis/palmer/r2r1_tm.html) to determine the local correlation time for each residue. Averaging the obtained values, it was possible to determine τ_c values.

We used both commercial and freshly laid eggs. When used within a few hours from deposition, eggs had a pH between 8.2 and 8.3. One week later, the pH reached values between 8.5 and 9.3, typically close to values of commercial eggs. A widely accepted explanation for this observation is that intact, freshly laid eggs contain a high concentration of CO₂, which is then progressively lost during time (http://newton.ex.ac.uk/teaching/CDHW/egg/). After this initial variation which can be up to 1 pH unit, the pH drift is minimal.

Beyond this, the 1D proton spectra of eggs from different sources are well reproducible with overall very common features [Fig. 1(A)]. The spectra are dominated by the resonances of a single protein, which by comparison with the known HEW composition and literature spectra can be easily identified as ovalbumin, as its concentration in egg white is estimated approximately to 1.5 mM [Fig. 1(B)]. Additionally, in all spectra there are some sharp peaks centered around 3.8 ppm, which have a pattern typical of carbohydrate spectra.20 A few other sharp but less intense peaks in the aromatic region seem to be specific only for free-range eggs of a given source. Although the intensities and/or the chemical shifts of some of the small molecular weight peaks changed on aging, the envelops originating from HEW proteins did not. This finding indicates that there is no appreciable protein degradation or spectral alternations as necessary in view of the possible use of HEW as a medium for long NMR experiments.

We then tested whether CyaY, a small globular protein of 106 amino acids with an α/β-fold and metal binding properties, yields acceptable NMR spectra in HEW. Comparison of the HSQC spectra of CyaY in buffer and in HEW shows that all peaks indicative of properly folded proteins are retained in HEW, albeit significantly broadened [Figs. 2(A,B)]. The superposition of the two spectra confirms that there is no significant chemical shift perturbation in going from the dilute solution to the crowded environment [Fig. 2(C)].

The absence of significant chemical shift changes indicates that our natural environment did not produce any chemical modification, for example, phosphorylation, of the side chains of the probe protein. Interestingly, the only significant differences between the two spectra (inany case ≤0.1 ppm and 1 ppm in the proton and nitrogen dimensions, respectively) map to residues, which are known to be highly susceptible to the divalent cation content,21 thus, suggesting that HEW is anyway a more suitable model system of the in cell situation than any nonspecialized buffer.

To better characterize the properties of CyaY in HEW we measured typical dynamical parameters by NMR ¹⁵N relaxation experiments. This technique has proven to be very successful in providing information about molecular internal motions.

Overall, the longitudinal (T₁) and transverse (T₂) relaxation rates, and the NOE values are rather uniform along the protein sequence, in agreement with what is expected for a compactly folded globular protein (Fig. 3). However, there is a clear lengthening of T₁ and a shortening of T₂ in ovo (red bars) as compared with the data in buffer, whereas no substantial difference is observed in the NOE values. The average value of T₁ goes from 602 ms in buffer to 775 ms in egg white, whereas the corresponding values of T₂ vary from 121 to 45 ms, respectively.

The overall rotational correlation time (τ_c) of CyaY at room temperature in HEW, as estimated from T₁/T₂ ratios, is 13.1 ns. For comparison, the value in diluted buffer calculated with the same approach is 6.4 ns. Although the latter value is in excellent agreement with what is expected for a monomeric globular domain of equivalent size,22 the value in HEW is appreciably longer. Consistent values were independently confirmed also by a recently devised, more rapid method for the direct determination of the ¹⁵N T₁/T₂ ratios (Kelly and Frenkiel, in preparation). This method estimates the correlation time from a series of 1D profiles, in a manner conceptually similar to the TRACT method,23 making the estimation of correlation times faster. Also according to this method, the rotational correlation time of CyaY doubles, going from ∼6 ns in the dilute buffer solution to ∼12 ns in the crowded environment of HEW.

The rate of rotational relaxation is commonly explained in terms of solvent viscosity. The dependence of the correlation time (τ_c) from viscosity (η) is given by the well-known Debye–Stokes–Einstein equation.24 According to the proportionality between τ_c and η, we could expect an increase of τ_c up to four times in going from a dilute buffer solution (with a viscosity close to 1 cp) to the thin portion of HEW, whose viscosity has been estimated as 4 cp.25 The doubling of τ_c observed experimentally, albeit smaller than the maximum possible increase, is consistent with the influence of viscosity as the main effect.

To check the effect of HEW on protein stability, we used the yeast ortholog of CyaY, Yfh1. One of the peculiarities of this protein is that it undergoes cold denaturation at temperatures above the freezing temperature of water and heat denaturation well below the coagulation temperature of HEW.13,14,21 This gave us the unique opportunity to check the thermal stability of a protein in HEW over the entire range of temperature comprising heat and cold denaturation. Both HSQC spectra (Fig. 4) and relaxation behavior of this protein in HEW (Fig. S1 of Supporting Information) are comparable to those found for its bacterial ortholog.

We ran spectra of Yfh1 in the temperature range 0°C to 45°C and observed both cold and heat denaturation (Fig. 4). Although the spectral quality prevents an accurate quantitative assessment of the thermodynamic parameters of the transition, it is very interesting to observe that the temperature dependence in HEW parallels that in dilute conditions. Altogether, it is fair to say that the thermal stability of Yfh1 in HEW is not higher than thatmeasured in dilute buffer, with a low temperature transition between 5 and 10°C and a high temperature transition temperature between 30 and 35°C. As for CyaY, the spectrum of the folded species at 25°C in HEW has broadened peaks with respect to that in dilute buffer solution. It does otherwise correspond well to the one at room temperature in Hepes previously reported by us.14 The same can be said for the cold denatured species: the spectrum in HEW at 0°C (upper left panel in Fig. 4) is very similar to that in Hepes.14 On the contrary, the spectra monitoring heat denaturation in HEW look intrinsically different with respect to those in Hepes. The spectrum at 45°C in HEW, (upper right panel of Fig. 4), shows only very few barely visible peaks, whereas the published spectrum at the same temperature, in buffer, is nearly indistinguishable from that of the cold denatured species.14 At first sight, such a behavior seems to speak of an irreversible precipitation of the unfolded species at high temperature. However, as shown in the right bottom panel of Figure 4 (spectrum labeled 25°C_back), the sample cooled down from 45 to 25°C has a spectrum identical to the starting one, a clear indication that also the high temperature unfolding in HEW, is a completely reversible process.

The disappearance of all cross peaks parallels the unfolding of Yfh1 as the temperature is raised from room temperature to 45°C. All cross peaks reappear as the protein folds back to its native state. Two possible explanations come to mind: an enhanced exchange rate due to the higher pH in HEW or a weak interaction of the unfolded species with HEW proteins, enhanced by high temperature. We checked the behavior of Yfh1 at a pH higher than that used in the published study14 (pH 7.0 in 20 mM HEPES). It is possible to see (in Fig. S2 of Supporting Information, showing representative spectra in the range 5 to 35°C) that even at pH 8.5 (in 20 mM TAPS) the spectrum of the unfolded species does not disappear at high temperature. Accordingly, we favor the hypothesis based on weak interactions with HEW proteins. We hypothesize that the binding of the high-temperature unfolded species to larger HEW proteins leads to broadening of its resonances so large as to make them unobservable. These interactions, possibly due to hydrophobic forces, weaken as the temperature is decreased (and the protein concomitantly refolds). The absence of a similar behavior at low temperature is consistent with the strong hydration of the cold denatured species, we have recently described.26 It is fair to hypothesize that hydration of amide groups and possibly also of side chains, as implied in the standard model of cold denaturation,27 can effectively shield the unfolded species from weak interactions with the proteins of HEW.