Relaxometry of lens homogenates. II. temperature dependence and comparison with other proteins

MAGNETIC RESONANCE IN MEDICINE 10,362-372 (1 989)

Relaxometry of Lens Homogenates. 11. Temperature Dependence and Comparison with Other Proteins

CHRISTOPHER F. BEAULIEU, * RODNEY D. BROWN III,? JOHN I. CLARK, * MARGA SPILLER,~ AND SEYMOUR H. KOENIG~

*The Department of Biological Structure, University of Washington, School of Medicine, Seattle, Washington 98195; and t l B M T. J. Watson Research Center, Yorktown Heights, New York 10598

Received July 8, 1988; revised October 19, 1988

We have extended our earlier work (C. F. Beaulieu, J. I. Clark, R. D. Brown 111, M. Spiller, and S. H. Koenig, Magn. Reson. Med. 8,45 (1988)) on the magnetic field dependence of 1 / T , (NMRD profiles) of calf lens nuclear homogenates, at 25"C, to 5'C, and to other protein systems as well. These include concentrated solutions of myoglobin and bovine serum albumin, both globular proteins, the first compact and roughly spherical, the other extended, flexible, and with weak internal bonding; chicken lens homogenate, for which the dominant crystallins (lens proteins) are -70% a-helical compared with calf crystallins, which are essentially all &sheet; and hen egg white, both native and heat- denatured. Our earlier conjectures regarding a reversible change in protein organization of the calf lens crystallins as a function of solute protein concentration is given added support. Our findings suggest that cytoplasmic homogenate can be characterized as a heterogeneous and polymorphic solution of crystallins. At high concentrations the NH moieties of the protein backbone become accessible to solvent with water (not NH proton) exchange rates > lo4 s - I . This conclusion is based on two aspects ofthe observed NMRD profiles. At low crystallin concentration, the profiles of calf and chicken lens homogenates are similar in form to those of myoglobin and native hen egg white, a form that has been studied previously for a range of diamagnetic globular proteins and has been demonstrated to arise from the rotational thermal motion of the solute molecules. At high crystallin concentrations, the NMRD profiles of the lens homogenates develop a monotonic background (high rates at low fields), much like that of the heat-denatured egg-white sample and those of most tissues. In addition, there is a set of peaks in the central part of the profiles of the concentrated crystallins, seen also in the denatured egg white and some tissues but not in the myoglobin sample, which is known to arise from cross-relaxation interactions between the water protons and (through the intermediary of the NH proton) the I4N quadrupolar levels. The magnitude of these peaks, which is larger by an order of magnitude for native calf lens homogenates than for any tissue, requires that the majority of the NH moieties be accessible to water. Finally, going to 5°C for the native calf lens homogenate takes the sample below the temperature of reversible phase separation, and it becomes opaque. However, there is no change in the profile that can be attributed to this transition and its associated opacity. o 1989 Academic Press. Inc.

INTRODUCTION

In a recent paper ( 1 ), we reported the magnetic field dependence of 1 /TI of water protons (NMRD profiles) for transparent homogenates of calf lens at 25°C. The samples included nuclear homogenates, with total crystallin (heterogeneous lens

0740-3194189 $3.00 362 Copyright @ I989 by Academic Press, Inc. AU rights of reproduction in any form R X K V ~ .

NMRD OF LENS HOMOGENATES, 11 363

proteins) content between 34% ( v / v ) (native) and 14% (diluted), and cortical homogenates with 2 1 % (native) and 34% (concentrated) total protein. The NMRD profiles had two contributions: a monotonic dispersive component (analogous to those of solutions of diamagnetic globular proteins (2, 3 ) and excised tissue ( 4 , 5 ) ) and a structured region (between about 0.6 and 3 MHz proton Larmor frequency) arising from cross-relaxation between water protons and the I4N quadrupolar levels of protein NH moieties. These 14N peaks had been reported previously only for tissue and dehydrated proteins ( 4 , 6 , 7 ) ; never for protein solutions. Moreover, the 14N peaks of lens cytoplasm were unusually large, 10-fold greater in peak height than, for example, those seen in rat heart muscle ( 4 ) . The behavior of both the monotonic background and the peaks, as a function of protein concentration, indicated that the crystallins in solution undergo a change in organization in the range 14 to 19% protein. For example, the peaks, not visible in the data below - 14% protein, increase essentially linearly in magnitude with volume fractionf at crystallin concentrations above this value. In addition, the functional form of the monotonic component of the profiles, at lower concentrations, varies with concentration in a way that indicates a dependence of the orientational relaxation time of the protein molecules on protein-protein interactions seen previously in solutions of other globular proteins (8) fl By comparison, at higher concentrations, this component has a fixed form that scales at all fields as f / 1 - f.

These results are suggestive of a dependence of protein conformation and association on concentration that may be unique to lens crystallins and related to specific interactions required to maintain lens transparency in vivo. In particular, the magnitudes of the peaks indicate unusual access of water to essentially all the NH moieties of the residues ofthe protein backbone ( 9 ) . In the present work, we extend our earlier observations to YC, below the phase separation temperature at which the samples become white and opaque ( l o ) , and also include measurements on highly concentrated solutions of myoglobin (a compact globular protein), bovine serum albumin (BSA, a rather elongated protein with three globular sections in a linear array and “internal bonding that is unusually weak” ( I ] ) ) , chicken lens (with crystallins that have a substantial component of a-helix, in contrast to the predominantly P-sheet structure of calf crystallins (12)) , and hen egg white in various states of polymeriza- tion (raw and cooked).

The results lend support to our earlier conjectures regarding the unique properties of lens proteins by confirming that the effects seen earlier in lens homogenates ( I ) are not found in solutions of compact globular proteins. First, at one extreme, no peaks are visible in (globular) myoglobin solutions even at the highest concentrations whereas peaks appear in the egg white as the protein is progressively denatured and crosslinked by heat. Second, the chicken lens protein, although highly helical, can undergo a conformational change that exposes the majority of its NH groups to solvent and, in that sense, can have an open conformation much like that ofcalf crystallins; by extension, the existence of peaks at sufficiently high crystallin concentrations ostensibly correlates with the ability of lens proteins to satisfy their physiological roles. Third, there is little difference in the NMRD profiles of the calf lens nuclear homogenates over the whole temperature range, from well above to well below the cold cataract transition, consistent with this phenomenon being a phase separation on a scale (13, 1 4 ) too large to affect solvent proton magnetic relaxation.

364 BEAULIEU ET AL.

MATERIALS AND METHODS

Samples. The calf lens samples were nuclear homogenates prepared for the earlier study ( I ) and reference should be made to that work. The chicken lens homogenate was prepared in an analogous fashion except that the starting material was about sixty 9-week-old chicken lenses (ACME poultry, Seattle, WA). The myoglobin was from horse heart (Sigma Chemical Co.) and prepared at 4 I % concentration (w/ w) ( 34%, v/v) in water, pH - 7, its isoelectric point. Bovine serum albumin (Sigma) was prepared at 40% (w/w) (33%, v/v), pH 7.35 (well above its isoelectric point of4.9). The egg white was used directly from a raw egg, and subsequently denatured, first by gentle heating for under 1 min near 60°C and then by heating near 80°C for another minute; neither the water nor the protein content of the sample was altered by this process.

Reluxation rate measurements. The I / T I was measured over the field range 0.0 1 to 50 MHz proton Larmor frequency, corresponding to 0.00024 to 1.2 T, using an automated field-cycling relaxometer (I5), as before (I). The sample, in a stoppered test tube, was surrounded by circulating Freon, allowing measurements in the range - 10 to 35"C, regulated within k O . 1 "C.

Data reduction. The data were reduced as in the past ( I-3): the Cole-Cole expression, known to describe the monotonic components of the NMRD profiles of tissue and most protein solutions very well, were fit to the data points excluding the structured regions between 0.6 and 3 MHz. (The results of the Cole-Cole fits are shown in one case only since numerous examples are given in the earlier work ( I ) ) . The Cole-Cole expression describes a profile much like a Lorentzian, flat at low fields and decreasing monotonically to a high field-limiting value, with a point of inflection at a Larmor frequency vc that is also the field at which the dispersive component has decreased to one-half its total excursion. It differs from a Lorentzian only in that it can account for a range of slopes at the inflection point. It is well documented (3) that v, is related to the orientational relaxation time 7 4 of (spherical) protein molecules (as measured by dielectric dispersion, for example) by v, N (&)/27rTd where, by Stokes' law, Td = 47rr3v/kT. Here, 77 is the (neat) solvent viscosity and Y is the protein radius. These fits are treated here as phenomenological, and used mainly as baselines to allow expanded displays (I, 9) of the region with structure.

RESULTS

Figures I A and IB show NMRD profiles of native (circles) and diluted (squares) calf lens nuclear homogenates at 5°C (solid symbols) and 25°C (open symbols). The samples are transparent at 25"C, whereas they are white and opaque at 5°C. In Fig. IA, the obvious presence of I4N peaks at the higher concentration (34%, v / v ) with but a trace remaining at one-half the concentration ( 17% ) is a feature now confirmed over the range 5 to 35"C, and shown in the figure for 5°C as well as at 25°C (I) . 'The solid curve associated with the 25°C data is an example of a Cole-Cole representation of the monotonic component of an NMRD profile. The form of the monotonic background of the diluted sample, with an inflection near 0.15 MHz and a change in amplitude with temperature at 0.01 MHz of -6:4, about equal to the change in solvent viscosity, is what would be expected for a solution of homogeneous globular protein of about lo6 Da (2, 3 ) , taking into account the dependence of these effects


MAGNETIC FIELD (T) MAGNETIC FIELD (T) A 0.001 0.01 0.1 1 B O 0.02 0.04 0.06 0.08

r\ r I Calf Lens a6

PROTON LARMOR FREQUENCY (MHZ) PROTON LARMOR FREQUENCY (MHz)

FIG. 1. (A) The 1 / T , NMRD profiles ofa homogenate of native calf lens nucleus (circles) and an aliquot diluted to half the protein content (squares), at 5°C (solid symbols) and 25°C (open symbols). The protein contents ofthe samples were 34 and 17% (v/v), respectively. At 5”C, the native sample, below the temperature for reversible “phase separation cataract formation,” was white and opaque. The solid curve through the 25°C data is an example of a fit ofthe Cole-Cole expression to the monotonic part ofan NMRD profile. (B) The structured region of ( A ) shown on an expanded scale, after subtraction of the Cole-Cole fit to the monotonically decreasing background (see text). The peaks are associated with level crossing ofthe solvent proton Zeeman energy levels with the electric quadrupolar levels of I4N nuclei of NH moieties of the protein backbone.

on protein concentration (8). It is not possible, however, because of the high (and instrumentally inaccessible) rates of the native sample at low field, to generalize about the monotonic part of the higher concentration profile.

The 14N peaks are shown expanded and on a linear field scale in Fig. 1 B. (A quantitative theory of the positions and amplitudes of the peaks is given in (9) .) Here it is seen that there is no demonstrable difference in the peaks at 25 and 5°C for the native sample, despite the transition from clear to opaque. (Indeed, NMRD profiles have been measured at 5°C intervals from 5 to 35°C; the amplitudes of the peaks show essentially no variation with temperature over this range.) For the diluted sample, with one-half the protein concentration, the magnitude of the peaks is only about 20% that ofthe native sample. The apparent change with temperature is barely within experimental error, given the large monotonic component and the concomitant curved baseline.

Figure 2 shows the NMRD profiles for a solution of horse heart myoglobin, at 5 and 25°C at 34% (v/v), a concentration equal to that of the native calf lens homogenate, Fig. 1 A. The profiles are those expected for a solution of diamagnetic globular protein, judged from both their shape and temperature dependence, with no visible deviation of the profile from monotonic behavior. (Ferrous ions do not contribute to relaxation by myoglobin and hemoglobin ( 1 6 ) , even when paramagnetic, since their electronic relaxation time is too short. Moreover, the 50-MHz magnitude of any paramagnetic contribution, such as a possible contribution from (ferric) metmyo- globin contamination, is never less than 0.3 of its low field value. Thus the low rate at high field, Fig. 2, sets an upper limit on the maximum possible paramagnetic contribution such that the profiles in the figure can properly be taken as diamagnetic

366 BEAULIEU ET AL.

MAGNETIC FIELD (T) 0.001 0.01 0.1 1 - I ' , 1 3 1 I " ' 1 8 ) ' I !

7 60 v s Myoglobin

0

W !- Q w z ' O e 2

: 20

2

5 Z

0 (L n

0 0.01 0.1 1 10 100

PROTON LARMOR FREQUENCY (MHz)

FIG. 2. The I / T , NMRD profiles of a solution of horse heart myoglobin, pH 7, at 5°C (solid symbols) and 25°C (open symbols). The protein content is 34% (v/v) , equal to that of the native calf lens homogenate, Fig. 1A. Note that there is no structure in the profiles comparable to the peaks of Figs. 1A and IB.

contributions. ) The value for Y, at 25°C -0.6 MHz, is rather low in that, by comparison with earlier work (2, 3, 8), it implies a protein mass of about 1 X lo5 Da, or oligomerization of the myoglobin at this high concentration to pentamers, on average. Nonetheless, the lack of structure in the I4N peak region of myoglobin solutions, compared with the same concentration of (heterogeneous) crystallins, Fig. 1 A, is remarkable.

Figure 3A shows data for homogenates of chicken lens, to be compared with the data of Fig. 1A. (These samples remain transparent at 5°C.) The concentration of crystallins in native chicken lens is relatively low (18%), much less than that of the calf lens sample (34%), so that the preparation had to be concentrated for the comparison. Chicken lens proteins, b-crystallins, are tetramers of 5 X lo4 Da monomers and are -70% a-helical, compared with almost total @-sheet for the majority fraction of calf lens proteins ( 12, p. 1 58 ). The magnitudes of the peaks in the calf lens homogenates, the comparably large peaks in the chicken lens sample, and the absence of peaks from myoglobin all suggest that backbone NH moieties are accessible to solvent in both the a-helical chicken crystallins and the @-sheet conformation of the calf crystallins, but not in myoglobin at comparable concentrations.

Figure 3B, analogous to Fig. 1 B, shows that the peaks for the concentrated chicken lens homogenate are about two-thirds the magnitude of those of native calf lens at 5'C and decrease to half that value upon heating to 25°C. The more dilute chicken lens (native) sample has no observable peak contribution.

Figure 4 shows the NMRD profiles of BSA at 5 and 25"C, 34% (v/v), the same concentration as the myoglobin and native calf lens samples. BSA is a rather elongated protein ( 17) composed of three in-line, roughly spherical regions. This tertiary structure tends to maximize the surface-to-volume ratio, and the net negative surface charge at pH 7.35, well above the isoelectric point, tends to expand these relatively flexible protein molecules by Coulomb forces ( 11 ). Both of these effects, quite the


MAGNETIC FIELD (T)

0.01 0.1 1 10

0.02 0.04 0.06 0.08

0 1 2 3 4

PROTON LARMOR FREQUENCY (MHz) PROTON LARMOR FREQUENCY (MHz)

FIG. 3. ( A ) The 1 / T , NMRD profiles of a homogenate of native chicken lens (squares) and an aliquot concentrated by dehydration (circles) to increase the protein content, at S'C (solid symbols) and 25°C (open symbols). The protein contents of the samples were 18 and 30% (v/v), respectively. There is no phase separation cataract in chicken lens homogenates. ( B ) The structured region of ( A ) shown on an expanded scale, after subtraction of the Cole-Cole fit to the monotonically decreasing background (see text).

opposite situation from myoglobin, tend to increase exposure of BSA residues to solvent; it is perhaps not surprising that I4N peaks are seen to appear in the highly concentrated BSA sample (although they could not be identified in an earlier search ( 6 ) in a more dilute solution). In addition, the monotonic part of the BSA profiles, which inflect about a decade lower in field than the myoglobin sample, indicates a protein weight oforder lo6 Da, much higher than the monomeric weight of6.5 X lo4

MAGNETIC FIELD (T) 0.001

v

w I-

d 60 z I- 3 40

w K

z 20 0 I-

e

S

z L n

0.01 0.1 1 10 100


FIG. 4. The 1 / T , N M R D profiles of a solution of bovine serum albumin (BSA), pH 7, at 5°C (solid symbols) and 25°C (open symbols). The protein content is 34% (v/v), equal lo that ofthe native calf lens hornogenate, Fig. 1A. Note that there is structure in the profiles ofthe sort seen in Figs. 1A and 1 B.

368 BEAULIEU ET AL.

Da, and therefore oligomerization of the BSA. (The geometric anisotropy of the BSA molecule does not change this estimate substantively (2, 18) .)

Figure 5 shows the NMRD profiles of raw and heat-denatured hen egg white at physiological concentration ( - 1 1%, v/v) and 25"C, the intent being to demonstrate the effect of a drastic alteration of tertiary protein structure on the profiles. The profiles clearly change progressively from those characteristic of a solution of globular protein, with weight of -7 X lo4 Da if assumed spherical ( 3 ) (and with the appropri- ate temperature dependence between 5 and 25"C, not shown), to that of a polymeric, extended structure upon denaturation. This conclusion follows from the behavior of the monotonic part of the profiles. In addition, the appearance of I4N peaks concur- rently with the denaturation indicates that heating exposes NH groups of the protein backbone and makes them accessible to solvent. Interestingly, the variation of the amplitudes of the peaks over the range 5 to 35°C (not shown) is comparable in magnitude but opposite in sign to that of the chicken lens (Fig. 3B) .

DISCUSSION

Relaxation of the magnetization of an ensemble of solvent protons in solutions of macromolecules is possible only because each proton experiences a local magnetic field that fluctuates in time (in a manner that, on average, is the same for all solvent protons). Those Fourier components of the fluctuations at frequencies that coincide with allowed "spin-flip'' transitions of the protons contribute to relaxation to an extent that, in liquids, is well understood ( 5 , 19, 20) and (usually) readily calculated. In diamagnetic systems, the fluctuating field arises from two sources: that produced by other solvent protons (mainly by the neighboring proton on the same water molecule) and which fluctuates due to thermal motion of the water; and the fields generated by solute nuclei, including protons and I4N, which fluctuate because of

MAGNETIC FIELD (T) 0.001

V Hen Egg Whi te 25 "C w

I- 4 CL 12

0.001

V Hen Egg Whi te 25 "C w

I- 4 CL 12


FIG. 5. The 1 /TI NMRD profiles of hen egg white, at 25"C, initially raw (0). then heat-denatured for about 1 min near 60°C (A) , and, finally, heated for another minute near 80°C (V). The transition from a profile with an inflection and no I4N peaks, for the globular form, to a profile with a qualitatively different monotonic background and with clearly visible I4N peaks, for the heatdenatured protein, is quite dra- matic. (R. N. Muller reported similar results in San Miniato, 1987 (unpublished).)


relative translational Brownian motion as well as possible solute-solvent complex formation. We will call these “self-relaxation” and “cross-relaxation” processes.

Self-relaxation in protein solutions has two components, both involving modula- tion of the random thermal motion of the solvent water molecules by the solute protein (2, 3, 8): one arises from the rotational, relaxational (dissipative), Brownian motion of globular solute molecules (which dominates relaxation processes below - 10 MHz); the other is related to alterations of the motion of water molecules as they collide with the solute surface (which typically requires - 10-los, during which time the rotational motion of the protein can be ignored). Cross-relaxation is also dominated by two components: interactions of solvent protons with protein protons (which has been considered in some detail (21 ); it roughly mimics the rotational self- relaxation contribution and can be included in it), and energy exchange with level splittings of I4N nuclei of solute NH groups (which arise from interactions of the nuclear quadrupole moment of I4N with local electric field gradients; these splittings are little influenced by the static magnetic field (6, 7, 9 ) ) .

For solutions of compact globular proteins at moderate concentrations, it is well established that v,-which, for protons, characterizes a dispersive component of the NMRD profile with both self- and cross-relaxation contributions-relates directly to the rotational motion of solute protein molecules. Its dependence on temperature, solvent viscosity, and molecular shape (2 , 3 ) verifies this quantitatively, even though the details of the molecular interactions that produce these contributions are a subject of some debate (3,22). Less is known about solutions containing high concentrations of macromolecules. We find here that the monotonic behavior observed for the native calf lens homogenate, Fig. 1 A, and the uppermost data in Figs. 3A and 5 resembles that found for polymerized sickle hemoglobin ( 15, 23), various polysaccharide and polyvinyl alcohol gels (unpublished), cellulose suspensions (unpublished), and diamagnetic tissues ( 4 , 5 ) . (There are few data available that show the extent of any structured region arising from cross-relaxation with I4N nuclei for extended macro- molecular conformations.) This class of profiles generally does not inflect above 0.0 1 MHz, the low field limit of the present instrumentation, and tends to have significantly less temperature dependence than solutions of globular proteins (for reasons that remain to be understood).

From the monotonic components of the profiles of native and diluted calf lens homogenate in Fig. IA, we suggest that there is a shift from polymer-like to globular protein-like NMRD profiles upon dilution. Calf lens homogenates are a mixture of crystallins that differ widely in molecular weight: 20% y-crystallins of -20 kDa; and the rest a mixture of a (>5 X lo5 Da)- and P-crystallins (( 5-20) X lo4 Da). It is not clear from the present data whether the associated states in the native preparation result in the main from a--a! or a-P interactions, on the one hand, or from interactions of the minority y component with the much larger a and p majority fractions, on the other; this question can be addressed only by using samples of separated, homogeneous solutions. Nonetheless, the theory of the I4N peaks ( 9 ) is sufficiently quantitative to argue effectively that solvent water must have access to the majority of solute NH moieties at high concentrations of calf and chick lens homogenates, with exchange rates b lo4 s-l .

Our earlier discussions of the 14N peaks in the NMRD profiles of calf lens homogenates (1, 9 ) presented a model that incorporated the extensive &sheet structure in

370 BEAULIEU ET AL.

the proteins and could account for the magnitudes of the peaks. On this view, one could picture extended planar sheet structures, comprising either one or two layers, with exchanging waters hydrogen bonded to the same carboxylate oxygens as those linked to NH protons (nonexchanging on our NMR time scale), thereby stabilizing the sheet structure. Individual sheets were regarded as loosely associated so that solvent had ready access to the majority of the NH locations. This model, though by no means established, follows naturally from the known repeated “Greek key” motif of the y-crystallins: the intramolecular multisheet structure so evident in the solid state (24 ) might well reorganize at high solute protein concentration when intermolecular interactions could predominate in solution. Assuming an extensive P-sheet structure, then, was sufficient to form the basis of a working model to explain the peaks. How- ever, since the data for chicken lens homogenates show similar behavior, it is clear that extensive @-structure is not a necessity; the sample of concentrated chicken crystallin also contains solute protein organized in a manner that allows access of solvent to the majority of the protein NH moieties despite the preponderance of a-helix in their secondary structure. Access of solvent and increasing association at high concentration must then follow as for the calf lens homogenates.

For the compact, conformationally stable, globular protein myoglobin (which, at the high concentration used here, appears oligomeric) there is no hint of I4N peaks. We attribute this to the majority of the backbone NH moieties being inaccessible to solvent; an alternative-conceivable, but hard to model-would be broadening of the lines (by some unknown mechanism) to the point of obscurity. The behavior of BSA, with a less compact, less globular, and more flexible tertiary structure, is quite different from that of myoglobin-but consistent with the framework that is being established here.

The behavior of hen egg white, Fig. 5 , a system in which the protein conformation can be altered drastically (but irreversibly) at fixed protein concentration, is intrigu- ing. The change in its NMRD profiles upon heating demonstrates both aspects of protein reorganization discussed here: the change in the functional form of the monotonic background and the change in the I4N peaks of a magnitude that indicates access of solvent to the majority of the NH moieties. The data, which resemble both calf and chicken lens homogenates at the higher concentrations, are consistent with a relatively open network of protein in the heat-denatured, polymerized, hen egg white (at physiological concentrations).

It remains to address the temperature dependence of the NMRD profiles in somewhat more detail, first to ascertain whether it is consistent with the mechanisms pro- posed here and, second, to determine whether these data contain new information. For nonassociating globular proteins, the low field limiting magnitude of the dispersive component of the NMRD profiles varies as q / T which, for the limited range of absolute temperature studied here, is dominated by the temperature dependence of the viscosity of water. In addition, correlated with the increase of this amplitude is a comparable decrease of v,. This behavior should also hold for associating proteins as long as the extent of association changes but slightly over the temperature range studied. Otherwise, the NMRD profiles will be a superposition of the contributions of each oligomer and will deviate from the nonassociating case with a sign that depends upon whether the interaction is dominated by energy or entropy. As already noted, the profiles of diluted calf lens and native chicken lens homogenates behave approxi-

NMRD OF LENS HOMOGENATES, I1 37 1

mately as solutions of fixed-mass globular proteins between 5 and 25"C, whereas the oligomeric myoglobin (Fig. 2) varies somewhat more; these variations are part of the argument establishing the compact globular nature of these solutions. By contrast, the temperature (and concentration) dependence of the monotonic component of profiles of the most concentrated lens homogenates, as well as that of egg white (not illustrated), is significantly less. However, the interaction mechanism that generates these profiles is not yet understood.

The temperature dependence of the peaks, explored here for the first time, has its own intrinsic interest. For the calf lens nuclear homogenate, Fig. IB, the 5 and 25°C NH contributions are essentially identical. From the theory of the peaks (9), there are three temperature-dependent parameters that should be considered and which may well compensate in this system, a not unreasonable view given that the chicken crystallin and denatured egg-white samples, as noted, have opposite dependences on temperature. One parameter is the lifetime of a water molecule, hydrogen bonded and in rapid exchange, on a carboxylate near an NH group. This lifetime is the correlation time for the cross-relaxation interaction that causes the peaks; higher tempera- tures should reduce the lifetime and reduce the height of the peaks, as observed for the a-helical concentrated chicken lens homogenate. The second parameter is the intrinsic linewidth of the l4N quadrupolar levels (the I4N 1 / T 2 ) since this determines the rate at which magnetic energy is transferred from the (nonexchanging) NH proton to the I4N nucleus (which is part of the relaxation pathway ( 9 ) ) . However, since little is known about I4N relaxation in macromolecules and, in particular, whether the field range of 0.6 to 3 MHz is above or below the correlation time for this interaction, one cannot even be certain of the sign of the temperature dependence of this contribution. A third parameter is the number of accessible NH moieties, which in turn relates to the temperature-dependent pattern of association of the crystallins at any given concentration. Whether the difference in the temperature dependence of the peaks in the calf and chicken lens homogenates is a correlate of the @-sheet vs a- helix structure of the different crystallins, however, will require further investigation.

Finally, one aspect of the model theory of the I4N peaks ( 9 ) as it applies to the present results bears reemphasis: peaks of the magnitude seen in Figs. 1B and 3B can be explained only if solvent molecules can approach essentially all the NH protons of the residues of the protein backbone within - 1.7 A and remain there for - s. The longer this time, the fewer the number of accessible residues needed; however, lo-'' s is already a relatively long time, given that the only mechanism available for establishing this configuration is hydrogen bonding, and a longer lifetime for such bonds is unlikely. This point has been discussed in depth recently (25) and need not be reiterated except to note that a longer lifetime should engender a temperature dependence greater than that observed here. Thus, one can say with considerable confidence that the presence of large I4N peaks sets considerable constraints, both geometric and dynamic, on the possible configurations of lens crystallins near physiological concentrations.

SUMMARY

We have extended our earlier work on the NMRD profiles of calf lens homogenates at 25°C ( I ) to 5"C, and to other protein systems as well. The results, now for both calf and chicken lens homogenates, are similar: for homogenates of lens tissue (which

372 BEAULIEU ET AL.

are essentially heterogeneous protein solutions), we infer that a reversible conformational reorganization of the solute protein takes place, near 20% (v/v) protein, that involves a transition of the tertiary structure from globular molecules (at low concentrations) to an extended, highly open, network of protein chains. By contrast, solutions of myoglobin over the same range of concentration show no hint of such behavior, whereas heat denaturation of hen egg white mimics much of the behavior (irreversibly) of lens protein. The results give added support to the conclusions and conjectures in our earlier work ( 1 ). More about the details of the basic interactions that underly the unusual behavior of lens protein, and its relevance of both lens function and disease, will require similar measurements on better-defined, homogeneous, systems.

ACKNOWLEDGMENTS

This work was supported in part by NEI Grant EY-04542 (J.I.C.), and an MSTP scholarship GM-07266 and award from the CM Poncin Charitable Trust Fund (C.F.B.).

REFERENCES

1. C. F. BEAULIEU, J. 1. CLARK, R. D. BROWN 111, M. SPILLER, AND S. H. KOENIG, Mugn. Reson. Med.

2. S . H. KOENIG AND W. E. SCHILLINGER, J. Biol. Chem. 244,3283 (1969). 3. K. HALLENGA AND S. H. KOENIG, Biochemistry 15,4255 (I 976). 4. S. H. KOENIG, R. D. BROWN 111, D. ADAMS, D. EMERSON, AND G. C. HARRISON, Invest. Rudiol. 19,

5. S. H. KOENIG AND R. D. BROWN Ill, Invest. Radiol. 20,270 (1985). 6. F. WINTER AND R. KIMMICH, Biochim. Biophys. AcIu 719,292 (1982). 7. F. WINTER AND R. KIMMICH, Mol. fhys. 4,33 (1982). 8. S. H. KOENIG, “Water in Polymers” (S. P. Rowland, Ed.), p. 157, ACS Symposium Series, No. 127,

9. S. H. KOENIG, Biophys. J. 53,91 (1988).

8,45(1988).

76(1984).

Amer. Chem. SOC., Washington, DC, 1980.

10. T. TANAKAANDG. BENEDEK, Invest. Ophthalmof. 14,449 (1975). I I. C. TANFORD, “Physical Chemistry of Macromolecules,” p. 5 18, Wiley, New York, 196 I . 12. cf. P. F. LINDLEY, M. E. NAREBOR, L. J. SUMMERS, AND G. J. WISTOW, in “The Ocular Lens: Struc-

13. T. TANAKA, C. ISHIMOTO, AND L. T. CHYLACK, JR., Science 197,1010 (1977). 14. G. B. BENEDEK, J. I. CLARK, E. N. SERRALLACH, C. Y. YOUNG, L. MENGEL, T. SAUKE, A. BAGG,

AND K. BENEDEK, fhilos. Trans. R. SOC. London A 293,119 (1979). 15. cf. S. H. KOENIG AND R. D. BROWN 111, in “NMR Spectroscopy of Cells and Organisms’’ (R. K.

Gupta, Ed.), Vol. 11, p. 75, CRC Press, Boca Raton, FL, 1987, for a recent review and extensive references.

ture, Function, and Pathology” (H. Maisel, Ed.), p. 123, Dekker, New York, 1985.

16. S. H. KOENIG, T. R. LINDSTROM, AND R. D. BROWN 111, Biophys. J. 34,397 (198 I ) . 17. T. PETERS, JR., in “Advances in Protein Chemistry,” Vol. 37, p. 161, Academic Press, New York,

18. S. H. KOENIG, Biopolymers 14,2421 (1975). 19. I.So~o~~~,fhys. Rev.99,559(1955). 20. N. BLOEMBERGEN, E. N. FURCELL, AND R. V. POUND, Phys. Rev. 73,697 (1948). 21. S. H. KOENIG, R. G. BRYANT, K. HALLENGA, AND R. D. BROWN 111, Biochemistry 217,298 (1982). 22. cf. B. HALLEANDH. WENNERSTROM, J. Chem.Phys. 75,1928(1981). 23. T. R. LINDSTROMANDS. H. KOENIG, J. Mugn. Reson. 15,344(1974). 24. G. WISTOW, B. TURNELL, L. SUMMERS, C. SLINGSBY, D. Moss, L. MILLER, P. LINDLEY, AND T.

25. H. F. BENNETT, R. D. BROWN 111, S. H. KOENIG, AND H. M. SWARTZ, Mugn. Reson. Med. 4, 93

1985.

BLUNDELL, J. Mol. Biol. 170, 175 (1983).

(1987 ).

Documents

Relaxometry of lens homogenates. II. temperature dependence and comparison with other proteins