DigitalCommons@University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln A Circular Dichroic Study of Cu (II) -Ribonuclease Complexes A Circular Dichroic Study of Cu (II) -Ribonuclease Complexes

SUMMARY The visible and ultraviolet circular dichroic (CD) spectra resulting from the interaction of ribonuclease with successive Cu(I1) ions have been recorded under a variety of conditions. At pH 7 in the presence of 0.16 M KC1 a broad, negative band was found in the visible region. This band increased in intensity and changed in shape as successive coppers were added. The circular dichroic spectra could be analyzed in terms of two kinds of binding sites: a single strong site with CD minimum at about 710 nm, and four weaker sites with CD minimum at about 600 nm. The binding constants observed are close to those obtained by more conventional means. Carboxymethylation of one histidine results in loss of one of the weaker sites. In 0.01 M salt, only the 600-nm band is seen. Binding at pH 9.6 differed in that saturation did not occur until about 33 sites had been filled. The presence of tetra coordination at this pH was indicated by the shift of the primary d-d transition down to 530 nm. Additional structure in the visible and near ultraviolet CD was now present in the form of a negative band at 355 nm and, for the first two Cu(II)‘s added, a positive one at 480 nm. Strong positive bands were observed

A large number of proteins are known to associate with metal ions. These complexes may be divided into metnlloproteins and metal-protein complexes (l-3).
In metalloproteins the metal is usually found with the protein in the native state, and often a biological function can be ascribed to it. The binding is frequently so strong as to appear irreversible by such procedures as dialysis.
On the other hand, the formation of metal-protein comnplexes involves both specific and nonspecific binding of one or more metal ions to a protein which does not normally require *  these ions for biological function. The binding constants are characteristically orders of magnitude smaller than those of metalloproteins.
Because the metal-binding sites in metalloproteins SO often constitute active or oxygen-binding sites, they have been frequently studied with regard to their specificity of attachment and functional importance.
To answer these questions, it is desirable to choose a physical tool which concentrates on those binding sites to t,he exclusion of the rest of the protein.
An alternative tool is circular dichroism.
Circular dichroism possesses a distinct advantage over absorption spectroscopy in that metal ions are optically inactive in solution but may become optically active when bound.
One is thus able to look specifically at the ions bound to a protein in the absence of any background contribution from free ions.
This laboratory has lxeviously used CIY measurements to study the binding of copper to hrmocyanins (5). These results were lxovocative in that a single visible absorption band was observed to be split into two and three CD bands for arthropod and molluscan hemocyanins, respectively. Furthermore, not only these bands, but CD bands in the ultraviolet (at 350 and 250 nm) were fouud to be sensitive to the degree of oxygenation. L'nfortunately, we were unable to interpret these data in terms of specific amino acid residues at the binding site. Thus, we are attempting to investigate cases in which the copper-binding site is already better characterized, in the hope that such studies will help us to understand the hemocyanin spectra.
HoI\-ever, each of these systems has disadvantages as a model for the systems of interest to US. A protein of known conformation is needed, with dell defined copper-binding sites and minimal absorption from other chromophores in the near-ultraviolet and visible regions. Ribonuclease seemed an excellent choice (9,10).
That copper does interact with ribonuclcase was initially shown by cupric ion inhibition of both enzymatic activity (11) and active site histidine carbosymethylation (12). Since then the copper-ribonuclease system has been characterized by gel filtration (9,13), equilibrium dialysis in the presence of a metal ion buffer (14, 15)) proton magnetic resonance (16)) s-ray crystallography (17), potentiomctric titration (18)) and chemical modification of the ribonuclease prior to binding (10,15,16,19). These various techniques indicate the presence of up to five binding s&es at pI1 values Oscar Ilcutrality.
It now seems clcal that at pII 7 a site containing in part the oc-amino group is filled first (lo), followed by four sites involving histidine imidazole nitrogens.
The previous uncertainty arose from the cstcnt to which individual site affinities arc a function of pI1. 12inding strength decreases with decreasing $1 since the cupric ions must compete with protons for the available protein sites, but the rate of this dccreasc varies with the type of site inrol\-cd. Iiy pTI 5.5 the ar-amino site is no longer the most avid. This was clearly shown by Girotti and l~reslow (10) through spcrific modification of the cr-amino group with nitrous acid.
The conclusion drawn is that ribonuclcasc provides firr initial points of attachment for copper, the Ilitrogcn atoms of the cu-amill h mroup and the four imidxzolc rings. This is followed by a varying amount of chelation to the pcptidc bac~kbonr Ilitrogcns. The cstcnt of chelation, and consequently the magnitude of the association constant, depends on the ability of the copprr to labilize adjoining peptide hydrogen atoms. This model of riboiillclcase-billding: sites is bascti on Sidgwick's observation (20) tllat copper prefers nitrogen lignnds to oxygen ligands, al~d it gains great support by analogy frorn Curd's studies ((i-8, 21-23) on small pcptidcs as sufficient models for the interaction of cupric ions with myoglobin and bovine serum albumin.
For example, the visible CD spectra of a 1 :1 complex of Cu(I1) with bovine serum albumin is reproduced almost exactly (6) by a 1: 1 coml)les of Cu(I1) &h aspartyl-threonyl-histidyl-lysinc, the S&tcrminal tetrapeptide of bovine serum albumin. Similarly, such simple peptides as acetyl-diglycyl-histidyl-plycine have been shown (22) to complex to (sopper such that they represent an adequate model with regard to titration behavior and visible absorption spectrum for the Cu(II)-sperm whale myoglobin complex.
While circular dichroism has been used as a tool in many of t.hc studies of the relevancy of small peptidcs as adequate and suficient models (6~8), no Cl) spectra have yet been reported for the Cu(II)-ribonuclease complex. WC have chosen to study it, just because it presents a succession of fairly well charactnized col)l)er-binding sites. In addition to providing data for coml)arison with the CD spectra of metalloproteins, these csperiments provide a test for the applicability of the Cl> method to quantitative binding studies.
I)ovine pancreatic ribonuclease A (code RASE, chromatographically pure, aggregate free) was purchased from Worthington Iliochemicals Inc. 9ny phosphates present were removed by dialyzing five times against distilled water and three times against 0.16 RI KCl. Union Carbide cellophane casing was used. Ribonuclease concentrations were routinely determined spcctrol)hotometrically, assuming 1 mg per ml to yield an absorbance at 280 nm of 0.695 (24). A stock solution of 20.7 mg per ml in 0.16 M KC1 was diluted as needed.
The carbosymethylated derivative of ribonucleasc was prepared by the method of Crestfield el al. (25). One hundred milligrams of phosphate-free ribonuclcase in 8 ml were brought to pI-I 5.5, at which point 1.2 ml of iodoacetic acid (10 mg per ml), already at pH 5.5, were added. After 9 hours at room temperat,urc in a p&stat, the reaction mixture was placed on a Rio-Rex 70 column (1 X 10 cm) (100 to 200 mesh) equilibrated with 0.1 s acetic acid. The column was washed with 60 ml of 0.1 N acetic acid and the ribonuclcase eluted with 40 ml 1.0 x sodium acetate. The eluate was dialyzed against 4 liters of distilled water, folloncd by 1 liter of 0.16 RZ KCl.
No attempt was made to separate ribonurlcase carbosymethylated at histidine-12 from t,hat at histidine-119, and the &inction cocficient was assumed to remain unaltcrcd.
To check that only the desired histidine csarbosymethylation reaction had occurred, a lyophilized sample was hydrolyzed for amino acid analysis. The results differed from those of authentic ribonuclcasc only by the presence of three rather than four hi&dines.
'l'hc basic tc&liquc by whic?h n-e followed Cu(II)-ribonuclease binding with circular diclhroism is as follows.
'l'hr plrl was adjusted after (,a& addition of CuCL by a Radiometer type 7"lTl p&stat filled with "Acculute" 0.2 I ITaO from Anachemia Chemicals Ltd. Depending on the ribonuclcasc concentration l)resent, 20. or IOO-~1 Eppendorf pipettes \vcre used to add the stoichiometric amounts of CuC12. Xftcr each addition and pH adjustment the sample was transferred to a IO-cm cylindrical cucettc and the circular dichroism spectrum recorded at room temperature on a Durrum-Jasco model ClMl' apparatus. This procedure was repeated as often as required.
A jacketed, constant tcrnpcrature cell was used occasionally to insure that room temperature variations over 22-28" were without effect. l'hc number of protons released due to binding was estimated from the amount of iTaO needed to restorc a given 111-1 after each aliquot of CuCIZ had been added. Our values were very similar to those observed by Breslon-and Girotti (18).
In all of the CJ) spectra reported hcrc the ordinate is the observed ellipticity, 0. IloxercAr, &en Cl) intensities are compared on a per copper basis t,he molecular cllipticity [0] in deg .cm2 per dmolc is used, where [0] = 0 AI/10 I c'. Here 0 is the obserl-ed cllipticity in degrees, 1 the path length in cm, c' the COW centration in g per cm3, and M the gram molecular weight of the protein.
It should bc noted that these arc not rcsiduc cllipticities.
There has been some evidence (9, 13) that copper binding may be accompanied by association of the ribonuclease to dimers or higher aggregates.
Accordingly, average molecular weights were determined by sedimentat'ion equilibrium after the addition of up to 8 copper eq. A Spinco model E analytical ultracentrifuge with ultraviolet absorption optics and scanner was used. All solutions for these experiments n-ere 1. mg per ml in 0.16 M KC1 at 1'1-1 7.

RESULTS
Binding at pll 7.0-The visible Cl> slxctra observed when successive equimolar inrrements of CuCls.211Z0 were added to ribonuclease in 0.16 hi KC1 arc shown in Fig. 1. A single broad band is observed; that it may be composite is suggested by the fact that the wave length (X,i,) of the Cl) minimum shifts with addition of more copper ( Fig. 1 and Fig. 2, Curve a). The increase in the negative CD value at a representative wave length (600 nm) is shown in Fig. 3. Initially, we thought that the leveling off observed at high Cu : RXAse ratios indicated that saturation had been reached. Xow, for reasons given below, we feel that this is not the case, and that the lower value observed at 9 eq is spurious.
The variation in Xlllin with addition of copper ( Fig. 2) suggests that w-e are in fact observing a spectrum with contribution from more than one CD band.
This might be expected from earlier studies (10) which indicated that under these conditions copper binds to two kinds of sites: an NHz-terminal site with a binding constant of 5.0 X lo5 (moles per liter)-I, and four histidine sites with roughly equal constants of about 1.9 X lo4 (moles per liter)-'.
The wave length variation in Curve B of Fig. 2 suggests that the st,ronger site must give rise to a CD band centered near 700 nm and the weaker site to a band centered below 620 nm. Accordingly, we attempted to resolve the observed CD spectra, using a DuPont curve resolver, into two such bands. It was found that all of the spectra shown in Fig. 1 would be accurately fitted by combination of two bands of fixed wave length and breadth but of varying intensity.
One of these is centered at 710 nm and the other at 600 nm.
The band at 710 nm reaches saturation at quite low Cu: RNAse ratios (Fig. 4). Therefore, we have identified it as corresponding to the strong, NH*-terminal binding site. The 600~nm band, which ultimately becomes much more intense, has been assigned to the four histidine sites. We then ask whether the data are consistent with the binding constants obtained by Girotbi and Breslow (10). We have approached this question in the following ways. Given the binding constants and num-  bers of groups involved, one may calculate, for a given value of e = vlel + v2ez (5)  This allows construction (Fig. 4) of graphs of v1 and v2 versus Cu"/Po, where Cue is the total copper concentration.
Comparison of the data with the theoretical curves (Fig. 4) indicates that the bands at 710 and 600 are measuring binding by the two kinds of sites in good agreement with the results of Girotti and Breslow.
A more direct test is possible. If one rewrites Equat'ion 2 in terms of the total copper concentration (CuO) one obtains: corresponding ratio used to define & in Equation 4. While this method of fitting the data leaves much to be desired, it is supported by a comparable analysis of the results for carboxymethylated ribonuclease (Fig. 3). The curve drawn through these points utilized the same parameters 19~ and &, the same binding constants, but the assumption that one strong site and only three histidine sites were present.
The agreement is moderately good.
Taken together, these results strongly support the contention of Girotti and Breslow that in 0.16 M KCI, pH 7.0, ribonuclease possesses one strong site and four weaker sites for copper bindnmcuo -bl + YZP) (3) ing. Our results go somewhat further, however, in t.hat the lin-y2 = 1 + K2(Cu0 -(Y, + YP)PO) ear relation observed in Fig. 5 indicates that the weaker sites are, in fact, nearly equivalent and that they do not. exhibit strong If a single CD band can be attributed to the type 2 sites, we cooperativity. Extreme non-equivalence or cooperativity should may define the "reduced" dichroism of this band (0,) as the lead to non-linearity in such a graph. While strict equivalency ratio of the observed dichroism to the saturating value. Then is not to be edxpected, in view of the different environments of v2 = no,, and Equation 3 can be rearranged to: the histidines, whatever differences exist are not, detectable by 0, cue VI 1 -8, nK2po[--er-- However, as can be seen from Fig. 4, the first site, by our measurements, behaves approximately as expected from the data of Girotti and Breslow. The data shown in Fig. 3 can also be interpreted in the same manner.
The curve drawn through the points is predicted from the values of VI, v2 calculated from Girotti and Breslow's data. . In the calculation, ~1 was taken from our data in Fig. 4. See the text. our methods.
Part of the reason the identity of the single, strong site has created so much controversy (9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)) is that it is not uniformly the most avid site. This is known to be true if the pH is lowered to 5.5 (10, 16). Fig. 6 shows that it also appears to be the case at pH 7 if the ionic strength is lowered from 0.16 to 0.01. Under these conditions, the wave length of maximum ellipticity is constant, and at a bit below 600 nm (Fig. 2s). If we assign the 710.nm CD band to the first site occupied at high salt, we are forced to conclude that either (a) that site is no longer active at low ionic strength, possibly because of electrostatic repulsion between Cuff and the doubly charged KHZ-terminal lysine; or (b) the nature of the binding at the NHz-terminal site has been modified such that it too now has a 600-nm band.
Unfortunately, our data in 0.01 M salt are not complete enough to yield an unambiguous answer to this question. The data in Fig. 6 have been extended to cover the range from 300 to 800 nm. At the low wave length end we find the positive band hinted at in Fig. 1 more clearly expressed. It too must be a copper band, for ribonuclease alone has no absorption or CD in this region.
The spectra were extended to 800 nm here to assure that we were looking at all of the Cotton effects resulting from copper d-d transitions.
If copper is present in its preferred, Jahn-Teller distorted, square planar configuration, it is supposed (3) to give a maximum of three transitions.
However, four can be observed under less symmetrical conditions. Yasui et al. (26) report four d-d Cotton effects for complexes of Cu(L-serine)2, with the last one at 830 nm. We have no clear indication of any further bands in this region. Fig. 7 shows data obtained at pH 7.0 when 0.05 M Tris buffer was used to stabilize the pH. Since Tris is also known to bind Cu(I1) (27), the concentration of free Cu(I1) is kept at levels very much lower than would be indicated by the equivalents of CuC12 added.
Thus, only the most avid sites would be expected to be filled.
The data of Fig. 7 are also graphed in Figs. 2 and 3. A glance at Fig. 3 suggests that only the single, strongest site is being occupied at the Cu(I1) concentrations achieved in Tris.
However, Fig. 2 shows that the situation is more complex.
Here too a shift in the wave length of CD minimum occurs, and the resultant band position indicates that when 7 eq of Cu(I1) have been added, the 710-nm site and at least one of the 600~nm sites are about equally occupied. We believe this situation represents an intermediate stage between that found at 0.16 and 0.01 ionic strength.
The 710-nm site is still the strongest, but the margin by which its affinity is greater than the 600-nm sites has decreased. Thus, with decreasing ionic strength, the 710-nm site binds more weakly.
Finally, we must raise the question whether changes in the conformation or state of aggregation of ribonuclease, or both, occur as a consequence of Cu(I1) binding at pI1 7. Fig. 8 presents some data relevant to the first question.
It is evident that no appreciable change in CD over the range from 215 to 240 nm is caused by the addition of up to 6% eq of Cu(I1). This may be taken as presumptive evidence that the secondary The short horizontd lines result from the recorder pen sticking.
The number of equivalents added is given beside each curve. Ellipticity in millidegrees.
for a change in CD due to the aromatic amino acids as a consequence of Cu(I1) complexing.
However, small changes would be difficult to detect.
From gel filtration studies, other workers (9, 13) have suggested that ribonuclease undergoes association on Cu(I1) binding. We have carried out a few sedimentation equilibrium experiments at pH 7.0 and 0.16 M KCl.
Although association obviously does occur, our scanner data are not especially precise and we do not feel they warrant a detailed analysis.
The log c versus r2 graphs are presented in Fig. 9. They clearly show that until more than 2 eq of Cu(I1) have been added the material remains homogeneous, with a molecular weight very close to that expected for monomeric ribonuclease. The solid line passing through the 0-Cu control of Fig. 9 corresponds to a molecular weight of 13,683. Some aggregated material is evident in the samples to which 5 and 8 eq of Cu had been added. Their limiting slopes do correspond approximately to that expected for dimer (broken line), but the solutions are heterogeneous. This might be explained most simply by assuming that # .* the association is reversible, although that point is certainly not proven by our data. Data were taken with the ultraviolet scanner at wave lengths near 280 nm. All data are at 20,000 rpm and 22". The set of points have been arbitrarily displaced on the log c axis. The abscissa gives the square of the distance from the meniscus. The solid line through the 0-Cu set of points is that predict,ed for molecular weight 13,683. The broken line is of the slope predicted for a ribonuclease dimer .
The most interesting point is that absolutely no association is seen until more than two coppers have been bound.
It is reasonable to expect the NHS-terminal site, which is filled preferentially, not to be involved in association since the lysine can form an intramolecular chelate.
Occupation of one or more of the histidine sites must then confer the ability to dimerize.
In connection with this, it should be noted that hiitidine-105 is known to be a surface site (28) and might therefore be involved in dimerization. structure is unchanged.
The situation in the 240-to 300-nm region is more complex.
Changes are evident, but we feel they can be largely accounted for by the emergence of two positive, charge-transfer bands involving copper.
The one at 305 nm has already been mentioned, and the one at 251 nm will be discussed in detail later.
With these provisos, there is no definite evidence Binding at pH 9.6 and 5.5-It is evident from Figs. 10 and 11 that the CD spectra observed when Cu(I1) is bound to ribonuclease at pH 9.6 are more complex than those found at pH 7.0. The positive band at 320 nm seems invariant (note that it is truncated by the negative aromatic bands and is probably centered at about 305 nm), whereas the strong, negative band in the visible has been shifted down to a much lower wave length. The pH 9.6 CD spectra also reveal a new negative band at 355 nm and, for the first few Cu(IJ)'s bound, a weak, positive band around 470 nm. The interpretation of these spectra is aided by comparison with the high pH CD spectra of model peptides (6)(7)(8).
The CD spectra of some model Cu(II)tet'rapeptide complexes that contain histidine (6, 7) closely resemble the curVes shown in Fig. 10 for the first few coppers added in that they both have a positive band at 470 to 500 nm and a negative one at 560 to 600 nm. This indicates that a portion of the first copper ions added at pH 9.6 go into such histidine-containing sites. When the data of Fig. 10 are plotted in Fig. 2, it is seen that t,he wave length of the CD minimum remains relatively constant at 570 nm for the first four Cu(II)s added before falling off. When Cu(I1) is tetraliganded, in nonhistidine-containing model peptides (by three peptide nitrogens and an ar-NHB), a single CD minimum in the 515 to 560 nm range is observed (6)(7)(8).
Thus, both the NHS-terminal site and sites involving four peptide nitrogens should be expected to absorb in this region at high pII.
Figs. 2, 10, and 11 show that after the four presumed histidine-containing sites on ribonuclease have been filled, the addition of further Cu(II)s causes a shift toward wave length minima characteristic of such totally peptide liganding.
The enhanced lability of the peptide protons at pH 9.6 makes many more sites now available.
Just how many can be seen in Figs. 11 and 12. The increase in CD intensity with added Cu(II)'s saturates very sharply at about 33 f 3 copper ions added.
Because of errors which are bound to accumulate with so many transfers and pH adjust,ments, the exact number is somewhat imprecise, but it is clear that after a certain point the addition of 10 more copper eq produces zero further CD change. This limit of 33 copper-bindin g sites corresponds very nicely to that expected if almost all of the peptide nitrogens were involved.
Kibonuclease contains 124 peptide backbone nitrogens, In all cases they are pH 9.6, no buffer, 0.16 M KCl.
four imidazole rings, and 10 lysine t-amino groups.
Together these would give about 34 tetradentate-binding sites. The possible participation of the six tyrosine and nine carboxyl side chains might provide about four more. In any event, it is clear that the protein must unfold completely to provide so many sites. In accordance with this requirement that the native secondary structure must be lost, we have found that in this case the 215 to 240 nm CD is drastically decreased, as expected.
We may expect that this phenomenon is general for proteins since it is in essence the biuret reaction.
Whereas the visible CD band was used in Figs. 11 and 12 to determine that pH 9.6 saturation occurred at, 33 Cu(II)-binding sites, it should be pointed out that other bands of interest are present also. Fig. 13 shows a scan at $1 9.6 from 240 up to 750 nm in the presence of saturating amounts of Cu(I1).
The band at 251 nm clearly dominates the spectrum.
Hartzell and Gurd (8) have used CD to study the copper conlplexes of nonaromatic pentapeptides. They consistently observed both a positive band at 305 nm, which corresponds nicely to that we have found truncated at 320 nm, as well as a 275-nm band.
They att,ributed these bands to charge-transfer complexes between copper and the deprotonated peptide backbone. We have no indication of a 275.nm band, but we believe our dominant 251.nm CD band represents a comparable charge-transfer band.
It is about six times as intense as the visible copper CD band at 525 nm. Although there is no exact correlation between the intensities of CD and absorption bands, this agrees rather well with the generalization (I, 3) that charge-transfer bands are about 10 times more intense than d-d transitions.
The 251nm band must be characteristic of a great variety of binding sites. This is a reasonable expect,ation for the deprotonated peptide backbone.
The posit,ive CD bands at 251 and 305 to 320 nm are unique among those we have observed in that their position is not sensitive to pH, showing very little change between pH 7.0 and 9.6. This insensitivity of X to changes in environment above pH 7 was also noted for the 305-nm band by Hartzell and Gurd (8). Their use of peptides containing no aromatic residues allowed them to study the latter band at' 305 nm instead of truncated at 320 nm. It also completely eliminates the possibility that the 305.nm band is derived from an ionized tyrosine. In our experiments, the position and shape of the 251. and 320.nm 13. Visible and ultraviolet CD spectrum of ribonuclease at 1 mg per ml with 37 eq of Cu(I1) added. One-millimeter path length, pH 9.6, no buffer, 0.16 M KCl. Ellipticity in millidegrees.
bands were invariant, with numbers of copper added at pH values of 7.0 or 9.6. This means that tetra coordination is not a prerequisite.
Here the charge-transfer bands differ from both the positive 470-nm histidine band and the negative 355-nm band, both of which are only present at pH 9.6 ( Fig. 10). In these cases tetra coordination appears to be necessary. The 355-nm band is very close to known bands in hemocyanin (5) and other copper-containing metalloproteins (3) that have often been attributed (3) to copper dimer formation. This is thought to be in a manner similar to that found in many polynuclear copper complexes.

cyanins.
Metalloproteins are known to differ from metalprotein complexes in that they are characterized by tight binding (Kdiss is often 5 lo-$ M (l-3)) and large visible extinction coefficients; that for octopus oxyhemocyanin is 500 liters per g.atom copper cm at 570 nm (29). In contrast, the Cu(II)ribonuclease system appears to be typical of metal-protein complexes in that (at neutral pH) its values are only about lo-* and 85, respectively (18). Both affinity for copper and the resultant visible extinction coefficient are known to increase as the solution becomes more alkaline, but even at pH 11, when maximum coordination has been achieved, the visible absorption is still only one-third that of a true copper-containing metalloprotein such as oxyhemocyanin. These differences become even larger when we consider that for a copper-containing metalloprotein, oxyhemocyanin's visible absorption is weak enough for it to have been classified as a "non-blue" copper protein (3). Certain of the "blue" copper proteins absorb 10 times more intensely, again at neutral pH.
In conclusion, we will make brief mention of our attempted studies on copper binding at pH 5.5. We have not been able to compare our results with those of others, either in crystal (17) or solution (9, 10, 13-16, 18) since we get virtually no CD ellipticity at this pH. This is unfortunate since so much of the previous work was concentrated here, but it is to be expected since any binding that does occur at pH 5.5 should be unidentate. The copper is not yet able to labilize an adjoining peptide backbone proton.
No protein chelate is formed. We are measuring optical activity, and only when the added copper forms a chelate, as it does at pH 7 or above, do we get a strong signal. The weak ellipticity we do observe at pH 5.5 is broadly centered above 700 nm, which would agree with the suggestion (10, 16) that copper is bound to a single imidazole.
This situation is almost ident.ical with that observed for pH 5.5 complexes of Cu(I1) with model peptides (8). Our data are essentially negative.
If CD were the only criterion of binding, we would have little indication that binding had actually occurred.
Similarly, in the aromatic (240 to 300 nm) and peptide (215 to 240 nm) regions no differences at all result from the addition of 6% eq of Cu(I1). This is further indication that low pH binding must be unidentate.
The strong 251~nm band resulting from copper binding to the peptide backbone is totally absent. DISCUSSION We are now in a position to compare our CD results on ribonuclease with those previously obtained (5) for the oxyhemo-In this light, it is not too surprising to find that whereas visible CD intensities for oxyhemocyanins range from 2300 to 3000 deg.cm2 per dmole (5), that for the first copper added to ribonuclease at neutral pH (Fig. 3) is only about 1000 deg.cm2 per dmole.
On a per copper basis these values become more comparable at pH 9.6, by which time the Cu(II)-ribonuclease ellipticity has risen to about 2000 deg. cm2 per dmole. These values approach those of true copper metalloproteins more nearly than do the extinction coefficients referred to earlier, but they are still decidedly less.
What is remarkable, however, is the definite qualitative resemblance between the alkaline CD spectrum of Cu(II)ribonuclease and that exhibited by the oxyhemocyanins at both neutral and alkaline pH. The following similarities should be noted.
1. In both cases visible CD bands arc observed at 450 to 500 nm and between 500 and 600 nm. In particular, the hemocyanin CD spectra resemble those of histidine-containing peptides and ribonuclease at low levels of added copper.
2. A negative band around 350 nm is common to both, al-at UNIV OF NEBRASKA -Lincoln on September 11, 2007 www.jbc.org Downloaded from though its relative intensity is much grcatcr in the osyhemocyanins. 3. Both systems exhibit a strong, positive Cl1 band near 251 nm. 4. A weak, negative band at 305 to 310 nm has been observed in dcosygenated samples of hemocyanins (30). Such a band would be hard to detect in osyhemocyanin, because of overlap from the illtense, negative 3%run band. ;1 suggestion of a slioulder near 305 nm is seen in some spectra.
~Uthough the above similarities arc of inter&, we must also emphasize some significant differences bctneen the two systems.
1. Cu(II)-ribonuclease does not require 02 to exhibit visible and irear-ultraviolet CD and absorption bands. 2. Some of the CD bands already shown to be in the same position are in reality of different sign. In arthropod hemocyanins the 550-to 600.nm band is positive, unlike both the molluscan hemocyanins and Cu(II)-ribonucleasc.
The hemocyanin 305 to 310-nm bands have always been found to be negative.
3. XIolluscan hemocyanins all show an additional positive CD band above 700 nm.
In view of the nature of the similarities, it would seem reasonable to suggest that the copper-binding site in hemocyanins is tetradentatc, involving at least 1 histidine residue. We may further infer from the common appearance of presumed chargetransfer bands involving the peptidc backbone that at least a portion of the remaining ligands is peptidc nitrogens. One unique aspect of these metalloprotein metal-binding sites is then seen to be a conformation that allows sufficient peptide proton labilization to achieve tetradentate binding at neutral pf1.
ilcknowledgrrzent-We thank Robert Howard for performing the amino acid analyses.