Quantitative Determination of Carbamino Adducts of a and p Chains in Human Adult Hemoglobin in Presence and Absence of Carbon Monoxide and 2,3=l%phosphoglycerate*

The principal component of normal adult human hemoglobin was equilibrated under various conditions with ‘WO,. Quantitative analysis of the carbamino resonance intensi-ties over the pH range of 6.5 to 9.0 shows that the effects of conversion from the deoxy to the liganded state in reducing the carbamino adduct formation occur predominantly at Val-18. Analysis of the pH dependence of carbamino formation at constant total carbonates yields values of pK, and pK, for Val-lb and Val-1cY in the deoxy and liganded conditions. In contrast to the Val-l/3 role as the allosteric site for COz, the Val-la site is shown to be primarily an alkaline Bohr group. 2,3-Diphosphoglycerate is shown to reduce sub-stantially the Val-l/3 carbamino resonance intensity in deoxyhemoglobin. Evidence for 2,3-diphosphoglycerate effects in carbon monoxide hemoglobin at both Val-la and Val-lj3 sites is presented. Enhanced carbamino formation in

The principal component of normal adult human hemoglobin was equilibrated under various conditions with 'WO,. Quantitative analysis of the carbamino resonance intensities over the pH range of 6.5 to 9.0 shows that the effects of conversion from the deoxy to the liganded state in reducing the carbamino adduct formation occur predominantly at Val-18. Analysis of the pH dependence of carbamino formation at constant total carbonates yields values of pK, and pK, for Val-lb and Val-1cY in the deoxy and liganded conditions. In contrast to the Val-l/3 role as the allosteric site for COz, the Val-la site is shown to be primarily an alkaline Bohr group. 2,3-Diphosphoglycerate is shown to reduce substantially the Val-l/3 carbamino resonance intensity in deoxyhemoglobin.
Evidence for 2,3-diphosphoglycerate effects in carbon monoxide hemoglobin at both Val-la and Val-lj3 sites is presented. Enhanced carbamino formation in carbon monoxide hemoglobin at Val-l/3 is observed at pH values less than 7.8. Finally, chemical exchange analysis of the spectra shows the release rate of the deoxy Val-la carbamino adduct to be greater than that for deoxy Val-l/3. At pH 7.47 k;;s,s.P N 1.0 and k;&m "11.0 s-l.
The identification in human deoxyhemoglobin of Val-l/3 as the dominant site of formation of carbamino adduct has been achieved by several methods (2)(3)(4)(5). The present report extends the range of observation to yield quantitative estimates of CO, binding parameters for both subunits in liganded and unliganded states and in the presence and absence of the effector 2,3-diphosphoglycerate.
Observations are also made concerning the rates of release of the carbamino CO, from the individual subunits.
The pH dependence of carbamino formation, provided the total carbonates vary little, is generally constrained to a bellshaped form dictated by the equilibria * This work was supported by qublic Health Service Grants HL-05556 and 14680. This is the 80th paper in a series dealing with coordination complexes and catalytic properties of proteins and related substances (see Ref. 1 Here, K, is the dissociation constant of the amino group in question and K, is the formation constant of the carbamino adduct, expressed so as to include the step of dissociation of the relatively strong carbamic acid (6). At low pH, the concentration of the nonprotonated amino form will tend to be limiting, and at high pH, the concentration of dissolved CO? will tend to be limiting.
The assignments of carbamino resonances are again based principally on the observation of changes resulting from specific blockage of amino groups by modification with cyanate (5, 7-9). By these means it is possible to show a stabilization of the Val-1P adduct in the liganded form at pH values near 7.8 and below. Evidence is also obtained showing that 2,3-diphosphoglycerate affects carbamino adducts in both the liganded and unliganded states.

EXPERIMENTAL PROCEDURES
Normal Adult Hemoglobin -Hemoglobin A,, was prepared by DEAE-Sephadex chromatography following the procedure of Huisman and co-workers (10,ll) or that of Williams and Tsay (12). Other procedures such as removal of phosphates and paramagnetic ions, reduction of ferric forms when necessary, and equilibration with 'Xenriched CO, and bicarbonate were done as previously described (5,13).
Carbamylated Derivatives of Hemoglobin -The method of Williams et al. (9) was applied for the separation of specifically carbamylated derivatives of hemoglobin A,, with modifications and analyses made as previously described (5). The presence of the carbamyl form of the NH,-terminal residue in a given subunit is designated by a superscript, e.g. 01~~& to indicate the form blocked on the a subunit. Determination of 2,3-Diphosphoglycerate -The concentration of 2,3-diphosphoglycerate as the sodium salt, derived by ion exchange chromatography from the pentacyclohexylammonium salt (Calbiochem), was determined by an adaptation of the method of Lowry et al. (14).' NMR Measurements -NMR measurements were made as previously described (5) with the following changes. All spectra were accumulated in 16,384 and 8,192 memories giving digital resolution of 0.03 or 0.06 ppm. In several cases a high field instrument operating at 67.899 MHz was used (1, 15). For this instrument a recycle time of 5.5 s was used. Samples of the carbamyl hemoglobins were prepared with an additional internal reference compound, N-2-hydroxyethylpiperazine-N'-2-ethanesulfonate.
Two '% resonances in this compound with acid limits at 137.49 and 143.63 ppm, respec-such effects in terms of CO, release from hemoglobin. The acidtively, titrated with the expected pK value of 7.54 (16) and thus catalyzed release of CO, from model carbamate compounds is known provided a direct measure of the pH of the sample.
to be the dominant pathway although an uncatalyzed pathway also Data Handling*-The mole fraction of carbamino adduct of a exists. The scheme given by Caplow (24) Fig. 1 by plotting dZldK, and dZ/dK, as functions of pH for the typical case in which pK, is 5.5 and pK, is 7.0. It is seen that 2 is more sensitive to changes in Kc than K, for all relevant values of pH, and that only values of 2 for pH less than 7.5 are important for evaluating K,.
The effects of chemical exchange on NMR linewidths and resonance frequencies are well understood (22, 231. Here we examine ' Data and sample calculations are presented as a miniprint supplement immediately following this paper. (Tables IV through VIII will be found on p. 2244.) For the convenience of those who prefer to obtain this supplementary material in the form of 7 pages of full size photocopies, these same data are available as JBC Document No. 76M-1393. Orders for supplemental material should specify the title, authors, and reference to this paper and the JBC Document Number, and the number of copies desired. Orders should be addressed to The Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, Md. 20014, and must be accompanied by a remittance to the order of the Journal in the amount of $1.05.
The subscripts a, p, and CO, refer to Val-lo and Val-1P chain carbamates and free CO,, respectively.
Since the reaction is studied at equilibrium not all the fractional populations, fi, and rate constants are independent. The equilibrium relations between the t; and rate constants are k oba.a fe = k& fco;l k ohs.,, f,i = k&s ho.
Thus, for example, one can describe the system by the three fractional populations and two rates, k&, and k,&,. The desired steady state solution of these equations for the sum of the three magnetizations in the xy plane was obtained using the matrix method described by Johnson and Moreland (23). A fortran program utilizing the above analysis has been written. In Fig. 2 the effects of increasing k& on the linewidth and chemical shift of the carbamate resonances are shown. Both carbamate resonances are placed at 29.8 ppm in the absence of exchange (as in carbon monoxide hemoglobin), free CO, is at 68.4 ppm and the two values for k,-d,,, and k,-d, p are the same. The population of free CO,, f co*, is assumed to be 10 times greater than both fa and fu as is typical 1. The dependence of 2 (mole fraction carbamino) on the parameters K, and K, as a function of pH is illustrated.
The c~rue.s are based on typical values of pK, of 5.5 and pK, of 7.0, and show that 2 is much more sensitive to K, over the entire pH range of interest and K, exerts a negligible effect on observed 2 above pH = 8.0. 2. Shows the predicted line width (Hz) and chemical shift dependence on release rate (parts per million from external CS,) of two carbamino adducts exchanging with CO,. The line width in the absence of exchange is assumed to be 5 Hz and the spectrometer frequency 25.2 MHz.
in these experiments for acidic values for pH. Since the difference in chemical shift for the carbamate resonances in deoxyhemoglobin is small (0.6 ppm) with respect to the difference in chemical shift between free CO, and carbamate (39 ppm), Fig. 2 is also relevant to the discussion of exchange in deoxyhemoglobin (5). 3. Shown here are 25.2 MHz 'W NMR spectra of human adult hemoglobin A,, equilibrated at various pH values with '"CO,. A, deoxyhemoglobin; B, carbon monoxide hemoglobin. The pH values are listed at the rigght of each spectrum. The parts per million scale on the abscissa is referred to external CS,. Total carbonates were 35 to 57 mM, and WO, (mole fraction WO, 0.82 to 0.921 was equilibrated at pressures of 4 to 374 torr. The hemoglobin concentration was usually 11.0 to 16.6 rnM expressed as heme concentration in 0.05 M NaCl. The bicarbonate-carbonate resonance appears near 33 ppm, usually with spinning side bands in evidence. Measurements were obtained between 29-31". exchange and is not shown in Fig. 3 since it occurs near 68.38 wm.

General
For deoxyhemoglobin at higher values of pH, three carbamino resonances at 28.4,29.2 and 29.8 ppm are observed. They are assigned, respectively, to carbamino adducts to c-amino groups, Val-lot, and Val-1P. As the pH is lowered the intensities of all three resonances decrease. The e-amino adduct is not observable at pH 7.4. For all pH values the Val-l/3 adduct is more prominent than the Val-la adduct which is in turn more prominent than the e-amino adduct. Fig. 3B shows representative spectra of carbon monoxide hemoglobin equilibrated with '"CO, in the same manner under conditions specified in the legend. At high pH values only two carbamino resonances are observed. The resonance at 28.4 ppm again is assigned to the e-amino adducts and the 29.8 ppm resonance is assigned to the adducts of both Val-la and Val-1P 6).
As the pH is lowered (Fig. 3B)  with manometric measurements at this pH (4). The data points included in Fig. 4 were obtained at the three different spectrometer frequencies of 15.1, 25.2, and 67.9 MHz, all yielding equivalent chemical shifts at a given pH. These results eliminate the possibility that the difference in chemical shift values for the Val-lo and Val-l/3 adducts in carbon monoxide hemoglobin at low values of pH has its origin in any intermediate NMR rate process involving chemical exchange. Quantitation of Carbamino Adduct Formation-The fraction, 2, of a-amino group in each subunit in the carbamino form under given conditions was measured over the broad pH range from 6.5 to 9.0. The observed 2 values are reported here as corrected to 55 mM total carbonates (5). Fig. 5 shows the values of 2 as a function of pH for deoxyhemoglobin.
In this case the chemical shift values distinguish the Val-1P adduct (29.8 ppm), the Val-la adduct (29.2 ppm), and the e-amino adducts (28.4 ppm). At the high pH values, as previously reported (4, 51, the individual subunits exhibit nearly equal affinity for CO, in the carbamino form. Fig. 5 shows that in the physiological pH range, however, Val-l/? forms the predominant carbamino adduct. Table I summarizes the experimental conditions for the results in Figs. 5 and 6, and gives the values for pK, and pK, derived in each case from the two-parameter fit according to Equation 5. The curves are drawn according to Equation 5 using the values for pK, and pK, given in Table I. The insets in Figs. 5 and 6 show the per cent change in xz when pK, and pK, are caused to vary about their fit values. As expected from the computations illustrated in Fig. I, x2 is much more responsive to variations in pK, than in pK, for all data examined. Note that a unique solution exists for each parameter. The In the case of carbon monoxide hemoglobin, in contrast, the individual carbamino adduct resonances are clearly resolved only below pH 7.2, the Val-la! adduct shifting upfield of the Val-l/3 adduct. Accordingly, the specifically carbamylated carbon monoxide hemoglobin derivatives, (Y&~(CO) and a2('P2(CO), were used to evaluate the contribution of each liganded chain over the entire pH range of interest. Fig. 6 shows the 2 values as a function of pH for the unmodified carbon monoxide hemoglobin and for the two liganded, carbamylated derivatives.
It may be seen that the sum of the 2 values for the modified species is closely equivalent to the 2 value at a given pH of the unmodified carbon monoxide hemoglobin.
In the limited number of cases in which the Val-lcr and Val-l/3 adduct resonances in the unmodified carbon monoxide hemoglobin could be distinguished at pH 7.2 and below, the results corresponded well with those for the modified derivatives. The inset is included to show the sensitivity of the fitting criteria x2 when the best fit values of pK, and pK, are varied. Per cent change in x2 versu.s pK, and pK, for Val-Icu and Val-lp are shown by solid and broken lines, respectively. The deoxy Val-la acts with a pK, of 7.83 f 0.19 and pK, of 4.89 f 0.10 while deoxy Val-l/3 shows a pK, of 6.91 2 0.32 and pK, of 4.64 ? 0.08. The error limits represent 2 standard deviations. Measurements were determined at three field strengths, 14.1 kG (01, 23.5 kG (A), and 63.7 kG (0). All spectra were accumulated at conditions similar to those described in Fig. 3. Error bars reflect the digital resolution of the respective instruments. to Z over the pH range of interest. By the fitting criteria described in the inset it is seen that a&' conforms to a pK, of 7.16 f 0.18 and pK, of 5.5 + 0.04. The Z versus pH curve for (Yap& does not yield a two-parameter fit, yet it is easily seen that (~*~/3* and CY&~ do corroborate the behavior of the unmodified form, at/&. where the hemoglobin concentrations are expressed per olfi dimer, and the expression is applied to discrete value of [H+l taken directly from the pH reading. In Table I  of the carbamino adducts to Val-la and Val-l/3, 29.2 and 29.8 ppm, respectively, are unchanged by the inclusion of the organic phosphate. The predominance of the Val-l/3 adduct is much less pronounced than in the absence of the 2,3-diphosphoglycerate (Fig. 3A). Fig. 7B shows the carbamino resonances of carbon monoxide hemoglobin in the presence of 8.0 mM 2,3-diphosphoglycerate. The most striking observations about the carbamino adducts to Val-la and Val-1P have to do with the pH values in the physiological range and below. Here there are two points of distinction from the results in the absence of the organic phosphate (Fig. 3B). First, the resonance at 29.8 ppm, which reflects the sum of the two adduct contributions, is reduced in intensity.
Second, the Val-la adduct resonance does not undergo the characteristic upfield shift in the lower pH range. Fig. 8 shows very clearly the equivalence of the 2 values computed for the Val-la and Val-1P carbamino adducts to deoxyhemoglobin over a range of pH. This result is in clear contrast to the dominance of the Val-1P adduct in the absence of the organic phosphate. Fig. 9 shows the effect of 2,3-diphosphoglycerate in reducing the overall 2 value for the combined Val-la and Val-lp carbamino adducts to carbon monoxide hemoglobin.
The points are observed in the presence of organic phosphate.
The lower curve shows a two-parameter fit of these data according to Equation 5. The upper curve is taken from Fig. 6 and refers to 2 values obtained in the absence of organic phosphate. Table II lists the conditions of the experiments described by Figs. 8 and 9. Since the presence of 2,3-diphosphoglycerate affects the carbamino formation in most cases, the effective pa-dependent formation constants, h (Equation 12), are given in preference to computed values of pK, and pK, which would lack general meaning under these conditions. The values of h given in Table II can be compared directly with those computed in the absence of the 2,3-diphosphoglycerate which were listed in Table I. At pH 7.4, for example, the presence of the organic phosphate reduced h for the deoxyhemoglobin Val-lb site from 435 to 167 M-' whereas for the Val-la site there actually appeared to be a slight enhancement of carbamino formation.
At the present time the details of the effect of 2,3-diphosphoglycerate on the Val-la and Val-I@ carbamino adducts are not fully known. The decrease in overall 2 value shown in Fig. 9 fits nearly quantitatively with a loss of the stabilization seen for the Val-1P adduct below pH 8 (Fig. 6). This strong indication will be explored with suitably carbamylated preparations in the presence of 2,3-diphosphoglycerate.
An effect of the organic phosphate on the Val-la carbamino adduct in carbon monoxide hemoglobin is clearly demonstrated by the suppression of the shift to higher field observed in the lower pH range, which may be seen by comparing Fig. 7B with Fig. 3B.

Roles of a-Amino
Groups of a and p Subunits-The results in Table I confirm that the a-amino groups of the a and /3 subunits are adapted to clearly different roles with respect to the heterotropic effecters (51, in addition to any discrimination with respect to the homotropic effector, OI (27-29). One role is that taken by Val-la in responding to the change in heme ligand state by undergoing a change in hydrogen ion binding, thereby contributing to the so-called alkaline Bohr effect (17,30). The mechanism can be seen in the sharp change in pK, for the Val-la, a change that is not observed with Val-lp. The other role is that taken by Val-l/3 which undergoes a B, carbon monoxide hemoglobin and 8 m&r 2,3-diphosphoglycerate.
The pH values are listed at the right of each spectrum.
The parts per million scale on the abscissa is referenced to external CS,. Total carbonates were 48 to 60 mM and WO, (mole fraction WO, 0.82 to 0.92) was equilibrated at pressures of 4 to 374 torr. The average hemoglobin concentration were 11.97 m&r heme and 11.32 rnM heme for A and B, respectively. Conditions are more fully described in Table II. change in binding of CO, in the carbamino form, thereby making the major contribution to the so-called Haldane effect (17, 31). The mechanism in this case depends again on the differences in pK, values between the subunits and, in addition, on the reduced stability of the carbamino derivatives in both subunits in the ligand state as expressed by the pK, values in Table I. Since Val-lp undergoes the change in pK, but not in pK, it experiences a clear decrease in carbamino derivative that is mirrored in the values for h in Table I. On the other hand, the same trend of pK, values for Val-la is nearly compensated by the differences in pK, between the liganded and unliganded forms. In effect, Val-la in the deoxyhemoglobin is primarily protonated at physiological pH and is much less free to form the carbamino derivative than is Val-1P which is primarily unprotonated.
Since the two pK, values in deoxyhemoglobin lie about 0.5 unit on either side of the physiological pH of 7.4, Val-lb forms and discharges much more carbamate than does Val-la. The pK, values of 4.64 and 4.89 listed in Table I are  9. The effects of 8 rnM 2,3-diphosphoglycerate (DPG) on the urnmodified carbon monoxide hemoglobin resonance at 29.8 ppm (referenced to external CS,). The computer-calculated curve (from Fig. 6) for the unmodified carbon monoxide hemoglobin in the absence of organic phosphate is included for comparison as the upper solid curue. This consequence for the a subunit would run counter to the allosteric effector role of CO, with respect to the p subunit. Secondly, the maintenance of a low level of carbamino formation with Valla under both liganded and unliganded conditions allows this site to serve its function as essentially a Bohr group without significant interference. Dependence of Carbamino Formation on pH, pcoz, and Bicarbonate- Fig.  10 shows plots for the a-amino group of both a and /3 subunits in the liganded and unliganded states. The plots show values of 2 versus pH corresponding to various values of the concentration of bicarbonate and pcoz. For reference, each panel is marked with the symbols 0 and $J to indicate the conditions corresponding to arterial and venous blood, respectively.
The plots in Fig. 10 are useful for observing the response of 2 values for the individual chains to acidosis and alkalosis of respiratory or metabolic origin (32, 33).

Relationships between Bohr and Haldane
Effects- Fig.   11 shows results of computations describing the changes in bound hydrogen ions and carbamino derivatives at the a-amino groups between the two ligand states under given conditions of pH and pCo2. Fig. 1lA shows plots of AZ uersus pH for the carbamino formation with the a-amino groups of both sub- In these equations (Equations 13 and 14, below) the subscripts 1 and 2 refer to the initial and final ligand states and K is the Henry's law proportionality constant. The change in hydrogen ions bound, representing the contribution to the alkaline Bohr effect, is given in Fig. 11B for Valla! only since the corresponding plot for Val-1P would not be meaningful (Table I) where Zj = hj k P co /(l+hjkPco 2 2) Fig. 11A shows how much more marked the change in carbamino formation is for Val-l/3 and also how marked is the effect ofp,,, at physiological pH. The results in Fig. 11B show the increasing effect of pcol in limiting the proton uptake by the o-amino group of Val-lo, an effect that is relatively modest at pH 7.4.
The relationship between the Bohr and Haldane effects is most simply seen from the computations shown in Fig. 12. For each pH the bars show changes that accompany the transition from the liganded to the unliganded state of the hemoglobin with pcoZ equal to 40 torr. The left hand member of each pair shows the uptake by the two identified alkaline Bohr groups, His-146p and Val-la.
The pK, change experienced by His-146p has been found to be virtually identical with that of . It is assumed that His-146P is not directly affected by COe, whereas allowance is made for this effect on Val-la as shown in Fig. 11B. The right hand.member for each pH represents the net release of protons due to the Haldane effect from the two groups Val-la and Val-l/?. Here the total release consists of the contribution given by Equation 2, and also the effect of the carbamino formation in removing from the equilibrium of Equation 1 the conjugate base form of the amine.
The results in Fig. 12 show that the process of carbamino formation in the absence of organic phosphates can provide a substantial fraction of the protons taken up by these two Bohr groups that are responsible between them at pH 7.4 for approximately 80% of the alkaline Bohr effect (17,34,35 FIG. 13. Shown is a simulation of the carbon monoxide hemoglobin Val-l/3 carbamino formation data by incorporation of a third pHdependent parameter (in addition to K, and K,). The dotted curues show the enhancement of the low pH 2 values as the strength of the stabilizing parameter (K,) is increased from 0 to 600 cal/mol. Effect of 2,3-Diphosphoglycerate-Comparison of Table II  with Table I and of Fig. 8 with Fig. 5 shows that 2,3-diphosphoglycerate reduces the formation of carbamino adduct to Vall/3. As expected, the effect is most obvious at the lower pH values, in the range where the organic phosphate binds most strongly to the deoxyhemoglobin (38, 39). A more detailed interpretation will require direct measurement of the binding of the 2,3-diphosphoglycerate to the hemoglobin. The present experiments involve higher ionic strength in the presence of the organic phosphate; furthermore, bicarbonate ion may act somewhat like chloride ion in reducing the binding of the polyanion (39, 40). The results in Fig. 9 show that total carbamino formation at Val-1P and Val-la in carbon monoxide hemoglobin is suppressed by 2,3-diphosphoglycerate.
Here again direct binding measurements are planned for the future on specifically carbamylated hemoglobin preparations.
Since the carbamino formation in Fig. 9 shows a suggestion of dropping off relatively sharply below pH 7.5, it is tempting to conclude that the adduct to Val-l/3 is preferentially suppressed (cf. Fig. 6). The results in Fig. 7B showing that the upfield chemical shift of the Val-la carbamino adduct resonance at lower pH values (Fig. 4)  Also present in these solutions, with the exception of the measurement at pH 8.47, was an enzymatic reducing system (41); thus the ionic strength of these solutions is higher than those of spectra reported in Fig. 3. Each spectrum is accompanied by its computer simulation. environment of Val-la. From the kinetic analysis given below portion of the digitized spectrum (4096 data points) obtained and the treatment presented under "Data Handling" it is clear for deoxyhemoglobin was fit to the case of the three-site exthat this chemical shift has its origin not in an exchange change process. The actual spectra and the corresponding best process but in a conformational or electrostatic change con-Iit simulations are shown paired in Fig. 14. The values used trolled by proton binding that appears to be suppressed in the for the independent population and release rate parameters presence of 2,3-diphosphoglycerate.
In this regard it is inter-are listed in Table III Val-la and Val-1P was presented in the data-handling section and in terms of the populations of the sites and of the two first order release rate constants as the independent variables u,, = k,(R-NHCO,H) = k,(R-NHCOzJ(H+)/K, (Equations 7 to 11). The analysis deals with the observed where K, is the acid dissociation constant of the carbamate.4 variables of line width, resonance frequency, and resonance intensity.
The line widths of carbamino resonances in Figs. 3, A and B, are typically less than 12 Hz (25.2 Hz = 1 ppm in these spectra) with the Val-la adduct showing the greater broadening at low pH. In terms of Fig. 2 these observations mean that the first order rates of release of CO, from the hemoglobin are typically less than 25 s-l. This result is in agreement with the studies of Caplow on release of CO, from model carbamate compounds (24).
To estimate the exchange rates over a range of conditions a 3 Since the rates studied here are generally low, broadening effects due to magnetic field inhomogeneity and spectrometer drift are nearly of the same degree as the exchange broadening itself.
* The form of Equation 2 takes into account an estimate of pK,, of approximately 5.0 (6). This value is reasonable in view of the invariance of chemical shift of carbamino resonances in deoxyhemoglobin examined down to pH 6.6. The pH dependence of 13C chemical shifts in carboxyl or carboxamide groups normally covers a range of the order of 8 to 10 ppm (13). With a digital resolution of 0.03 ppm applying in the present studies, it follows that the pK, almost certainly falls at least 1.3 units below the pH of observation, 6.6. constants for the carbamino adducts of deoxyhemoglobin were calculated on the basis of a least squares fit to the observed spectrum based on three site exchange with CO, and the two NH&erminal carbamino adducts. The apparent T, used for the fitting procedure was 0.064 s. The values reported below must be regarded as only semiquantitative, due to the uncertainties involved in the fitting procedures. The column designated populations gives the relative populations of the resonances at 29.2, 29.8, and 68.3 ppm (CO,) as estimated by the fitting procedures. Taking this value at pH 7.47 and assuming that pK, is 4.6, corresponding to the value (Table I) for the Val-1P adduct that would be dominant under the experimental conditions (42), k& is found to be 17 s'. This value is somewhat higher than those listed in Table III for the Val-lp site, but within the errors inherent in the methods.