The Alkaline Reaction of Nicotinamide Adenine Dinucleotide, a New Transient Intermediate*

In basic solutions NAD+ reversibly produces a 370 nm absorbing transient intermediate which irreversibly decomposes to a 340 run absorbing species. The concentration of the transient produced is proportional to the hydroxide concentration up to 5 M NaOH. NAD+ also undergoes an ionization reaction of the carboxamide NH, group, the K, of which is 7.3 x 10-13~. NMN undergoes spectral changes in alkaline solutions entirely analogous to those of NAD, and has a K, of 5.6 X lo-l3 M.

NMN undergoes spectral changes in alkaline solutions entirely analogous to those of NAD, and has a K, of 5.6 X lo-l3 M.

NAD+
in alkaline solutions reversibly increases its absorbance in the 290 nm region (1). A reaction also takes place to produce an intermediate which absorbs maximally in the 340 nm region. This intermediate is replaced by a final fluorescent product which absorbs maximally at 360 nm in its basic form and at 340 nm in its acidic, nonfluorescent form (2). The pK, of the 360 nm product as measured by acid quenching of fluorescence (2) and spectrally (3) is about 9.6. This fluorescent product is the basis of an analytical assay for NADf (4). N-Methylnicotinamidc cation has been studied in alkaline solutions as a model for the alkaline NhDf reaction and has been found to undergo reversible ionization at the carboxamide group with a dissociation constant of 6.8 X lo-l4 M (5). We have recently found that A-methylnicotinamide cation undergoes reversible changes in solutions more alkaline than 2 1\1 NaOH to produce materials absorbing maximally at 320 nm (3). We have interpreted this additional equilibrium reaction to be due to a ringopening reaction of N-methyhlicotinamide cation.
A more readily accessible model of ring-opening reactions is available in the alkaline ring-opening reaction of N, N-dimethylcarbamoylnicotinamide cation (3) or of N,N-dimethylcarbamoylpyridinium ion (6). These reactions, which are kinetically second order in hydroxide, are essentially irreversible in that the reversal reaction is so much slower than the rate of formation under * This investigation was supported by Public Health Service Grant GM 168586 from the National Institutes of Health and bv Health Research and Services Foundation (Pittsburgh) Gran"t, K21.
alkaline conditions that the ring-opening product can be isolated.
In exploring the possibility of the occurrence of ring-opening reactions of NBD+ we have looked very closely at the initially produced spectrum when NAT)+ is introduced into alkaline solutions.
A the kinetic plots of the 370 nm data. A Durrum stopped flow spectrophotometer was used to study the rate of production of the transient 370 nm intermediate.
The maximum excursion of the oscilloscope trace is recorded and the resulting transmittance value is converted to A,,.
The transmittance values resulting from the build-up of the transient are converted to absorbance values and a plot of log (Amax -At) against time is made and the rate constant is obtained from the slope. All plots obtained in the stopped flow work are linear to 90% reaction.
A Cary 16 spectrophotometer thermostated at 25.0 f 0.1" was used for the determination of the equilibrium constants of NAD+ and NMN in dilute alkali. A previously measured amount of sodium hydroxide solution is placed in a l-cm glassstoppered cuvette and the cuvette is allowed to come to temperature equilibrium in the thermostated cell compartment.
Manual measurements are made at 290 nm, giving the absorbance value of the cell and its contents with air as a reference. The cell is always placed in the same position in the cell holder. A measured quantity of NAD+ or NMN is added to the cell and the total contents are mixed. The cell is then carefully replaced and measurements of the absorbance at 290 nm are recorded as a function of time for 3 min. Linear extrapolation to zero time gives Ao. Acid quenching of basic NAD+ solutions was carried out in the following manner: 10 ~1 of 0.2 M NAD+ were added with stirring to 2.0 ml of 5 M NaOH or to 2 ml of 1 M NaOH. After measured time intervals the reactions were quenched in an icesalt bath with 1.7 ml of a mixture of 23 ml of concentrated HCI and 10 ml of 1 M Tris, or with8 ml of 0.6 M TriseHCI, respectively. The final pH is near 8.2. The spectrum is taken and the NAD+ content is measured by reducing the remaining NAD+ to NADH with ethanol catalyzed by alcohol dehydrogenase and measuring the absorbance change at 340 nm (8). The amount of absorbance change at 340 nm obtained in the quenched samples is compared with that obtained when the same amount of NAD+ is added to the acid and basic components previously mixed together. From these data the percentage of NAD+ remaining is calculated.

RESULTS
The spectrum of NAD+ or NMN in alkaline solutions, if taken immediately after the NAD+ or SMN is introduced into the alkaline medium, shows an unstable intermediate absorbing maximally at 370 nm. This intermediate is more easily seen in concentrated alkali because more of it is formed. A typical sequence of events is as follows. In 3.0 ml of 5.0 M NaOH solution 10 ~1 of 0.028 Y NAD+ are introduced, the solution is mixed, and scanning is started 11 set after mixing. Scanning downward from 400 nm, the 370 nm region is reached 23 set after mixing and a large absorption band is apparent, as well as a second small band at 331 nm. The second scan is started 123 set after the original mixing and shows very little of the distinct 370 and 331 nm bands. Instead, a 342 nm band is building up, reaching maximum absorbance 6 min after the initial mixing. Thereafter the 342 nm band slowly shifts to longer wave lengths and decreases in intensity. A stable final spectrum, which has a 360 nm maximum, is obtained in 2 hours. This material is highly fluorescent as described by Kaplan, Colowick, and Barnes (2), and is maximally excited at 360 nm and maximally emits at 450 nm (3). Similar results are obtained in solutions more alkaline and less alkaline than above 5 M NaOH solution. Even in 0.3 M NaOH a small amount of the 370 nm transient can be seen in the very first scan if a large amount of NAD+ is used. In solutions less alkaline than 5 M NaOH, after the disappearance of the 370 nm intermediate, a material absorbing at 330 nm appears to be formed and subsequent spectral scans show increasing absorption as well as increasing wave lengths of maximal absorption until the reasonably stable 342 nm intermediate is obtained. The 342 nm band always subsequently slowly decreases in intensity and the wave length of maximal absorption is shifted to higher values until the final 360 nm product is formed.
Rapid repetitive scanning of reacting basic solutions of NAD+ or NMN shows the presence of an isosbestic point between the 370 nm transient and the 340 nm product, as shown in Fig. 1. The isosbestic condition lasts for about 2 min, corresponding to about 757, reaction in the cases studied here, and thereafter the isosbestic point shifts to a slightly lower wave length. The instability of the isosbestic point after 75% reaction is probably due to the decomposition of the 340 nm product to the final 360 nm product. The position of the isosbestic point varies with basicity as follows: (given is molarity of NaOH, isosbestic point for NAD+, isosbestic point for NMN) 2.5 M NaOH, 368 nm, -: 5 M NaOH, 363.5 nm, 366 nm: 10 M NaOH, 355 nm, 356 nm: 15 M NaOH, -335 nm. Semilog plots for the decomposition of the 370 nm material and for the formation of the 340 nm material show similar or identical rates of decomposition or formation.
These results are included in Table I. The amount of 370 nm material initially formed was determined by extrapolating to zero time in a semilog kinetic plot or directly with the stopped flow method. These results indicate a linear increase in yield of the 370 nm product with increasing alkalinity up to 5 M NaOH for both NAD and NMN as shown in Fig. 2. The lower limit of the extinction coefficient of the 370 nm product from NAD is 25,000 M -l cm -l and from NMN, 20,000 M+ cm-r. Above 5 M NaOH a leveling in the yield of 370 nm product occurs which may reflect either an approach to complete conversion of NAD+ to the 370 nm material or specific salt effects. The yield data in Fig. 2, if plotted against the alkalinity function of Schwarzenbach and Sulzberger (9), show a definite leveling off in the yield of the 370 nm intermediate as shown in Fig. 3. The maximal yield of the 340 nm product also levels off above 5 36 NaOH, as seen in Fig. 2. A similar plot is obtained with the NMN yield data. The rates of formation of the 370 nm intermediate, decomposition of the 370 nm transient, formation of the 340 nm absorbance, and decrease of the 340 nm absorbance are given in Table I. The rate of decrease of the 340 nm absorbance is identical with the rate of decrease of 360 nm absorbance and probably reflects the decay of the 340 nm product to the 360 nm product. The 370 nm transient is formed much more rapidly than it decomposes to the 340 nm product and the 340 nm product decomposes at an even slower rate to the 360 nm product.
Rapid Kinetic Stud+NAD+ when rapidly mixed with weakly alkaline solutions and followed at various wave lengths shows two distinct kinetic phenomena. At high wave lengths a transient increased absorption is observed: at low wave lengths a nontransient increased absorption is observed. For example, in 0.05 rd NaOH solutions, NAD+ shows a "slow" (h/r s 0.5 set) increased absorbance in the 340 to 390 nm range with maximal transient phenomena occurring at 360 to 370 nm. After its formation the transient slowly produces more highly absorbing materials. At wave lengths below 335 nm no transient phenomena are encountered: the increased absorbance of NAD+ at low wave lengths first noticed by Burton and Kaplan (1) is produced with a half-life less than 0.012 set, which is the limit of the stopped flow cuvette used in these experiments. This nontransient behavior was observed from 335 to 290 nm.
Nontransient Increase in Abaorbo~pectral observations of NAIY in dilute alkali have led to the conclusion that NAD+ increases its absorbance maximally at 290 nm (1). The difference absorption of NAD+ and NMN in alkaline solutions versus water at 290 nm is shown in Fig. 4. The maximal change in absorbance comes in the 0 to 0.04 M NaOH range of concentrations for both NAD+ and NMN.
Since the behavior of these two compounds is so strikingly similar, the possibility of an adenyl ionization may be discounted in the case of NAD+. The increased absorbance may be interpreted as an ionization of the carboxamide NHz group in analogy with the carboxamide ionization of N-methylnicotinamide cation (5, 10). The tr&,ment of the data in Fig. 4 is as follows. The ionization constant of NAD+ is given by Equation 2 in which N+ is the positively charged species of NAD+, and N is its neutral species resulting from ionization of NAD+. No is the total of N+ and N. K, The absorbance of a solution of N is given by: where el and ~1 are the extinction coefficients of N+ and N, respectively. It follows that: W=A* where Ao is c,No or the absorbance of an aqueous solution of Nf. Substituting this value of (N) into Equation 2 yields: A plot of the left-hand side of Equation 5 against l/(OH-) should yield a straight line with 1 /(a -cl) as the intercept and 1 /(e -edK~ as the slope. The value of K1 is then given as the intercept divided by the slope. The data in Fig. 4 when plotted in this way give a linear regression. A least squares analysis of these data yields Kr values of 72.7 k 3.3 ~-1 for NAD and 56.4 f 6.2 M-I for NMN.
This corresponds to acid diasociation constants of 7.3 x 10-u M and 5.6 x lO+a M for NAD+ and NMN, respectively.
Quenching Ezperimenls -With increasing time in basic solutions, increasing amounts of a 375 nm product (seen at pH 8.1) are formed and decreasing amounts of NAD+ remain. Semilog plots of the percentage of NAD+ remaining against time are linear (Fig. 5), as are semilog plots for the appearance of the 375 nm product. The 375 nm material is the acidic form of the 340 nm product (pK -lo), as determined by reversible titration of this material. The rate constants for the disappearance of NADf in 1 M NaOH and 5 M NaOH are 0.31 and 0.69 min-*, respectively.
The rate constants for the appearance of the 375 nm product in the quenching experiments are 0.68 and 0.92 min-i. Although the rate const.ants for the disappearance of NAD+ and appearance of the 375 nm product are not exactly the same, upon consideration of t.he room for error in the quenching experiments, it is probable that the rates are equal. In basic solution the disappearance of NAD+ parallels the appearance of the 340 nm product.

DISCUSSION
Kinetics and Stoichiometry--At the alkalinity levels at which the appearance of the 370 nm material is studied, the N.4D+ molecule exists primarily as the amide anion. The formation of this amide anion undoubtedly protects SAD+ from hydrolysis at t,he amide group as in the case of N-methylnicotinamide because the amide hydrolysis in highly alkaline media is retarded and other alkaline processes which are not inhibited by the amide equilibrium can become predominant (10). another reaction that must be considered with NADf in alkaline solution is cleavage at the nicotinamide-riboside bond, which has been shown to occur by Kaplan et al. (2), and to be first order in hydroxide in the pH 8 to 9 region (11). The riboside cleavage reaction is more sensitive to alkali than is the pyrophosphate cleavage in 0.1 to 1 1~ NaOH (1). In 0.17 M KOH, the rate of disappearance of the riboside-nicotinamide linkage determined by the cyanide addition reaction is about equal to the rate of the loss of functional SAD+ as determined by ethanol oxidation catalyzed by alcohol dehydrogenase (1). This means that the cleavage of the nicotinamide riboside bond is the predominant process which determines the destruction of NAD+, as this is the process which destroys the posit.ive charge on the nicotinamide ring nitrogen necessary for the cyanide addition reaction.
The behavior of NAD+ in basic solutions 1 to 15 M in NaOH can be summed up as follows.
1. NAD+, in alkaline solutions in which it exists mainly as the amide anion, undergoes a very rapid reaction (tl,, 4.5 set) to form a highly absorbing species absorbing maximally at 370 nm, a >25,000 r1 cm-l.
2. The formation of the 370 nm species is reversible and becomes complete at high alkalinities.
3. The 370 nm species* undergoes an irreversible conversion * Any species in equilibrium with this species could be undergoing the irreversible reaction. Because of known ring-opening reactions of analogous compounds (3,6) we believe that the 370 nm species is undergoing the irreversible reaction. to a 340 nm product and to another substance (or substances) which is transparent in the near ultraviolet.
The half-life for this process is about 1 min. 4. The 340 nm material undergoes an irreversible conversion to a highly fluorescent material which absorbs maximally at 360 nm. The half-life for this conversion is about 20 min.
The behavior of NMN in base is very similar to that of NAD+. Equation 6 illustrates the above points.
[Anionic X4D'J Reversibility is supported by the fact that in 5 x NaOH, in which the conversion of NAD+ to the 370 nm substance is nearly complete, quenching with acid 5 set after t.he introduction of NAD+ to alkali nearly quantitatively regenerates NAD+.
The variation with basicity of the isosbestic point between the 370 nm and 340 nm species, not accountable by t.he pK, values of these substances at pH values greater than 14, supports the contention that the 370 nm substance is convert.ed to the 340 nm species as well as optically transparent materials. If the 370 nm species (I) decomposed according to Scheme 6 by concurrent kinetics (12) to the absorbing 340 nm species (II) and to transparent materials (T), t.lie ratio of (T):(II) at any time is z, which is the ratio of the corresponding rate constants for the formation of these materials from I, ka/k,.
The value of z will be constant at constant pH, and an isosbestic will be obtained at the wave length that err = ~(1 + z). The optical density at this wave length of a reacting mixture equals ~(11 + 11~ + I) zz eICI where Ct is the total concentration of components. The higher x is, that is, the lower the yield of the 340 nm species from the 370 nm species, the lower the wave length at which the isosbestic point occurs. A plot of the ratio of the maximal absorbance of the 370 nm species to absorbance of the 340 nm products gives a linear relationship with hydroxide concentration, indicating that the transition state for the production of the 340 run species contains 1 more proton that the transition state for the production of the transparent products from the 370 nm species.
The 340 nm product appears to be converted to the final fluorescent 360 nm product, because the rate of decomposition of the 340 nm product is independent of the wave length at which this reaction is studied. The wave lengths used were 340 and 360 nm. There are, however, no material balance data to prove this point with certainty. Further support for the idea that the 340 nm material converts to the 360 nm product is obtained from the data of Kaplan et CL (2) which show that the yield of the 360 nm product is linear in hydroxide up to 5 M, and after 5 M NaOH the yield of the 360 nm product levels off (4). This behavior closely follows the behavior of the yield of the 340 nm material, suggesting that the amount of the final 360 nm material produced depends upon the amount of 340 nm material produced under a given condition.
The fact that a product with spectral, fluorescence, and pK, properties identical with the 360 nm product of NAD+ is obtained from the ring-opening reaction of N,iVdimethylcarbamoylnicotinamide cation (3) suggests strongly that NAD+ is also undergoing a ring-opening reaction.
The previously studied pyridinium ring-opening reactions (3, 6) involve a pseudo-base intermediate and follow the mechanism given in Equation 7. The rate constant as a function of pH would be expected to be quite complex. The f&t, second, and third terms in Equation 12 describe processes originating .from .NAD+, anionic NADf, and I (370 nm), respectively. The last term has more higher order hydroxide terms than does the second. The frrst has none.' At higher alkalinities, therefore, destruction reactions originating from the 370 nm intermediate will be more important than reactions originating from NAD+ or anionic NAD+.
Structural Consider&ms-The effect of the riboside group in increasing the acidity of the carboxamide moiety over that of N-methylnicotinamide cation IV is at least lo-fold. Martin and Hull (5) report a value of 6.8 & 1 x 10-l' M for the acid dissociation constant of N-methylnicotinamide cation at 25", whereas the value of Brooke and Guttman (10) extrapolated to 25" is 3.53 Z!Z 0.2 x 10-r4 M. The riboside linkage at its aldehyde level of oxidation is more electron-withdrawing than the N-methyl group and is responsible for the acidity of the carboxamide group of NAD+. It is possible that the riboside linkage plays an important role in NAD+ reactions and that N-alkylated nicotinamide cations are very poor models for NAD+.
The value of kr, the rate constant for hydroxide addition to NAD, is 200 times smaller for the same process with N ,N-dimethylcarbamoylnicotinamide cation V (3), but 5 times larger than the same process with the unsubstituted N, N-dimethylcarbamoylpyridinium ion VI (6).
COXHt CONHr Q 0 IV V VI An important difference between NAD and V is that the ring-opened form of NAD is rapidly reversed to NAD, as is also the ring-opened form of IV easily closed to IV, where ae the ring-opened form of V is only very slowly acid-closed (3). The ease of reversibility is undoubtedly related to the b&city of the amino nitrogen in the ring-opened form. The more basic the nitrogen, the more rapidly the nitrogen adds to the aldehyde group, and the more rapidly the reversal reaction takes place. V is not a reversible system whereas NAD and IV are. In this respect IV is a good model for NAD. * The first term will be lowered at high alkalinities; the last term will be enhanced. The yields corresponding to these terms vary accordingly (12). Analogies are the ring-opened product of N,Ndimethylcarbamoylnicotinamide cation which absorbs maximally at 392 nm (3) and the quinolinic acid precursor 2-acroleyl-3-amino fumarate, VII, which absorbs maximally at 360 nm at pH 7 (13). o= VII Lyle and Gauthier (14) have shown that reactions of N-substituted nicotinamide cations with nucleophiles such as cyanide result in the kinetically controlled 1,2-or 1,6addition products which then more slowly rearrange to the thermodynamically more stable 1 ,Paddition products. With nucleophiles which contain two replaceable hydrogens on the same nucleophilic atom, ring opening by a process similar to that shown in Equation 7 is possible because, after the initial attack at position 2, ring-opening occurs before the nucleophile can migrate to position 4. Biologically important nucleophiles of this class are lysine and arginine. Such nucleophiles on an enzyme surface can act to store NAD+ in its ring-opened form which is less susceptible to degradation reactions such as ribose-N cleavage reactions. We have shown that ring-opening reactions are subject to general base catalysis (3, 6) so that it is very feasible that enzymes can catalyze such reactions. Enzymes such as triose phosphate dehydrogenase which tightly bind NAD+ in complexes which absorb maximally at 360 nm (15) and can form an enzyme-bound "alkali-stable NAD" (16) might well be storing NAD+ in its ring-opened modification.