NMR Parameters in Some Nitrodiphenylmethanes Coupling Chemical shifts constants Compound (c.P.s.from TRIS) (C.P.S.)
Table 111.
V
b'
a'
b
a
a'
b
a
LITERATUN CITED
(1) Abraham, R J., Bernstein, H. J , Can. J . Chenz. 39, 905 (1961). (2) Bothner-By, A . A,, Xaar-Colin, C., J . Am. C'hem. SOC.83, 231 (1961). 13) Cavanauch. J. R.. Ilailev. R P k. ('hem. P'hys. 34, 1099 (1961). (4) Ilailey, B. P., Shoolery, J. S . , J . Am. Chem. SOC.77, 3977 (1955). ( 5 ) Farbenfabriken Baver A,-(;,< Brit, Pat. 764,633 (Ilec. 28, 1956). (6)yGoldstein, J. H., Reddy, G. S., J . ( h e m . Phys. 36, 2644 (1962). ( 7 ) Giltowsky, H. S.,McCall, D. R., AIcGarvey, B. R., Meyer, L. H., J . A m . Chem. SOC. 74, 4809 (1952). (8) Sarasimhan, P. T., Rogers, 11. T., J . Chem. Phys. 31, 1303 (1959). (9) Reddy, G . S., iinpublished work. (10) Reddy, G. S., Goldstein, J. H., J . A m . ('hem. Soc. 83, 2045 (1961). (11) Ibid., p. 5020. (12) Reddy, G . S.,Goldztein, J. H., J . Chem. Phys. 38, 2736 (1963). (13) Ibid., 39, 3509 (1963). (14) Reddy, G. S., Hobgood, R. T., Jr., Goldstein, J. H., J . i l m . ('hem. Soc. 84, 336 (1962). (15) Tillieu, J., Ann. Phys. 2, 471, 631 (1957). ~~
These values are approximate since detailed analyses of the spectra werc not carried out.
methane have proved very useful in identifying and quantitatively estimating the various components and isomers in a mixture. Quantitative estimation of the components is both rapid and reasonably accurate. Obviously, these effects can be applied to, and should
for supplying the samples and carrying out the chromatographic separations.
prove useful in, the analysis of similar organic systems. ACKNOWLEDGMENT
The authors thank 11. 0. Halvorson of Repauno Development Laboratory
RECEIVED for review February 11, 1965 Accepted llarch 19, 1965.
r
Determination of a Protonation Scheme ot Tetracycline Using Nuclea r M a gnetic Resonance NEIL E. RIGLER,' SASWATI P. BAG,' DONALD E. LEYDEN, JAMES L. SUDMEIER, and CHARLES N. REILLEY Depurfmenf o f Chemisfry, Universify o f North Carolina, Chapel Hill, N . C.
b Complete microscopic dissociation schemes of three members of the tetracycline antibiotic series have been determined using nuclear magnetic resonance. The compounds investigated are tetracycline, its 4-epimer, &d its quaternary methyl iodide. All studies were carried out in a solvent mixture of 50-50 w./w. methanol-water. Chemical shift data of selected nonlabile protons as a function of pH are applied to determine the distribution of protons among various sites on the molecule. These distributions, along with macroscopic dissociation constants obtained in the same media, are applied to calculate microscopic dissociation constants.
T
MACROSCOPIC pR, values of tetracycline antibiotics in aqueous solution are agreed to be approximately 3.3, 7 . 7 , and 9.7 ( I , 223, 1 2 ) . However! there is lack of agreement as to the assignment of the particular functional group3 giving rise to the observed HE
872
ANALYTICAL CHEMISTRY
macroscopic constants. For example, Stephens, Murai, I h n i n g s , and Woodward ( I S ) proposed a protonation scheme of tetracycline hydrochloride(1) in which successive dissociations occur at sites labeled A , B , and a combination of C and C'.
0 0
I
Leeson, Krueger, and Nash (9) lroDosed a reversal in assignments of the second and third pKa values of tetracycline on the basis of comparison with macroscopic I I K , ~values of the quaternary methylammonium salt (also referred to as methiodide). In addition, these authors pointed out the need for a more detailed study of the microscopic
dissociation constants. Determination of t'hese microscopic dissociation constants is of great value in a fundamental understanding of the chemical behavior of these compounds, particularly in the important area of metal complexation (S). Garrett (6) investigated the effect of dielectric constant on the apparent pK, values of tetracycline in dimethylformamide-water mixtures and suggested that the observed decrease in pK, values of neutrally-charged acids with increasing dielectric constant supports the reversal in assignment of the second and third pK, values of tetracycline. Because microscopic constants are directly related to the fraction of time that. a particular site is protonated, an experimental technique capable of monitoring the state of protonation of individual sites is required. Recent 1 Present address, Lederle Lab(iratories, Pearl River, N. Y. 2 Present address, Department of Chemistry, Jadavpur Vniversity, Calcutta, India.
7
2
3 8 , PPm
Figure 1. Partial NMR spectra tetracycline
of
htudies (IO, 1 4 ) have shown the particular alqilicability of nuclear magnetic resonance to the determination of protonation schemes EXPERIMENTAL
The proton magnetic resonance spectra were recorded using a Varian X-60 high resolution K X R spectrometer. Chemical shift data were obtained relative to t-butyl alcohol as an internal itandard. The reported values of the chemical shifts, however, are related to sodium 3-(trimethylsilyl)-propane sulfonate. abbreviated tnis*. The ambient temperature of the sample compartment' was 30' f 2' C. The precision of the chemical shift measurements was estimated to be better than 10.01 p.1i.m. In cases of low signal intensity, a repetitive scan technique was employed in order to differentiate more clearly between signal and background noise. pH values were measured a t 26" f 1' C. using a Leeds and Northrup Model 7661 line-operated pH meter equipped with Model 124138 microelectrode assembly and standardized using Sational Bureau of Standards buffers.
The pK,, values of compounds under investigation were determined potentiometrically in 50-50 w;w. methanolwater, using a semiautomatic recording device previously described (14). The titrant used was a standard 6JI C02-free potassium hydroxide solution (J. T. Baker, A R grade). The tetracycline compounds employed were analytical standards supplied by Lederle Laboratories and shown by paper chromatography and other techniques to be free of extraneous tetracyclines except, for less than 1% epimer. Solvents and other chemicals used for this investigation were 4~ grade. The particular solvent mixture used in these determinations was found to overcome best' the problems imposed by limited solubility of tetracyclines o\-er various p H values. Because the AY-methylgrouli resonance of the methiodide was obscured by the methyl resonance of methanol, this resonance was observed in aqueous solution. The concentration of solutions ranged from 0.015111 to 0.1M in tetracycline. RESULTS
A compleie assignment of the entire
NMR spectrum of tetracyl.ine was not attempted. Typical spe:(n as shown in Figure 1 exhibit two resonance signals which were of prime interest in this study. The most intense of these signals is a singlet occurring a t 3.17 p.p.m. for tetracycline hydrochloride in methanol-water. This sigr.al is attributed to the protons in the 4dimethylamino group. The other resonance signal of interest is a pattern appearing at, 6.6-7.6 p.p.rLi. and resulting from the i , 8, and 9-protons in the aromatic D-ring. The observed chemical shifts of these two resonance patterns depend upon pH for reasons previously discussed ('7, 10). X signal from the &methyl group is readily observed but its chemical shift value is not appreciably pH dependent. Figure 2 shows a plot of chemical shift values of the 4-dimethylamino singlet and of the first prominent resonance signal occurring a t low field in the D-ring pattern. Figures 3 and 4 show similar plots for the methiodide and the biologically inactive 4-epimer of tetracycline (fl), respectively. The solid lines were drawn by placing calculated curves symmetrically about the potentiometrically determined pK, values.
'3
r
36
34
32
B,pprn vs tms'
Figure 3. Chemical shift data of tetracycline methiodide at various pH values
chemical shift is a weighed average of chemical shifts of the two varieties. To correlate observed chemical shift data with fractional extent of protonation of individual sites, it is first necessary to ascertain the influence of complete protonation of each individual site upon the chemical shifts of adjacent nonlabile protons. The three acidic functional groups commonly observed in potentiometric titration of tetracycline have been proposed (13) to consist of the tiicarbonyl group labeled d in I, the dimethylammonium group labeled B , and the phenolic-@ diketone system, ivhich is a combination of C and C'. From potentiometric titrations of the methiodide, the presence of a fourth ionizable proton was demonstrated in the course of these investigations and
P"
DISCUSSION
Correlation of Chemical Shift Data with Microscopic Equilibria. Dif2
76
75
14 I
73 1
I
1
26
24
Figure 2. Chemical shift data tetracycline at various pH values
of
1
32 8,ppm
1
I
30 20 u5 t m s *
ferences in chemical shift of nonlabile protons adjacent to functional groups in protonated and nonprotonated forms arise from different electronic environments caused by inductive, solvent, steric, and other factors. Because of rapid exchange of labile protons with the solvent, the observed
76 I
7I 5
114
1 I3
1
1
1
32 30 2 8 8,ppm vs tms'
I
1
26
2.4
Figure 4. Chemical shift data of 4epi-tetracycline at various pH values VOL. 37, NO. 7, JUNE 1965
873
the pK. value of 10.67 confirmed by spectrophotometric measurements a t 385 mp. From structural considerations, the groul) labeled C‘ was assigned as the probable acidic site. Figures 2, 3 and 4 show that the total chemical shift change is approximately 0.65 1i.p.m. for the dimethylamino protons in tetracycline and 4-epi-tetracycline and 0.28 p,p.m. for the D-ring protons in the three compounds. Provided that the chemical shift change of dimethylamino protons is caused only by protonation of site B and that the chemical shift change of the D-ring protons is caused only by protonation at site C, the fraction of chemical shift change which occurs during addition of each equivalent of acid readily yields the concentrations of various species participating in the microscopic equilibria. From tetracycline data alone, the presence or absence of chemical shift contributions caused by protonation of remote sites is not readily ascertained. However, the methiodide proves to be valuable in elucidation of these effects by elimination of any protonation a t site B. Figure 3 shows that the total chemical shift change of the quaternary methylammonium group resonance is very small. Because this change could not be caused by dissociation a t site B , it was subtracted as a small correction from the observed chemical shift changes of the dimethylamino protons in the remaining two compounds. The magnitude of the change indicates the small effect of dissociation of sites A , C, and C’ upon the electronic environment of the dimethylamino protons and permits the use of chemical shift changes of dimethylamino protons as a monitor of the state of protonation a t site B. Conversely, dissociation of site B would be expected to have little influence upon the electronic environments of sites A , C, and C’. Thus, it is assumed that chemical shift changes of I)-ring protons are not influenced by dissociation of site
B. Similarly, it is important to know the magnitude of effects caused by dissociaTable 1.
A
1
0 33 19
4
tion a t sites A , B , and C’ upon the chemical shift of D-ring protons. Although the D-ring protons of the methiodide undergo chemical shift changes a t the first and second pK, values, the absence of any appreciable chemical shift change in the pK3 region corresponding to dissociation of site C’ is a useful observation. Because dissociation of site C’ has no appreciable effect upon the chemical shift of D-ring protons, the assumption is made that ionization of more distant groups such as A will also have no effect. I t follows from the above discussion that the chemical shift of methyl protons in the dimethylamino group is a direct measure of protonation a t site B , and the chemical shift of D-ring protons is a direct measure of protonation a t site C. I t is assumed that dissociation of site C’ of tetracycline and 4-epi-tetracycline is negligible in the pH range studied. This assumption is supported by two considerations. First, because the Dring protons are insensitive to dissociation of C’, the latter could only be detected by a corresponding decrease in dissociation of C. But the total chemical shift change of 0.28 1i.p.m. observed for D-ring protons of tetracycline is equal to that observed upon complete dissociation of site C in the methiodide. Secondly, the microscopic pK, value of the C’ site is expected to be somewhat greater in the case of tetracycline than in the case of the methiodide because of electrostatic repulsion from the more highly localized positive charge in the latter compound. Knowing the number of equivalents of acid added, the extent of protonation at site A can be found by
Per Cent protonation at Various Values of n
n
2 3
Figure 5. Microscopic acid-base equilibria of tetracycline
100
B Tetracycline 0 16 90 100
C
C‘
0 51
100 100 100 100
91 100
difference. In summary, dissociations a t sites B and C are determined directly from fractional chemical shifts of dimethylamino and D-ring protons, respectively, dissociation at site C‘ is neglected, and dissociation at site A is obtained by difference. Table I shows the per cent protonation of tetracycline, 4-epi-tetracycline, and tetracycline methiodide a t sites A , B , C, and C’ for various values of n, the number of equivalents of acid added to the completely dissociated anion. Differences in microscopic ionization schemes of the above compounds reflected by the data in Table I are often more readily evaluated by conversion of data into microscopic ionization constants. Calculation of Microscopic Dis-4s demonsociation Constants. strated by Edsall and Q7yman ( 5 ) , microscopic dissociation constants are readily calculated from per cent of protonation a t various basic sites and macroscopic dissociation constants. The primary advantage of microscopic constants is that, they provide a more familiar frame of reference for evaluation and comparison of acidic st,rengths. Figure 5 depicts all possible dissociation equilibria of tetracycline, the superscripts indicating the electrostatic charge on corresponding sites in I . Site C’ is omitted because it is assumed to be negligibly dissociated under the experimental conditions. The following calculation of kl illustrates the method of calculating microscopic ionization constants from fract’ional protonation data. At the pH value corresponding to pK,, the following relationships apply :
[AoB+Co]= [‘l-B+Co1 f [ A O B O C O ] [A0B+C-1 (1)
+
kl
=
[ H + ] [ A - B + C O ][AoB+CoJ
~~
Ki [ A-B +Co] [ AOB+C0 ]
[iiOB+CO] = 0.5A [.4-B+Co] = 0.5(1 - f 2 ~ ) i l
(2) (3) (4)
wherefZa is the fraction of the time the A site is protonate! a t n = 2 and A is the analytical concentration of solute. Substitution of (3) and (4) into (2) gives
which allows calculation of microscopic constants directly from data in Table I. Table I1 summarizes the microscopic constants calculated in this manner. GENERAL DISCUSSION
Tetracycline methiodide 1 2
3
874
0 17 100
ANALYTICAL CHEMISTRY
, . .
...
...
0
83 100
100 100 100
The influence of the state of protonation of a given site upon the dissociation of a neighboring site may be represented as a Apk value. As shown in Figure 5,
To ble II.
Microscopic Dissociation Constants
Tetracycline 4.49 5.40 5.45 8 00 8 51 7 09 7 29 7 55 7 24 9 11 8 60
Pkl
Pk2 wkr
Epitetracycline 4.91
Tetracycline methiodide 3.98
5.44
4.67
ing which serves to link sites B , A , C', and C in the manner indicated by 11.
+
a
...
I1
...
(I
7.41 8.36
TetraEpi- cycline Tetra- tetra- methcycline cycline iodide Apkascc") 2.60 . . .b Apkascc -1 1.37 1.09 , . b 3.06 2.50 3.05 APkAc(s+ ) 1.83 .. b APkAc(Bo) ApkBcu 1.84 . . .b APkBcca -) 0.60 . . .b a Indeterminate (negligibly dissociated). b Inapplicable due to absence of site B. O)
7.72
0
Apk Values
,
, . .
7.93
Table IIt.
7.03
A considerable degree of intramolecular H-bonding has been shown by means of x-ray data (4) to exist in the 9.45 . . . 10.80 crystalline hydrochloride salt and has 8.50 , . . 8 92 been found by infrared studies (8) to a Indeterminate (negligibly dissociated). exist in solution. An example of the resultant influence is that dissociation of site C is expected to strengthen the H-bond between the oxygen atoms pk12 is direct,ly proportional to the attached to Cll and CI2 carbon atoms niagnit,ude of free energy of dissociation and thereby decrease the acidity of site of site B when site A is dissociated, and C'. Another consequence of such an pk2 is directly proportional to the H-bonded system would be to make any magnitude of free energy of dissociation simple correlation between Apk values of site B when site A is undissociated. and coulombic interactions impossible, Therefore, the difference, pkl* - pkz, is as evidenced in the present study. a measure of the influence of the state I n addition, the magnitude of error of protonation of site A upon the discaused by neglecting any contribution sociation of a prot!on from site B . of H-bonding to observed chemical This measure of the interact'ion of site shift changes is uncertain. Therefore, A with site B is designated as A $ A B ( c o ) , further discussion must await more the superscript in ( C O ) indicating the direct evidence concerning intramocase where site C is protonated. Conlecular H-bonding in tetracyclines. versely, the effect of the state of proA detailed understanding of tetratonation of site B on the dissociation of cycline acid-base reactions must take a proton from site A is given by pkZl due account of the keto-enol equilibria p k l (when site C is undissociated) in the tricarbonyl system of ring A . and is identical in magnitude to the The relative abundance of the two effect of site A on site B . This may be enolic forms depicted in I11 and IV pk12 = deduced from the fact that p k l are expected to influence acidity of p k z p h ; hence, pk12 - pkz = pk21 various sites by altering distribution of phi. When site C is dissociated, the A-B interaction, designated A p k ~ ~ p ) , charges and strength of H-bonds. is given by the difference, pkI32 - pk32. Various interactions between sites are a
+
+
- pki A p k ~ ~ ( c -=1 pk321 - pk31
A P ~ A B ( C=~ p) h i
A ~ ~ A c ( B= + )pk31 - pkl
Apkaccao) = pk3a - pkz A ~ ~ B c ( A=- pk132 ) - pklz
0-
A P ~ A c ( B= ~ )pk321
These ApK values are the result of a number of effects including inductive, field (electrostatic), and resonance effects. I n the case of tetracycline, the magnitudes of interactions, as given in Table 111, follow the order A-C > A - B > B-C. This is a t variance with the order expected from a simple electrostatic model. The magnitudes of Apk values, particularly those related to A-C interactions, are surprisingly large in view of the large distance between these sites (greater than 9A). This is attributed to the high degree of conjugation between certain acidic sites and the extensive intramolecular hydrogen-bond-
I11 normal tetracycline I
I+
I IV epi-tetracycline I n addition, steric factors may influence acid-base behavior b y shifting this keto-enol equilibrium. X-ray crystallographic data have established that the A-ring of normal tetracycline is folded about the axis indicated by the dotted line in I11 (4). Considerations of Courtaulds and Dryding molecular
models have indicated that in 4-epitetracycline, the axis of folding shown by the dotted line in IV is preferred because it allows the bulky dimethylamino group to occupy an equatorial position, resulting in considerable decrease in the amount of steric repulsion by neighboring atoms and ring strain. The condition of planarity as a requirement for double-bonding would seem to indicate a shift in the enolic equilibrium from I11 toward IV as normal tetracycline is converted to epi. Therefore, a difference in conformation of the A-ring may be partly responsible for the striking difference in basicities of the nitrogen atoms (site B ) of the two epimers. ACKNOWLEDGMENT
The authors thank Lederle Laboratories for supplying the tetracycline compounds used in this work. LITERATURE CITED
(1) Albert, A., Nature 172, 201 (1953). (2) Albert, A., Reese, C. W., Ibzd., 177, 433 (1956'1. (3) Ddluisid, J. T., Ilartin, A. N., J . Med. Chem. 6 , 16 (1963). (4) Donohue, J., Dunite, J. D., Trueblood, K. X . ) Webster, 31. S.,J . Am. Chem. SOC.85, 851 (1963). (5) Edsall, J. T., Wyman, J., "Biophysical Chemistry,'' Vol. I, p. 495, Academic Preas, New York, 1908. (6) Garrett, E. R., J . Pharm. Sci. 52, 797 (1963). (7) Grlinwald, E., Loewenstein, A., Meiboom, S. J., Chem. Phys. 27, 641 (1957). (8) Kalnin'sh, K. K., Belen'kii, B. G., Dokl. Akad. Nauk. S S S R 157. 619 (1964). (9) Leeson, L. J., Kraeger, J. E., Kash, R. A , , Tetrahedron Letters, So. 18, 1155 (1963). (10) Loewenstein, A., Roberts, J. D., J . Am. Chem. SOC.82, 2705 (1960). (11) McCormick, J. R. D., Fox, S. M., Smith, L. L., Bitler, B. A., Keichenthal, J., Origoni, 1.. E., Muller, W. H., Winterbottom, R., Doerschuk, A . P., Ibid., 79, 2849 (1957). (12) Parke, T. Y.,Davis, W. W.,ANAL. CHEM.26, 642 (1954). (13) Ste hens, C. R., hlurai, K., Brunings, J., Woodward, R . B., J . Ani. Chem. SOC.78, 4155 (1956). (14) Sudmeier, J. L., Reilley, C. E., ANAL.CHEM.36, 1698 (1964). RECEIVED for review February 18, 1965. Accepted A ril 8, 1965. Ilivision of Analytical Ciemistry, 148 Jleeting, ACS Chicago, Ill., September 1964. Investigation sup orted by Public Health Service Research Erant RG-08349 from the Xational Institutes of Health. \ - - - - ,
E.
VOL. 37, NO.
9,
JUNE 1965
a
875
