Spectrophotometric Determination of 8-Hydroxypenillic Acid

complex ion with bromine at pH 8 may be due to the steric effect offered by the two ethyl groups and two large bromine atoms in the region of the nitr...
0 downloads 0 Views 552KB Size
Diethylamine EtzNH

+ -OCl+ Et&Cl+

EkNCl

-OH

+ H20 + 21- + + EhKH + C1- + -OH 12

Triethylamine

+

EtsN HOC1 Et&Cl+ 21-

+

+EtaNClf + -OH +I2 + Et3X + C1-

The excew hypochlorite is destroyed by reaction with nitrite ion: -0C1

+ NOz-

+

i C1Nos-

and the color development reactions may be represented by the reaction:

I-

+ Ip + SKI + 13---EIKI complex (blue)

The ratio of molar absorptivities for ethylamine, diethyls mine, and triethylamine predicted from the above reaction scheme is 1,5:1.0:1.0, and the observed ratio is 31,000:22,500: 14,000 or 1.4:1.0:0.62. The low value for triethylamine and for ethylamine at p H 8 is probably due to incomplete reaction with hypochlorite because of some prior equilibrium. The very high molar absorptivity for ethylamine at p H 8 in the presence of bromide ion may be accounted for by postulating the formation of the ionic species EtNBr3+ whose molar absorptivity should be 1.5 times t h a t for EtNCI?. Table I gives a value of 44,600:31.000 or 1.44 for this ratio, which is considered good agreement. At pH 6.6, apparently some equilibrium prevents the formaticn of the ethyltribromammonium ion. The fact t h a t diethylamine does not form a similar

complex ion with bromine a t p H 8 may be due t o the steric effect offered by the two ethyl groups and two large bromine atoms in the region of the nitrogen. Ammonia is not readily detected in the procedures a t p H 6.6 and 8.1 due to the destruction of the chloramines by hypochlorite and not due t o their reduction by nitrite ion. Evidence for this view is found when the hypochlorite reaction time is decreased to 15 seconds or less and the nitrite reaction is held a t 1 minute. The samples determined under these conditions showed strong positive tests, indicating the presence of some chloramine, while blanks under the same conditions were colorless. Ammonia is probably oxidized to nitrogen at the longer hypochlorite reaction times. The procedures described here are easily carried out and the results are reproducible. I n several determinations on each amine with different technicians preparing their own reagents and carrying out their own procedures, the molar absorptivities were reproducible to within 1% and an average error on amine analysis of less than 2y0 was obtained. The method has been used very successfully in the study of the kinetics of the hydrolysis of N-substituted maleamic acids (3) in which ethylamine and diethylamine were determined in the presence of their respective maleic acid amides. ACKNOWLEDGMENT

The author gratefully acknowledges the technical assistance of Nancy L. Simmerman and the valuable comments of Fred Zitomer.

LITERATURE CITED

(1) Barnes, H., “Apparatus and Methods of Oceanography,” p. 140, Interscience,

New York, 1959. (2) Clark, S. J., Morgan, D. J., Mikrochim. Acta, 1956, 966. (3) Dahlgren, G., Simmerman, N. L. (unpublished data). See Dahlgren, G. and Simmerman, N. L., Science 140, 485 (1963). (4) Diamond, L. H., Audrieth, L. F., J . Am. Chem. Soc. 77, 3131 (1955). (5) Ellis, A. J., Soper, F. G., J . Chem. Soc., 1954, 1750. (6) Hawkins, W., Smith, D. M., Mitchell, J., Jr., J . Am. Chem. SOC.66, 1662 (1944).’ (7) Kolthoff. Kolthoff, I. M.. M., Belcher. Belcher, R.. R., “Volu‘[Voliimetric Analysis, Andysis, ’ Vol. 111,” ’p, p. 581, 58i, Interscience, New York, 1957. (8)Kolthoff. (8) Kolthoff, I. M.. M., Sandell. Sandell, E. B., “Textbook of ’Quantitative Quantitative Analvsis.” Analysis,” p. n. 597, Macmillan, Macmdlan, New York, 1952. 1k2.’ (9) Lacy, W., Shemwell, R., Huckaby, J., Proc. Louisiana Acad. Sci. 18, 94 (1955). (10) Lambert, J. L., ANAL. CHEM. 23, 1247 (1951). (11) LiebhafRky, H. A., Bronk, L. B., Zbid., 20, 588 (1948). (12) Makris. K. G.. 2. Anal. Chem. 84. . 241 ( m i j . (13) Mitchell, J., Jr., Hawkins, W., Smith, D. M..J . Am. Chem. SOC. 66, 782 (1944). ’ (14) Noller, C. R., “Chemistry of Organic Compounds,” p. 233, Saunders, Philadelphia, 1951. (151 Rowe. R. A.. Audrieth. L. F.. J . Am. Chem. SOC.78, 563 (1956). (16) Van Hoogstmten, C. W., Rec. Trav. Chim. 51, 414 (1932). (17) Weil, I., Morris, J. C., J . Am. Chem. SOC.71. 1664 (1949). (18) Zitomer, F:, Lambert, J. L., ANAL. CHEM.34, 1738 (1962). (19) Zuman, P., Nature 165, 485 (1950). \

,

- I



RECEVIEDfor review June 6, 1963. Accepted November 7, 1963. Work supported by the American Cancer Society through grant P-272.

S pect rophot o metric Dete rminatio n of 8-Hydroxypenillic Acid DAVID A. HULL, JEAN M. SHEPP, WILLIAM J. WEAVER, ROBERT F. RESCHKE, and JOSEPH BOMSTEIN Brisfol Laborafories, Clivision of Brisfol-Myers Co., Syracuser N. Y.

b A method has been developed for determining 8-hydroxypenillic acid, based on isolating the silver salt, followed by reacting with vanillin to produce a visible chromophore. Precision is better than 3%. Effects of operating variables and interferences are discussed. Detailed procedures are given for determinations in fermentation media.

R

(1, 4 ) demonstrate that 8-hydroxypenillic acid (HPA) [The name 8-hydroxypenillic acid, for the enolic form of 3,3-diECENT REPORTS

-

-

methyl 8 - oxo - 4 - thia 1,7 - diazabicyclo - (3.3.0) - octane - 2,6dicarboxylic acid, has been used in conformity with Johnson and Hardcastle ( 4 ) ] can be isolated from Penicillium chrysogenum fermentations in which 6-aminopenicillanic acid (ilPA) is produced. In these reports, and in a paper appearing simultaneously (d), the authors show t h a t APB reacts with COz in water to form HPA in high yields. Thus, substantial degradation of .4PA may occur in the presence of dissolved C 0 2 in aqueous solutions and fermcntation media. For this reason a reliable method is required for determining

HPA in the presence of APA and impurities likely to be encountered. Ballio et al. ( I ) reported that HPA takes up six equivalents of iodine per mole, and that this reaction can be used for quantitative analysis. An iodine titration method is subject to serious limitations, however, because of the possible presence of other iodineabsorbing species, for example, the penicilloic acids of XPA and Penicillin G. A search of the literature revealed no other applicable methods. The technique we have developed depends on isolating HPA as the silver salt, followed by reacting with vanillin VOL. 36, NO. 3, MARCH 1964

599

Table I.

Precision of HPA Determination

Concn., pg./ml. 540 1080 1630

2170

Table 11.

HPA added, pg./ml. 560

2240

Repli- Std. dev. Rel. std. cations pg./ml. dev., % 13 15.5 2.9 5 20.6 1.9 6 14.9 0.9 12 14.1 0.7

Recoveries of Added HPA

HPA found, pdml. 560 570 2245 2220

Error, pg./ml. 0

+10 +5 -20

Rel. error,

7%

0 +1.8

+0.2 -0.9

to generate a chromophore which can be measured spectrometrically. It is applicable in the presence of APA, Penicillin G, and commonly occurring impurities. EXPERIMENTAL

Details given here are for determining HPA in experimental fermentation media : these represent the most complicated situation. For simpler mixtures the procedures may be shortened appropriately. Apparatus. A Beckman DU spectrometer equipped with 1-em. cuvettes was used throughout. Reagents. STANDARD HPA. The preparation of H P A for standardization follows the method outlined by Johnson and Hardcastle (4). APX (0.001 mole), NaHC03 (0.002 mole), and 5 ml. of water are placed in a Parr micro combustion bomb, which is then pressurized with 30 p.s.i.g. of C 0 2 overnight. At the end of this time the contents are transferred to a 150-ml. beaker, and the bomb is washed with 13 ml. of absolute methyl alcohol, and the washings are added to the beaker. Normal propyl alcohol is added to the beaker until a slight turbidity develops; the suspension is cooled in a n ice-bath for 30 minutes; then additional n-propyl alcohol is added until a total of 7 5 ml. has been used. The suspension is allowed to crystallize at room temperature for several hours, and is then filtered under vacuum, washed successively with n-propyl alcohol and acetone, and air-dried. The recovered solids are recrystallized from absolute methyl alcohol and n-propyl alcohol. Elemental analysis and Karl Fischer titration of a typical preparation showed C, 30.46; H, 4.81; N, 7.90; H20, 14.8%. Theoretical values for Na2. H P A . 3 H z 0 are C, 30.17; H, 4.47; N, 7.82; H20, 15.1%. n'ORKING STANDARD. Twenty mil& grams of HPA per 100 ml. of absolute methyl alcohol, prepared fresh daily. COLOR REAGENT.One half gram

600

ANALYTICAL CHEMISTRY

of vanillin (cap.) per 100 ml. of 50 ' HzS04, prepared fresh daily. v./v. % SILVER NITRATE. Ten grams of AgN03 (c.P.) per 100 ml. of water. Procedure. STANDARDCURVE. Two milliliters of HPA-free broth (obtained either by fractionating a n HPA-containing sample on Sephadex G-25, or by using broth in a n early stage of growth) is pipetted into each of four 50-ml. centrifuge tubes. Five, 15, and 20 ml. of working standard are pipetted into three of t h e tubes, and all four are diluted to 25 ml. with absolute methyl alcohol to precipitate protein. After 15 minutes, the tubes are centrifuged a t 1100 X G for 5 minutes, then decanted into 50-ml. beakers. Each tube is washed once with 5 ml. of alcohol. The suspension is centrifuged and the liquid decanted, combining the clear supernatant with the previous alcohol solutions, and discarding the precipitated solids. The apparent p H of each solution is adjusted to 3.0 with 1% nitric acid; 2 ml. of silver nitrate solution are added; and the solutions are stored in the dark for 15 minutes. At the end of this period of time, the suspensions are transferred completely to 50-ml. centrifuge tubes, washing with a few ml. of water, and then are centrifuged a t 1100 x G for 15 minutes. The supernatant layers are discarded, and the precipitates are washed twice with 5 ml. of water, centrifuging and decanting each time. The moist precipitates are dissolved in 10 ml. of 50 v./v. % sulfuric acid, transferred completely to 100-ml. volumetric flasks, and made up to volume with additional 50% acid. Ten milliliters of the acid solutions and 10 ml. of color reagent are pipetted into 25-ml. volumetric flasks, mixed, and heated in a boiling-water bath for exactly 5 minutes. The aolutions are cooled quickly to room temperature (ice-bath), and are diluted to volume with additional 50% acid. The absorbances of the resulting cherry-red solutions are measured within 30 minutes a t 490 mp, using as a blank the broth to which no HPA was added. -4plot of absorbance us. concentration of HPA is approximately linear a t least u p to 2000 pg. per ml., and passes through the origin. SAMPLES.Two milliliters of centrifuged sample are pipetted into a 50-ml. centrifuge tube containing 23 ml. of absolute methyl alcohol. The solution is mixed and allowed to stand for 15 minutes. The determination is continued as described for standardization, and the absorbance obtained is used to determine the concentration of HPA4 by interpolation in the standard curve. RESULTS

Reproducibility d a t a were obtained by replicating calibration standards prepared in a n HPAfree substrate and assayed on different days. Table I lists t h e d a t a , and shows that the precision varies from 2.9% a t 540 pg. per ml. to 0.7% at 2170 pg. per ml. Precision.

ok

Figure 1.

8-Hydroxypenillic Acid

Accuracy. T o establish accuracy in a completely satisfactory manner requires a blank containing APA a n d all normal contaminants, but free of HPA. The extreme complexity of fermentation media makes this a virtual impossibility. To obtain an estimate of accuracy, however, standard HPA was added to samples of mature broth, and recoveries were determined. The data of Table I1 show that the relative error is less than 2%. I n addition, because the standard curves are prepared in broth, their deviations from linearity constitute a check on recovery. One may plot the straight line passing through the origin and the first two points. The deviation of the line from the third point (highest level), computed from a large number of standard curves, showed a relative error of -1.8% at 2000 pg. of HPA per ml. I n a second approach toward establishing accuracy, we carried out radiochemical determinations by standard isotope dilution techniques. The radioactive HPA was prepared and purified as described for nonradioactive HPA, starting with .APA-S35. The latter compound was prepared by fermentation, and was isolated and purified by extraction and recrystallization. For six samples analyzed both spectrophotometrically and radiochemically, the radiochemical results were 20 to 30% higher than the spectrophotometric. The cause of this apparent bias was determined to be a conversion of APA to HPA by means of reaction with dissolved C02 during the extraction process. To prove this point, a solution with 2000 pg, of APA per ml. was prepared in water, then carried through the radiochemical procedures devised for broth. Of the APA present, 21% appeared as HPA. Identity of the HPA was confirmed by infrared spectrometry. It does not follow, however, that spectrophotometric procedures yield high results, because the radiochemical and spectrophotometric methods are necessarily different. The radiochemical technique, requiring preparation of a pure product, involves successive extractions with methyl isobutyl ketone, a ater, and butanol, followed by multiple recrystallization. During the shaking periods, exposure to COz must occur. The spectrophotoniptric method requires no extraction, and as shown above by recoveries on spiked samples of mature

APX broth, apparently does not lead to the conversion of APA to HPA. Based on the recovelies of added H P A and the extent of A P h conversion during extraction, the accuracy appears to be about 3 to 4%. DISCUSSION

The nature of the silver-HPh salt and of the vanillin brtsed chromophore have not been investigated in detail. Numerous workers, however, have studied and reported silver salts of nitrogenous bases. The conditions of formation of the chromophore suggest either a Schiff base car a condensation of the type encountered with tryptophan and vanillin. Streuli (8)has recently reviewed the subject of the formation of stable complexes of monov,tlent silver with nitrogenous bases in nonaqueous media, pointing out the stability of 2: 1 and 1:1 ligand: metal adducts. His potentiometric data demonstrated that 2: 1 complexes are generrtlly soluble, and that primary amines are soluble even at 1:l ratios. Secondary amines precipitate at the 1:l ratio, and tertiary amines precipitate a t all ratios. H e concluded that ,he formation of such complexes was related to the proton affinity of nitrogen compounds and depended strongly on pH. Katritzky and Lagowski ( 5 ) have pointed out that imidazoles behave as weak acids, forming insolu hle silver derivatives. The HP44 molecule shocvn in Figure 1 as the keto-enol tautomers suggested in (4),is a substituted imidazolidone, and appears to behaw according to the pattern described by Streuli and by Katritzky and Lagmski. Wet-ash determinations performed on R ashed silver-HPA precipitates show a 1:1 salt (25.1% of silver, compared to 28.2?, of silver theoretical). The salt is formed in a n essentially nonaqueous medium and the formation is pHsensitive. Mercury (Hg+*) also precipitates with HPA, but the precipitate is more difficult to l- andle physically. I n this case the recovered solids contained 42.17, of H g , compared to a theoretical value of 43.6% for the 1: 1 salt. Stability cohetants have not been determined. The development of the cherry-red color with vanillin mag be explained by one of t\yo well established reactions: splitting out of COz from the imidazolidone ring, as has bevn shown to occur by Nulvaney and Evans (?), followed by Schiff base formal ion between the primary amine thus formed and the aldehyde; or, condensation between the aldehyde and the umaturated imidazolidone ring (in the enol form) to form a quinoidal system, as has been reported

for the reaction of tryptophan with p dimethylaminobenzaldehyde (3). Operating Variables. Solvents evaluated for precipitating protein were methyl alcohol, ethyl alcohol, acetone, acetonitrile, a n d dioxane. Of these, t h e cleanest separations a n d best recoveries were achieved with the alcohols. Methyl alcohol was

Table IV.

HPA concn., rg./ml. 3.44 6.88 10.32

Table 111.

Effect of Volume of Color Reagent

vel. added, 2 4 6 10 20

Absorbance 0.462 0.695 0.818

1.290 1.240

Chromogens Evaluated

Vanillin 490 mp 0.260 0.525 0.780

preferred t o ethyl alcohol for reasons of economy. Silver nitrate solutions ranging from 10 to 40y0were evaluated as precipitants a t p H 3.0. Absorbances obtained after reacting with vanillin did not differ on passing from 10 to 20%, but showed 5 and 8% decreases a t the 30 and 40% levels. The p H was varied over the range 2.0 to 5.0, holding the silver nitrate concentration a t 10%. Maximum absorbance was found a t pH 3.0, but no substantial differences occurred between 2.5 and 3.5. The concentration of the vanillin reagent was based on the level chosen for the tryptophan-vanillin reaction by Kraus (6). The temperature for color development, 100°C., was chosen arbitrarily as providing a n easily reproducible condition. Heating times up to 1 hour were considered, but within this period no absorbance plateau wae reached. Color intensity was linear with time for a solution of purified H P A in water and of H P A in broth. Five minutes was chosen for convenience, adequate sensitivity, and good reproducibility. Under these conditions of time, temperature, and concentration, the absorbances developed from solutions of H P A (20.85 pg. per ml.) were measured as a function of the volume of vanillin reagent added. I n each case the final volume was adjusted to 25 ml. with additional 50QJ,sulfuric acid. The data are listed in Table 111, and show that a maximum is reached a t about 10 ml. The colors developed are stable for at least 30 minutes, but stability was not tested beyond that period of time. The actual increment in absorbance after 30 minutes at room temperature was 1.5%. Other aldehydes were evaluated for color formation with HPA, using the same conditions as for vanillin. Table IV lists the data obtained for standard

Absorbance p-DAB 430 mp 464 mp 0.327 0.257 0.667 0.505 0.997 0.760

Table V.

p-HBA 393 mp 0,223 0.457 0.683

Substances Tested for Interference

Material APA Penicillin G Chloride ion Kitrate ion Penicilloic acids Silver ion excess

Color produced h-one None Iione Red-purple None None

solutions of HP.l reacting with vanillin, p-dime t h ylamino benzald e h y d e ( p D AB) and p- h y d r o x y b en z al d ehyde (p-HBA). I n addition to these chromogens, three others were con-

sidered-p-dimethylaminocinnamaldehyde, 4-hydroxybenzaldehyde-3-sulfonic acid, and 2,4,6-trihydroxybenzaldehyde. Of these, the first two caused precipitation, and the third was very insensitive. Among the three apparently satisfactory reagents, p-HB-4 showed a small advantage in sensitivity, and vanillin had the advantage of an absorption maximum furthest removed from the absorption usually encountered with broth impurities. All gave linear plots of concentration us. absorbance. There appeared little reason to prefer either vanillin or p-HBA. Interferences. Substances possibly present in samples, due either t o their natural occurrence or to introduction during t h e analysis, were tested for interference. Table V lists t h e results, showing t h a t serious errors will be incurred unless nitrate ion is removed during t h e washing of t h e silver-HPA precipitate. As detailed in the method described above, two washes are sufficient for this purpose. ACKNOWLEDGMENT

The authors express their appreciation to D. H. Phillips for preparing XPA-S35, the starting material for HPA-S35. VOL. 36, NO. 3, MARCH 1964

0

601

LITERATURE CITED

(1) Ballio, A., Chain, E. B., Dentice diAccadia. F., Mauri. M.. Rauer, K.. Schlesinger, h. J., ‘ Schlesinger,’ S.; Nafure 191. 909 (1961). (2) Batchelo;, F. R., Gazeard, D., K’ayler, J. H. C., Ibzd., 191,910 (1961). (3) Feigl, F., “Spot Testa in Organic

Analysis,” 5th ed., p. 279, Elsevier, New York, 1956. (4) Johnson, D. A., Hardcastle Jr., G. A., J. Am. Chem. SOC.83. 3534 (1961). (5) Katritzkv, A. R., J.agowiki, J. M., “Heterocvclic

Chemistrv.” I,

D. r

(7) Mulvaney, J. F., Evans, R. L., Ind. Eng. Chem. 40, 393 (1948). (8) Streuli, C. A., “Titrimetric Methods,” DD. 97-120. Plenum Press. New York. 1961.

__

227.

Wiley, A$w York, 7960. (6) Kraus, I., J . Biol. Chem. 63, 157 (1925).

RECEIVEDfor review August 22, 1963. Accepted December 16, 1963.

Micro and Semimicro Differential Thermal Analysis (PDTA) CHARLESMAZI~RES Ecole Nafionale SupGrieure de Chide, Universitd de Paris, France

b Through a critical study of the various factors involved in differential thermal analysis, apparatus has been designed which permits examinationin controlled atmosphere and between -180” and +1200° C.-of samples to weighing from 1 to 200 pg. 2 X 10-4 gram) or, alternatively, from 0.1 to 10 mg. (10-4 to 10-2 gram). Thermal effects as small as 1 O+ cal. can be detected. Examples are given showing uses of the apparatus.

the peak, even in the ideal case of an isothermal transformation. (2) The thermocouples generally used as sensors only indicate their own temperature, which may be notably different from the temperature of the sample, especially if the thermocouple is chemically insulated. The approach described herein consists essentially in using a sample as

3 2

C

mineralogists, and solidstate physicists have shown increasing interest in differential thermal analysis (DTA) (7). The advantages and disadvantages of the method, as well as the experimental solutions to various problems, have been the object of numerous articles (4). The present author has given particular attention (6) to the following points: (1) the existence of a thermal gradient between the surface and the core of the sample investigated, resulting in a variable degree of completion of the physicochemical phenomenon giving rise to the thermal effect. This, in turn, results in a broadening of

.lmm L

Figure 1. Micro DTA detecting head

602

I

HEMISTS,

ANALYTICAL CHEMISTRY

difficulties mentioned above, enclosing the sample within the thermocouple was an improvement on conventional design; as pointed out by one of the referees of the present paper, the effect of sample shape and/or sample thermal diffusity on the area of the peak or on its shape is thus minimized. On the other hand, Wittels (9) showed that by using the vacuum apparatus of Whitehead and Breger (8), it was possible to analyze samples of calcite as small as 300 pg. However, the thermal effect involved in the decomposition to CaO is quite large; once more, the sensitivity was probably 0

1

2

3

4

4

:

:

:

:

5cm. i

not to scale

Figure 2. head

Semimicro DTA detecting

1, platinum junction crucible; 2, removable lining; 3, lid

small as possible (down to 1 X gram) to obtain a better definition of its temperature. The micro sample is placed completely inside the junction of the detecting thermocouple which results in optimum use of the thermal effect involved and in excellent identification of the temperature of the sample and of that of the thermocouple. Because of the small mass of the samples involved, a very high sensitivity is essential: this is obtained by foregoing the conventional metal or ceramic block often used to homogenize the temperature. The thermocouple sample cup was first suggested by Herold and Planje ( 8 ) ; the samples they investigated probably weighed about 1 gram (no data are given) ; the thermocouples were embedded in a refractory block. Though this design did not solve all the

11Figure 3.

-

12

High temperature design

1 , 2, water cooled base; 3, refractory sheath; 4, refractory support; 5, 6, platinum base and hood; 7, furnace; 8, refractory six-duct sheath; 10,13, O-rings; 1 1,12, gas inlet/outlet