CLIN. CHEM. 20/7, 794-801 (1974)

794 CLINICAL CHEMISTRY, Vol. 20, No.7, 1974

Mechanisms of the Liebermann-Burchard and ZakColor Reactions for Cholesterol

R. W#{149}Burke, B. I. Diamondstone, R. A. Velapoldi, and 0. Menis

Correlation of SO2 and Fe2+ measurements withnew spectral data indicates that the Liebermann-Burchard (L-B) and Zak color reactions for choles-terol have similar oxidative mechanisms, each yield-ing, as oxidation products, a homologous series ofconjugated cholestapolyenes. These studies furthersuggest that the colored species observed in thesetwo systems are enylic carbonium ions formed byprotonation of the parent polyenes. Thus, the red(Amax, 563 nm) product typically measured in theZak reaction is evidently a cholestatetraenylic cat-ion, and the blue-green product in the L-B reaction(Amax, near 620 nm) is evidently the pentaenyliccation. The effects of rate of carbonium ion forma-tion and sulfuric acid concentration on sensitivity andcolor stability are discussed. A solvent extractionprocedure Is described for specifically convertingcholesterol to 3,5-cholestadiene. Incorporating thisstep into the typical L-B method can increase theL-B sensitivity for cholesterol by several fold.

Additional Keyphrases: cholestapolyenes #{149}conjugateddouble bonds #{149}carbonium ions from polyenes #{149}fac-tors affecting color #{149}SO2 and Fe2+ measurementequivalent ratios

Cholesterol reacts with various strong acids of theBronsted and Lewis types to give colored products.Although these reactions have been used empiricallyfor many years for the qualitative and quantitativedeterminationof cholesterol,their mechanisms stillare not clearly understood.

Among the many color reactions for cholesterol,the Liebermann-Burchard probedure is perhaps themost widely used. This reaction was described ini-tially by Liebermann (1) in 1885 and applied to cho-lesterol analysis shortly after by Burchard (2). Chlo-roform was used as a solvent in the early studies, but

Analytical Chemistry Division, Institute for Materials Research,National Bureau of Standards, Washington, D. C. 20234.

Received Jan. 4, 1974; accepted May 7, 1974.

the Liebermann-Burchard (L-B) reaction, as per-formed today, is carried out in an acetic acid-sulfU-ric acid-acetic anhydride medium. The other widelyused method, the Zak reaction, which was first ap-plied to cholesterol analysis by Zlatkis, Zak, andBoyle (3) in 1953, is carried out in acetic acid-sulfu-ric acid in the absence of acetic anhydride. In thisreaction, however, Fe3+ must be added to obtain thedesired colored species.

In one of the earliest mechanistic studies, Lange eta!. (4) showed that treating chloroform solutions ofcholesterol with equivalent amounts of perchioric orhexafluorophosphoric acid resulted in the formationof colorless sterolium salts. These salts hydrolyzedinstantaneously on contact with an excess of water,givingquantitativerecoveryofthe added cholesterol.Addition of acid in excess of the stoichiometricamount resulted in slow dissolution of the colorlesscrystals, followed by formation of purple productsthat were considered to be halochromic salts ofcholestadiene. Degradation of these products wasthought to occur through formation of polymerizeddieneoid hydrocarbons, with the trimer being thehighest polymer observed. Subsequent studies byDulou et al. (5) and by Brieskorn and Capuano (6)suggestedthat the initialstep in the L-B reactionwas the dehydration of cholesterol to form cholesta-diene, which dimerized to bis-cholestadiene. The finalproduct was believed to be the monosulfonateddimer. Later studies by Watanabe (7) supported thishypothesis; he isolated a 3,3’-bis-3,5-cholestadieneby column chromatography from a concentrated L-Breaction mixture. However, a subsequent paper byBrieskorn and Hofmann (8) indicated that dimer for-mation was not the principal reaction in the L-.B sys-tem and that a more probable mechanism involvedthe oxidation of cholesterol to a conjugated penta-ene. It is at this stage that efforts to elucidate themechanism of the L-B reaction appear to havestopped.

N2.

CLINICAL CHEMISTRY. Vol. 20, No.7, 1974 795

In contrast to the investigations of the L-B reac-tion, little, if any, systematic study has been made ofthe mechanism of the Zak reaction. The brief men-tion (9) that the reaction appears to be oxidative isapparently the only reported information on the na-ture of this color reaction.

We have quantitatively measured SO2 and Fe2+formation as a functionof reactiontime,to demon-strate that the L-B and Zak reactions have similaroxidative mechanisms. These data, in conjunctionwith solvent extraction, uv-visible spectrophotome-try, and mass spectrometric measurements showthat these oxidation reactions lead to the formationof a series of conjugated cholestapolyenes. Moreover,evidence is presented that shows that the coloredspecies comprising these two systems are the corre-sponding enylic carbonium ions of the respectiveconjugated polyenes. On the basis of these studies,we propose that the red product typically measuredin the Zak reaction with Amax at 563 nm is a cho-lestatetraenylic cation while the blue-green product

in the L-B reaction with Amax near 620 nm is apentaenylic cation.

Materials and Methods1

Reagents

Solutions of cholesterol were prepared from Na-tional Bureau of Standards high-purity cholesterol(NBS SRM 911).

3,3’-bis-3,5-Cholestadiene was obtained from Co-lumbia Organic Chemicals Co., Columbia, S. C.29209.

2,4- and 3,5-Cholestadiene, 5,7-cholestadien-3fl-ol,and 7-cholesten-3i3-olwere purchased from Steral-oids, Inc., Pawling, N. Y. 12564.

Purified pararosaniline hydrochloride for the de-termination of SO2 was obtained as a 2 g/liter con-centrate (Item No. 64327) from the Hartman-LeddonCo. (Harleco), Philadelphia, Pa. 19143.Bathophenanthroline was obtained from the G.

Frederick Smith Chemical Co., Columbus, Ohio43223.Cyclohexane used in the extractionstudieswas

“Spectro”quality.All other chemicals were ACS reagent grade.

Apparatus

Absorbances were measured with a Cary Model 14recording spectrophotometer (Varian/Instrument

Division, Palo Alto, Calif. 94303).Mass-spectrometry was done with a Du Pont

Model 21-491 medium-resolution mass spectrometer(E. I. du Pont de Nemours, Instrument Products Di-vision, Wilmington, Del. 19898).

1 In order to adequately describe materials and experimental

procedures, it was occasionally necessary to identify commercialproducts by manufacturer’s name or label. In no instances doessuch identification imply endorsement by the National Bureau ofStandards, nor does it imply that the particular product orequipment is necessarily the best available for that purpose.

Eppendorf pipets (Brinkmann Instruments, West-bury, N. Y. 11590) were used for all microliter-scaletransfers.

Extractions were performed in 60- and 125-ml sep-aratory funnels equipped with Teflon stopcocks andstoppers (Kontes Glass, Vineland, N. J. 08360).The 30-ml midget impingers forcollectionof SO2

were also obtained from Kontes Glass, Vineland, N. J.08360.

Procedures

Determination of sulfur dioxide. Sulfur dioxideproduced in the L-B reaction was determined spec-trophotometrically with pararosaniline. This pro-cedure, developed by West and Gaeke (10), wassubsequently perfected by Scaringelli et al. (11) fordeterminations of SO2 in air. Figure 1 shows a sche-matic diagram of the apparatus used in our study forgenerating and collecting SO2.

The pear-shaped reaction vessel was made from a10-ml separatory funnel to which was attached aside-arm bubbling tube that extended to approxi-mately 3 mm from the bottom of the vessel. Thisvessel was connected in series with two midget im-pingers, each filled with 10 ml of potassium tetra-chloromercurate S02-absorbing solution (40 mmol/liter) (11). To determine SO2, we added 5 ml offreshly prepared L-B acid mixture [acetic anhydride:glacial acetic acid:sulfuric acid (10:5:1 by vol),cooled to 25#{176}C]to the reaction vessel followed by100-300 tl aliquots of a 5 mg/ml solution of choles-terol in acetic acid. The vessel was quickly closedand its contents thoroughly mixed by swirling. Sul-fur dioxide was then continually swept from the L-Bmixture by purging with nitrogenat the rate of 25ml/min. At designated intervals ranging from 0.5 to22 h, purging was discontinued and the contents ofthe two midget impingers were combined and dilut-ed with the absorbing solution to 25 ml, in volumet-ric flasks. Appropriate aliquots were then analyzedspectrophotometrically for SO2 by the procedure ofScaringelli et al. (11).

Fig. 1. Schematic diagram of apparatus for generationand collection of SO2 in the Liebermann-Burchard reac-tion

Table 1. Rate and Stoichiometry of SO Formationin the Liebermann-Burchard Reaction at 25 #{176}C

(505 zg of Cholesterol)

SO,, g

348711

1396

Zak

Equiv. ratio

SO,/chol.

4.184.244.17

Equiv. ratio

Fe’ ‘/chol.5.045.024.95

50.5101202

796 CLINICAL CHEMISTRY, Vol. 20, No.7, 1974

Initial experiments in which three impingers wereused indicatedthat 98% of the SO2 was collectedinthe firstimpinger while no SO2 was found in thethird impinger. Accordingly, we used only two im-pingersinsubsequentexperiments.

Determination of Fe2+ formed in the Zak reaction.The acidconditionsand Fe3+ concentrationused forcarrying out the Zak reactionwere thosespecifiedbyBoutwell (12): a glacialaceticacid to sulfuricacidratio of 1.5:1 and a 4.4 x 10 mol/liter Fe3 con-centration. A 50-200 tl aliquot of a standard solu-tion of cholesterol (1 mg/ml in acetic acid) wasadded to 13 ml of this mixture in a 60-ml separatoryfunnel. The contents of the funnel were quicklymixed by vigorous shaking. One-milliliter aliquotswere taken at prescribed intervals over a period of0 .5-5 h, to be analyzed for Fe2.

The spectrophotometric bathophenanthroline pro-cedure described by Pollock and Miguel (13) wasused to determine Fe2+ in the presence of Fe3+. Thekey featureinthismodifiedprocedureisthe additionof phosphate ion which complexes Fe3+ and so effec-tively removes it from the Fe2+_Fe3+ equilibrium.Failureto use thismasking reaction will lead to highresults for Fe2+, because the reagents used in thespectrophotometricprocedure also reduce FeS+ toFe2.

Extraction studies. After initial tests with a seriesof solvents, cyclohexane was selected for subsequentstudy of the extraction of cholesterol and its reactionproducts because of its superior transmission in theultraviolet. We used two variations in the fundamen-tal procedure. In the first, cholesterol in acetic acidwas added directlyto variousmixturesof aceticandsulfuric acids, which were then extracted with cyclo-hexane. In the second, cholesterol was dissolved incyclohexane and this solution was equilibrated withthe acid mixtures by mechanical shaking.

Results and Discussion

In both the L-B and Zak reactions, our data indi-cate that cholesterol is oxidized in steps, with eachoxidationstepyieldinga cholestapolyenehaving onemore double bond than the compound from which itwas derived. Further evidence is presented to showthat the active oxidants in these two systems areSO3 and Fe3+, respectively, and that measurementof the corresponding reduction products-i.e., SO2and Fe2+_can yield valuable information regardingthe rates and relative efficiencies of these two reac-tions. Such measurements have now been carried outusing the analytical procedures and conditions de-scribed in the Materials and Methods section andconstitute one of the significant aspects of this re-port.

Some typical results obtained for SO2 and Fe2are presentedinTables 1,2,and 3.The lastcolumnin each of these tables is headed “equivalent ratio,”which is defined as the ratio of the number of equiv-alents of SO2 or Fe2+ found to the number of equiv-

Reaction time, Eqizlv. ratioh SO,, zg SO,/chol.”

0.5 52 0.621 144 1.732 2213 2584 2796 304

22 349

SO, blank 10,g/22h

2.653.103.343.65

4.18

#{176}In Tables 1-3, “equivalent ratio” refers to the ratio of number of equivalents of SO, or Fe2 found to the number of equiv.alents of cholesterol added.

Table 2. Rate and Stoichiometry of Fe2 Formationin the Zak Reaction at 25 #{176}C(101 hgof Cholesterol)

Reaction time, Equiv. ratio

h Fe’, ,g Fe2’/choI.

0.5 90.1 3.071.0 102.6 3.501.5 114.2 3.922.0 122.5 4.192.5 131.7 4.503.0 136.3 4.694.0 143.0 4.925.0 146.5 5.02

Fe2 blank = 0.5 1hg

Table 3. SO, and Fe’ Formation as a Function ofCholesterol Concentration (25 #{176}C)in the

Liebermann-Bu rchard and Zak ReactionsCholesterol, Time,

Reaction h

505 22L-B 1010 22

2020 22

,g Fe”

5 73.55 146.55 289

alents of cholesterol initially added. Because a two-electron transfer is required for generation of a dou-ble bond, the equivalent weight of cholesterol in thisinstance is one-half its molecular weight. The calcu-lated values of the equivalent ratios are numericallyequal to the average number of double bonds formedper cholesterol molecule, and they will be used inthis context throughout this paper. In Table 1, forexample, it is seen that after 0.5 h the equivalentratio obtained for the L-B reaction with cholesterolis 0.62. As shown later, although this reaction time isoptimum for spectrophotometric measurement under

478nm

HOAc/H2 $04

LIEBERMANN-BURCHARD

Ac20

($03)

Co,’bonlum Ion of 3.5-Die,,.

563nm

ZAK

Fe’3

Dienylic ColionXm0n4I2NM (CoIc. 396NM)5

+ Fe +2TrenyIic Cation

X,n,,n47SNM (Cole. 473 NM)5

Tetroenylic Cation

X,, 563ttM (CaIc.549NM)5

WAVELENGTH, rim

SOZOHd +s02Ct,oI.stah.no.n. Sulfonic Acid

X,, 4IONM (CoIc. 418NM)5 *

Fig. 2. Absorbance spectra of a Zak reaction as a func-tion of time, showing, initially, decrease in the 412-nmpeak as a peak appears at 478 nm, and subsequent con-version of 478-nm absorbing material to the product ab-sorbing at 563 nm where the color is typically measured

Fig. 3. Proposed mechanisms of the L-B and Zak reac-tions* Sorensen, T. S., J. Amer. Chem. Soc. 87, 5075 (1965); ** Fieser, L. F.,and Fieser, M., Steroids. Reinhold. New York, N. Y., 1959. chap. 1

CLINICAL CHEMISTRY, Vol. 20, No. 7, 1974 797

the conditions used, the conversion of cholesterol tothe requisite cholestapolyene (four generated doublebonds) is, at best, only one-seventh complete. Simi-lar data are shown for the Zak reaction in Table 2.Again, a 0.5-h reaction time is optimum for spectro-photometric measurement. In this case, however, theconversion of cholesterol to the appropriate cholesta-polyene (three double bonds generated) is potentially100% efficient.

The data in Table 3 show that the equivalent ra-tios calculated for the L-B and Zak reactions areconstant over fourfold changes in cholesterol concen-tration. Such reproducibility definitely implies thatthe reaction pathways are more explicit than pre-viously expected. In addition, the constancy of theseequivalent ratios also indicates that these color sys-tems should obey Beer’s law; that they in fact do sohas been well established.

As further evidence of the stepwise nature of theoxidation processes, it will be helpful to examinemore closely the absorbance characteristics of theZak reaction. This reaction was chosen in preferenceto the L-B system because it better illustrates thesignificant points. A series of absorbance spectra, ob-tained over a period of about 1 h for a typical Zakreaction, is shown in Figure 2. Three absorbancepeaks are observed, with Amax at 412, 478, and 563nm. The peaks at 478 and 563 nm have been de-scribed in an earlier paper (12); the 412-nm peak,however, has not been reported previously. This ini-tial peak disappears after the first few minutes of thereaction and can best be observed by making eitherof the following modifications to the fundamentalprocedure: (a) by carrying out the reaction in a solu-tion containing a lower proportion of acetic to sulfu-ric acid, i.e., 1:1 HOAc/H2S04 rather thap the 1.5:1ratio usually used, or (b) by lowering the tempera-ture of the reaction to 10#{176}C.The serial nature of

peak formation, together with the well-defined isos-bestic points at 427 and 503 nm, clearly show thatthe final steps in the Zak color reaction consist of aproduct that absorbs maximally at 412 nm convert-ing to a 478-nm absorbing species which, in turn,yields the peak at 563 nm that is routinely mea-sured. As discussed in a subsequent paper (14), therates of these reactions are strongly temperature de-pendent, increasing rapidly with increasing tempera-ture.

The data presented thus far suggest that the prod-ucts observed in the Zak reactions are carboniumions formed by the successive protonation of the ho-mologous series of cholestapolyenes proposed pre-viously in this report. This theory was strengthenedby noting the close similarity between our observa-tions and the spectral data reported by Sorensen(15) for a homologous series of aliphatic polyenyliccations. In the latter study, the structures of the par-ent polyenes and their corresponding carbonium ionswere proven unequivocally by NMR and ultravioletspectroscopy. These data in conjunction with ourequivalent ratio, solvent extraction, ultraviolet-visi-ble and mass-spectrometric measurements have ledto the mechanisms proposed in Figure 3 for the Zakand L-B color reactions with cholesterol. Accordingto these two mechanisms, both reactions have acommon initiation step, i.e. protonation of the -OHgroup in cholesterol and subsequent loss of water to

798 CLINICALCHEMISTRY, Vol. 20, No. 7. 1974

give the carbonium ion of 3,5-cholestadiene. Evi-dence will be presented subsequently that indicatesthat formation of this species is the first step in thecolor reactions. Serial oxidation of this allylic carbo-nium ion by either Fe3 or SO3 yields the cholesta-polyene carbonium ions shown, together with equiv-alent amounts of Fe2+ or SO2. Notably, no cationicspecies are given in Figure 3 for the L-B system, ingoing from the allylic to the pentaenylic cation, be-cause none were observed spectrally under the exper-imental conditions. The acidity of the L-B reactionis relatively low, however, and thus would not nor-mally favor stable carbonium ion formation. Theability to observe the pentaenylic species can there-fore be attributed to two factors: (a) sufficiently ex-tended conjugation, and, perhaps more importantly(b) the stabilizing effect of the terminal cyclopentylring, as reported by Sorensen (15).

Observed and calculated values of the absorbancemaxima are reported for each of the structuresshown in Figure 3. The calculated values were ob-tained from Sorensen’s empirical equation (16): Xmax= 319.5 + 76.5n where, for our purposes, n rangesfrom 1 (dienylic cation) to 4 (pentaenylic cation).

Having established the sequences of reactions inFigure 3, it is worthwhile to re-examine the useful-ness of the equivalent ratio measurements discussedpreviously. To do so it must be noted that maximumabsorbances of the peaks at 563 and 620 nm of theZak and L-B reactions are obtained typically in30-40 mm at 25#{176}C.In the Zak reaction it is seen thatthe generation of the desired tetraenylic cation re-quires the formation of three additional doublebonds in the cholesterol molecule. Reference toTable 2 shows that after the reaction has proceededfor 30 mm, an equivalent ratio of Fe2 to cholesterolof 3.07 is obtained, indicating that within this timethe formation of the tetraenylic cation is potentiallyquantitative. On the other hand, the equivalent ratioof SO2 to cholesterol obtained in the L-B reactionafter 30 mm is only 0.62 (Table 1). This ratio is sub-stantially less than the minimum value of 4 requiredfor complete conversion of cholesterol to the pen-taenylic cation. On this basis the L-B procedurewould be expected to yield about one-seventh theconcentration of the product desired for spectropho-tometric measurement, as compared to the Zak reac-tion. Assuming similar molar absorptivities for thetwo colored species, this difference would account forthe well-known fact that the Zak procedure is aboutsevenfold as sensitive as the L-B method. This in-creased sensitivity may be attributed largely to thestabilizing effect on enylic cation formation of thehigher sulfuric acid concentration. In general, in-creasing the sulfuric acid concentration would be ex-pected to improve the stability of each of the carbo-nium ions formed in the stepwise oxidation of choles-terol, thereby making it much more likely that onecould observe carbonium ion formation in the Zakreaction than in the L-B reaction. This hypothesis

agrees well with the fact that we saw no absorbancepeaks in the visible spectrum of the L-B mixturepreceding the appearance of the 620-nm peak nor-mally measured. The subsequent conversion of thecompound absorbing at 620 nm to the product ab-sorbing at 410 nm (Figure 3) can be reasonably justi-fied on the basis of the agreement of the experimen-tal absorption peak with the value calculated by theFiesers’rules (17) and from the SO2 data obtained.The formation of this type of structure is furthersupported by the observation that the intensity ofthe 410-nm peak increases continually with time andthe yellow product formed is not extractable into im-miscible organic solvents.

Recently obtained mass-spectrometric data furthersupport the mechanisms proposed in Figure 3. Inthis study, we quenched the reaction of cholesterolwith Zak reagent by rapidly dispersing the acid mix-ture in excess alkali. The sample was then extractedwith cyclohexane and mass spectra were obtained di-rectly on the extract. Characteristic peaks werefound at mass/charge ratios of 368, 366, 364, and362-peaks that have been assigned to cholestadiene,cholestatriene, cholestatetraene, and cholestapen-taene, respectively. The largest fractions presentwere the triene and the tetraene. This distributionwas to be expected, because the color reaction wasquenched when the absorbance of the 478-nm peakwas nearlymaximal.

A finalobservationsupportingthe proposed mech-anisms is found in a recent investigation of the ki-netics of the Zak reaction (14). In that study theconsecutive pseudo-first-order rate constants areshown to be directly proportional to the square ofthe Fe3+ concentration.This isexactlythe expectedrelationshipifthe functionof Fe3+ is to generatedouble bonds for,in doing so,two electronsmust betransferred for each new bond produced.

Finally, we also studied the general applicabilityof equivalent ratio measurements in determining theextent of the reactions of other steroids in the L-Band Zak procedures. A series of Fe2+ and SO2 deter-minations were made on 5,7-cholestadien-3-ol and7-cholesten-3f3-ol, two compounds that react quite dif-ferently in these color systems. Cook (18) has shownthat these steroids give about twice and four timesas much color, respectively, with the L-B reaction asdoes cholesterol. Henry (19), on the other hand, hasreported that these same compounds give only abouta third as much color in the Zak reaction as doescholesterol.As expected,SO2 was generatedsignifi-cantlyfasterforthesecompounds than forcholester-ol while,conversely,the ratesof formation of Fe2+were much slower. In the latter system, minimun

reaction times of 5 h are required to obtain the req-uisite values of 3 for the equivalent ratios. Thesetimes can be compared to the 30 mm necessary forcholesterol (Table 2). Thus, measurement of equiva-lent ratios may provide a useful tool for studyingwhy compounds similar to cholesterol react differ-

with a synthesized dimer, however, showed that itsTable 4. Absorbance Characteristics of absorbance peaks were at longer wavelengths

Cholesterol Solutions (100 ma/liter) in VariousAcetic Acid-Sulfuric Acid Mixtures at 25 #{176}C = 298, 312, and 323 nm) than the finger-like absorbance peaks observed (Xmax = 269, 280,Isosbestic

HOAc: Peak Shoulder point 295, and 312 nm). Moreover, in these and similarH,SO,

(by vol) studies, the dimer was characterized by its very low1:1 ‘-..‘300(very strong); - solubility and reactivity in all of the acid mixtures

412 used. In view of these properties and the fact that1.5:1 300, 348#{176} - 325, 377 previousmechanisticstudies (4,5,7) were undertaken

2:1 300 peak showing - 380 by using gram amounts of cholesterol, it is not diffi-269, 280, 295 and cult to understand how the dimer was readily isolat-312 fine structure; ed and thus considered to be a reaction intermediate348#{176},412,438

rather than a degradation product. Furthermore, our3:1 269, 280, 295, 312 438 -

412, 500 proposal that the colored species measured in theL-B and Zak procedures are carbonium ions of cho-

5:1 267, 280, 296, 500 312, 385, 412 -450, 610 lestapolyenes is also consistent with the isolation of

10:1 385, 440, 480 - - dimer, because one of the expected modes of disap-pearance of these ions would be through dimeriza-

Peak continually decreasing.

CLINICAL CHEMISTRY, Vol. 20, No.7, 1974 799

nanometers

-‘450, 500

ently in terms of the colored products formed.Although aceticand sulfuricacidsareused in dif-

ferent proportions in the L-B and Zak procedures,the study of cholesterol reactions in these simpleacid systems provided valuable insight into themechanisms of these reactions. One of our first ex-periments was to measure the absorbance spectra ofcholesterol in different mixtures of these two acids.Relative HOAc/H2SO4 ratios ranging from 1:1 to

10:1 were used. In general, those mixtures in whichthe proportion of sulfuric acid was high led to prod-ucts that absorbed in the ultraviolet, while those inwhich the proportion of acetic acid was high showedincreased absorbance in the visible range. These ob-servations are summarized in Table 4.

One of the most noteworthy features of this studywas the similarity between the absorbance spectra ofcholesterol in 1.5:1 HOAc/H2S04 and in the corre-sponding Zak mixture, in which the same ratio ofthese acids is used. Thus the decreasing peak at 348nm and the increasing peak at 412 nm in HOAc/H2S04 behave similarly to the decreasing 412-nmand increasing 478-nm peaks in the Zak procedure(Figure 2). According to the mechanisms proposed inFigure 3, the conversion of the compound absorbingat 348 nm to the product absorbing at 412 nm is in-terpreted as the cation of 3,5-diene being oxidized todienylic cation. Because no Fe3+ is present, we canonly suggest that the oxidation is due to residualconcentrations of SO3 present. However, we did notattempt to measure SO2 formation in this system.

A second distinguishing feature of this study wasthe finger-like peaks observed near 300 nm, whichwere especially apparent in 3:1 and 5:1 HOAc/H2SO4 mixtures. Initially,we thought that thesepeaks reflected the possible formation of the 3,3’-bis-3,5-cholestadiene dimer. Subsequent experiments

tion.To determine if any intermediate products could

be isolated by solvent extraction and identified by uvspectrophotometry, we extracted the pre-reactedacidmixtures ofcholesterolwith cyclohexane.Aceticacid, which is also partially extracted and which hasan ultraviolet cutoff at about 250 nm, was removedby backwashing the extracts several times withwater or dilute base. The absorbance spectra werethen observed to have several characteristic absorb-ance peaks. Only 2,4-cholestadiene (Amax = 266, 275,and 287 nm) was identified with certainty. More-over, the orange color present in the prereacted acidlayer before extraction was also retained in thatlayer after extraction.

In a modification of this procedure, a second seriesof extraction experiments was performed in whichcholesterol was first dissolved in cyclohexane andthese solutions were equilibrated with the acetic-sul-furic acid mixtures. In this case, spectrophotometricexamination of the extracts showed that only 3,5-cholestadiene was formed (Xmax = 228, 235, and 243nm), and we saw no color in the acid phases. In bothof these studies the excellent review by Dorfman (20)on the ultraviolet absorbance of steroids proved ex-tremely valuable. These two extraction experimentssuggest that the precursor to color formation in boththe Zak and L-B reactions is the initial formation ofthe carbonium ion of 3,5-cholestadiene, as shownbelow:

[+c] c#{244}

The nonclassical resonance structure involving thebridgehead C-5 carbon contributes significantly tothe stability of this ion. The fact that this dehydrox-

z0(0

w>z0f-fI-.zwU

Lu0.

EQUILIBRATION TIME (hrs.)

Fig. 4. Specific conversion of cholesterol to 3,5-cholesta-diene by a two-phase equilibration technique; cholesterolin cyclohexane (20 mg/liter; phase ratio 1:1; temp., 25#{176}C

tO 12 8 22

800 CLINICAL CHEMISTRY, Vol. 20, No.7, 1974

ylation reaction goes easily is clearly demonstratedin the second series of extraction studies. Eventhough cholesterol is dissolved initially in cyclohex-ane, the -OH group is sufficiently polar to be proton-ated at the interface. Subsequent loss of water and aproton yields the 3,5-diene that is observed.

The carbonium ion of 3,5-cholestadiene is shownconclusively to be the reactive intermediate by thefollowing two experiments: (a) If the reactions ofcholesterolin acetic-sulfuricacid mixtures arequenched and the cyclohexane extracts then exam-ined spectrophotometrically, at least 80% of thediene formed isapparentlythe 3,5-derivative.(Thequenching was done by rapidlydispersingthe acidmixtures into excess base 2-3 mm after the choles-terol was added.) (b) Similar studies, starting with2,4-cholestadiene, show that it also is largely con-verted to the 3,5- compound, or its carbonium ion,upon dissolution in strong acid. Such acid-catalyzedre-arrangements are not unexpected, however, as evi-denced by the work of Allan et al. (21) in the relatedtriterpenoid series, where the mobility of doublebonds under acidic conditions has been clearly dem-onstrated.The acid-catalyzeddehydroxylationof cholesterol

yields predominantly the carbonium ion of 3,5-cholestadiene, together with smaller amounts of 2,4-cholestadiene. All evidence obtained thus farsuggests that this carbonium ion is the precursor tothe color reactions observed in the Zak and L-B sys-tems. In strongly acidic media such as are used inthe Zak reaction,cholesterolisrapidlydehydroxylat-ed, and thisis not a limitingfactorin the overallrate of the reaction.Consequently,cholesterol,3,5-cholestadiene,and 2,4-cholestadieneallform colorand generate Fe2+ at similar rates under these con-ditions. If the acidity is decreased, however, as is thecase in the L-B system,the rateof dehydroxylationbecomes significant.Startingwith 3,5-cholestadiene,for example, the amount of product absorbing at 620nm formed in the L-B reactionisabout fourfoldthatobtainedwith cholesterol.To demonstrate the effectof acidityon the rateof

dehydroxylation, we equilibrated cyclohexane solu-tionsof cholesterolwith acetic-sulfuricmixtures ofvarious proportions and spectrophotometrically de-termined the 3,5-cholestadiene formed. The resultsobtained are shown in Figure 4. The rate of forma-tion and the yield of 3,5-diene are greatest at thehighest sulfuric acid concentration. The use of anHOAc/H2SO4 ratioof 10:1,typicalofthe L-B condi-tions,produces dramatic decreasesin the rate andthe yieldof 3,5-cholestadiene.The relativeratesob-tained on these two-phase systems are believed to besignificant, although the absolute rates are not di-rectly applicable to the reactions occurring in theacid phase. Moreover, it should be noted that thistwo-phase equilibrationtechnique providesa simplemeans of improving by several fold the potential sen-

sitivity of the L-B reaction. From direct studies of

3,5-cholestadienein the L-B reaction,we concludethatthe maximum increasein sensitivity-assumingcomplete conversion of cholesterol to the 3,5-deriva-tive-is fourfold.

In conclusion,new evidencehas been presentedtoindicatethat the Liebermann-Burchard and Zakcolor reactions have very similar mechanisms. Fol-lowing an acid-catalyzeddehydroxylation,the 3,5-cholestadiene and (or) its cation are subjected to anumber of step-wise oxidations by either SO3 orFe3+, therebyforming a homologous seriesofcholes-tapolyene carbonium ions. The stability of each of

these new cationicspeciesisprimarilya functionofthe number of conjugated double bonds in the par-ent polyene and the sulfuricacidconcentration.Thesteroid concentration is also important, because highconcentrations can lead to competing reactions suchas the dimerization reported by Watanabe (7).Therefore, throughout this study, we tried to usecholesterolconcentrationstypicalof those used inspectrophotometric analyses. This limitation pre-cluded the use of NMR spectroscopy, a techniquecommonly used to identify intermediates and verifythe existence of carbonium ions.Further work isunderway to synthesizeand char-

acterize the cholestapolyene intermediates discussedin this paper. Additional studies are intended to ex-amine the relative reactivities and the mechanismsof the reactions of the cholesterol-like compoundsroutinely found in sera. Of special interest will bethose compounds which yield variable amounts ofcolor in the L-B and Zak reactions.

We express our appreciation to Dr. Harry S. Hertz, of the NBS,for making the mass-spectrometric measurements.

CLINICAL CHEMISTRY, Vol. 20. No.7, 1974 801

References

1. Liebermann, N. C., Uber das Oxychinoterpen. Ber. 18, 1803(1885).2. Burchard, H., Beitraegezur Kenntnis des Cholesterins. Chem.Zentralbl. 61, 25(1890).

3. Zlatkis, A., Zak, B., and Boyle, A. U., A new method for directdetermination of serum cholesterol. J. Lab. Clin. Med. 41, 486(1953).

4. Lange, W., Folzenlogen, R. G., and Koip, D. G., Colorlesscrystalline perchloric and hexafluorophosphoric acid salts of ster-ols.J. Amer. Chem. Soc. 71,1733(1949).5. Dulou, R., Chopin, J., and Raoul, Y., Structures of the two bi-cholestadienes formed in the Salkowski test for cholesterol. Bull.Soc. Chim. Fr. 18, 616 (1951).

6. Brieskorn, C. H., and Capuano, L., Color reaction with triter-penes and sterols. Chem. Ber. 86, 866(1953).

7. Watanabe, T., Coloured products and intermediates of Lieber-mann-Burchard reaction with cholesterol and its mechanism.Eisei Shikenjo Hokoku 77,87(1959).8. Brieskorn, C. H., and Hofmann, H., Beitrag zum Chemismusder Farbreaktion nach Liebermann-Burchard. Arch. Pharm. 297,37(1964).

9. Zak, B., Simple rapid microtechnic for serum total cholesterol.Amer. J. Clin. Pathol. 27,583(1957).

10. West, P. W., and Gaeke, G. C., Fixation of sulfur dioxide asdisulfomercurate(II) and subsequent colorimetric estimation.Anal. Chem. 28, 1816 ( 1956).

11. Scaringelli, F. P., Saltzman, B. E., and Frey, S. A., Spectro-photometric determination of atmospheric sulfur dioxide. Anal.

Chem. 39, 1709 (1967).12. Boutwell, J. H., Jr., Serum cholesterol. Clin. Chem. 10, 1039(1964).13. Pollock, E. N., and Miguel, A. N., Determination of iron(II)in the presence of thousand-to-oneratio of iron(ffl)using batho-phenanthroline. Anal. Chem. 39,272(1967).

14. Velapoldi, R. A., Diamondstone, B. I., and Burke, R. W.,Spectral interpretation and kinetic studies of the Fe3 -H2S04(Zak) procedure for determination of cholesterol. Clin. Chern. 20,802 (1974).15. Sorensen, T. S., The preparation and reactions of a homolo-gous series of aliphatic polyenylic cations. J. Amer. Chem. Soc.

87,5075(1965).16. Sorensen, T. S., Dienylic and polyenylic cations. In Carboni-um Ions, II, G. A. Olah and P. v. R. Schleyer, Eds., Wiley Inter-science, New York, N. Y., 1670, p815.17. Fieser, L. F., and Fieser, M., Steroids, Reinhold, New York,1959, chap. 1.18. Cook, R. P., Reactions of steroids with acetic anhydride andsulphuric acid (the Liebermann-Burchard test). Analyst (Lon-don) 86,373(1961).19. Henry, R. J., Clinical Chemistry: Principles and Technics.Hoeber, New York, N. Y., 1964, p 860.20. Dorfman, L., U’traviolet absorption of steroids. Chem. Rev.53,47(1953).21. Allan, G. G., Fayez, M. B. E., Spring, F. S., and Stevenson,R., Triterpenoids. Part XLVII. The constitution of some com-pounds obtained by the dehydration of fl-amyrin and related al-cohols: Further observations on the stereochemistry of a-amyrin.J. Chem. Soc. 456(1956).