Detection and Estimation of Steroidal Sapogenins in Plant Tissue MONROE E:. U L 1 . L C . KOLASD EDDY, .\IhRIAN L. McCLENSLK, 4 \ i ~U 4 R Y E. KLU\IPP Eastern Regional Research Laborutory. Philadelphia 18, Pa. 1ri the course of screening hundreds of plant samples for steroidal sapogenins, it became imperative to
find a procedure that would rapidly detect and estimate these compounds. 1 literature search indicated that a rapid and specific procedure was not available. It was found that a negative hemolytic test on alcoholic plant extracts was definite proof of the absence of steroidal saponins. Samples giving positive hemolytic tests were acid hydrolyzed and acetylated, yielding a crude residue. Steroidal sapogenins coiild he detected, and the quantity present estimated from the highly specific infrared absorption spectra of all sapogenins. By this procedure, a large number of plant samples (40 to 50 by tw-o workers in a 40-hour week) can be screened rapidly for steroidal sapogenins. Consequently, the rcarch for these valuable precursors for cortisone and sex hormones has been conducted at a great]?iricrcased rate.
I
S RECEKT years steroidal sapogenins have attracted con-
siderable attention as precursors ior sex and cortical hormones ( 3 - 5 ) . This laboratory has been engaged for some time in a comprehensive survey of the plant kingdom in an attempt to find good sapogenin sources. The sapogenins are not found tree in nature, but occur in a combined glycosidal form called saponins ( 1 ) . Steroidal saponins are not n-idespread. Many plants contain little or no saponin; in others the triterpenoid saponins predominate. Because the methods available for large scale isolation and identification of steroidal sapogenins are tedious ( 4 , 6 ) . a procedure u hich would rapidll- eliminate samples containing little or no steroidal sapogenin v-ould be useful. Although numerous reports on hemolytic methods for "saponins" have been published, a literature search shon ed that no specific method \?-as availalile (1).
I rapid microprocedure is reported here. Saponins are detected by hemolytic methods (1). If the sample gives a pofiitive test, the saponin? are hydrolyzed, and steroidal sapogenins in the crude mixture are detected and estimated hy means of their specific infrared absorption spectrn. PHOC~DIJRE
Extraction. WET S ~ W L F S . -1 400- to 500-gram saniplr of wet plant tissue is put through a Ball and Jeivell grinder with a 1-inch screen. T n o hundrcd prams of the ground tiwie arc covered with appro\imately 500 ml. of 95% ethyl alcohol ancl reflu\;& for 15 to 30 minutes. The sample is thcii cooled. filtixied. washed, and made t o a volump of I liter n i t h 9 5 5 ethyl alcohol. DRYSAMPLE&. Fifty grams, ground to pass an 80- to 100mesh screen, are covered Kith 300 ml. of 80% ethvl alcohol and refluxed for 1 hour. The sample is then treated the same as the et sam le, except that it is adjusted to a final volunie of 500 nil. with 8 O k ethyl alcohol. Much smaller quantities of both wet and dry samples can be used with proportionately smaller volumes of solvent. Sampling errors can be serious with small quantities of plant tissue, and hence larger samples should be used if available. Hemolytic Detection of Saponins. BLOOD STASDARDIZ {TIOY Ten t o 20 ml. of whole human blood are suspended in 100 ml. of 0.857 aqueous sodiuin chloride solution. thc, suslwn-ion is
centrifuged and thc sunrrnatant liquid is dccanted. The process The blood corpuscles are then suspended in sodium chloride solution. Ten milliliters of the turbid suspeiision and 1 ml. of digitonin solution (10 mg. of pure digitonin in 100 nil. of 80% ethyl alcohol) are mixed in a 15-mI. conical centrifuge tube. The mixture is kept a t room temperature for 5 minutes, and then visually compared with a tube of untreated blood suspension. If complete hemolysis has occurred, the tube containing digitonin will be entirely clear. This can he checked by centrifuging. The hemolyzed solution should give little or no red corpuscle precipitate. Within the authors' experience, complete hemolysis by 1 ml. of digitonin does not occur under these experimental condit,ions. Hence the stock blood suspension is progressively diluted with 0.85% sodium chloride solution until 10 ml. of the blood s u p pension are completely hemolyzed a t room temperature within 5 minutes. I n this manner, different blood samples can be roughly standardized to give equivalent hemolysis with digitonin. The stock solution can be kept for 1 to 2 weeks under refrigeration. One milliliter of the plant extract DETECTIOS O F S~POXIKS. described above is added to 10 ml. of standardized blood suspension. After 5 minutes. the presence or absence of hemolysis is observed as discussed under blood standardization. Samples giving a negative tcst are discarded. Positive extrarts are used for detection of steroidal sapogenins. Isolation of Crude Sapogenins. An aliquot of the alcoholic estract equivalent t,o 5.0 grams of original plant material (moisture-free basis) is evaporatcd just to dryiicw on a steam hath. The product is transiwred with a total of 5 ml. of 50cc ethyl alcohol t o a reaction tube ivith an over-all length of 13.0 rm. and an outside diamctw of 1.6 cni. The reaction tube is essentially a test tube provided with a 3 14/35 outer joint. Five milliliters of henzcne saturated with 50% ethyl alcohol are added to the concentrated estract, in the reaction tube. The ayers are vigorously shaken, and the tube is centrifuged a t 1500 to 2000 revolutions per minute for 1 minute. The benzene layer is drawn off, and the benzene extraction is repeated? The benzene extracts containing fats, pigments. etc., are discarded. Sufficient concentrated hydrochloric acid is added to the residual estract to make t,he resultant, acid concentration approsimately 4 S. One or 2 nil. of benzene (nlcohol-saturated) are added, and the tube is at,tached to a sniall condenser proinner joint,. The tube is immersed in a vided with a 141'35 water bath kept a t 78" to 80" C. for 2 hours. The tube is then withdrann from the bath and cooled, 5 mI. of benzene containing 10% ethyl alcohol ;ire addcd, and the layers are vigorously mixed. The tube is centrifuged, and the benzene layer withdrawn. The benzene est,raction is repeated twice. The combined benzene extracts containing crude sapogenin are placed in B small beaker and evaporated to dryness on the steam bath, 2 nil. of acetic anhydride are added, and the inixture is gently boiled for severn.1 minutes. The acetylated crude sapogenin is transferred to a 15-niI. centrifuge tube n i t h 5 ml. of benzene, 5 ml. of methanol saturated with potassium hydroxide are added, and the contents are mixed vigorously. Immediately 5 ml. of water are added and rnixetl well, and the tube is centrifuged. The henzene layer which separates is withdrawn, and the residual aqueous methanol is tq-ice re-ext,racted Kith benzene The combined benzene layers containing the crude sapogenin acetate are placed in a tared beaker and eviiporated to drynees on the steam bath. The beaker is then dried to constant \%-eightin a vacuum oven at 110". Samples Kith a repidue weight of less than 10 mg. are discarded. Samples betn-een 10 and 100 mg. are dissolved in carbon disulfide, 5 nil. of carbon disulfidr are added, and the beaker is cautioudy warmed until complete solution of tmhecrude acetate takes place. The solution is cooled and filtered through a microfilter into a 5-in]. volumetric flask. The filter is xaehed with carbon disulfide until the solution is up to volume. If the crude acetate v-eight is in excess of 100 mg., a larger volume (10 ml.) is used. Occasionally if a sample is difficultly soluble in carbon disulfide. chloroform is used as the solvent. Infrared Determination. Tht: infrared spectrum of thc ~ o l u -
1337
ANALYTICAL CHEMISTRY
1338 tion, relative t o the pure solvent, is obtained from 800 to 1050 em.-’ (12.5 t o 9.5 microns) in a 1-mm. cell and examined for the presence of the four characteristic sapogenin absorption bands. (-4Reckman llodel IR-3 infrared spectrophotometer was used in these studies.) Barids near 852, 900, 922, and 987 em.-.’ (11.75, 11.1, 10.85, and 10.14 microns), with the 922 hand stronger than the 900 band, indicate a “normal” sapogenin. Bands near 866, 900! 922, arid 982 c m - l (11.55) 11.1, 10.85, and 10.18 microns), \.rith the 900 hand stronger than the 022 haiid. indicate an ‘liso” sapogenin. If the bands a t 900 and 922 c1n-l are not much different in strength and the remaining Imnds lie between the values given above, the sample contains sapogenins of both configurations. All four bands should he present hefore it is concluded that the sample contains significant amounts of sapogenin, because isolated Innds of other substances ran occur in this region of the spectrum.
IO
WAVELENGTH, MICRONS 10.5 II 11.5
amount of original plant material--for esample, if the extract from 5 grams of plant material is contained in 5 ml. of final s o h tion, the concentration is called 1000 grams per lit’er. If the sample is found to contain primarill- “iso” sapogenins, the corrected absorptivity is divided b>- 0.53 liter per gram per centimeter (the corrected absorptivity of diosgenin acetate) to obtain the n-eight fraction of is0 sapogenins in the original plant material. If the sample is found to contain primarily “normal” sapogenins, its corrected allsorptivity is divided by 0.63 liter per gram per centimeter (the corrected absorptivity of sarsasapogenin acet,ate),t o obtain t,he weight fraction of normal sapogenins in the original plant material. If the sample contains both types of sapogenins (rarely the case in the authors’ esperiencej, a rough estimate of sapogenin content is obtained by dividing the corrected absorptivity by 0.58 liter per gram per centimeter. Because the straight-line backgrouiid correction, described above, cannot he an exact representation of the conhibution of the nonsapogenin material t,o the ahsorption curve, this estimate of sapogenin content must be considered as only approximate if the plant estract contains ~ul)stantialamount,s of nonsteroidal materials.
As a numerical example, curve n of Figure 3 has an absorption hand a t 982 em.-’ with a transniittance of 22.30j0. On opposite sides of this band are transmittance maxima at 970 and 997 em.-’ A straight line drawn between these two maxima crosses the 982 em.-’ coordinate at a transmittance of 66.3%. Corrected absorbance
=
log 66.3 - log 22.3
=
0.473
Cell thickness was 0.109 cm. The extract from 5 grams (moisture-free basis) of original plant material had been tlisWAVE LENGTH, MICRONS
IO
I
1000
I
950 900 WAVE N U M B E R , cm-1
I
a 50
10.5 I
I1 I
11.5 I
12
J
Figure 1. Infrared Spectra of Sarsasapogenin ( a ) and Smilagenin ( b ) Acetates Concentration 10.0 grams per liter; solvent. carbon disulfide
For an approximate determination of the quantity of sapogenin in the sample, the 982-987 mi.-’ ahsorption band is used, with a correction for background due to impurities and nearby absorption bands. .?, straight line is drawn between the two points of maximum transmittance on opposite sides of the 982-987 em.-’ band. .knother straight line perpendicular to the frequency axis is drawn through the point of minimum transmittance of this band. The absorbance value a t the intersection of these two straight lines is subtracted from the absorbance of the point of minimum transmit’tance, and the corrected absorptivity is calculated from this at)soi,liance difference. [iibsorptivity (or estinction coefficient) equals ahsorbance (or optical density) divided by concentration and by cell thickness.] I n calculating the absorptivity, the concentration used is t,hat of the equivalent
2otI
I I 900 0 50 WAVE NUMBER, cm-I Figure 2. Infrared Spectra of Diosgenin ( a ) , Dihvdrodiosgenin ( b ) , and Cholesterol (c) Acetates
I O 1 983 1 1000 950
Concentration approximately 10 grams per liter; solvent carbon disulfide
J
V O L U M E 24, NO. 8, A U G U S T 1 9 5 2
1339
rolved in 10 ml. of final carbon disulfide solution. Hence concentration = 5/0.010 = 500 grams per liter. Corrected absorptivity =
WAVELENGTH, MICRONS
9.5
0.473 = 0.00868 0.109 x 500
10.5
IO
I
11.5
II I
I2
I
Since the 900 em.-' band is qtionger than the 920 em.-' band, this is an is0 sample and should be compared with the reference :ilxorptivity of 0.53 liter gram-' cm.-l
Thu$ the original plant m;iterial contained I .6y0ipo snliogeiiiii~ on a nioipture-free basis. caI(:ulxted as diosgenin acetate. DISCUSS103
Iletection and estimation of steroidal sapogenins in plant tissue are essentially a three-phase process: detection of t,he steroidal saponins viliich are the glycosidal precursors of the sapogenins, ipolntion of the crude sapogenin acetate, and infrared detection of steroidal sapogenins. These three phases are discussed in turn. Detection of Saponins. Steroidal saponins are soluble iri aqueous alcohols. Under experimental conditions, which a r e i):i,setl on larger scale studies at t'his laboratory ( 6 ) , TO to 80%
I
982 * O V
I
lo'ld,0
Id00
9 h WAVE N U M B E R ,
I'igirre 3.
a. b.
68 1 284 13!2 393 s2.4
.50
8li.i
11
!IOU
206
80
dl1
113
18
$182
393
11,an 11 11 10 X'? 10 17
8fi.i
!I00 924
72 4 230 140 34 I
11 II 10 10
Sot
887 'J0 I Ylll 9 RL'
i u r r e r t i x l for backpir,:,n,l al~.t>ri)ti. Sat,urated, noncarbonyl sapogenins lack any specific cheniical character’lstic useful for ident,ification purposes, and their unsaturated or ketonic analogs are but slight,ly bett,er in this respect. The infrared absorption spectra of sapogenin ac:.ates, however, reveal some striking characteristiw. -1s shown in Figures 1 and 2 and Table I, all is0 sapogenins have characteristic absorption bands near 866, 900, 922, and 982 c m - l (11.55, 11.1, 10.85, and 10.18 microns). Sarsasapogenin (spirostaii-:3&ol), which has a “normal” configuration, hns bands at 852, 900, 922, and 987 c111.-~(11.75, 11.1, 10.85, and 10.14 microns). \\-it,h normal sapogenins, the 922 band has a stronger abeorptivit\- t,haii the 900 hand. I n is0 eapogenins this relationship is reversed. Figure 1 demonstrates this relationship for sarsasapogenin and iimilagenin, n-hich differ only in the configuration at ( 2 2 2 . The four characteristic infrared absorption bands are apparently a property of ring F. This is demonstrated by Figure 2, which gives the infrared absorpt,ion of the acetates of diosgenin ( A6-22-isospirosten-3p o l ) , dihydrodiosgenin ( A6-22-furosten-3@26-diol), and cholesterol (16-cholesten-38-ol). Thwe compounds
9.5
WAVELENGTH, MICRONS 10.5 II 11.5
IO I
I
I
12
I
Y
z“4
60
I-
sr 5z
50
4
0:
+
40
30
20 10
1050
Figure 3.
U
I050
Figure 4.
I I 1000 950 900 WAVE NUMBER, Cm-I
850
Infrared Spectrum of Crude Normal Sapogenin .4cetate, 0.45%
I 1000
950 900 WAVE N U M B E R , Cm-l
850
Infrared Spectra of Nonsteroidal Substances
sample. The four cliaracterist’ic sapogenin bands do riot depend upon the presence of the acetate group. The free sapogeriiiis themselves also show these four bands. From the data it is apparent that the characteristic infrared sapogenin absorption is due to the presence of ring F. No other known naturally occurring compounds have the sapogenin ring st,ructure. Hence the method is specific. The bands show u p in a clear-cut manner in the crude sapogenin preparations. The wave lengths are exactly the same as those of the pure reference compounds. The absorption spectra of most crude preparations indicate whether the sapogenin is is0 or normal. Because sarsasapogenin is the only common normal sapogenin, the absorption spectrum is a fairly good test foi, the presence or absence of this compound. Figures 3 . 1 , and 5 show some typical infrared absorption curves that indicate positive and negative tests for sapogenins. Curves a, h, and c (Figure 3) are is0 sapogenins; Figure 3, a, is a high s:iniple (1.6% moisture free basis), b intermediate (0.2%), and c low (0.08%). The typical sapogenin peaks are clear even with low samples such as Figure 3, c. Figure 4,a, shows a similar normal sapogenin (0.45’%moisture-free basis). Curves a, b, and c (Figure 5 ) show typical spectra for plant estracte containing no st.eroida1 sapogenins. Index marks show
V O L U M E 24, NO. 8, A U G U S T 1 9 5 2 the wave nunibers a t nhich the is0 and normal sapogenin band systems should occur. It can be seen that none of them possess the characteristic system of four sapogenin absorption bands, although some contain one or t n o bands a t nearly the same frequencies. Curve c is the spectiurn of a triterpenoid sapogenin. Such a compound could not be mistaken for a steroidal sapogenin because it does not have the entire system of four bands. However, if a substantial amount of such substance occurs mixed with a small amount of qteioidal sapogenin in a plant extract, it can lead t o an erroneouslv high estimate of the amount of steroidal sapogenins because of its absorption a t 982 cni.-l Certain other substances found in plant materials also have xeak absorption bands near 982-987 cm.-’ but do not have the complete system of four sapogenin bands. Therefore, although the infiared method appears to give positive evidence of the presence or absence of steroidal sapogenins, the estimatc of percentayc of sapogenin is only semiquantitative if considerable amount. of nonsteroidal material are present. Severtheless, it is the ouly method now available which will give even an approximate a ~ a on y a crude material. Two worhers, onp czrr?ing out the heiiioly,is and crude sapo-
1341 genin isolation steps and the other the infrared procedure can handle 40 t o 50 samples in a &hour n eek. ACKNOW LEDG.\IEYT
The authors wish to thank A. E. Jones for assistance in various phases of this research, and T. D. Fontaine for the sample of dihydrodiosgenin. LITERATURE CITED
(1) KoAer, L., “Die Saponine,” Vienna, J. Springer. 1927. (2) Kofler, L., and Raum, H., Hiochem. Z.. 219, 335 (1930). (3) Marker, R . E.. and Applerweig. S . , C‘liem. Eng. S e w s , 27, 3348 (1949). (4) Narker, R. I-:., Wagner, R . R.. L-lshafer. P. K., JVittbecker, E. L., Goldsmith, D. P. J., a n d Iluof, C. H., J . A m . C’hem. Soc.. 69, 2167 (194ij. (5) Roserikranz. C;., Pataki, J., and Djcrassi, C., Ibid., 73, 4055 (1951). (6) Wall, hI. E., Krider, RI. hl., liothman, E. S., and E d d y , C . R.. S.Bio2. Chem., in press.
R E C ~ I \for T Dreview Febriiary 10. 1052. Accepted I I a y 23, 1952. in a series on steroidal rapugenins: for paper I see (6).
.Second
Microdetermination of Chromium in Blood H. J. CAHNMA” AND RUTH BISEN Cancerigenic Research Laboratories, .Yational Cancer Institute, 4ictional Institutes of Health. Rethesda, .\Id. In the course of investigations of lung cancers among chromium workers it became necessary to hav e a method for the quantitativ e deterinination of traces of chromium in blood. As the chromium coiicentration of most blood saniples which had to be analyzed w-a8 in the order of magnitude of 0.01 to 0.1 p.p.m., a method with a high degree of sensiti\ity and precision was required. A spectrophotometric micromethod is described, based o n the w-ell-known reaction of hexavalent chromiuni with s-diphenylcarbazide, which results i n the formation of a pink color. This reaction is sensitive to 0.005 p.p.m. if the color is read in a Beckman spectrophotometer, Model DU. The standard deviation was determined for chromium concentrations ranging from 0.02 to 0.32 p.p.m. and found to be 0.005 p.p.m., when 10-1111. blood samples were used.
A
LTIIOUGH a great, numbtxr of toxic maiiifestntions of chromium have heen described in t81ieliterat,ure (1, 7 , 20, 39), very litt,le is known about the nirtabolism of chromium in the human body. The authors became interested in chromiumnietabolisin because of the high incidenw of cancer anioiig chromate and chromite workers (3-6, 18, 20, 21. Z+,2;). -1s the production of chromium iii the United Statw hns incrcvtsed t,rcmendousl>during the past tn-o decades, thv study of certain biochemical aspect.s of chroniium cancer beconies n i u r c ~arid niore iniport,ant,. In the course of the investigations it 1)ecanit~necessary t’ohave a method for the quantitative determination of very small aniounts of chroniium in blood. The method must be sensitive and it must have a high degree of reprodueihilit~-for such low blood chroniiuni concentrations as 0.1 p.p.ni. mid Iew. I t must also permit the determination of chroniium prewnt as soluble chromium salts or as organic chroniiuni componnds, as well as chromium in chromite which might be present in the blood of chromite workers in the form of very fine particles (cf. 2 3 ) . Among the various possible methoda (including spectrophotometric, spectrographic, polarographic, and radioisot’ope proce-
dures) a spectrophotometric method was first investigated, n-hich is based on the format,ion of an int,ensely colored compound when s-diphenylearbaxide is added to a dichromate solution. [The nature of this colored substance is still unknown. Therefore, t,he authors prefer to call it a compound rather than a complex, following a suggestion made by Feigl (Is)]. Many ot,her color reactions have been suggested (cf. I), but none of them is so sensitive and so specific as the diphenylcarbazide reaction. The principle of this method has been known since the beginning of this century (9,SO), and a great number of publicat,ions have dealt with its application to the detection and niicrodeterminat,ion of chromium in minerals, steels, soils, and water (29, s3, 34). Relat,ively lit,t,lework has been done on biological material and only a fen- methods have dealt, with the colorimetric niicrodetermination of chromium in blood ( 1 , 7 , 28, 38). Akatsuka and Fairhall ( 1 ) stat(, that t,heir method permits only a rough estiniat,ion of t h e chloiiiium content of blood and tissues. Brard ( 7 ) uses large blood samples in his determination. Mitchell and Gray’s niethod (28) cannot be used if the sample contains less than 3 micrograms of chromium. rill three methods are r:rt,her tedious and not easily adaptable to the analysis of a large number of blood samples containing minute amounts of chroniium. Urone and Andere’ method (38)is by far superior to both Braid’s and Mitchell and Gray’s methods. Vsing this procedure, sat,&factory results were obtained when blood samples cont’aining chromium in concentrations great,ert,han 0.1 p,p.ni. were analyzed. However, insufficient precision x-as obtained for the wthors’ purposes a t chromium concent,rations below 0.1 p.p,m. In view of the fact that the chromium concentrations of most samples t o be analyzed in this laboratory ranged from about 0.01 to 0.1 p.p.m. it appeared desirable to develop an alternative procedure which gives an increased precision and can therefore be used advantageously for these extremely lo\\- chromium concentrations. [Determinations carried out in this laborat,ory and in the 1aborat.ories of the Division of Industrial Hygiene, Ohio Department of Health ($6), shon- that blood of chromiuni workers contains about 0.01 to 0.16 p.p.m. of chromium, whereas normal human blood usually contains less than 0.01 p.p.ni. Somewhat similar results are re-
