Ultracentrifugal Method for the Determination of Serum Lipoproteins LUlGl DEL GATTO, F.
T. LINDGREN,
and A. V. NICHOLS
Donner laboratory o f Medical Physics and Biophysics, Department o f Physics, University o f California, Berkeley, Calif.
b A method was developed for ultracentrifugal analysis of all classes of serum lipoproteins, which requires a minimum of time, work, and materials. It utilizes the principle that lipoproteins experience flotation in a medium the density of which is greater than that of their own hydrated density. In this procedure the isolation and analysis of lipoproteins are made in a sodium bromide medium of density 1.20 grams per ml.
T
importance of serum lipoproteins in relation to lipide transport and metabolism has repeatedly been emphasized. Thus far the ultracentrifuge has provided one of the most useful techniques for studying the serum lipoproteins. Gofman, Lindgren, and Elliott (3) were first to apply the basic principle of ultracentrifugal flotation to the study of lipoproteins. Several improvements and modifications of this ultracentrifugal methodology have been made (2, 6, 7). Many laboratories have adopted the procedures described by deLalla and Gofman ( 2 ) for complete serum lipoprotein analysis. However, none has allowed quantitative study of the complete human lipoprotein system in one procedure requiring only 2 ml. of serum. HE
MATERIALS AND EXPERIMENTAL
Two milliliters of serum, 0.94 ml. of sodium chloride solution of density, 20 "
1.0060 grams per ml. (which corresponds to the amount of this salt solution present in 1 ml. of serum), and 3 ml. of sodium bromide solution ( p _ - 1.3895 rt 0.0005 grams per ml.) p~
Too
are pipetted into a 6-ml. Lusteroid preparative tube, and, after capping, thoroughly mixed. The background solution density of this mixture (exclusive of serum proteins, lipoproteins, and other macromolecular constituents) 20
is p~ = 1.2052 =!= 0.0005 grams per ml. After ultracentrifugation (Spinco Model L ultracentrifuge) a t 40,000 r.p.m. for 24 hours a t a temperature between 18" and 20" C., all the knon-n classes of serum lipoproteins are concentrated in a 1-ml. volume a t the top
of the preparative tube. Analyses of the second milliliter reveal the presence of neither lipoproteins nor serum proteins. The final density of this lipo20 O
protein fraction is p~ = 1.1977 rt 4 0,0005 grams per ml. (representing the average value and standard deviation of six determinations on one serum sample). The corresponding background sodium chloride and sodium bromide salt mixture consisting of 2.82 20 "
ml. of sodium chloride solution (m= 4 1.0060 grams per ml.) and 3 ml. of 20 O
sodium bromide solution (p: 4 = 1.3895 grams per ml.), centrifuged under the same conditions, yielded a 20 O
top-milliliter final density of p4O ' 1.1965 rt 0.0003 grams per ml. (six determinations). This drop in density from 1.2052 grams per ml. to approximately 1.1965 grams per ml. reflects the approach toward sedimentation equilibrium of the salts in the preparative tube that occurs during the centrifugal conditions used for this lipoprotein isolation. The concentrations and flotation rates of the various lipoproteins present in the total lipoprotein fraction are determined in a Spinco Model E analvtical ultracentrifuge at - operating . 26" + 0.5" C. If desired, the ultracentrifugal determination of total serum protein (albumin, globulin, etc.) may be carried out simultaneously. For this analysis the 5-ml. subnatant fluid in the preparative ultracentrifuge tube (after thorough remixing) is diluted with distilled water in the ratio of 1 part of subnatant fluid to 3 parts of water. The final density of this mixture is 1.0613 + 0.0045 grams per ml. (six determinations), Figure 1 shows the schlieren patterns of a typical lipoprotein and protein analytical run.
-
Several different ultracentrifugal rotor speeds have been tested (52,640, 50,740, 47,760, 44,770, 42,040, 39,460, and 37,020 r.p.m.). Successive pictures have been taken a t appropriate time intervals to allow complete quantitative study of all the major serum lipoprotein classes, which under these conditions have flotation rates of from 0 to 6 and 16 t o 802 Svedberg units (9). The recommended ultracentrifugal rotor speed of 42,040 r.p.m. has a centrifugal
field of 133,63i X g (at a radius of 6.50 cm.). For this speed, appropriate timcs for obtaining schlieren patterns ar(1: 120 seconds after starting (during acct.1eration at approximately 22,000 r.p.ni.) ; 242 seconds after starting (0 time up to speed) and a t 2, 12, 14, 44, 64; and (or) 76 minutes after achieving full speeds. Uniform acceleration is achieved by maintaining a constant current (13 to 15 amperes) with Variac manipulution such that full speed (42,040 r.p.m.) is reached in 242 seconds. The proper amperage value necessary will depend upon the characteristics of each individual drive mechanism. In order to measure accurately lipoprotein concentrations in this procedure, it is necessary to use a double-sectored analytical cell (I), into one sector of which is introduced a base line solution equivalent to the background solution 20" ( p = ~1.1965 grams per ml.). This is necessary because during the analytic run there occur appreciable changes in concentration gradient of the sodium bromide and sodium chloride salts, as well as rotor speed changes which would frequently prevent accurate placement of the baseline. The analytical ultracentrifuge film, therefore, is a simultaneous and superimposed plot of the refractive index gradient of thc sodium bromide base line solution and the lipoprotein solution as a function of the distance from the center of rotation. Analysis of the film is made using an enlarged (5X) tracing of each frame. Measurement of any given area under the base line is ordinarily obtained b! planimetry. The area obtained by planimetry is converted to milligrams per 100 ml. by using the relationship derived by Pickels ( 8 ) . Because of the sector shape of the analytical cell and the varying centrifugal field, a radial correction is applied. This radial correction may be used in lipoprotein analysis, as it has been shown to hold for multicomponent systems (10)*
In the film analysis the total serumlipoprotein spectrum has been subdivided into five flotation-rate classes: S/ci 20) (185-802), (61-185), (44-61), (1&44), and (0-6). The subdivision of the four faster-migrating lipoprotein classes has been made to correspond as VOL. 31, NO. 8, AUGUST 1959
1397
closely as possible to the Gofman lowdensity lipoprotein classification (see Table I). This correspondence was determined by analysis of nine sera by both procedures in which the ultracentrifugal background salt solution 26" density, '40, for the sodium chloride low-density lipoprotein run was 1.063 grams per ml. and for the sodium bromide total lipoprotein run was 1.198 grams per ml. The flotation rates in the sodium bromide run were standardized to a medium of density, p E 1.200 grams per ml. The error in 49 ' the determination of Sfo.zol rates as
Table I. Slo.as, and Sl,~.2o)Relationship S, in NaCl S, in NaBr Medium Medium 26"
1.063 0.0005 G./M1.) p~
*
26" 1.2000 4 0.0005 G./Ml.)
( p a
Experimental Paints 135 84.0 81.0 53.2 46.9 43.3 42.0 15.0 14.3 6.7 6.7 5.9 5.9 5.8 5.7 5.6 5.6 5.6 5.6 5.4 5.4 4.8 4.8 4.1
'
1.1
0.9
244 154 149 105 106 99.5 105 48.5 47.5 30.4 30.5 28.3 28.9 27.4 28.3 27.3 28.1 27.3 26.9 27.4 26.6 26.9 26.6 25.2 15.8 16.5
Estimated Points 44.4 61.3 ""
12.0 20.0 nm
.
.
~~
si,,.,,
%
Figure 1.
Patterns obtained in a typical lipoprotein and protein run
a result of the variation in density, temperature, and location of the bottom of the cell on the film is approximately 2% for all classes. However, it can increase considerably if the standard conditions are not carefully maintained. If there is any significant dependence of flotation rate on concentration ( S f us. C), the concentration of any particular class of lipoprotein requires correction because of the JohnstonOgston effect (6). This effect results in the apparent decrease in concentration of the faster molecular class bccause of a negative concentration gradient of the slower component in the boundary region of the faster component. At the same time, there is a corresponding apparent increase in concentration of the slower molecular class. Johnston and Ogston have treated this phenomenon theoretically for two components. Such apparent changes in concentration are a function of the concentration and flotation rate of each component. If the Johnston-Ogston correction is to he applicable to a spectrum of components (as in dealing with a lipoprotein spectrum), certain simplifications are necessary. The lipoprotein spectrum is arbitrarily divided into the five component classes previously described. The effective flotation rate of each class is calculated by considering the d i s tance traveled by the class to the midarea position of the schlieren pattern. Thus, the mid-area position of each of these classes is designated as the effective flotation rate of that class. (The
s;,,.,,,
2426 2485 231 101 1875 50.0 1651 26.5 565 1.80 1798 226 1380 52.4 1252 27.0 356 2.05 107 1733 1662 226 1193 49.5 1044 27.4 330 1.85 119 1609 943 224 708 55.6 629 27.5 161 2.25 113 911 781 210 584 53.0 517 27.6 133 123 768 1.so 267 27.4 117 455 70.6 2.20 505 205 319 52.5 266 28.5 61.6 2.25 475 210 315 52.8 117 415 2.25 435 211 314 52.8 261 28.0 113 397 61.6 114 217 2.08 Av. 1136 218 1089 836 52.5 736 27.6 All concentrations calculated &ssume a specific refractive increment of 0.00145 g./lOO ml. for all lipoprotein olssses in NaBr medium of density 1.20 g./ml. [Estimated from
Hanig's data. ( 4 ) ] . rate a t mid-area. pos S ~ C , . ~Flotation ~).
1398
ANALYTICAL CHEMISTRY
accuracy of the measurement of the mid-area position of the schlieren pattern is better than that of the second moment, although theoretically the latter is a better dcfined property of the system than the former.) Equation l expresses this mid-area flotation rate for a particular lipoprotein class as a function of the concentration.
This equation also assumes that., for the same concentrations of any lipoprotein class, the slowing effects on a particular lipoprotein class are equivalent. Thus, for any class under consideration, this equation becomes a straight line where If is the slope. The data for the S f us. C plot for the different classes of lipoproteins are given in Table 11. The values of K for the five major lipoprotein classes are as follows:
The K values are all positive, except ~ lipoprotein for the S I O . (185802) class. This unexpected finding is being studied on lipoprotein fractions of the class Sla.wl (185-802), properly isolated. It is of interest in this study to find such small K values for all lipoprotein classes. These small K values lead to minimal Johnston-Osston effects on the observed lipoprotein spectrum, and on this basis the Johnston-Ogston correction is omitted in this procedure. Because the K value of each lipoprotein class has been obtained in the presence of other lipoprotein classes, the actual K values for isolated lipoprotein classes may be different from those given here. However, these values correspond more closely to the actual experimental conditions 01' the procedure. DISCUSS1ION AND CONCLUSIONS
This methiid provides an ultracentrifm a i n r c71.0. nf ugal analysiLa nf all ..._ serum lipoproteins isolated from 2 ml. of serum.
_.-..
Jv.
_- -.
The ultracentrifugal procedure used by Lewis and coworkers (6) employing potassium bromide-sodium chloride (1.21 grams per ml.) and which involves ultracentrifuging a t 30,000 r.p.m. for 13 t o 17 hours a t temperatures between 16” and 18’ C. does not quantitatively recover all the serum lipoproteins. The lower viscosity of potassium bromidesodium chloride solution than that of sodium bromide solution at a density of 1.20 grams per ml. does not sufficiently compensate for the shortness of their preparative ultracentrifugal run, ahich also occurs at a lovc-er centrifugal field. Because of viscosity differences, lipoproteins in potassium bromidesodium chloride (1.21 grams per ml.) float a t a rate about 2OyO faster than in sodium bromide (1.20 grams per ml.). Considering this factor in potassium bromidesodium chloride (1.21 grams per ml.), it is somewhat more difficult to study lipoproteins with flotation rates in the neighborhood of S , ( I 20) 802 (in 1.20 grams per ml. of sodium bromide) in the analytical ultracentrifuge without using an accurately determined acceleration picture taken considerably below 52,640 r.p.m. In choice of salts for increasing solution densities, it is worthy of note that sodium bromide has a much higher solubility than potassium bromide.
Saturated aqueous sodium bromide a t 20” C. has a density of 1.543 grams per ml., whereas saturated potassium bromide has a density of only 1.371 grams per ml. Compared to the deuterium oxide-sodium nitrate system of deLalla ( 2 ) sodium bromide represents a simplification, as well as a further advantage of not interfering with a Kjeldahl nitrogen analysis, which may be desired. The analytical ultracentrifuge field a t 42,040 r.p.m. represents a reduction of 3GY0 in centrifugal force compared with 52,640 r.p.m., the speed a t 1%-hich lipoprotein runs have been customarily made. This change will certainly reduce analytical-cell breakage. Because flotation rates remain nearly constant within the concentration ranges normally encountered, the analysis is simplified because the Johnston-Ogston correction becomes negligible and therefore unnecessary. Moreover, it is possible to follow the previously established flotation classification of the lowdensity serum lipoproteins ( 2 ) . In this procedure the S,(l.BO)0-8 class lipoproteins correspond to the total of the two principal high-density lipoproteins (HDL2 and HDL-3). This method provides a simplified procedure for complete ultracentrifugal analysis in essentially a one-salt system (sodium bromide) and uses only 2 ml. of serum.
ACKNOWLEDGMENT
The authors express appreciation to John W. Gofman for his many helpful criticisms and suggestions given during the course of this investigation. LITERATURE CITED
(1) Beams, J. W.,Dixon, H. hI., 111, R e v . Sci. I n s t r . 24, 228 (1953). (2) delalla, O., Gofman, J. I\’., in “Methods of Biochemical Analysis,” Vol. I (D. Glick, ed.), Interscience, Ti-ew York. - , 1954. ~ ~ (3) Gofman, J. W.,Lindgren, F. T., Elliott, H. A . , J . Bid. Chem. 179, 973 (1949): (4) Hanig, M., Shainoff, T. R., Ibid., 219, 479 (1956). (5) Johnston, J. P., Ogston, A. G., Trans. Faraday Soc. 42, 789 (1946). (6) Lewis, L. A., Green, A. A., Page, J. H., Am. J . Phys. 171,391 (1952). ( 7 ) Lindgren, F. T., Elliott, H. A , , Gofman, J. W., J . Phys. & Colloid Chem. 5 5 , 80 (1951). (8) Pickels, E. G., Methods of M e d . Research 5, 107 (1952). (9) Svedberg, T., Pedersen, K. O., “The Ultracentrifuge,” Oxford Univ. Press, Oxford, 1940. (10) Trautman, R., Schumaker, V., J . Chem. Phys. 22, 551 (1954). RECEIVED for review September 29, 1958. Accepted March 26, 1959. Luigi Del Gatto is a fellow of the Jane Coffin Childs Memorial Fund for Medical Research. Investigation aided by a grant from The Jane Coffin Childs Memorial Fund for Medical Research.
Determination of Titanium and Iron in Titaniferous Materials by Cerate Titrimetry J. 0.PAGE and A. 8. GAINER’ Agriculfural and Mechanical College of Texas, College Station, Tex.
b As all commercial titanium ores contain iron, it is frequently desirable to determine these two metals in a single titration. A simple, rapid, and accurate volumetric method has been developed for the simultaneous determination of titanium and iron coexisting in solution. Seven titaniferous ore samples were fused with potassium bisulfate and the melt was extracted with dilute sulfuric acid. The solutions containing titanium(1V) and iron(ll1) ions were reduced with aluminum foil or liquid zinc amalgam in a simple apparatus which provided for flushing of the solution with carbon dioxide to remove air, and quick, easy removal of the zinc amalgam before titration. After reduction, a quantitative determination of titanium and iron in the solution was made with cerium(lV) sulfate as oxidant, and with methylene
blue and N-phenylanthranilic acid as the internal indicators for the titanium and iron end points, respectively. Separate determinations were made for chromium and vanadium. Results compare favorably with those obtained on the same ore using two modified methods.
T
for a simple, rapid, and accurate method for the simultaneous determination of titanium and iron in titaniferous ores prompted the development of a method which answered these specifications. Before 1949, all methods for the determination of the titanium and iron content of titanium ores required separation of the iron from the titanium. I n 1949, Shippy (3) presented a procedure for the simultaneous determination of these metals HE NEED
coexisting in solution, using potassium permanganate as oxidant with methylene blue as internal indicator for the titanium end point and 1,lO-phenanthroline ferrous complex as the internal indicator for the iron end point. Reduction was accomplished by use of the Jones reductor. Certain volumetric methods were evaluated with the objective of replacing potassium permanganate by cerium(IV) sulfate, the Jones reductor by aluminum or liquid zinc amalgam, and 1,IO-phenanthroline ferrous complex by AT-phenylanthranilic acid. Reduction of titanium(1V) and iron(II1) by aluminum foil or liquid zinc amalgam is recommended by ease of handling and 1 Present address, University of Oklahoma, Norman, Okla.
VOL. 31, NO. 8, AUGUST 1959
1399
