Tolerance to Opioid Narcotics, II. Cellular Tolerance to Levorphanol in Mouse Brain* Judith A

Prnocedings of the National Academy of Scictnces Vol. 66, No. 3, pp. 944-951, July- 1970

Tolerance to Opioid Narcotics, II. Cellular Tolerance to Levorphanol in Mouse Brain* Judith A. Richtert and Avram Goldstein

D)EPARI'MENTr OF PHARMACOLO)GY, STANFORi) UNIVERSITY SCHOOL OF MEDICINE, STANFOIRI), CALIFORNIA Commnunicated by Seymour S. Ktly, April 20, 1970 Abstract. M\lice were made tolerant to a large dose of levorphanol, a congener of morphine. Then 3H-levorphanol was given. The concentration of free, unchanged levorphanol in the brain water (ultrafiltrate) was found to be much higher than required to produce pharmacologic effects in nontolerant animals. The result indicates that tolerance arises from a diminished sensitivity to the drug at cellular or subcellular sites of drug action in the brain.

The first paper of this series' described the development and loss of tolerance to levorphanol in mice, using the "running fit" as a measure of the action of the opioid narcotics. The tolerance under investigation was simply a reduction in pharmacologic effect with repeated dosage of the drug. Obviously, two very different mechanisms may be involved. Metabolic tolerance would entail a reduction in the blood and brain levels at a given dose, through increased metabo- lism or excretion of the drug.2 Brain tolerance, on the other hand, would entail a diminished sensitivity of brain to an otherwise sufficient drug concentration. It has been shown by others that although the brain concentration of an opiate is sometimes lower ill tolerant animals than in controls, the reduction is too small to account for the tolerance.3-6 What has never been demonstrated, however, is that the opiate in brain of tolerant animals is really free, thermodynamically active, unchanged drug, capable of interacting at receptor sites; and that it is not inactivated by binding to some other molecule, as suggested by the work of Kornetsky and Cochin7 and Ungar and Cohen.8 We show here that in the brain water of mice made tolerant to levorphanol there is a high enough concentration of unchanged, free, ultrafiltrable drug to produce intense pharmacologic effects in nontolerant mice. We conclude that brain tolerance to opioid narcotics repre- sents a decreased sensitivity to drug at the cellular or subcellular level. Methods. Tolerance: Swiss-Webster male mice were used. The methods for recording running activity automatically in photocell cages, for measuring analgesia on the hot plate, and all other procedures except as noted have been described.' After an initial dose of 20 mg/kg levorphanol intraperitoneally and recording of analgesia 30 min later and running activity for the next 8 hr, the animals were returned to ordinary cages with sawdust bedding. There they were given 5 more injections of the same dose at 8-hr intervals, then between 8 and 29 injections of 50 mg/kg at the same intervals until the running activity had fallen below 25% of the initial value and analgesia had disappeared (reaction within 30 see on the hot plate). An injection of 50 mg/kg of 3H- levorphanol was then given. The mice were tested for analgesia again just before and 944 Downloaded by guest on September 30, 2021 VOL. 66, 1970 MEDICAL SCIENCES: RICHTER AND GOLDSTEIN 945

30 mill after this injection, then immediately (lecapitated. Control mice were treated i(leiltically except that they were given sodium tartrate in doses equivalent to the tartrate received as levorphanol tartrate by experimental animals; their final injection was 10 mg/kg of 1H-levorphanol. In some experiments untreated mice were used as controls. Tissue work-up: For trichloroacetic acid extraction fresh brains were homogenized in a Dounce homogenizer in 10 vol of cold 5% trichloroacetic acid. After centrifugation at 8000 X g for 10 min in the cold a sample of the supernatant solution was taken for determination of radioactivity. Blood levels of 3H-levorphanol were determined on trichloroacetic acid supernatants of whole blood obtained from mice heparinized (100 units i.p.) 20 min before injection of levorphanol. For ultrafiltration studies the following procedure was performed at 50C. A number of fresh brains (usually 6-8) were pooled and homogenized without addition of any fluid in a Potter-Elvehjem homogenizer with a ground glass pestle driven at 800 rpm. At each stroke the pestle was withdrawn and the tissue was scraped from the sides of the homogenizer and pestle and returned to the bottom of the homogenizer. After 5-10 strokes a paste was obtained in which no tissue fragments were visible. This material was transferred to a 2-ml Spinco tube and centrifuged overnight at 105,000 X g. The supernatant solution (about 0.3 ml from six brains) was placed in the dependent portion of a segment of dry dialysis tubing hung down in the shape of a U into a centrifuge tube. The open ends of the tubing were secured over the rim of the centrifuge tube, which was covered tightly with Parafilm. Centrifugation was conducted at 50C for about 2 hr at 8000 X g, to obtain a small amount of ultrafiltrate (never more than 25% of the total fluid volume). Because of the small volumes used it was essential for the dialysis tubing to be absolutely dry. Supernatant solutions were also prepared from dilute brain homogenates. In these cases 8-10 brains were pooled in an equal volume (or in 4 vol) of 0.005 M Tris hydrochlo- ride buffer at pH 7.0 and homogenized with a ground glass homogenizer with 25 strokes at 800 rpm. The homogenate was treated in a Mullard 9-kc sonic oscillator for three 30-sec bursts. Centrifugation and preparation of the ultrafiltrate was as described, except that 1.5 hr at 1000 X g was sufficient for expressing a small amount of ultrafiltrate. Determination of blood volume in mouse brain: Three heparinized mice were injected with 10 mg Evans Blue in 0.1 ml of water in the tail vein and decapitated 5 min later. Whole blood and a brain homogenate were extracted with 70% ethanol, the extracts were evaporated to dryness, and the residues were redissolved in water. From the optical densities at 610 nm the dye concentrations were obtained, whence the amount of blood in the brain was readily determined. Results were corrected for losses of 10-20% observed when known amounts of dye were added to blood or brain. Normal blood and brain were treated identically, to provide appropriate blanks. Ethylene dichloride extraction and silica gel chromatography: For these experiments larger amounts of ultrafiltrates were obtained. To each milliliter of ultrafiltrate were added 0.5 ml of saturated Na2CO3 and 3 ml of ethylene dichloride. After a period of mixing on a Vortex mixer for 2 min, the layers were allowed to separate and the water layer was removed. The ethylene dichloride was evaporated to dryness in a test tube, resuspended in 50 Il of distilled water, and chromatographed on silica gel (Eastman Chromatogram Sheet 6061, without fluorescent indicator) by the ascending method with methanol: chloroform: ammonia (67:51:2). Levorphanol and norlevorphanol, included as reference standards, were revealed as blue spots by spraying with Kiefer's reagent (10 ml of 1% potassium ferricyanide and 2-3 drops of 1 N ferric chloride). The sample track (not sprayed) was cut into 1 X 3-cm strips for counting in 10 ml of scintillation mixture. Efficiency of 3H counting in this system was 14%, as determined by applying known amounts of 3H-levorphanol to the gel. Elution of ultrafiltrates from Biogel: Ultrafiltrates were layered on a 0.8 X 12-cm column of Biogel P-2, which had been equilibrated with 0.01 M Tris buffer, pH 7.0. The exclusion size of this column is stated to be approximately 1600 daltons. The void Downloaded by guest on September 30, 2021 946 MEDICAL SCIENCES: RICHTER AND GOLDSTEIN PROC. N. A. S.

volume, determined with blue dextran, was 2.5 ml. Elution was performed with the same buffer at about 0.4 ml/min. Fractions of 1 ml were collected. Other methods: Brain water was found to be 77% of the wet weight by drying to constant weight. Protein was determined by the method of Lowry, Rosebrough, Farr, and Randall.9 Radioactivity was determined by liquid scintillation counting using a xylene-dioxane mixture10 and hydroxide of hyamine (Packard) or ethanol, as appropriate, for solubilization. All samples were counted to a standard error of 3% or less (1000 counts). Efficiencies were determined by addition of 3H20 as internal standard, in a volume negligible relative to Water already in the system. Drugs: Levorphanol tartrate and 3H-levorphanol tartrate (hydroxy-N-methyl- morphinan-6,7(8)-3H,2-3; 5.0 /Ci/mg) were generously donated by Hoffmann-LaRochc, Inc. Stock solutions of levorphanol tartrate and sodium tartrate were prepared in distilled water and diluted With 0.9% NaCl. All concentrations and doses of drugs refer to the free base. Results. Table 1 shows that in nontolerant mice given 10 mg/kg of levor-

TABLE 1. Levorphanol in trichloroacetic acid extracts of brains from tolerant and control mice. Control Tolerant No. mice 13 10 Final running activity (% initial) * 13 No. mice displaying analgesia/total mice 13/13 0/10 Levorphanol concentration (1ag/g brain) 0.51 (0.36-0.69) 2.05 (1.10-2.70) Control mice injected every 8 hr with sodium tartrate, then given 10 mg/kg 3H-levorphanol and decapitated 30 min later. Tolerant mice injected (see Methods) on an 8-hr schedule with 20 mg/kg then 50 mg/kg levorphanol, then given 50 mg/kg 3H-levorphanol and decapitated 30 min later. Ranges are given in parentheses. * Not measured. Vigorous running at 100% of initial rate observed in independent experiments. phanol, a total concentration of 0.51 Mug/g, in brain was established, and this was more than sufficient to produce vigorous running activity and analgesia in all the mice tested. In the tolerant mice given 50 mg/kg, the total brain concentration was 2.05 lug/g, yet very little running and no analgesia resulted. The tolerance, therefore, cannot be attributed to reduction in the drug concentration in brain. Nor, since the levorphanol measured here was trichloroacetic acid- soluble, could the drug be covalently bound to any macromolecule in brain to an extent sufficient to account for the tolerance. Table 2 shows the results of experiments in which levorphanol was measured in brain homogenates, in supernatant solutions after 105,000 X g centrifugation, and in ultrafiltrates. In all cases the control mice ran vigorously and displayed analgesia, whereas the tolerant mice ran little and displayed no analgesia. The drug concentrations in the homogenates from tolerant mice were 3-5 times higher than the controls, as would be expected from the higher dose. Between half and two thirds of the drug was bound to subcellular particles, de- pending upon the extent of dilution in homogenization; but there were no dif- ferences between control and tolerant animals in this respect. In most experiments 90% or more of the drug that was not particle-bound was ultrafiltrable,1I so that the levorphanol concentrations (i.e., the free levorphanol in brain water) Downloaded by guest on September 30, 2021 VOL. 66, 1970 MEDICAL SCIENCES: RICHTER AND GOLDSTEIN 947

TABLE 2. Levorphanol in brain homogenates and ultrafiltrates from tolerant and control mice. Tissue Volume:Buffer Volumie No Buffer Added , 11-1:4-_ Control* Tolerantt Controlt Tolerantt Controlt Tolerantt No. experiments 4 1 2 1 1 1 No. mice per experiment 6 7 10,8 8 3 3 Final running activity (% initial) § 15 § 10 § 10 No. mice displaying analgesia 22/24 0/7 10/10, 8/8 0/8 3/3 0/3 Levorphanol in homogenate (,sg/ml) 0.83 (0.75-0.93) 4.04 1.28,1.17 3.15 1.12 4.13 Levorphanol in supernatant (g/ml) 0.26 (0.24-0.28) 1.36 d .62, 0. '56 1.50 0.53 2.12 Percentage not bound to particles 32 (26-35) 34 48, 48 48 47 51 Levorphanol in ultrafiltrate (;&g/ml) 0.24 (0.22-0.24) 1.25 0.44, 0.44 1.41 0.47 2.06 Percentage ultrafiltrable 92 (86-101) 92 72, 79 94 93 90 Ultrafiltrate volume (% original in dialysis bag) 21 (15-25) 18 9, 5 6 7 7 Ultrafiltrate protein (% protein in dialysis bag) 3.5 (0.7-6.5) 0.3 1.4, 0.9 2.2 1.8 1.6 Control and tolerant mice as in Table 1. Homogenates, supernatant solutions, and ultrafiltrates were prepared as described in Method8. When more than one experiment was done, ranges are given in parentheses. Concentrations are given per ml of brain water, determined to be 77% of the brain wet weight. * Tartrate injected as described under Methods. Received 10 mg/kg 3H-levorphanol. t Tolerant to 50 mg/kg levorphanol. Received 50 mg/kg 3H-levorphanol. Untreated controls. Received 10 mg/kg 3H-levorphanol. § Not measured. Vigorous running at 100% of initial rate observed in independent experiments. were 3-5 times higher in tolerant than in control brains. The low protein con- tent in the ultrafiltrates indicates that the dialysis bags (which could not be tested because they were used dry) remained intact. These results show that tolerance cannot be attributed to binding, reversible or irreversible, to any constituent large enough to be retained by the dialysis bag, about 6,000-12,000 daltons. 12,13 We had to consider the possibility that a major part of the levorphanol in brain was present in blood trapped there at the time of decapitation, for then the preceding data would refer principally to drug in blood rather than in brain water. By the dye technique described under Methods, the amount of blood in mouse brain was found to be 16-20 1d. The mean blood concentration of levorphanol was 1.88 pg/ml in control mice after 10 mg/kg and 7.52 ,g/ml in tolerant mice after 50 mg/kg. Thus, only 10-16% of the levorphanol radioactivity in brain was in blood; 84-90% was in brain tissue."4 Table 3 shows that when 3H-levorphanol was added to ultrafiltrates from control or tolerant mice in vitro, virtually all of it was extracted into ethylene dichloride. However, after administration of 3H-levorphanol in vivo, larger proportions remained in the water phase, especially in ultrafiltrates from tolerant Downloaded by guest on September 30, 2021 948 MEDICAL SCIENCES: RICHTER AND GOLDSTEIN PROC. N. A. S. TABLE 3. Extraction of ultrafiltrates with ethylene dichloride. Ultrafiltrate Levorp)haniol ill volume ultrafiltrate Distribution of Radioactivity (%) (ml) (DPM X 10-3) Water EDC Recovery Levorphanol added to ultrafiltrate in vitro Control* 0.15 157 2 105 107 Tolerant* 0.15 124 2 114 116 Controlt 0.50 377 2 99 101 Controlt 1.00 755 2 105 107 Levorphanol given in vivo, 50 mg/kg Control* 0.15 3.85 13 94 107 Tolerant* 0.15 1.81 47 56 10t3 Tolerant* 0.20 1.82 62 -$ t Controlt 1.00 10.0 8 86 94 Tolerantt 1.00 3.51 27 66 93 Tolerantt 1.50 6.88 19 -t -t A sample of ultrafiltrate was removed for radioactivity determination, then the volume indicated was alkalinized and extracted as described under Methods. Radioactivity was determined in the water phase and in the ethylene dichloride phase. Controls were untreated mice; tolerant mice were tolerant to 50 mg/kg. * Ultrafiltrate of homogenate with no added buffer. t Ultrafiltrate of homogenate with 4 vol added buffer. * Not determined. mice. A polar metabolite is evidently present, and occurs in increased amounts in brains of tolerant animals. Since it has been shown that only 10-16% of the levorphanol radioactivity in brain could be in blood, and since much higher fractions of the ultrafiltrate levorphanol in tolerant mice were present as the metabolite, it seems unlikely that the metabolite could be wholly contained in blood. Yet the most probable candidate, levorphanol glucuronide, would not be expected to pass from blood into brain, and is not known to be formed within brain tissue. In the experiments summarized in Table 3, and also in those to be described in Table 4 and Figure 1, the control mice were given the same large dose of drug

TABLE 4. Silica gel chromatography of ethylene dichloride extracts of ultrafiltrates. Total Total apparent recovery levorphanol At At from in Levorphanol Norlevorphanol ultra- ultrafiltrate Position* Positiont filtrate (Ag/ml) (.g/ml) MtO (jg/n11) (% (S 3H-Levorphanol added to tolerant ultrafiltrate in vitro 75.6 54.5 72 3.5 5 94 Control in vivo 6.09 2.74 45 0.34 6 70 Tolerant invivo 2.10 0.76 36 0.05 2 46 Tolerant invivo 2.77 1.08 39 0.09 3 47 Homogenates were prepared with 4 vol of buffer. Ultrafiltrates were extracted with ethylene dichloride, and the extracts were chromatographed on silica gel plates, as described under Methods. Controls were untreated mice; tolerant mice were tolerant to 50 mg/kg levorphanol. All mice were given 50 mg/kg 3H-levorphanol 30 min before decapitation, except for a group of tolerant mice used for in vitro addition of 3H-levorphanol to ultrafiltrate. * Rf = 0.68; strip included Rf 0.40-0.75. t Rf = 0.27; strip included Rf 0.19-0.40. Downloaded by guest on September 30, 2021 VOL. 66, 1970 MEDICAL SCIENCES: RICHTER AND GOLDSTEIN 949

A. :N VITRO B. CONTROL C. TOLERANT

03- t z

002~~~~~~~~~~~

II ~~~~~~~~~~50'~100 it O2>152 0.I zI

C ~~~~~~~~~~~~0 0 10 ~~20 10 20 I0 20 ELUTION VOLUME (ml) FIG. 1.-Characterization of levorphanol radioactivity from ultrafiltrates by means of Biogel P-2. (A) Levorphanol tartrate (0.2 ml of a 1 mg/ml solution in 0.01 M Tris buffer, pH 7.0) was placed on the column and eluted with the same buffer at room temperature. Eluate fractions were read at 279 nm (broken curve L). In a separate experiment 3H-levorphanol was added to ultrafiltrate prepared from tolerant mouse brain without added buffer and applied to the column in the same way (solid curve L). Recovery of radioactivity was 97%. Finally, ferredoxin (0.2 ml of a 2 mg/ml solution, mol wt 6012) was applied in the same way; fractions were read at 285 nm (curve F). (B) Ultrafiltrate was prepared without added buffer from control mice decapitated 30 min after injection of 50 mg/kg 3H-levorphanol. The ultrafiltrate (0.17 ml con- taining 2.69 gg/ml levorphanol) was applied and eluted as above. Recovery of radioactivity was 95%. (C) As in (B), but mice were tolerant to 50 mg/kg. 0.14 ml of ultrafiltrate contained 1.13 pg/ml. Recovery was 105%.

(50 mg/kg) as the tolerant animals, in order to obtain sufficient radioactivity in the various fractions. Under such conditions of equal dosage the brain levels of levorphanol are consistently lower in tolerant than in nontolerant mice. This is a reflection of metabolic tolerance-the more rapid conjugation and excretion of the drug.' The amounts of levorphanol in brains of control mice given 50 mg/kg of levorphanol are at least five times higher than required to produce running activity and analgesia. The ethylene dichloride phases were dried, taken up in water, and chromatographed on silica gel. The results are shown in Table 4. When 3H-levorphanol was added in vitro to ultrafiltrate from tolerant mice, most of the radioactivity behaved as authentic levorphanol. The over-all recovery through the whole procedure was 94%. In experiments in vivo with both control and tolerant animals (but more so in the latter) the losses in the procedure were considerably greater, at least partly because the polar metabolite was not extracted by ethylene dichloride, as already shown (Table 3). If there was any formation of norlevorphanol in vivo it was very small in extent, as demonstrated by M\Iilthers15 in the rat. With respect to the question of brain tolerance, the main finding is that at least 0.76-1.08 jsg/ml of authentic levorphanol was present in the ultrafiltrates of the tolerant mice, whereas ultrafiltrate concentrations of at most 0.47 #g/ml (Table 2) were pharmacologically effective in nontolerant animals. Ultrafiltrates were analyzed on Biogel P-a, a molecular sieving column that excludes substances larger than about 1600 dalto". Figure 1A shows that Downloaded by guest on September 30, 2021 950 MEDICAL SCIENCES- RICHTER AND GOLDSTEIN PROC. N. A. S.

ferredoxin (mol wt 6012) appeared in the void volume, as expected, whereas levorphanol was eluted at 10 ml, whether it was applied in buffer or added to ultrafiltrate from tolerant mouse brain. The shape of the levorphanol peak is essentially Gaussian, indicating homogeneity. Incubation of 3H-levorphanol with ultrafiltrate from control or tolerant mouse brain for 1 hr at 370C did not change the elution pattern. Figure 1B and C show that the ultrafiltrates from animals given 50 mg/kg of 3H-levorphanol contain an additional radioactive component of larger molecular weight than levorphanol itself. The proportional contribution of this secondary component is greater in the tolerant than in the control mice. It may well be the same as the polar metabolite, presumed to be a conjugate, that was described in Table 3. Analysis of the areas under the curves of Figure 1B and C led to the following results. In the ultrafiltrate from control animals 5% of the total radioactivity was associated with the component of greater molecular weight. The concentration of apparent levorphanol in the ultrafiltrate was 2.69 ,ig/ml; thus authentic levorphanol was 9570 of 2.69, or 2.56 lug/ml. In the ultrafiltrate from tolerant animals the corresponding figure for the heavier component was 23%, and the apparent levorphanol concentration was 1.13 .g/ml; thus authentic levorphanol was 77% of 1.13, or 0.87 ,ug/ml. This concentration is at least twice that demonstrated to be sufficient for pharmacologic activity in the brain of nontolerant mice. Discussion. The main question considered in this paper is whether or not tolerance to an opioid narcotic entails a diminished sensitivity to the drug at receptor sites in the brain. The experimental method was to determine the free, unchanged concentration of levorphanol in brain water of tolerant and nontolerant mice. This free drug concentration (thermodynamically active drug) is available for interaction with drug receptors; if it is sufficiently high and yet no pharmacologic effects are produced, the tolerance must reflect a changed responsiveness at the cellular or subcellular level. The answer is about as clear as can be obtained. Sufficiently high levorphanol concentrations are indeed present in brain ultrafiltrates-at least twice (and sometimes five times or more) the concentration that suffices to produce vigorous running activity and analgesia in nontolerant mice. The radioactive drug was shown to be largely unchanged levorphanol by ethylene dichloride extraction, silica gel chromatography, and elution from Biogel P-2. When analyzed by ultrafiltration it was clearly not bound to any macromolecule retained by the dialysis membrane. A reversible interaction is nearly impossible to rule out, since all the procedures entail some dilution, which could promote dissocia- tion of a putative complex. Homogenization itself, for example, involves sub- stantial dilution as ordinarily performed. In an attempt to detect such reversible binding we carried out some experiments by homogenizing brain without any added buffer, and others with addition of various amounts of buffer. No indication of reversible binding was obtained. The binding and inactivation of the drug by an antibody, as suggested by Kornetsky and Cochin,7 is rendered unlikely by these results. A reversible interaction with a molecule small enough to be ultrafiltrable is difficult to rule out. The possibility that a peptide Downloaded by guest on September 30, 2021 VOL. 66, 1970 MEDICAL SCIENCES: RICHTER AND GOLDSTEIN 951

(or other substance formed during development of tolerance) interacts with the drug receptor and so prevents the drug action in the tolerant animal, as proposed by Ungar and Cohen,8 can neither be refuted nor supported by these experiments. An incidental finding has been the detection of a polar metabolite of levorphanol, probably a conjugate, formed in vivo but not on incubation with brain ultrafiltrate in vitro. It is unlikely, however, that the metabolite has anything to do with tolerance. Although its amount relative to levorphanol was considerably greater in tolerant than in nontolerant brains, this was because the concentration of unchanged levorphanol -was lower in the brains of the tolerant mice. In nontolerant and tolerant mice given the same dose of levorphanol, the actual amounts of the metabolite were virtually the same. We are grateful to Hoffmann-LaRoche, Inc., for generous supplies of levorphanol and 3H-levorphanol, and to Patricia Sheehan for expert assistance with some of the experiments. The investigations described here owe their inception to a discussion with Linus Pauling. *-Supported by research grant MH13963 from the National Institute of Mental Health and training grant GM322 from the National Institute of General Medical Sciences. t Present address: Department of Biochemistry, University of Cambridge, England. Reprint requests should be directed to Stanford. Material in this paper was used in Dr. Richter's dissertation, which was submitted to Stanford University in partial fulfillment of requirements for the Ph.D. degree, June 1969. 1 Goldstein, A., and P. Sheehan, J. Pharmacol. Exp. Therap., 169, 175 (1969). 2 Goldstein, A., in preparation. 3 Mellett, L. B., and L. A. Woods, J. Pharmacol. Exp. Ther., 116, 77 (1956). 4Cochin, J., J. Haggart, L. A. Woods, and M. H. Seevers, J. Pharmacol. Exp. Ther., 11, 74 (1954). 5 Szerb, J. C., and D. H. McCurdy, J. Pharmacol. Exp. Ther., 118, 446 (1956). 6 Mule, S. J., and L. A. Woods, J. Pharmacol. Exp. Ther., 136, 232 (1962). 7Kornetsky, C., and J. Cochin, Fed. Proc., 23, 283 (1964). 8 Ungar, G., and M. Cohen, Int. J. Neuropharmqcol., 5, 183 (1966). 9 Lowry, 0. H., N. J. Rosebrough, A. L. Farr, and R. J. Randall, J. Biol. Chem., 193, 265 (1951). 10 Carey, N. H., and A. Goldstein, Biochim. Biophys. Acta, 55, 346 (1962). 11 The data in the third column, 72% and 79% ultrafiltrable, are presumed to be in error, for they are not part of any systematic effect of dilution. In any case, they do not alter any of the conclusions. 12 Solutions of ferredoxin (mol wt 6012) and horse heart cytochrome c (mol wt 12,400), 5 mg/ml and 1 mg/ml, respectively, in 0.005 M Tris buffer pH 7, were subjected to ultrafiltration as described under Method&. The ultrafiltrates contained, respectively, 20% and 4% of the original concentrations, as determined by optical density at 285 and 400 nm. 13 The conclusion about binding to a macromolecule might not be correct if proteolysis occurred during the preparative procedures. Samples were kept below 50C throughout to minimize proteolysis. 14 Sample calculWion: 1.88 ,g/ml X 0.018 Ml = 0.034 gg in brain; brain weight 400 mg X 77% brain water = 309 Il brain water; 0.034/309 = 0.110 Mg/ml due to drug in blood; this is 13% of 0.83 /g/ml in homogenate (Table 2, col. 1). '6Mllthers, K., Ada Pharmacol. Toxicol., 19, 235 (1962). Downloaded by guest on September 30, 2021