The effect of detergent selection on retinal outer segment A280/A500 ratios.

A280/A500 purity ratios for 5 different retinal rod outer segment (ROS) preparations were determined without prior centrifugation, in 1-cm light path cuvettes, in 1% concentrations of 5 detergents. Mean ratios were 3.66 + 0.23 for Brij 96, 3.56 ± 0.20 for cetyltrimethylammonium bromide, 3.12 ± 0.14 for Na-cholate, 5.62 ± 0.34 for digitonin, and 1.78 ± 0.07 for Triton X-100. Ratios in Triton were depressed by a large UV absorbance of the detergent, which prevented accurate photodetection of the UV absorbance of the protein. Brij and digitonin also displayed significantUV absorbance. Purity ratios in digitonin were also elevated by UV lightscattering associated with undissolved, sedimentable protein. Differences evidently attributable to interaction of a particular detergent with ROS protein or lipoprotein persisted even when these UV effects were minimized by use of 1-mm light path cuvettes and centrifugation of nonsolubilized material prior to spectrophotometry. Such differences were minimal, however, among the 3 nonionic detergents octyl glucoside, Emulphogene and Ammonyx LO. At 11 gM rhodopsin and detergent concentrations, the 3 nonionic detergents displayed negligible A280 and no significant sediment upon centrifugation. In the presence of excess ROS, Emulphogene extracted more protein from the ROS than did Ammonyx LO or octyl glucoside. Despite differences in color of the unextracted, bleached pellets resulting from extraction by these detergents, Emulphogene spectra of the pellets revealed qualitatively similar contents. We have selected Emulphogene for A280/A500 ratios because (a) it does not absorb in the 278-nm range; (b) it does not interfere with Lowry protein determinations while all ofthe other detergents except octyl glucoside do; (c) it apparently extracts more ROS protein than do Ammonyx LO or octyl glucoside when ROS is in excess; and (d) it is much cheaper than octyl glucoside.

A280/A500 purity ratios for 5 different retinal rod outer segment (ROS) preparations were determined without prior centrifugation, in 1-cm light path cuvettes, in 1% concentrations of 5 detergents. Mean ratios were 3.66 + 0.23 for Brij 96, 3.56 ± 0.20 for cetyltrimethylammonium bromide, 3.12 ± 0.14 for Na-cholate, 5.62 ± 0.34 for digitonin, and 1.78 ± 0.07 for Triton X-100. Ratios in Triton were depressed by a large UV absorbance of the detergent, which prevented accurate photodetection of the UV absorbance of the protein. Brij and digitonin also displayed significant UV absorbance. Purity ratios in digitonin were also elevated by UV lightscattering associated with undissolved, sedimentable protein. Differences evidently attributable to interaction of a particular detergent with ROS protein or lipoprotein persisted even when these UV effects were minimized by use of 1-mm light path cuvettes and centrifugation of nonsolubilized material prior to spectrophotometry. Such differences were minimal, however, among the 3 nonionic detergents octyl glucoside, Emulphogene and Ammonyx LO. At 11 gM rhodopsin and detergent concentrations, the 3 nonionic detergents displayed negligible A280 and no significant sediment upon centrifugation. In the presence of excess ROS, Emulphogene extracted more protein from the ROS than did Ammonyx LO or octyl glucoside. Despite differences in color of the unextracted, bleached pellets resulting from extraction by these detergents, Emulphogene spectra of the pellets revealed qualitatively similar contents. We have selected Emulphogene for A280/A500 ratios because (a) it does not absorb in the 278-nm range; (b) it does not interfere with Lowry protein determinations while all ofthe other detergents except octyl glucoside do; (c) it apparently extracts more ROS protein than do Ammonyx LO or octyl glucoside when ROS is in excess; and (d) it is much cheaper than octyl glucoside.
The purity of rhodopsin and of retinal rod outer segment preparations from which it is derived have long been estimated in detergent solutions by comparing absorbance at the absorption maximum of rhodopsin (about 500 nm, depending on species) with absorbance at some wavelength which indicates the presence of contaminants, typically 280 nm for protein or 400 nm for hemoglobin, cytochromes, or non-rhodopsin pig-* This research was supported by a grant from the National Eye Institute (to D. M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ments of retinal origin (1)(2)(3)(4). Thus an elevated A280/A500 ratio would betray the presence of extraneous protein, and an elevated A400/A500 ratio would betray a non-rhodopsin pigment. If one is elevated, the other is likely to be also, except in cases where an elevated A400/A500 ratio arises from bleaching of rhodopsin rather than contamination. If detergent solutions are to be bleached to determine difference spectra, 0.1 M hydroxylamine is typically present at neutral pH to trap free retinal as its oxime rather than permit it to rebind to protein as adventitious Schiff base pigments.
During the early days of rhodopsin purification, A280/A500 ratios almost always were determined in digitonin solutions (4). In recent years, a proliferation of detergents has arisen, each with its own special advantages for extraction, bleaching, regeneration, or reconstitution of rhodopsin or ROS' fragments. Rigorous kinetic and thermodynamic analysis of rhodopsin behavior in solution requires as pure rhodopsin as possible. However, in the last decade active soluble enzyme complexes, such as cGMP phosphodiesterase (5)(6)(7)(8), and two or more kinases (9)(10)(11)(12) have been discovered to co-purify with rhodopsin, at least until washing of the ROS. This has levied even more stringent requirements on preparation purity than did rhodopsin work, because of the possibility of extraneous origin of these mobile proteins. In attempting to reconcile preparative procedures and results from different laboratories, we have often been confused by reported (or absent) purity ratios. Since in recent years, most investigators have used A280/A500 ratios rather than A400/A500, we focus here on the former. We examined nine papers appearing in the Journal of Biological Chemistry in the last 18 months reporting observations on rhodopsin or ROS preparations. Of the nine, two gave the A280/A500 ratio in 1% Ammonyx LO, one gave the ratio in 50 mM CTAB, and one gave the ratio without specifying the detergent, as one of us has also done (7). Five gave no ratio, but referred to earlier reports, of which one gave the A280/A500 ratio in 45 mm CTAB, one gave it in 1% Ammonyx LO, one (referenced twice) gave it in 3% Ammonyx LO, and one gave no A280/A5w ratio but did provide data comparing rhodopsin and protein content. A broader search of other journals revealed a diversity comparable to that found in this journal. Two inferences can be made from this admittedly limited survey: first, that a significant minority of investigators do not use this measure of purity, or report none at all; and second, that a significant number (including ourselves until quite recently) think that the detergent used for determininig 1.12-1.14) or discontinuous (densities, 1.12, 1.14, and 1.16) sucrose gradients as previously described (14). All operations were performed in dim red light unless otherwise specified. The top band from the continous gradient is usually purer, but the top band from the discontinous gradient has a higher rhodopsin yield with less fragmentation of the ROS. Some of these preparations were made in the presence of 0.15 M KCl, to increase retention bv the ROS of cGMP phosphodiesterase activity (15). The ROS preparations, taken directly from the gradient, were stored in liquid N2 prior to use. In the initial experiments, aliquots of 5 to 100 /gl (approximately 50 gg-l mg of Lowry (13) protein) of thawed ROS suspension were dispersed in neutral solutions of 1 of 5 detergents, without sedimentation, to a final volume of 1 ml in a thick-walled quartz cuvette. Final concentration of detergent was 1% w/v. The ROS-detergent complex also contained previously neutralized 0.1 M NH20H to prevent formation of adventitious visual pigments during rhodopsin bleaching. The matched reference cuvette contained detergent and NH20H but no ROS (nor buffered sucrose, in which ROS were suspended). Absorption spectra were taken before and after bleaching of the cuvette contents by fluorescent laboratory light. In the second set of experiments, three of the same detergent systems were subjected to a more detailed examination of spectral effects on a single ROS preparation. CTAB, digitonin, octyl-f8-D-glucoside, Emulphogene, and sodium cholate were procured from Sigma, and Triton X-100 originated from Rohm and Haas. Ammonyx LO was a gift from Onyx Chemical Co. of Jersey City, NJ. Digitonin solutions free of turbidity were prepared by the method of Bridges (16). Fig. 1 shows representative dark and bleached spectra for the same ROS preparation in Triton X-100, digitonin, and CTAB. None of these solutions was centrifuged prior to spectrophotometry. The ratios of A2s0/A500 determined on the 3 dark spectra for this ROS preparation were 2.0 for Triton, 6.2 for digitonin, and 3.9 for CTAB. Except for digitonin, these ratios were largely independent of ROS concentration within a range of 5-100 gl/1 ml final volume in the cuvette. In the case of digitonin, it was not possible to increase ROS concentration to the same levels as other detergents without also increasing turbidity. Table 12 expands the ratios of Fig. 1 to 2 additional detergents and 4 additional ROS preparations. The means of the ratios for each detergent appear in the right-hand column, along with the standard error calculated for the 5 ROS preparations examined in the presence of that detergent. It is quite evident that Triton X-100 produced uniformly lower A28o/A50" ratios, and digitonin produced higher ratios than the other 3 detergents (p < 10-4 by Student's t-test, comparing Triton X-100 with cholic acid, or digitonin with Brij). Table II presents AA500 (dark minus bleached) divided by the size of the aliquot of ROS, for the same spectra on which the purity ratios in Table 1 were determined. Again, within the limited range of ROS aliquots used (5-100 ,l) the size of 2Portions of this paper (including Tables I-V  Dark ( , with absorption maxima at 500 nm) and bleached (---, with maxima around 365 nm) spectra of the same ROS preparation (B52K) in the presence of 1% Triton X-100 (T), 1% digitonin (D), and 1% CTAB (C). None of the solutions was centrifuged prior to spectrophotometry. Aliquots of ROS/1-ml cuvette were 40 /ul for Triton X-100, 30 /21 for CTAB, and 5 ,ul for digitonin. Larger aliquots for digitonin resulted in noticeable turbidity; 0.1 M NH20H was always present, and pH was 7. A2m was taken from the dark spectrum for purity ratio. Length of light path in Cary 14 spectrophotometer was 1 cm. the ratio did not depend upon aliquot size. However, in digitonin, this normalized AAwoo was consistently higher than the mean of the 5 detergents together, and the variance of the measure among the 5 ROS preparations in digitonin was also elevated. Thus the elevated A2s8o/A5o00 ratio in digitonin arose despite a somewhat higher rhodopsin extraction/mi of ROS. Fig. 2 shows that digitonin has a considerable UV absorbance (A250 in H20 0.18 at 1% concentration versus H20 blank), although less than that of Brij 96 (A276 0.14 at 0.5%, 0.28 at 1%) or Triton X-100 (A275 0.24 at 0.01%, 24 at 1%). Minor differences in effective detergent concentration between the sample and reference cuvettes may thus disproportionately alter A280 of ROS preparations, and in fact neither sample nor reference photodetector probably detects enough light for a valid comparison to be made in the case of Triton X-100.

RESULTS
Studies of the solubilization of ROS disc membranes by octyl glucoside indicate that only the detergent in the micellar form is effective in solubilization. A critical ratio of micellar octyl glucoside to rhodopsin of 270:1 was required for complete solubilization of the disc membrane (17). A rough meas-  High UV and visible spectra of the 5 detergents. The reference cuvette contained water for all but CTAB, for which the reference cuvette and the sample cuvette contained 1 M citrate, pH 6. All other sample cuvettes were at pH 7. ure of the solubilizing power of the detergent levels employed in the present study can be obtained by comparing the ratios of micellar detergent to rhodopsin used here, to the comparable ratio for octyl glucoside (270:1). The micellar detergent to rhodopsin ratios revealed in Table III are at least a factor of 10 greater for the detergents reported in this study than is required for complete solubilization of the disc membrane by octyl glucoside. While this is likely an appropriate comparison for Triton X-100 (22) and the structurally related detergents Brij-96 and CTAB, it may not be appropriate for sodium cholate and digitonin, which appear to solubilize by a different mechanism than either the previous detergents or octyl glucoside (21).
To pursue these problems further, we repeated experiments with digitonin, Triton X-100, and cholate, centrifuging the detergent-dispersed ROS preparations to reduce turbidity, and employing 1-mm light path cuvettes to reduce UV absorbance by detergents. Once again, the detergent systems were compared using the same ROS preparation, making an effort to keep both the initial ROS aliquot size and the range of rhodopsin concentrations comparable in each system. Fig. 3A shows a digitonin solution of the ROS examined in a 1-cm light path after a 1:1 dilution of an earlier, very turbid solution. The purity ratio in A was estimated at 4.08, at a rhodopsin concentration of 7.88 gM. Upon centrifugation (B) a large pellet of unsolubilized ROS sedimented, bringing about a 67% reduction in A280, but only a 33% reduction in rhodopsin. The resulting purity ratio was 3.09. Use of a 1-mm light path (C) slightly increased A280, with little effect on the visible spec-trum. In D, the same aliquot size which resulted in a highly turbid digitonin solution was brought to 1% Triton and examined in 1-mm light path. A2wg was off scale, rhodopsin was 24 gM, and no unsolubilized ROS sedimented upon centrifugation. Upon dilution (E) the purity ratio was found to be 4.0, which was not appreciably changed by centrifugation (F) since no ROS sedimented. However, use of a 1-cm light path drastically reduced A280 (G) without an effect on the visible spectrum. In H, the same initial ROS aliquot was solubilized in 1% cholate (top right-hand spectrum) and then diluted 1:2 with cholate. The purity ratio for the dilute ROS was 2.76. Centrifugation (I) resulted in a small pellet, but only a small reduction in purity ratio. Use of a 1-mm light path (J) resulted again only in a small lowering of the ratio. Additional experiments revealed that up to 4-fold dilutions of centrifuged solutions of ROS in these 3 detergents had no significant effect on the purity ratios, although, in the 1-mm cuvettes, there were limits to the accuracy of measurement because of noisy records.
In Table IV are presented ratios of A2wi/Aooo and micellar detergent/rhodopsin for cholate, octyl glucoside, Emulphogene, and Ammonyx LO, at rhodopsin and detergent concentrations comparable to the previous experiments. Octyl glucoside, Emulphogene, and Ammonyx LO are powerful nonionic detergents often used in contemporary studies of the functional dependence of rhodopsin upon its lipid environ- Table 3. Ranges of rhodopsin concentrations typically eniployed in the experiments described, together with estiwates of critical micellar concentrations and ranges of the ratios of micellar detergent to rhodopsin concentrations (woO/mol). The micellar detergent concentration is derived by subtracting the CMC from the total detergent covicentration. The corresponding ratio provided by Stubbs (17,23). Unlike Triton X-100, they have no appreciable absorbance in the 280-nm region.3 After the initial dispersion in 1% detergent solution, at approximately double the rhodopsin concentration shown in Table IV, centrifugation produced only minimal clearing of these three detergent solutions, while a small pellet and demonstrable clearing resulted in the case of cholate. The difference probably arises from the greater ratios of micellar detergent to rhodopsin estimated for the nonionic detergents as compared to cholate.  cuvettes with 1 cin l ight path. The reference cuvette contiined the sanie solution without ROS. The original spectrumw (not shown) of this preparation was highly turbid, displaying an A280 greatly in excess of 2, and an A500 of 1.36. This was diluted 1:1 with 1i digitonin, Q.1 M NH20H before recording A. A280 (dark) was estimated as 2.08 frtni the dark spectrLni at upper Teft (arrow), and A500 (lower right arrow) was 0.51. The estionated purity ratio was 4.08. Rhodopsin concentration was 7.88 pM determined by dividing the difference0A500 between dark (arrow) and bleached spectra, by the 1 ml niolar extinction coefficient 40,600. B -Unbleached A was centrifuged at 30,000 x 9 for 20 a in. prior to the same detenilnations as in A. A large pellet sediniented, rhodopsin concentration decreased to 5.0 MM, and the new ratio of A280 to A500 (arrows) was 3.09. C -Saie as A, but using 1 nii light path cuvettes. Rhodopsin was 5.47 pM, and the A280 to A500 ratio was 3.27. D -0.5 ml of ROS was dispersed by Oh. nil of a solution of 10t Ho pellet was obtained and the spectrair was essentially unchanged. Rhodopsin wes 9.gt5 IM. G -F was examined with 1 cm light path, rhodopsin was 9.61 1M, but decrease in A280 (absorption max. actually occurred at 290 nn) brought purity ratio down to 1. of ROS. Emulphogene and CTAB demonstrated the highest quantitative extraction of ROS protein, but the unextracted residue from CTAB evidently contained little or no rhodopsin. The first three nonionic detergents in Table V left an unextracted pellet which upon bleaching exhibited a 310-nm peak of unidentified origin. Differences in color of the bleached pellets were also obvious to the naked eye.

DISCUSSION
Since the purity of each ROS preparation was a constant, the large differences observed in purity ratio when the preparation was dispersed in the different detergents of the initial experiments (Tables I and II; Fig. 1) have a clear implication for purity estimates. If differences exist not only in preparative procedures but in the detergent used to determine the ratio, the ratio becomes essentially meaningless. To dramatize this, we point out that the ratio of 1.78 ± 0.07 we demonstrated above for Triton-dispersed ROS membranes is precisely what Heitzmann reported (29) for very pure rhodopsin (1.7-1.8) in Emulphogene solutions of ROS prepared by our method and subsequent chromatography.
Differences among purity ratios in various detergents arise from 3 origins. First is the degree of solubilization of the ROS in the detergents, without respect to selectivity among different protein moieties. At 1%, digitonin obviously is the poorest agent among those examined. If it is to be used, centrifugation of preparations is imperative. This is not always feasible when very small aliquots of widely different ROS preparations are stored in large numbers in the freezer for subsequent determinations of rhodopsin and of purity ratios. In contrast, Triton, Emulphogene, and CTAB are very efficient solubilizers, even at ROS concentrations clearly in excess of the capacity of other detergents ( Table V).
The second origin of differences is the intrinsic UV absorbance of the detergents. This is manifestly intolerable in the case of Triton, undoubtedly a problem with Brij, and possibly with digitonin. It does not appear to present a difficulty with cholate, although its small UV absorbance may contribute to a slightly higher A280/A500 ratio (Table IV) than was found in three nonionic detergents with even less UV absorbance. To minimize contributions to the purity ratio by intrinsic UV absorbance of Triton, Brij, digitonin, or other detergents, use of 1-mm light paths is not an attractive alternative as a routine laboratory procedure because of required changes in cuvette racks and slide-wires, noisiness of recordings at low concentrations and the fact that even at 1 mm or even 0.5 mm, UV absorbance of the detergent may still significantly exceed that of the protein, as was the case in Fig. 3, E and F. Thus the recorded UV absorbance in those spectra is still not necessarily accurate.
A third origin of differences is apparently attributable to the kinds and amounts of proteins solubilized by each detergent. After minimizing UV absorbance contributions from turbidity or the detergent itself, substantial differences among the ratios in digitonin (3.18, average of Fig. 3 Fig. 3, H, I, and J) still remain. Their interpretation is uncertain, but some differences in solubilizing properties are known. For example, digitonin has been shown by analysis of the behavior of an electron spin probe attached to rhodopsin to exert a smaller effect on rhodopsin conformation than Triton X-100, CTAB, and other detergents (30). This observation was reinforced by determinations that digitonin altered thermal stability of rhodopsin markedly less than Triton X-100, CTAB, or Ammonyx LO, and also increased the rate of metarhodopsin II production less than the other detergents (23).
In the present study, the differences in extraction of excess ROS by 8 detergents (Table V) also appear to imply important differences among the detergents with respect to selectivity for protein or lipoprotein moieties in the ROS or its inevitable contaminant membrane fractions. CTAB, for example, apparently leaves behind unextracted protein of non-rhodopsin origin and would consequently appear to be the detergent of choice for purification of rhodopsin, although not for determining ROS preparation purity. Why it would produce a higher A280/A500 ratio than cholate (Table I) is unexplained. In contrast with CTAB, the 3 nonionic detergents Emulphogene, Ammonyx LO, and octyl glucoside left behind an unextracted protein residue clearly containing rhodopsin, although differences were observed in pellet color to the naked eye as well as in extractive efficiency. We make no pretense in the present study of rigorous interpretation of such differences. In other detergents, they may very well contribute to observed differences in A280/As00 ratios even when UV absorption and turbidity have been minimized.
To resolve confusion among purity ratios in different detergents, we propose the following criteria. Select detergents with low UV absorbances, good extraction efficiences for excess ROS, and A280/A500 ratios in the middle range (2.5-3.5) for ROS purified on sucrose density gradients. Further purification to rhodopsin will bring the ratio down to less than 2.0. The nonionic detergents octyl glucoside, Emulphogene, or Ammonyx LO would seem acceptable on the basis of Table  IV ratios and the differences observed in Table V. Of these three, Emulphogene may be most convenient. Octyl glucoside is quite expensive, and Ammonyx LO can be bought at present only in 50-gallon drums, although the Onyx Chemical Co. has thus far been generous in providing pint volumes of it. Other detergents may also meet these criteria. For our own use we have selected Emulphogene because (a) it does not absorb in the 278-nm range; (b) it does not interfere with Lowry protein determinations while all of the other detergents except octyl glucoside do (The Biuret method of Gornall et al. (31) was also unsuccessful with the same detergents.); (c) it apparently extracts more ROS protein than Ammonyx LO or octyl glucoside when ROS is in excess, and (d) it is much cheaper than octyl glucoside.
Since visible difference spectra do not appear to be significantly affected by differences in intrinsic UV absorbance of the detergents (Fig. 3), one might also use detergents only to determine rhodopsin concentration by difference spectrum and determine protein by the Lowry method, expressing rhodopsin purity directly as nanomoles/mg of Lowry protein. However, Lowry estimates may render high values in the presence of elevated lipid content of membranes (32), a caution which is especially applicable to the ROS. Thus we do not recommend this alternative.
The yeast his4 multifunctional protein. Immunochemistry of the wild type protein and altered forms. The third author's name should be Raj K. Sehgal Ramunas Bigelis, Joseph K. Keesey, and Gerald R. Fink Page 5150, Fig. 10: Fig. 10 is reprinted here from a new original print supplied by the author:

I
We suggrst that subscribers photocopy these corrcctions and insert thc photocopies at the appropriatc places where the article to bc corrcctctl originally appeared. Authors are urgrd to introduce these corrections into :my reprints thry tlistributc. Sccondary (abstract) services ore urgcd to carry notice of these corrcctions as prominently as thcg carried the original abstracts.