Membrane Receptors as General Markers for Plasma Membrane Isolation Procedures THE

Specific cell surface membrane receptors, labeled by forming a complex with low concentrations (about 10--9 M to 10--10 M) of a highly radioactive (125-I, carrier-free) ligand, can serve as simple, reliable, sensitive, and quantitative markers for plasma membranes in fractionation procedures. 125-I-Labeled insulin, cholera toxin and the plant lictins, wheat germ agglutinin (WGA), and concanavalin A are the receptor ligands used for labeling plasma membranes. Plasma membranes are labeled before homogenization by incubating intact cells briefly at 24 degrees or 4 degrees, or by very brief in situ perfusion of the organ, with the 125-I-Labeled marker. After removing the free 125-I-labeled ligand from the medium by washing (at 4 degrees), the membrane-marker complex remains intact over prolonged (days) periods of time at 4 degrees. Labeling occurs nearly exclusively on the cell surface, the specificity of this plasma membrane reaction is maintained through homogenization and fractionation, and little dissociation of the complex, detectable exchange of label, or aggregation occur even upon prolonged incubation of the homogenates. When desired, the complex can be dissociated deliberately by manipulating experimental conditions such as temperature or by adding specific simple sugars. The most generally suitable marker appears to be WGA. At least in certain tissues (e. g. fat cells) labeling of the plasma membrane with 125-I-WGA and 125-I-isnulin can be performed equally well and selectively in homogenates as in the intact cell. 125-I-Cholera toxin cannot be used in homogenates because of significant binding to nuclei. The use of 125-I-labeled WGA as a specific plasma membrane marker is illustrated in following the course of fractionations, and in quantitating the yield and purity, of plasma membranes from fat cells, lymphocytes, and liver. The results are compared with simultaneous measurements of the plasma membrane enzyme "markers," ATPase, 5-nucleotidase, and basal as well as hormone-stimulated adenylate cyclase activities. The fractionation of liver plasma membranes by aqueous dextran-polyethylene glycol two-phase polymer systems and by conventional differential centrifugation procedures arealso quantitated with the marker, 125I-WGA. Substantial quantities of plasma membrane material are no recovered in the interphase of the two-phase polymer system. Conventional liver fractionation procedures which retain, for further purification, only the readily sedimented pellet (2000 times g, 15 min) discard a very large (at least 70%) questenal hy


SUMMARY
intact cells briefly at 24" or 4", or by very brief in situ perfusion of the organ, with the 1251-labeled marker. After removing the free '21-labeled ligand from the medium by washing (at 49, the membrane-marker complex remains intact over prolonged (days) periods of time at 4". Labeling occurs nearly exclusively on the cell surface, the specificity of this plasma membrane reaction is maintained through homogenization and fractionation, and little dissociation of the complex, detectable exchange of label, or aggregation occur even upon prolonged incubation of the homogenates. When desired, the complex can be dissociated deliberately by manipulating experimental conditions such as temperature or by adding specific simple sugars.
The most generally suitable marker appears to be WGA. At least in certain tissues (e.g. fat cells) labeling of the plasma membrane with l*jI-WGA and *261-insulin can be performed equally well and selectively in homogenates as in the intact cell.
12"I-Cholera toxin cannot be used in homogenates because of significant binding to nuclei. The use of lZ51-labeled WGA as a specific plasma membrane marker is illustrated in following the course of fractionations, and in quantitating the yield and purity, of plasma membranes from fat cells, lymphocytes, and liver. enzyme "markers," ATPase, 5'-nucleotidase, and basal as well as hormone-stimulated adenylate cyclase activities. The fractionation of liver plasma membranes by aqueous dextran-polyethylene glycol two-phase polymer systems and by conventional differential centrifugation procedures are also quantitated with the marker, 12"1-WGA. Substantial quantities of plasma membrane material are not recovered in the interphase of the two-phase polymer system. Conventional liver fractionation procedures which retain, for further purification, only the readily sedimented pellet (2000 x g, 15 min) discard a very large (at least 70%) fraction of the total plasma membrane. These methods are compared with fractionation procedures which utilize isotonic sucrose.
These studies also identify a special liver plasma membrane fraction which in isotonic sucrose sediments slightly less readily than mitochondria. This fraction constitutes about one-third of the total plasma membrane and is highly enriched in glucagon-sensitive adenylate cyclase activity.
The m&hods currently in use for the specific identification and quantitative assay of plasma membranes are generally considered to be cumbersome and inaccurate (l-9). Although electron and phase contrast microscopy are very useful for the purposes of identification and for detecting certain impurities, these are not quant,it,ative or very discriminating tools. Biochemical markers are thereforc csscrnial for the isolation of plasma mcmbrancs.
A satisfactory marker should be present only in the plasma membrane and it should be easily and accurately measured.
>lost studies have relied on the use of enzymes (e.g. 5'.nucleotidase, Na+K+-dependent ATPase, and phosphatases) as markers for plasma membranes, although only in exceptional cases have such enzymes been shown to be located uniquely in these organelles.
The USC of enzymes as specific quantitative markers for membranes is beset with other difficulties which limit their usefulness (l-9).
The basic approach in the present study is to label surface re-cell membranes was performed by rapidly (about 10 s) injecting, ceptors with very low concentrations (about 10h9 M to lo-" M) as a single bolus, 5

RESULTS
Labeling Plasma Membranes with lz51-Labeled Insulin, WGA, and Cholera Toxin in Fat Cells 1251-lns-ulin-Idmt fat cells were incubated (30 min) at 24" with 1251-labeled insulin (8 x 1OWo M) followed by washing at 4" to remove the free 1251-insulin in the medium.
The pattern of radioactivity that results when the total particulate material of the homogenates of these cells is applied on linear (54.1~~ to 27.6%, w/v) sucrose density gradients (Fig. 1B) reveals that the lz51-insulin bound to the plasma membrane is readily separated from a very small amount of free 12%nsulin, which is retained in the 8.5y0 isotonic sucrose fractions (above arrow, Fractions 28 to 36). TWO distinct peaks of radioactivity ( Fig. IB), which correspond to the two fractions enriched in plasma membranes (Fig. lA), arc discerned.
The existence of these two separate plasma membrane fractions was described previously by Avruch and Wallach (36). All of the radioactivity disappears from the gradients if the cells are labeled in the presence of 1 PM native insulin (Fig. lB), indicating that the radioactive peaks probably represent specific, saturable binding sites that have previously (18,34) been identified as cell surface insulin receptors. Little radioactivity is observed in the nuclear (Fraction I), mit,ochondrial (Fractions 8 to la), or soluble fractions (Fig. 1, A and B). The small amount of free ?-insulin detected is consistent with the extremely stable nature of the insulin-receptor complex at 4' (20, 27). When the 1251-irlsulirl-labeling reaction is performed with whole homogenates or with the total particulate pellet of homogenates rather than with intact cells, the radioactivity pattern observed on sucrose density gradient's is nearly identical with that described in Fig. 1B. The results of these studies are consistent with the suggestions that insulin recept,ors are present exclusively in the plasma membrane (18,19). Thus, specific labeling of surface membranes with '2"1-insulin can be performed before or after the cells arc disrupted.
The small amount of radioactivity detected in the mitochondrial fractions ( Fig. 1B) almost cert,ainly represents residual contamination with plasma membranes, as is described. '2"Z-WGA-As described above for iodoinsulin, plasma membrancs can be specifically labeled by incubating intact fat cells for very brief periods with very low concentrations of '2"I-WGA (Fig. 1). Under these conditions, gross membrane physical properties (e.g. aggregation) are not significantly affected. The unbound 12"I-WGA is removed by washing (at 4") before homogenization.
The lectin bound to the cells adheres extremely t,ightly to the plasma membrane during the mashing, homogenization, and ccntrifugation procedures provided these are performed at 4O.
On linear sucrose density gradients, the distribution of radioactivity ( Fig. 1C) is highly reproducible and is very similar to that drscribrd for 1251-insulin. However, a significantly greater quant,ity of radioactivity is present in the isotonic sucrose fractions (above the arrow in Fig. 1) when 1251-insulin is used com- Isolated fat cell suspensions, prepared from 10 rats (120 g), were divided equally into 3 parts (10 ml each) and incubated at 24" with 1261-WGA (5 X lo6 cpm, 50 ng per ml) and 1251-insulin (7 X 10" cpm, 8 X lo-l0 M) for 10 (WGA) or 30 (insulin) min in KRB-lyO albumin.
The free, unbound radioactivity in the medium was removed by washing twice with ice-cold isotonic sucrose-Tris buffer containing 0.25 M sucrose, 10 mM Tris-HCl, and 1 mM EL)TA. The pattern obtained from cells exposed (5 min, 24") to 1 PM native insulin before labeling with 1261-insulin is shown in B (0); a similar suppression of radioactivity occurs if native WGA is added before lz51-WGA (not shown).
Labeled fat cells were homogenized (30 s) with a Polytron PT.10 (Brinkman) set at 2.5. The entire particulate material obtained by centrifugation at 40,000 X 9 for 30 min was suspended in the same buffer using a Polytron PT.10 at a setting of 2.5 for 10 s and layered on top of a 27.ml linear sucrose gradient (54.10/, to 27.6%, w/v) containing 10 mM Tris-HCl and 1 mM EDTA, pH 7.4, and centrifuged at 25,000 rpm for 60 min at 0" in a Beckman SW 27 rotor. One-milliliter fractions were collected from the bottom of the gradients.
Radioactivity is counted in a well-type gamma counter. The interphase of the linear sucrose gradient and the isotonic sucrose-Tris buffer is indicated by the arrow. pared to 1251-Vi~GA. Virtually no radioactivity is found on t,he isotonic sucrose fractions for '2%WGX.
The small amount of radioactivity present in this region in the 1251.insulin experiments disappears when the cells are incubated with native insulin before labeling (Fig. 1B). It probably represents free 12"I-insulin which has spontaneously dissociated from t,he membrane during the procedures.
These results arc consistent \\ith the much slower rate of dissociation (20, 24) of RGA-membrane complexes (to be described in more detail) compared to the insulin-receptor complexes provided t,hat low concentrations of the iodolectin are used for labeling.
liecause the spontanrous rate of WGA-or insulin-membrane dissociation is highly trmprrature-dependent, great care must be taken to maintain the temperature at 4" and to avoid contact with warm glassware or exposure to warm air. Experimental conditions are the same as described in Fig. 1 except that the whole homogenate (after removal of the fat vake) rather than the total particulate fraction is layered on the linear sucrose gradient.
3. Comparison of the distribution of 1261-cholera toxin on sucrose density centrifugation when membrane labeling is performed in intact fat cells and in whole particulate preparations from homogenates. Isolated fat cells from 8 rats (150 to 175 g) were divided into 2 equal parts. One part (A) was incubated with 12~I-cholera toxin (5.1 X lo6 cpm) for 5 min at 24" in 5 ml of KRB-1% albumin.
The other (B) was treated the same way but without 12"1-cholera toxin.
Both sets were washed and homogenized as described in Fig. 1

491
The 1251-WGR marker also permits detection of a small amount of plasma membrane contaminationin the mitochondrial fractions (Fig. IC).
i%WGA can also be used to label selectively plasma membranes in fat ctll homogenates (10 min, 4") because the sucrose den&y centrifugation patterns obtained (Fig. 6) arc nearly idcntical wit,h those obt,ained when the cell is labeled before homogenizat,ion (Fig. 1C). Apparently this lectin does not bind in significant quantity to various intracellular organellcs of fat cells, at least under the conditions of low WGA concentrations, depressed temperature, and short incubation periods. 125~.Cholera To&n-Labeling of plasma membranes in intact fat cells with very low concentrations (about lo-i0 RI) of the cell surface glycolipid marker, '""I-cholera toxin, results in the same sucrose density gradient pattern of distribution ( Fig. 2B) as described for iZ51-WGA.
Thr pattern of radioactivity is nearly the same when the entire, unfractionated homogenate is applied to the sucrose gradient as when only the particulate fraction of the homogenate is examined (as in Fig. 1). These results, similar to those obtained when WGA and insulin are used (not shown), indicam again t,hat negligible dissociation of the complex has OCcurred and that very little, if any, ligand is bound t,o soluble cytoplasmic prot,eins. Very substantial radioactivity is detected in the nuclear fractions when labeling with i2jI-choleragen is performed with the particulate material obtained from homogenates rather than with intact cells (Fig. 3). Furthermore, the second, lighter plasma membrane peak is more highly labeled under these con ditions, indicating t,he presence in this fraction of binding structures not accessible t,o the toxin in the intact cell. Thus, iZeIcholera toxin is not a suitable marker for plasma membranes if it is to be used after the cells have been disrupted.

E$ect of Sucrose on Dissociation of Membrane Receptor Jlarkers
The possibility t,hat the absence of radioactive markers in the nuclear and mitochondrial pellets results from an adverse effect on binding of t,he higher sucrose concentrations present in these fractions is excluded by t,he data shown in Table I from the markers appears in the nuclear or mitochondrial fractions ( Figs. 1 and 2). However, more significant plasma membrane contamination, as judged by the presence of 12%WGA, occurs when large quantities of particulate material are applied on the gradients.
In the patterns described in Fig. 4, A and B, for example, small but significant '2%WGA peaks can be seen in the nuclear and mitochondrial fractions. When these two fractions from the gradients are suspended (at 4") in 10 times their volume of isotonic sucrose-Tris buffer and centrifuged at 600 X g and 12,000 X g for 15 min, respectively, most of the radioactivity remains in the supernatant. This again suggests that the small amount of radioactivity in the nuclear and mito- chondrial fractions is indeed due to plasma membrane contamination.
These conclusions are supported by measurements of the distribution of adenylate cyclasc activity in the sucrose density gradients (Fig. 4, C and D). The distribution of catccholaminestimulated enzyme activity is nearly identical with that of the basal (unstimulatcd) enzyme activity. Furthermorc, this distribution is very similar to that of 12"I-WGA (Fig. 4B). These results also indicate that all of the adenylate cyclase activitypresent in these cells is at least grossly associated with a susceptibility to hormonal regulation and is associated with plasma membranes. Very little adenylate cyclase activity is found in the soluble fractions (above arrow in Fig. 4). Furthermore, a 7-fold stimulation of cnzymc activity is observed in the heavier plasma membrane peaks (Fractions 17 to 23), whereas only 2.5.fold stimulation is seen in the lighter peak (Fractions 24 to 27). This indicates functional differences between the two plasma mcmbrane fractions.
The lighter fraction may contain damaged membrane vesicles.
Alternatively, inside-out plasma membrane vesicles, the adcnylate cyclase activity of which is known to be less sensitive to catecholamines (27)) may be present and prcfcrcn tially localized in this fraction. the possibility also exists that dissociation of the 12"1-WGA plasma membrane with the cschange of the label to other intracellular organellcs could occur. These possibilities can bc safely excluded bccausc incubation of whole cell homogonatcs, obtained from cells labeled with iz61-WGA (and washed), for 4 hours at 4' does not alter the distribution of radioactivity on sucrose density gradients (Fig. 5). Lit.tle radioactivity is found in the nuclear or mitochondrial fractions and the relative size and position of the two plasma membrane fractions remain essentially unchanged.
The abscncc of interorganellc "exchange" is probably explained by the extraordinarily slow rate of dissociation of the WGA-membrane complex, whereas the apparent lack of WGA-induced organellc cross-linking and aggregation is probably explained by the very small quantit,ies of 12SI-WGA which arc bound to the cells (generally less than 4 X lo3 molecules per cell) in the labeling reactions.

Reversibility of '251-WGA-Plasma dlembrane Complex
The lectin-mcmbranc complex spontaneously dissociates at a very slow rate, whether measured after the addition of a large excess of the native protein or after washing the cells to remove the free lectin in the medium (24, 27). Howcvcr, dissociation occurs very rapidly and profoundly when a specific simple sugar N-acetyl-nglucosamine (24, 27), or ovomucoid, is added. This "unusual" (24) interaction between the simple sugar and the lectin-membrane complex provides a simple method to remove the membrane-bound lectin under mild conditions. Although 90% of the radioactivity bound to t,he major plasma membrane fraction can be readily displaced, nearly 30~~ of the 1221.WGA present in the lighter fraction cannot be dissociated (Fig. 6) homogenization, 90% of the radioactivity can be dissociated from all of the fractions (Fig. 6, C and D).
These studies, together with the adenylatc cyclase results described above, suggest that the lighter plasma membrane fraction contains a population of insidc-out plasma membrane vesicles that probably form spontaneously during t,he homogenization procedures. However, the lighter mcmbrauc fraction is probably not composed predominantly of insidc-out vesicles because labeling of this peak with i*"l-KGA can be achieved easily after homogenization (Fig. GD), because the major part of the radioactivity can be tlisplacotl by the simple sugar and bccauscl it is believed (23, 55) that WGA does not bind to the inner aspclct of t,lie plasma mc'mbranc.
The rnajor part of the second, lighter ptak may not basically represent a tlistiilvt l)opulntioii of vcsiclcs but is rather the lower end of the normal distribution raiigc of density which is sharptucd by the ending gradient and thus fort,uitously gives the appearance of a distill& peak. The '2"I-iVGA bound to intact cells can be nearly completely (more than 90%) dissociated c~wn after relatively long (e.g. 60 min at 24") periods of incubation.
Tlicst~ fiiiclirigs are corisistt~nt with t,lie data presented carlicr in this report, as well as with previous studies (24), whicli indicat,c that thcrc is negligible intcrnalization of "%WGA by the intact cells. This is an important condition for the use of such reagents as plasma membrane markers.

Quanfitafion of Purify and Recovery of Fat Cell Plasma Membranes
The radioactivity iii prclabcled intact cells is followed during the course of membrane fractionation as described in Table II. On discontinuous sucrose gradients of homogenates, the nuclear fraction was collected from the bottom of the gradients, the mitochoudrial fraction from thtx interface of 54.1 LT and 40y0 sucrose, and the plasma membrane fraction from the interface of 40y0 sucrose and the isotonic sucrose buffer.
Resolution of the vari- ous particulate fractions by these methods is inferior to that achieved by linear sucrose density centrifugation (Figs. 1 to 4). An approximately B-fold purificat,ion (compared to the whole particulate) of plasma membrane is achieved judging by the two different membrane receptor markers used, whereas 2-fold and 'i-fold purification factors are estimated on the basis of 5'.nucleotidasc and ATl'ase activities, respectively (Table II). The data indicate that 5'-nucleotidasc is not a suitable marker for plasma membranes because all of the fractions contain very significant activity and the specific activity in the nuclear fraction is in fact increased 3.fold.
ATl'ase is also not a very specific marker because very substantial amounts of activity, especially when compared to the 12S1-WGA or 1251-insulirl data, are present   (Figs. 1 to 4). The the interfaces of 8% and 40% (plasma membranes), 40% and54.1% intent of these experiments was to evaluate the relative merit of (mitochondria), and the bottom pellets (nuclear) were aspirated these separation methods by using various plasma membrane with a Pasteur pipette, diluted with 6 volumes of isotonic-sucrose-markers. only by one-half and that in the nuclear fraction is increased I .5-fold.
Labeling Plasma Alembranes of Lymphocytes with '25Z-WGA As in isolated fat wlls, over 90y0 of the 12zI-WGA bound to the cells during a short period of incubation can be immediately dissociated by N-acctyl-wglucosaminc and ovomucoid at 4" (Fig.  7). The '?5I-KGA-labeled lymphocytes were incubated for another 24 hours at 4' before studying the reversibility of the bound WGA.
About 20~~ of the radioactivity is lost during this period of time, presumably as a result of spontaneous dissociation or ccl1 breakage, and the half-life of the label is calculated as to be at least 3 days at 4". Equally important, nearly all (over 906-/0) of the bound lQ"I-WGA can be dissociated rapidly with N-acctyl-n-glucosaminc and ovomucoid even after incubating for 24 hours at 4". This data point to the remarkable stability of the membrane complex and to the negligible intracellular sequestration or internalization of t,hc label even during very exaggerated periods of incubation.
Thus, the use of very highly radioactive '29-WGA at very low concentrations should also be an excellent plasma membrane marker for lymphocytes.

Homogenization and centrifugation of lymphocytes labeled
Cth '*"l-WGA rcvcals that nearly all (over 95%) of the radioactivity is recovered in the pellet sedimenting at 40,000 X g for 30 min (Table 111)   present in a single plasma membrane peak (Fig. 8). The small amount of radioactivity in the nuclear fraction probably represents contamination with plasma membranes because most of this can be removed by resuspcnding the material in a lo-ml volume of isotonic sucrose-Tris buffer followed by ctntrifugation at 600 X g for 10 min.
All of the i*bI-WGA bound to particles can be dissociated by the incubation of the particulate sample wit'h Nacetyl-nglucosaminc and ovomucoid prior to ccntrifugation (Fig. ES), suggesting that virtually no inside-out membrane vesicles form with these cells, at least by the use of this method (nitrogen pressure) of cell disruption.
The use of i%WGA can facilitate monitoring the course of purification of lymphocyte plasma mcmbrancs. For example, it is readily determined that by standard methods of differential centrifugation using isotonic sucrose-l'ris buffer the microsomal fraction contains about 40% of the plasma membrane, and that these membranes are purified about 4-fold (Table IV).
Labeling Plasma Membranes of Liver with lz5Z-WGA The plasma membranes of liver cells can be specifically labeled (set "h4atcrials and Llethods" and figure legends) by briefly perfusing the liver in situ with ice-cold buffers containing izSI-WGA or by incubating liver slices for 5 min with '2"1-WGA. The free 1%WGA is removed by soaking and washing (slices or minces of intact, perfused liver) with ice-cold buffers before homogenization.
Upon centrifugation of the homogenates, 90% of the radioactivity sediments in t,hc total part,iculate fraction (Table III).
The free unbound lZKI-WGA in the medium was removed by washing twice with ice-cold buffer.
Labeled lymphocytes were homogenized with a nitrogen bomb in ice-cold isotonic sucrose-Tris buffer.
Three milliliters of homogenate were layered on a 34.ml linear sucrose gradient (54.17, to 27.67,, w/v) and centrifuged at 25,000 rpm for 60 min at 0" in a Beckman SW 27 rotor. The profiles of radioactivity are presented for homogenates incubated (10 min, 4") in the absence (0) and presence (0) of 50 mM N-acetyl-n-glucosamine and 5 mg per ml of ovomucoid before application on the gradients.  Little (1%) radioactivity is rcxcovt>rcld in the mitochondrial fraction, although a substantial amount (14%) is present in the nuclear fraction. The latter, which probably contains the "sheet" plasma membrane material frcqucrnly &scribed for liver homogenates (3,5,7,37,38,40), is much less rich in insulin binding activity relative to total plasma membrane (by the WGA marker) compared to the microsomal and heavy membrane fractions. The distribution of adcnylate cyclasc activity on differential centrifugation confirms the hcterogcneous nature of the liver "plasma membrane" ( Table V). The basal activity of this enzyme is surprisingly high in the mitochondrial pellet, which on the basis of the WGA marker, insulin binding, and glucagon stimulation of adenylate cyclase activity, has very minimal (about lv/,) contamination of plasma membrane The presence of a substantial quantity of glucagon-stimulated adenylate cyclase activity in the nuclear fraction is consistent with the 's51-WGA and insulin binding data described above which show that this fraction is significantly contaminated with plasma membranes. The heavy membrane fraction, which contains about one-third of the total plasma membrane material, is highly enriched with respect to glucagonstimulated adenylate cyclase (Table V). Although adcnylate cyclasc is a useful plasma membrane marker, especially wheel a hormonally stimulated function is specifically examined, it is not a reliable quantitative means of estimating or even monitoring the total dist,ribution of liver plasma mem-branes. Certain fractions (e.g. mitochondrial) contain hormonally insensitive activity, other fractions (e.g. heavy membranes) arc enriched in hormonal responsivcncss out of proportion to t,heir content of "total" plasma membrane, and the total recovery of activity is poor due to the lability of this enzyme.
The presently available conventional procedures (37,39,40) for the isolation and purification of plasma membranes from liver all utilize initially only the material present in the low speed nuclear fraction, all of the other fractions being discarded.
Because the data in Table V suggest that only a small portion of the total plasma membrane material is present in this fraction, studies were performed to quantitate the yield of the plasma membrane obtained in the first step of those conventional methods by using the plasma membrane receptor marker, 1251-WGA (Table Vl). Less than 307, of the total plasma membrane is recovered during this first, starting step in the methods (39, 40) which utilize low ionic strength.
When CaClz is omitted from the sucrose solution, a relatively small proportion (about 1070) of t'he total plasma membrane is present in the nuclear fraction (Table VI). Isotonic sucrose, in the presence of CaClz (37), results in a much higher yield of total membrane in the nuclear pellet, probably as a result of aggregation facilitated by Ca2+. As indicated earlier (Table V), under these conditions most of the plasma membrane is present in the microsomal and heavy membrane fractions.
The plasma membrane material present in the IIUclear fractions is generally believed to represent relatively intact, large sheet membranes (37,39,40). Aqueous dextran-polyethylene glycol two-phase polymer systems have recently been used to purify plasma membranes from nuclear fractions prepared in 1 mM NaHC03 buffer, pH 7.4, containing 0.5 mM CaC& (46). The use of 1251-WGA-labeled plasma membranes from liver in such studies reveals that about 15% of the plasma membrane is sedimented in the nuclear fraction (Fig. 9). About 75% of the radioactivity sedimenting in the nuclear fraction is retained in the interphase of the two-polymer system.
Nearly 45% of the total plasma membrane material sediments between 500 x g and 12,000 x g, but less than 10% of this material is retained at the interphase of the twopolymer system; most of the material sediments to the bottom of the tube.
Of the plasma membrane which sediments between 12,000 x g and 45,000 x g (about 20% of the total), only 14y0 1s retained by the two-phase polymer system; the remaining either sediments to the bottom (48%) or is dispersed in the bottom phase.
Notably, very little (about 10%) plasma membrane Recovery of plasma membranes of liver homogenates during jkst step of commonly used fractionation procedures Y-WGA-labeled liver minces (1.5 g for each) prepared as described in Table V were  The liver minces (1.5 g) were homogenized with a Polytron (setting at 3, 90 s) when 0.25 M sucrose was used (38).
When 1 mM NaHCO$ plus 0.5 mM CaCls solution was used, the homogenate was further diluted to 100 volumes (39). The homogenates were stirred gently for 5 min at 4". The unbroken tissue fragments were separated by filtering through four layers of silk screen. The crude "nuclear" pellets were obtained by centrifugation (4') for 15 min at 2,000 X g. In separate portions of the homogenates, "total particulate" fractions were obtained by centrifugation (4') for 30 min at 40,000 X g; in cases containing sucrose, NaCl (0.1 M), and MgSOl (0.5 mM) were added before centrifugation (38).
These total particulate fractions contained the microsomal membranes.  58 17" a When these pellets are centrifuged at 1,500 X g for 10 min, as described in the methods of Neville (39) and Ray (38), the recoveries are less than 30%. b Crude nuclear pellets were resuspended by 3 strokes in a Dounce homogenizer. c Crude nuclear pellet was resuspended with a Polytron for 10 s (setting 3).
remains in the supernatant after centrifugation for 30 min at 45,000 x g.

Use of 1251-Concanavalin
A as Plasma Membrane

Receptor Marker
The plant lectin, concanavalin A, has many, if not all, of the desirable features described in this report for WGA.
The commercial, low cost availability of this lectin could also facilitate its widespread use as a membrane marker. Excellent binding of low concentrations (0.1 pg per ml) of very high specific activity (about 10 PCi per ~ys) of 1251-labeled concanavalin A ran be demonstrated with a variety of cells, the rate of spontaneous dissociation of the 12%labelcd concanavalin A-membrane complex (22,24) at 4" is nearly as slow as that of RGA, the complex can be dissociated rapidly (and at 4") with cu-mothyl-o-mannopyrarloside (24), and virtually no internalization (24) of the lcctin by the cells occurs undrr conditions similar to those described for WGA.
A serious disadvantage of concanavalin A, however, is the rapid dissociation observed in solutions cont.aining sucrose. In thr present studies it was not possible to utilize this mcmbrane marker in differential centrifugation or in any sucrose gradient system. Concanavalin A should, however: prove useful in studies not requiring homogenization or separation procedures which utilize sucrose or related sugars.
Wallach et al. (56) have recently used coliphage-coupled *2Qoncanavalirl A to label and perturb the plasma membrane density in CsCl density gradients.
The widespread use of enzymic act,ivitiw as specific plasma membrane "markers" suffers from a numb(xr of uncertainties and difficult.&. Seldom have such cnzymcs been shown to be localized exclusively in the plasma membrane (9,57), different patterns of localization may occur betwcn the same tissue in different species or between different tissues in the same species (9,58), and dramatic changes may even occur during the cell cycle (59). The present studies confirm these ambiguities of localization.
For example, t,otal 5'.nuclcotidasc activit,y in fat cell homogenates was found to change during fractionation and very substantial activity was present in norl-plasma membrane fractions.
A serious difficult,y with marker enzymes is that their act,ivity cannot bc accurately measured, quantitated, or conclusively localized in t,he intact cell before homogenization.
Furthermore, the latter procedure may bc injurious or may potentially change the catalytic properties of the enzymes.  Table  III, were centrifuged at 500 X g, 12,000 X g, or 45,000 X g and 100,000 X g. The pellet obtained from each step was suspended in the top phase of the two-polymer phase, mixed gently with the bottom phase, and centrifuged at 3.000 rpm in an Intertional centrifuge Pg-J at 4". The interphase and bottom pellets were collected and radioactivity was determined.
especially that, port,ion which is stimulated by hormones, may accurat,ely reflect a plasma membrane act,ivity, at least in fat cell preparations (Fig. 4). The use of this enzyme, however, suffers from the, difficulties described above for enzyme markers. In addition, spcc'ial problems 1yliich limit its use 011 a more routine basis in&&~ the technically difficult and costly nature of the current assay proccdurcs and its intrinsic lability, especially wit'h respect to hormonal sensit,ivity. Furthermore, its use may be restricted to known hormonally sensitive tissues, especially because in most tixsucs the exclusive localization of this enzyme to the plasma mcrnbrane of the cell remains to be proven. For example, basal as well as hormonally stimulated adenylate cyclasc activity has been described in isolated nuclei of rat liver (62) and ventral prostate (63). The application of membrane receptor markers to evaluate this enzyme in membrane fractionation procedures in liver illustrates some of thcsc problems (Table V) In the latt,rr, the problem of acccssibility of the ccl1 plasma membrane to t,hc 1251-mark(,r in solution is compounded by the inherent ldrrogrncitg of the "cdl surface" due to the existence of a variety of cell types as well as the asymmetry and morphological polarity of unique ccl11 types. Despite this, the useful assessment of fractionation procedures can nevertheless be obtained in tissues like liver with 1251-WGA (Tablcs III, V, and VI and Fig. 9).
In fat cells (Fig. 6), the plasma membrane can be labeled with '2SI-WGA as selectively in homogenates or in total particulate fractions as in the intact cell. This method of labeling may be of pracbical importance in large scale fractionation procedures or for solid tissues where perfusion may be difficult or impossible. In these cases, it may be especially valuable to first prepare a total particulate fraction.3 Thcx USC of lrctins in broken cell preparations, however, must be approached with caution, and in each case the absc~nce of int,racellular binding structures must be demonstrated rigorously before the lcctins arc usc~d in this way. Hormone rcccptors, because of t,hrir localizat,ion in cell surface membranc,s (65, 66), can also be useful markers.
The principal advantage over the other markers described above is t,hat, as demonstrated hcrc for l*%labeltd insulin (Fig. I, Table II), the distribution of label is vcxry similar whether labeling is performed before or after the homogenization of the cells. The 1251-labeled insulin-membrane complex can bc readily dissociated by an excess of native hormone or by warming t.o 24 or 37" (20, 27). On the other hand, the disadvantages of hormones include the restriction of their use to hormone-srnsitivc tissues, the uneven or sclcct,ivc labeling which occurs in crlls whcrc functionally asymmetrical surfaces of the cell exist (e.g. liver), the faster (compared to WGA and choleragen) rate of spontaneous dissociation of the hormone-receptor complex at, 4" (Fig. I), t,lic much greater scarcity of available binding sites on the cell surf&cc, the much larger proportion of label which binds nonspecifically to structures other than rtccptors, the less ficxibility in t,he cell-labcling reaction, and the much greater lability of the radioiodinated hormones compared to the plant lectins and toxins4 The use of 12SI-labelt~d hormone s to label plasma mcmbranc>s in homogcnatcs must bc used with special care and rcquiros proof of selectivity in cvc>ry cast because considcrablc "nonspc,cific" binding to nonreceptor structures may be a serious complication. Y-Labeled cholera toxin cannot be used to label plasma membranes in broken cell preparations because very substantial labeling of nuclei and altered labeling patterns in sucrose density gradients occur under these conditions (Fig. 3). However, this protein should be very valuable when used to label intact cells because under these conditions binding is extremely fast and virtually irreversible (25,26), because the binding sites are ubiquitous and relatively abundant in most cell surfaces, and because no intracellular label can be detected even after very prolonged periods of incubation with intact cells. The principal limitation of the use of this marker is that no mild methods exist for dissociating the extraordinarily tenacious toxin-membrane complex once it has formed. This should not, however, be a serious limitation in the use of this marker because only a very small number of toxin molecules need be bound to the plasma membrane to obtainsatisfactory labeling, and because membrane function will very likely not be altered if the procedures described in this report are followed.
Activation of adenylate cyclase, which is the specific membrane function altered by this protein, requires incubation of the toxin with the intact cell for at least 30 min at temperatures above 20" (25).
3 The present studies demonstrate (e.g. Table III) that over 90% of the plasma membrane material can be recovered in various cell types by centrifugation at 40,000 X g for 30 min (in the presence of NaCl, Cs2+, and Mg2+) to obtain a "total particulate" fraction.
4 Whereas l*bI-labeled WGA, concanavalin A (24), and cholera toxin (25) are very stable and retain excellent binding properties for at least 1 month when stored frozen, the specific binding properties of 1251.insulin and '251.glucagon deterioriate seriously after 1 week. Furthermore, much more care is required in preparing 1251-labeled hormones with good specific binding properties compared to the plant lectins and cholera toxin.
Despite the widespread and accepted use of procedures which utilize the material present in the low speed crude nuclear fractions as the starting material for liver plasma membrane purification (3,37,39,40), and despite the awareness and warning by ly'eville (40) that these methods select only for sheet-type membranes, it has not been possible in the past to quantitate yields or to determine what specific proportion of the total or even of a specialized type of plasma membrane is saved or thrown away by these procedures.
The present studies demonstrate that in the standard methods (37,39,40) most (at least 70%) of the total plasma membrane material of homogenates does not sediment (Fig. 9, Tables V and VI) at the low centrifugal forces used and is thus discarded in the first step of the fractionation procedures.
Furthermore, these methods select a type of plasma membrane that is specialized not only with respect to morphology (sheets) but also to function (Table V). For example, the plasma membrane sedimenting in the nuclear pellet has relatively low insulin binding activity (relative to the WGA marker) when compared with the microsomal or heavy membrane fractions.
Another complication of methods utilizing hypotonic solutions for homogenization is the fragmentation of nuclei, which results in the formation of nuclear membrane material and the release of large quantities of nucleic acids. In contrast, with isotonic sucrose (without added Ca2+) only about 15y0 of the plasma membrane is present in the nuclear fraction (Table VI) and more than 70% of the total plasma membrane recovered is obtained in the less rapidly sedimenting microsomal and heavy membrane fractions (Table V).
A special liver plasma membrane fraction (heavy membranes) has been identified by differential centrifugation in isotonic sucrose (Table V). This fraction, which sediments on top of the mitochondrial pellet and which constitutes a distinct layer which can be readily removed separately, contains large vesicular structures and is highly enriched inglucagoll-sensitiveadenylatecyclase activity.
The relative absence (less than 1%) of plasma membrane (by 12SI-WGA) in the mitochondrial fraction, along with the large quantity of basal activity of adenylate cyclase and its lack of sensitivity to glucagon, and the poor insulin binding capacity suggest that this cellular fraction contains some material with adenylate cyclase activity which is not associated with the plasma membrane (Table V). Another interesting illustration of theuse of the receptor marker method was presented in the studies of liver membrane fractionation by aqueous two-polymer phase systems (46). As described by Lesko et al. (46) these methods clearly do not quantitatively separate sheet membranes (and perhaps large vesicles) from nuclear fractions.
Very substantial portions of the plasma membrane are not recovered in the polymer interphase (Fig. 9). These and the above described studies on liver plasma membrane isolation point again to the great heterogeneity which exists in such membranes with respect to gross physical properties, morphology, and functional characteristics. '1 his is, of course, consistent with the complicated cell composition and morphology of the liver.
Although in the light of this it is difficult to refer very strictly to the plasma membrane in this tissue, the use of the general plasma membrane marker, WGA, can be helpful in evaluating in quantitative terms the proportion of the total plasma membrane material which is being purified or examined by any given procedure.
Labeling of plasma membranes with iodinated hormones, lectins, and toxins cannot be used to elucidate membrane chemical structure because the complexes are dissociated by membrane denaturants such as sodium dodecyl sulfate. For these purposes, other more useful methods have recently been developed for labeling exposed chemical structures at the cell surface. For example, cell surface tyrosyl residues can be labeled with lactoperoxidase and radioactive iodine (67,68), amino groups with nonpenetrating irreversible agents (10, II), sialic acid residues with periodate and tritiated sodium borohydride (69), and galactose and galactosamine with galactose oxidase and tritiated sodium borohydride (70). These methods have proven very useful in the labeling and elucidation of exposed chemical structures at the cell surface (71, 72). Many of these procedures are not generally suitable as plasma membrane markers for fractionations because of their possible interference with function (direct chemical modification or secondary, as with sodium borohydride), relatively low specific radioactivity, uncertainty as to the localization of the reaction, irreversibility, or minimal control of the extent and nature of the reaction. It is anticipated that the use of specific membrane receptor structures as described in this report will help in the rational use of plasma membrane fractionation, isolation, and purification procedures which have for years lacked suitable specific assays.