Permeability of Canine Cardiac Sarcoplasmic Reticulum Vesicles to K+, Na+, H', and C1-*

Cardiac muscle sarcoplasmic reticulum appears to contain channel-like structures that render the mem- brane permeable to small univalent ions. Canine heart microsomes fractionated according to buoyant density were examined by Millipore filtration, light scattering, and membrane potential measurements. Enzymatic analysis and measurement Of D-glUCOSe permeation and Na/Ca exchange systems indicated two membrane fractions suitable for the permeability studies, one en- riched in surface membranes with a buoyant density of 1.04-1.11 (10-25% sucrose) and one enriched in sarco- plasmic reticulum with a buoyant density of 1.13-1.15 (30-34% sucrose). Surface membrane vesicles imperme- able to [3Hlsucrose were largely impermeable to K', Na+, and C1-, while sarcoplasmic reticulum vesicles impermeable to [3~sucrose were readily permeable to K+, Na+, H+, and C1-. Sarcoplasmic reticulum vesicles were essentially impermeable to Ca2+, M&+, choline+, gluconate-, 1,4-piperazinediethanesulfonic acid (Pipes-), and D-glUC0Se. These results suggest that cardiac muscle sarcoplasmic reticulum contains structures that facilitate the movement of small univalent ions. A possible function of these putative ion-conducting structures may be to allow rapid ion fluxes to counter electrogenic Ca2+ fluxes across sarcoplasmic reticulum during cardiac muscle

Sarcoplasmic reticulum is a highly specialized intracellular membrane system that plays the essential role of releasing and reabsorbing Ca2' during cardiac muscle contraction and relaxation (I). Few clues are available as to how a muscle action potential results in the release of Ca2+ from sarcoplasmic reticulum (2-4). Studies with membrane vesicle fractions isolated from canine heart muscle have, however, indicated that cardiac muscle sarcoplasmic reticulum, like skeletal muscle sarcoplasmic reticulum, accumulates Ca2' at the expense of ATP (5-13). The slower relaxation rate of cardiac muscle, compared with skeletal muscle, is reflected by relatively lower rates of ATP hydrolysis and Ca2' transport. In skeletal muscle sarcoplasmic reticulum, appreciable Ca2+ fluxes are compensated by counter movement of H+, K+, and Na' via H'-and K+,Na+-permeable structures (14). It is of interest that liver endoplasmic reticulum lacks highly active ion-releasing or sequestering transport systems, coincident with the absence of highly conducting structures for H' , K', and Na' (15). The reticulum structures of skeletal muscle and liver are permeable to C1- (15)(16)(17). In this report, we present * This research was supported by United States Public Health Service Grant AM 18687. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. evidence that cardiac muscle sarcoplasmic reticulum, like skeletal muscle sarcoplasmic reticulum, is permeable to K' , Na' , H' , and Cl-, suggesting the presence of univalent ionconducting channels.
Isolation ofMembranes-Microsomal membrane fractions derived from canine ventricular tissue were prepared as follows. 50 g of muscle were minced and homogenized in 375 ml of 0.3 M sucrose and 20 mM 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid, pH 7.7, at 4 "C for 60 s in a Waring blender. The homogenate was centrifuged for 12 min at 6,000 rpm (5,800 X g) in a GSA rotor in a Sorvall RC-2 centrifuge. A crude microsomal fraction was obtained from the supernatant by centrifugation for 90 min at 33,000 rpm (90, OOO X g) in a Beckman type 35 rotor. The pellet was resuspended in 0.3 M sucrose containing 25 m~ KC1 and 2.5 m~ Hepes, pH 7.2, and placed on a discontinuous sucrose gradient consisting of 10 ml of 25% (w/w), 10 ml of 30%, 10 ml of 3496, and 2.5 ml of 42% sucrose. Sucrose gradient solutions contained 25 m~ KC1,l m~ MgC12, and 2.5 mM Hepes,' pH 7.2. After centrifugation for 16 h at 23,000 rpm (95,000 X g) in a Beckman SW 27 rotor, membranes present at the 0.3 ~/ 2 5 % (Fraction I), 25-30% (Fraction 2), 30-34% (Fraction 3), and 3442% (Fraction 4) sucrose interfaces were collected, diluted with 1.5 volumes of 0.6 M KCl, and sedimented by centrifugation for 90 min at 33,000 rpm in a Beckman 35 rotor. The membranes of Fractions 1-3 were resuspended in 0.3 M sucrose and 1 m~ Hepes, pH 7.1. In Fraction 4, only the upper less brownish part of the pellet was recovered. The fractions were quickly frozen and stored at -65 "C before use.
Biochemical Assays-Protein was determined by the procedure of Lowry et al. (18) using bovine serum albumin as a standard.
Isotope Flux and Membrane Potential Measurements-Apparent 86Rb', and 36Cl-were determined at 22 "C by Millipore filtration as isotope spaces and efflux rates of microsomes to [3H]sucrose, "Na+, previously described (15). Membrane potentials were generated by gradients of permeant ions between the intravesicular cavity and the medium into which the vesicles were diluted. Membrane potentials

7704
This is an Open Access article under the CC BY license.
(negative inside) were detected by the use of the fluorescent dye 3,3'dipentyl-2,2'-oxacarbocyanine iodide as described by McKinley and Meissner (16). The sign of the membrane potential is reported according to standard convention, i.e. reference (ground) is extravesicular. Fluorescence assays were carried out at 15 'C under stirring in a Farrand model 801 fluorometer. Excitation was at 470 nm and emission was recorded at 495 nm. Vesicle concentrations (approximately 15 pg of protein/ml) were used which produced negligible perturbation of the fluorescence emission during dilution with incubation medium. Light-scattering Measurements-Osmotically induced volume changes in microsomes were detected by monitoring the changes in light-scattering intensity at 400 n m at right angle to the incoming beam using a Farrand model 801 fluorometer. Microsomes (0.1-0.2 mg of protein/ml) were equilibrated at 21 "C for about 30 min in 3 lll~ sucrose and 10 mM K-Pipes (1,4-piperazinediethanesulfonic acid), pH 7. Osmotic volume changes were induced at 21 "C by adding to the microsomal suspension, under rapid stirring, %o of a volume containing 10 lll~ K-Pipes, pH 7, and 800 mosm of the test compound.

RESULTS
Properties of Cardiac Microsomal Fractions-Microsomes were separated into four fractions of differing buoyant density (Table I) (23,24). The levels of these two activities indicated that the sarcoplasmic reticulum content of Fractions 2 and 3 was 2to 3-fold higher than in Fractions 1 and 4.
Estimation of surface membrane content of the four gradient fractions was complicated by the lack of agreement in the distribution of enzymatic activities characteristic of this membrane. All four membrane fractions, with highest levels being present in Fractions 2 and 3, possessed Mg2+-or Ca2+activated ("basic") ATPase activity (Table I), an enzymatic activity previously shown to be associated with canine cardiac muscle surface membrane structures with the exception of the nexus (19). On the other hand, Na/Ca exchange (cf Fig. 4) and D-glUCOSe permeation (Fig. 5) rates suggested a higher surface membrane vesicle content in Fraction 1 than in Fraction 3. Surface membrane-associated enzymes with varying distribution have also been observed in liver (25) or skeletal muscle (26,27). Fractions 1-3 contained small amounts of inner mitochondrial membranes, as indicated by the succi- nate-cytochrome c reductase activities. Sarcoplasmic reticulum and surface membranes of Fraction 4 were appreciably contaminated with inner mitochondrial membranes. Thin section electron microscopy revealed that Fractions 1 and 3 consisted mainly of membranous, vesicle-like structures with diameters ranging from approximately 750-3000 d; (Fig.   1). "Open" membranous structures were, however, seen in Fraction 1. In Fraction 3, most of the membranes appeared to be present in the form of enclosed, "sealed" compartments.
The above data suggested to us that Fraction 3 was best suited for studying the permeability properties of cardiac sarcoplasmic reticulum. K+, Na+, and C1-permeability of the surface membrane-enriched vesicle Fraction 1 was studied for comparison. . Permeability to =Rb+, "Nu+, and 36CZ--The ability of the vesicles to trap small ions and solutes was assessed by Millipore filtration.
[3H]Sucrose and [3H]choline were used to determine the internal volume of intact vesicles. They are considered impermeant and are retained by the vesicles to a similar extent (see Table 11).
[3H]Sucrose (see Fig. 3) and [3H] choline equilibrate across the vesicle membranes during prolonged incubation. Using the [3H]sucrose or [3H]choline spaces as a measure of total internal volume of intact vesicles, the proportion of vesicles permeable to other ions or solutes can be determined. Vesicles are considered to be permeable to a particular ion if an isotope of that ion can be released within seconds. These vesicles are devoid of radioactivity within 20-30 s, i.e. the first time point in the Millipore filtration experiments. Fig. 2 shows that for vesicles of Fraction 3 after preincubation for 6 h at 22 "C, the apparent isotope spaces for *'Rb+, the [3H]sucrose space. Comparison of the slopes on the semilogarithmic plots indicated that "Na+ (t1/2 = 2.5 min) passed through the ion-impermeable vesicles somewhat faster than and the three ion spaces was independent of the time of preincubation with the radioisotopes (Fig. 3). Preincubation for 6-12 h or longer was usually required to obtain maximal radioisotope spaces. Preincubation for 24 h resulted in uniformly lower isotope spaces, suggesting a general breakdown of the permeability barrier.
The apparent [3H]sucrose, [3H]choline+, "Na+, =Rb+, and C1-spaces and efflux rates of Fractions 1 and 3 are summarized in Table 11. In Fraction 1, "Na+, *'Rb+, and 36Cl-spaces approached or exceeded the [3H]sucrose and [3H]cholinef spaces. In contrast, 22Na+, %Rb+, and 36Cl-spaces accounted for a quarter or less ofthe [3H]sucrose and [3H]choline+ spaces of Fraction 3. Our interpretation of the different isotope spaces is that 22Na+, 86Rb+, and 36Cl-were able to efflux within 20-30 s from a majority of Fraction 3 vesicles because of the presence of mechanisms that facilitated their rapid release. For the residual vesicles which trapped these ions, efflux was on the time scale of minutes, which we believe is characteristic of nonmediated permeation. Surface Membrane Vesicle Content of Fractions 1 and 3-The different permeability behavior of the membrane vesicles in Fractions 1 and 3 to %Rb+, "Na+, and 36Cl-would be consistent with what is known about their permeability in skeletal muscle (16,17,28) if Fraction 1 was mainly surface membranes and Fraction 3 largely sarcoplasmic reticulum.
We tested for the presence of surface membrane vesicles by measuring the ability of the two fractions to exchange Na' or Ca2+ for 45Ca2+ and "Na+ (29), as well as to preferentially take up D-glucose over L-glucose (30). The effect of external Ca2+ on 22Na+ efflux was measured by diluting "Na+-filled vesicles into KC1 media containing either 0.25 m Ca2+ or the Ca2+  , "Rb', 22Na', and "Cl-spaces and efflux rates were determined as described in Figs. 2 and 3. Apparent isotope spaces give the maximal spaces determined between 6 and 12 h (cf Fig. 3) and indicate vesicle spaces that both form a permeability bamer and are accessible to the isotopes. Isotope spaces may have been underestimated since slowly penetrating solutes such as ["HI sucrose may not have fully equilibrated before the membrane barrier deteriorated. t 1 / 2 is the time in which the apparent isotope space, extrapolated to time of vesicle dilution (cf Fig. 2), decreased to onehalf of its initial value. D-["C]Glucose spaces were determined as described in Fig. 5. Data are the average of three determinations. S. E. = +20%.

Apparent isotope space
Efflux rate (fl,*) chelator EGTA at a concentration of 1 m~. Increased initial "Na' efflux rates on dilution of the vesicles into the Ca2'containing medium (Fig. 4) suggested that both vesicle fractions were capable of Ca/Na exchange. Vesicles of both fractions also appeared to be capable of Na/Na exchange since an enhanced "Na' efflux rate was noted when vesicles were diluted into a medium containing 100 m~ NaCl instead of 100 m~ KC1. Both Ca2+-and Na'-mediated "' Na' efflux could be blocked by prior addition of 1 mM La3+, a potent blocker of the cardiac surface membrane Ca/Na exchange system (29). Thus, the **Na+-impermeable vesicle populations in Fractions 1 and 3 were capable of Ca/Na exchange. However, as shown above (Table II), an important difference was that in Fraction 1 more than two-thirds of the vesicles were impermeable to "Na', whereas in Fraction 3, "Na'-impermeable vesicles made up less than a quarter of the total vesicle population.

Radioisotope Fraction Fraction Fraction Fraction
In the reverse experiment, we studied the effect of extravesicular Na' on 45Ca2+ efflux from vesicles. Vesicles were equilibrated at 22 "C in a medium containing 1 mM ' "Ca", 1 mM M2', 200 m~ K gluconate, and 10 mM K-Pipes, pH 7 .
Vesicles were then diluted 250-fold into a 200 m~ Na or K gluconate medium containing 10 m~ K-Pipes, pH 7 , 2 mM M $ ' , and 1 mM EGTA to maintain external Ca" at a concentration of less than 10" M. Initial 45Ca2+ efflux from vesicles of Fraction 1 was appreciably greater in Na' medium than K' medium.  (Fig. 5). In Fraction 3, D-and L-glucose uptake rates differed by a factor of 1.25 or less, suggesting that most of the D-glucose was nonspecifically taken up. At longer time intervals, similar D-and L-glucose isotope spaces were found. Both were in reasonable agreement with the ["Hlsucrose space (2.7 and 3.0 pl/mg of protein for Fractions 1 and 3, respectively, Table 11). Together, the "Na' and 45Ca2+ efflux and glucose uptake measurements were in  Fig. 2. Vesicles were then diluted and processed as described in Fig. 2. Apparent isotope spaces obtained by back extrapolation to the time of vesicle dilution (cf Fig. 2) indicate vesicle spaces that both form a permeability barrier and are accessible to the isotopes at the indicated time. support of the notion that Fraction 1 was enriched in vesicles derived from the surface membrane of canine heart. Fraction 3 also contained surface membrane vesicles but at a lower proportion than Fraction 1.
Light-scattering Measurements-The permeability of  vesicle size was similar using D-glucose, K gluconate, or choline C1 (Fig. 6). In all three cases, light-scattering signals reached a maximum within 2-3 s, the experimental limit of detection. This rapid initial increase reflected decreased vesicle volumes due to outflow of water (17). The signal then slowly returned to the one observed for control vesicles diluted with %o volume of the initial vesicle medium. The gradual return of the lightscattering signals and thus vesicle volumes to control values was due to inflow of the solutes followed by water (17). Reestablishment of equilibrium was similar for D-glucose, K gluconate, and choline C1 (tlIz -40-50 s). Since net movement of a salt requires that both ions move across the membrane, the slower ion will be rate limiting. The addition of the K' ionophore valinomycin did not affect the rate of re-establishing equilibrium with K gluconate (Table 111), indicating that gluconate was the rate-limiting species.
impermeable to C1-since the K+ ionophore valinomycin did not alter the signal (Table 111). K' , Na+, monoethanolamine+, and dimethylaminoethanol' pass rapidly through the K,Na channel of skeletal sarcoplasmic reticulum vesicles, whereas the somewhat larger cations choline and Tris are impermeable (16). Table I11 shows that the cardiac vesicles possess an ion pathway comparable to the one in skeletal muscle sarcoplasmic reticulum. Tris-C1 and choline C1 elicited nearly maximal light-scattering signals, indicating that most of the vesicles were relatively impermeant to these two cations. NaC1, monoethanolamine C1, and dimethylaminoethanol C1 behaved like KC1 in that these three ion pairs elicited an intermediate signal with a relatively long lifetime, suggesting that a large portion of the vesicles was readily permeable (tip < 2-3 s) to Na+, monoethanolamine+, dimethylaminoethanol+, and C1-. K-Pipes in the presence of valinomycin and sucrose elicited maximal light-scattering signals with a long lifetime (t1/2 > 200 s), suggesting that Pipes and sucrose permeated slowly across all of the vesicle membranes (Table 111). Light-scattering data of Table 111 are in good general agreement with radioisotope space and efflux measurements which had also suggested that Fraction 3 contains two vesicle populations with differing permeability to K' , Na+, and C1-.
Membrane Potential Measurements-Membrane polarization measurements are another useful means of probing the permeability of vesicles to small cations such as K' , Na+, or H' (16,31). The permeability of Fraction 3 to K+ was determined in the presence of the impermeable anions gluconate and Pipes (cf Table 111) by measuring the potential that formed during dilution of K'-loaded vesicles into media containing low defined K' concentrations. The developed K' diffusion potentials were visualized with the use of the fluorescent dye 3,3'-dipentyl-Z,2"0xacarbocyanine. A characteristic property of this dye is the decrease in fluorescence emission when vesicles become negatively charged inside. The magnitude of fluorescence decrease is roughly proportional to the fraction of vesicles that is polarized (16).
K'-loaded vesicles of Fraction 3 elicited no significant change in fluorescence emission when diluted into K' or Na' medium ( Fig. 7), indicating that no membrane potential (negative inside) was formed. Dilution of these vesicles into Tris'  technique The per cent increase and the lifetime of the light-scattering intensity signals were determined as described in Fig. 6. Light-scattering increases were obtained by extrapolating back to the time of vesicle addition (as indicated by the broken lines in Fig. 6) and by taking into consideration that the light-scattering intensity decreased by a factor of 0.9 due to solute addition. tllP was the time in which the extrapolated light-scattering signal returned to one-half of the value obtained with the buffer control. Data are the average of three determinations. S. E. = 20% or less. Glycerol elicited a suboptimal signal with a lifetime of 2.5 s or less, suggesting that glycerol equilibrated rapidly across the vesicle membranes (Fig. 6). A different light-scattering signal was recorded when the osmolality of the vesicle medium was increased by the addition of KC1 (Fig. 6). In this instance, a signal of intermediate size, but of long lifetime, was observed, indicating the presence of two vesicle populations in Fraction 3 with differing permeability. One appeared to be permeable to KCl, with a transient volume change lasting less than 2-3 s. The remaining vesicles were impermeable to KC1, as indicated by the long lifetime of the signal. These vesicles were  valinomycin (val). Fluorescence decreases were roughly proportional both to the logarithm of the K' gradient that formed across the vesicle membranes as well as the fraction of vesicles that was actually able to maintain a membrane potential for 1-2 s, the experiment h i t of detection (16). The K' of the vesicle medium served to establish an initial 60-fold K' gradient throughout the experiments. medium caused a rapid decrease in fluorescence followed by a slow return to the control value. Addition of the K' ionophore valinomycin to Tris' or Na' dilution media resulted in only small increases in the magnitude of the dye signals. The gradual return of the signals and, therefore, the breakdown of the membrane potentials were likely due to the slow inward movement of " I % '

Increase
(see Table 111). In agreement with this interpretation, a rapid breakdown of the potential was observed when the vesicles were rendered permeable to T r i s ' by the addition of the ionophore X537A (not shown).
Fluorescence measurements suggest that cardiac microsomal vesicles can form a K'-diffusion potential, negative inside, when diluted from a K' medium to a Trk' medium containing a low K' concentration. This, in turn, indicated that the vesicles were more permeable to K' than Tris'.
Absence of a fluorescence signal in Na' medium suggested the presence of vesicles which did not form a K'-dfision potential because of the rapid exchange of K' for Na' within 1-2 s, the experimental limit of detection. A majority of vesicles of Fraction 3 appeared, therefore, to be permeable to K' and Na+, suggesting a K',Na'-permeable structure comparable to the one identified in skeletal muscle sarcoplasmic reticulum (16). In support of a similar, or identical, channel structure was that the organic cation monoethanolamine behaved like a highly permeant cation in skeletal (16) and cardiac (not shown) sarcoplasmic reticulum.
Increases in fluorescence signal seen in the presence of valinomycin are doe to the presence of a K+,Na+-impermeable vesicle fraction. In the presence of the ionophore, these vesicles were rendered selectively permeable to K' and could, therefore, form a K'-diffusion potential. The small increase in fluorescence signals suggested that greater than 85% of the canine heart membrane vesicles in Fraction 3 (as compared to 70% in rabbit skeletal muscle, Ref. 16) were highly permeable to K' . Membrane potential measurements are in reasonable agreement with the radioisotope experiments (cf Table 11) considering that the two measurements likely indicate the surface area and the intracellular space of heterogeneously sized vesicle populations, respectively. Permeability of Fraction 3 to Ca2+ and Mg' was probed by transferring K+-filed vesicles to isoosmolal Ca-Pipes or Mg-Pipes media. Fluorescence decreases and return rates were similar to those seen in Tris-Pipes medium (not shown). In control experiments, we observed a rapid breakdown of the membrane potential in Ca-Pipes medium on addition of the Ca2' ionophore A23187 (1 pg/ml). Thus, membrane potential measurements indicate that Ca2+ and M P , like Tris', slowly pass across the membranes of the K'-permeable vesicles.
A characteristic property of skeletal muscle sarcoplasmic reticulum vesicles is that the fluorescence signals return faster to the base-line when vesicles actively transport Ca" (14). The faster decay indicates an increase in the rate of K+ release during Ca2' transport. In the present study, we initiated Ca2+ uptake at 22 "C by transferring K'-fiied microsomes of Fraction 3 to Tris' media containing 1 m~ Mg2+, 5 ~L M Ca", and 0.5 mM ATP. In a control, ATP was omitted from the dilution medium. Addition of ATP shortened the half-lifetime of the fluorescence signal from 9 to about 3 s (not shown). The effect of ATP on shortening the lifetime of the membrane potential was not eliminated in the presence of 1 p~ V04-, a potent inhibitor of the surface membrane Ca2+ pump (32). These results support the suggestion that the K'-permeable vesicle population of Fraction 3 originates from sarcoplasmic reticu-lUm.
Proton Permeability-Proton permeability of Fraction 3 was investigated as previously described for skeletal muscle and liver microsomal fractions (31). Vesicles present in Tris-Pipes buffer at pH 6.2 were transferred to Tris-Pipes buffer at pH 7.8. Fig. 8 shows that the H' gradient thus established generated a transient membrane potential. Addition of the H+-carrier carbonyl cyanide p-trifluoromethoxyphenylhydrazone did not sigmfkantly affect the overall magnitude of the fluorescent signal, suggesting the presence of membranes intrinsically permeable to H' (or OH-).
Additional evidence for H' permeability was provided by investigating the effect of opposing H' and K+ gradients on fluorescence emission of di0-Cs-(3). In Experiments 1-4 of Table JV, we determined the ability of an opposing H' gradient to nullify membrane potentials generated by a K+ gradient. Dilution of K+-containing vesicles into Tris' media of equal pH yielded fluorescence decreases of 18% at pH 6.2 and 12% at pH 7.8. Dilution from K' medium at pH 6.2 to Trisf medium at pH 7.8 resulted in an increased fluorescence signal (25%). In the latter case, K' and H' gradients were of equal

diO-G-(3)
In Experiments 1-4, microsomes of Fraction 3 were incubated in 210 mosm K-Pipes and 210 mosm Tris-Pipes at pH 6.2 or 7.8 before diluted 100-fold into isoosmolal Tris-Pipes media at pH 6.2 or 7.8. The external K+ in the dilution media was derived from the vesicles such that the initial K' gradient was maintained constant (100-fold) in Experiments 1-4. Part of the K+ in the vesicles was replaced by Tris' to obtain optimal conditions of K'-H+ exchange (31). In Experiments 5-7, microsomes containing 420 m o m Tris-Pipes at pH 6.2 were diluted 250-fold into isoosmolal Pipes media at pH 8 containing the indicated cation. The pH of the dilution media was not sign& cantly changed by the addition of the sample. Fluorescence decreases were obtained by extrapolating back to the time of vesicle addition (cb Fig. 8). direction and similar magnitude. No significant K' -H' exchange could, therefore, occur in vesicles permeable to both K' and H' , and an optimal membrane potential was formed (31). In contrast, an opposing pH gradient (K+, pH 7.8 -+ Tris', pH 6.2) essentially eliminated the fluorescence signal, indicating that vesicles were capable of rapid K'-H+ exchange and of lowering their internal K' concentration to that of the surrounding medium within 1-2 s, the experimental limit of detection.

Expeti-Vesicle medium
Experiments 5-7 of Table IV describe the reverse experiment, namely, the effect of opposing K+ and Na' gradients on a membrane potential generated by a H+ gradient. Dilution of vesicles from Tris' medium at pH 6.2 to Tris' medium at pH 7.8 generated a H'-induced membrane potential which decreased the fluorescence emission of diO-C5-(3) by 14%. Replacement of external Tris' by K' or Na' nullified the membrane potential, indicating that K+ and Na' could rapidly enter all of the H'-permeable vesicles. Thus, together the data of Table IV suggest that the majority of the vesicles in Fraction 3 were permeable to K' , Na', and H+.

DISCUSSION
We have used three techniques to measure the membrane permeability of a cardiac muscle microsomal fraction enriched in sarcoplasmic reticulum. The Millipore filtration technique using radioactively labeled compounds is ideal for determining vesicle spaces and exchange or efflux rates of relatively impermeant solutes. Light-scattering measurements provide similar information and have the additional advantage of allowing us to measure ion pair and solute fluxes on a time scale of 2-3 s, as compared to 20-30 s using Millipore filtration. Finally, membrane polarization measurements prove useful in evaluating the free cation permeability of isolated membrane vesicles. If should be noted, however, that different vesicle parameters were measured by the three techniques. Radioisotope and membrane polarization measurements reflect internal vesicle spaces and surface areas of the polarized vesicles, respectively.
As summarized below, radioisotope flux measurements indicated that Fraction 1 is predominantly made up of surface membrane vesicles, whereas Fraction 3 consisted of about 80% sarcoplasmic reticulum vesicles and 20% surface membrane vesicles. Fractions 1 and 3 differed appreciably in their permeability to =Rb+, 22Na+, and 36Cl-. About 80% of [3H]sucroseand [3H]choline+-impermeable vesicles of Fraction 3 were readily permeable to these ions, while a similar portion of Fraction 1 was impermeable to =Rb+, 22Na+, and 36Cl-. The ability of Na'-impermeable vesicles to exchange 22Na+ for Ca2+ or Na+ suggested that the Na+-impermeable vesicles in Fractions 1 (80%) and 3 (20%) were mainly derived from the surface membrane structures of cardiac muscle. In agreement with this interpretation was the enhanced ability of the Na+impermeable enriched vesicle Fraction 1 to differentiate between D-and L-glucose uptake.
A majority of skeletal muscle surface membrane vesicles are impermeable to C1- (28), whereas both skeletal muscle sarcoplasmic reticulum (16,17) and liver endoplasmic reticulum (15) membranes are permeable to C1-. The presence of a significant portion of C1-impermeable vesicles in Fraction 1 is in accord with our contention that this fraction is enriched in surface membrane vesicles. The higher "basic" ATPase activity of Fraction 3, an enzyme characteristic of the surface membrane (19), may be due to an uneven in vivo distribution resulting in vesicle fractions with different enzyme contents. A similar phenomenon has been observed in other tissues such as liver (25) and skeletal muscle (26,27). Na+-impermeable vesicles did not originate from inner mitochondrial mem-branes since Fractions 1 and 3 had low succinate cytochrome c reductase activities.
Recent studies have contrasted the ion permeability of skeletal muscle sarcoplasmic reticulum and liver endoplasmic reticulum vesicles (15,16). About two-thirds of skeletal muscle sarcoplasmic reticulum vesicles contain a K,Na channel, rendering them highly permeable to K+, Rb' , Na' , and small organic cations such as monoethanolamine'. The remaining one-third lack the K,Na channel and are, therefore, relatively impermeable to these cations. K',Na'-impermeable vesicles are thought to arise as a consequence of a limited number of channels in the in vivo skeletal muscle reticulum structure (about 50/pm2) (16). Both K+,Na+-permeable and -impermeable vesicles are permeable to protons (37) and chloride (16), suggesting a higher density of H' -and C1"conducting structures.
Liver microsomes resemble sarcoplasmic reticulum vesicles in that they are permeable to small univalent ions. They appear to lack, however, H+ and K,Na channels characteristic of skeletal muscle sarcoplasmic reticulum (15). Another important difference between these membranes is that liver microsomes are readily permeable to several small biologically relevant solutes and ions including D-and L-glucose, gluconate-, and choline+.
In this report, we have demonstrated that Fraction 3 contains cardiac muscle sarcoplasmic reticulum vesicles which possess permeability properties comparable to those of skeletal muscle sarcoplasmic reticulum but unlike those of liver endoplasmic reticulum. About 80% of the vesicles in Fraction 3 were highly permeable to K' . This same fraction was found to be highly permeable to C1-and other small monovalent cations including Na+, monoethanolamine', and H' . It also contained a highly active Ca2+ transport system characteristic of sarcoplasmic reticulum.
K' , NaC, and C1-equilibrated within 3 s, the experimental limit of detection of the light-scattering and membrane potential techniques. Using an average vesicle diameter of 0.15 pm (Fig. l ) , the permeability coefficient of the vesicles for K+, Na+, and C1-can be calculated to be greater than cm/s. The greater proportion of K+,Na'-permeable vesicles in cardiac (>85%) versus skeletal muscle (-70%) sarcoplasmic reticulum preparations indicates that heart reticulum has at least as many K,Na channels as skeletal reticulum.
The physiological function of the H' -and K+,Na+-permeable structures in the sarcoplasmic reticulum membrane of cardiac and skeletal muscle is unclear at present. Miller (33) made the interesting observation that the K,Na channel displays single channel conductance fluctuations between an "open" and "closed" state when skeletal muscle sarcoplasmic reticulum vesicles were fused with planar bilayers. Ca2+ transport and ion flux measurements have led us to propose that the H' and K,Na channels of skeletal muscle sarcoplasmic reticul-act as a complementary part of the Ca2+ transport system by allowing rapid H' and K' movement to counter Ca2+ fluxes during muscle relaxation (14). In addition, rapid H' and K' fluxes may allow cardiac and skeletal muscle sarcoplasmic reticulum to release Ca2+ during muscle contraction.