Heat shock proteins (HSPs) are essential for normal cellular stress responses. Absolute amounts of HSP72, HSP25, and αB-crystallin in rat extensor digitorum longus (EDL) and soleus (SOL) muscle were ascertained by quantitative Western blotting to better understand their respective capabilities and limitations. HSP72 content of EDL and SOL muscle was only ∼1.1 and 4.6 μmol/kg wet wt, respectively, and HSP25 content approximately twofold greater (∼3.4 and ∼8.9 μmol/kg, respectively). αB-crystallin content of EDL muscle was ∼4.9 μmol/kg but in SOL muscle was ∼30-fold higher (∼140 μmol/kg). To examine fiber heterogeneity, HSP content was also assessed in individual fiber segments; every EDL type II fiber had less of each HSP than any SOL type I fiber, whereas the two SOL type II fibers examined were indistinguishable from the EDL type II fibers. Sarcolemma removal (fiber skinning) demonstrated that 10–20% of HSP25 and αB-crystallin was sarcolemma-associated in SOL fibers. HSP diffusibility was assessed from the extent and rate of diffusion out of skinned fiber segments. In unstressed SOL fibers, 70–90% of each HSP was readily diffusible, whereas ∼95% remained tightly bound in fibers from SOL muscles heated to 45°C. Membrane disruption with Triton X-100 allowed dispersion of HSP72 and sarco(endo)plasmic reticulum Ca2+-ATPase pumps but did not alter binding of HSP25 or αB-crystallin. The amount of HSP72 in unstressed EDL muscle is much less than the number of its putative binding sites, whereas SOL type I fibers contain large amounts of αB-crystallin, suggesting its importance in normal cellular function without upregulation.
- heat shock protein
- inducible heat shock proteins
- stress response
- skinned fiber
heat shock proteins (HSPs) enable skeletal muscle to cope with physiological stresses, such as glycogen depletion (6), Ca2+ increases (42), heat (39), and exercise (21). HSPs play essential roles in maintaining cellular homeostasis by acting as molecular chaperones and as stress sensors and by conferring direct cytoprotection. They show a high degree of homology across species but are a diverse family of molecules, including both constitutively expressed and stress-inducible members. Three HSPs known to have significant roles in cellular protection and adaptation in skeletal muscle are HSP72 (the inducible HSP70 isoform), HSP25 (murine isoform, homologous to human HSP27), and αB-crystallin. These HSPs bind to and stabilize damaged proteins and thus protect against protein degradation (14). Specifically, HSP72 has been shown to stabilize both the structure and function of the sarcoplasmic reticulum Ca2+ pump (SERCA1a and SERCA2a) in skeletal and cardiac muscle subjected to heat stress (8, 43). HSP25 and αB-crystallin on the other hand are thought to bind predominantly to cytoskeletal/myofibrillar proteins, including desmin, titin, α-actinin, actin, and myosin, protecting them against denaturation following potentially stressful insults (1, 11, 16, 19, 21, 22, 31, 36).
Data from various studies suggest that HSP72, HSP25, and αB-crystallin are each expressed at relatively higher levels in type I muscle fibers than in type II muscle fibers (1, 11, 12, 18, 29). However, the actual amounts of each HSP present in skeletal muscle are unknown; even the relative abundance of the three HSPs is not known. Skeletal muscle fibers likely contain numerous potential binding sites for the various HSPs, and it is currently unclear whether the expression levels of any of the HSPs in unstressed muscle are sufficient to protect all or only a fraction of the relevant target molecules. It is well known that the levels of expression of the HSPs increase to various degrees with stress (5, 21, 29, 31, 33), but to quantitatively interpret the relevance of such increases and their implications for cellular function it is important to know how much of the given HSP is initially present and how much the increase actually represents.
Most previous studies have suggested that, in unstressed muscle fibers, the majority of each of the three HSPs is located in the cytosol (2, 11, 16, 21, 32). This has been concluded largely from centrifugation of muscle tissue into “soluble” and “insoluble” fractions, with the cytosolic constituents thought to partition predominantly into the soluble fraction. However, there is some uncertainty as to whether the cytosolic and noncytosolic constituents are completely separated in this way, particularly since cytosolic components might become bound, and noncytosolic components unbound, during the homogenization and centrifugation procedures. Additionally, to permit truly quantitative interpretation of the data, no fraction at all should be discarded during the procedures, and the proportion of HSP present in a given fraction should be quantified by directly comparing the HSP content across all of the fractions in absolute terms, rather than by normalizing the HSP content of a fraction by its protein content, a measure that differs for each individual fraction.
In this study, we performed quantitative Western blotting on entire constituents of muscle samples to determine the absolute amounts of HSP25, HSP72, and αB-crystallin present in muscles composed primarily of either type I or type II fibers, and also that present in individual muscle fibers of known type. By peeling off the sarcolemma from a muscle fiber, it was also possible to determine the proportion of HSP readily diffusible in the cytoplasm and whether any of the HSP associated with the sarcolemma. It was hypothesized that 1) the absolute amounts of the various HSPs differ considerably in unstressed muscle, possibly in inverse relation to their relative increases following stress, and also differ between fiber types, 2) each HSP is present at much lower density than its potential binding sites, 3) the majority of each HSP is readily diffusible within unstressed fibers and hence able to reach and bind to numerous possible targets as needed, and 4) heat stress causes tight binding of all three HSP types, but differentially so to membranous and cytoskeletal targets.
MATERIALS AND METHODS
Materials and antibodies.
All chemicals were obtained from Sigma (St. Louis, MO) unless otherwise stated. Antibodies used were against HSP25 (1:2,000 rabbit polyclonal, SPA-801; Stressgen), HSP72 (1:500 mouse monoclonal, SMC100A; Stressmarq), αB-crystallin (1:1,000 mouse monoclonal, SPA-222; Stressgen), actin (1:300 rabbit polyclonal, A2066; Sigma), myosin heavy chain (MHC) I (1:200 mouse monoclonal, A4.840), MHCII (1:200 mouse monoclonal, A4.74), SERCA1 (1:1,000 mouse monoclonal, Ca F2–5D2; from Developmental Studies Hybridoma Bank), and SERCA2a (1:5,000, gift from Dr. F Wuytack, Leuven, Belgium). Purified proteins used were HSP25 (mouse recombinant, ADI-SPP-510; Stressgen), HSP72 (rat recombinant, ADI-SPP-758; Stressgen), and αB-crystallin (bovine native, ADI-SPP-226; Stressgen).
With approval of the La Trobe University Animal Ethics Committee, male Long-Evans hooded rats (∼6–10 mo old) or C57/BL10 mice (∼5–7 mo) were killed by overdose with inspired isoflurane (4% vol/vol). The extensor digitorum longus (EDL) and soleus (SOL) muscles were quickly excised. Bovine muscle, supplied through a butcher, was obtained from the upper and lower hind limb of an animal ∼24 h postdeath and carcass storage at 4°C. To verify that the level of αB-crystallin was not greatly altered by the postmortem storage, additional experiments were carried out comparing SOL muscle excised from a rat carcass that had been stored at 4°C for 24 h after death; there was no significant difference in the amount of αB-crystallin found in muscle excised immediately after death and muscle excised 24 h after death and 4°C storage (ratio: 1.1 ± 0.2, n = 3).
Muscle homogenates were prepared (1:10 wt/vol) in a physiological-like solution with free Ca2+ buffered at very low levels with EGTA: buffer A (in mM): 126 K+, 36 Na+, 1 free Mg2+ (10.3 total Mg), 90 HEPES, 8 ATP, 8 creatine phosphate, and 50 EGTA (pH 7.1), with protease inhibitor cocktail (COMplete, Roche Diagnostics, Sydney, NSW, Australia). Muscle homogenates were diluted further 1:20 (vol/vol) using the same solution and then 2:1 (vol/vol) with SDS loading buffer (0.125 M Tris·HCl, 10% glycerol, 4% SDS, 4 M urea, 10% mercaptoethanol, and 0.001% bromphenol blue, pH 6.8). Samples were stored in −20°C until analyzed by Western blotting.
Single fiber experiments.
For single fiber experiments, rat muscles were pinned immediately after excision at resting length under paraffin oil and then either slowly cooled and maintained at ∼10°C on an ice pack (unstressed, control muscle) or first heated for 30 min to 45°C and then slowly cooled and maintained at ∼10°C (heated muscle). Fibers from each heated muscle were collected over the same period as fibers from the unheated contralateral muscle (control) from the same rat. Single fibers were separated by dissection, and segments ∼3 mm in length were collected for Western blotting or first “skinned” by microdissection to remove the sarcolemma (i.e., surface membrane). The latter was achieved by pulling a small number of myofibrils away from the rest of the fiber, causing the sarcolemma to roll back along the fiber (27). Only one segment was examined from any given fiber to avoid sampling bias. A small piece of suture thread was tied around each fiber segment to allow ready transfer between solutions as required. To examine the diffusibility of αB-crystallin, HSP25, and HSP72, single mechanically skinned fiber segments (each ∼10–15 nl in volume) were immersed in 5 μl of buffer A in a microcentrifuge tube for a specified time (30 s, 2 min, 10 min, 30 min, 60 min, or 120 min). In a subset of experiments, fiber segments were immersed in buffer A with 1% Triton X-100 for 10 min. Most fibers were vortexed briefly (∼2 s) at least one time during the exposure time, in particular just before removing the fiber from the bathing solution. It was subsequently found that there was no detectable difference in results between fibers that were or were not vortexed. After the required time, the fiber was removed and placed in another microcentrifuge tube containing the same volume of buffer A, and 2.5 μl of 3× SDS loading buffer was then added to both tubes, thus obtaining a matched set with the fiber segment and corresponding wash solution in separate tubes [as previously described (24, 28)].
Western blotting for protein diffusibility and absolute quantification.
As described previously (20), protein samples were loaded and separated on 12% SDS polyacrylamide gels, or in some cases on Criterion Stainfree gels (26). Proteins were then transferred onto nitrocellulose membrane and blocked with blocking buffer (5% skim milk in Tris-buffered saline with Tween 20) for 2 h. Following blocking, primary antibodies were applied for 2 h [room temperature (RT)] and overnight (4°C) on a rocker. Membranes were cut into two sections at a molecular mass position of ∼40 kDa, and the sections were probed as required for αB-crystallin, HSP25, HSP72, MHCI, MHCII, SERCA1, SERCA2, and actin, diluted in 1% BSA PBS with 0.025% Tween (23). Following washes, secondary antibodies were applied for 1 h (RT), either goat anti-mouse horseradish peroxidase (HRP)-conjugated (Pierce) or goat anti-rabbit HRP-conjugated (Pierce) (both diluted 1: 20,000 in blocking buffer). Following transfer, BioSafe Coomassie Stain (Bio-Rad) was used to stain the SDS-polyacrylamide gel for the detection of MHC, which was used as an indicator for the presence of myofibrils and also of the relative amounts of sample loaded. Images of the membrane were collected following exposure to chemiluminscence substrate (Thermo Scientific SuperSignal West Femto; Pierce) using a charge-coupled device camera attached to ChemiDoc XRS (Bio-Rad), and Quantity One software was used for detection as well as densitometry. The relative positions of molecular mass markers were captured under white light prior to chemiluminescent imaging without moving the membrane. For the examination of protein diffusibility, the individual fiber segments and their corresponding wash solution were run side by side and analyzed for the relevant proteins by Western blotting (24, 28).
For analyses of the absolute amount of specific proteins, whole muscle homogenates (2.5–50 μg total muscle mass) were loaded onto a gel together with known amounts of purified recombinant proteins (mouse HSP25, rat HSP72, and bovine αB-crystallin), the latter allowing a calibration curve to be generated (see Fig. 1). When quantifying absolute amounts of αB-crystallin, HSP25, or HSP72, the density of the relevant band was converted to an equivalent protein amount, according to the calibration curve derived from the pure protein samples run on the same Western blot. This amount was expressed relative to the mass of muscle loaded in that lane, and the average was calculated for all repeated samples run on the same gel. The validity of this Western blot quantification procedure was verified as described in results (see Fig. 2).
Data are expressed as means ± SE, with the number of samples studied denoted as n, being muscles or individual fibers as indicated. Statistical significance was examined using Student's t-tests (paired or unpaired as appropriate), or, in cases where data were not normally distributed (e.g., data in Figs. 3 and 9), significance was tested using the nonparametric Mann Whitney unpaired rank test; a probability value (P) <0.05 was deemed as significant. All statistical analyses and data fits were performed using GraphPad Prism version 4.
Measurement of absolute amounts of HSPs in rat muscle.
Absolute amounts of αB-crystallin, HSP25, and HSP72 present in rat EDL and SOL muscle were each assessed using the methodology shown in Fig. 1. Very small amounts of unfractionated muscle homogenate were run on a given gel together with a range of amounts of the relevant purified HSP. A calibration curve was generated for each gel by plotting the density of the Western blot band for each purified HSP sample vs. the amount of HSP loaded (e.g., Fig. 1B), and this was used to assess the amount of HSP present in homogenate samples run on the same gel. The entire muscle homogenate was run without any spinning or removal of cell debris to ensure that there was no omission of any muscle proteins. The relative amount loaded was also verified for each muscle type by reprobing the membrane for actin and also by the MHC signal after Coomassie staining of the SDS gel posttransfer. Such analysis also suggested that rat SOL muscle likely has ∼10% less MHC and actin per unit muscle wet mass than does rat EDL muscle (mean: 90 ± 6%, n = 8; 90 ± 12%, n = 3, respectively), as expected given the greater mitochondrial content per unit volume in SOL muscle.
This HSP quantification method is only valid if the HSP present in the muscle homogenate is detected with the same efficacy as the pure HSP run alone as the standard. This was tested by comparing band intensities in lanes containing just purified HSP, or just muscle homogenate, or purified HSP mixed together with another sample of the same homogenate (Fig. 2). Such data were obtained on three to five independent gels for each HSP. For all three HSPs, the amount of HSP gauged as present in the muscle homogenate with added HSP was virtually identical to the sum of that found when running the homogenate and pure HSP samples separately (ratio: 1.0 ± 0.1 for HSP72, n = 3; 1.0 ± 0.1 for HSP25, n = 5; and 1.1 ± 0.1 for αB-crystallin, n = 3), verifying the reliability of the quantification method.
Table 1 lists the absolute amounts of the three HSPs present in rat EDL and SOL muscles in the unstressed state, deemed as such because the animals were healthy and had not been subjected to any heat or exercise stress. Of note, in EDL muscle, there was only ∼1.1 μmol HSP72/kg wet muscle mass, whereas SOL muscle contained approximately fourfold more HSP72 (∼4.6 μmol/kg). In both muscle types, there was approximately two to three times more HSP25 than HSP72 (∼3.4 and 8.9 μmol/kg in EDL and SOL, respectively). Perhaps the most surprising finding was the very large absolute amount of αB-crystallin present in rat SOL muscle, ∼139 μmol/kg, with EDL muscle containing only ∼4.9 μmol/kg. Thus, in SOL muscle, the amount of αB-crystallin is ∼16 times greater than the amount of HSP25 and ∼30 times greater than the amount HSP72.
HSP measurements in murine and bovine muscle.
The pure proteins used to quantify HSP25 and αB-crystallin were murine and bovine proteins, respectively. To validate their use as standards for quantifying the HSP content of rat muscle, and additionally to enable comparison between different species, these same pure proteins were also used to assess the HSP content of murine and bovine muscle. The absolute amounts of HSP25 in mouse EDL and SOL muscles were determined, respectively, to be 3.5 ± 0.7 and 11.1 ± 8 μmol/kg muscle mass (n = 4 mice), which were similar to the amounts determined in the corresponding rat muscles (Table 1). The amount of αB-crystallin in bovine muscle was 132 ± 17 μmol/kg in the three muscle samples examined from the distal hind limb of one animal and 69 ± 17 μmol/kg (n = 3) for samples from the proximal hind limb. The density of MHCI was approximately two times greater in the distal hind limb than in the proximal hind limb (2.3 ± 0.2-fold, n = 3), indicative of a higher proportion of type I fibers in the distal muscles. These data suggest that the αB-crystallin content of bovine type I fibers is quantitatively similar to that in the predominantly type I SOL muscle of the rat.
Relative amounts of HSPs in individual fibers.
The above HSP measurements were made on muscle homogenates that consisted of a range of different fiber types. It was thus important to also consider whether individual SOL and EDL fibers showed the same disparity as the homogenate data because the total HSP in a given muscle might be determined predominantly by higher HSP levels in some subset of fibers in that muscle. To investigate this, single segments of individual muscle fibers were obtained by microdissection and analyzed in their entirety by the same Western blotting procedure used for the muscle homogenate measurements. Because it was not possible to accurately measure the weight of the individual fiber segments (each ∼3 mm long and likely ∼10 to 15 μg wet wt), the Coomassie MHC band density was taken as a measure of the relative sample mass and used to normalize the corresponding HSP band density for each fiber sample run on a given gel. This normalized HSP amount for each fiber was then expressed relative to the mean of that for all SOL type I fibers run on the same gel (mostly 4 to 6); in effect, this defined the HSP amount in an “average” SOL type I fiber as unity and indicated how the HSP amount in a given fiber compared with that level (see Fig. 3). SOL fiber segments were denoted as being either type I (slow-twitch) or type II (fast-twitch) based on the MHC type present (as assessed by Western blotting of the same membranes). As seen in Fig. 3B, the EDL fibers, which were all type II fibers, formed a relatively discrete population with much less of each HSP than present in the SOL type I fibers (P < 0.05 in all cases); in fact, there was no overlap in the amount of HSP present in the EDL type II fibers and SOL type I fibers. Rat SOL muscle of this rat strain typically is composed of ∼80% type I fibers and ∼20% type II fibers (40), and, of the 22 SOL fibers examined here, 20 were assessed as being type I and the other 2 as type II. In both of these SOL type II fibers, each of the three HSPs was present at lower levels than in the SOL type I fibers, levels indistinguishable from that present in the type II fibers from EDL muscle (Fig. 3B).
Proportion of HSPs associated with sarcolemma.
Immunostaining of muscle fibers has indicated that some HSPs appear to be localized at relatively high levels at or just beneath the sarcolemma (surface membrane) (11). To determine whether particular HSPs are actually directly associated with the sarcolemma and related structures, segments of individual SOL and EDL muscle fibers were skinned by microdissection to completely remove the sarcolemma, and the amount of the given HSP present in the skinned segments was compared with that in “intact” segments where the sarcolemma was still associated. In both EDL and SOL fibers, the amount of αB-crystallin found in the skinned segments was significantly lower than in the intact segments (see Table 2), indicating that some αB-crystallin had been closely associated in some way with the sarcolemma and removed by the skinning. This amount was only ∼10% of the total in SOL fibers, but proportionately much more (∼60%) in EDL fibers, consistent with a substantial amount of αB-crystallin being associated with the sarcolemma in both types of muscle fiber and the total present being larger in SOL fibers (Table 1 and Fig. 3B). Skinned segments from SOL fibers also contained ∼20% less HSP25 than comparable intact segments (Table 2). In EDL fibers, however, the difference in HSP25 amount (∼12%) was not statistically significant. It was nevertheless apparent that some HSP25 is associated with the sarcolemma in at least some EDL fibers, because in a few cases the sarcolemma excised from a segment of fiber was analyzed by Western blotting alongside the skinned segment, and, in one of the EDL fiber cases shown in Fig. 4, a substantial amount of HSP25 was found to be present with the sarcolemma. In the case of HSP72, the mean amount present in skinned fiber segments was not significantly different from that in intact segments for either EDL or SOL fibers (Table 2); no attempt was made to directly measure whether HSP72 was associated with the sarcolemma owing to the low absolute amounts involved and detection limitations of the Western blotting.
Diffusibility of HSPs in unstressed muscle fibers.
To gauge how much of a given HSP is freely diffusible in the cytoplasm in unstressed muscle fibers, segments of individual fibers from rat SOL and EDL muscles were skinned by microdissection under paraffin oil and then transferred for a set period into a comparatively large volume of an aqueous solution that broadly mimicked the normal cytosol (see materials and methods). Figure 5 displays representative Western blots of the proteins remaining within the fiber and those that diffused out into the wash solution; note that there was no detectable loss of either actin or myosin to the wash solution. Each of the fiber and wash samples was run in its entirety without any spinning or fractionation. For each HSP, the total of that found in the wash solution and fiber lanes together was not noticeably different from that seen when running either unskinned fiber segments or skinned segments taken straight from paraffin oil without any washing. (Note all such comparisons of HSP amounts between different fiber samples took into account the relative mass of the fiber sample, based on the associated MHC signal.) Figure 6 plots the extent and rate of diffusional loss of each of the HSPs in SOL and EDL fibers. Most of the HSP72 diffused out into the wash solution within 10 min, but ∼5–15% still remained within the fiber even after 2 h (Fig. 6A). With HSP25, most washed out of the fiber within several minutes, but there was a nondiffusing component of ∼24% in SOL fibers and ∼50% in EDL fibers (Fig. 6B). Similar to HSP25, most αB-crystallin was readily diffusible in SOL fibers, with a maximum washout of ∼89% attained within ∼2 min (Fig. 6C), whereas, in EDL fibers, the washout was only ∼30%. Interestingly, comparison of the rates of diffusional loss of the three HSP in SOL fibers (where values are probably more accurately determined than in EDL fibers because of the larger absolute amounts of HSP involved) showed that HSP72 diffused much more slowly out of fibers (time constant ∼2.2 min) than did the HSP25 and αB-crystallin (time constants ∼0.25 and 0.34 min, respectively) (see Fig. 6). This difference was not explained simply by the relative mass of the HSP proteins (see discussion).
The above diffusional data were obtained from fibers that had been skinned at various different times after removing the muscle from the rat (up to ∼150 min). To ascertain whether this time period in vitro had any influence on HSP diffusibility, the percentage HSP washout value found in each fiber washed for 10 min or more was plotted against the time delay before skinning the fiber (data not shown). Linear regression analysis showed no significant relationship in any case except for αB-crystallin washout in SOL fibers, which displayed somewhat reduced washout if the time before skinning the fiber exceeded 80 min. Consequently, for the latter case, the HSP washout data presented here were restricted to fibers examined <80 min after muscle removal.
To further investigate the binding properties of the HSPs in unstressed fibers, we examined the effects of supplementing the wash solution with 1% (vol/vol) of the detergent, Triton X-100. Membrane-bound proteins become diffusible in the presence of the detergent, whereas those bound to sarcomeric structural proteins evidently remain within the fiber (24). The membrane-bound protein examined was the sarcoplasmic reticulum (SR) calcium pump protein, SERCA, which is a known binding site for HSP72 and situated in the SR membranes around each myofibril throughout a muscle fiber. The SERCA1 isoform could be detected not only in SOL type II fibers but even in a proportion of type I SOL fibers, although at comparatively low levels. (In some experiments, membranes were probed only for SERCA1, but, in later experiments, they were probed for both SERCA1 and SERCA2.) As seen in the example in Fig. 7, left, after 10 min exposure to the standard wash solution without Triton X-100, all of the SERCA1 still remained within the fiber. Neither SERCA1 nor SERCA2 was ever seen to partition into the wash solution in any fiber in the absence of detergent. In contrast, when Triton X-100 was present, 80 ± 5% of SERCA1 was lost to the wash solution after 10 min in the three SOL fibers examined here with detectable levels of SERCA1, consistent with the expected effect of the detergent on membrane-bound proteins. This Triton X-100 treatment did not, however, detectably increase the extent of diffusional loss of any of the HSPs in those fibers nor in the three other SOL fibers examined; the percentage remaining after 10 min was 9 ± 3% (n = 6) for HSP72, 16 ± 4% (n = 6) for HSP25, and 15 ± 4% (n = 6) for αB-crystallin, values that do not differ significantly from the respective percentage remaining after the same wash time without detergent (HSP72, P = 0.6; HSP25, P = 0.8; and αB-crystallin, P = 0.4). Thus the great majority of all three HSPs are readily diffusible in the cytoplasm of SOL fibers from unstressed muscles, and the small proportion of each remaining within the fibers is not detectably altered by disrupting and dispersing at least some of the internal membranes.
Diffusibility of HSPs in fibers from heat-stressed muscle.
Finally, HSP amounts and diffusion were examined in fibers from heat-stressed muscles. Rat SOL and EDL muscles were heated in vitro to 45°C for 30 min, and the fibers were compared with those from the nonheated contralateral muscles. This heat treatment caused no acute change in the amounts of the HSPs; the amounts measured in homogenates from three heated SOL muscles expressed relative to that in the corresponding three contralateral muscles were 1.0 ± 0.1 for HSP25, 0.9 ± 0.1 for αB-crystallin, and 0.9 ± 0.1 for HSP72 (no significant change in any case).
Individual fibers from heated and nonheated muscles were skinned and bathed for 10 min in physiological intracellular solution as described earlier, and the fiber and wash solution samples were run in adjacent lanes, as seen in Fig. 8A. The HSPs in the fibers from the unheated control muscles were largely lost to the wash solution, consistent with the results in the previous section, whereas, in the fibers from the heated muscles, all three types of HSPs were found to remain almost entirely within the fiber (e.g., Fig. 8A, wash solution and fiber data on right), even though aldolase, a 40-kDa cytoplasmic protein, still readily diffused out of the fiber in every case examined. The mean data for the three HSPs are shown in Fig. 9. The HSPs all still remained in the fibers even if the wash time was increased to 30 min (data not shown). Very similar tight binding was also seen in all the fibers examined from heated EDL muscles (data not shown, only HSP25 and HSP72 were examined). Consistent with the homogenate data reported above, there was no significant difference in any case between the HSP amounts in the fibers obtained from the control and the heated SOL muscles; the relative amounts for the fibers from the control and the heated muscles were 1.0 ± 0.1 (n = 16) and 1.0 ± 0.1 (n = 12), respectively, for HSP25, 1.0 ± 0.1 (n = 16) and 1.0 ± 0.1 (n = 14) for αB-crystallin, and 1.0 ± 0.1 (n = 10) and 1.1 ± 0.1 (n = 15) for HSP72. (These values were calculated from the sum of the HSP in each individual fiber and wash set, each normalized by respective MHC amount, expressing all values relative to the mean for the control fibers run on the same gel.)
Importantly, when 1% Triton X-100 was present in the wash solution used to bathe SOL fibers from the heated muscles (e.g., Fig. 8B), 52 ± 12% of the HSP72 and 70 ± 7% of the SERCA were lost to the wash solution within 10 min in the six fibers examined, whereas virtually all of both the HSP25 and the αB-crystallin still remained in the fibers (mean data shown in Fig. 9).
The absolute amounts of the three major HSPs found here in unstressed skeletal muscle provide novel insights into their relative capabilities and limitations. The absolute amounts (Table 1) were in the order HSP72 < HSP27 < αB-crystallin in both EDL and SOL muscle, which are composed primarily of type II (fast-twitch) and type I (slow-twitch) fibers, respectively. In EDL muscle, the amount of αB-crystallin (∼4.9 μmol/kg muscle mass) was approximately fourfold greater than the amount of HSP72 but only slightly more than the HSP25. The most striking finding was that SOL muscle contained a very large amount of αB-crystallin (∼140 μmol/kg muscle mass), almost 30-fold higher than in EDL muscle, whereas the amount of HSP25 was only 2.5-fold higher in SOL fibers compared with EDL fibers (∼8.9 vs. 3.4 μmol/kg). The large amount of αB-crystallin found in the SOL muscle was not a peculiarity of rat, because bovine hind limb muscles were also found to contain very high amounts (∼130 and 70 μmol/kg in distal and proximal muscles). Like rat SOL muscle, these bovine muscles would be expected to have relatively high densities of type I fibers, and the relative amounts of αB-crystallin found in the distal and proximal muscles approximately matched the relative density of MHCI in those muscles.
The relative amounts of the HSPs found here in rat SOL muscle compared with EDL muscle (Table 1) are in good accord with previous studies in rat and mouse, where it was found that SOL muscle had ∼3-fold more HSP25 (35), ∼6-fold more HSP72 (12), and ∼15- to 40-fold more αB-crystallin (1, 11, 13, 35) than did EDL or tibialis anterior muscles (which both consist predominantly of type II fibers). The absolute amounts of the HSPs present in the muscles, however, are not apparent from those earlier studies because HSP content was typically examined only in enriched fractions rather than in total homogenates, and the amounts of the HSPs were reported relative to protein contents, which could not be related back to muscle amounts. One early report (9), however, did determine the αB-crystallin content of rat total heart homogenate as being ∼0.6% of total cellular protein, and, if the latter is assumed to be ∼230 g/kg muscle mass, this indicates that the αB-crystallin content of cardiac muscle is ∼70 μmol/kg. Furthermore, the same group later found (11) that the αB-crystallin content in total homogenates of rat SOL muscle was very similar to that of rat cardiac muscle, consistent with the very high absolute levels of αB-crystallin found here.
Using our recently described methodology for quantitative Western blotting of single fiber segments (20, 25, 28), the present study also provided the first quantitative assessment of the HSP levels present in individual muscle fibers. This single fiber analysis demonstrated that the type I fibers in rat SOL muscle are a broadly homogenous group, in all cases having greater levels of each of the HSPs than found in any of the EDL fibers (which were all type II fibers) (Fig. 3). Interestingly, both of the two type II fibers from SOL muscle examined in this study had lower levels of all three HSPs than was present in any SOL type I fiber, with the amounts being indistinguishable from those present in the EDL type II fibers (Fig. 3). The MHCII antibody used in this study detected all MHCII isoforms, and the fibers were not classified into type II subclasses. Nevertheless, it can be confidently assumed that the two type II SOL fibers examined here were not IIB fibers because previous studies have established that type II fibers in rat SOL muscle are all either IIA or less often IIX/D fibers and never IIB fibers (1, 4), in contrast to rat EDL muscle where the great majority are either IIB or IIX fibers. The findings here are in general accord with conclusions of immunohistochemical staining of muscle cross sections from rat muscle, which, although not truly quantitative, indicated that most if not all type I fibers contain much higher levels of αB-crystallin than type II fibers (1, 11), and also higher levels of HSP72 (3), whereas the differences between type I and type II are less marked with respect to HSP25 (11).
Comparison of HSP72 and SERCA amounts.
HSP72 binds to both SERCA1 and SERCA2 (8, 43), as well as to many other proteins, including the Na+-K+-ATPase (34) and K+ channels (7). SERCA proteins are present in very significant amounts, far more than the Na+-K+-ATPase, particularly in fast-twitch muscle fibers, and it appears that the SERCA must be an important target in quantitative terms, at least in rodent muscle. Specifically, in rat EDL muscle, there are ∼110 μmol of SERCA1 molecules/kg, with SERCA2 below detection limits, and, in rat SOL muscle, there is ∼20 and 15 μmol/kg of SERCA1 and SERCA2, respectively (44). Similarly in mouse muscle, the amounts equate to ∼100 μmol/kg muscle for SERCA1 and ∼1 μmol/kg of SERCA2 in EDL, and ∼20 and 1.5 μmol/kg, respectively, in SOL (30). Thus, the amount of HSP72 present in unstressed EDL muscle (∼1.1 μmol/kg, Table 1) is ∼100-fold lower than the density of the SERCA (∼100 to 110 μmol/kg), one of its putative targets. This disparity between HSP72 and SERCA densities apparent in EDL whole muscle homogenates also evidently pertains at the single fiber level because individual EDL fibers were all found to have both relatively high SERCA1 density (e.g., Fig. 3A and Ref. 25) and low HSP72 density (Fig. 3B). Even in SOL muscle fibers, where there is more HSP72 (Table 1) and fewer SERCA molecules, the number of SERCA molecules is still on average three- to fourfold greater than the number of HSP72 molecules present. Previous studies have shown that, following heat shock, at least some of the HSP72 molecules present become tightly bound to SERCA, which evidently helps protect both SERCA1 and SERCA2 molecules from thermal inactivation (8, 42, 43). It was found here that, in fibers from heat-treated EDL and SOL muscles, virtually all of the HSP72 remained tightly bound at sites within the fibers even with extensive washing, but, when the internal membranes were disrupted with the detergent Triton X-100, the HSP72 dispersed into the surrounding solution in tandem with the SERCA (Figs. 8 and 9). If HSP72 does need to be bound to SERCA to protect it from thermal inactivation, it is evident that the amount of HSP72 present in unstressed EDL muscle would be sufficient to bestow such protection only on at most ∼1% of the SERCA present, even without taking into account HSP72 binding to any other target molecules. This points to one reason why greatly increasing HSP72 content, particularly in type II fibers, might help enable them to cope in stressful situations and reflect why HSP72 protein expression in muscles has been observed in some cases to increase 10-fold or more following heat stress (17) or acute exercise (15).
Binding sites for HSP25 and αB-crystallin.
In contrast to HSP72, αB-crystallin and HSP25 are reported to bind in stressed conditions to structural and/or contractile proteins, and this seems consistent with the findings here that disruption and dispersion of intracellular membranes by Triton X-100 had no apparent effect at all on the tight binding of αB-crystallin and HSP25 in heat-stressed fibers (Figs. 8 and 9). It is unclear, however, whether the target proteins differ for these two HSPs, perhaps also differing with the type of stress involved. αB-crystallin and HSP25 (HSP27 in humans) have been reported to both accumulate at the Z-disk and with intermediate structures (thought to be desmin) in both mouse (16) and human (31) muscle following damaging eccentric exercise. On the other hand, with heat stress, αB-crystallin and HSP25 have been both reported to bind and protect actin (22, 36), and αB-crystallin was found to bind at least transiently with myosin and help preserve its ATPase function (19). Other studies, in contrast, found that, following ischemic stress, αB-crystallin and HSP25 bound to titin and desmin and not at all to myosin, actin, or α-actinin in cardiac muscle (10), and, in skeletal muscle, αB-crystallin showed diffuse binding across the whole I-band but none at the Z-disk (11). Hence, it is difficult to quantitatively compare the densities of these HSPs and their putative targets. Nevertheless, the following offers some important perspective. There is ∼60 μmol of α-actinin/kg muscle (38), and ∼2–3 μmol/kg of titin (assuming a total of 8 titin molecules tethering each myosin filament and 300 myosin molecules/filament). Thus, the amount of αB-crystallin present in EDL muscle (∼5 μmol/kg) would be just sufficient to bind at all the N-line sites (11) on the titin molecules but could bind to only a small proportion of the α-actinin molecules present in the muscle. The amount of αB-crystallin present in rat SOL muscle (∼140 μmol/kg muscle), however, would be sufficient to bind in a 1:1 manner on all the titin and all alpha-actinin molecules present, and even on many myosin molecules [∼94 μmol/kg (45)], although only on a fraction of the total actin molecules [620 μmol/kg (45)]. The fact that virtually all of the large amount of αB-crystallin present in rat SOL fibers (∼140 μmol/kg muscle) became bound in the heat-stressed fibers here (Figs. 8 and 9) indicates that it indeed likely targets many of the structural and sarcomeric proteins. In the case of HSP25, however, the absolute amount present in unstressed muscle fibers of either type (∼3 to 9 μmol/kg) is insufficient to bind and protect even a very limited range of the suggested target molecules. Furthermore, if HSP25 and αB-crystallin compete for the same binding site on particular target molecules, which seems possible given their homology, the binding of αB-crystallin would be expected to predominate, at least in type I fibers.
In any case, the very high αB-crystallin content of type I fibers stands out as an anomaly and is strongly suggestive that the role of αB-crystallin in such fibers differs substantially in some way from that of HSP25 and, furthermore, that it is important for its role in the type I fibers that large amounts are present even in the absence of overt stresses. The findings here also make it clear that the relative importance of the various HSPs should not be judged simply by the relative extent to which each is upregulated with stress. In the case of αB-crystallin in rat type I fibers, where there is ∼140 μmol/kg present in the unstressed state, even a 10% upregulation would represent a very considerable increase in absolute terms, in fact by an amount equal to the summed total of all HSP25 and HSP72 present in the muscle.
HSPs and the sarcolemma.
The results of the sarcolemma removal experiments here (Table 2 and Fig. 4) also demonstrated that some of the HSP25 and αB-crystallin present in unstressed muscle fibers is closely associated with the sarcolemma in some way, possibly bound or otherwise localized there. In the case of αB-crystallin, the proportion associated with the sarcolemma appeared to be much greater in the EDL fibers (∼60%) than in the SOL fibers (∼10%) (all type II and all type I fibers, respectively). This relative difference likely simply reflects that some appreciable amount of αB-crystallin is associated with the sarcolemma in both fiber types, but, in the SOL fibers, this amount represents only a comparatively small proportion of the total present, which is large in absolute terms (Table 1). These observations and explanation appear to fit well with the immunohistochemical staining patterns of rat muscle fibers, in which αB-crystallin was seen in type II fibers predominantly as a relatively bright ring around the fiber edge, whereas in type I fibers there was quite bright staining throughout the whole cross section (11). Furthermore, with HSP25, there was less apparent difference in the staining between type I and type II fibers, with an appreciable amount present around the edges in both types (11), which is consistent both with the much smaller disparity in HSP25 levels between type I and type II fibers seen here (Table 1) and the comparatively large amount (∼20%) apparently associated with the sarcolemma (Table 2), at least in some cases (Fig. 4). In contrast, no evidence was found here that HSP72 associated with the sarcolemma in unstressed fibers from rat (Table 2), and HSP70 immunohistochemistry of human fibers in control conditions shows no obvious sarcolemmal staining (32, 41). Such findings could result from HSP72 binding primarily at intracellular sites, in particular on SERCA (8, 43), or alternatively from the absolute levels of HSP72 in fibers being so low that the sarcolemmal component fell below detection limits.
Diffusibility of HSPs in unstressed muscle.
In skinned muscle fibers, proteins that are freely diffusible in the cytoplasm are readily lost to the bathing solution within ∼1–2 min (28, 37). Such experiments here (Figs. 5 and 6) demonstrated that, for all three HSPs, the majority of that present in rat SOL fibers was readily diffusible within the fiber. Taking into account the amounts associated and removed with the sarcolemma (Table 2), this diffusible component accounts for ∼60–70% of the total HSP25 and αB-crystallin and ∼90% of the total HSP72 present in SOL fibers. In EDL fibers, a similar proportion of the total HSP72 is diffusible, but, in the cases of HSP25 and αB-crystallin, <40% of the low total amount present is diffusible (Table 2 and Fig. 6). It is interesting to note that the HSP72 diffused out of the SOL fibers ∼8 to 10 times more slowly (time constant ∼2.2 min) than did the HSP25 and αB-crystallin (time constants ∼0.25 and 0.34 min, respectively). This disparity is not seemingly explained by ∼3-fold greater molecular mass of the HSP72, since, theoretically, this would be expected to slow diffusion only ∼1.5-fold (since diffusion rate depends on the cubed root of molecular mass if spherical shape is assumed). Moreover, we have previously shown that μ-calpain, a protein with a slightly larger molecular mass than HSP72 (80 kDa), washes out of rat skinned fibers with a time constant of ∼0.4 min (28). There are a number of possible explanations for the apparently slow diffusion rate of the HSP72. One is that much of the HSP72 might actually be loosely bound, coming off these sites with a time constant of ∼1–2 min. An alternative possibility is that HSP72 is continually binding and unbinding rapidly on various target molecules it encounters, considerably slowing its rate of diffusional loss. Presumably, the latter must happen to some degree in any process in which HSPs must rapidly recognize and strongly bind to damaged or unfolded proteins. Irrespective of this relatively minor slowing in the apparent diffusional rate of HSP72, a key finding here is that much of all three HSPs is freely diffusible or in rapid equilibrium with the cytoplasm in unstressed fibers, and hence could be expected to rapidly reach any cytoplasmic-assessable sites as needed.
In conclusion, this study highlights how it is necessary to know the absolute levels of each HSP present in skeletal muscle to fully understand the possible roles and limitations on each. Of particular note, it was found that the amount of HSP72 present in type II fibers in unstressed EDL muscle is much lower than the density of its putative binding sites, indicating why substantial upregulation would be advantageous in stress. On the other hand, type I SOL fibers contain very large amounts of αB-crystallin, even in the unstressed state, suggesting its importance in normal cellular function in such fibers.
This study was supported by National Health and Medical Research Council of Australia Grants 541938 and 602538.
No conflicts of interest are declared by the authors.
We thank Maria Cellini and Heidy Latchman for technical assistance and Dr. Frank Wuytack (Katholieke Universiteit, Leuven, Belgium) for the anti-SERCA2a antibody.
The monoclonal antibodies directed against adult human MHC isoforms (A4.84 and A4.74) were developed by Dr. Blau and those directed against SERCA1 were developed by Dr. D. Fambrough. All were obtained from the Development Studies Hybridoma Bank, under the auspices of the NICHD and maintained by the University of lowa, Department of Biological Science, Iowa City, IA 52242.
- Copyright © 2012 the American Physiological Society