Background

The formation of oligomers and aggregates of β-amyloid peptides derived from amyloid precursor protein (APP) is a causative early factor giving rise to Alzheimer Disease (AD) [1–3]. Neuritic plaques are the pathological feature associated with deposition of large aggregates of β-amyloid peptides. Mutations in the human APP gene can give rise to autosomal dominant early-onset forms of AD, which resemble late-onset AD (hereafter simply AD) both clinically and at the histopathological level [4–6]. Furthermore, some mutations in APP also confer protection against AD [7]. The deleterious APP mutations driving early-onset AD appear sufficient to promote, within humans, the formation of the other classical pathological hallmark of AD known as the neurofibrillary tangles (NFTs). NFTs are formed as a result of the self assembly of hyperphosphorylated microtubule-associated protein tau into paired helical filaments (PHFs) which then aggregate into larger structures known as NFTs [8–11]. Human mutations in the tau gene (MAPT), on their own, can also give rise to a group of neurodegenerative diseases referred to as frontotemporal dementia linked to chromosome-17 (FTDP-17), one group of a number of different types of tauopathy [12, 13]. However, human MAPT mutations do not cause the formation of neuritic plaques, thereby lending significant support for β-amyloid peptide formation as a factor upstream of tau in AD.

Studying the relationship between NFTs and neuritic plaques in AD is complicated by the fact that rodent models that carry human APP mutations do not generally display detectable NFT pathology [14, 15], even in the presence of neuritic plaques and cognitive impairment. Conversely, mouse models that express human MAPT mutations, such as the P301L mutant expressing JNPL3 mice, do not produce neuritic plaques [16]. For this reason several groups have developed mouse models of AD that recapitulate both pathological features of AD. One model, generated by Lewis et al., referred to as the TAPP mouse, resulted from crosses of JNPL3 tau mice with mice expressing the most common APP mutant (Swedish mutation K670N/M671L, APPSwe; Tg2576) [17]. Differing from the parent single transgenic models, TAPP mice brains contain both plaques and NFTs. Notably, these TAPP mice display enhanced tau pathology as compared to JNPL3 mice, suggesting that β-amyloid peptides accelerate the formation of NFTs [17], a proposal supported by other studies involving intracerebral injection of β-amyloid Aβ42 fibrils into P301L tau transgenic mice [18]. The TAPP mouse model is therefore well suited to study therapeutic strategies that might impact neuritic plaques or NFT formation in a setting that captures these two synergistic pathologies.

Previously, we reported a potential disease modifying approach aimed at reducing toxicity associated with tau aggregation and NFT formation. This approach involved increasing global levels of a little-studied post-translational modification known as the O-GlcNAc modification [19]. Chronic increases in the levels of O-GlcNAc modification in JNPL3 mouse brains and spinal cords slowed neuronal loss and reduced the number of NFTs formed over a treatment period of several months [20]. O-GlcNAc modification of proteins involves the attachment of single N-acetyl-D-glucosamine residues to the hydroxyl side chains of serine and threonine residues of proteins [19]. The O-GlcNAc modification differs from classical forms of glycosylation found on the outside of the cell and within the secretory pathway because O-GlcNAc is found in the nucleocytoplasm and it does not have additional sugar residues attached to it to form more complex structures [19]. The fact that O-GlcNAc can be added or removed from a particular protein multiple times during the lifespan of the protein makes it a dynamic modification [21] somewhat akin to protein phosphorylation, which is likewise reversible. Installation of O-GlcNAc on serine or threonine is carried out by a single glycosyltransferase referred to as O-GlcNAc transferase (OGT), which uses the high-energy donor sugar uridine 5’-diphospho-N-acetyl-D-glucosamine (UDP-GlcNAc) as its substrate [22, 23]. A single glycoside hydrolase, O-GlcNAcase (OGA), is tasked with the hydrolytic cleavage of O-GlcNAc from modified proteins [24, 25]. To increase global O-GlcNAc levels in JNPL3 mice we made use of a potent (Ki =21 nM) and selective (37000-fold for human OGA over functionally related human β-hexosamindases) inhibitor of OGA referred to as Thiamet-G, which blocks removal of O-GlcNAc from modified proteins [26]. Thus, even as OGA is inhibited, OGT can continue adding O-GlcNAc onto to modified proteins resulting in elevated O-GlcNAc within cells [26, 27].

We previously showed that increasing O-GlcNAc levels has beneficial effects in the JNPL3 mouse model of tauopathy [20]. Recently APP has been found to be O-GlcNAc modified and one study suggested that O-GlcNAc might alter β-amyloid production by regulating APP processing [28, 29]. To study the role of O-GlcNAc on APP and β-amyloid production in mice exhibiting both tau and β-amyloid pathologies, we undertook a long-term study using Thiamet-G to increase the global levels of O-GlcNAc in TAPP mice. Here we show that Thiamet-G can increase O-GlcNAc levels in the TAPP mouse brain, leading to reductions in levels of both neuritic plaques and amyloidogenic β-amyloid peptides. Consistent with these findings we also find that Thiamet-G treatment blocks the onset of cognitive impairment in these animals. We also show, using cell models of β-amyloid peptide formation, that Thiamet-G does not alter the release of β-amyloid peptides from cells, suggesting that increased O-GlcNAc levels in mice mediate protection against β-amyloid peptides through a mechanism that is likely independent of Aβ42 release.

Results

Thiamet-G treatment improves performance in the Morris water maze (MWM)

To assess whether increased O-GlcNAc can influence amyloid deposition or cognitive impairment in bigenic TAPP mice, we divided 60 double transgenic TAPP mice into three groups (n = 20) receiving either 0, 200, or 500 mkd of Thiamet-G in their drinking water. We have previously shown oral Thiamet-G treatment of mice over a period of months leads to sustained O-GlcNAc increases in the brains of mice [20]. Parental transgenic Tg2576 mice, which harbor only the APPSwe mutation, develop pronounced memory impairment starting at 6 months of age [15, 30] as judged by performance in the Morris water maze (MWM). Therefore to determine whether increased O-GlcNAc can influence disease progression in TAPP mice we started dosing with Thiamet-G at 10-13 weeks of age and continued until 44-47 weeks of age. Effects of OGA inhibition on cognitive impairment were assessed by performing MWM testing on each of the animals in the three groups starting between 28-32 weeks of age. All data acquisition and data entry was performed blinded with the coding being held by non-experimenters. We find that in the acquisition phase, during which animals engage in spatial learning, the latency to solve the maze was not significantly different between the groups as judged by RM-ANOVA (F5,78 = 0.668, p =0.649) (Figure 1A). Differences in the distance traveled within the maze during this learning phase were significantly different (F5,78 = 2.389, p =0.045), though a Tukey’s post-hoc analysis did not reveal any specific differences between groups (Figure 1B). To clarify whether behavioural changes that might be observed in these studies are due to cognition per se and not due to motor differences between groups, we examined the time spent in the outer ring as a measure of anxiety and swim speed during these acquisition trials. The swim speed was significantly different between groups (F5,78 = 2.817, p =0.022), with the Tukey’s post-hoc analysis indicating that the 500 mkd Thiamet-G treated group was faster than the untreated wild-type control group (p =0.042), perhaps because treatment protects against neurodegeneration of motor neurons as previously observed in JNPL3 mice [16]. There was no difference between the groups in the time spent in the outer ring (F5,78 = 0.758, p =0.582) indicating the animals showed no apparent anxiety effects.

Figure 1
figure 1

Thiamet-G prevents cognitive decline in the TAPP mice. A, B. Beginning at 30-32 weeks of age 0, 200 or 500 mkd Thiamet-G treated TAPP mice were tested for cognitive performance in the Morris water maze (MWM). Learning curves were recorded during five consecutive days of training. No significant difference was observed between any of the groups in latency to solve the maze (A) and while a significant main effect was noted by the ANOVA for distance travelled, the post-hoc Tukey’s analysis revealed that there were no significant differences between groups. (B). C. During the probe trial, the latency to solve the maze was recorded and the control TAPP mice (0 mkd Thiamet-G) show significant cognitive impairment compared to untreated age-matched wild-type mice. Conversely, 500 mkd resulted in better performance than the 0 mkd Thiamet-G treated TAPP mice while the performance of 200 mkd TAPP mice is indistinguishable from the untreated age-matched wild-type mice. D. No differences were observed in the distance travelled during the probe trial. Error bars represent standard deviation (± S.D) and p-value result from a one-way analysis of variance (ANOVA) For all panels, n =8 for 0 mkd wild-type mice, n =17 for 0 mkd TAPP mice, n =17 for 500 mkd TAPP mice and n =19 for 200 mkd TAPP mice.

Full size image

During the probe trials, where memories formed during the acquisition phase are tested, the 500 mkd Thiamet-G group spent significantly more time over the original platform location (F5,57 = 2.835, p =0.024) than did the untreated TAPP mice (p =0.03, Tukey’s post-hoc analysis) and the same amount of time as compared to wild-type control mice (Figure 1C), indicating increased O-GlcNAc blocks cognitive impairment in TAPP mice. Because the goal of the probe trial is for the animals to locate and remain within a very small area within the pool (~25 cm2), the differences between the 500 mkd Thiamet-G and the untreated TAPP mice are unlikely to result from motor differences between these groups. Consistent with this view, we did not observe any difference in the distance travelled during the probe trial (Figure 1D). These data indicate treatment with Thiamet-G resulted in cognitive enhancement compared to untreated animals, to a level matching that of wild-type animals.

Thiamet-G increases O-GlcNAc but does not impact tau phosphorylation

To better understand how Thiamet-G prevents impairment of cognitive performance, we first verified that O-GlcNAc levels were increased in treated animals (Figure 2). Immunoblotting with O-GlcNAc antibodies CTD110.6 and RL2 revealed dramatically increased O-GlcNAc levels in both the 200 and 500 mkd treated TAPP mice compared to the 0 mkd control TAPP mice (Figure 2A). We performed densitometric quantification of the O-GlcNAc immunoreactivity of the CTD110.6 antibody based on either all of the immunoreactive bands or only the low molecular weight (<50 kDa) immunoreactive bands (Figure 2B). This analysis revealed a trend toward lower O-GlcNAc levels in the 200 mkd TAPP mice group compared to the 500 mkd TAPP mice group when considering all of the immunoreactive bands. When only the low molecular weight bands are considered, O-GlcNAc levels were roughly 20% lower in the 200 mkd TAPP mice (p <0.05). These analyses reveal that there is a dose dependent effect of Thiamet-G on the levels of O-GlcNAc. We also performed immunohistochemistry (IHC) using the O-GlcNAc antibodies (CTD110.6 and RL2) on animals from the 0 and 500 mkd Thiamet-G treatment groups. The hippocampus and the cerebelleum have been previously shown to have high levels of expression of both OGA and OGT [31, 32] and O-GlcNAc levels in these structures, along with the pons and the amygdala, were particularly increased in the 500 mkd Thiamet-G treatment group (Figure 2C). Previously, we have shown that Thiamet-G does not block tau hyperphosphorylation in the transgenic parental JNPL3 mouse model [20]. We confirmed that this was also the case in the double transgenic TAPP mice by immunoblot (Figure 3A) and IHC (Figure 3B) analyses using various phosphorylation state-specific tau antibodies. Even though dramatically increased O-GlcNAc levels were observed (Figure 2) immunoblot analyses revealed that Thiamet-G treatment actually slightly increased the total amount of tau (92e) and the extent to which tau was phosphorylated but this did not reach statistical significance. In our previous study of JNPL3 mice [20], we found that Thiamet-G treatment reduced the amount of sarkosyl insoluble tau, which is known to correlate with the amount of fibrillar tau in 9 month old mice JNPL3 mice [33]. In a blinded experiment we therefore probed whether Thiamet-G can also reduce the amount of sarkosyl insoluble tau in the double transgenic TAPP mice. We observed a strong trend of less sarkosyl insoluble tau in the 500 mkd Thiamet-G treatment group (32% reduction), which was consistent in magnitude with our earlier JNPL3 study, although this difference did not reach statistical significance in the TAPP mice (Figure 4).

Figure 2
figure 2

O -GlcNAc levels are increased in the TAPP mouse brain. A. Western blots of total brain homogenates from 0, 200 and 500 mkd Thiamet-G treated TAPP mice reveals that O-GlcNAc levels are vastly increased (RL2 and CTD110.6) while actin indicates equal protein loading. B. Quantification of O-GlcNAc immunoreactivity (CTD110.6) normalized to actin by densitometry of all bands (left panel) or only the low molecular weight (MW) bands (bands <50 kDa, right panel). N =10 in each group. *indicates p <0.05, ***indicates p <0.001, unpaired two-tailed t-test) C. Immunohistochemical (IHC) analysis of 0 and 500 mkd Thiamet-G treated TAPP mice brain tissue reveals that O-GlcNAc levels are increased in all of the hippocampus (A’, E’), cerebellum (B’, F’), pons (C’, G’) and the amygdala (D’ , H’).

Full size image
Figure 3
figure 3

Thiamet-G does not alter total tau or phospho-tau levels in the TAPP mice. A. Total tau (92e) blots indicate that levels of total tau are not altered by Thiamet-G treatment. Western blots using phosphorylation state sensitive tau antibodies, pS199/pS202, pT205, pT212, pT214, pT217, pT231, pS262/pS356, pS396, and pS396/pS404 indicate that levels of phospho-tau are also not significantly altered. Quantification by densitometry of each total or phospho-tau epitope is shown to the right. Phospho-tau immunoreactivity is normalized to total tau (92e) in each case. N =10 in each group. B. IHC analysis using phosphorylation state sensitive tau antibodies, pS199/pS202, pS396, pS396/pS404, pT231, pS262/pS356 indicate that levels of phospho-tau are also not significantly altered in the hippocampus, the amygdala, or the pons in either of the 200 or 500 mkd Thiamet-G treated TAPP mice groups. Error bars represent standard error of the mean (± S.E.M).

Full size image
Figure 4
figure 4

500 mkd Thiamet-G may reduce sarkosyl-insoluble tau in the TAPP mice. Levels of sarkosyl-insoluble tau were assessed by Western blot analysis in both the 0 and 500 mkd Thiamet-G treated TAPP mice groups and indicated a strong trend towards less sarkosyl-insoluble tau in the 500 mkd group. Error bars represent standard error of the mean (± S.E.M) and p-values result from a student’s unpaired one-tailed t-test. N =18 (Control) and n =17 (500 mkd Thiamet-G).

Full size image

Thiamet-G treatment reduces the number of amyloid plaques and reduces Aβ levels

Previous work has shown that at 7-8 months of age the single transgenic JNPL3 mice generally perform no worse than age matched wild-type control animals in the MWM [34]. For this reason, and the absence of significant effects on tau phosphorylation, we felt that the cognitive effects of OGA inhibition in this TAPP model reflected in the MWM results may stem from effects of Thiamet-G treatment on accumulation of β-amyloid peptides and be reflected in the extent of amyloid plaque formation. To address this possibility we performed a number of experiments in which the experimenter was blinded using coded samples. The first study established whether Thiamet-G had any impact on the quantities of the amyloidogenic forms of β-amyloid 1-40 and 1-42 (Aβ40, Aβ42). Using widely used commercially available ELISA assays for Aβ40 and Aβ42, we determined that administration of 500 mkd Thiamet-G significantly reduced the quantity of Aβ42 while the 200 mkd treatment had no significant effect on this measure (Figure 5). We also observed a trend toward less Aβ40 in the 500 mkd treatment group that was not present in the 200 mkd group (Figure 5). As described above, O-GlcNAc levels in the 200 mkd group are slightly lower than the 500 mkd group and thus may indicate that a sustained increase in O-GlcNAc levels above a certain threshold must be reached in order to influence the levels of Aβ peptides within brain.

Figure 5
figure 5

Aβ 40 and Aβ 42 levels. ELISA assays for Aβ40 and Aβ42 were used to assess the quantity of each of these species in the 0, 200 or 500 mkd Thiamet-G treated TAPP mice groups. A. 500 mkd was sufficient to reduce the amount of Aβ42 by half while 200 mkd was ineffective. B. 500 mkd Thiamet-G also showed a trend toward less Aβ40 which was not evident in the 200 mkd group. Error bars represent standard error of the mean (± S.E.M) and p-values result from student’s unpaired two-tailed t-tests For all panels, N =14 for 0 mkd TAPP mice, N =16 for 500 mkd TAPP mice and N =13 for 200 mkd TAPP mice.

Full size image

Finally, we evaluated whether the reduced Aβ42 levels correlated with fewer amyloid plaques within these animals. By quantitative analysis, we observed fewer amyloid plaques in both cortical and hippocampal regions of the brain in the 500 mkd Thiamet-G treated group (Figure 6). Interestingly however, in the cortex we found that 200 mkd was sufficient to reduce the number of amyloid plaques to the same level as seen in mice treated with 500 mkd Thiamet-G despite this dose being unable to significantly reduce levels of Aβ42 as assessed by ELISA (Figure 6B). It is interesting to note that O-GlcNAc levels are generally higher in the hippocampus (Figure 2C) than in the cortex, yet we observe larger reductions in the number of plaques in cortex in both dose groups than observed in the hippocampus. Perhaps different brain regions have differing capacities to increase O-GlcNAc levels and this may be coupled to tissue dependent differences in amyloid deposition that depend on O-GlcNAc levels. Further, perhaps differences in the levels of Aβ42 within certain brain regions are also present yet were not detected here because the ELISAs are conducted on homogenates obtained from complete brain hemispheres. Using whole brain homogenates might mask small differences in Aβ42 levels present in certain brain regions. These observations suggest that increased O-GlcNAc could influence both β-amyloid peptide production or clearance as well as assembly/clearance of amyloid plaques arising from Aβ42.

Figure 6
figure 6

Thiamet-G reduces plaque load in the TAPP mice. A. Representative 6E10 immunohistochemical analysis used to assess amyloid plaque load in both the cortical and hippocampal regions. B. Quantitative assessment of 6E10 IHC analysis reveals that both 200 and 500 mkd Thiamet-G is sufficient to reduce the number of amyloid plaques in the cortex. C. 500 mkd Thiamet-G significantly reduced the number of amyloid plaques in the hippocampus whereas 200 mkd Thiamet-G was ineffective. Error bars represent standard error of the mean (± S.E.M) and p-values result from student’s unpaired two-tailed t-tests. For all panels, N =12 in each group.

Full size image

Thiamet-G treatment does not influence release of Aβ42 from cells

In an effort to address the question of how O-GlcNAc might affect β-amyloid peptide formation we turned to using a well-established cell culture model used to evaluate molecular pathways influencing APP processing [35–57] and cultured in primary neuron growth media (PNGM; Lonza). The astrocyte feeder layer for the neuronal co-culture was generated using neural progenitor cells as described [58]. A 50% volume media replacement was performed before neurons were treated with vehicle or 100 μM Thiamet-G for 24 hrs.

Immunohistochemistry

Free-floating brain sections were rinsed with PBS (pH 7.4) three times for 45 min first, then permeabilized with PBS containing 0.3% Triton X-100 (PBST) for 30 min. After blocking with 10% normal goat serum (NGS) and 2.5% BSA in PBST for 60 min, sections were incubated with mouse IgG monoclonal anti-β-amyloid antibody (6E10) at 4°C overnight, which reacts to human Aβ peptide (1-16 aa) and was generated in-house at the NYIBR and used at a dilution of 1:1000. After washing three times with PBS for 45 min, sections were incubated with Alexa488 goat anti mouse IgG (Invitrogen) secondary antibody at 1:1000 for 90 min. After 45 min washing, the sections were mounted on pre-coated slides (Adhesion superfrost plus, Brain Research Laboratories, Newton, MA), and cover-slipped with Vectashield MountingMedium (H-1000, Vector Laboratories). Sections examined in parallel but without being exposed to primary antibody served as experimental controls. The stained sections were examined using a Nikon C1 confocal system equipped with Nikon 90i fluorescent microscope. The images were acquired by the Nikon digital camera and Nikon C1 imaging system.

Aβ40 and Aβ42 ELISAs

Commercially available enzyme linked immunosorbant assay (ELISA) kits for amyloidogenic β-amyloid 1-40 and 1-42 (Aβ40 and Aβ42) (Invitrogen) were used to determine the total human Aβ40 and Aβ42 levels in the brain homogenates. Aβ42 was detected in neuronal cell culture media exactly as described by the manufacturer’s instruction using an ELISA kit from Wako (Human/Rat β-Amyloid(42)). Protein concentrations of the brain homogenates were determined using the BCA method (Pierce) and the Aβ40 and Aβ42 concentrations were expressed as ng Aβ40 or Aβ42/g of protein in the homogenate.

Quantitative immunohistochemistry of amyloid plaques (APs)

One sagittal section per mouse was immuno-stained with 6E10 which recognizes amino acids 1-16 of β-amyloid. Immuno-positive APs were counted manually and expressed as the number of APs/section.

Immunohistochemical staining

Frozen mouse brain sections (50 μm) stored in DeOlomos buffer at -20°C were washed with TBS and treated with 1% H2O2 and 50% methanol for 30 min to inactivate endogenous peroxidase. After being blocked with 5% normal goat serum in TBS for 1 hour, sections were incubated with the appropriate primary antibodies (see Table 1), followed by biotin-labeled secondary antibodies (Vector Laboratories), HRP-labeled avidin-biotin complex (Vector Laboratories), and the substrate DAB (3,3'-Diaminobenzidine, Sigma). The specimens were observed under a microscope (Nikon) and images captured using a digital camera DS-L2 (Nikon).

Table 1 Antibodies used in this study
Full size table

Antibodies

All antibodies used in this study are described in the Table 1.

Quantitative image analysis

Phosphorylated tau was visualized in sagittal brain sections from 6-9 mice each group by IHC staining. Using the identical microscope and camera settings, at least four digital images per sample were taken to reflect the overall staining in the pons region of brain. For the cortex and hippocampus, images at 20X were used. Only one image for hippocampus and one for frontal cortex were taken. All images were analyzed using the commercially available software program Image-Pro Plus version 4.0 for Windows (Media Cybernetics, Silver Spring, MD). The total immuno-intensity of the selected immuno-positive area was divided by the area size, and the values relative to that of the 0 mkd TAPP group are presented in the graphs. The proposed method allowed numerical analysis of the immunostaining intensity. The data from each group was normalized by the 0 mkd TAPP group.

Statistics analysis

For the MWM testing repeated measures analysis of variance (RM-ANOVA) was used to examine the effects of Thiamet-G on the latency of mice to complete the maze, their distance traveled, their speed of movement, and their time in each of the quadrants, with Tukey’s post-hoc to determine group differences where appropriate. The probe trial was analyzed using a one-way ANOVA with Tukey’s post-hoc tests. For all other data two-tailed unpaired student’s t-tests were used except for the sarkosyl insoluble tau data where we had data from previous studies to inform the predicted direction of change thus allowing the use of a one-tailed test in this case.

Western blotting

Brain tissue was homogenized at 4°C in 9 volumes of buffer (Buffer H) containing (50 mM Tris-HCl (pH 7.4), 20 μM UDP, 0.5 μM PUGNAc, 2 mM sodium othovanadate, 0.1 M NaF, 1.0 mM EGTA, 0.5 mM AEBSF, 2.0 μg/ml aprotinin, 10 μg/ml Leupeptin, 2.0 μg/ml pepstatin A) and the protein concentrations of the homogenates were determined using the BCA method (Pierce). The levels of global O-GlcNAcylation (CTD110.6 and RL2) and tau phosphorylation at various sites were determined by Western blot analysis using the antibodies listed in Table 1. Primary neurons were lysed in RIPA buffer containing Complete Protease Inhibitor Cocktail (Roche). Samples (10 μg) were resolved on 16% SDS-polyacrylamide gels and transferred to nitrocellulose membranes for blotting of APP-CTF. 30 μg of protein lysates was loaded per lane of a 4-20% SDS-polyacrylamide gel for blotting of O-GlcNAcylation (CTD110.6). Membranes were incubated with the following primary antibodies overnight at 4°C: anti-APP-CTF (1:2500, Abcam), anti-O-GlcNAc antibody (CTD110.6, 1:3000, Covance) and anti-actin (1:1000; Li-Cor). Quantitative analysis of immunoreactivity was carried out using images obtained by Li-Cor fluorescence and quantified using ImageJ. Quantitation of the lysate concentrations was performed using the BioRad QuickStart Bradford assay for both neurons and the 20E2 cells and levels of proteins established as a ratio of protein to actin.

Sarkosyl extraction

Whole tissue homogenates described above were first centrifuged at 10,000 × g in an Eppendorf 5417C centrifuge for 20 min. Supernatants were adjusted to have a final concentration of 1% N-laurylsacrosinate (Sarkosyl, Sigma) and were incubated at 37°C for one hour with shaking. Samples were then centrifuged at 100,000 × g in a Beckman TLA-45 rotor in a Beckman TL-100 centrifuge at 4°C for 45 min. The supernatant was removed and the sarkosyl insoluble pellet was resuspended in SDS-PAGE loading buffer.