JOURNAL OF NEUROCHEMISTRY

| 2008 | 106 | 2170–2183

doi: 10.1111/j.1471-4159.2008.05545.x

*Laboratory of Neurobiology, Campus Gasthuisberg, Leuven, Belgium  Biosignalling and Therapeutics, Campus Gasthuisberg, Leuven, Belgium àProtein Phosphorylation and Proteomics, Campus Gasthuisberg, Leuven, Belgium

Abstract Amyotrophic lateral sclerosis (ALS) is a chronic, adult-onset neurodegenerative disorder characterized by the selective loss of upper and lower motor neurons, resulting in severe atrophy of muscles and death. Although the exact pathogenic mechanism of mutant superoxide dismutase 1 (SOD1) causing familial ALS is still elusive, toxic protein aggregation leading to insufficiency of chaperones is one of the main hypotheses. In this study, we investigated the effect of overexpressing one of these chaperones, heat shock protein 27 (Hsp27), in ALS. Mice over-expressing the human, mutant SOD1G93A were crossed with mice that ubiquitously over-expressed human Hsp27. Even though the single transgenic hHsp27 mice showed protection against spinal cord ischemia,

the double transgenic SOD1G93A/hHsp27 mice did not live longer, and did not show a significant delay in the onset of disease compared to their SOD1G93A littermates. There was no protective effect of hHsp27 over-expression on the motor neurons and on the mutant SOD1 aggregates in the double transgenic SOD1G93A/hHsp27 mice. In conclusion, despite the protective action against acute motor neuron injury, Hsp27 alone is not sufficient to protect against the chronic motor neuron injury due to the presence of mutant SOD1. Keywords: amyotrophic lateral sclerosis, chaperones, HSPB1, motor neuron, protein aggregation, small heat shock proteins. J. Neurochem. (2008) 106, 2170–2183.

Amyotrophic lateral sclerosis (ALS) is an adult-onset neurodegenerative disorder characterized by severe and selective death of motor neurons in the cortex, brain stem, and spinal cord. The progressive motor neuron loss results in spasticity, hyperreflexia, general weakness, muscle atrophy and paralysis of the patient (Rowland 1998). Collapsing of the respiratory muscles ultimately leads to death of the ALS patients in about 1–5 years after disease onset. The etiology of 90% of ALS cases is unknown and hence is termed sporadic, while 10% have a familial history. In 20% of familial ALS cases, mutations in the superoxide dismutase 1 (SOD1) gene were identified (Rosen et al. 1993). As in other neurodegenerative disorders like Alzheimer’s, Huntington’s and Parkinson’s disease where protein aggregation overrides the protein quality control system (Stefani and Dobson 2003; Ross and Poirier 2004), SOD1 inclusions and aggregates have been suggested to be the primary cause of toxicity in ALS (Lindberg et al. 2005). These inclusions are also known to sequester chaperones or heat shock proteins (Hsp’s) – molecules that aid in proper refolding of

misfolded proteins preventing their aggregation – thereby making them unavailable to perform other functions like nascent folding, transport and degradation of proteins, signal transduction, cell growth, differentiation and protection against apoptosis (Jolly and Morimoto 2000; Nollen and Morimoto 2002). Thus, chaperone insufficiency is suggested to be one of the main causes that toxic SOD1 aggregates are not efficiently cleared from the cell’s milieu leading to a ‘selfpropagating’ cycle of events ultimately leading to its demise. Many in vitro studies have documented the protective effects of various Hsp’s, notably Hsp70, Hsp40, aB crystallin

2170

Received December 5, 2007; revised manuscript received May 22, 2008; accepted June 11, 2008. Address correspondence and reprint requests to Dr Ludo Van Den Bosch, Laboratory of Neurobiology, Campus Gasthuisberg O&N2 PB 1022, Herestraat 49, Leuven B-3000, Belgium. E-mail: [email protected] Abbreviations used: BSA, bovine serum albumin; CAA, citrate synthase aggregation assay; GFAP, glial fibrillary acidic protein; PBS, phosphate-buffered saline; PVDF, polyvinylidene difluoride.

 2008 The Authors Journal Compilation  2008 International Society for Neurochemistry, J. Neurochem. (2008) 106, 2170–2183

Hsp27 and amyotrophic lateral sclerosis | 2171

and Hsp27 in models of ALS (Bruening et al. 1999; Takeuchi et al. 2002; Batulan et al. 2003, 2006; Patel et al. 2005; Wang et al. 2005), while we reported that a general heat shock response protected against mutant SOD1-dependent cell death (Krishnan et al. 2006). In vivo protective effects of compounds like arimoclomol and celastrol in ALS mice (Kieran et al. 2004; Kiaei et al. 2005) have raised the impetus for chaperone inducers as therapeutic molecules (Klettner 2004; Traynor et al. 2006). We previously showed an increase in the levels of the small, stress inducible, Hsp27 (HSPB1) in the spinal cord homogenates of ALS mice as disease progresses (Vleminckx et al. 2002). Subsequently, Maatkamp et al. (2004) reported that a loss of Hsp27 from motor neurons precedes their degeneration in ALS suggesting that retaining Hsp27 in the motor neurons might be protective. In this study, we investigated whether Hsp27 over-expression could protect motor neurons in the ALS mice against the devastating effects of mutant SOD1.

Materials and methods Experimental animals Mutant [B6SJL-TgN (SOD1-G93A) 1Gur] SOD1G93A transgenic mice were purchased from the Jackson Laboratory (Bar Harbor, ME, USA). Mice over-expressing hHsp27 transgene were obtained from Dr. Dilmann, University of California, CA, USA. The generation of these mice has been described previously (Hollander et al. 2004). The hHsp27 transgenic mice were back-crossed into C57/BL6 background for seven generations. Female hHsp27 mice were then crossed with the SOD1G93A males and offspring was genotyped to obtain double transgenic mice that contained both mutant SOD1G93A and the hHSP27 transgene (SOD1G93A/hHsp27). The grip strength of the forelimbs was measured using a Grip Strength Meter (Columbus Instruments, Columbus, OH, USA) as described by Smith et al. (1995). The measurements were performed twice a week with the apparatus in the tension mode using a triangular bar. The maximal force (in gm) of five attempts was averaged. A rotarod (SciPro Inc., North Tonawanda, NY, USA) rotating at 15 rpm was used to evaluate motor performance. Each mouse was given five trials of 180 s maximally and the average time on the rotarod was used as a measure of motor performance. Onset of disease was determined as the age at which mice were not able to stay on the rotarod for more than two standard deviations below the average performance during the period between 65 and 85 days of age. Death was defined as the age at which a paralyzed animal was no longer able to roll over to the normal position within 10 s after being pushed on its back. All experiments were approved by the ethical committee of the University of Leuven. Genotyping of mice Genotyping of mice for the presence of mutant SOD1G93A was performed according to the protocol provided by the Jackson Laboratory with minor modifications using the Ready-To-Go PCR beads (Amersham, Piscataway, NJ, USA). Four primers used were as follows: 5¢-CTAggCCACAgAATTgAAAgATCT-3¢; 5¢-gTAgg-

TggAAATTCTAgCATCATCC-3¢; 5¢-CATCAgCCCTAATCCATCTgA-3¢; 5¢-CgCgACTAACAATCAAAgTgA-3¢. The PCR protocol used was as follows: 95C for 5 min, [95C for 15 s; 55C for 15 s; 72C for 30 s] for 29 cycles, 72C for 5 min and 4C pause. This procedure yielded a 236 bp band in the transgenic animals and a 324 bp band as internal control. Primers for genotyping the hHsp27 transgene were: 5¢-ATTACggggTCATTAgTTCATAgCC-3¢ and 5¢-TgTggAATACCCTgAAAggATg-3¢ The PCR program was as follows: 95C for 5 min, [95C for 10 s; 60C for 10 s; 72C for 25 s] for 30 cycles. This reaction amplifies a band of 290 bp in transgenic animals. Electrophoresis and immunoblotting For sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) analysis, spinal cords were isolated and homogenized in a buffer (pH 7.5) containing 50 mM Tris, 150 mM NaCl, 1% NP40 (Sigma, St. Louis, MO, USA), 0.5% Sodium deoxycholate, 0.1% SDS and protease inhibitors (Complete, Roche Diagnostics, Basel, Switzerland), while for native gel analysis, spinal cords were homogenized in phosphate-buffered saline (PBS) and protease inhibitor (Complete). Insoluble debris was pelleted by centrifugation (1000 g) and protein concentration of the supernatant was determined using the Micro BCA Protein Assay (Pierce Endogen, Rockford, IL, USA). Equal amounts of protein were resolved on a 12% SDS–PAGE gel or 12% native gel and transferred to a polyvinylidene difluoride (PVDF) (Millipore Corp., Bedford, MA, USA) membrane. The non-specific binding was blocked by incubation in 5% bovine serum albumin for 1 h at 23C [bovine serum albumin (BSA); Serva, Heidelberg, Germany] and incubated with primary antibodies overnight. The antibodies were: Hsp22 (1/2000, a kind gift of Dr. Benndorf, MI, USA), Hsp27 (1/250, Santa Cruz, Santa Cruz, CA, USA), Hsp40 (1/5000), Hsp70 (1/1000), Hsp90 (1/10000) all from Stressgen (Victoria, Canada), caspase 3 and caspase 9 (1/1000) from Cell Signalling (Danvers, MA, USA). Secondary antibodies coupled to alkaline phosphatase (1/5000, Sigma) were used for 1 h at 23C. To minimize intergel variabilities, all the samples were loaded on the same gel and the experiment repeated at least twice. Blots were then scanned on a STORM 840 scanner (Blue Fluorescence, Molecular Dynamics, Sunnyvale, CA, USA) and quantified using IMAGEQUANT version 1.1. Blots were stripped with methanol and an anti-b-actin (Sigma) staining was performed to check equal protein loading. The intensity of the protein bands were normalized to the intensity of b-actin staining and compared with controls to evaluate the difference in expression levels. Co-immunoprecipitation Spinal cord extracts (10 lL) in a total volume of 25 lL PBS were incubated with either 5 lL of Hsp27 or 3 lL of SOD1 antibody for 3 h at 4C. 10 lL of resuspended Protein G beads (Santa Cruz) were then added to this mixture and the total volume made up to 1 mL with PBS. This was then allowed to rotate slowly at 4C for 1 h, followed by precipitation of the beads by centrifugation at 420 g for 2 min. The beads were washed four times with PBS containing 0.85 M NaCl and 0.01% NP40 detergent and precipitated by centrifugation as above, followed by two washes in PBS alone. The beads were then resuspended in 20 lL of 1X SDS sample buffer and boiled at 95C for 5 min. The supernatant from

 2008 The Authors Journal Compilation  2008 International Society for Neurochemistry, J. Neurochem. (2008) 106, 2170–2183

2172 | J. Krishnan et al.

this was loaded onto SDS gel and western blotting carried out as described earlier. Filter trap assay Spinal cords were isolated and homogenized in PBS containing protease inhibitor (Complete). After centrifugation at 800 g, equal amounts of the supernatant protein (40 lg) was filtered under vacuum through a 0.22 lm cellulose acetate membrane (Schleider and Schuell, Dassel, Germany) followed by two washes in PBS. The membrane was incubated in 5% BSA and then processed similar to the PVDF membranes as described above. Gel filtration Spinal cords from symptomatic SOD1G93A/hHsp27 and SOD1G93A mice were homogenized in 1 mL of PBS + protease inhibitor (Complete). The lysates were cleared by centrifugation at 17 900 g for 15 min at 4C. 100 lL of the supernatant was then loaded onto a Superdex 200 (HR 16/50; Pharmacia, Amersham, Buckinghamshire, UK). The column was pre-equilibrated with 50 mM Tris (pH 7.5) buffer containing 150 mM NaCl and the samples were chromatographed in the same buffer. The eluting fractions (flow rate 1 mL/min) were collected as 500 lL per tube. The absorbance value (at OD 280 nm) of each fraction was used to draw the elution profile. Protein standards were run prior to the samples and a standard curve was generated based on their elution volume (Thyroglobulin: 669 kDa, Ferritin: 440 kDa, Aldolase: 158 kDa, BSA dimer: 134 kDa and BSA: 67 kDa). A fraction of the odd-numbered sample eluates (100 lL) was blotted onto PVDF membrane using a slot blot machine under vacuum, and the membrane processed for Hsp27 and SOD1 staining as described before. Immunochemistry and histology SOD1G93A/hHsp27 and SOD1G93A mice in different age groups, pre-symptomatic (aged 70–90 days), symptomatic (100–120 days) and end stage (130–140 days), were anesthetized using Nembutal (10 mg/kg) and transcardially perfused with 4% paraformaldehyde. After further processing in 20% and 30% sucrose, spinal cords were embedded in Tissue-Tek (Sakura Finetek Europe, The Netherlands). Cryosections (16 lm thick) were cut and collected on pre-coated glass slides. The sections were washed in PBS and blocked with 10% normal serum (Novocastra, Newcastle, UK), 0.1% Triton X-100 (Sigma) in PBS for 1 h at RT followed by incubation with primary antibodies overnight at 4C in PBS. Primary antibodies used were anti-Hsp27 (1/250), anti-SMI-32 (1/1500, Sternberger Monoclonals, Inc., Luthersville, MD, USA) and anti-glial fibrillary acidic protein (GFAP) (1/500, Dako Cytomation, Denmark). SMI-32 and GFAP were detected using Alexa488-labeled secondary antibodies (Vector Laboratories, Burlingame, CA, USA) and Hsp27 was detected using CY3 labeled secondary antibody (Sigma). Sciatic nerve pieces fixed in 4% paraformaldehyde and embedded in paraffin were used for immunohistochemistry. Hsp27 primary antibody was used on these sections followed by incubation with biotin coupled secondary antibody. Visualization was performed with routine 3, 3¢-diaminobenzidine tetrahydrochloride staining (Sigma). Paraffin-embedded spinal cords were used to perform hematoxylin and eosin (H&E) staining. For counting, 10 lumbar sections per animal with a 35 lm interval between each were used. In electronic pictures of every slide, the area of normal appearing

motor neurons with nucleoli in the ventral horn was calculated using LUCIA IMAGE (Version 4.60, Laboratory Imaging, Prague, Czech Republic) and the number of neurons in different size groups was determined. Spinal cord ischemia Spinal cord ischemia reperfusion was carried out as described previously (Lang-Lazdunski et al. 2000a,b), by clamping the aortic arch, left subclavian artery and internal mammary artery for 10 min (36C) which causes persistent paraplegia in mice. Motor function in the hind limbs was scored 0, 1, 3, 24, and 48 h after reperfusion using a rating scale previously published (Lambrechts et al. 2003) by an experimenter blinded for the genotype of the mice: 0 (normal function), 1 (hind limbs are stretched, but remain rigid when the mouse is lifted by its tail, normal walking), 2 (no stretching of hind limbs when lifted by tail, normal walking), 3 (rigid hind limbs when lifted by tail, ataxic or knuckle walking), 4 (rigid hind limbs when lifted by tail, hind limb movement but no walking) and 5 (rigid hind limbs when lifted by tail, total absence of movement). After 48 h the spinal cords were processed for histology stainings using H&E as described above. Citrate synthase aggregation assay Assays were performed essentially as described by Hook and Harding (1997) and modified based on Lindner et al. (2000) and Stromer et al. (2003). This assay is dependent on the ability of chaperones present in cytosolic extracts to prevent heat denaturation of an exogenously added protein (citrate synthase in these experiments). In brief, tissues were lysed in 50 mM HEPES (pH 7.4), 100 mM KCl, 5% glycerol, 1 mM MgCl2, and 0.1% NP40 [citrate synthase aggregation assay (CAA) buffer] containing Complete protease inhibitor. Forty micrograms of extract was mixed with 400 lg/mL citrate synthase (Sigma) dissolved in PBS to a final volume of 500 lL CAA buffer. The reaction mix was heated at 43C, and aggregation of the denatured protein was followed for 20 min by measuring light scattering at 360 nm with a spectrophotometer. Statistics Students t-test was used to check for statistical significance for the quantification of western blots. Repeated measures ANOVA was used on grip strength and aggregation assays. Log-rank analysis was performed on disease onset and survival data.

Results Hsp27 expression in mutant SOD1G93A mice In our previous study, an increase in the level of endogenous Hsp27 in total spinal cord homogenates of SOD1G93A mice as disease progresses, was observed (Vleminckx et al. 2002). In this study, we performed immunohistochemical staining on lumbar sections of spinal cord from SOD1G93A mice of different ages. As shown in Fig. 1, few motor neurons express Hsp27 in pre-symptomatic mice (70–90 days; Fig. 1a). Moreover, there is little Hsp27 staining in the rest of the ventral spinal cord at this stage. Similarly, not all

 2008 The Authors Journal Compilation  2008 International Society for Neurochemistry, J. Neurochem. (2008) 106, 2170–2183

Hsp27 and amyotrophic lateral sclerosis | 2173

(a)

(b)

(c)

motor neurons express Hsp27 in the symptomatic mice (100– 120 days; Fig. 1b). However, the few motor neurons that remain in the end-stage SOD1G93A mice all stain positive for Hsp27 (130–140 days; Fig. 1c), as do the large number of stellar-shaped GFAP-positive glial cells at this stage as we have previously shown (Vleminckx et al. 2002). This could indicate that the motor neurons negative for Hsp27 in the pre-symptomatic and symptomatic stages are

Fig. 1 Hsp27 expression as a function of disease in SOD1G93A mice. Representative micrographs showing the expression of endogenous Hsp27 (red) in SOD1G93A mice. Motor neurons are identified by their large, round shape and their staining with the neurofilament marker SMI32 (green). (a) Spinal cord sections of the pre-symptomatic mice (70–90 days) demonstrate large motor neurons (broken arrow) that are positive for Hsp27, while others do not stain positive (arrow). No other cells are Hsp27 positive at this stage. (b) Sections of symptomatic mice (100–120 days) reveal some motor neurons that are negative for Hsp27 (arrow). (c) Spinal cord sections of end-stage mice (130–140 days) reveal no motor neurons that are devoid of Hsp27. The motor neurons remaining at this time point are all positive for Hsp27 (broken arrow). Scale bar: 50 lm; WM: white matter; GM: gray matter.

the ones that die first. Alternatively, it could be that motor neurons that originally were negative up-regulate Hsp27 in an attempt to remain healthy. In either case, sustained Hsp27 expression in motor neurons could protect against the toxicity of mutant SOD1. Hence, by constitutively overexpressing Hsp27 in mutant SODG93A mice we aimed to evaluate whether this could be beneficial for motor neurons. Characterization of hHsp27 over-expressing mice Transgenic mice over-expressing human Hsp27 (hHsp27) have been described before and large increases in the hHsp27 protein levels in heart tissue of these mice were reported (Hollander et al. 2004). With an antibody that recognizes human and mouse isoforms of the protein, this overexpression in heart was confirmed and a more than 25-fold over-expression of the human protein was detected in other organs (including kidney and liver; data not shown). The expression level of hHsp27 in spinal cord was determined by western blot. Total spinal cord homogenates of the hHsp27 mice showed at least 40-fold higher expression level of the combined protein (mouse Hsp27 and human Hsp27) compared to the non-transgenic controls (Fig. 2a, upper blot). Other brain tissues including cortex, cerebellum, hippocampus showed an increase up to 25-fold (Fig. 2a, lower blot). Immunohistochemistry for the localization of hHsp27 showed that the hHsp27 protein is over-expressed in all motor neurons of the ventral spinal cord of these mice (Fig. 2c-i) compared to the non-transgenic control (Fig. 2bi). Given its non-cell autonomous nature, over-expression of hHsp27 in motor neurons alone might not be sufficient to ameliorate ALS. Although not many GFAP-positive glial cells are seen in these phenotypically normal hHsp27 mice, we confirmed that the over-expression occurs also in resident glial cells stained with GFAP antibody (Fig. 2c-ii, iii) in contrast to the non-transgenic mice tissue (Fig. 2b-ii, iii). A clear over-expression of hHsp27 was found in the axons of the transgenic mice compared to the non-transgenic littermate, as was revealed by paraffin sections of sciatic nerve stained for Hsp27 using the 3, 3¢-diaminobenzidine

 2008 The Authors Journal Compilation  2008 International Society for Neurochemistry, J. Neurochem. (2008) 106, 2170–2183

2174 | J. Krishnan et al.

Fig. 2 Characterization of the hHsp27 transgenic mice. (a) Western blot to study the over-expression of hHsp27 protein in the CNS of hHsp27 transgenic mice. Lanes 1–3 contain total spinal cord homogenates (upper blot) from three different non-transgenic mice (Ntg) and lanes 4–6 contain hHsp27 transgenic mice (hHsp27) samples. Lanes 1, 4 in the lower blot contain protein samples from cortex (Ctx), lanes 2, 5 from cerebellum (Crb) and 3, 6 from hippocampus (Hpc) of Ntg and hHsp27 mice. The blot was stripped with methanol and reprobed with an antibody to b-actin to demonstrate equal protein loading. The intensity of the protein bands were normalized to the intensity of b-actin staining and compared to the Ntg controls to evaluate the fold of increase in the transgene expression (p = 0.0024). (b and c) Representative micrographs of immunohistochemistry on lumbar spinal cord sections. Not all neurons staining positive for SMI32 (green, arrow) are positive for Hsp27 (red, broken arrow) in the non-transgenic control (b-i). The hHSP27 transgenic sample shows that all neurons (broken arrow) express the hHsp27 protein (c-i). Scale bar: 50 lm. The nuclei are stained with DAPI (blue). The smaller

micrographs (ii: GFAP stained in green; iii: merge with Hsp27 staining in red and nucleus in blue with DAPI) in both b and c, demonstrate that hHsp27 over-expression in the transgenic mice (c-ii,iii) is also seen in the glial cells (Scale bar: 20 lm; WM: white matter; GM: gray matter). (d) Representative micrographs showing the staining pattern of Hsp27 in sciatic nerve from the transgenic mice compared to the Ntg littermate (Scale bar: 50 lm). (e) Western blot of other Hsp’s in the spinal cord homogenates of hHsp27 transgenic mice. Lane 1 is a sample from an Ntg mouse, lanes 2–4 are samples from three different hHsp27 transgenic mice. The blots were probed with antibodies to Hsp22, Hsp40, Hsp70, and Hsp90 followed by re-probing with b-actin to demonstrate equal protein loading and intensities quantified as in Materials and methods (p = 0.043). (f) Average neurological scores from hHsp27 (white bar; n = 6) and control mice (black bar; n = 4) at 48 h post-spinal cord ischemia reperfusion (p = 0.039). (g) Representative micrographs of H&E staining on paraffin-embedded spinal cord tissue demonstrating healthy motor neurons in the hHsp27 transgenic mice 48 h post-spinal cord ischemia (Scale bar: 50 lm).

tetrahydrochloride method (Fig. 2d). Moreover, accompanying changes in the levels of other Hsp’s were detected in these transgenic mice. Western blots of spinal cord homogenates were probed with antibodies against Hsp22, Hsp40, Hsp70 and Hsp90 (Fig. 2e). Increases in levels of Hsp90 and Hsp70, up to 1.5–2-fold, were found, while the Hsp22 levels were increased 5–7-fold in the hHsp27 transgenic mice. Since these mice have been shown to be protected against cardiac ischemia previously, we evaluated whether such protection could be seen in a paradigm affecting motor neurons. Hence, spinal cord ischemia experiments were performed on these mice. As shown in Fig. 2f, hHsp27 transgenic mice recovered partially from ischemia compared to the non-transgenic littermates, which remained paralysed persistently 48 h post-reperfusion. Histology demonstrated that most motor neurons remained healthy in the hHsp27 mice compared with controls (Fig. 2g). These results indicate that hHsp27 over-expression can offer protection against motor neuron injury.

(upper blot), while the level of other Hsp’s (e.g., Hsp40 is shown in the lower blot) did not differ between both the groups at end-stage. Immunohistochemical staining of the lumbar spinal cord sections from symptomatic samples also confirms that the expression of Hsp27 in the SOD1G93A/ hHsp27 mice spinal cord is higher than the endogenous mouse Hsp27 levels in control SOD1G93A animals (Fig. 3c). The inset in Fig. 3c shows that GFAP-positive glial cells also stain for Hsp27. These results demonstrate that there was no decrease in the level of Hsp27 expression in our SOD1G93A/ hHsp27 mice at any age. Grip strength and weight reduction have been reported as reliable markers of disease onset and progression in the ALS mice, especially in earlier stages. Hence, decline in muscle strength in the forelimbs of the SOD1G93A cohort was compared with that of the SOD1G93A/hHsp27 littermates. Fig. 4(a) demonstrates that the grip strength in both groups remained similar throughout the disease process (p = 0.9). Similarly, no difference in weight reduction between both the groups was observed (data not shown). Disease onset was also determined for SOD1G93A/hHsp27 and SOD1G93A littermates using the rotarod test. As shown in Fig. 4b, rotarod performance was similar in both groups on all days evaluated. Onset of disease as determined by failure on the rotarod was also similar in both groups with a mean (± SEM) of 105.8 ± 3.5 days for the SOD1G93A/hHsp27 mice and 106.1 ± 3.7 days for the SOD1G93A littermates (Fig. 4c). Compared to the SOD1G93A littermates, survival in the SOD1G93A/hHsp27 mice was prolonged by 4.2 days, increasing the mean survival (± SEM) from 139.6 (± 1.9) to 143.8 (± 3.2) days, which did not reach statistical significance (Fig. 4d). There was a similar mild increase in disease duration of the SOD1G93A/hHsp27 mice (37.0 ± 3.9 days) compared with the SOD1G93A littermates (33.5 ± 4.9 days) also not reaching statistical significance (p = 0.5).

Effect of hHsp27 over-expression in SOD1G93A mice Mutant SOD1G93A mice were crossed with the hHsp27 transgenic mice and offspring that over-express both mutant SOD1G93A and hHsp27 (SOD1G93A/hHsp27) were identified by PCR. As shown in Fig. 3a, human Hsp27 is clearly expressed in the SOD1G93A/hHsp27 mice even in the presymptomatic age when mouse Hsp27 is negligible in the control SOD1G93A littermates (upper blot). Also overexpression of hHsp27 did not affect the expression of mutant SOD1 (middle blot). Since up-regulation of endogenous mouse Hsp27 is observed in end-stage SOD1G93A mice (Vleminckx et al. 2002), we also checked if the human transgene was expressed above this level. Western blot shown in Fig. 3b demonstrates that Hsp27 expression in the SOD1G93A/hHsp27 littermates is higher than the normal ‘up-regulation’ of Hsp27 in the control SOD1G93A mice

 2008 The Authors Journal Compilation  2008 International Society for Neurochemistry, J. Neurochem. (2008) 106, 2170–2183

Hsp27 and amyotrophic lateral sclerosis | 2175

(b) (i)

(a)

(ii)

(c) (i)

(iii)

(ii)

(iii)

(d) (e)

(f) (g)

In order to evaluate the effect of Hsp27 over-expression on motor neuron survival, an H&E staining on paraffin sections of symptomatic mice from both SOD1G93A/hHsp27 and SOD1G93A groups was performed (Fig. 5a). Systematic counting of the motor neurons demonstrated no difference in their numbers between both groups (Fig. 5b). Accumulation of age-dependent, tissue-specific, mutant SOD1 aggregates (large enough to be trapped by 0.2 lm cellulose acetate filters) are suggested to be the primary cause

of pathology in ALS (Wang et al. 2002). Using cellulose acetate membranes, we were able to be trapped such aggregates in the SOD1G93A mice as age progresses, compared to SOD1WT and non-transgenic controls (data not shown). However, no change in the age-dependent accumulation of aggregates in the SOD1G93A/hHsp27 mice was observed (Fig. 5c), as also confirmed by quantification of the intensity of aggregates (Fig. 5d). We also found no changes in the levels of caspase 3 between both the groups

 2008 The Authors Journal Compilation  2008 International Society for Neurochemistry, J. Neurochem. (2008) 106, 2170–2183

2176 | J. Krishnan et al.

(a)

(c)

(b)

Fig. 3 Expression of hHsp27 in the SOD1G93A/hHsp27 double transgenic mice. (a) Western blot of spinal cord homogenates from presymptomatic mice (70–90 days) stained with Hsp27 antibody (upper blot), demonstrating the overexpression of hHsp27 in SOD1G93A/ hHsp27 mouse (lane 1) compared to SOD1 G93A littermate (lane 2). Lanes 3 and 4 contain samples from hHsp27 and Ntg mice. The same blot was stripped and re-stained with SOD1 antibody (middle blot), followed by re-staining with b-actin for equal protein loading. (b) Western blot of spinal cord homogenates from three different endstage mice (130–140 days) stained with Hsp27 antibody (upper blot).

The blot was stripped and stained with Hsp40 antibody (lower blot; which also serves as the internal control for equal protein loading). (c) Representative micrographs of spinal cord sections from symptomatic mice (100–120 days) stained with SMI32 (green) and Hsp27 (red) antibodies. The SOD1G93A sample reveals a motor neuron negative for Hsp27 (arrow), while some are positive (broken arrow). The SOD1G93A/hHsp27 sample shows that all motor neurons stain positive for Hsp27 (broken arrow). Scale bar: 50 lm; WM: white matter; GM: gray matter. Inset shows glial cell stained with GFAP antibody (green) and Hsp27 (red). Scale bar: 50lm.

before or after onset of disease (Fig. 5e). The activated form of caspase 9 increased during the end-stage concomitant with a decrease in the full-length protein (Fig. 5e; arrow head and arrow, respectively). However, this was not different between SOD1G93A and SOD1G93A/hHsp27 mice. Chaperone activity of the spinal cord tissue in SOD1G93A mice has been shown to be altered compared with SOD1WT mice as the mice aged (Bruening et al. 1999; Tummala et al. 2005). To assess if over-expression of hHsp27 has an effect on the chaperone activity of the SOD1G93A mice, we performed aggregation assay as described by Lindner et al. (2000) with modification (see Materials and methods). Although the spinal cord extracts from single transgenic hHsp27 mice were able to inhibit thermal aggregation of citrate synthase, there was no difference in the chaperoning activity between the SOD1G93A and SOD1G93A/hHsp27 samples at any age. Fig. 5f shows the chaperoning activity of hHsp27 transgenic samples compared with non-transgenic littermates and end-stage SOD1G93A and SOD1G93A/hHsp27 samples. hHsp27 transgenics are known to be protective against acute ischemic insult (Hollander et al. 2004), similar to our current study using spinal cord ischemia experiments. To

understand the lack of protection against chronic toxic effects of mutant SOD1, we investigated whether there was interaction between mutant SOD1 and Hsp27, which can counteract the abilities of Hsp27 to function efficiently. Co-immunoprecipitation experiments showed that incubation of spinal cord extracts with SOD1 antibody pulled down Hsp27 protein indicating an interaction between these two proteins (Fig. 6a). Although the amount of proteins pulled down seemed moderate, this experiment clearly demonstrates the propensity of SOD1 and Hsp27 to interact. We also checked if the over-expressed hHsp27 was able to form oligomers. Similar to the observation by Hollander et al. (2004) the over-expressed hHsp27 from the different groups migrated almost exclusively as oligomers on native gels (Fig. 6b). To precisely size these oligomers gel filtration chromatography was used. Spinal cord eluates from the column were tested on a slot blot and stained for Hsp27. After comparing with standards run prior on the column, we observed that in physiological conditions, the majority of soluble hHsp27 eluted around 64 kDa (Fig. 6c). The corresponding fractions on the slot blot stained for Hsp27 from both SOD1G93A and SOD1G93A/hHsp27 mice are shown in Fig. 6d.

 2008 The Authors Journal Compilation  2008 International Society for Neurochemistry, J. Neurochem. (2008) 106, 2170–2183

Hsp27 and amyotrophic lateral sclerosis | 2177

Fig. 4 Effect of hHsp27 over-expression on disease onset and survival. (a) Absolute grip strength measurements (in gm) of SOD1G93A/ hHsp27 and SOD1G93A mice as tested on a dynamometer from day 40 until death, obtained after averaging five trials for each mice tested twice weekly (n = 10 in each group; p = 0.9, repeated measures G93A ANOVA). (b) Average rotarod performance of SOD1 /hHsp27 and SOD1G93A mice. The mice were tested for rotarod performance twice weekly starting from 85 days until they could no longer stay on it. The average performance shown here is until the time point when the first

Discussion The stress inducible, ATP-independent Hsp27 can be protective against neuronal death (Wagstaff et al. 1999; Latchman 2005). In vitro models of Parkinson’s and Huntington’s disease, caused by mutated forms of a-synuclein and huntingtin respectively, have demonstrated the higher neuronal benefits of Hsp27 in comparison to the other stress inducible chaperone, Hsp70 (Wyttenbach et al. 2002; Zourlidou et al. 2004). Hsp27 has been reported to be important for neuronal survival in both the CNS and the PNS (Costigan et al. 1998; Lewis et al. 1999; Wagstaff et al. 1999; Benn et al. 2002; Kalmar et al. 2002; Kalwy et al. 2003; Dodge et al. 2006). Long-term motor and muscle

animal in the study died (day 131) and was obtained after comparing the values to the learning period (60–85 days) which was normalized to 100% (n = 10 in each group). (c) Kaplan–Meier curve showing average disease onset in both the SOD1G93A/hHsp27 and the SOD1G93A cohort (n = 10 in each group; p = 0.9, log-rank analysis). (d) Kaplan–Meier curve demonstrating the probability of survival in the SOD1G93A/hHsp27 and the SOD1G93A cohort (n = 10 in each group; p = 0.09, log-rank analysis).

recovery have also been attributed to the protective role of Hsp27 in motor neurons suggesting that sustained Hsp27 availability can help in repairing adult mammalian brain (Sharp et al. 2006). In ALS models, mutant but not wild-type SOD1 protein was shown to interact with several Hsp’s, notably Hsp27, aB crystallin, Hsp40 and Hsp70 in N2a and NIH3T3 cells (Bruening et al. 1999; Shinder et al. 2001; Okado-Matsumoto and Fridovich 2002; Takeuchi et al. 2002). Although cell culture-based models demonstrated a protective effect of Hsp27 (Patel et al. 2005), no direct evidence exists as to the exact potential of Hsp27 in attenuating disease in the ALS mice. Decrease in Hsp27 levels in motor neurons preceding disease onset suggests that the availability of Hsp27 during

 2008 The Authors Journal Compilation  2008 International Society for Neurochemistry, J. Neurochem. (2008) 106, 2170–2183

2178 | J. Krishnan et al.

progression of ALS may be one of the limiting factors, thus making it an eligible candidate to investigate (Maatkamp et al. 2004; Wang et al. 2008). Also, since motor neurons in culture lack the ability to activate HSF1 and mount a stress response (Batulan et al. 2003), it is unlikely that motor

(b) 100

G93A

Motor neurons per ventral horn

(a)

neurons can up-regulate Hsp27 to counteract the toxicity of mutant SOD1. Hence, it is plausible that the motor neurons lacking Hsp27 in the pre-symptomatic and symptomatic phase of the SOD1G93A mice could be the ones that die first because of the toxicity of mutant SOD1. Thus, supplement-

G93A/hHsp27

G93A/hHsp27

90

G93A

80 70 60 50 40 30 20 10 0 400

Area of motor neuron (µm2)

sp hH A/

3A

93 G

G

Symp

G9

hH 93

A/

3A G9

93 G

27

27 sp

27 sp hH A/

A 93 G

Pre-symp

End-stage

Density of staining (arbitrary units)

(d)

(c)

9 8

G93A

1

Symp

End-stage

*

*

7 6 5 4 3 2 1 0

G93A

G93A/hHsp27

G93A/hHsp27

2

3

(f)

4

1.4

Cas3

hHsp27

Ntg

G93A/hHsp27

G93A

1.2

Cas9

β − actin Pre-symp 1

2

3

4

Cas3 Cas9

β − actin

O.D (360 nm)

(e)

Pre-symp

1.0 0.8 0.6 0.4 0.2 0 6

End-stage

11

16

20

Time (min)  2008 The Authors Journal Compilation  2008 International Society for Neurochemistry, J. Neurochem. (2008) 106, 2170–2183

Hsp27 and amyotrophic lateral sclerosis | 2179

(a)

(c)

(b)

(d)

Fig. 6 Co-immunoprecipitation and oligomerization pattern of Hsp27 (a) Co-immunoprecipitation of spinal cord extracts from pre-symptomatic and symptomatic mice with antibodies against Hsp27 (+Hsp27 Ab lane) and SOD1 (+SOD1 Ab lane). Spinal cord sample incubated with only the beads was used to check non-specific binding to the beads (- Ab lane), while the extract itself is added as a positive control. Note that Hsp27 already present in the pre-symptomatic SOD1G93A/ hHsp27 mice is pulled down by SOD1 antibody (*). (b) Native gel analysis of spinal cord homogenates from hHsp27, SOD1G93A and

SOD1G93A/hHsp27 mice stained for Hsp27. (c) Spinal cord extracts from SOD1G93A and SOD1G93A/hHsp27 mice were chromatographed on gel filtration column (see Materials and methods) and the oddnumbered eluates were tested on a slot blot for Hsp27 elution pattern. Quantification of the intensity of staining from these blots is shown along with the position of molecular weight standards eluted prior on the same column. (d) Fractions 7–17 show no elution of Hsp27 in both SOD1G93A and SOD1G93A/hHsp27 samples, while fractions 61–71 in the SOD1G93A/hHsp27 sample stain intensely for Hsp27.

Fig. 5 Effect of hHsp27 over-expression on histology, aggregation and chaperoning ability (a) Representative micrographs of paraffinembedded spinal lumbar sections from symptomatic SOD1G93A/ hHsp27 and SOD1G93A mice (100–120 days) stained with H&E (Scale bar: 50 lm). (b) Graph showing quantification of the number of motor neurons in the lumbar spinal cord from SOD1G93A/hHsp27 and the SOD1G93A mice (n = 3 in each group). No differences were found in the numbers of either small ( 400 lm) motor neurons. (c) Cellulose acetate filter trap assay of spinal cord homogenates from different age groups of SOD1G93A and SOD1G93A/hHsp27mice stained for SOD1. (d) Quantification of the intensities of aggregates filter trapped from different age groups of SOD1G93A and SOD1G93A/

hHsp27mice (n = 5, *p < 0.05 when compared to pre-symptomatic or symptomatic ages). Average intensity of staining from SOD1WT mice from different ages was