Retinoid signaling alterations in amyotrophic lateral sclerosis

Abstract

Amyotrophic lateral sclerosis (ALS) is a fatal neuromuscular disease for which effective therapeutic interventions and an understanding of underlying disease mechanism are lacking. A variety of biochemical pathways are believed to contribute to the pathophysiology of ALS that are common to both sporadic and familial forms of the disease. Evidence from both human and animal studies indicates that expression of retinoid signaling genes is altered in ALS and may contribute to motor neuron loss. Our goals were to examine the expression and distribution of proteins of the retinoid signaling pathway in spinal cord samples from patients with sporadic and familial ALS and to evaluate the role of these proteins in motor neuron cell survival. In sporadic ALS, the cytoplasmic binding protein that facilitates nuclear translocation of retinoic acid, cellular retinoic acid binding protein-II (CRABP-II), was localized to the nucleus and retinoic acid receptor β (RARβ) was significantly increased in motor neuron nuclei when compared to either familial ALS patients or non-neurologic disease controls. Motor neurons with increased nuclear RARβ were negative for markers of apoptosis. Pre-treatment of primary motor neuron-enriched cultures with a pan-RAR or RARβ-specific agonist decreased motor neuron cell death associated with oxidative injury/stress while a RARβ-specific antagonist enhanced cell death. Our data suggest retinoid signaling is altered in ALS and increased nuclear RARβ occurs in motor neurons of sporadic ALS patients. Activation of RARβ protects motor neurons from oxidative-induced cell death.

Introduction

Amyotrophic lateral sclerosis (ALS), the most common adult onset motor neuron disease, is a progressive, fatal neurodegenerative disorder which is poorly understood and for which effective therapeutic interventions are lacking. The prevalence of ALS is approximately 4-6 cases per 100,000, the average age at disease onset is approximately 55 years and the majority of cases (over 90%) are idiopathic in nature. For familial forms of ALS, the list of associated genes continues to grow [reviewed in [1]; however, no mechanistic differences between sporadic and familial forms have been identified to date.

The clinical manifestations of ALS include both upper motor neuron (enhanced and pathological reflexes and spasticity) and lower motor neuron (focal and multifocal muscle weakness and atrophy, fasciculations and cramps) involvement [2]. A variety of processes (i.e., oxidative stress, mitochondrial dysfunction, abnormal axonal transport and protein aggregation) are believed to play a role in motor neuron degeneration although it remains unclear how each contributes to the pathogenesis of ALS.

Retinoic acid (RA), a member of the retinoid family of lipids and mediator of vitamin A activity, is an essential morphogen in vertebrate development and is critical for neural development/specification and neural plasticity and regeneration [3,4]. Studies also suggest a role for this molecule in adulthood with evidence that retinoid signaling can promote regeneration of the adult rodent nervous system [5-7] or impact the process of neurodegeneration [8,9]. In addition, RA has been shown to modulate the antioxidant defense system and superoxide dismutase-1 (SOD1) gene expression [10,11] further implicating a role for RA in ALS.

Prior studies have demonstrated that genes for retinoid pathway proteins and genes regulated by retinoid signaling are differentially expressed in post-mortem tissues from ALS patients [12-16]. Moreover, in transgenic animal models of familial ALS (FALS), spinal cord gene expression profiling has revealed altered expression of genes of the retinoid signaling pathway [17]. A study in the mutant G93A SOD1 transgenic rat model of ALS indicates that changes in the retinoid receptor expression profile occur in the spinal cord of early pre-symptomatic and terminal stage disease both at the RNA and protein levels [18]. Furthermore, Corcoran and colleagues observed astrocytosis, accumulation of neurofilaments and motor neuron loss in the lumbar spinal cord of vitamin A deficient animals [19]. As systemic vitamin A deficiency in patients with ALS has not been observed [20], it remains unclear how this signaling pathway plays a role in this disease.

Within the cell, the level of free versus bound retinol and RA are regulated by cellular retinol-binding protein (CRBPI) and cellular retinoic acid-binding proteins (CRABP-I and II), respectively [21]. CRBPI binds retinol in the cytoplasm and promotes its metabolism to retinaldehyde [22] and esterification to the retinyl ester for storage [23]. CRABP-I has been suggested to promote the catabolism of RA [24] whereas CRABP-II is thought to mediate the transport of RA to the nucleus for association with its receptors [25]. RA mediates its effects on gene transcription through the retinoic acid receptors (RARs) and the retinoid X receptors (RXRs) [26,27]. Each receptor has three isotypes (α , β and γ) with multiple isoforms of each subtype generated by alternative splicing and differential promoter usage [26]. RARs are activated by either all-trans retinoic acid (ATRA) or 9-cis-retinoic acid (9-cis-RA) and mediate gene expression by forming heterodimers with RXRs whereas RXRs are activated only by 9-cis-RA and modulate gene expression either as homodimers or by forming heterodimers with RARs or a variety of orphan receptors [28-30]. In the absence of ligand, RA nuclear receptors function as transcriptional repressors [4].

The goal of this study was to investigate the role of proteins involved in the RA signaling pathway in ALS. Our overall hypothesis was that retinoid signaling is altered in ALS and thereby contributes to motor neuron cell death. We analyzed post-mortem lumbar spinal cord tissue from individuals with sporadic ALS (SALS), FALS, and non-neurologic disease controls to determine the levels and subcellular localization of proteins involved in this pathway. We observed increased nuclear localization for both CRABP-II and RARβ in motor neurons of patients with SALS. In primary motor neuron-enriched cultures, pan-RAR or RARβ-specific agonists were neuroprotective indicating that signaling through RARβ can promote motor neuron cell survival. These data indicate a role for retinoid signaling in ALS and support further investigation of this signaling pathway as a therapeutic opportunity in ALS.

Materials and methods

Subjects

Lumbar spinal cord tissue was obtained from the ALS Tissue Bank at the University of Pittsburgh Medical Center (RB, Director). Approval was obtained from the University of Pittsburgh Institutional Review Board. In selecting cases for this study, efforts were made to control for age, gender and post-mortem interval with cases with available frozen spinal cord tissue whenever possible. SALS subjects (n = 20) included in this study were clinically diagnosed using the revised El Escorial criteria. FALS patients (n = 4) had a family history of ALS. Two of the SALS and one of the FALS cases harbor C9ORF72 repeat expansions (Table 1), as determined using the repeat-primed polymerase chain reaction (PCR) method previously described [31]. The genetic cause of the remaining FALS cases remains undefined. Control subjects (n = 9) lacked clinical or neuropathological evidence of neurologic disease. The age range for all individuals was 40-95 years old. The average age at death was 60.25 ± 12.66 years for SALS (range, 40 to 82 years) and was not significantly different from FALS (52.75 ± 8.02 years; range, 45 to 63 years; p = 0.27) or control subjects (64.78 ± 15.60 years; range, 51 to 95 years; p = 0.41). The average post-mortem interval time for SALS subjects was 5.50 ± 2.12 hours (range, 2 to 11 hours) and not significantly different from FALS (6.50 ± 5.07 hours; range 3 to 14 hours; p = 0.51) or control subjects (7.89 ± 5.71 hours; range 2 to 20 hours; p = 0.11).

Table 1.

Cases utilized for this study

	Age	Gender	PMI (hours)	Percentage of Motor Neurons with Nuclear CRABP-II Localization	Percentage of Motor Neurons with Increased RARβ Nuclear Localization
SALS1	45	M	11	40% (4 of 10)	55% (6 of 11)
SALS2	51	M	7	50% (4 of 8)	60% (6 of 10)
SALS3	45	M	7	82% (9 of 11)	64% (7 of 11)
SALS4	62	M	7	70% (7 of 10)	14% (1 of 7)
SALS5	73	M	6	82% (9 of 11)	55% (6 of 11)
SALS6	80	F	4	64% (9 of 14)	88% (22 of 25)
SALS7	82	M	5	64% (9 of 14)	73% (22 of 30)
SALS8	72	F	9	50% (4 of 8)	70% (7 of 10)
SALS9	56	M	3	63% (5 of 8)	62% (8 of 13)
SALS10	63	F	5	67% (6 of 9)	18% (2 of 11)
SALS11	59	M	4	67% (8 of 12)	33% (5 of 15)
SALS12*	43	M	6	70% (7 of 10)	86% (6 of 7)
SALS13*	67	M	4	41% (7 of 17)	13% (2 of 16)
SALS14	51	M	4	78% (7 of 9)	58% (7 of 12)
SALS15	40	M	6	38% (3 of 8)	44% (4 of 9)
SALS16	77	F	4	89% (8 of 9)	50% (5 of 10)
SALS17	69	M	5	80% (8 of 10)	56% (5 of 9)
SALS18	60	F	2	70% (7 of 10)	27% (3 of 11)
SALS19	50	F	7	88% (7 of 8)	73% (8 of 11)
SALS20	60	F	4	78% (7 of 9)	50% (5 of 10)
FALS1	45	M	3	17% (2 of 12)	8% (1 of 12)
FALS2	48	F	5	17% (2 of 12)	20% (2 of 10)
FALS3	55	M	14	56% (5 of 9)	20% (2 of 10)
FALS4*	63	F	4	50% (4 of 8)	20% (2 of 10)
CON1	54	M	6	26% (6 of 23)	13% (2 of 15)
CON2	51	F	5	38% (3 of 8)	18% (2 of 11)
CON3	76	M	13	31% (4 of 13)	29% (9 of 31)
CON4	57	M	2	25% (3 of 12)	4% (2 of 47)
CON5	58	F	5	73% (8 of 11)	25% (6 of 24)
CON6	82	F	5	11% (1 of 9)	19% (4 of 21)
CON7	57	F	11	ND	21% (7 of 34)
CON8	95	M	20	50% (5 of 10)	0% (0 of 7)
CON9	53	F	4	20% (2 of 10)	6% (1 of 16)

The age and post-mortem interval (PMI) in hours (average SALS 5.50 hours; average FALS 6.5 hours; average control 7.89 hours) are indicated for the SALS (60.25 years), FALS (52.75 years) and control (64.78 years) cases. 17 of 20 SALS cases exhibited increased nuclear localization of RARβ as compared to controls and FALS.

^*signifies presence of the C9ORF72 repeat expansion within the subject.

ND = not determined

Antibodies

Monoclonal antibodies were used to detect CRBPI (G4E4; Santa Cruz Biotechnology, Santa Cruz, CA), CRABP-I (C-1; Novus Biologicals, Littleton, CO), RARα (763; Chemicon International, Temecula, CA) and RARγ (1371; Chemicon International). Polyclonal antibodies were used to detect CRABP-II (K-13; Santa Cruz Biotechnology) and RARβ (Chemicon International). For immunohistochemistry, these antibodies were used at dilutions of 1:50 (RARα and RARβ), 1:100 (CRABP-I and RARγ), 1:250 (CRBPI) and 1:2500 (CRABP-II). Western blots were probed with the antibodies listed above at 1:500 (CRABP-II) or alternate polyclonal antibodies to detect RARα (C-20; Santa Cruz Biotechnology) and RARβ (C-19; Santa Cruz Biotechnology) used at a dilution of 1:500.

Tissue homogenates and protein extraction

Lumbar spinal cord tissue homogenates were prepared from a subset of patients (9 SALS; 4 FALS; 5 control) for co-immunoprecipitation studies. The age range and post-mortem interval for this patient subgroup did not differ from that of all subjects nor across the subgroups.

Spinal cord frozen tissue samples from control and both sporadic and familial ALS cases were homogenized and analyzed as previously described [32]. For total cell lysates, spinal cord tissue samples were homogenized using an Omni Tissue Homogenizer (Omni International, Marietta, GA) set at 15,000 rpm for 45 seconds in lysis buffer containing 25 mmol/L HEPES (pH 7.4), 50 mmol/L NaCl, protease inhibitor cocktail II (Sigma, St. Louis, MO) and 1% Triton X-100. The homogenized product was spun at 14,000 rpm at 4°C and the supernatant saved as the total cell lysate. Nuclear and post-nuclear extracts were prepared as described previously [33]. Following homogenization with a Potter-Elvehjem grinder (Omni International) in buffer containing 10 mmol/L Tris (pH 8.0), 10 mmol/L MgCl₂, 15 mmol/L NaCl, 0.5 mmol/L phenylmethyl sulfonyl fluoride (PMSF), 2 μg/mL pepstatin A and 1 μg/mL leupeptin, nuclei were collected via low-speed centrifugation at 800g for 5 minutes. The resulting supernatant was saved as the post-nuclear extract and the nuclei further extracted with high-salt buffer (20% glycerol, 20 mmol/L HEPES (pH 7.9), 0.42 mol/L NaCl, 0.1% Nonidet P-40, 0.5 mmol/L PMSF, 2 μg/mL pepstatin A and 1 μg/mL leupeptin) on ice for 10 minutes. Remaining insoluble material was removed via centrifugation at 14,000 rpm for 5 minutes. The resulting supernatant fraction was collected as the nuclear-enriched fraction. Protein concentrations were determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Rockford, IL) per the manufacturer’s instructions.

Immunohistochemistry and TUNEL staining

Paraffin-embedded tissue sections (8 μM) were deparaffinized and a hydrogen peroxide/methanol mixture used to block endogenous peroxidase activity. Sections were steamed for 20 minutes in Antigen Retrieval Citra (pH 6.0) (BioGenex, San Ramon, CA) or Target Retrieval Solution (pH 9.0) (Dako North America, Carpinteria, CA). After washing, blocking was performed with Protein Blocking Agent (Thermo Fisher Scientific) or Power Block Universal Blocking Reagent (BioGenex, San Ramon, CA). Primary antibodies were diluted in phosphate-buffered saline (PBS) or Super Block (ScyTek Laboratories, Logan, Utah) and incubated for 1 hour at room temperature in a humidified chamber. After washing, sections were incubated with the appropriate biotinylated secondary antibody (1:200 dilution; Vector Laboratories, Burlingame, CA) for 30 minutes at room temperature in a humidified chamber. The signal was amplified using Vectastain ABC Reagent (Vector Laboratories) according to the manufacturer’s protocol. NovaRED (Vector Laboratories) served as the chromagen and all sections were counterstained with hematoxylin.

For terminal deoxynucleotidyl transferase (TdT)-mediated dUTP-biotin nick end-labeling (TUNEL), sections (4 μM) were deparaffinized, treated with proteinase K and endogenous peroxidase activity blocked with a hydrogen peroxide/methanol mixture. TdT enzyme was then added for 25 minutes at 37°C. Anti-digoxigenin conjugate (rhodamin) was added for 30 minutes at room temperature. After washing, sections were incubated in 3,3'-diaminobenzidine (DAB) solution and counterstained with hematoxylin.

Motor neurons located within the ventral horn of the spinal cord were counted for each case by two independent investigators. The number of motor neurons with nuclear CRABP-II or RARβ was reported as a percentage of the total number of motor neurons counted for each case (Table 1) and by grouping (i.e., control, SALS or FALS). Only motor neurons for which the nucleus could be observed were counted. One slide was counted for each case although consistent staining patterns were observed over multiple slides. SALS cases (n = 6) with demonstrated nuclear RARβ were utilized to evaluate the co-localization of nuclear RARβ and TUNEL staining.

Protein-protein interactions

Co-immunoprecipitation (co-IP) of nuclear-enriched fractions was performed using the MultiMACS Protein A/G Kit and μMACS Protein A/G MicroBeads (Miltenyi Biotech, Inc., Auburn, CA). Lysates were incubated with primary antibody for two hours at 4°C with gentle rocking. Protein A/G MicroBeads were then added to the immune complex. After washing, extracts were added and the magnetically labeled protein retained on the column. The eluted immunoprecipitates were analyzed by immunoblot as described below.

Immunoblot

Immunoprecipitated proteins from nuclear-enriched fractions were fractionated on 12% gels by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Proteins were transferred to a polyvinylidene difluoride (PVDF) membrane (PerkinElmer, Boston, MA) and blocked with 5% non-fat milk/Tris-buffered saline with Tween-20 (TBST). Blots were probed with primary antibody overnight at 4°C in 5% milk/TBST and then washed with TBST. Horseradish peroxidase-conjugated secondary antibodies specific for each primary antibody were added for 2-3 hours at room temperature after which blots were washed with TBST. The final reaction products were visualized using enhanced chemiluminescence (PerkinElmer) and the density of bands measured using NIH ImageJ software (National Institutes of Health).

Motor neuron-enriched primary cultures

Motor neurons were isolated from the spinal cords of Sasco Sprague-Dawley rat embryos at gestation day 14 (E14) as previously described [34-36]. Briefly, following removal of the spinal cords, tissue was trypsinized and the dissociated cells spun over a metrizamide density gradient. Cells above the metrizamide layer (larger, dense motor neurons) were collected and motor neurons plated in 24-well culture trays coated with poly-D lysine (PDL) and then human placental laminin (HPL) at a density of approximately 12,000 cells/well in neurobasal base media (Invitrogen, Carlsbad, CA) supplemented with B27 (variations +/- RA) (Invitrogen), BDNF, CNTF and GDNF (Peprotech, Rocky Hill, NJ), and as described [34,35]. Cultures were incubated at 37°C at 5% CO₂ and half of the culture supernatant exchanged with fresh medium after 24 hours and every 2 to 3 days thereafter. For RNA extractions, cultures were treated with the Qiagen RNeasy Kit (Valencia, CA) per the manufacturer’s instructions. RNA (1.0 μg) was reverse-transcribed to complementary DNA (cDNA) using SuperScript III reverse transcriptase (Invitrogen) prior to analysis by real-time PCR. Products were resolved on 1.5% agarose gels and visualized with ethidium bromide.

RA pathway modulators

All-trans retinoic acid (ATRA) was obtained from Sigma (St. Louis, MO). Adapalene (RARβ and RARγ agonist) and LE-135 (RARβ antagonist) were obtained from Tocris Bioscience (Ellisville, MO). Stock solutions were prepared in dimethyl sulfoxide (DMSO) such that a constant concentration of 0.1% DMSO was present for all treatment groups. Motor neuron-enriched cultures were pre-treated with these agents for approximately 1 hour prior to the addition of hydrogen peroxide (H₂O₂).

Cell viability/live cell imaging

After 24 hours in culture, motor neurons were treated with the RA pathway modulators (ATRA final concentrations of 1.0, 10 or 100 μM; adapalene final concentrations of 0.25, 2.5 or 25 μM; LE-135 final concentrations of 1.0, 10 or 100 μM) in combination with 400 μM H₂O₂. Propidium iodide (PI) was added at a final concentration of 1 μL/500 μL medium. Cells were kept in 24-well plates in a previously described time-lapsed microscopic imaging system (Automated Cell, Pittsburgh, PA) where bright-field and fluorescent images were acquired every 10 minutes [37]. These images were analyzed using NIH Image J software and the number of cells in the bright-field and PI-positive cells quantified at 0, 4, 6, 12, 18, 24, 36, 48, 60 and 72 hour time points. The percentage of cell death at these time points was then determined for each treatment. For a typical experiment, each treatment was added to three wells and three separate fields averaging 10-20 cells each were counted per well. The percentage of cell death was determined for each field by binning at the indicated time points, and each experiment repeated at least twice.

Statistical methods

Comparisons between any two groups of data were accomplished using the unpaired t test at the 95% confidence interval. Multi-group comparisons (SALS vs. FALS vs. controls) were performed using ANOVA followed by the Tukey post-test. Comparisons between treatment groups were accomplished using two-way ANOVA followed by Bonferroni post-tests. A p value ≤ 0.05 was considered statistically significant.

Results

Subcellular distribution of CRABP-II is altered in SALS spinal cord motor neurons

We first examined the localization of cytoplasmic retinol and retinoic acid (RA) binding proteins in the lumbar spinal cord of ALS and control subjects using commercially available antibodies. CRBPI, the cytoplasmic protein that mediates metabolism and storage, was not detected in spinal cord motor neurons of SALS, FALS or control subjects (Figure 1A, A-C). CRABP-I was noted in all three subject groups and the immunostaining pattern and intensity were not significantly different across subject groups (Figure 1A, D-F). However, CRABP-II, the protein responsible for mediating nuclear translocation and facilitating nuclear receptor interactions, exhibited enhanced nuclear and perinuclear immunoreactivity in spinal cord motor neurons of SALS cases (Figure 1A, H). CRABP-II was observed in a more diffuse, cytoplasmic pattern in both FALS and control subjects (Figure 1A, G and I). When morphologically distinct motor neurons from multiple lumbar spinal cord sections of SALS, FALS and control cases were counted, the percentage of motor neurons with nuclear staining for CRABP-II was 66% (135 of 205) for SALS cases, 32% (13 of 41) for FALS cases and 33% (32 of 96) for control cases. These differences were statistically significant between and across subject groups (Figure 1B). The percentage of motor neurons with punctate nuclear CRABP-II for each individual subject is shown in Table 1.

Figure 1.

Immunohistochemical analysis of cytoplasmic binding proteins of the retinoic acid signaling pathway in human spinal cord tissue. A: Lumbar spinal cord sections were immunostained for CRBPI (A-C), CRABP-I (D-F) and CRABP-II (G-I) in red and counterstained with hematoxylin. Insets represent a high power magnification of each panel. Original magnifications: 200X (A to I); 400X (insets in A to I). B: The localization of CRABP-II was assessed for morphologically distinct motor neurons from each patient group. The percent of motor neurons with nuclear CRABP-II was significantly greater in patients with SALS (white bars) as compared to FALS (black bars; p < 0.01) or controls (stippled bars; p < 0.001). There was no significant difference between FALS patients and controls (p > 0.05).

RARβ immunoreactivity is significantly increased in motor neuron nuclei of SALS patients and does not correlate with TUNEL staining

We next examined the levels and localization of the three retinoic acid receptor (RAR) isotypes (RARα, RARβ and RARγ) in lumbar spinal cord motor neurons as these receptors are activated by the predominant RA isoform, all-trans RA. RARα was observed in motor neurons of all ALS cases as well as controls and was localized predominantly to the cytoplasm (Figure 2A, A-C). The level of this protein did not differ significantly between motor neurons of the three patient groups but was noted in distal processes of motor neurons of SALS and, to a lesser extent, control subjects. For RARβ, there was a significant increase in the immunostaining of motor neuron nuclei in SALS subjects (Figure 2A, E). In FALS and non-neurologic disease controls, RARβ exhibited a predominantly cytoplasmic distribution in lumbar spinal cord motor neurons (Figure 2A, D and F). The percentage of motor neurons with nuclear RARβ immunostaining was 55% (137 of 249) for SALS cases, 17% (7 of 42) for FALS cases and 16% (33 of 206) for control cases (Figure 2B). This difference in nuclear RARβ was statistically significant when motor neurons from SALS were compared to FALS patients (p < 0.01) and controls (p < 0.001) (Figure 2B). The percentage of motor neurons with increased nuclear RARβ per subject is shown in Table 1. We did not detect any correlation of CRABPII or RARβ distribution to the presence of C9ORF72 repeat expansions that were seen in three patients (Table 1). RARγ was not detected in spinal cord motor neurons of SALS, FALS or control subjects (Figure 2A, G-I).

Figure 2.

Immunohistochemical analysis of retinoic acid receptor (RAR) proteins in human spinal cord tissue. A: Lumbar spinal cord sections were immunostained for RARα (A-C), RARβ (D-F) and RARγ (G-I) in red and counterstained with hematoxylin. Insets represent a high power magnification of each panel. Original magnifications: 200X (A to I); 400X (insets in A to I). B: Nuclear RARβ immunoreactivity was determined by counting morphologically distinct motor neurons from each patient group. The percent of motor neurons with increased nuclear RARβ was significantly greater in patients with SALS (white bars) when compared to FALS (black bars; p < 0.01) or controls (stippled bars; p < 0.001). There was no significant difference between FALS patients and controls (p > 0.05).

RARβ has been implicated in molecular mechanisms regulating cell death in multiple cell types [38-40]. Conversely, this RAR isotype has been shown to mediate both neurotrophic and neuritogenic effects of retinoic acid [41]. To determine whether this response was apoptotic or pro-survival, we used immunohistochemistry to evaluate the co-localization of TUNEL with RARβ in motor neurons. Serial sections of lumbar spinal cord from six SALS patients were immunostained for RARβ and TUNEL, and the same motor neurons evaluated from both sections as denoted by tissue landmarks. We did not observe TUNEL staining in motor neurons with RARβ nuclear localization (Figure 3A) but did observe TUNEL-positive nuclei in motor neurons lacking nuclear RARβ (representative image in Figure 3B). Overall, 79% (26 of 33) of motor neurons with nuclear RARβ immunostaining were TUNEL negative across multiple SALS subjects.

Figure 3.

Immunohistochemical analysis of RARβ and TUNEL in human spinal cord. Serial lumbar spinal cord sections from SALS cases were immunostained for RARβ and TUNEL and representative cases are shown. Identical motor neurons were imaged as determined by anatomical hallmarks and location within the tissue. A: Motor neurons with intense nuclear RARβ immunostaining (A) lacked TUNEL staining (B). B: Motor neurons within the spinal cord without nuclear RARβ immunostaining (A) were TUNEL positive (B). Insets represent a high power magnification of each panel. Original magnifications: 200X; 400X (insets).

Protein interactions between CRABP-II and the nuclear receptors do not differ between ALS and control groups

We next examined protein-protein interactions between CRABP-II and either RARα or RARβ to determine the functional consequences of CRABP-II localization to the nucleus. We immunoprecipitated CRABP-II from nuclear-enriched fractions and determined whether the nuclear receptors co-immunoprecipitated with CRABP-II. We observed an interaction between CRABP-II and both RARα (Figure 4A) and RARβ (Figure 4B) although there was no significant difference between ALS and control groups (Figure 4C; p = 0.9042 for RARα in ALS versus controls; p = 0.8881 for RARβ in ALS versus controls).

Figure 4.

Protein-protein interactions in nuclear-enriched fractions from human lumbar spinal cord. Protein extracts from ALS and controls were prepared as described in the Materials and Methods. Protein (500 μg) from each extract was immunoprecipitated with CRABP-II and resulting blots were probed with antibodies specific to RARα (A) or RARβ (B). A: Immunoblots of nuclear-enriched fractions from lumbar spinal cord co-immunoprecipitated with CRABP-II and probed for RARα. B: Immunoblots of nuclear-enriched fractions from lumbar spinal cord coimmunoprecipitated with CRABP-II and probed for RARβ. C: Graph representing the results from densitometry quantification. RAR Receptor levels were normalized to that of CRABP-II for each sample. Black bars represent control cases (n = 2) and checkered bars are ALS cases (n = 6). Statistical analyses were performed using an unpaired t-test with a 95% confidence interval. The p values for RARα and RARβ were 0.9042 and 0.8881, respectively. Letters and numbers correspond to patients in Table 1.

Motor neuron-enriched cell cultures and retinoic acid treatment

To further explore the functional role of RAR signaling in motor neurons, we utilized a primary motor neuron-enriched culture system. The procedure for purifying spinal motor neurons was modified from that of Camu and Henderson [42,43] and utilized a density gradient to separate low-density cells from the spinal cord cell suspension. PCR was used to determine which of the RARs were expressed in this in vitro system. Using the primers listed in Table 2, we detected expression of RARα and RARβ but not RARγ (Figure 5).

Table 2.

Primer sets for PCR analysis

	Forward Primer	Reverse Primer
RARα	5’-GGCTACCACTATGGGGTTAGCGC-3’	5’-GGTTCCGGGTCACCTTGTTGAT-3’
RARβ	5’-AGTCATCGGGCTACCACTATGGC-3’	5’-AGCACTTCTGCAGGCGGCA-3’
RARγ	5’-CATTTGAGATGCTGAGCCCCA-3’	5’-CATACAAAGCACGGCTTATAGACCC-3’
Cyclophilin	5’-CCCCACCGTGTTCTTCGACA-3’	5’-CCCAGTGCTCAGAGCACGAAA-3’

Figure 5.

In vitro expression of the RARs. Expression of the three RAR isotypes was evaluated in the motor neuron-enriched cell culture system. Polymerase chain reaction (PCR) analysis of RNA isolated from primary motor neuron cultures after 24 hours in culture. RARα and RARβ but not RARγ were expressed. “MN” denotes RNA from primary motor neurons; “+” denotes RNA isolated from JM2 cells used as a positive control. Cyclophilin was used as a PCR positive control for each cell type.

As vitamin A deprivation has been characterized by increased reactive oxygen species (ROS) formation and oxidative stress has been implicated in ALS pathogenesis [reviewed in [44]], H₂O₂ was used as a model of oxidative stress/injury to determine the effect of stimulating the RARs on motor neuron cell survival. After 24 hours in culture, cells were treated with increasing concentrations of H₂O₂ to establish a dose-response curve (data not shown), and the concentration used for subsequent experiments (400 μM H₂O₂) resulted in approximately 50% cell death after 24 hours.

Eliminating RA from the media reduced cell viability at the 24 hour time point and thereafter (Figure 6A) as assessed by live cell imaging/propidium iodide staining. The presence of RA increased motor neuron survival in the presence of H₂O₂ as significant protection was observed as early as 6 hours post-treatment (Figure 6B).

Figure 6.

The presence of RA modulates cell death. The percentage of motor neuron cell death was evaluated using live cell imaging and propidium iodide staining over time. A: The absence of RA in the B27 media supplement (blue curve) resulted in significantly increased cell death in the absence of any insult. B: When treated with H₂O₂ to model oxidative stress/injury, the absence of RA in the B27 media supplement (blue curve) resulted in significantly increased percentages of cell death beginning at the 6.0 hour time point. C: Motor neuron-enriched cultures were pre-treated with ATRA for 1.0 hour prior to addition of H₂O₂ to induce cell death. In the presence of H₂O₂, ATRA significantly decreased the percentage of cell death as compared to the control treatment (DMSO + H₂O₂). Control treatments without H₂O₂ (ATRA or DMSO alone) were included for comparison (orange and black curves). Error bars represent the mean percentage of cell death ± standard error of the mean from at least two separate experiments. *p < 0.05; **p < 0.01; ***p < 0.001.

Stimulating RAR-mediated signaling delays oxidative-induced cell death

The ability of ATRA, a pan-RAR agonist, to protect motor neurons from H₂O₂-induced cell death was evaluated. When cells were pretreated with 10 μM ATRA, the percentage of cell death was significantly decreased as compared to control-treated cells (Figure 6C). The localization of RARα and RARβ during ATRA treatment was also assessed, although no difference in RARα or RARβ protein localization was observed in the presence of ATRA (data not shown).

Specific agonists and antagonists of RARβ were then used to determine if modulating this specific nuclear receptor could recapitulate the protective effects observed with ATRA. When pretreated for 1.0 hour with adapalene (RARβ agonist) prior to the addition of H₂O₂, the percentage of motor neuron cell death was significantly decreased as compared to the control treatment (Figure 7A). However, when motor neuron-enriched cultures were pre-treated for 1.0 hour with a RARβ-selective antagonist (LE-135), the percentage of cell death was significantly increased over control treatment (Figure 7B).

Figure 7.

RARβ activity regulates H₂O₂ induced cell death. Motor neuron cultures were pre-treated for 1.0 hour with agents that specifically target RARβ prior to the addition of toxin. A: In the presence of adapalene, a RARβ agonist, the percentage of cell death induced by H₂O₂ was significantly reduced when compared to the control treatment of DMSO plus H₂O₂ (compare blue and green curves). Controls of DMSO alone or adapalene alone measured baseline level of cell death over time (black and orange curves). B: Pre-treatment of cells with a combination of ATRA and LE-135, a selective RARβ antagonist, eliminated the protective effects of ATRA and exacerbated the toxicity in the presence or absence of H₂O₂ (green and orange curves). Control treatments of DMSO alone (black curve) or DMSO plus H₂O₂ (blue curve) are shown. Error bars represent the mean percentage of cell death ± standard error of the mean from at least two separate experiments. *p < 0.05; ***p < 0.001.

Discussion

The retinoid signaling pathway is critical for neuronal development with increasing evidence that it has equally important functions in adulthood. Our prior studies identified altered post-translational modifications of transthyretin in the CSF of ALS patients [45,46], suggesting that regulated delivery of retinol throughout the CNS may be abnormal in ALS. Although gene expression studies suggest that the RA signaling pathway and genes regulated by this pathway are altered in ALS, the role of this pathway in ALS remains unclear. The goal of the current study was to examine cytoplasmic binding proteins and nuclear receptors of the retinoid signaling pathway in lumbar spinal cord motor neurons of control and ALS patients. We also utilized a primary motor neuron-enriched cell culture system to demonstrate that RARβ-mediated signaling events modulate oxidative stress-induced cell death.

Surprisingly, we noted differences between SALS and FALS. More specifically, we observed increased nuclear immunoreactivity for both CRABP-II and RARβ in motor neurons of SALS patients. In SALS cases with increased CRABP-II in motor neuron nuclei, 85% (17 of 20) also had increased nuclear RARβ in motor neurons. The lack of RARβ nuclear immunostaining in FALS suggests differences in underlying disease mechanism between familial and sporadic forms of the disease. However, further studies are necessary to validate these initial findings.

We did not observe alterations between ALS and controls with respect to CRABP-II and nuclear receptor interaction via coimmunoprecipitation. Due to low cellular CRABP-II levels and multiple RAR binding partners, the reverse co-immunoprecipitation experiment was not successful. We suggest that downstream interactions between RA (ligand), RARβ (receptor) and nuclear response element (RARE) may explain the altered subcellular distribution we observed in motor neurons of SALS patients rather than upstream protein-protein interactions.

Although RA signaling has been shown to mediate apoptosis in a variety of cells [38,47-49], this pathway has also been shown to inhibit apoptosis in specific cell types [50-52]. We propose that redistribution of RARβ to motor neuron nuclei represents a regenerative, protective response as demonstrated previously [41]. Our inability to co-localize nuclear RARβ with TUNEL or activated caspase-3 (data not shown) suggests that RA signaling through RARβ is a non-apoptotic response. In addition, our in vitro data indicates that RARβ mediates the neuroprotective properties of RA to oxidative damage. RARβ-mediated pro-survival and regenerative responses have been reported previously in models of spinal cord and axonal injury [5,7,53,54].

The intracellular distribution of the RARs is an indicator of the transcriptional activity of RA [55,56]. This localization is believed to be determined, in part, by the presence of ligand. Previous reports suggest that over 95% of RARβ translocated to the nucleus in the presence of RA [57]. Although both ligand binding and receptor heterodimerization with RXRs are required for transcriptional activation, only heterodimerization is necessary to ensure binding to retinoic acid response elements. Based on the tissue-specific expression of the RXRs [58], we would expect RXRβ (expressed in most tissues) and possibly RXRγ (expressed mostly in muscle and brain) to heterodimerize with RARβ in the spinal cord. However we did not observe any change in the localization of RXRβ in spinal cord motor neurons between control and ALS subjects (data not shown). While the overall effect of RARβ on gene expression during ALS remains to be determined, several genes known to be modulated by RA have been implicated in motor neuron disease including various cytokines, growth factors, matrix metalloproteases (MMPs), and the amyloid precursor protein (APP) [59-65] (summarized in Figure 8).

Figure 8.

Schematic representation of retinoid signaling in SALS. Increased nuclear localization of RARβ occurs in SALS. Nuclear CRABP-II and cytoplasmic RARα localization were also observed. Signaling through RARβ in vitro was neuroprotective in the presence of oxidative stress. Processes shown to be mediated by the RARs are provided as potential downstream targets mediating this protection. Pharmacologic agents that stimulate RARβ may be promising therapeutic targets that could delay or prevent motor neuron cell death in ALS.

Our findings contradict prior studies investigating the role of retinoid signaling proteins, RARs and RXRs in animal models of chronic spinal cord degeneration. For example, Jokic and colleagues observed diffuse RARα and RARβ immunostaining in lumbar spinal cord motor neurons at pre-symptomatic stages of a rat model of ALS, which declined in end-stage disease [18]. A similar decline in RARα immunoreactivity was reported in rats fed a vitamin A-deficient diet [19]. Although we are unable to perform a longitudinal study in human tissue, we did not observe a difference between ALS and control subjects with respect to the level or subcellular localization of RARα. We believe this is due, at least in part, to species and mechanistic differences between the transgenic SOD1 animal models of ALS and the human disease.

We acknowledge that protein lysates generated from spinal cord tissue include multiple cell types that contribute to our immunoblot and coimmunoprecipitation data. However, these cells impact motor neuron survival [66] and are able to respond to RA signaling [67]. In fact, cells within the white matter were immunopositive for RARα and RARβ suggesting that they are responding to the disease process. Further studies using a combination of laser capture microdissection and quantitative real-time polymerase chain reaction (qPCR) will examine protein and mRNA levels on an individual cell basis.

Our cell culture experiments support a direct role for RARβ in motor neuron survival and neuroprotection. The fact that motor neuron viability was decreased in the absence of RA (Figure 6A) was surprising considering the multitude of growth factors and supplements included in the culture media. However, this suggests that RA modulated signaling is critically important for motor neuron survival and this is further supported by the neuroprotective properties of RA in the presence of oxidative damage. Moreover, RARβ specific agonists mediate the neuroprotective effects in this model system, indicating a particular nuclear receptor in the retinoid signaling pathway for motor neuron survival. Inhibition of RARβ exacerbated motor neuron cell death, suggesting that up- or down-regulation of specific downstream RARβ targets may prevent this response.

As noted above, ATRA and adapalene were neuroprotective even following toxic insult, suggesting that these agents could be useful therapeutically. Although the post-insult neuroprotective effects were modest and occurred at early time points, it may be that stimulating RARβ may be most effective when initiated early in the disease process, and could possibly protect motor neurons that have not yet been affected by the disease. Adapalene is an FDA-approved agent for the topical treatment of acne with few and relatively minor side effects [68]. An adapalene delivery system has been used for the treatment of cervical intraepithelial neoplasia without side effects [69] although an ALS-relevant therapy may require an analog that could cross the blood-brain barrier for optimal long-term treatment.

In conclusion, we have determined that the retinoid signaling pathway exhibits distinct alterations in SALS but not FALS patients. In particular, altered localization of two key components of the retinoid signaling pathway (CRABP-II and RARβ) were observed in spinal cord motor neurons of patients with SALS. Increased nuclear RARβ in lumbar spinal cord motor neurons may increase the ability of these cells to survive in the spinal cord microenvironment during ALS as it did not co-localize with markers of apoptosis. Furthermore, our in vitro data indicates RARβ signaling mediates neuroprotection of motor neurons under oxidative stress. Our findings support a role for retinoid signaling in the pathophysiology of ALS and suggest new potential therapeutic targets for ALS.