Supporting Information

Lee et al. 10.1073/pnas.1300894110 3 2 SI Materials and Methods 10 cells/cm on glass coverslips coated with poly-D-lysine (Sigma- Human Subjects. Postmortem specimens of human spinal cord Aldrich) in six-well culture dishes and were cultured under fi were obtained from Boston University School of Medicine. a humidi ed atmosphere containing 5% (vol/vol) CO2 at 37 °C in Cervical spinal cord sections from nine sporadic amyotrophic Neurobasal medium (Invitrogen) supplemented with B27 solu- lateral sclerosis (sALS) patients and eight age-matched control tion, 0.5 mM glutamine, glial cell line–derived growth factor (10 individuals without neurological disorders were examined. The ng/mL; R&D Systems), ciliary neurotrophic factor (10 ng/mL, average age of the subjects was 61.3 y (range 41–78 y) for sALS R&D Systems), and brain-derived neurotrophic factor (1 ng/mL, patients and 59.3 y (21–84 y) for control individuals. The post- R&D Systems). mortem interval (PMI) ranged from 4 to 22 h, with a mean of Cell Culture and DNA Transfection. Mouse motor-neuron-like 14 h. All studies with the use of human samples were approved fi by the Boston University School of Medicine Ethics Commission NSC34 cells and 293T cells were cultured under a humidi ed in accordance with Helsinki Declaration. For immunohisto- atmosphere of 5% (vol/vol) CO2 at 37 °C in DMEM (Gibco) – supplemented with 10% heat-inactivated FBS (Gibco). For chemical analysis of phospho mammalian sterile 20 (STE20)- × 6 like kinase 1 (MST1), human specimens that had been fixed in DNA transfection, 293T cells (1 10 ) were plated in a 100-mm dish, cultured for 24 h, and transfected for 48 h with appropriate 10% formalin and embedded in paraffin were depleted of par- vectors with the use of polyethyleneimine (Sigma-Aldrich). For affin by treatment with xylene, incubated for 10 min at room construction of NSC34 cells stably expressing human super- temperature in methanol containing 0.3% H O to inactivate 2 2 oxide dismutase 1 (WT) (hSOD1) or SOD1(G93A), the cells endogenous peroxidase, and then blocked with 5% (wt/vol) BSA were transfected with pcDNA3 vectors encoding the Flag- for 1 h. The sections were incubated first overnight at 4 °C with tagged human proteins, and stable transfectants were selected antibody to phospho-MST1 and then consecutively for 1 h with in culture medium containing G418 (500 μg/mL, Gibco). biotin-conjugated goat antibody to rabbit IgG (Vector Laborato- – – Heterogeneous populations of the stable transfectants were ries) and an avidin biotin peroxidase complex (Vector Labora- studied to avoid clonal variation of the neomycin-resistant cells. tories). The sections were then exposed to 0.05% 3,3′- diaminobenzidine tetrahydrochloride dehydrate (DAB) (Sigma- Antibodies. Mouse monoclonal antibodies to Flag and to α-tubulin fi Aldrich) and 0.003% H2O2 before examination with a bright- eld as well as rabbit polyclonal antibodies to human SOD1 were microscope (Olympus). The intensity of phospho-MST1 im- obtained from Sigma-Aldrich. Mouse monoclonal antibodies to munostaining in ventral motor neurons in each sample was quanti- non-phosphorylated neurofilaments (SMI32) were from Ab- fi ed with the use of Image Gauge V4.0 software (Fuji Photo Film). cam, and those to neuronal nuclei (NeuN) and to glial fibril- lary acidic protein (GFAP) were from Millipore. Rabbit Stereological Analysis of Viable Motor Neurons. Coronal sections – μ – polyclonal antibodies to apoptosis signal regulated kinase 1 (thickness, 40 m) of the lumbar spinal cord (L1 L5) were (ASK1) and to p38 as well as goat polyclonal antibody to MST1 prepared with the use of a Cryocut Microtome (Leica Micro- were from Santa Cruz Biotechnology. Rabbit polyclonal anti- systems), and every 10th section was stained with 0.5% cresyl bodies to thioredoxin-1 (Trx1) was from Lab Frontier, and those violet (Sigma-Aldrich). The total number of viable motor neu- to MST1, to phospho-MST1, to beclin-1, to autophagy-related rons in the lumbar spinal cord was estimated by the optical protein 12 (ATG12), to light chain 3 (LC3), and to the cleaved fractionator method as described (1). The region of ventral forms of caspase-9 or caspase-3 were from Cell Signaling. segments for stereological counts of motor neurons was outlined with the use of an Olympus BX-51 microscope fitted with a 4× Immune Complex Kinase Assay. Cultured cells were lysed with a lysis objective, and a 100× oil immersion objective was applied to buffer (3), and the extracts of the lumbar spinal cord were pre- count the number of motor neurons. The number of viable pared by homogenization in the lysis buffer with a Dounce ho- motor neurons was then quantified with the use of Visiopharm mogenizer. Cell lysates or the spinal cord extracts were subjected Integrator System software (Visiopharm) connected to a bright- to immunoprecipitation with the appropriate antibodies, and the field microscope (Olympus) and according to the following resulting precipitates were assayed for the indicated kinase equation: activities as described previously (4). GST fusion proteins of X forkhead box protein O3 (FOXO3), MAP kinase kinase 6 1 1 1 N = Q × × × (K82A) [MKK6(K82A)], and activating transcription factor 2 hsf asf ssf (ATF2) were used as substrates for MST1, ASK1, and p38, re- spectively. where ΣQ is the total number of counted motor neurons, hsf is the height sampling fraction (3/4), asf is the area of the sampling Immunoprecipitation and Immunoblot Analysis. For immunoblot fraction (1/20), and ssf is the slice sampling fraction (1/10). analysis, lysates of cultured cells or extracts of the lumbar spinal cord were subjected to SDS–polyacrylamide gel electrophoresis, Primary Culture of Motor Neurons. The spinal cord of E13 embryos and the separated proteins were transferred to a polyvinylidene was dissected to obtain the ventral segments, which were then difluoride membrane (Millipore). The membrane was exposed to incubated in L15 medium (Invitrogen) containing 0.05% trypsin 5% (wt/vol) nonfat dried milk and then probed with appropriate for 15 min at 37 °C and dissociated by gentle trituration with primary antibodies, after which immune complexes were de- a Pasteur pipette. The resulting single-cell suspension was lay- tected with horseradish peroxidase–conjugated goat antibodies ered on top of a cushion of 10.4% (vol/vol) Optiprep (Sigma- to mouse or rabbit IgG (Millipore) and with enhanced chem- Aldrich) (2) in a 15-mL centrifuge tube and centrifuged at 800 × g iluminescence reagents (Pierce). The intensity of bands was for 15 min at 4 °C. Motor neurons were collected from the white quantified with Image Gauge V4.0 software. For immunopre- turbid band that formed at the interface of the two layers. cipitation, lysates of cultured cells or extracts of the lumbar Primary motor neurons (PMNs) were plated at a density of 2 × spinal were incubated for 3 h at 4 °C with appropriate antibodies

Lee et al. www.pnas.org/cgi/content/short/1300894110 1of8 and then for an additional 1 h with protein G–conjugated aga- goat IgG) that had been conjugated to unique DNA probes rose. Immune complexes were isolated by centrifugation at (rabbit PLUS and goat MINUS) and then for 30 min with 12,000 × g for 3 min at 4 °C and subjected to immunoblot DNA ligase to induce circularization of DNA probes located in analysis with the indicated antibodies. close proximity to each other. Polymerase-mediated rolling-circle amplification (RCA) was then performed with the ligated circles Measurement of Intracellular Reactive Oxygen Species Production. as a template. For bright-field analysis, a horseredish peroxidase- Intracellular reactive oxygen species (ROS) generation was de- fl conjugated DNA linker was incorporated into the RCA step to termined by dichloro uorescin (DCF) assay (5). Cultured NSC34 detect the site of MST1–Trx1 interaction as a brown dot. Spinal cells in a 24-well plate were incubated with 100 μM dichloro- cord sections were then counterstained with Mayer’s hematoxylin, dihydro fluorescein diacetate (DCDHF-DA) (Molecular Probe) images were acquired with an Olympus BX53 microscope, and the plus 2% (vol/vol) Pluronic F-127 in Hepes-buffered (HCSS) number of spots was counted in ventral motor neurons in each buffer (120 mM NaCl, 5 mM KCl, 1.6 mM MgCl2,2.3mMCaCl2, fi 15 mM glucose, 20 mM Hepes, and 10 mM NaOH) for 1 h at 37 °C captured eld. under the dark condition and washed three times with HCSS Fluorescence Analysis of Autophagy. PMNs in culture were moni- buffer. The relative fluorescence of oxidized DCDHF was tored for autophagosome formation with the use of Premo monitored using a FL600 microplate fluorescence reader (Bio- Tek Instruments) equipped with filter for the oxidized DCDHF Autophagy Sensor (LC3B-FP) system (Invitrogen). PMNs were (excitation, 485 nm; emission, 530 nm). All data were normalized cultured for 3 d and then transduced for 24 h with GFP-LC3 ’ fi with respect to the fluorescence intensity of SOD1(WT) cells. according to the manufacturer s protocol. The cells were xed, permeabilized, stained with DAPI (5 μg/mL), and then examined Immunohistofluorescence Analysis. Mice were perfused trans- by fluorescence microscopy. The intensity of GFP–LC3 fluores- cardially first with PBS and then with 3% (vol/vol) para- cence in the perinuclear region of >50 cells in each group was formaldehyde in PBS with the use of a peristaltic pump. The quantified with the use of Image Gauge V4.0 software. spinal cord was isolated from the mice, subjected to cry- oprotection with 30% (wt/vol) sucrose, and dissected into the Electron Microscopy. The spinal cord obtained from mice subjected cervical, thoracic, and lumbar segments, all of which were frozen to transcardial perfusion with 3% (wt/vol) paraformaldehyde in with dry ice. The frozen lumbar segments were embedded in PBS was cut into cervical, thoracic, and lumbar segments. The optimal cutting temperature (OCT) compound (Sakura), cut lumbar spinal cord was further fixed overnight at 4 °C with 2% coronally at a thickness of 30 μm with a cryostat, and then (vol/vol) glutaraldehyde in 0.05 M sodium cacodylate buffer (pH mounted on gelatin-coated slides. The sections were exposed at 7.2) and was then exposed for 2 h at 4 °C to 1% osmium te- room temperature first for 10 min to a solution containing 0.3% troxide in the same buffer. The tissue was briefly washed with H2O2 and 0.25% Triton X-100 and then for 1 h to 10% horse distilled water, stained en bloc overnight at 4 °C with 0.5% uranyl serum. They were then incubated first overnight at 4 °C with acetate, dehydrated with a graded series of ethanol solutions, primary antibodies and then for 2 h at room temperature with treated with propylene oxide, and embedded in Spurr’s resin that fl – uorescein isothiocyanate conjugated goat antibodies to mouse had been polymerized in a 70 °C oven for 24 h. Ultrathin sections – IgG (Vector Laboratories) or Texas red conjugated goat anti- were prepared from the specimens with a diamond knife bodies to rabbit IgG (Vector Laboratories). Fluorescence images (MT-X ultramicrotome; RMC), and the dorsal region of each were acquired with an Olympus BX53 fluorescence microscope was trimmed off. The sections were mounted on bare copper equipped with an Olympus DP70 digital camera and were ana- grids, stained first with 2% (vol/vol) uranyl acetate for 7 min and lyzed with the use of Image Gauge V4.0 software. then with Reynolds’ lead citrate for 7 min, and then examined In Situ Proximity Ligation Assay. In situ proximity ligation assay with a transmission electron microscope (JEM-1010; JEOL) at (PLA) was performed with the use of a Duolink II In Situ PLA kit an accelerating voltage of 80 kV. A total of at least 15 micro- (Olink Bioscience). In brief, spinal cord sections from mice were graphs were collected for each experimental group and were fixed, blocked with 10% (vol/vol) horse serum, and incubated examined for the presence of autophagic vacuoles by morpho- overnight at 4 °C with goat antibodies to MST1 and rabbit an- metric analysis according to criteria described previously (6, 7). tibodies to Trx1. The samples were then incubated first for 1 h The number of autophagic vacuoles was counted in each cap- with corresponding secondary antibodies (to rabbit IgG and to tured field.

1. Lu XH, et al. (2009) Bacterial artiﬁcial chromosome transgenic mice expressing a truncated 4. Ryoo K, et al. (2004) Negative regulation of MEKK1-induced signaling by glutathione mutant parkin exhibit age-dependent hypokinetic motor deﬁcits, dopaminergic neuron S-transferase Mu. J Biol Chem 279(42):43589–43594. degeneration, and accumulation of proteinase K-resistant alpha-synuclein. J Neurosci 5. Kim SH, et al. (2002) Brain-derived neurotrophic factor can act as a pronecrotic factor 29(7):1962–1976. through transcriptional and translational activation of NADPH oxidase. J Cell Biol 2. Wang PY, et al. (2005) Mullerian inhibiting substance acts as a motor neuron survival 159(5):821–831. factor in vitro. Proc Natl Acad Sci USA 102(45):16421–16425. 6. Eskelinen EL (2008) New insights into the mechanisms of macroautophagy in 3. Park HS, Lee JS, Huh SH, Seo JS, Choi EJ (2001) Hsp72 functions as a natural inhibitory mammalian cells. Int Rev Cell Mol Biol 266:207–247. protein of c-Jun N-terminal kinase. EMBO J 20(3):446–456. 7. Koike M, et al. (2008) Inhibition of autophagy prevents hippocampal pyramidal neuron death after hypoxic-ischemic injury. Am J Pathol 172(2):454–469.

Lee et al. www.pnas.org/cgi/content/short/1300894110 2of8 Primary motor neurons

Control WT G93A G93A + Trolox G93A + NAC pMST1 SMI32 Merge

Fig. S1. Effect of SOD1(G93A) expression on MST1 phosphorylation in primary motor neurons. PMNs prepared from E13 embryos of control B6 mice or human SOD1(WT) or SOD1(G93A) transgenic mice were cultured for 3 d and then incubated in the absence or presence of 100 μM Trolox or 3 mM NAC for 12 h. The cells were then ﬁxed, permeabilized, and subjected to immunoﬂuorescence staining with antibodies to phospho-MST1 (pMST1, red) and to SMI32 (green). (Scale bar, 200 μm.)

Phospho-MST1

Normal #2 Normal #3 Normal #4 Normal #5

Normal #6 Normal #7 Normal #8

sALS #2 sALS #3 sALS #4 sALS #5

sALS #6 sALS #7 sALS #8 sALS #9

Fig. S2. Patients with sALS show higher activity of MST1 in ventral motor neurons. Postmortem spinal cord sections from normal individuals and sALS patients (data for Normal #1 and sALS #1 cases are shown in Fig. 1A) were subjected to immunohistochemical analysis with antibodies to phospho-MST1. (Scale bar, 20 μm.)

Lee et al. www.pnas.org/cgi/content/short/1300894110 3of8 A Lumbar sections (16wk) 3 P<0.05 NeuN / 2 Control G93A

(% of control) (% of 1 Phospho-MST1

MST1-/- G93A/MST1-/-

Relative intensity of phospho-MST1 of intensity Relative 0 Control G93A MST1-/- G93A/MST1-/-

B Phospho-MST1/GFAP Control G93A 8 wk 12 wk 16 wk

Fig. S3. Immunofluorescence staining of phospho-MST1 and NeuN (A)orGFAP(B). (A) Sections of the lumbar spinal cord of 16-wk-old control, SOD1(G93A), − − − − MST1 / , and SOD1(G93A)/MST1 / mice (n = 4 per group) were subjected to immunofluorescence staining with antibodies to phospho-MST1 (red) and to NeuN (green). (Scale bar, 10 μm.) (B) Sections of the lumbar spinal cord of 8-, 12-, 16-wk-old control, and SOD1(G93A) mice (n = 4 per group) were subjected to immunofluorescence staining with antibodies to phospho-MST1 (red) and to GFAP (green). (Scale bar, 20 μm.)

Lee et al. www.pnas.org/cgi/content/short/1300894110 4of8 P P 300 <0.05 <0.05

250

200

150 (% of WT)(% of 100 DCF fluorescence intensity 50

0 --trolox NAC

Flag-SOD1: (WT) (G93A)

Fig. S4. Effect of SOD1(G93A) on ROS generation in intact cells. NSC34 cells stably expressing Flag-tagged human SOD1(WT) or SOD1(G93A) were incubated in the absence or presence of 100 μM Trolox or 3 mM NAC for 12 h, then with 100 μM DCDHF-DA for 1 h. Intracellular ROS production was monitored by the ﬂuorescence signals of oxidized DCDHF-DA with a ﬂuorescence reader. The data were means ± SEM from three independent experiments. Statistical analysis was performed with one-way ANOVA followed by Student Newman–Keuls test.

NSC34 Flag-SOD1: WT G93A IP: α-MST1 IB: α-Flag IP: α-MST1 IB: α-Trx1

α-Flag

α-Trx1 Lysates α-MST1

Fig. S5. MST1 does not physically associate with either wild-type SOD1 or SOD1(G93A). NSC34 cells stably expressing Flag-tagged human SOD1(WT) or SOD1 (G93A) were lysed and subjected to immunoprecipitation (IP) with antibodies to MST1. The resulting precipitates were examined by immunoblot analysis (IB) with antibodies to Flag and to Trx1. Cell lysates were also examined directly by immunoblot analysis with antibodies to MST1, to Trx1, or to Flag.

Lee et al. www.pnas.org/cgi/content/short/1300894110 5of8 PLA: Trx1-MST1 (Lumbar sections ) A P 20 <0.05

Control G93A 10

5 No. of spots/motor neuron

MST1-/- G93A/MST1-/- 0 Control G93A MST1-/- G93A/MST1-/-

B PLA: Trx1-MST1 (Lumbar sections)

(i) (ii) (iii) (iv) (v)

MST1 Ab +-+++ Trx1 Ab ++-++ MST1 ---+- Trx1 ----+

Fig. S6. SOD1(G93A) induces dissociation of MST1 from Trx1 in ventral motor neurons. (A) PLA analysis of the interaction between MST1 and Trx1 in lumbar spinal cord sections from 8-wk-old mice of the indicated genotypes is shown (Left). (Scale bar, 10 μm.) The number of brown spots (denoting interaction of the two proteins) in motor neurons was also determined (Right); data are means ± SEM (n = 4 mice of each genotype). (B) The interaction between MST1 and Trx1 in the lumbar sections of 8-wk-old control mice was examined by PLA assay as in A. To verify the speciﬁcity of the PLA signal indicating MST1–Trx1 interaction, the lumbar sections were incubated with the indicated combinations of anti-MST1 antibody, anti-Trx1 antibody, and puriﬁed hexahistidine-tagged MST1 or Trx1 proteins. (Scale bar, 10 μm.)

A B Flag-SOD1(WT) Flag-SOD1(G93A) Flag-SOD1(WT) Flag-SOD1(G93A) - - zVAD SB - - zVAD SB MST1 assay GST-FOXO3 Cleaved caspase-9 Fold: 1.0 2.2 1.9 1.9

Cleaved caspase-3 ASK1 assay GST-MKK6(K82A)

Flag-SOD1 Fold: 1.0 2.3 2.7 2.8 p38 assay GST-ATF2 Tubulin Fold: 1.0 3.5 3.2 1.2

ASK1

p38

Lysates MST1

Flag-SOD1

Tubulin

Fig. S7. SOD1(G93A) induces sequential activation of p38 MAPK and caspases in NSC34 cells. NSC34 cells stably expressing Flag-tagged human SOD1(WT) or SOD1(G93A) were incubated in the absence or presence of 50 μM carbobenzoxy-valyl-alanyl-aspartyl-[Omethyl]-ﬂuoromethylketone (zVAD-fmk) or 30 μM SB203580 (SB) for 12 h, after which cell lysates were subjected to immunoblot analysis with antibodies to the cleaved forms of caspase-9 or caspase-3, to Flag, or to α-tubulin (A) or were subjected to immune complex kinase assays for MST1, ASK1, and p38 (B).

Lee et al. www.pnas.org/cgi/content/short/1300894110 6of8 A Flag-SOD1(WT) Flag-SOD1(G93A) - - zVAD SB

Beclin-1

ATG5-ATG12

LC3-I

LC3-II

Flag-SOD1

Tubulin B Primary motor neurons P<0.05 P<0.05 Control G93A 300

250 DAPI

/ 200

G93A+zVAD G93A+SB 150

GFP-LC3 100 (% of control)

50 Reletive intensity of GFP-LC3 0 Control G93A G93A G93A + + C zVAD SB Primary motor neurons

Control G93A 400 *

300 DAPI / † MST1-/- G93A/MST1-/- 200 GFP-LC3 (% of control) 100 Reletive intensity of GFP-LC3

Fig. S8. Genetic ablation of MST1 mitigates the SOD1G93A-induced accumulation of autophagosomes in motor neurons. (A) NSC34 cells stably expressing Flag- tagged human SOD1(WT) or SOD1(G93A) were incubated in the absence or presence of 50 μM zVAD–fmk or 30 μM SB203580 for 12 h, lysed, and subjected to immunoblot analysis with antibodies to beclin-1, to ATG12, to LC3, to Flag, or to α-tubulin. GFP–LC3 was introduced into PMNs that had been prepared from − − − − E13 embryos of SOD1(G93A) transgenic or control mice (B) or from control, SOD1(G93A), MST1 / , or SOD1(G93A)/MST1 / mice (C) and then cultured for 3 d. In B, the cells were then incubated further for 12 h in the absence or presence of 50 μM zVAD-fmk or 30 μM SB203580. All cells were then fixed, permeabilized, and stained with 4′,6-diamidino-2-phenylindole (DAPI). Representative images of GFP–LC3 (green) and DAPI (blue) fluorescence are shown (Left). (Scale bar, 100 μm.) The fluorescence intensity of GFP–LC3 in the cytoplasm of >50 cells for each experimental group was quantified and expressed as a percentage of the value for the control group (Right); data are means ± SEM from three independent experiments. Statistical analysis was performed with one-way ANOVA followed by Student Newman–Keuls test. *P < 0.05 versus control, †P < 0.05 versus SOD1(G93A).

Lee et al. www.pnas.org/cgi/content/short/1300894110 7of8 Lumbar sections

Control G93A MST1-/- G93A/MST1-/- p62 SMI32 Merge

Fig. S9. SOD1(G93A)-induced accumulation of p62 in motor neurons of the lumbar spinal cord and its attenuation by genetic ablation of MST1. Sections of − − − − the lumbar spinal cord prepared from 16-wk-old control, SOD1(G93A), MST1 / , and SOD1(G93A)/MST1 / mice (n = 4 mice per group) were subjected to immunoﬂuorescence staining with antibodies to p62 (red) and to SMI32 (green). (Scale bar, 1 μm.)

Lee et al. www.pnas.org/cgi/content/short/1300894110 8of8