1. Volume 15, Issue 5, Pages 753-766 (September 2004)
GM1-Ganglioside-Mediated Activation of the Unfolded Protein Response Causes Neuronal Death in a Neurodegenerative Gangliosidosis 
Alessandra Tessitore, Maria del P. Martin, Renata Sano, Yanjun Ma, Linda Mann, Angela Ingrassia, Eric D. Laywell, Dennis A. Steindler, Linda M. Hendershot, Alessandra d'Azzo 
Molecular Cell 
Volume 15, Issue 5, Pages (September 2004)
DOI: /j.molcel

2. Figure 1
Morphologic Changes and Apoptotic Events Are Evident in β-gal−/− Spinal Neurons
(A) Electron microscopy of spinal cords from wild-type (β-gal+/+) mice and β-gal−/− mice both of 3 months of age revealed the presence of numerous enlarged lysosomes (L) filled with membranous material, ribosomes (Ri), and compressed ER (lower, middle, and right panels). Scale bars: upper panel, 1 μm; lower panels, 0.5 μm.
(B) Immunolabeling with anti-calnexin antibody showed that the level of expression of this ER-membrane protein is higher and more localized in 5-month-old mutant mice than in wild-type mice. Magnification, 60×.
(C) TUNEL assays revealed sporadic apoptotic cells in the spinal cord of β-gal−/− mice. TUNEL+ cells were counted on serial sections spanning 6 mm of spinal cord. The number of apoptotic cells present in β-gal−/− knockout (ko) mice is reported as fold of increase over that in age-matched wild-type (wt) mice. Substantially more apoptotic cells were detected in the sections from mutant mice than in those from wild-type mice.
Molecular Cell  , DOI: ( /j.molcel )

3. Figure 2 GM1-Ganglioside Loading of Wild-Type Cells Activates the UPR
(A) Double-immunolabeling of neurospheres with markers for astrocytes (GFAP, green) and neurons (β-III tubulin, red) showed that both types of cells are present in wild-type and β-gal−/− neurospheres. Nuclei were visualized with Hoechst stain (blue).
(B) Thin-layer chromatography of total glycolipids extracted from β-gal−/− and β-gal+/+ neurospheres and MEFs (with or without GM1 loading) revealed the accumulation of the ganglioside in both β-gal−/− and GM1-loaded wild-type neurospheres. Asterisk indicates the GM1 derivative, sialoglycolipid GD1a. Contrary to what was seen in the neurospheres, there appeared to be more GM1 in wild-type GM1-loaded MEFs than in β-gal−/− MEFs, a finding justifiable by the young age of the mice from which these cells were derived (E13.5). Asterisk indicates GD1a.
(C) Real-time PCR was performed on mRNA extracted from wild-type (with or without GM1 loading) and β-gal−/− MEFs and neurospheres. The levels of BiP, CHOP, Jnk2, and caspase-12 (casp-12) mRNAs increased in GM1-loaded and β-gal−/− neurospheres (red bars) and in MEFs (green bars). The reactions were standardized to the level of GAPDH mRNA.
(D) Western blot analysis of total lysates of neurospheres using XBP-1, Jnk, and caspase 12 antibodies. The levels of the 50 kDa stable form of XBP-1 were higher in GM1-loaded wild-type and β-gal−/− neurospheres (lanes 2 and 3) than in wild-type neurospheres (lane 1). Cells accumulating GM1 also showed a clear increase in the phosphorylated form of Jnk2, but not Jnk1, with a phospho-Jnk-specific antibody (lanes 2 and 3). The Pan-Jnk antibody was used to assess the total amount of Jnk proteins. The amounts of the mature 42- and 20 kDa caspase–12 was also increased in both GM1-loaded wild-type neurospheres and β-gal−/− neurospheres (lanes 2 and 3). 293 cells treated or not with UV were used as control for Jnk phosphorylation (lanes 4 and 5). Hsc70 was included as loading control. The asterisk indicates an unspecific band.
Molecular Cell  , DOI: ( /j.molcel )

4. Figure 3 GM1-Ganglioside Accumulation Results in Apoptosis
(A) FACS analysis of annexin-V-labeled MEFs was performed. Wild-type (β-gal+/+) cells loaded with GM1 showed an increase in two cell populations: early-apoptotic cells (annexin-V+; lower-right quadrant) and late-apoptotic cells (propidium iodine+, upper-right quadrant). The percentage of GM1-treated wild-type MEFs in early- and late-stage apoptosis was substantially higher than that in untreated wild-type cells and almost comparable to that observed in untreated β-gal−/− MEFs.
(B) TUNEL assay performed on neurospheres revealed the presence of apoptotic cells (green) in wild-type neurospheres loaded with GM1 and in β-gal−/−neurospheres, but not in wild-type untreated neurospheres. DAPI (blue) was used to label nuclei. Magnification, 20×.
Molecular Cell  , DOI: ( /j.molcel )

5. Figure 4
BiP Overexpressing CHO Cells Do Not Undergo Apoptosis following GM1 Loading
(A) Untreated (green bars) wild-type Chinese hamster ovary (CHO) cells and BiP-overexpressing CHO (CHO-BiPoe) cells were examined for apoptosis. After GM1 loading (blue bars), wild-type CHO cells showed a clear increase in apoptosis, as indicated by the increased number of Annexin-V+ cells. A comparable result was observed after tunicamycin (Tm; 1 μg/ml; red bars) treatment. In contrast, BiP-overexpressing CHO cells did not show an increase in Annexin-V staining after either treatment. Staurosporine (1 mM; yellow bars) induced apoptosis in both cell lines.
(B) The calcium concentration in ER calcium stores was measured by using INDO-1/AM fluorescence (ratio, 400 ± 20 nm: 480 ± 20 nm) in wild-type CHO cells and BiP overexpressing CHO cells. The assay was performed on either untreated cells, GM1-loaded cells, or Tg pretreated cells. Calcium release was measured before and after the addition of 1 μM thapsigargin (Tg) during FACS. In both cell types GM1-loading resulted in the release of ER calcium.
Molecular Cell  , DOI: ( /j.molcel )

6. Figure 5
ER Accumulation of GM1 Activates the UPR via Depletion of ER Calcium Stores
(A) Wild-type, GM1-loaded, and β-gal−/− MEFs were double labeled with anti-GM1 antibody (Alexa-green; left panels) and anti-calnexin antibody (Alexa-red; middle panels). Similarly to β-gal−/− MEFs, GM1-loading of wild-type MEFs resulted in a clear redistribution of the ganglioside and its abnormal accumulation at the ER membranes, as assessed by confocal laser scanning microscopy (right panels, merge). Magnifications, 40×.
(B) The calcium concentration in ER calcium stores was measured in untreated wild-type (black trace), GM1-loaded wild-type (blue trace), and β-gal−/− (red trace) neurospheres by using INDO-1/AM fluorescence (ratio, 400 ± 20 nm: 480 ± 20 nm). Calcium release was measured before and after the addition of 1 μM thapsigargin (Tg) during FACS.
(C) Direct measurement of ER calcium content was obtained by FRET analysis of wild-type and β-gal−/− MEFs transduced with cameleon constructs expressing ER-targeted YC3er. FACS analysis of these cells revealed a drastically reduced ratio between YFP-FRET and CFP in β-gal−/− cells (red bar) compared to wild-type cells (green bar). Calcium levels in mutant cells were similar to that of cells treated with Tg (blue and yellow bars). The results are the average of two independent experiments.
Molecular Cell  , DOI: ( /j.molcel )

7. Figure 6
BiP, CHOP, Jnk2, and caspase-12 Are Activated at the Transcriptional Level
(A) Thin-layer chromatography of total glycolipids extracted from the brain stem of wild-type mice and mutant mice and sprayed with resorcinol (left panel) or orcinol (right panel) revealed the abnormal accumulation of GM1 (asterisk) and its asialo-derivative, GA1 (closed circles), in the β-gal−/− sample. These glycolipids were not visible in the β-gal−/−/GalNAcT−/− double-knockout mice. The ganglioside standards are indicated by arrowheads. GM1: monosialoganglioside GM1; GM2: monosialoganglioside GM2; GM3: monosialoganglioside GM3; GD1a: disialoganglioside GD1a; GD3: disialoganglioside GD3; O-acetyl-GD3: ortho-acetyl-disialoganglioside GD3.
(B) Real-time PCR analyses revealed that the levels of CHOP and Jnk2 were consistently much higher in the β-gal−/− samples than in wild-type samples. The level of BiP was increased substantially only in the older animals. No transcriptional activation of BiP, CHOP, or Jnk2 was observed in β-gal−/−/GalNAcT−/− (dko) mice or in Neu1−/− mice at 5 months. The amounts of BiP, CHOP, or Jnk2 mRNAs in the mutant (ko) samples is reported as folds of increase over that detected in the wild-type (wt) samples. Values are the average of three independent experiments.
(C) An RNase protection assay revealed that caspase-12 was up-regulated in β-gal−/− mice (lanes 2, 4, 6) compared to wild-type (β-gal+/+) mice (lanes 1, 3, 5) . The relative levels of RNA induction were normalized against L32 RNA. The amount of caspase-12 in β-gal−/− samples is reported as folds of increase over that detected in age-matched wild-type mice.
Molecular Cell  , DOI: ( /j.molcel )

8. Figure 7
Levels of ATF6, CHOP, Jnk2, and Caspase-12 Proteins Increase over Time in β-gal−/− Spinal Cord
(A) Western blots were probed with anti-ATF6 antibody. The 50 kDa nuclear form of ATF6 was dramatically elevated in β-gal−/− mice (lanes 2 and 4) compared to wild-type animals (lanes 1 and 3).
(B) Western blots using the anti-CHOP antibody also revealed that CHOP was consistently higher in β-gal−/− mice (lanes 2, 4, 7) than in wild-type animals (lanes 1, 3. 6). However, the level of CHOP normalized in the β-gal−/−/GalNAcT−/− double knockouts (lane 5); CHO cells, untreated or treated with tunicamycin (Tm) (lanes 8 and 9, respectively), were used as a control of CHOP activation (Ma et al., 2002).
(C) Western blots were probed with anti-Jnk antibodies. The level of phosphorylated Jnk2 increased in β-gal−/− mice (lanes 2, 4, 6) compared to age-matched wild-type mice (lanes 1, 3, 5), while the level of phosphorylated Jnk1 did not vary significantly in the same samples. A similar increase in the level of Jnk2 total protein was also observed with a pan-Jnk antibody.
(D) Immunoprecipitation with anti-caspase-12 antibody revealed an age-dependent increase in both the 62 kDa procaspase-12 precursor and the 42 kDa mature form of caspase-12 in mutant mice (lanes 2 and 4). The mature form of caspase-12 was not seen in wild-type samples (lanes 1 and 3). In the β-gal−/−/GalNAcT−/− mice (lane 5), the level of the procaspase-12 was normalized, and the mature form was not detected.
Molecular Cell  , DOI: ( /j.molcel )