Advances in APBD Research: High Content Screening, Antisense Oligonucleotides and Computational Drug Design
Advances in APBD Research: High Content Screening, Antisense Oligonucleotides and Computational Drug Design Or Kakhlon Department of Neurology Hadassah University Hospital 20 June 2012 APBD Research Foundation Annual Meeting, New York, NYFunding: APBD Research Foundation
Glycogen biosynthesis involves chain elongation by Glycogen Synthase (GS) andchain branching by Glycogen Branching Enzyme (GBE). If chain elongationoutbalances chain branching, glycogen could form starch-like precipitates madeup of long, non-branched chains called polyglucosans. Polyglucosan, non-branched Normal glycogen, branched GS/GBE activity ratio
What causes Adult Polyglucosan Body Disease (APBD)?Glycogen build up is normally suppressed in neurons by a well-regulated system,which inactivates (phosphorylates) and degrades Glycogen Synthase (GS)
Glycogen build up is suppressed in neurons by a Nevertheless, overwell-regulated system time glycogen could precipitate as polyglucosan bodies if chain elongation is not adequately balanced by its branching.Vilchez et al (2007) Nat Neurosc APBD Striano et al (2008) Nat ClinWierzba-Bobrowicz et al (2008) Pholia Neuropathol Pract Neurol
Two experimental approaches for curing APBD: 1. Ameliorating APBD, or slowing down the progress of the disease byreducing the GS/GBE activity ratio.2. Clearance of polyglucosan bodies.
1. Ameliorating APBD, or slowing down the progress of the disease by reducing the GS/GBE activity ratio. There are pre-existing PG which cannot be removed by any GS/GBE modulating strategy. Y329S Gluc/Co Y329SPG Intensity/cell Gluc/Co 3d Gluc/Rap Y329S Keto 3d Gluc/Rap 3d
Rapamycin can induce autophagy and also reverse polyglucosan accumulation. However, only in neurons transduced with shGBE1 lentiviruses.Therefore, rapamycin could only suppress de novo PG synthesis, not degrade pre-existing PG GFP LC3 Gly shGBE1 hGBE1/Rap/3-MA shGBE1/Rap
1. Ameliorating APBD, or slowing down the progress of the disease byreducing the GS/GBE activity ratio.Nevertheless, there are three therapeutic strategies for reducing the GS/GBE ratio:1A. Injection of Antisense Oligonucleotides againstPTG & GS in collaboration with ISIS Pharmaceuticals.This approach is already in Phase I clinical trials fortreating other disorders such as Spinal MuscularAtrophy (SMA)Plan:1. In vitro screening for the identification of antisenseoligonucleotides (ASO) to PTG is currently in progress(expected to be completed by July 2012).2. Scaling up the drug and screening it in vivo both byinntracerebroventricular and systemically bysubcutaneous injection (a few months).3. Lead ASOs identified by the in vivo screens will betested in the APBD mouse model (expected by fall of2012). Hua et al (2010) Genes Dev
1. Ameliorating APBD, or slowing down the progress of the disease byreducing the GS/GBE activity ratio.1B. Candidate Testing
Candidate testingTesting three types of compounds known to reduce the GS/GBE ratio:1. GS inhibitors (AMP Kinase (AMPK) and GSK3β activators). Examples: 5-Aminoimidazole-4-carboxamide-1-β-D-ribofuranoside (AICAR), PI3K inhibitors (e.g., wortmanin, Akt inhibitor IV), Berberine (herbal drug) etc.2. GBE stabilizers.3. Compounds predicted by solvent mapping to replace mutated Tyr329 in GBE1, or to destabilize GS and PTG.
Candidate testing Compounds predicted by solvent mapping (in collaboration with Dima Kozakov, Boston University) to replace mutated Tyr329 in GBE1, or to destabilize GS and PTG.Designing drugs using protein solvent mappingGBE, GS and PTG were checked for "druggability" (activators for GBEand inhibitors for GS and PTG).Kozakov et al (2012) PNAS
The Y329S site is druggable, as it is not conserved (less off target binding) and isconcave and with a hydrophobic functional group.
Plan:a. Testing candidate binding using calorimetry.b. Testing the effect of bound drugs on GBE and GS activity using our establishedbiochemical assays.c. Testing library compounds for binding and activity modulation.
1. Ameliorating APBD, or slowing down the progress of the disease byreducing the GS/GBE activity ratio.2. Clearance of polyglucosan bodies.1C. Screening both for compounds which reduce GS/GBE ratio, or clearPG: High Content Screening using the IN CELL 2000 Analyzer
Ctrl Optimization by epifluorescence and confocal microscopes: Some representative photosY329S keto keto/RapGluc Gluc/Rap Gluc/Co
Integrated intensity of punctate PAS fluorescence (PAS/IP) is significantly higher inAPBD patient (Y329Shomozygous) skin fibroblasts than in control fibroblasts.PAS/IP in fibroblasts treated for 3 days with a ketogenic and diastase pre-digestedis higher in patient-derived fibroblasts as compared to control fibroblasts. Thismeans that patient-derived fibroblasts have pre-existing polyglucosans. Patient-derived fibroblasts have pre-existing polyglucosans.
There is no significant difference in PAS/PI between fibroblasts treated with ketogenic medium and ketogenic medium supplemented with rapamycin. Rapamycin couldnt degrade pre-existing polyglucosans (but only reduce de novo synthesized). Diastase digestion amplifies the difference between fibroblasts treated with glucose and ketogenic medium. PAS/PI intensity in glucose fed fibroblasts is not significantly different between diastase digested and not digested conditions. Most PAS/PI staining is attributable to polyglucosans. Cobalt increases PAS/PI staining in glucose fed fibroblasts with or without diastase digestion.Rapamycin decreases PAS/PI. This reduction also only becomes statistically significant afterdiastase digestion.Fibroblasts treated with glucose, rapamycin and cobalt have lower PAS/PI than fibroblasts treated withglucose and cobalt only, but equal PAS/PI to fibroblasts treated with glucose and rapamycin only. The effect of rapamycin in reducing PAS/PI overrides the effect of cobalt in increasing it.
Polyglucosan analysis in control andY329S fibroblasts by HCA using the IN CELL 2000 Analyser Leonardo J. Solmesky, Ph.D. Cell Screening Facility for Personalized Medicine, Lab for Neurodegenerative Diseases and Personalized Medicine, Department of Cell Research and Immunology, Wise Faculty of Life Sciences, Tel Aviv University, Israel
Control Y329S/Gluc Y329S/Gluc/Co Y329S/Gluc/Rap•High level of diffuse, non granular staining in control. Need more aggressive diastasetreatment.•Co increases PG intensity and rapamycin decreases it, reverting PAS staining to the diffusepattern observed in control cells Leonardo J. Solmesky, Ph.D. Cell Screening Facility for Personalized 19 Medicine
An example of an improved diastase treatment:ctrl Y329S
Polyglucosan granules count/cell. Populational distribution in different treatments Gluc Gluc/Co Gluc/Rap•Rapamycin skews distribution to the left, i.e., reduces PG number.•Rapamycin increases area under the curve, i.e., increases cell count suggesting rescue against PGtoxicity. •We need to analyze PG integrated intensity! 21•Co appears to be toxic in this experiment.
Polyglucosan granules mean area populational distribution in different treatments Gluc Gluc/Co Gluc/Rap•Conclusion: Treatments do not affect mean PG size. Consistent with rap blocking de novo synthesis 22
Polyglucosan granules total area populational distribution in different treatments Gluc Gluc/Co Gluc/Rap•As compared to mean PG size, total PG size correlates better with PG number 23
Nuclear area populational distribution in different treatments Gluc Gluc/Co Gluc/Rap•Treatments did not affect nuclear size 24
Nuclear IxA populational distribution in different treatments Gluc Gluc/Co Gluc/Rap•Treatments did not affect nuclear size 25
Population distribution among different phases of cell cycle under different treatments Gluc Gluc/Co Gluc/Rap •Gluc/Co treatment led to growth arrest, while rapamycin slightly accelerated growth 26
High Throughput Screening pharmacophoreReadouts: PG integrated intensity (PAS/PI).Positive hits analysis1. If positive hits are suspected activators of AMPK, or GSK3β, testing activation of purified enzymes.2. Testing for undesirable chemical liability (covalent protein binding, thioether adduct formation) consequent to metabolic processing.3. Ranking compounds according to their effectiveness. Clusters of highest and lowest scorers will be aligned to predict a pharmacophore and toxicophore, respectively. in silico screening of pharmacophore and toxicophore (using Similarity Ensembl Approach (SEA))Similar structures with knownmechanism of action (future lead Molecules with liabilitycompounds) Possible targetsCandidates for therapeutic modulation.
Summary•APBD can be ameliorated by reduction of theGS/GBE activity ratio.•This approach is tested by a. AntisenseOligonucleotide injection; b. Solvent mappingof GBE, GS and PTG in order to find bindingdrugs; c. High Throughput Screening.•High Throughput Screening is also used tofind potential compounds which would clearPG and thus could not only ameliorate, butalleviate APBD pathogenesis.
Genomic DNA indicates a heterozygous p.Y329S mutation. However, cDNA reveals only onemutated allele, indicating that the second allele is missing.cDNA sequencing: Manifesting Manifesting heterozygote heterozygote Wt ctrl Homozygote whole blood lymphoblasts
Manifesting hetreozygotes (MH) phenomenon – is it due to differential allelic expression?First and foremost need control: Test MH carrier parents for mRNA homozigosity of cDNA v heterozygosity of gDNA Yes No Need to explain the manifesting Is there reduced/no expression of WT allele? heterozygotes phenomenon by other ways – alternative to differential allelic expression Quantitate GBE expression by RT-PCR GBE expression in MH is roughly 50% GBE expression in MH is of normal roughly the same as normal Suggests WT allele is suppressed Suggests compensation by overexpression Polymorphism: Check: Methylation Collaboration: Corroborate results by (Bisulfite sequencing), compare haplotypes between MH and Look for post transcriptional PAS-based genome-wide siRNA WT, Linkage analysis to MH modifications of GBE in MH screen in wt MEFs trait, whole gene analysis, exome analysis, CGH to check copy number differences, exome analysis, whole genome analysis, transcription factor binding, (exclude small deletions?)
ThanksTel Aviv University Hadassah Medical Center Boston UniversityLeonardo Solmesky Alexander Lossos Dima KozakovMiguel Weil Columbia University ISIS Pharmaceuticals Orhan Akman Tamar Grossman Salvatore DiMauro
Apoptosis is also reported in glycogen synthase-activatedneurons Vilchez et al (2007) Nat NeuroscSuggestion: Polyglucosan accumulation induces apoptosis
Using the model to test therapeutic approaches: Induction of autophagyInclusion bodies (PBs) formed.Can induction of autophagyfacilitate their clearance? Can autophagy be cytoprotective against apoptosis? Maiuri et al (2007) Nat Rev Mol Cell BiolSarkar et al (2009) Cell Death DifferTest autophagy enhancers as a therapeutic strategy against APBD
Autophagy can be stimulated and inhibited in neurons by rapamycin and 3-methyl adenine, respectively Jaeger & Wyss-Coray (2009) Mol Neurodegen
How can PB be cleared by autophagy? Classical mode of action – autophagic engulfment followed by autolysosomal degradation (by acidRapamycin maltase?) mTOR (through Ulk1/2 IM inhibition) MVB Autolysosome Autoph. Amphisome =LC3
Can the effects of rapamycin be reproduced if autophagosome maturation to autolysosomes is inhibited?Rapamycin mTOR (through Ulk1/2 IM inhibition) MVB Autolysosome Autoph. Amphisome Vin
Saponized neurons: Autophagic flux is fast. Rapamycin induces autopahgy and thus slows down autophagic flux. Vinblastine blocks autophagic maturation, further slowing down autophagic flux DMSO Rap Rap+Vin 300 200 100 0 LC3 LC3 LC3Confirmation of vinblastine’s effect:1. Reproduction of the block in autophagic flux induced by lysosomal protease inhibitors (PI).2. Blunting the sensitivity of rapamycin induced neurons to PI. * Fibroblasts Rap * Rap Rap Vin Starved DMSO Rap PI Vin PI Vin -LC3 I -LC3 II 1.1±0.2 1.5±0.1 2.0±0.2 2.2±0.1 1.9±0.3 1.7±0.2 6.2±0.5
Blocking autophagic maturation by vinblastine did notreverse down-modulation of polyglucosan Eaccumulation and apoptosis by rapamycin. Rapamycin protection of GBE1-knocked-down neurons did not depend on autophagic maturation and polyglucosan degradation in autolysosomesQ: Why does 3-MA antagonize Rap?A: Probably not via activation of autophagy or GSK3.Perhaps via PFK inhibitionC GFP Gly merge Rap+Vin D nt/Rap shGBE1/Rap shGBE1/Rap/3-MA shGBE1/Rap/VinFL-2 PI FL-1 Annexin V Fig. 5
As opposed to neurons, in APBD patient derived fibroblasts Rap probably does clears PB by autophagy Control Gbe1Y329S/Y329S patient Gbe1Y329S/Y329S/Rap Uranyl-Et-OH
Suggestion: Rap-stimulated PB clearance is mediated by exosome release, bypassing autophagosomal maturationRapamycin mTOR (through DGKα ? Ulk1/2 IM inhibition) MVB Autolysosome Autoph. Amphisome Vin
PBs were not found in either MVB (A), amphisomes (B&C), apparent exosomes (C), or cytosol (A-C) in GBE1-knockeddown neurons treated with rapamycin.A B Relatively small PBs were observed in untreated GBE1-knocked down neurons (D). Suggestion: Neurons not living with other cell types might succumb to cell death onceshGBE/Rap shGBE/Rap glycogen deposits appear Ca b DshGBE/Rap shGBE
?Rapamycin P active GSK3β GS mTOR X IM Autolysosome Autoph. MVB X Amphisome X Vin =LC3 =Polyglucosan Body
1.2Main conclusion: -G6PRap positive +G6P GS activity (nmol/min/mg Protein) 1effect wasprobably 0.8mediated by GSinhibition. 0.6 0.4None of thetreatments affected 0.2G-6-P-stimulated GSactivity, suggesting it 0overrode GS nt shGBE1 shGBE1 nt Rap shGBE1phosphorylation Rap Rap 3-MAstate. GS -75 -50 Tubulin
Candidate testing Testing three types of compounds known to reduce the GS/GBE ratio: 1. GS inhibitors (AMP Kinase (AMPK) and GSK3β activators). Examples: 5- Aminoimidazole-4-carboxamide-1-β-D-ribofuranoside (AICAR), PI3K inhibitors (e.g., wortmanin, Akt inhibitor IV), Berberine (herbal drug) etc. 2. GBE stabilizers. 3. Compounds predicted by solvent mapping to replace mutated Tyr329 in GBE1, or to destabilize GS.
Another direction: Microtubule-mediated transport :A nerve biopsy from Hereditary Spastic Paraplegia reveals PB.Linkage analysis along chromosome 2 shows the locus where markers for the diseased stateco-segregated only in patients and not in healthy family members. This locus encodes theMT motor KIF1A Do microtubule motors LOD Score mediate PB clearance?
Summary•We have shown that GBE1 knockdown causespolyglucosan accumulation and apoptosis in apure neuronal model.•These phenotypes can be rescued byrapamycin via inhibition of Glycogen Synthaseand not via induction of autophagy•We conclude that polyglucosan accumulationis causal for APBD. Therapeutic search shouldtherefore focus on restricting polyglucosanaccumulation.
Main objectives•Establishing a neuronal model of APBD in which GBE1 isrepressed and PB are observed.•Using the model to test pharmacological and biochemicalmethods for correcting adverse phenotypes associated withGBE1 deficiency.
APBD Neuronal Model Produced by transduction with lentiviral particles encoding for shRNA against GBE1 Demonstration of GBE1 knockdown: A RT-PCR & activity 300 1.2 GBE1 mRNA (relative values) GBE1 activity 250 1(nmol/min/mg protein) GBE1 mRNA Indirect GBE1 Activity 200 0.8 immunofluorescence 150 0.6 C GFP GBE1 merge 100 0.4 50 0.2 0 0 nt nt shGBE1 Western: Reduction comparable to that found in patients homozygous shGBE1 shGBE1 B Control Y329S nt GBE1 -75 kDTubulin -50 kD
Polyglucosan accumulates in GBE1 knocked down neurons Glycogen detected as polyglucosan punctae similar to GSK3 inhibition or PTG over expression. Lower expressers of shGBE1 lentiviruses are less affected: Polyglucosan is the culprit GFP Gly merge A nt B shGBE1 C shGBE1 GFP DIC Gly mergeVilchez et al (2007) Nat Neurosc
A GBE1 knockdown increases apoptosis B ntC FL-2 PI shGBE1D SNP FL-1 Annexin V