Embed Size (px)
Citation preview
Human-based Seizurogenic Assays: Using iPSC-derived Glutamatergic Neurons to Assess Drug-induced Changes in Network Activity
Blake Anson, PhD
Global Bus iness D i rector
Tox ico logy and Safe ty Pharmacology
Chr is S t rock , PhD
Associa te D i rector, B io logica l
Screening and Assay Deve lopment
Cyprotex US, LLC
September 26 , 2017
October 9, 2017 2
Outline – Cellular Dynamics and iCell Glutaneurons
• CDI Company Overview
• iCell Glutatneurons
Cell Identity
Use on the MultiElectrode Array
October 9, 2017 3
FCDI Driving Expertise and Technology
Bringing relevant human biology and diversity to drug discovery
October 9, 2017 4
Company Overview
• Cellular Dynamics International (CDI) is a leader in production and application of human iPS cells and iPS cell-derived cell types
• Acquired by FUJIFILM (4/2015); International presence
Headquartered in Madison, WI (additional site in Novato, CA)
Application / Distribution sites in Japan, South Korea, and Tilburg
Local Sales and FAS support
• Currently employs ~175 total staff w/ >900yrs cumulative stem cell and differentiation experience
• >100 patents (owned or licensed)
• Portfolio includes off the shelf products, as well as, custom cell production and assay services.
October 9, 2017 5
CDI: Life Sciences and Regenerative Medicine
2 Business Divisionso Life Sciences
o Cellular Therapeutics
Life Sciences Division serves 4 major market areaso Basic and Translational Sciences
o Safety Pharmacology & Toxicity
o Drug Discovery and Bioengineering
o Specialty Markets
Cellular Therapeutics Division has 2 focal areaso Internal cell therapy programs
o Ocular, Cardiac, Neurodegenerative, and Oncology
o Contract Development and Manufacturing partnerships
An unyielding commitment to consistent and robust iPSC-based
solutions for current and future research and therapeutic applications
October 9, 2017 6
Life Science Research: Current Product Portfolio
iCell
Cardiomyocytes
iCell
Cardiac
Progenitor Cells
iCell
Neurons
iCell
DopaNeuronsiCell
Astrocytes
iCell
Motor NeuronsiCell
GlutaNeurons
iCell
HepatocytesiCell
Macrophages
iCell
Hepatoblasts
New Products
in development:
iCell RPEs
Early access availability
iCell
HPCs
iCell
Endothelial Cells
iCell
Skeletal MyoblastsMSCs
October 9, 2017 7
Human iPSC ModelsFunctional recapitulation
Neurite outgrowth / retraction
Synaptogenesis / pruning
Ion channel and synaptic activity
Seizurogenesis
Neurons
Electrical activity
Ca2+ handling
Contractility
Cardiomyocytes
October 9, 2017 8
Human Glutamatergic Neurons for Basic and Applied Research
Neurotoxic and protective compound screening
Safety pharmacology
Seizurogenic models
Formation of predominately excitatory neural networks make iCell GlutaNeurons a
powerful tool for drug screening, excitotoxicity testing and basic research in the context
of understanding and treating neurodegenerative and seizurogenic disorders.
Mechanisms of neurodegenerative diseases
Potential utility of iPSC-derived GlutaNeurons
Popoli et al., 2012
October 9, 2017 9
80K Cells/well80K Cells/well
High viability with bipolar or multipolar neurite outgrowths within days of plating
iCell GlutaNeuronsCharacterization
October 9, 2017 10
Characterization of iCell GlutaNeuronsPurity and Identity
Nestin
MA
P2
Nestin
TU
J
iCell GlutaNeurons are > 95% neuronal purity with ≥70%
Glutamatergic identity
Neuronal Identity Subtype Identity
October 9, 2017 11
Characterization of iCell GlutaNeuronsElectrophysiology on a Multi-Electrode Array System
iCell GlutaNeurons fire spontaneous action potentials
and form neural networks that burst synchronously
Single well
Individual
Electrodes
Periods of synchronized ‘firing’ across the well
October 9, 2017 12
iCell GlutaNeurons CDI exampleRobust Activity
48-well Plate (5X8: no data from the bottom row)
‘Velocity Graphs’
Ra
ste
r P
lot
Ra
ste
r P
lot
Ve
loc
ity
Gra
ph
Ve
loc
ity
Gra
ph
= all-points histogram
binned @ 500 msec
October 9, 2017 14
iCell GlutaneuronsMonitoring electrical activity
Intercellular
(synaptic)
Activity
Intracellular
(ion channel)
Activity
iCell GlutaNeurons can be
used as a model to assess
changes in ion channel,
synaptic, and network activity
October 9, 2017 15
Astrocytes significantly altered the network burst phenotype of GlutaNeurons (data from a CytoView MEA 48).
Picrotoxin
(100 µM)
Changes in network burst phenotype with 100
µM Picrotoxin for GlutaNeuron + Astrocyte co-
culture (n=6).
GlutaGluta
+Astro
iCell Glutaneurons
October 9, 2017 16
Vehicle66.7 33.3 16.7 8.3 4.1 2.2 1.1
Pre
-Dru
giCell GlutaNeurons: Gabazine [µM]
48-well Plate (5X8: no data from the bottom row)
- Velocity Graphs Presented from 2 rows -1 H
rP
ost
GB
Z
October 9, 2017 17
iCell GlutaNeurons DIV28: Gabazine [µM]
% Change from Baseline
(WITHIN Subjects Design)*
-100
-50
0
50
100
150
200
250
300
350GBZ 1Hr
Ctrl 1 µM 2 µM 4 µM 8 µM 17 µM 33 µM 67 µM
% C
ha
ng
e F
rom
Ba
se
lin
e
Ctrl
2 µM
8 µM
33 µM
-100
-50
0
50
100
150
200
250
300
350
GBZ 1Hr
Ctrl 1 µM 2 µM 4 µM 8 µM 17 µM 33 µM 67 µM
Note Hyper-Sensitivity:
- 1 µM (lowest dose) displays
significant change from
Baseline in Mono-cultures
‘NA’ = In-House Network Bursts statistics
‘chan’ = Axion channel ‘Poisson’ statistics
‘n’ = Axion Network Burst statistics
‘Synchrony’ = Axion ‘synchrony index’ statistic
= Axion & HESI’s most notable metric
3D Version
October 9, 2017 18
-200 -150 -100 -50 50 100 150 200
-200
-100
100
200
R2 = 0.76
Post WashoutPTX
PTX
D22
D200
D1800
-100
-50
0
50
100
150
200
250
300
mnFRs ActEsNA BPM
NA ConnchanB
chanBspkchanBDur
nBnBurstPerc
nBurstDurnElect
Synchrony
-100
100
300% Change
from Baseline
SpikesNetwork
‘Velocity’
Network‘ISI’
Channel‘Poisson’
SynchronyIndex
% C
ha
ng
e f
rom
Baselin
e
Network Burst
Frequency (BPM)
Poisson Burst
Intensity (Hz)
Post WashoutPTX (5 Minutes)
Vehicle66.7 33.3 16.7 8.3 4.1 2.2 1.11 H
ou
r P
ost
GB
Z (
µM
)
iCell GlutaNeurons: Seizurogenic Assay
October 9, 2017 19
iCell GlutaNeuronsExtended Pharmacology
iCell GlutaNeurons
Excitatory neuronal population
Seizurogenic model
Tested across chemical space
Utilized across multiple sites
Under evaluation within HESI
Translational Biomarkers of Neurotoxicity
October 9, 2017 20
iCell Glutaneurons
Recapitulate native behavior
Exhibit bursting, synchronized electrical activity
Can be mixed with other neuronal and support cell types
Useful model for predicting drug-induced seizurogenic effects
www.cellulardynamics.com
Human-Based Seizurogenic Assays: Using iPSC-Derived Glutamatergic Neurons to Assess Drug-Induced Changes in Network Activity
Cyprotex, Safety Pharmacology Society– Tuesday, September 22, 2017
PAGE
Cyprotex Overview
2
Focused on support for pharmaceutical industry, agrochemical industry,
tobacco industry, cosmetics and personal care industry and chemical industry
• Founded in 1999
• Acquired in 2016, wholly owned subsidiary of Evotec AG
−Cyprotex laboratories in Alderley Park (UK), Kalamazoo and Boston (US)
−Over 1,600 customers worldwide with exceptional client retention rate
−Focus on ADME-Tox/PK services
In vitro ADME-Tox screening
In vitro regulatory ADME-Tox services
Contract Research Organisation providing ADME-Tox, DMPK & Biosciences services
PAGE
Cyprotex is now part of Evotec AG
3
Evotec’s worldwide operations
San Francisco, Branford & Princeton, Watertown, Kalamazoo, USA
~105 employees
• Compound ID, selection and acquisition
• Compound QC, storage and distribution
• Cell & protein production
• ADME-Tox, DMPK
Abingdon, Manchester, Alderley Park, UK
~400 employees
• Medicinal chemistry
• ADMET-Tox, DMPK
• Structural biology
• In vitro & in vivo anti-infective platform / screening
Hamburg (HQ), Göttingen and Munich, Germany~405 employees
• Hit identification
• In vitro & in vivo biology
• Chemical proteomics & Biomarker discovery and validation
• Cell & protein production
• Antibody discovery
Toulouse, France
~230 employees
• Compound management
• Hit identification
• In vitro & in vivo oncology
• Medicinal chemistry
• ADME & PK
• Early drug formulation andsolid form screening
• Cell, protein & antibody production
PAGE
Cyprotex Capabilities
4
• Consultative interactive approach for small biotech and start-ups – guidance
• Combination of in vitro data and in silico modelling to maximise value of data and
predict clinical outcome – not just a data provider
• Responsive, agile, focused on delivery and quality
− Response to enquiries within 2 hours
− Full study proposal provided within 24 hours
− Fast turnaround times
• Knowledge sharing and research
− Internal R&D, publications and posters
− Development of novel approaches and solutions in conjunction with clients
Unique Technologies …
− Cyprotex free educational guides (ADME, Tox, DDI and Chemical and Cosmetics
Testing guides) www.cyprotex.com/guides
− Presentations, webinars, training sessions
Our Differentiators
PAGE
eCiphrNeuro: Predicting Neurotoxicity
• Assay for neurotoxicity developed using primary rat cortical neurons
− Capable of predicting seizure inducing compounds as well as other
neuroactives and neurotoxins
− Phenotypic assay based on predicting seizure causing compounds
by burst and firing pattern analysis.
• The primary rat cortical neurons have sufficient baseline activity, can
network with each other, and are inducible with test agents
− Able to measure spontaneous action potentials in neurons and
determine what changes a compound may cause
− Same distribution of cells and neurochemistry as found in native
brain.
− Cells isolated with a native dissociation so the cell distribution is
not altered
− In addition to neurons, you have astrocytes(glia) and microglia also
present
PAGE
Maestro Platform
The Maestro
Axion’s 768 channel system768 stimulating and recording channels
SBS-Compliant Multiwell MEA plates
Accommodates 12, 48 and 96 wells
Fast MEA insertion with auto MEA-plate identification
Ultra-thin, transparent, & affordable MEA plates
Fully integrated heater with software controls
Automated electrode characterization & diagnostics
Higher throughput 48-well configuration
16 low-noise microelectrodes per well
4 integrated ground electrodes per well
Polymer (Kapton) insulation
Nano-textured Gold electrodes
ANSI compliant well plates
Evaporation-reducing lid
48-Well
PAGE
Characteristics of Axion Biosystem’s Array
•Grid of tightly spaced electrodes, and each
electrode is capable of simultaneously
monitoring and stimulating the activity of >12
cells
•The arrangement of multiple electrodes in a
grid extends the recording range across a
relatively large area, providing concurrent
access to both single cell and network-level
activity
•The control and monitoring of this cellular
activity is made possible by the electronics,
which impart multiple functions to each
electrode
Axion Biosystem’s Array (electrodes are 30 μm
in diameter and are spaced 200 μm apart; this
image is representative of a 12-well system)
PAGE
Cells growing on MEA
PAGE
Criteria for monitoring of neural activity
The advantages of MEA
• Label-free, non-invasive operation to observe natural cell function
• Preservation of cellular interconnectivity
• Real-time measurement of the phenotypic signal: voltage
• Sufficient resolution to capture action potentials
• Multiple recording sites to assay action potential propagation
PAGE
Can we make a human seizure prediction assay?
• Although rat cortical neurons are highly predictive of neurotoxicity, there is a desire in the industry to have a human model for
prediction
• With the advent of stem cell derived neurons, this became a possibility although early hiPSC derived neuronal models lacked
complex burst organization, making electrophysiological neurotoxic prediction challenging when utilizing an MEA platform
• Recent advancements in hiPSC neuronal models have addressed these disadvantages with the introduction of multiple neuronal
cell types.
• CDI has created iCell Glutaneurons, iPS cell-derived human cortical neurons consisting primarily of 70% glutamatergic (excitatory)
neurons
• We will show here our initial evaluation of these cells, both alone as well as with iCell Astrocytes and iCell Neurons.
PAGE
Burst Endpoint Characterization with custom Matlab script
Individual spikes
Irregular burst duration
and organization
Irregular interburst
intervals
•Longer burst durations
•Regular interburst intervals
•Increased burst organization
•Increase in burst duration regularity
Most spikes occur
within burstsPicrotoxin
Post-Dose
PAGE
Synchrony Endpoint
Changes in synchronous activity
are observed below Picrotoxin Post-Dose
PAGE
Endpoints and what they mean
•Mean Firing Rate (MFR) is a simple and effective MEA metric to identify neurotoxicity: Proves not to be very effective
•Burst Characteristics−Burst Rate - Number of bursts normalized by time of the recording.
−Number of Spikes in Burst.
−Coefficient of variation (CV) of the inter-spike intervals (ISI) - Burstiness.
−Normalized IQR Burst Duration – Burst Duration Regularity
−Burst Duration
− Interburst Interval
−Normalized MAD Burst Spike Number – Indicator of statistical dispersion of the spikes in bursts
−Median ISI/Mean ISI– Measure of spike organization within bursts. Increases as spike/burst organization deteriorates.
•Synchrony of spikes and bursts− ISI-distance - ISI-distance is calculated by the Kreuz et al. method for spike
train synchrony.
PAGE
Methods
48-well MEA plates were pre-coated with a 0.07% PEI solution, rinsed and allowed to dry overnight.
iCell GlutaNeurons and iCell Astrocytes were rapidly thawed and slowly diluted (to avoid osmotic shock) with BrainPhys Neuronal Medium
supplemented with iCell DopaNeurons Supplement, iCell Nervous System Supplement, N2 supplement, Laminin and Penicillin/Streptomycin.
After a gentle centrifugation step, the cells were resuspended at the appropriate density with cell dotting medium (complete BrainPhys
Medium supplemented with additional laminin).
A 10µL droplet of a cell suspension containing 120K iCell GlutaNeurons and 20K iCell Astrocytes was dispensed directly over the electrode
grid of each well of a 48-well MEA plate.
The cells were incubated, humidified at 37°C in 5% CO2 for 1 hour.
500 µL of complete BrainPhys medium was slowly added to each well in a 2-step process to avoid detaching the cells.
Cells were maintained for 14-18 days by changing 50% medium 3 times a week.
Recordings were acquired on the Axion Biosystems’ Maestro periodically throughout the maintenance period to document the maturation
process.
Recordings were also acquired immediately before compound treatment (baseline) and 1 hour post treatment. All experiments were
performed 14-18 days after plating.
Custom MATLAB scripts were used to analyze the spike trains. Endpoints reported include: firing rate, burst rate, number of spikes in bursts,
percent isolated spikes, ISI CV, normalized IQR burst duration, burst duration, interburst interval, mean of ISI-distance, IQR/median ISI,
skewness ISI, normalized MAD burst spike number, median/mean ISI and median ISI.
Raster plots were generated with Axion BioSystems’ Neural Metric Tool.
PAGE
Maturation: iCell GlutaNeurons (90%)
Day 4,12 and 15
PAGE
Maturation: iCell GlutaNeurons (90%) / iCell Astrocytes
Day 6, 9 and 15
PAGE
Maturation: iCell GlutaNeurons (70%) / iCell Astrocytes
10 days3 days 14 days
PAGE
Maturation of Firing Summary
• iCell Glutaneurons show great promise as a human iPS cells with good burst organization and network organization
− They start out with significant firing activity from day 1 with no real network organization
− After 9 days you start to see some bursting
− Soon after this networks start to form and you see synchrony develop
• Addition of astrocytes appears to help to speed up the maturation process
− Firing patterns appear to be more mature at same time point
− Firing rate may be a little lower but in our experience organization of the firing is the real key
• Analysis of data on MEA
− Due to the significant firing rates and the long bursting phenotype, adjustments needed to be made to analysis
− Need to set the spike threshold to a fixed amplitude because the long bursts tend to cause the adaptive threshold to slide up
because it begins to see them as noise
− We have had to focus on identifying some new endpoints as our original endpoints developed for rat neurons are not always
effective with the patterns generated with iCell Glutaneurons
• Comparison of the 70% Glutaneurons to the 90% Glutaneurons
− The 70% appear to mature in a way that makes them look more similar to the maturation process of the rat neurons.
PAGE
Pharmacology: Neurotoxicity Prediction
PAGE
Compounds and Mechanisms
• Picrotoxin – GABA A antagonist
• Bicuculine – GABA A antagonist
• Strychnine – glycine receptor
• 4-aminopyridine – Potassium channel blocker
• SNC80 – delta opioid receptor
• Pilocarpine - Muscarinic acetylcholine receptor M3.
PAGE
Pharmacology: iCell GlutaNeurons:Astrocytes
0.2% DMSO (vehicle)
Baseline 0.2% DMSO
PAGE
Pharmacology: 4-aminopyridine
PAGE
Pharmacology: SNC80
PAGE
Pharmacology: Strychnine
PAGE
Pharmacology: Picrotoxin
PAGE
Pharmacology: Bicuculline
PAGE
Pharmacology: Pilocarpine
PAGE
Optimization of the assay
• Further Optimization may be required for timing of the assay
• Most of the assays have been run between days 14 and 17 depending on maturation.
• Even on fully synchronized plates, the GABA A antagonists still had measurable effects. Compounds that break up synchrony may
benefit from more synchronized cells. We will further characterize and optimize these responses
• Our analysis tools may need to be modified. The amplitude and dynamics of the firing are somewhat different that the rat neurons
that we developed our Matlab scripts on. This may require further adjustments of our scripts.
• We may look again at the newer versions of the Axion software to see if this can be a fit for purpose.
PAGE
GABA A Antagonists
• In the cells that were 90% glutamatergic, we couldn’t definitively identify a seizure type response with these cells. Probably
because they didn’t have enough Gabaergic cells.
• The 70% glutamatergic cells now respond well to the gabaergic compounds and give a robust reproducible response
PAGE
Pharmacology summary
• iCell Glutaneurons have significant and robust responses to many compounds with seizure liability
• GABA A antagonists have a robust response consistent with what we see in rat cortical neurons
− In the cells that were 90% glutamatergic, we couldn’t definitively identify a seizure type response with these cells. Probably
because they didn’t have enough Gabaergic cells.
− The 70% glutamatergic cells now respond well to the gabaergic compounds and give a robust reproducible response
• For other non GABA A antagonist targets, we get robust responses with all combinations of cell types tested.
− Astrocytes appear to enhance the response to the compounds
− Different compounds give different firing phenotypes. It will be interesting to see if these phenotypes correspond to a target that is
engaged
• Steps moving forward
− Test more compounds
− Repeat compounds already tested to verify the robustness of the responses
− Modify our analysis software to make sure we catch all of the finer details of the responses to compounds
PAGE
Acknowledgements
• Cyprotex
− Jenifer Bradley
• CDI
− Blake Anson
− Kile Mangan
