Utilization of iPSC-derived human neurons for high-throughput drug-induced peripheral neuropathy screening

Toxicology in Vitro 45 (2017) 111–118

Contents lists available at ScienceDirect

Toxicology in Vitro journal homepage: www.elsevier.com/locate/toxinvit

Utilization of iPSC-derived human neurons for high-throughput druginduced peripheral neuropathy screening

MARK

Payal Ranaa, Gregory Luermanb, Dietmar Hessb, Elizabeth Rubitskia, Karissa Adkinsa, Christopher Sompsa,⁎ a b

Drug Safety Research and Development, Pfizer, Eastern Point Road, Groton, CT, United States Axiogenesis AG, Cologne, Germany

A R T I C L E I N F O

A B S T R A C T

Keywords: Neurotoxicity Stem cells Screening

As the number of cancer survivors continues to grow, awareness of long-term toxicities and impact on quality of life after chemotherapy treatment in cancer survivors has intensified. Chemotherapy-induced peripheral neuropathy (CIPN) is one of the most common side effects of modern chemotherapy. Animal models are used to study peripheral neuropathy and predict human risk; however, such models are labor-intensive and limited translatability between species has become a major challenge. Moreover, the mechanisms underlying CIPN have not been precisely determined and few human neuronal models to study CIPN exist. Here, we have developed a high-throughput drug-induced neurotoxicity screening model using human iPSC-derived peripheral-like neurons to study the effect of chemotherapy agents on neuronal health and morphology using high content imaging measurements (neurite length and neuronal cell viability). We utilized this model to test various classes of chemotherapeutic agents with known clinical liability to cause peripheral neuropathy such as platinum agents, taxanes, vinca alkaloids, proteasome inhibitors, and anti-angiogenic compounds. The model was sensitive to compounds that cause interference in microtubule dynamics, especially the taxane, epothilone, and vinca alkaloids. Conversely, the model was not sensitive to platinum and anti-angiogenic chemotherapeutics; compounds that are not reported to act directly on neuronal processes. In summary, we believe this model has utility for high-throughput screening and prediction of human risk for CIPN for novel chemotherapeutics.

1. Introduction Due to advances in cancer diagnosis and treatment, there were an estimated 12 million cancer survivors in 2012 in the United States (Elena et al., 2013) and 67% of U.S. cancer patients are surviving at least 5 years (Park et al., 2013). Therefore, addressing side effects associated with cancer treatment is of significant importance due to the potential impact of these effects on quality of life (Fehrenbacher, 2015). Chemotherapy-induced peripheral neuropathy (CIPN) is a major side effect which can result in irreversible symptoms and disability in up to 40% of cancer survivors (Wolf et al., 2008). CIPN has been observed following treatment for a number of cancers, including breast, colorectal, and testicular cancers, and is associated with a number of chemotherapeutic mechanisms of action (Balayssac et al., 2011; Pike et al., 2012). Symptoms associated with peripheral nerve damage often include numbness in the limbs, as well as more severe symptoms such as stabbing and burning pain and extreme sensitivity to touch (Argyriou et al., 2011; Binda et al., 2013). Due to these side effects, patients may

⁎

Corresponding author. E-mail address: christopher.j.somps@pfizer.com (C. Somps).

http://dx.doi.org/10.1016/j.tiv.2017.08.014 Received 3 April 2017; Received in revised form 17 July 2017; Accepted 21 August 2017 Available online 24 August 2017 0887-2333/ © 2017 Elsevier Ltd. All rights reserved.

decide to discontinue the chemotherapeutic regimen or be forced to switch to less effective treatments. As a consequence, there is a critical need to understand pathophysiological mechanisms of CIPN, optimize clinical assessment, and develop neuroprotective strategies. Additionally, within the pharmaceutical industry, there is a need to identify and develop new chemotherapeutics that minimize risk of peripheral neuropathy. Preclinical studies in animals may be used to evaluate peripheral neuropathy risk during early drug development. However, peripheral neuropathy is not easily detected in animals, especially with standard endpoints such as histopathology assessment of nervous system tissue. Functional assessments, such as behavioral and specialized neurophysiological endpoints (Marmiroli et al., 2012), are time consuming and have limited translatability to humans. Limited translatability of animal models may be due to lack of consistency in dose or mode of delivery of drugs, or the sex, age, and genetic background of the animals used in the studies (Hoke and Ray, 2014). Additionally, in vivo models are labor intensive, require significant amounts of drug for testing, and are

Toxicology in Vitro 45 (2017) 111–118

P. Rana et al.

A

DIV 2

B

Hoechst

MAP2

Peripherin

Merge

DIV 4

Fig. 1. Peri.4U neurons were cultured up to 14 days in vitro (DIV) and light microscope images show the typical neurite morphology for 2, 4 and 14 DIV (A). 7 DIV Peri.4U neurons were characterized by fluorescence microscopy for MAP2 and peripherin expression. Hoechst staining shows nuclear localization. Merged images demonstrate largely overlapping expression of MAP2 and peripherin in the soma and neurites (B).

DIV 14

Peri.4U neurons. Our work confirms recent reports (Hoelting et al., 2016; Wing et al., 2017) that stem cell-derived neurotoxicity models can be used to predict CIPN liability, and includes results for additional positive and negative chemotherapeutic drugs (N = 19). Our results suggest that this assay could be positioned early in the drug development process to identify compounds with risk of peripheral neuropathy (Argyriou et al., 2011; Argyriou et al., 2014).

very low throughput. Consequently, various cell lines, such as human neuroblastoma cells (SH-SY5Y) and rat PC12 pheochromocytoma cells, have been used for in vitro neurotoxicity assessment (Slotkin et al., 2007). However, these cell line models, which are often differentiated during culture into “neuronal like” cells, lack a “true” neuronal phenotype (Marmiroli et al., 2012). The U.S. National Research Council and major regulatory agencies have recommended new strategies for toxicity testing (Tox21) based on high-throughput testing in more physiologically relevant cell-based systems, ideally of human origin (Collins et al., 2008; Tice et al., 2013). Consistent with this strategy, a human embryonic stem cell-derived, peripheral neuron model has been recently shown to predict peripheral neurotoxicants with high sensitivity and specificity (Hoelting et al., 2016). In vitro strategies for predicting CIPN implicitly require a reproducible source of well-characterized human neurons. Large quantities of peripheral-like neurons from a single human inducible pluripotent stem cell (iPSC) line are commercially available for assay development and safety screening (Peri.4U neurons, Axiogenesis, Cologne, Germany). These neurons are derived using a protocol (modified from Chambers et al., 2012) that produces neurons expressing a number of peripheral markers. However, Peri.4U neurons have not been thoroughly characterized in the literature. We therefore performed a limited characterization of their neuronal protein expression including markers such as peripherin, microtubule-associated protein 2 (MAP2), and beta III tubulin, as well as their electrophysiological properties. In the assay described here, we used the Peri.4U neurons with high content image analysis to predict CIPN related toxicities associated with multiple classes of chemotherapeutics, including platinum agents, taxanes, vinca alkaloids, proteasome inhibitors, and anti-angiogenic drugs. For comparison, we also evaluated a “neuronal” cell line, the NG108-15 cell, using the taxane vincristine to demonstrate the benefit of using the

2. Materials and methods 2.1. Materials Cisplatin (Cat no 1134357), Carboplatin (Cat no C2538), Lenalidomide (Cat no CD5022536), Paclitaxel (Cat no T7402), Docetaxel (Cat no 01885), and Vincristine (Cat no V8879) were purchased from Sigma-Aldrich (St. Louis, MO). Sagopilone (cat no U83388) was purchased from Adooq biosciences (Irvine, CA). Ixabepilone (Cat no HY-10222/CS-0551) was purchased from Medchem express (Monmouth Junction, NJ). The rest of the chemicals were obtained from Pfizer's chemical inventory (Groton, CT) and were > 95% purity and in free base form. All cell culture reagents were purchased from either Sigma or Thermo Fisher Scientific (Waltham, MA) unless otherwise noted. Human Peri.4U neurons, growth media, and supplements were obtained from Axiogenesis, AG (Cologne, Germany). A hybrid mouse neuroblastoma and rat glioma cell line, NG108-15, was purchased from ATCC (Manassas, VA). 2.2. Methods 2.2.1. Cell culture Peri.4U neurons derived from human induced pluripotent stem cells were cultured and maintained per the manufacturer's protocol. Briefly, 112

Toxicology in Vitro 45 (2017) 111–118

P. Rana et al.

Fig. 2. Raw data traces from single Peri.4U neurons manually patch clamped at times between 8 and 19 DIV demonstrate a voltage-dependent TTX-sensitive inward (sharp downward deflections on left of top panel) and at least two types (transient and non-inactivating) of outward currents (longer duration upward deflections on right of top panel) (A). The representative cell currents were achieved using the described voltage step protocol in voltage clamp mode (lower panel). TTX (100 nM) blocked the entire inward current (middle panel), leaving outward currents unaffected. Evoked action potentials were elicited with current injections (1 s, 20 pA) (B). Spontaneous action potentials were also readily observed when neurons were plated at high density (C).

prepared in a 5-fold dilution scheme. The concentration ranges of cisplatin, oxaloplatin, carboplatin, thalidomide, lenalidomide, pomalidomide, K252b, bortezomib, suramin, amoxicillin, sorbitol, and hydroxyurea were 0.0013 to 100 μM. The concentration ranges of paclitaxel, docetaxel, ixabepilone, and sagopilone were 0.00013 to 10 μM. The concentration ranges of nocodazole and colchicine were 0.0004 to 30 μM. Finally, the concentration range of vincristine was 0.00001 to 1 μM. The final concentration of DMSO in the culture media was 0.3% (non-toxic dose; data not shown). Compounds were screened in duplicate at each concentration on the same plate.

96-well plates (Corning, Inc., Corning, NY) were coated with matrigel (100 μg/mL) in DMEM:F12 media for 2 h (37 °C). After thawing at 37 °C, cells were suspended in serum-free DMEM:F12 basal medium supplemented with Peri.4U supplement cocktail (Axiogenesis). Cells were plated at densities ranging from 10,000 to 15,000 cells/well based on the number of live cells counted. Cells were maintained in a humidified incubator at 37 °C with a 95% air/5% CO2 atmosphere for 48 h prior to compound treatment. NG108-15 cells were maintained in growth medium composed of DMEM (Invitrogen, Carlsbad, CA) with 10% fetal bovine serum (FBS; Invitrogen) and 1% penicillin-streptomycin (P/S; Invitrogen), and supplemented with 0.1 mM hypoxanthine, 400 nM aminopterin, and 16 μM thymidine (HAT; Sigma-Aldrich, St Louis, MO). The 96-well plates were coated with a solution of poly-L-lysine (15 μg/mL) in PBS for at least 2 h at 37 °C. NG108-15 cells were plated at densities ranging from 2000 to 2500 cells/well based on the number of live cells counted. Cells were maintained in a humidified incubator at 37 °C with a 95% air/5% CO2 atmosphere for 48 h prior to compound treatment.

2.2.3. Immunocytochemistry For initial characterization using light microscopy and immunofluorescence (IF) experiments, cells were plated as above for up to 14 DIV (days in vitro) or 7 DIV, respectively. Cells were then fixed and incubated for 20 min with 0.1% Triton X-100 in PBS, followed by blocking with 10% goat serum in Tris-buffered saline containing 0.1% Tween-20 for 1 h. Fixed cells were incubated overnight at 4 °C with Hoechst 33342 nuclear stain (Thermo Fisher, Cat no 62249), anti-MAP2 (Millipore MAB3418) and anti-peripherin (Millipore AB1530). After two washes in Tris-buffered saline containing 0.1%; Tween-20, Alexa Fluor conjugated secondary antibodies (Thermo Fisher, goat anti-mouse A11031 or goat anti-rabbit A11008) were added at 1:500, again in antibody diluent for 2 h at RT. Finally, two washes in Tris-buffered saline containing 0.1% Tween-20 and one wash with Tris-buffered

2.2.2. Preparation of compounds All compounds were dissolved in 100% DMSO in 30 mM stocks. At 30 mM, we did not observe insolubility and precipitation with any compounds upon visual inspection. Compound plates containing stocks were stored at room temperature in inert nitrogen boxes for up to 2 weeks or at − 20 °C for 3 months. Test compounds for assays were 113

Toxicology in Vitro 45 (2017) 111–118

P. Rana et al.

A

B

120

120

100

100 80

% C o n tr o l

% C o n tr o l

80 60 40

60 40

20

20

0

0

-2 0 0 .0 0 0 0 0 1

0 .0 0 0 1

0 .0 1 V in c r is tin e ( M )

-2 0 0 .0 0 0 1

1

0 .0 0 1

N e u ro n C o u n t p e r S e le c te d F ie ld (% )

0 .0 1

0 .1 V in c r is tin e (

1

10

100

M)

N e u ro n c o u n t p e r s e le c te d fie ld (% )

N e u rite T o ta l L e n g th p e r N e u ro n (% )

N e u rite T o ta l L e n g th p e r N e u ro n (% )

C

D

Fig. 3. Representative images of untreated Peri.4U (A) and NG108-15 (B) cells stained with DAPI (blue, nuclei) and Beta 3-tubulin antibody (red, neuronal cell bodies and neurites). Image analyses of neurites (Cellomics, Neuronal Profiling application) are shown in green. Note the greater number of neurites for the Peri.4U cells. The percentage change in neuron count per selected field and neurite total length per neuron are shown for Peri.4U (C) and NG108-15 (D) cells treated with increasing concentrations of vincristine for 24 h. Curves were fit using a non-linear regression algorithm (GraphPad Prism) constrained between 0 and 100% (of time-match controls).

Assay Schematic Peri.4u cells

48 hrs

Exposure Vehicle (0.33% DMSO) or each Compound (N=2)

10,000 cells/well

24 hrs

Inhibition of Neurite Outgrowth High Content Imaging of Beta III tubulin and Neuronal cells

Neuronal Cytotoxity ATP level (PromegaCell titer glo)

Fig. 4. Peri.4U neurons were seeded at 10,000 cells per well, incubated for 48 h, then treated with test articles for 24 h and finally evaluated for neurite length and cytotoxicity endpoints.

blocked with 3% BSA (Bovine Serum Albumin) for 30 min. Blocking buffer was then gently aspirated and primary antibody βIII-tubulin diluted in 1% BSA (1:500) was added for 2 h at room temperature. Following incubation of primary antibody, cells were washed three times with PBS and incubated with 2 μg/mL AlexaFluor 488-conjugated rabbit anti-goat secondary antibody in 1% BSA for 1 h at room temperature, protected from light. After washing the cells once with PBS, 2 μM DAPI (4′, 6-diamidino-2-phenylindole) was added for 15 min at

saline were performed prior to imaging. For neurite length experiments, cellular media was removed from the plates and cells were fixed in situ with a 100 μL solution of 8% paraformaldehyde in PBS for 20 min. Fixative was then gently aspirated and cells were washed three times with PBS (Dulbecco's phosphatebuffered saline). Cells were then permeabilized using 0.1% triton X-100 in PBS for 20 min. Permeabilization reagent was gently aspirated and cells were washed three times with PBS. Unreactive binding sites were 114

Toxicology in Vitro 45 (2017) 111–118

P. Rana et al.

120

120 100

100

% C o n tr o l

A

% C o n tr o l

80

80 60 40

60

B

40 20

20 0

0 -2 0

-2 0

0 .0 0 0 1

0 .0 0 0 1

0 .0 1 P a c lit a x e l (

1

0 .0 1 I x a b e p il o n e (

100

N e u r ite T o ta l L e n g th p e r N e u r o n ( % )

120

120

100

100

80

80

% C o n tr o l

% C o n tro l

100

N e u r o n C o u n t p e r S e le c te d F ie ld ( % )

N e u r o n C o u n t p e r S e le c te d F ie ld ( % ) N e u rite T o ta l L e n g th p e r N e u ro n (% )

C

1 M)

M)

60 40

D

60 40 20

20

0

0

-2 0

-2 0 0 .0 1

1 C o lc h ic in e ( M )

0 .0 1

100

1 S o r b it o l (

100 M)

N e u r o n C o u n t p e r S e le c te d F ie ld ( % )

N e u r o n C o u n t p e r S e le c te d F ie ld ( % )

N e u rite T o ta l L e n g th p e r N e u ro n (% )

N e u r ite T o ta l L e n g th p e r N e u r o n ( % )

Fig. 5. The percentage change in neuron count per selected field and neurite total length per neuron of Peri.4U cells treated with paclitaxel (A), ixabepilone (B), colchicine (C) and sorbitol (D) for 24 h. Each data point represents the mean ± SEM, n = 3 separate experiments.

Table 1 The dose response values (IC50) of cell viability (ATP depletion), neuron count per field, and neurite total length per neuron for Peri.4U cells treated with various classes of chemotherapy drugs for 24 h. Each data point represents the mean ± SEM; n = 3 for independent experiments. Compound name

Compound class

Cell viability (ATP) in μM ± SEM

Neuron count IC50 in μM ± SEM

Neurite length IC50 in μM ± SEM

Cisplatin Oxaloplatin Carboplatin Thalidomide Lenalidomide Pomalidomide Paclitaxel Docetaxel Ixabepilone Sagopilone Vincristinea Nocodazole Colchicine K252b Bortezomib Suramin Amoxicillin Sorbitol Hydroxyurea

Platinum

> 100 ± 0 > 100 ± 0 > 100 ± 0 > 100 ± 0 > 100 ± 0 > 100 ± 0 > 10 ± 0 > 10 ± 0 > 10 ± 0 > 10 ± 0 >1 ± 0 > 30 ± 0 > 30 ± 0 > 100 ± 0 > 100 ± 0 > 100 ± 0 > 100 ± 0 > 100 ± 0 > 100 ± 0

> 100 ± 0 94.55 ± 5.46 > 100 ± 0 > 100 ± 0 > 100 ± 0 > 100 ± 0 > 10 ± 0 > 10 ± 0 > 10 ± 0 > 10 ± 0 >1 ± 0 > 30 ± 0 > 30 ± 0 > 100 ± 0 > 100 ± 0 > 100 ± 0 > 100 ± 0 > 100 ± 0 > 100 ± 0

> 100 ± 0 74.08 ± 20.37 > 100 ± 0 > 100 ± 0 > 100 ± 0 > 100 ± 0 0.00563 ± 0.0007 0.00066 ± 0.0002 0.00426 ± 0.0013 0.00971 ± 0.0013 0.00548 ± 0.0047 0.09462 ± 0.0369 0.01038 ± 0.0042 > 100 ± 0 > 100 ± 0 55.287 ± 21.62 > 100 ± 0 > 100 ± 0 > 100 ± 0

a

Anti-angiogenic/immunomodulatory Agent

Taxanes Epothilones Microtubule interfering agents

Protein kinase inhibitor Proteasome inhibitor Growth factor inhibitor Negative controls

Positive control.

room temperature to stain for nuclei. Cells were then washed twice with PBS, and stored at 4 °C prior to image acquisition and analysis.

(+)-glucose 10. The pH value of the extracellular solution was set to 7.4 using NaOH and the osmolarity was set to 308 mOsm/l. The intracellular solution was prepared fresh every day from frozen (− 22 °C) aliquots containing (in mM): K-aspartate 145, MgCl2 1, HEPES 10, EGTA 5, Na2-ATP 5 and Na2-GTP 0.5. The osmolarity was set to 303 mOsm/l and the pH to 7.2 using KOH.

2.2.4. Manual patch clamp assessment The extracellular solution for patch clamp recordings consisted of (in mM): NaCl 145, KCl 4, MgCl2 1, CaCl2 1.8, HEPES 10 and alpha-D 115

Toxicology in Vitro 45 (2017) 111–118

P. Rana et al.

Treatment with the potent sodium channel blocker tetrodotoxin (TTX) eliminated the inward (downward) current suggesting that neurons are expressing a TTX-sensitive Na+ current (Fig. 2A). Action potentials could be elicited under current clamp mode (Fig. 2B). Spontaneous action potentials were observed (Fig. 2C) in all wells when plated at densities > 15,000 cells/μL. These observations are in line with previous reports of primary peripherally-derived neurons (Study and Kral, 1996). Notably, when cells were plated at high density on microelectrode arrays, all wells from a 96 well Axion Maestro (Axion Biosystems, Atlanta, GA USA) MEA plate routinely demonstrated spontaneously active neuron cultures within 72 h (data not shown).

Manual patch clamp recordings were performed on adherent Peri.4U neurons seeded at low density on glass cover slips coated with 0.1% polyethylenimine (in boric acid/sodium tetraborate in water, Sigma P3143). All experiments were performed at room temperature with cells in culture from 12 to 20 days. To minimize fluctuations of the room temperature, the temperature of the bath solution was set to 21 °C using a temperature control system. The patch clamp amplifier (EPC10, HEKA, Lambrecht, Deutschland) was controlled by the recording software Patchmaster (v2x35, HEKA, Lambrecht, Deutschland). Patch clamp recordings were performed in voltage clamp as well as in current clamp mode using electrodes with an electrode resistance between 5 and 8 MΩ.

3.3. Comparison of NG108-15 and Peri.4U cells 2.2.5. Cell viability assays Cell viability was assessed after 24 h of drug treatment using the CellTiter-Glo Luminescent Cell Viability Assay (Promega, Madison, WI, USA) according to manufacturer instructions. Luminescence was measured using a Spectramax M5 plate reader (Molecular Devices LLC, Sunnyvale, CA, USA). At least three biological replicates for the viability assays were performed, with 2 wells per drug concentration measured for each replicate.

We initially compared Peri.4U cells with NG108-15 cells (a hybrid mouse neuroblastoma and rat glioma cell line) for their neurite characteristics and response to our positive control, vincristine, a welldocumented microtubule destabilizer (Morris and Fornier, 2008) that causes neurite retraction in vitro (Hoelting et al., 2016; Wing et al., 2017) and peripheral neuropathy in vivo (Argyriou et al., 2014). As shown in Fig. 3(A & B), Peri.4U cells had significantly more neurite processes, branching and branch density as well as more neuronal bodies extending neurites when compared with NG108-15 cells. Further, when vincristine was tested in concentrations ranging from 0.00001 to 1 μM in both NG108-15 and Peri.4U cells, there was a clear dose-dependent separation between cellular viability (neuronal count) and neurite length values in Peri.4U cells (Fig. 3C). The NG108-15 cells failed to separate cellular viability from neurite length (Fig. 3D). Thus evaluations of remaining chemotherapeutic compounds were conducted using Peri.4U neurons only.

2.2.6. High content imaging and neurite length analysis Images were attained on a Cellomics ArrayScan VTI (Thermo Scientific, Waltham, MA) and analysis was performed after 24 h of drug treatment and drug treated cultures were compared to time-matched, untreated controls. Image analysis was performed using the Neuronal Profiling version 4 bioapplication. A 10× 0.45NA objective was used for image acquisition and all 25 fields were evaluated for each well. DAPI was used to identify and count individual nuclei in channel 1. βIIItubulin was used in channel 2 to identify cell bodies and neurites by intensity. Cell bodies were not used for neurite detection if they contained > 1 nucleus. The critical well level output parameters reported were Neuron Count per Valid Field (provides relative cell counts for cytotoxicity measurement), and Mean Neurite Total Length Ch2. Detailed protocols for neurite outgrowth analysis using high content imaging technology were previously established (Harrill et al., 2013).

3.4. Assessment of various classes of chemotherapy compounds in Peri.4U cells Compounds associated with CIPN from a variety of chemotherapy classes (16 in total) were tested for their acute (24 h) effects on Peri.4U cells as described in materials and methods. Endpoints included neuron count and neurite total length, as well as an independent measurement of neuronal viability using ATP as shown in assay schematic Fig. 4. Fig. 5 shows representative neuron count and neurite total length data for paclitaxel, ixabepilone and colchicine, as well as the negative control sorbitol. See Supplementary material for an example of raw images with paclitaxel treatment. Table 1 lists all 16 compounds tested, their chemotherapy classes and the IC50s for the 3 in vitro end points evaluated. The microtubule stabilizers (taxanes and epithilones) exhibited similar profiles of reduced neurite length in the absence of cytotoxicity (based on neuron counts and viability). The IC50s for neurite length reductions ranged from approximately 1–10 nM, with docetaxel being the most potent (IC50 = 0.66 nM) in this model. The microtubule destabilizers (vincristine, nocodazole, and colchicine) also reduced neurite length in the absence of cytotoxicity, although at slightly lower potencies, with Nocodazole being the least potent (IC50 = 95 nM). In contrast, the platinum compounds (cisplatin, oxaloplatin, and carboplatin), along with K252b (a protein kinase inhibitor), bortezomib (a proteasome inhibitor), and suramin (a growth factor inhibitor), all associated with CIPN in humans, were not active in this model; only suramin and oxaloplatin showing inhibition of neurite length (IC50s of 55.3 and 74.1 μM respectively). Interestingly, when we tested some of these compounds (cisplatin, oxaloplatin or bortezomib) by exposing Peri.4U neurons to compound for the entire 72 h period, starting when the neurons were seeded, as was recently done by others (Hoelting et al., 2016), we also can detect a clear effect on neurite length at low drug concentrations (see Supplementary materials). For the anti-angiogenic compounds (thalidomide, lenalidomide and pomalidomide) tested, no effect on cell viability, neuron count, or

2.2.7. Data analysis Concentration-response plots were expressed as percent of timematched controls (equivalent endpoints in untreated cultures) plotted as log of concentration vs. normalized response. IC50 values for each compound were generated in GraphPad Prism 6 (San Diego, CA) using a non-linear regression model of the form Y = 100 / (1 + 10^ ((LogIC50 − X) ∗ Hillslope))), constrained between 0 and 100% of the time matched controls. 3. Results 3.1. Morphological and cellular characterization of Peri.4U cells To demonstrate the neuronal nature of commercially available Peri.4U neurons, we first examined their basic morphological features. Peri.4U cells demonstrated robust, interdigitated neurites within two days of culture (Fig. 1A). These physical interactions were highly dynamic, which was characterized by tracking cells using light microscopy over time using an IncuCyte Zoom (Essen Bioscience, see videos in Supplementary materials). Peri.4U neurons exhibited largely homogenous MAP2 and peripherin expression (Fig. 1B), which is consistent with previous reports of human iPSC-derived peripheral-like neurons (Chambers et al., 2012). 3.2. Electrophysiological characterization of Peri.4U cells Under manual voltage clamp conditions, nearly all individual Peri.4U neurons demonstrated the expected Na+ and K+ currents. 116

Toxicology in Vitro 45 (2017) 111–118

P. Rana et al.

shown to be active in our assay, which is just over 50%. When compound class is considered, there is much higher predictive value. Testing 2–4 compounds per class isn't sufficient to robustly evaluate the predictivity of this assay and additional compounds from each class should be evaluated in future studies. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.tiv.2017.08.014.

neurite length was observed. Negative controls amoxicillin, sorbitol and hydroxyurea also showed no effect on Peri.4U viability or neurite length. 4. Discussion We report here on a high throughput, in vitro peripheral neuropathy assay that uses commercially available, highly reproducible, human iPSC-derived peripheral-like neurons (Peri.4U cells) in a 96-well, high content imaging format. These cells express protein markers that are expected to be present in neuronal cells (e.g. peripherin, MAP2, & beta3 tubulin), as well as electrophysiological properties that are consistent with functional neurons. Using these iPSC-derived human neurons, we have shown that treatment with representative drugs associated with clinically observed CIPN can affect neurite length in culture. Other in vitro models that use transformed cells, or primary neurons from animals, are not able to reliably detect drugs that cause peripheral neuropathy in humans (Golden and Johnson, 2004). This may be due to the altered genotype and phenotype of transformed lines, species differences, or differences between cells derived from central verses peripheral nervous system. For many of these cell models effects on neurites cannot be distinguished from non-specific effects on general cell health or viability, making it difficult to distinguish mechanisms of toxicity. Interestingly, the finding that chemotherapeutic drugs can impact neurite length in the absence of cytotoxicity was also recently demonstrated in cultures using human embryonic stem cell-derived neurons (Hoelting et al., 2016), further supporting the physiological relevance of human stem cell-derived models, at least for some classes of chemotherapeutics. Notably, untreated Peri.4U cells do not show obvious differences in neurites between 48 and 72 h (Fig. 1A and control video in Supplementary materials). However when treated with vincristine during this period Peri.4U cells showed a nearly complete loss of neurites (see vincristine (1 μM) video in Supplementary materials) suggesting that reductions in total neurite length are the result of neurite degeneration and not merely inhibition of neurite extension. A major finding of this work was that treatment of the Peri.4U cells with representative drugs from the taxane, epothilone and vinca alkaloid classes of chemotherapeutics, known to cause peripheral neuropathy in humans, reduced neurite length in the absence of cytotoxicity. Moreover, these effects were observed in low nanomolar drug concentration ranges; concentrations at or below systemic exposures used clinically. These observations are consistent with reports that axonal microtubules are the primary site of toxic action of these drugs classes (Argyriou et al., 2014). In contrast, representative drugs from the platinum, anti-angiogenesis, and proteasome and growth inhibitor chemotherapeutic classes, which do not act directly on axonal structures or maintenance of axonal processes (Argyriou et al., 2014), were not observed to affect neurite length. However, when we added drug at the same time that cells were seeded, similar to methods recently described (Hoelting et al., 2016), we also detected inhibition of neurite outgrowth for classes of drugs whose primary site of action is not axonal processes, such as the platinum drugs cisplatin and oxaloplatin, and the proteasome inhibitor bortezomib (Supplementary materials). Under these conditions we are likely detecting mechanisms that impact the ability of neurons to extend processes, and are affecting the cells ability to make and transport proteins critical to neurite extension. Clearly, one limitation of this assay is that it cannot detect drugs that may cause CIPN via indirect effects on blood flow, immunomodulation or non-neuronal glial cells. In summary the Peri.4U model developed here has the potential to be a useful tool for mechanistically differentiating drugs that work directly on axonal processes from those that affect the neuronal cell body. Furthermore, it has the sensitivity to serve as a screening tool for novel chemotherapeutics to evaluate the risk of peripheral neuropathy for patients. However, only 9 of the 16 positive compounds tested were

Transparency document The http://dx.doi.org/10.1016/j.tiv.2017.08.014 associated with this article can be found, in online version. Acknowledgements The authors would like to thank Jim Finley and Chang-Ning Liu for their neuronal cell expertise and advice, and Colleen Doshna and Lisa Marroquin for QC of raw data and the final manuscript. Additionally the authors thank Elke Guenther and the NMI-TT in Reutlingen for their patch clamp contributions. All research described here was funded by Pfizer, Inc. and Axiogenesis, AG. References Argyriou, A.A., Assimakopoulos, K., Iconomou, G., Giannakopoulou, F., Kalofonos, H.P., 2011. Either called “chemobrain” or “chemofog”, the long-term chemotherapy-induced cognitive decline in cancer survivors is real. J. Pain Symptom Manag. 41, 126–139. Argyriou, A.A., Kyritsis, A.P., Makatsoris, T., Kalofonos, H.P., 2014. Chemotherapy-induced peripheral neuropathy in adults: a comprehensive update of the literature. Cancer Manag. Res. 6, 135–147. Balayssac, D., Ferrier, J., Descoeur, J., Ling, B., Pezet, D., Eschalier, A., Authier, N., 2011. Chemotherapy-induced peripheral neuropathies: from clinical relevance to preclinical evidence. Expert Opin. Drug Saf. 10, 407–417. Binda, D., Vanhoutte, E.K., Cavaletti, G., Cornblath, D.R., Postma, T.J., Frigeni, B., Alberti, P., Bruna, J., Velasco, R., Argyriou, A.A., Kalofonos, H.P., Psimaras, D., Ricard, D., Pace, A., Galie, E., Briani, C., Dalla Torre, C., Lalisang, R.I., Boogerd, W., Brandsma, D., Koeppen, S., Hense, J., Storey, D., Kerrigan, S., Schenone, A., Fabbri, S., Rossi, E., Valsecchi, M.G., Faber, C.G., Merkies, I.S., Galimberti, S., Lanzani, F., Mattavelli, L., Piatti, M.L., Bidoli, P., Cazzaniga, M., Cortinovis, D., Lucchetta, M., Campagnolo, M., Bakkers, M., Brouwer, B., Boogerd, W., Grant, R., Reni, L., Piras, B., Pessino, A., Padua, L., Granata, G., Leandri, M., Ghignotti, I., Plasmati, R., Pastorelli, F., Heimans, J.J., Eurelings, M., Meijer, R.J., Grisold, W., Lindeck Pozza, E., Mazzeo, A., Toscano, A., Russo, M., Tomasello, C., Altavilla, G., Penas Prado, M., Dominguez Gonzalez, C., Dorsey, S.G., 2013. Rasch-built Overall Disability Scale for patients with chemotherapy-induced peripheral neuropathy (CIPN-R-ODS). Eur. J. Cancer (Oxford, England: 1990) 49, 2910–2918. Chambers, S.M., Qi, Y., Mica, Y., Lee, G., Zhang, X.J., Niu, L., Bilsland, J., Cao, L., Stevens, E., Whiting, P., Shi, S.H., Studer, L., 2012. Combined small-molecule inhibition accelerates developmental timing and converts human pluripotent stem cells into nociceptors. Nat. Biotechnol. 30, 715–720. Collins, F.S., Gray, G.M., Bucher, J.R., 2008. Toxicology. Transforming environmental health protection. Science (New York, N.Y.) 319, 906–907. Elena, J.W., Travis, L.B., Simonds, N.I., Ambrosone, C.B., Ballard-Barbash, R., Bhatia, S., Cerhan, J.R., Hartge, P., Heist, R.S., Kushi, L.H., Lash, T.L., Morton, L.M., Onel, K., Pierce, J.P., Robison, L.L., Rowland, J.H., Schrag, D., Sellers, T.A., Seminara, D., Shu, X.O., Thomas, N.E., Ulrich, C.M., Freedman, A.N., 2013. Leveraging epidemiology and clinical studies of cancer outcomes: recommendations and opportunities for translational research. J. Natl. Cancer Inst. 105, 85–94. Fehrenbacher, J.C., 2015. Chemotherapy-induced peripheral neuropathy. Prog. Mol. Biol. Transl. Sci. 131, 471–508. Golden, J.P., Johnson, E.M., 2004. Models of chemotherapy drug-induced peripheral neuropathy. Drug Discov. Today Dis. Model. 1, 186–191. Harrill, J.A., Robinette, B.L., Freudenrich, T., Mundy, W.R., 2013. Use of high content image analyses to detect chemical-mediated effects on neurite sub-populations in primary rat cortical neurons. Neurotoxicology 34, 61–73. Hoelting, L., Klima, S., Karreman, C., Grinberg, M., Meisig, J., Henry, M., Rotshteyn, T., Rahnenfuhrer, J., Bluthgen, N., Sachinidis, A., Waldmann, T., Leist, M., 2016. Stem cell-derived immature human dorsal root ganglia neurons to identify peripheral neurotoxicants. Stem Cells Transl. Med. 5, 476–487. Hoke, A., Ray, M., 2014. Rodent models of chemotherapy-induced peripheral neuropathy. ILAR J. 54, 273–281. Marmiroli, P., Nicolini, G., Miloso, M., Scuteri, A., Cavaletti, G., 2012. The fundamental role of morphology in experimental neurotoxicology: the example of chemotherapyinduced peripheral neurotoxicity. Ital. J. Anat. Embryol. 117, 75–97. Morris, P.G., Fornier, M.N., 2008. Microtubule active agents: beyond the taxane frontier. Clin. Cancer Res. 14, 7167–7172. Park, S.B., Goldstein, D., Krishnan, A.V., Lin, C.S., Friedlander, M.L., Cassidy, J.,

117

Toxicology in Vitro 45 (2017) 111–118

P. Rana et al.

ganglion neurons from rats with a painful neuropathy. Pain 2-3, 235–242. Tice, R.R., Austin, C.P., Kavlock, R.J., Bucher, J.R., 2013. Improving the human hazard characterization of chemicals: a Tox21 update. Environ. Health Perspect. 121, 756–765. Wing, C., Komatsu, M., Delaney, S.M., Krause, M., Wheeler, H.E., Dolan, M.E., 2017. Application of stem cell derived neuronal cells to evaluate neurotoxic chemotherapy. Stem Cell Res. 22, 79–88. Wolf, S., Barton, D., Kottschade, L., Grothey, A., Loprinzi, C., 2008. Chemotherapy-induced peripheral neuropathy: prevention and treatment strategies. Eur. J. Cancer (Oxford, England: 1990) 44, 1507–1515.

Koltzenburg, M., Kiernan, M.C., 2013. Chemotherapy-induced peripheral neurotoxicity: a critical analysis. CA Cancer J. Clin. 63, 419–437. Pike, C.T., Birnbaum, H.G., Muehlenbein, C.E., Pohl, G.M., Natale, R.B., 2012. Healthcare costs and workloss burden of patients with chemotherapy-associated peripheral neuropathy in breast, ovarian, head and neck, and nonsmall cell lung cancer. Chemother. Res. Prac. 2012, 913848. Slotkin, T.A., MacKillop, E.A., Ryde, I.T., Tate, C.A., Seidler, F.J., 2007. Screening for developmental neurotoxicity using PC12 cells: comparisons of organophosphates with a carbamate, an organochlorine, and divalent nickel. Environ. Health Perspect. 115, 93–101. Study, R.E., Kral, M.G., 1996. Spontaneous action potential activity in isolated dorsal root

118