Effects of neuroactive agents on axonal growth and pathfinding of retinal ganglion cells generated from human stem cells

We recently established a novel method for generating functional human retinal ganglion cells (RGCs) from human induced pluripotent cells (hiPSCs). Here, we confirmed that RGCs can also be generated from human embryonic stem cells (hESCs). We investigated the usefulness of human RGCs with long axons for assessing the effects of chemical agents, such as the neurotrophic factor, nerve growth factor (NGF), and the chemorepellent factors, semaphorin 3 A (SEMA3A) and SLIT1. The effects of direct and local administration of each agent on axonal projection were evaluated by immunohistochemistry, real-time polymerase chain reaction (PCR), and real-time imaging, in which the filopodia of the growth cone served as an excellent marker. A locally sustained agent system showed that the axons elongate towards NGF, but were repelled by SEMA3A and SLIT1. Focally transplanted beads that released SLIT1 bent the pathfinding of axons, imitating normal retinal development. Our innovative system for assessing the effects of chemical compounds using human RGCs may facilitate development of novel drugs for the examination, prophylaxis, and treatment of diseases. It may also be useful for observing the physiology of the optic nerve in vitro, which might lead to significant progress in the science of human RGCs.


Axonal transport observation
The time series images of anterograde axonal transport was generated as described previously 1 by the injection of Alexa Fluo-555-conjugated cholera toxin subunit B (Life Technologies) into the centre of attached OVs, followed by imaging the Alexa Fluo-555-conjugated cholera toxin subunit B using an IX71 inverted research microscope (Olympus).

Electrophysiological recording
Whole-cell patch-clamp recordings were performed as previously described 1 . Briefly, EBs were attached and cultured on mixed cellulose ester filter paper (0.2-µm pore size; ADVANTEC, Houston, TX) from D27. Whole-cell patch-clamp recordings were performed on RGCs located at the peripheral portion of attached OVs. RGCs exhibited repetitive action potentials. Recording pipettes (6−8 MW Mrding pipettes (6e action plar solution, as described previously 1 . The morphology of the recorded cells was visualised using Lucifer Yellow (LY). Liquid junction potentials (-11 mV) were corrected and R s was compensated at 40%. Cells with R s > 50 MW0 M Ronot included in the analysis. Average membrane capacitance was 8.3 ± 3.8 pF in hESC-derived RGCs (n = 6). Current and voltage data were acquired using pCLAMP 9.2 software and saved on a custom-built personal computer (Physio-Tech, Tokyo, Japan). Analyses were performed with Clampfit 9.2 (Molecular Devices, Sunnyvale, CA) and OriginPro 2015 (OriginLab, Northampton, UK). Images of LY-filled cells were captured using a high-gain colour camera (HCC-600; Flovel, Tokyo, Japan) and saved using INFO.TV Plus software (Infocity, Gandhinagar, India). All data are presented as mean ± SD.

Local application of glutamate
We used a puffer pipette whose diameter was the same as the recording pipette (» 1 µm) for local application of glutamate to the recorded RGCs. The puffer pipettes were filled with an external solution containing (in mM): 135 NaCl, 3 KCl, 2.5 CaCl2, 1 MgCl2, 10 glucose, 10 HEPES, and 1 glutamate (pH 7.4) and positioned near the soma of recorded RGCs. We applied a pressure pulses continuously (10 kPa) via an electromagnetic valve (M5136-NB-M5; CKD, Japan) which was controlled using pCLAMP 9.2 software.

Physiological function of retinal ganglion cells generated from human embryonic stem cells
Furthermore, we confirmed that hESC-derived RGCs possess the same physiological functions as those reported for hiPSC-derived RGCs 1 . Firstly, we tested axonal transport in the axons apparent on D30. Anterograde axonal transport was assayed by injecting Alexa Fluor-conjugated cholera toxin B into the central portion of the attached OVs.
Within 60 min, cholera toxin B had been transported through the axon by anterograde transport (Fig 2a and Supplementary Movie 1). Secondly, we evaluated whether hESC-derived RGCs were able to fire action potentials. We selected RGCs on the peripheral margin of attached OVs for whole-cell patch-clamp recordings (Fig. 2b). A representative RGC showing a long axon process is shown in Figure 2c. The sample of cells we recorded from the generated repetitive action potentials in response to current injection in current-clamp mode is shown in Figure 2d. Resting membrane potential was determined to be −67 ± 15 mV and the amplitude of the first action potential was recorded as 58 ± 14 mV (n = 6). Cells that generated action potentials exhibited tetrodotoxin-sensitive (TTX-sensitive) Na currents with maximum amplitudes of 958 ± 355 pA (n = 6), followed by outward currents (Fig. 2e). Using puff application of 1 mM glutamate, we also observed an inward current of 28 ± 29 pA (n = 5) and repetitive action potentials with an amplitude of 69 ± 9.9 mV (n = 5) (Fig. S3). Combining the results of our current and previous studies confirms that hESC-and hiPSC-derived RGCs expressed characteristic molecular and electrophysiological markers of RGCs.
Therefore, hESC-derived RGCs were ready for use in the in vitro evaluation of the effects of neurotrophic and chemorepellent agents.

Supplementary Figures
Figure S1        In order to quantify the amount of axons, we use the algorism indicated from (a) to (c).         (a)Radiated axons of hESC-derived RGCs on D31, stained for NFL, are converted in gray scale. In the control, few axons take a path through the 100-µm-wide area, parallel to the irradiated axons (red lines). (b) Compared to the control, bent axons are clearly identified and take a path through the 100-µm-wide area by the local sustained release of 5 µg/ml SLIT1, placed immediately in front of the irradiated axons. (c) The number of bent axons that take a path through the 100-µm-wide area is significantly increased by local SLIT1 release (t-test, p≤0.05). Error bars indicate ± standard deviation (SD).
Each column shows an average value for the studied samples (n = 4). Scale bars, 200 µm.