In Vitro Generation of Posterior Motor Neurons from Human Pluripotent Stem Cells

The ability to generate spinal cord motor neurons from human pluripotent stem cells (hPSCs) is of great use for modelling motor neuron–based diseases and cell‐replacement therapies. A key step in the design of hPSC differentiation strategies aiming to produce motor neurons involves induction of the appropriate anteroposterior (A‐P) axial identity, an important factor influencing motor neuron subtype specification, functionality, and disease vulnerability. Most current protocols for induction of motor neurons from hPSCs produce predominantly cells of a mixed hindbrain/cervical axial identity marked by expression of Hox paralogous group (PG) members 1‐5, but are inefficient in generating high numbers of more posterior thoracic/lumbosacral Hox PG(8‐13)+ spinal cord motor neurons. Here, we describe a protocol for efficient generation of thoracic spinal cord cells and motor neurons from hPSCs. This step‐wise protocol relies on the initial generation of a neuromesodermal‐potent axial progenitor population, which is differentiated first to produce posterior ventral spinal cord progenitors and subsequently to produce posterior motor neurons exhibiting a predominantly thoracic axial identity. © 2021 The Authors. Current Protocols published by Wiley Periodicals LLC.


INTRODUCTION
During embryonic development, motor neurons (MNs) arise from a set of progenitors within the ventral spinal cord, from which they mature and project axons to innervate target muscles. Multiple MN subtypes are specified across the anterior-posterior (A-P) axis of the spinal cord, allowing for innervation of the diverse axial-level-dependent muscle targets (reviewed in Sagner & Briscoe, 2019). The specification of MNs across the A-P axis of the spinal cord is largely regulated by a family of homeobox genes known as HOX genes (Dasen, Liu, & Jessell, 2003;Dasen, Tice, Brenner-Morton, & Jessell, 2005Jung et al., 2010). Hox genes are arranged as paralogous groups (PG) 1-13 across four distinct chromosomal clusters (A-D) and are expressed along the post-cranial A-P axis in a strict spatiotemporal manner reflecting their 3 -to-5 genomic order: hindbrain/cervical MNs are marked by Hox PG(1-5), whereas more posterior thoracic and lumbosacral MNs are marked by Hox PG(6-9) and Hox PG(10-13), respectively.
Damage or degeneration of MNs results in a large array of movement-based disorders. In vitro production of MNs from human pluripotent stem cells (hPSCs) has provided a useful tool for in vitro modelling of these neurodegenerative diseases as well as cell therapy applications, e.g., for treatment of spinal cord injuries. Most conventional hPSC differentiation protocols for generation of MNs rely predominantly on initial induction of an anterior neural identity that is successively patterned to a ventral spinal cord/MN fate through exposure to sonic hedgehog (SHH) and retinoic acid signals (Amoroso et al., 2013;Lee et al., 2007;Li et al., 2008;Peljto, Dasen, Mazzoni, Jessell, & Wichterle, 2010;Wichterle, Lieberam, Porter, & Jessell, 2002). However, MNs produced in these protocols predominantly exhibit a Hox PG(1-5)+ hindbrain/cervical character, with low yields of the more posterior Hox PG(8-13)+ thoracic/lumbosacral MNs.
We recently described an efficient protocol for generation of posterior motor neurons that mainly display a thoracic axial identity (Wind et al., 2021). Our strategy is based on initial induction of a neuromesodermal progenitor (NMP)-like population from hPSCs following treatment with WNT and FGF signaling pathway agonists (Frith et al., 2018;Gouti et al., 2014;Lippman et al., 2015). In vivo, NMPs are posteriorly located bipotent progenitors that give rise to neural and mesodermal cells of the post-cranial axis during embryonic development (Tzouanacou, Wegener, Wymeersch, Wilson, & Nicolas, 2009;Wymeersch et al., 2016;reviewed in Wymeersch, Wilson, & Tsakiridis, 2021). These are subsequently steered toward a ventral spinal cord progenitor state via active suppression of the pro-dorsal TGF-β and BMP signaling pathways in addition to SHH/retinoic acid signaling stimulation, which eventually gives rise to posterior MNs following culture in neurotrophic medium (Wind et al., 2021). Here we provide an in depth step-by-step description of this approach.

DIFFERENTIATION OF NEUROMESODERMAL PROGENITORS
The initial step of the differentiation protocol depends upon the high yield generation of NMP-like cells from hPSCs driven by simultaneous stimulation of the WNT and FGF signaling pathways and inhibition of BMP activity (Fig. 1A) to limit induction of neural crest progenitors (Frith et al., 2018). These are marked by the co-expression of TBXT and SOX2, transcriptional factors indicative of a mesodermal and neural identity, respectively. Moreover, NMP cultures exhibit expression of other posteriorly expressed genes, such as NKX1-2, CDX2 and Hox family members.
7. Transfer suspended cells to a 15-ml Falcon tube and spin 4 min at 200 × g.
Small molecules and recombinant proteins should be added to medium fresh each day.
10. Aspirate vitronectin from the culture plates and add 150 μl/cm 2 N2B27 medium with 10 μM Y-27632 to each well.
11. Plate cells at a density of 60,000 cells/cm 2 .
NMP differentiation is complete on day 3. Cells should appear as a confluent monolayer (see Fig. 1B).

Day 3: Assess differentiation
15. On day 3, confirm the NMP cell type by antibody staining.
Successfully differentiated NMPs should co-express TBXT and SOX2 (Fig. 1C). Efficient induction of CDX2 and other genes denoting a posterior axial identity (as reviewed by Wymeersch et al., 2021) can also be assessed by immunostaining/qPCR.

DIFFERENTIATION OF POSTERIOR VENTRAL SPINAL CORD PROGENITORS
This protocol encompasses the directed differentiation of human NMPs to a posterior ventral spinal cord identity. It relies on combined BMP and TGF-β inhibition and supplementation by retinoic acid and SHH agonist signals (Figs. 2A, 3A) to first promote generation of HOXC9 + /SOX1 + /PAX6 + posterior neurectoderm cells and subsequently induce differentiation of ventral spinal cord progenitors.

Wind and Tsakiridis
Geltrex should not be removed until cells are seeded.
4. Add excess DMEM-F12 (at least 2 vol. relative to Accutase) and triturate gently to create a single-cell suspension.
5. Transfer cell suspension to a 15-ml Falcon tube and centrifuge 4 min at 200 × g.
Small molecules and recombinant proteins should be added to medium fresh each day.
8. Aspirate Geltrex from the culture plates and add 150 μl/cm 2 of posterior neurectoderm induction medium with 10 μM Y-27632 to each well.
9. Plate cells at a density of 60,000 cells/cm 2 .
10. Rock gently to distribute the cells evenly across the wells and incubate overnight at 37°C/5% CO 2 .

Y-27632 is included only on days 3 and 7 when cells are replated.
Days 7-14: Perform posterior ventral spinal cord progenitor differentiation 12. On day 7, repeat steps 1-11 to replate cells at a density of 80,000 cells/cm 2 .
On day 7, before replating, successfully differentiated cells should appear as a confluent monolayer with clusters/clumps beginning to form (Fig. 2B).
14. Also on day 8, confirm posterior neuroectoderm cell type by antibody staining and qPCR.
Successful generation of a posterior neurectoderm identity, at day 8, can be identified by co-expression of the neural progenitor markers SOX1 and PAX6, together with expression of CDX2 and posterior HOX PG(6-9), indicating a posterior brachial/thoracic axial identity (Fig. 2C,D). Furthermore, maintenance of expression of SOX2 (a neural progenitor marker) and loss of TBXT (a marker of mesodermal progenitors) indicates efficient neural conversion of NMPs.

DIFFERENTIATION OF POSTERIOR MOTOR NEURONS
The final step of the procedure allows for promotion of a post-mitotic posterior motor neuron identity. Supplementation with the γ-secretase inhibitor DAPT promotes exit from a progenitor state, and the presence of neurotrophins promotes neuronal survival and axonal outgrowth (Fig. 4A).
3. Add excess DMEM-F12 (at least 2 vol. relative to Accutase) and triturate gently to create a single-cell suspension.
4. Transfer cell suspension to a 15-ml Falcon tube and centrifuge 4 min at 200 × g.

5.
Resuspend cell pellet in 1 ml of freshly prepared motor neuron maturation medium supplemented with 10 μM Y-27632.
Small molecules and recombinant proteins should be added to medium fresh each day.
7. Aspirate Geltrex from the culture plates and add 150 μl/cm 2 of motor neuron maturation medium with 10 μM Y-27632 to each well. 9. Rock gently to distribute the cells evenly across the wells and incubate 48 hr at 37°C/5% CO 2 .

Y-27632 is included only on day 14 when cells are replated.
Axonal projections should be clearly visible on day 16.
On day 24, posterior motor neuron differentiation should be complete. Cells should display clear clusters of neuronal cell bodies with fasciculated axonal projections to adjacent clusters (Fig. 4A). A mature motor neuron identity can be confirmed by expression of ISLET1, MNX1, cholinergic acetyltransferase (ChAT), and neurofilament (Fig. 4C,D). Expression of HOX PG(6-9) indicates a posterior motor neuron identity, while expression of nNOS suggests a high yield of thoracic-specific preganglionic motor column character (Fig. 4C,D). Electrophysiological functionality, displayed through whole-current patchclamp recordings, can also be used to demonstrate the functional maturity of the derived motor neurons (Wind et al., 2021). final 100 μM) 2 ml 100× penicillin-streptomycin (final 1×; optional) Store up to 3 weeks at 4°C On the day of use add: 20 ng/ml recombinant basic FGF 3 μM CHIR99021 100 nM LDN193189

COMMENTARY Background Information
Because damage or degeneration of MNs occurs in a wide array of movement-based disorders, the in vitro generation of MNs of defined A-P identities from hPSCs has been the focus of intense research.
Conventional MN differentiation protocols typically rely on induction of an early anterior neural identity through exposure to TGF-β and BMP inhibitors (Amoroso et al., 2013;Chambers et al., 2009). Subsequent treatment with SHH agonists has been globally utilized to promote a ventral character, whereas simultaneous exposure to retinoic acid is designed to promote more caudal identities. However, MNs produced through this approach exhibit a predominantly hindbrain/cervical/upper brachial character marked by expression of HOX PG(1-6) members and innervating the anterior musculature upon in vivo transplantation (Amoroso et al., 2013;Soundararajan, Miles, Rubin, Brownstone, & Rafuse, 2006).
During amniote embryonic development, posteriorly located neuromesodermal progenitors have been shown to give rise to cell lineages of the post-cranial axis (Cambray & Wilson, 2007;Tzouanacou et al., 2009;Wymeersch et al., 2016). Similarly, NMP-like cells generated in vitro from hPSCs can also be steered towards posterior neural, mesodermal, and neural crest lineages (Cooper et al., 2021;Frith et al., 2018;Gouti et al., 2014;Lippmann et al., 2015;Mouilleau et al., 2021;Turner, Rué, Mackenzie, Davies, & Martinez Arias, 2014;Verrier, Davidson, Gierliński, Dady, & Storey, 2018). Moreover, we have recently shown that the efficient generation of posterior thoracic HOX PG(6-9) motor neurons is best achieved by differentiating through a neuromesodermal progenitor intermediary state (Wind et al., 2021). This strategy relies on exposure of hPSC-derived NMPs to TGF-β and BMP inhibitors to promote a posterior neural tube identity, similar to that observed by Verrier et al. (2018). Further differentiation in the presence of Shh signals and TGF-β and BMP inhibitors coordinate more efficient ventralization to promote a MN identity, while continued high levels of FGF and Wnt signals are required to promote a posterior axial identity (Mouilleau et al., 2021;Wind et al., 2021).
Our differentiation strategy gives rise to cultures predominantly marked by high levels of HOX PG(8-9) transcripts and large numbers of HOXC9-ISLET1-nNOS co-expressing/MNX1-negative MNs, thus demonstrating a thoracic preganglionic columnar character. However, we have also detected the presence of more anterior (PG4-6) and posterior (PG10) HOX transcripts, indicative of potential co-emergence of MN subpopulations of a brachial and lumbar axial identity, respectively (Wind et al., 2021). Moreover, MNX1 positivity may reflect the induction of MNs of lateral, medial, or hypaxial motor column character.

Quality of hPSC cultures
Prior to use in differentiation protocols, it is essential to ensure that the hPSCs have retained pluripotency, which can be observed by screening for pluripotency markers using immunofluorescence and flow cytometry. A reduction in pluripotency and observed spontaneous differentiation in hPSC cultures can result in inefficient differentiation, and cultures with high levels of spontaneous differentiation or reduced pluripotency marker expression should be disposed. Screening for genetic abnormalities is also important, as specific aberrations can impair hPSC differentiation capabilities. Good practices such as regular culture medium changes, passaging cells once they reach 70%-80% confluency, and discarding cells that have a high passage number can ensure better hPSC quality. The protocol described here has been optimized using hPSCs cultured on vitronectin-and laminin-521-coated surfaces in Essential 8 medium. It is important to note that culturing hPSCs on different substrates and in different media can impact downstream differentiation efficiencies, and therefore further optimization may be required.

Titration of small molecule concentrations
We have found that levels of WNT, TGFβ, and BMP signals must be tightly regulated to allow both efficient neural induction and subsequent ventralization. Titration of TGFβ and BMP inhibitor and CHIRON concentrations is essential when using different cell lines, likely due to differences in endogenous signaling levels.

Cell density
It is important to ensure that the optimum number of cells is seeded at the indicated time points to ensure successful differentiation. At later stages of differentiations (day 18 onwards), the neurons will form foci with bundled axons, which can be fragile and prone to movement. To address this: Always apply medium gently to the wall of the plate wells; avoid applying it harshly or directly onto the cells Conduct half medium changes by gently aspirating only half of the medium at a time, as this can be gentler for the cells Supplement media with additional Geltrex during medium replacement to enhance neuron adhesion The cells are highly proliferative between days 3 and 10 of differentiation, so it is important to plate cells at the correct densities on days 3 and 7 to prevent over-confluency and cell death. We have recommended optimal densities at each stage, but these should be fully optimized by each user and for individual hPSC lines.

Troubleshooting
For troubleshooting guidelines, see Table 2.