The peripheral nervous system (PNS) is necessary to interact with our environment and to maintain body homeostasis (Saito-Diaz and Zeltner, 2019; Jakob et al., 2021). As part of the PNS, sensory neurons (SNs) connect and feed information from organs and tissues to the brain (Abraira and Ginty, 2013; Donnelly, Chen and Ji, 2020). SNs arise from multipotent neural crest (NC) cells that delaminate from the border between the ectoderm and the neural plate during neural tube formation (Marmigère and Ernfors, 2007). NC cells then migrate and cluster in ganglia (such as the dorsal root ganglia, DRG), where they further differentiate into SNs and innervate their target tissues (Marmigère and Ernfors, 2007; Lallemend and Ernfors, 2012). The DRG is composed of three main SN subtypes, which express different cell membrane proteins and detect various stimuli: 1) nociceptors (express TRKA and detect pain and temperature), 2) mechanoreceptors (express TRKB and detect touch), and 3) proprioceptors (express TRKC and detect body position relative to the environment) (Marmigère and Ernfors, 2007). However, during development, expression of different TRKs overlap in some SN subpopulations, for example, in TRKB⁺/C⁺ mechanoreceptors (Lallemend and Ernfors, 2012).

Due to lack of access to relevant tissues, it is difficult to study development and diseases of the PNS, including SNs. The human pluripotent stem cell (hPSC) technology can close this gap. Indeed, many protocols have been established to differentiate hPSCs into SNs, focusing on nociceptors (Pomp et al., 2005; Chambers et al., 2012; Young et al., 2014; Boisvert et al., 2015; Wilson et al., 2018; Holzer et al., 2022), mechanoreceptors (Schrenk-Siemens et al., 2015; Nickolls et al., 2020; Zhu et al., 2021), or proprioceptors (Dionisi et al., 2020). Other protocols can generate all SN subtypes in bulk DRG-like cultures (Alshawaf et al., 2018; Saito-Diaz et al., 2021). Each protocol has certain drawbacks, including the requirement of over expression of exogenous genes (Schrenk-Siemens et al., 2015), or relying on transdifferentiation (Holzer et al., 2022), or the fact that the number of neurons generated is low (Young et al., 2014; Schrenk-Siemens et al., 2015; Alshawaf et al., 2018; Wilson et al., 2018). A common issue is that the cultures are not pure and/or the overall efficiency of SN subtype generation is low. Thus, to further dissect the cellular mechanisms and understand the differences between neuronal subtypes it is necessary to develop methods to isolate and purify SN subtypes. Most current isolation methods rely on fluorescent activated cell sorting (FACS), however this method has drawbacks. While it enables protein or mRNA isolation followed by gene expression analysis (Guez-Barber et al., 2012; Miyazaki et al., 2020), dramatically reduced cell viability post sorting is a big hurdle that disables any further live cell analysis or cell manipulation, including drug testing.

Immunopanning is a method developed to isolate specific cell types and has been used, for example, to isolate astrocytes from murine brains (Zhang et al., 2016; Sloan et al., 2017). It relies on the presence of membrane proteins that can be used to isolate the desired cell types from a heterogenous mixture of cells. Antibodies against the selected protein are coated onto petri dishes (panning dishes). The mixture of detached cells is incubated in the panning dish and the cells expressing the membrane proteins bind to the antibodies. Then the panning dishes are washed to remove unbound cells, and the selected cells are dissociated and further cultured (Zhang et al., 2016; Sloan et al., 2017; Nolle et al., 2021). The gentle nature of the process promotes survival of the panned cells, which makes it an ideal method to isolate neurons for further long-term culture, manipulation, and analysis at various time points.

Here, we describe an updated method of immunopanning to isolate all three SN subtypes differentiated form hPSCs. We used the surface proteins TRKA, TRKB, and TRKC to select for nociceptors, mechanoreceptors, and proprioceptors, respectively. We show that the panned cultures are devoid of other SN subtypes by immunofluorescence and RT-qPCR. Finally, we show that they are electrically active and that they respond only to their expected stimuli. This work updates our previous report (Saito-Diaz et al., 2021) with new conditions to increase yield, purity, and viability of the panned SNs.

We have previously established a protocol to differentiate NC cells and SNs from hPSCs (Saito-Diaz et al., 2021; Saito-Diaz and Zeltner, 2022) (Figure 1A). Starting on day 8, we observe the formation of ridges (i.e., dark clusters) of NC cells (Figure 1B, arrows). Replating these cells on day 12, promotes differentiation into SNs, which are visible on day 16 and by day 20 (Figure 1B). By day 25 the neurons have acquired the morphology characteristic of SNs including axons connecting SN clusters (Figure 1B) (Saito-Diaz et al., 2021). We have shown that these cultures are composed of 70% nociceptors, 30% mechanoreceptors and proprioceptors (Saito-Diaz et al., 2021). Various research applications, including studies on mechanisms of development, axon regeneration, multi-omics, and characterization of neuronal subpopulations can enormously benefit from a method to isolate, healthy SN subtypes. The most common method to isolate and purify live cells is FACS. However, we have experienced that post FACS, sorted neurons do not survive. Similar results were shown in the literature (Guez-Barber et al., 2012; Miyazaki et al., 2020). Thus, more gentle and easy methods are required to isolate such cell types. We have previously shown that these neuron subtypes can be isolated via immunopanning (Saito-Diaz et al., 2021) (Figure 1C). Antibodies against the TRK family of surface proteins, provide an ideal way to isolate individual SN subtypes. TRKA can be used to isolate nociceptors, TRKB is expressed in mechanoreceptors, and TRKC is expressed by proprioceptors (Figure 1C). A subpopulation of SNs express both TRKB and TRKC, which is necessary to take into account when performing the immunopanning. It is important to note that for the success of this procedure the quality of the antibody is of utmost importance. Additionally, it is critical to start with healthy SNs, such as SNs showing clear cell bodies and low number of apoptotic cells (Figure 1D). Resuspended SNs can be observed under the microscope to monitor health of the cells during immunopanning. They should be circular, not forming aggregates and presence of debris should be minimal (Figure 1D). However, as the immunopanning progresses, it is normal to see some debris from dead cells. This does not negatively impact the results. To increase survival, it is critical to add Y- 27632 to the buffers and to the seeding media (SN differentiation medium). Cells that do not attach to the panning dish will move with the media if the dish is gently tapped (Figure 1D). The following day, live SNs are firmly attached to the bottom of the wells, and some are already growing axons. Five days post immunopanning, cells start showing large axon extensions. After 20 days, SNs aggregated and axons are clearly visible by brightfield microscopy (Figure 1D).

To confirm that the panned cells indeed had the identity of SNs, we performed IF for the markers TRKA, TRKB, and TRKC. We found that SNs isolated using the TRKA antibody, were negative for TRKB and TRKC (Figure 2A, top row), suggesting that nociceptors were successfully isolated. Additionally, cells that were panned using the TRKB or TRKC antibodies did not express TRKA by IF (Figure 2A and image quantification in Figures 2B–D) suggesting that the correct SN subtypes were successfully isolated. However, some non-neuronal cells were carried over to the final culture, highlighting the importance of the washes after the panning step (Figures 2A–D). We also found presence of TRKB⁺ cells in TRKC-panned SNs. (Figure 2D). This is possible because some TRKB⁺ mechanoreceptors are also positive for TRKC (Figure 2E).

To confirm the purity of the panned cultures, we measured expression of additional markers by RT-qPCR. We found all the panned SNs expressed the general SN markers BRN3A and PRPH (Figure 3A). Agreeing with our previous results, we found that TRKA-panned SNs but not TRKB- or TRKC-panned neurons expressed high levels of nociceptor markers (TRKA, SST, and TAC1) (Figure 3B). Conversely, TRKB-panned SNs showed the highest expression of mechanoreceptor markers (TRKB, KNCK2, ASIC1, PIEZO2) (Figure 3C). Interestingly, TRKC-panned SNs also expressed relatively high levels of TRKB, possibly coming from cells that are TRKB and TRKC double positive (Figure 3C). Finally, TRKC-panned SNs showed the highest expression of proprioceptor-markers (TRKC, SPP1, and PVALB) (Figure 3D). However, TRKC was also highly expressed by TRKB-panned SNs (Figure 3D), possibly due to presence of TRKB⁺/TRKC⁺ SNs. These results suggest that panned cultures consist of mainly the panned SN and few contaminant cells.

To assess whether the panned SNs are still functional, we measured their electrical activity using multielectrode array (MEA) (Figure 4A). After immunopanning, the cells survived and successfully attached to the MEA plates 10 days post panning (Figure 4B). Furthermore, we found that SNs were electrically active at this time point (Figures 4C–J). Additionally, only TRKA-panned SNs were responsive to capsaicin, which is known to activate nociceptors, but not mechanoreceptors or proprioceptors (Figure 4C). Furthermore, capsaicin increased the number and frequency of bursts, but not their duration from TRKA-panned SNs (Figures 4D–F). In contrast, when the panned cells were incubated with hypoosmotic media, which mimics touch by generating pressure on the plasma membrane, only TRKB-panned mechanoreceptors responded appropriately by increasing their firing rate (Figure 4G). This is consistent with increased electrical activity by mechanoreceptors upon touch (Ranade, Syeda and Patapoutian, 2015). Similarly, hypoosmotic media increased the number, duration, and frequency of bursts generated by TRKB-panned SNs (Figures 4H–J). We saw a similar response from TRKC-panned SNs, however the response was lower compared to TRKB-panned cells. Overall, these data suggests that the panned SNs are functionally active and they respond to their proper stimuli.

Finally, we tested whether the immunopanning can be used in other hPSC lines. We differentiated SNs from a previously characterized induced pluripotent stem cells line: iPSC-ctr-C1 (Zeltner et al., 2016; Wu et al., 2022), which underwent the developmental stages previously reported: NC and SN (Figure 5A). We found that SNs from iPSC-ctr-C1 cells can be successfully immunopanned to isolate TRKA⁺, TRKB⁺, and TRKC⁺ SNs (Figure 5B). Thus, these results show that the immunopanning protocol can be successfully applied to a broader range of cells.

Here, we have shown that immunopanning is a simple and straightforward method to isolate live SNs. We used the membrane proteins TRKA, TRKB, and TRKC as markers to isolate nociceptors, mechanoreceptors, and proprioceptors, respectively. However, theoretically, any cell surface marker can be applied to this method. We showed that neurons survive the process healthily, which is a major advance over other methods including FACS, where neurons do not survive well. The panned neurons express markers that are characteristic to each subtype. Furthermore, the neurons are electrically active within 10 days of panning and respond to specific stimuli.

It is important to note that a culture of healthy SNs with large round cell bodies, straight axons protruding from them, with no or minimal presence of apoptotic cells is critical for the success of the protocol. We have immunopanned SNs ranging from day 20–25 and as late as day 40. Interestingly, we found that SNs closer to day 30 tend be more fragile compared to older or younger neurons. However, the reason of this remains unknown. Regardless of their panning day, SNs project axons and fire action potentials around 10 days post panning. Furthermore, regardless of the day the immunopanning was done, our results show that we can enrich the subpopulations of each SN by 2- to 3-fold, compared to the original protocol (Saito-Diaz et al., 2021), highlighting the power of this approach.

The quality and concentration of the antibodies used are also critical for the success of the protocol. We found that variations in lots have a significant impact in the quantity of SNs that can be recovered. However, higher concentration of antibody does not necessarily correlate with increased yield. Thus, antibody titration and careful lot testing is necessary to achieve the highest recovery rate.

The simplicity of our immunopanning method contrasts to other alternatives. For instance, although FACS can be used to also isolate SNs, its application is limited to transcriptomic and proteomic analysis because the harsh protocol results in cell death (Guez-Barber et al., 2012; Miyazaki et al., 2020). A recent protocol (van Niekerk et al., 2022) used MACS columns to isolate live neurons from brain and spinal cord from rats. However, the protocol also resulted in some contaminant cells in the final suspension. Also, the costs of MACS columns and their associated supplies limit their applications. In contrast, our method uses the same principle (positive isolation using antibodies against surface proteins) with reagents and supplies that are commonly found in any research lab and are relatively economical.

There are two major points where our isolation technique can be further improved in the future: 1) the scalability of the protocol and 2) improved purity of cultures. From a 10 cm dish of SNs (with ∼10 × 10⁶ cells) we can isolate up to 2 × 10⁶ cells per subtype. This suggests that there is a large number of neurons that are not captured by the antibodies. A future solution to this might be testing different antibodies and/or using larger panning dishes (Barres, 2014). Such measures may also reduce the number of contaminant cells in each panning dish.

Overall, our immunopanning protocol provides a straightforward, simple, and cost-effective way to isolate live SNs. Isolated neurons can then be further cultured for downstream analysis of live SN. Furthermore, the method we describe can be easily adapted to other neuronal types. The gentle nature of the protocol ensures cell survival; thus we expect that it can be translated to other fields.