Morphological properties of ipRGC subtypes during development
In order to assess the morphological and physiological properties of ipRGC subtypes during development, we first needed to confirm that we could reliably identify each subtype at early postnatal stages using criteria available to differentiate the adult subtypes. We chose to focus on M1, M2, and M4 ipRGCs because the properties of these subtypes are well characterized and they have been previously shown to tile the retina [25,26,27]. M1, M2, and M4 ipRGCs can be differentiated by their dendritic stratification in the inner plexiform layer (M1: OFF stratifying and M2, M4: ON stratifying) and by presence (M4) or absence (M1, M2) of SMI-32 immunolabeling. We therefore first wanted to determine whether we could identify ipRGC subtypes during postnatal development using these same criteria: M1 ipRGCs, OFF stratifying and SMI-32 negative, M2 ipRGCs, ON stratifying, SMI-32 negative, and M4 ipRGCs, ON stratifying, SMI-32 positive. We targeted ipRGCs in Opn4-GFP mice for patch clamp recordings of ipRGCs at P6, P8, P10, P14, and Adult ages and filled cells with neurobiotin. We then performed immunohistochemistry for SMI-32 and choline acetyltransferase (ChAT), determined whether each cell was SMI-32 positive and whether it was ON or OFF stratifying (using ChAT bands as a reference). Using the aforementioned subtyping criteria, we find that we can indeed clearly identify these three ipRGC subtypes in our earliest time point, P6 (Fig. 1). We therefore continued to use this method to categorize all ipRGCs going forward.
Interestingly, when we mapped the lamination patterns of M1, M2, and M4 ipRGCs at P6 and adulthood, we found that all ipRGC subtypes had a different lamination pattern compared to their adult counterparts with the M1 subtype being most similar to adulthood (Fig. 1b). In contrast, the M2 and M4 subtypes seem to experience a bigger change in lamination pattern as cells mature. We observed that the M2 ipRGCs stratify closer to the middle of the IPL in early postnatal development before refining this dendritic lamination to the innermost portion of the IPL in adulthood (Fig. 1d) and that the M4 ipRGCs have dendrites stratifying closer to the ganglion cell layer in early postnatal development but then moving slightly closer to the middle of the IPL in adulthood (Fig. 1f), in agreement with previous observations of adult M2 and M4 ipRGC morphology [28]. These findings suggest that although the M1, M2, and M4 ipRGC dendrites broadly stratify within the correct layer early on, their dendritic stratification undergoes refinement in later parts of postnatal development.
The ability to define ipRGC subtypes early in development affords us the opportunity to characterize the progression of ipRGC structural and functional development in a way that is not possible for most RGC types. By using SMI-32 and dendritic stratification we then went on to assess ipRGC subtypes morphology and physiology through development. We first analyzed the morphological changes that occur in each ipRGC subtype during postnatal development. To do this, we filled M1, M2, and M4 ipRGCs with Neurobiotin at P6, 8, 10, 14 and Adult stages. We measured soma size, dendritic field diameter, and total dendritic length, and performed Sholl analysis to assess the complexity of the dendritic arbors (Fig. 2). We found that soma size remained constant across development in M1 and M2 ipRGCs, but increased in M4 ipRGCs (Fig. 2h). With regards to dendritic field size, we found that M1 ipRGCs exhibit adult dendritic field size and length by P10 (Fig. 2c, d), while M2 ipRGCs mature by P14 (Fig. 2f, g) and M4 ipRGCs continuing to expand their dendritic field size and complexity beyond P14 (Fig. 2i, j).
In adulthood, M1 ipRGCs have the smallest somata and smallest, least complex dendritic arbors amongst these three subtypes while M4 ipRGCs have the largest somata, as well as the largest and most complex dendritic arbors [1, 25, 28] (Fig. 3f). We therefore next examined whether the reported morphological differences between adult ipRGC subtypes could be detected at early postnatal stages (Fig. 3). Interestingly, at P6, we find that M1 ipRGCs have the largest dendritic field diameter while M4 ipRGCs have the smallest, which may be reflective of a faster rate of maturation for M1 ipRGC morphology (Fig. 3a). All three subtypes exhibit similar total dendritic length at this age while in adulthood M4 cells have the largest total dendritic length of the three subtypes (Fig. 3c, d). Of note, we found a large spread in the morphological measurements for the M1 and M4 subtypes the adult stage (Fig. 4), and so we did not find that the subtypes were significantly different in dendritic field diameter (Fig. 3b), as had been previously reported [1, 25, 28]. The M4 variation is likely a function of the large differences in M4 ipRGC arbors from nasal, where M4 cells are very large, to temporal retina where M4 cells are very small [29]. Additionally, M1 ipRGCs have been reported to show large variation in their morphological (and biophysical) properties [30]. Sholl analyses comparing morphological complexity between all three subtypes reveals that the M2 and M4 ipRGCs begin to exhibit more complex dendritic arbors than M1 ipRGCs at early postnatal stages (Fig. 3e, f).
Physiological properties of ipRGC subtypes during development
Following morphological analysis, we next characterized the intrinsic physiological properties of M1, M2, and M4 ipRGCs across development. In general, the intrinsic physiological properties of each subtype were relatively stable across development (Fig. 5). We observed that M1 cells have a downward trend in capacitance and input resistance as cells age (Fig. 5c, d) while M2 and M4 cells experience a drop in capacitance between P14 and adult, as well as a downward trend in input resistance as development progresses (Fig. 5g, h, k, l). The variation in capacitance and resistance in particular are likely to be a combination of changes in membrane surface area, intrinsic membrane properties, and electrical coupling with a surrounding network of cells [31]. When we directly compared the input resistance and resting membrane potential of M1, M2, and M4 ipRGC subtypes at P6 and Adult ages, we found that M1 cells have a more depolarized resting membrane potential and higher input resistance even early in development (Fig. 6a, c). These differences mimic those previously observed in light adapted tissue for adult M1 versus M2 and M4 ipRGCs [20, 32] as well as our own observations (Fig. 6b, d).
We next compared the spiking properties and action potential waveform of M1, M2, and M4 ipRGCs. We performed current clamp recordings from each of these subtypes and injected 1 s stepwise depolarizing current of 10 or 20pA until cells reached depolarization block. M1 ipRGCs show very few action potentials evoked by positive current (Fig. 7a, b), as reported previously for light-adapted M1 cells [9, 20]. In contrast, the M2 and M4 subtypes are much more excitable during development with the M4 subtype significantly increasing in excitability as cells mature (Fig. 7a, d, f). Somewhat surprisingly, the current density needed to reach the maximum spiking frequency was not significantly different across ages (Fig. 7c, e, g) for each of the subtypes. We next analyzed several components of individual action potentials from each subtype including width at half max, threshold, and fast after hyperpolarization (Fig. 8). Unsurprisingly, we find that action potential width at half-max decreases for all cell types across development (Fig. 8b, e, h) which is in line with typical progression of neuronal development [33, 34]. We also observe that threshold decreases for the M2 and M4 subtypes as cells mature (Fig. 8f, i).
In addition to the intrinsic properties of M1, M2, and M4 ipRGCs, we also examined the ipRGC light response across development. We performed current clamp recordings of ipRGC light responses to 30s of saturating blue light stimulus at 1 × 1017 photons/cm2 s− 1 at P6, P8, P10, P14, and Adult. In general, we found that all subtypes exhibited adult-like light responses by P14 (Fig. 9), consistent with the intact synaptic circuitry in the retina around the time of eye opening [15, 16]. Specifically, M2 and M4 ipRGCs, which are known to receive strong drive from the cone pathway [1, 28, 35], show faster and larger light responses as development progress (Fig. 9d-g). M1 ipRGCs, however, had statistically similar light responses throughout development (Fig. 9b, c). This is in line with previous reports that M1 ipRGCs are strongly driven by melanopsin phototransduction in bright light [35], and indicates that M1 ipRGCs show mature light responses from early developmental stages. M1 cells also showed strong depolarization block in their light responses, as reported previously [36]. We note that there is a large variability in the light response for the M2 subtype at early developmental timepoints (Fig. 9a, d, e), which could mean that the M2 population is quite diverse earlier in development. This may reflect a difference between M2 cells that project to image and non-image forming visual regions, though this remains to be tested [25]. Interestingly, when we compared ipRGC light responses early in development and adulthood, we observe that the maximum depolarization in response to light and onset time is statistically the same between all subtypes during development, although the M1 subtype tended to have a slightly larger depolarization and faster onset time on average (Fig. 10a, c). In adulthood, we see that again all subtypes have a similar maximum depolarization, however, the M1 subtype has the slowest onset time (Fig. 10b, d).
Assessing the embryonic birthdate of ipRGC subtypes
Overall, our results suggest that ipRGC subtypes mature at different rates during postnatal development. We next asked whether these differences in maturation rate might be reflected in differences in cellular birthdate. That is, do the M1, M2, and M4 subtypes terminally differentiate at different embryonic timepoints, and how does this compare to the birthdate of conventional RGCs? To answer this, we utilized 5-ethynyl-2′-deoxyuridine (EdU), a thymidine analog, to label cells that terminally differentiated on specific embryonic days, also known as birthdating. We first compared the birthdate of all ipRGCs, M1–3 ipRGCs, and Brn3a-positive RGCs (non-ipRGCs). To do this, we quantified the percentage of cells that were EdU and GFP positive in both Opn4Cre/+; Z/EG animals (where all ipRGCs are labeled with GFP; Fig. 11a) and Opn4LacZ/+; Opn4-GFP animals (where only M1-M3 ipRGCs are labeled with GFP and only M1 ipRGCs are labeled with LacZ; Fig. 11a) from Embryonic Day E11–14. We also immunostained for a non-ipRGC population of RGC, the Brn3a-positive RGCs (Fig. 11a), and counted the number EdU-positive, Brn3a-positive RGCs from E11–14. While ipRGCs appear to be born primarily on E11 and E12 (Fig. 11b, c), we observed that Brn3a positive RGCs continued to terminally differentiate at E13 and E14, suggesting that ipRGC birthdates differ from other RGC types.
We next wanted to assess and compare the birthdate of individual ipRGC subtypes (M1, M2/3, and M4 ipRGCs). To identify M4 ipRGCs from “non-M4” ipRGCs, we immunolabeled Opn4Cre/+; Z/EG retinas for SMI-32. M4 ipRGCs are easily identified as GFP positive, SMI-32 positive, while non-M4 ipRGCs are GFP positive, SMI-32 negative (Fig. 12b). In this line, OFF alpha RGCs can also be identified as GFP negative, SMI-32 positive. To differentiate M1 and M2/3 ipRGCs, we immunolabeled Opn4LacZ/+; Opn4-GFP mice for LacZ and GFP. M1 ipRGCs will be both GFP and LacZ positive (Fig. 12a), while M2/3 ipRGCs should be GFP positive, LacZ negative (though some M3 ipRGCs may be LacZ positive, see [37]; Fig. 12a). In agreement with our broad comparisons in Fig. 11, we find that M1, M2, and M4 ipRGCs are all primarily born on E11 and E12 (Fig. 12c, d). Interestingly, when we compared the birthdate of M4/ON alpha RGCs and OFF alpha RGCs, we find that the OFF alpha RGCs continue to be born through E13 (Fig. 12e, f), highlighting an important difference in birthdate between the ON and OFF alpha RGC population, despite these cells being considered part of the same class of (alpha) RGC.