A wealth of studies has unraveled the environmental factors, signaling, and neuron-intrinsic mechanisms of dendritic growth with an emphasis on growth-promoting effects of neural activity (Wong and Ghosh, 2002; Cline and Haas, 2008). Depolarization evokes calcium currents via voltage-gated channels and NMDA receptors (Konur and Ghosh, 2005), which can trigger local protein synthesis and the release of trophic factors acting in a paracrine or autocrine manner (Wong and Ghosh, 2002; Wirth et al., 2003; Redmond, 2008). Biologically meaningful synaptic activity, for instance, elicited by rearing in an enriched environment or by training, increases dendritic complexity (Volkmar and Greenough, 1972; Greenough et al., 1979). Neurons of the upper layers respond differently than those of the deep layers which could be related to sequential genesis, migration, and arrival in the layers. Moreover, apical and basal dendrites of pyramidal cells differ functionally (Schiller et al., 2000; Milojkovic et al., 2005; Nevian et al., 2007; Major et al., 2008; Antic et al., 2010; Chalifoux and Carter, 2011) and respond differentially to morphogenetic signals. For example, antagonizing the N-methyl-D-aspartate receptor (NMDAR) subunit GluN2B in organotypic cultures (OTCs) of the rat visual cortex from DIV 7–10 results in impaired growth of basal but not apical dendrites of pyramidal cells (Gonda et al., 2020). The overexpression of the kainate receptor subunit GluK2 results in the growth of apical dendrites in layers II/III but not in layers V/VI (Jack et al., 2019). Calcium signaling evokes expression of BDNF, and neurotrophins exert layer- and type-specific effects on dendritic maturation of cortical neurons (Redmond, 2008). When BDNF is overexpressed in OTCs from the rat visual cortex until DIV 10, only the dendrites of infragranular pyramidal cells show increased growth (Wirth et al., 2003). Early postnatal overexpression of certain AMPA or kainate glutamate receptor subunits or transmembrane AMPA receptor regulatory proteins and a NETO protein auxiliary to kainate receptors increases apical dendritic complexity because the amplitude of depolarizing events driven by spontaneous activity increases with more excitatory receptors (Hamad et al., 2011, 2014; Jack et al., 2019). Stimulation frequency also plays a role. A previous study has shown that neurite growth is at its best within an optimal calcium concentration and frequency of calcium transients (Kater and Mills, 1991; Gomez and Spitzer, 1999). In this study, we tested whether dendritic complexity of channelrhodopsin2-eYFP (ChR2-eYFP) expressing cortical pyramidal neurons can be altered with repetitive optogenetic stimulation. Indeed, stimulation at a biologically meaningful range for the selected developmental stage evoked remodeling of dendrites. Contrary to expectation, we observed a reduction of dendritic length of infragranular neurons.

Organotypic cortex cultures preserve the cortical architecture and develop spontaneous action potential activity and activity-dependent neurotrophin expression (Gorba et al., 1999; Klostermann and Wahle, 1999; Chattopadhyaya et al., 2004; Baho and Di Cristo, 2012). Our hypothesis was that we could evoke dendritic growth with an optogenetic stimulation at frequencies that are within the range of those that occur naturally during or shortly after the selected developmental time window.

The optogenetic stimulation elicited activity of the ChR2-expressing neurons as indicated by c-Fos protein expression and ChR2/c-Fos double-labeled neurons increased with increasing pulse frequency and pulse duration. One might ask why not all ChR2 neurons turned c-Fos positive. With the timeline used, we were near the half-life of c-Fos expression. Thus, apart from the technical aspect of antibody penetration in slice cultures, the expression might have already declined to below the detection level in some ChR2-positive neurons. It suggested that an acute effect could have yielded even higher numbers of c-Fos positive neurons. Furthermore, stimulation frequencies close to the range of naturally occurring frequencies can alter dendritic complexity. Against hypothesis, however, the 0.5 Hz stimulation reduced dendritic complexity selectively of infragranular pyramidal cells in both time windows.

Initially, during development, the network grows extensively, followed by a period of decreased growth and pruning of weak connections. Pruning of dendrites is important for the refinement of the network (Neely and Nicholls, 1995). The expression of constitutively active CaMKII in Xenopus optic tectum neurons restricts the growth of dendrites, whereas the inhibition of CaMKII enhances dendritic growth (Wu and Cline, 1998). The activation of cytosolic and membrane-associated CamKI is the essential first step of NMDA receptor and calcium influx-dependent dendritic elongation and branching of hippocampal neurons, involving MAP kinase signaling and CREB activation further downstream (Wayman et al., 2006). The growth of neurites follows a hormetic principle, in that factors which promote dendritic growth can also evoke the opposite effect when administered at too high doses (Kater and Mills, 1991; Gomez and Spitzer, 1999; Rubio-Casillas and Fernández-Guasti, 2016). For example, in pyramidal cells, GluN2B receptor signaling promotes basal dendritic growth (Gonda et al., 2020), but excessive NMDA receptor activation, e.g., by disinhibition, results in dendritic injury and retraction (Ikegaya et al., 2001; Nishimura et al., 2008; Martin and Wellman, 2011), and neurons which survive such excitotoxic activation show a decreased dendrite growth (Monnerie et al., 2003). Nevertheless, neurons are capable of preventing cell damage by homeostatically downregulating transmitter receptors which renders neurons less excitable (Wang et al., 2016) and subsequently less responsive to growth-promoting stimuli. Calcium is a key player in dendritic growth and maintenance (Wayman et al., 2006). The disruption of calcium signaling, e.g., upon expression of the indicator GCaMP3, leads to less complex apical dendrites (Gasterstädt et al., 2020) because calcium indicator proteins are sequestering calcium. For instance, chronic optogenetic stimulation of DIV 21–30 pyramidal cell populations in sparsely transfected Stoppini-type cultures of the mouse somatosensory cortex has been shown to reduce the frequency and amplitude of UP states of the transfectants via homeostatic mechanisms to reduce their excitability and to cause a synaptic decoupling between ChR2-expressing stimulated and neighboring ChR2-negative neurons (Liu et al., 2023). Assuming that such a functional shift might have happened already in our much younger cultures, such a cell-intrinsic effect could evoke the reduction of apical dendritic length of supra- and infragranular pyramidal cells.

Furthermore, inhibition may play a role. For instance, chemogenetic silencing of neurons reduces the growth of apical dendrites of layer II/III neurons and their axonal collateralization (Gasterstädt et al., 2022). In line, excessive hyperpolarization by axosomatic parvalbumin-positive interneurons has been implicated in the stunted dendritic growth typical for Rett syndrome (Armstrong et al., 1995; Durand et al., 2012; Tomassy et al., 2014; Patrizi et al., 2020). In Down's syndrome, overinhibition and a reduced BDNF expression have been implicated in the reduction of dendritic complexity (Dierssen et al., 2003; Zorrilla de San Martin et al., 2018). Strong inhibition weakens action potential backpropagation into dendrites, and this could reduce release of neurotrophins. Backpropagation is known to trigger BDNF release (Kuczewski et al., 2008) which can increase dendritic complexity (Wirth et al., 2003; Redmond, 2008). Overexpressing BDNF in P25 ferret visual cortex slice cultures evokes a massive sprouting of basal dendrites via autocrine signaling; however, these short dendrites and in particular their spines are highly unstable. In addition, dendritic branches of ~90 μm from the soma are reduced, indicating that excessive BDNF destabilizes dendritic growth (Horch et al., 1999). We could not see such an increase in the number of basal dendrites nor any significant reduction of higher order basal branches (except for in the 2.5 Hz experiment, possibly). Rather, between DIV 10/11 and DIV 15, the basal dendrites of supra- and infragranular pyramidal neurons grow substantially. We, therefore, suggest that the optogenetic stimulation did not elicit excessive BDNF production in the transfectants, and furthermore, the stimulation per se did not have any global negative effect on the neuronal structure.

Transcranial optogenetic stimulation (180 light pulses, each 3 s long with 7 s intervals) to activate ChR2-positive medial prefrontal cortex L2/3 pyramidal cells in mice from P7-12 elicits an increase of dendritic length and an increase in the power of endogenous oscillations (Bitzenhofer et al., 2021). The length increase is transient and normalizes to control levels at P23 and older (Bitzenhofer et al., 2021). Although postnatal age is comparable to our developmental time windows, we could not observe any length increase. Probably, in addition to the context (e.g., species, regional, and cell-type specific ion conductances, in vivo/ex vivo), the exact stimulation protocols are of critical importance. Bitzenhofer et al. (2021) also reported an increase in inhibition with more parvalbumin-positive neurons in the stimulated cortex which have longer dendrites and exert stronger inhibition as suggested by slightly larger amplitudes of inhibitory postsynaptic potentials. The optogenetic activation of pyramidal neurons in DIV 19 hippocampal slice cultures increases the number and the size of gephyrin-positive postsynaptic inhibitory domains along their proximal apical dendrites within 24 h in an NMDA receptor-dependent manner (Flores et al., 2015), and GABAergic terminals grow larger and release more GABA upon NMDA receptor stimulation (Fiszman et al., 2005). Alternatively, the optogenetic activation of layer II/III pyramidal cells could have contributions. For instance, in the somatosensory cortex, descending layer II/III projections suppress the spontaneous activity of layer V pyramidal cells via the activation of dendrite-targeting interneurons (Pluta et al., 2019). This could explain why our infragranular pyramidal cells were so strongly affected. Together, it remains to be shown if an engagement of inhibitory neurons via network activity elicited by the 0.5 Hz stimulation might have contributed to the substantial apical dendritic length reduction of infragranular pyramidal cells.

Repetitive ChR2-evoked depolarization might have compromised metabolism and energy allocated to neuronal compartments. Bioenergetics might deliver a simple reason: If neurons are substantially depolarized, they may suffer energy shortage and start to save on neurite outgrowth. Action potential burst activity reduces dendritic ATP levels and cortical dendrites are more vulnerable to energetic and redox fluctuations than somata (Hasel et al., 2015). Indeed, healthy mitochondria that are actively transported into dendrites are essential for dendrite growth and maintenance (Tsubouchi et al., 2009; Fukumitsu et al., 2015; Shen et al., 2019). Though worth a thought, the metabolic argument falls short of explaining why supragranular pyramidal neurons were only subtly affected. Moreover, one would expect to see an even stronger growth impairment after the 2.5 Hz stimulation. ChR2-mediated sodium currents render the neurons more active, and it has been shown in zebrafish spinal cord that electrically highly active neurons have the lowest dendritic filopodial dynamics (Kishore and Fetcho, 2013) and have high chances to survive (Warm et al., 2022). In the rodent retina, depolarization-evoked release of calcium from intracellular stores stabilizes dendrites (Lohmann et al., 2002). Over time, this may result in less complex dendritic trees than control neurons. In this view, the ChR2 stimulation would not trigger a regressive process but would rather stabilize the dendritic dimensions present at the beginning of the stimulation. Indeed, the total internode dendritic length present at DIV 15 (see Figure 5D, right) is close to the measure obtained at DIV 10 (see Figure 5C, right). For axons, excessive calcium influx evoked by ChR2 stimulation into growth cones stops axonal growth (Huang et al., 2017). In line, stimulating with 4 and 10 Hz, but not 0.1 and 1 Hz, has been shown to temporarily slow down axonal elongation (Malyshevskaya et al., 2013), suggesting that the higher frequencies resemble the spontaneous activity level at a later time period when axonal elongation has ceased, and axons instead elaborate terminal branches.

These arguments do not explain why the infragranular neurons were strongly affected by the 0.5 Hz stimulation, whereas supragranular neurons were only mildly affected. However, the study (Malyshevskaya et al., 2013) lends support to our interpretation that “more of the same” will not result in morphological changes. Indeed, our 0.05 Hz stimulation was morphogenetically ineffective at DIV 10 and at DIV 15. It seems as if morphological changes can be elicited with activity patterns which resemble but do not far exceed those adopted at “the next developmental level.”

The 2.5 Hz stimulation did not alter apical dendritic complexity and also did not evoke c-Fos expression higher than the proportion reached with 0.5 Hz stimulation, suggesting a saturation. Tentatively, the small differences toward less branched basal dendrites occurring in a few Sholl bins have been caused by interbatch variability since six independent preparations had to be exploited for this condition to obtain sufficient numbers of neurons. Furthermore, the higher incidence of neuronal damage of 2.5 Hz stimulated pyramidal cells asks for caution with interpretation. For survival, neurons depend on optimal activation levels via spontaneous activity (Warm et al., 2022). The suppression of spontaneous activity and synchronized burst discharges in organotypic neocortical slices of neonatal mice by application of TTX increases the apoptosis rate (Heck et al., 2008). Non-physiological activity patterns evoked by optogenetic stimulation in primary hippocampal cultures can also increase the cell death rate (Wong Fong Sang et al., 2021). Apparently, the 2.5 Hz stimulation with higher activity levels of the ChR2 transfectants and presumably also of numerous non-transfected, c-Fos positive neighboring neurons of the network is beyond of what the still immature DIV 11–15 neurons can cope with.

A biologically intriguing explanation of the layer-specific effect comes from looking at what happens in developing the visual cortex in vivo during the time window. Immature postmigratory pyramidal cells initially have their apical dendrites ending in layer I. Perinatally, a period of remodeling begins. Corticothalamic layer VI pyramidal cells retract their apical dendrites which now terminate in layer IV. Between P7 and P10 in the visual cortex, numerous pyramidal cells of layer V forming callosal projections as indicated by retrograde labeling retract the apical dendrites which thereafter end rather unconspicously in middle layers (Koester and O'Leary, 1992; Kasper et al., 1994). Concurrently, the large subcortically projecting layer V pyramidal neurons elaborate their apical tufts in layer I. This fits well with the apical length reduction we observed with the 0.5 Hz stimulation from DIV 5–10. Indeed, the total terminal length of infragranular pyramidal cells was reduced at DIV 10 but no longer at DIV 15, which could be interpreted as a pruning of distal dendritic elements. Regressive events are influenced by neuronal activity. Indeed, in the prefrontal cortex, cholinergic signaling via α5 nicotinic receptors has been implicated in the remodeling of layer VI apical dendrites (Bailey et al., 2012). The role of activity has been extensively demonstrated for spiny stellates of layer IV which remodel their initially polarized pyramidal-like appearance. The apical dendrite regresses, and the basal dendrites elaborate to form the typical stellate pattern with dendrites restricted to single barrels in layer IV, which at the same time becomes heavily innervated by thalamocortical afferents (Lund et al., 1991; Callaway and Borrell, 2011; Nakazawa et al., 2018). However, when deprived of thalamocortical input activity, spiny stellates of the mouse somatosensory cortex remain polarized and do not adopt radial basal dendritic fields (Li et al., 2013). The GluN2B receptor regulates this patterning in a cell-autonomous manner, and a lack of GluN2B results in stellate cells extending dendrites into multiple barrels (Espinosa et al., 2009). In OTC, the driving force of subcortical afferents is lacking, and so, dendritic remodeling may proceed more slowly in OTC. Suggestively then, the optogenetic stimulation accelerates the naturally occurring remodeling, and this way helps infragranular neurons to acquire their characteristic morphology. Although intriguing, infragranular pyramidal cells were not the only ones impaired by the 0.5 Hz stimulation.

As shown by our staining, ChR2-YPF labeled the neuronal membrane completely. Activation of ChR2 evokes a sodium influx which may subsequently recruit calcium signaling. An evoked depolarization and calcium influx is considered beneficial for plastic changes when it happens synaptically. In contrast, extrasynaptic signaling can be neurotoxic due to a shut-off of the CREB pathway and BDNF production (Hardingham et al., 2002). In hippocampal neurons, the shut-off emerges between DIV 7 and DIV 12, and at DIV 12, extrasynaptic NMDA receptor activation overrides the beneficial CREB-activating signal elicited by synaptic NMDA receptors (Hardingham and Bading, 2002). In the neocortex, the ontogenetically older pyramidal cells of layers V/VI are more mature than the supragranular pyramidal cells and possibly switch to CREB shut-off earlier during development. This assumption could explain why infragranular pyramidal neurons were strongly affected by optogenetic stimulation. In summary, our results suggest that optogenetic stimulation at a biologically meaningful frequency for the selected developmental stage can influence dendrite remodeling. Contrary to expectation, the stimulation decreased dendritic growth in a layer-specific manner. The underlying mechanisms remain to be elucidated.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

The animal study was reviewed and approved by Ruhr University Bochum Animal Research Board.

PW conceived the experiments. AR prepared the cultures. SG, IK, AH, KC, CR, and PW performed the experiments, reconstructions, and data management. SG, IK, and PW analyzed the results. PW wrote the manuscript. All authors contributed to the article and approved the submitted version.