El fundamento de validez en el enfoque crítico

Calcium is one of most important signaling molecules utilized by neurons. This fact is often overlooked due to the power of using calcium influx as a proxy for neuronal activity. Furthermore, while calcium influx is closely correlated with neuronal spiking, calcium indicators operate on a much slower time scale than do action potentials, increases in intracellular calcium can frequently occur independent of action potentials, and action potentials can even occur in the absence of significant calcium influx. These points warrant careful consideration. Thus, I will discuss two important points regarding the interpretation of calcium imaging data in this section. First, what are the

caveats/pitfalls of using calcium imaging as a proxy for neuronal activity/how should a calcium trace be interpreted? Second, what does measuring calcium influx (as opposed to electrical activity) mean for subsequent changes occurring within the cell? Though most of the following discussion pertains to all of the modern GECIs, the following discussion will focus on the GCaMP family of indicators.

The typical calcium transient corresponding to a single action potential has a fast rise time and slow decay (~0.1 second and 1 second respectively, see Chen et al., 2013); this is due to the dynamics of calcium binding with GCaMP (Nakai et al., 2001).

However, even the relatively fast rise time of a calcium transient, which occurs during spiking activity of a neuron, is much slower than the time-scale of the ion influx/efflux causing a typical action potential (Hodgkin & Huxley, 1952). Individual action potentials can be inferred from calcium activity in some cases (Chen et al., 2013). However, even with high signal-to-noise preparations, e.g. sparsely labeled neurons recorded using a two-photon microscope, deducing action potentials underlying the calcium signal can be difficult (Pnevmatikakis et al., 2016). This is further compounded by the fact that the amplitude and rise time of a calcium trace increases with the number of spikes occurring (Chen et al., 2013). This increased signal is likely due to the interaction of calcium ions entering the cell through voltage-gated calcium channels (VGCCs) at the soma and through calcium-permeable NMDA receptors in the dendrites (Magee et al., 1998; Markram, Helm, & Sakmann, 1995). While calcium entry into the soma occurs primarily through VGCCs (Mao, Hamzei-Sichani, Aronov, Froemke, & Yuste, 2001), large

amounts of calcium can induce additional calcium release from internal stores in the endoplasmic reticulum (Grienberger & Konnerth, 2012; Kano, Garaschuk, Verkhratsky, & Konnerth, 1995; Tsien & Tsien, 1990), further magnifying the detected calcium signal. The detection of single action potentials thus becomes difficult to impossible in a lower signal-to-noise preparation, e.g. single-photon imaging with a miniscope, since the amplitude of these events is unlikely to exceed the noise floor. The non-linear amplification of transient amplitude with increasing number of action potentials also makes interpreting the height of a calcium transient difficult. On the other hand, single action potentials can sometimes even occur in the absence of significant calcium entry

into the cell (Magee et al., 1998). Calcium imaging is therefore biased toward recording from high firing rate neurons and indicates significant amounts of calcium entry into the cell.

Bursting events require calcium entry through both the soma and dendrites, and the main source of dendritic calcium influx is through NMDA receptors (Bloodgood & Sabatini, 2009). Calcium entry through NMDA receptors is also vital for induction of long-term plasticity in synapses (Malenka & Bear, 2004; Shepherd & Bear, 2011). Thus neurons detected with calcium imaging are likely those which will undergo significant remodeling/plasticity at their afferents. Is this the primary reason why a number of recent calcium imaging studies have demonstrated significant turnover in the active cell

population over time (Cai et al., 2016; Kinsky et al., 2018; Mau et al., 2018; Rubin et al., 2015; Ziv et al., 2013)? Would we see no turnover/slower turnover using different methods that detected activity from less bursty cells which were subsequently less likely to undergo plasticity? Evidence suggests that the answer is no, since previous studies have found significant turnover in hippocampal activity patterns using electrophysiology over short to intermediate time-scales (Mankin et al., 2015, 2012; Manns et al., 2007). Furthermore, recent studies using different, non-calcium related, imaging techniques have found that plasticity is ubiquitous in hippocampal neurons even in highly familiar

circumstances (Attardo et al., 2018), presumably because afferent inputs to CA1 exhibit 100% turnover over the course of a month (Attardo et al., 2015). Still, this bias toward higher firing rate, more plastic neurons must be considered, especially since the rules regarding how downstream structures interpret hippocampal outputs are poorly

understood (Lisman et al., 2017). On the other hand, vesicle release due to a single action potential is unreliable (Korn, Triller, Mallet, & Faber, 1981); bursting of neurons may provide a method for reliable transmission of signal at downstream synapses (Lisman, 1997). Furthermore, calcium entry through low-threshold VGCCs is linked with BDNF mediated long-term cell survival in cultured neurons (Ghosh, Carnahan, & Greenberg, 1994; Lyons & West, 2011). Thus, calcium imaging’s bias toward the recording of more plastic neurons could also provide a readout of the more reliable signals transmitted to downstream regions.

Much of the early work utilizing calcium imaging focused on the interaction between calcium signal in different compartments of the neuron during spontaneous and evoked activity and how that contributed to firing activity/plasticity (Davie, Clark, & Hausser, 2008; Mao et al., 2001; Markram et al., 1995; Schiller, Schiller, Stuart, & Sakmann, 1997; Yuste & Denk, 1995). Recent work has continued this trend, using head- fixed two-photon imaging in conjunction with active behavior to elucidate how/why place cells emerge (Sheffield, Adoff, & Dombeck, 2017; Sheffield & Dombeck, 2014). Calcium entry through dendrites is intimately tied to the presence of plateau potentials – sustained, sub-threshold rises in membrane potential result from precisely timed input between ECIII and CA3 inputs to distal/proximal regions of CA1 dendrites (Bittner et al., 2015; Sheffield & Dombeck, 2019). Current injections during plateau potentials,

observed with whole-cell patch recordings, were sufficient to induce place field

formation (Bittner et al., 2015). Likewise, the prevalence of dendritic calcium transients ramped up immediately prior immediately prior to, and in the same location as, the

formation of a place field for a neuron (Sheffield et al., 2017). These two studies suggest that the detection of somatic calcium implies the existence of high calcium influx through dendrites. Furthermore, another study found that neurons with reliable activity in most of their dendritic branches also had the most stable place fields (Sheffield & Dombeck, 2014), suggesting that careful monitoring of the reliability of calcium activity could provide valuable information about the inputs to a given neuron. Though these studies all considered the formation of place fields, the same mechanisms could also apply to the emergence of hippocampal neurons responding to non-spatial features of a task (Aronov et al., 2017; Muzzio et al., 2009; Pastalkova et al., 2008; Robinson et al., 2017; Wood et al., 1999). Thus, though there are several caveats underlying the use of calcium imaging that warrant careful consideration, calcium imaging can also provide valuable

information about both the gross spiking activity of neurons as well as their inputs.