Ionic current correlations are ubiquitous across phyla

Ionic currents, whether measured as conductance amplitude or as ion channel transcript levels, can vary many-fold within a population of identified neurons. This variability has been observed in multiple invertebrate neuronal types, but they do so in a coordinated manner such that their magnitudes are correlated. These conductance correlations are thought to reflect a tight homeostasis of cellular excitability that enhances the robustness and stability of neuronal activity over long stretches of time. Notably, although such ionic current correlations are well documented in invertebrates, they have not been reported in vertebrates. Here we demonstrate with two examples, identified mouse hippocampal granule cells and cholinergic basal forebrain neurons, that ionic current correlations is a ubiquitous phenomenon expressed by a number of species across phyla.

Here we test the hypothesis that ionic current correlations are widely distributed across animal species, and demonstrate that ionic current amplitude correlations are also expressed in mammalian neurons. We conclude that this is a ubiquitous phenomenon observed in species across phyla.

Methods
We report observations from two different cell types: hippocampal granule cells Hippocampus granule cells C57BL/6 adult (8-10 weeks old) mice of both sexes were "entrained" for at least 2 weeks in light tight compartments with 12 hour dark/light cycles. For slice preparation, the mice were removed from their cages 15 minutes before the Light-to-Dark transition (scheduled at 10 AM). The mice were first deeply anesthetized with isofluorane and then perfused transcardially with cold dissection buffer (5 ml at 10 ml/min) containing 92 mM N-methyl-D-glucamine (NMDG), 2.5 mM KCl, 1.25 mM NaH 2 PO 4 , 30 mM NaHCO 3 , 20 mM HEPES, 25 mM glucose, 2 mM thiourea, 5 mM Na-ascorbate, 3 mM Na-pyruvate, 12 mM N-acetyl cysteine, 0.5 mM CaCl 2 and 10 mM MgSO 4 pH adjusted to 7.4. After decapitation brains were removed quickly, and acute hippocampal slices (300 µm) were made as described (Boric, Munoz, Gallagher, & Kirkwood, 2008) in ice-cold dissection buffer bubbled with a mixture of 5% CO 2 and 95% O 2 . The slices were allowed to recover for 15 min at 30°C in dissection buffer and then for one hour at room temperature in artificial cerebrospinal fluid (ACSF): 124 mM NaCl, 5 mM KCl, 1.25 mM NaH 2 PO 4 , 26 mM NaHCO 3 , 10 mM dextrose, 1.5 mM MgCl 2 , and 2.5 mM CaCl 2 bubbled with a mixture of 5% CO 2 and 95% O 2 .

Ionic currents and conductances
We measured the following currents: delayed rectifier K + (I Kd ), transient A-type K + current (I A ), inward rectifier K + current (I Kir ), a fast early inactivating transient inward current (which we label I in , see below), hyperpolarization-activated inward current (I h ), and the linear leak current (I leak ) (Fig. 1). GCs were clamped both at a holding voltage (V h ) of either -40 and -90 mV and voltage steps of 500-600 msec duration were applied in 10 mV increments at 0.33 Hz to measure I Kd , I in , and I leak . For I Kir we applied 800 msec pulses from V h =-40 mV in -10 mV increments. We measured I Kd (Fig. 1c, d) after leak subtraction as the current at the end of a step to +40 mV from V h = -40 mV in hippocampal GCs (in BF ChAT + cells we measured I Kd at +10 mV). In each case K + current conductances were calculated by dividing the current by the driving force using the calculated E K (-84 mV in hippocampal GCs, -99 mV in ChAT + BF cells). In BF ChAT + cells I A was measured by subtracting the currents obtained from V h = -40 mV from those measured from V h = -90 mV, and g A was calculated from currents measured at +10 mV. I Kir (GCs only, see Fig. 1a, b) and I h (BF ChAT + cells only) was measured after leak subtraction at the end of a voltage step to -120 mV, and g Kir was calculated by dividing the current by the driving force with E Kir measured from the same cell's current-voltage (I-V) curve (-70.5 ± 2.1 mV, Fig. 1b). g h was calculated assuming a reversal potential of -10mV. I in is the early transient inward current we observed in GCs (Fig. 1c) that peaks at around -20 mV (Fig. 1d) from either a V h = -40 or -90 mV. We did not observe a noticeable difference in amplitude as a function of V h . We refer to this current as I in because we did not attempt to further characterize the current for this project. We believe that it is likely to be dominated by a high threshold Ca ++ current, perhaps with a Na + current contribution. Finally, g leak was calculated as the slope of the I-V curve between -70 and -40 mV (Fig. 1a, d), which is predominantly a linear component.

Data analysis
Statistical analysis was performed using SigmaStat (Systat Software, Inc., San Jose, CA). Averages are represented as means ± SD and compared with t-Student tests for independent samples. Pearson product-moment correlation coefficients were calculated to reveal correlations between different variables. The Kolmogorov-Smirnov test was used to determine the normality of distributions.

Hippocampal granule cells
We recorded from 30 hippocampal GCs from the upper blade of the dentate gyrus (DG) from male (2) and female (3) mice at either the end of the "day" or the end of the "night" of 12h light-dark cycle entrained animals. Synaptic inputs were all blocked with APV, CNQX and bicuculline (Methods). We did not detect any significant differences between females and males and the data are thus pooled. To maximize the number of cells recorded we focused on four distinct ionic currents that can be studied without the need to add pharmacological agents (Fig. 1): the delayed rectifier (I Kd ), an inward rectifier K + current (I Kir ), a fast transient inward current (I in ), and a leak current (I leak ).

Figure 2.
Ionic conductance correlations in hippocampal GCs of 4 month-old mice. g Kd , g in , g Kir and g leak are plotted against each other for data obtained at the end of the day (red) and end of night (blue)12h light-dark cycle). Only those pairs that showed significant correlations are shown. Pearson-moment correlation coefficients and their statistical significance, as well as regression lines, are shown in each panel. Means and SD bars for each conductance are shown to the right and top of the plots in a and b. ** P<0.001 (t-test).
As shown in figure 2 the variability of the conductance values of all currents measured both during the day and night was very large, with conductance ranges (ratio of max/min) as low as 2.2 for g leak (night) and as high as 15.9 for g in (day): 2.3 for g leak (day), 14.2 for g Kd (day), 4.2 for g Kir (day), 3.5 for g Kd (night), 2.8 for g Kir (night), and 13.1 for g in (night). Notably, all 4 conductances show statistically significant correlations (Pearson product-moment correlation, at P<0.05) in some subset of pairwise combinations (Fig. 2). The pairs g Kd -g leak and g Kd -g in did not show a significant correlation either during the day or night cycle, and are thus not plotted (Personmoment correlation coefficients and P values are shown in Figure 2). On the other hand, the pairs g leak -g in , g leak -g Kir , g in -g Kir , and g Kd -g Kir showed highly significant correlations even after adjusting for multiple comparisons (false discovery rate method (Curran-Everett, 2000)). There was one notable exception: the pair g Kd -g Kir showed no correlation during the night (ρ=-0.008, P=0.981) while the correlation during the day was strong and highly significant (ρ=-0.649, P=0.012). We compared these two slopes (slope end of night = -0.035, n=14, slope end of day = 2.555, n=12) using a Fisher z statistic (z(12,14) = 1.739) and found it to be significant at P=0.082. Although the normal cut off to consider a difference statistically significant is P=0.05, given that this value is essentially arbitrary, we take this result as a strong indication that there is a likely circadian regulation in the correlation of this pair, and only this pair, of conductances from among those we measured in this study. Importantly, the mean values of the two K + conductances change between day and night (Fig. 2b side bars), while the mean values of g leak and g in do not significantly differ from each other (Fig. 2a, side bars).
Altogether the results indicate that ionic conductances in DG cells do vary in a correlated manner, and suggest a circadian-like control of some of these correlated pairs.

Basal forebrain ChAT + neurons
We recorded from a total of 17 BF ChAT + neurons. We analyzed three ionic currents, I A , I Kd and I h . I A and I Kd were expressed in all 17 cells (see example in Fig. 3a), but I h was measurable only in 10 of those cells. The range of conductances (ratio of max/min) was broad, similar to hippocampal GCs: 27.0 for g A , 14.0 for g Kd and 3.9 for g h . However, we detected a statistically significant correlation only for the g A -g Kd pair (Pearson product-moment correlation ρ=0.920, P=1.7x10 -7 ) (Fig. 3b). For the remaining two pairs the correlations were characterized by: g A -g h pair (ρ=0.393, P=0.261); g Kd -g h pair (ρ=0.580, P=0.079).
Although the cell types sampled in this study are a minimal set of possible neuronal types, our results are entirely in line with the phenomenon thus far only characterized in invertebrate neurons, namely, that ionic current or conductances appear to be co-regulated in a very broad range of neurons also in vertebrates, specifically in mammals. Furthermore, we show that although not all currents appear to be correlated in any given cell type, a substantial fraction of them are. We predict that this will be shown in the future in an even wider range of neuronal subtypes in many different species. We think that these correlations reflect the existence of common regulatory pathways (see (O'Leary et al., 2013)) that are important in establishing cell type-specific setpoints in conductance space. These setpoints are not immutable, but can shift as neurons respond to persistent stimuli or factors, thus enabling these neurons to behave in cell-type characteristic ways while allowing the individual currents to vary in amplitude (cf. (Liu, Golowasch, Marder, & Abbott, 1998;O'Leary & Marder, 2016)). Like in invertebrates, multiple ionic current types appear to be correlated in mammalian neurons, which may be a cell type-specific characteristic (Schulz et al., 2007). Furthermore, we have shown evidence of what appears to be a regulatory mechanism of the correlations themselves in granule cells, which appears to be tied to a circadian mechanism. This would indicate that these correlations serve an important functional role that shifts between day and night. The cellular mechanism underlying this Ionic currents and conductance correlations in adult mouse basal forebrain CHAT + neurons. a. Raw leak-subtracted I Kd (gray trace) and I A (black trace). Voltage steps used to elicit the currents are shown below the currents. b. g Kd and g A are plotted against each other. Pearson productmoment correlation coefficient, statistical significance, and regression line are shown. regulation is not known. However, consistent with this possibility, a regulation mechanism of ionic correlations has already been shown in crustacean neurons, in which the neuromodulatory environment determines the existence of correlations (Khorkova & Golowasch, 2007).
We conclude that ionic current amplitude correlations is a ubiquitous property of neurons across vertebrate as well as invertebrate species, and their functional significance remains to be determined in each case.