Male Wistar rats (26–36 days old) were anesthetized using isoflurane, decapitated and coronal slices (350 μm thick) were prepared from the whisker-related barrel field of the somatosensory cortex (BC). All experimental procedures were carried out according to the animal welfare guidelines of the Max-Planck Society. Brain slices were incubated for 20–30 min at 36°C and then stored at room temperature until used. The extracellular solution for cutting and incubation contained (in mM): 125 NaCl, 25 NaHCO₃, 2.5 KCl, 1.25 NaH₂PO₄, 6 MgCl₂, 1 CaCl₂, 3 myo-inositol, 2 Na-pyruvate, 0.4 ascorbic acid, 25 glucose. For the experiments, brain slices were transferred to the recording chamber and superfused with an extracellular recording solution that contained 125 mM NaCl, 25 mM NaHCO₃, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 1 mM MgCl₂, 2 mM CaCl₂, 25 mM glucose, and was saturated with 95% O₂/5% CO₂ (pH 7.4). Recordings were made at near-physiological temperature of 32–36°C.

L5A and L5B of the BC were visualized at low magnification (5×) under bright-field illumination, and the pyramidal neurons were visualized with infrared gradient contrast microscopy using a Leica upright microscope fitted with a 40×/0.8 numerical aperture water-immersion objective. In addition to their layer location, L5A-st and L5B-tt pyramidal neurons were selected based on the size of their cell body and apical dendritic trunk (L5A-st: soma diameter 17 ± 3 μm, apical dendrite diameter 3 ± 0.7 μm, n = 6; L5B-tt: soma diameter 21 ± 2.8 μm, apical dendrite diameter 5 ± 0.9, n = 8; de Kock et al., 2007). Recording pipettes (4–6 MΩ) were pulled from borosilicate glass and filled with (in mM): 135 K-gluconate, 10 HEPES, 10 Phosphocreatine-Na, 4 KCl, 4 ATP-Mg, 0.3 Na-GTP, pH 7.2 (adjusted with KOH). Signals were recorded using an Axon Instrument amplifier (Axoclamp-2B), low-pass filtered at 3 kHz and sampled at 10–50 kHz.

Physiological stimulation protocols were designed according to previously recorded trains of APs that were evoked by principle whisker stimulation in L5A-st (84 AP patterns, including 29 patterns where no spikes were elicited, n = 6) and L5B-tt (156 AP patterns, including 18 patterns with no APs, n = 6) pyramidal neurons in the BC of anesthetized rats (P25–30; de Kock et al., 2007). These 500 ms-long traces are comprised of spontaneous AP output and evoked AP firing (200 ms-long whisker stimulation between 145 ms and 345 ms, for details see de Kock et al. (2007). In this study, we randomly selected 12 different in vivo AP patterns for each recorded L5A-st (from a total of 84 patterns) and each L5B-tt (from a total of 156 patterns) cell (Figure 2). To better simulate the in vivo situation we included in the random selection also those trains that did not elicit an AP response. The data shown in Figures 2E–H, is thus the average calcium response that could naturally occur during a 500 ms period where spiking could range from zero to several spikes (in our data set: L5-st max 5 spikes; L5B max 11 spikes). The time point of each AP during these 500 ms sequences was taken and a replica trace consisting of 3 ms-long current pulses at the respective time points was used to stimulate the neurons. These “physiological” firing patterns were compared to “artificial” trains of three APs, where the interval between the first and third AP was fixed (50 ms) and the position of the second AP was shifted to occur at 5 ms, 25 ms or 45 ms after the first AP. The second AP was thus shifted to evoke either an AP burst early in the spike train (2nd AP at 5 ms) or at the end (2nd AP at 45 ms). This can be compared with the “physiological” AP pattern where the median interspike-interval (ISI), measured for the first three APs in the train, was for L5A-st 16.1 ms and for L5B-tt 12.9 ms. Furthermore, the L5B-tt patterns had 9% high-frequency-ISIs (defined as ISI ≤5 ms), whereas L5A-st patterns had significantly fewer (1%; Fisher’s exact test, P = 0.0091).

The method used for high-resolution, multi-photon imaging of Ca²⁺ fluorescence was as described previously (Krieger, 2009; Grewe et al., 2010). In brief, one system was a Leica TCS-SP2RS scanning unit attached to an upright Leica microscope (DMLFS) and equipped with a 40× objective (HCX APO, 0.8 NA, Leica). Two-photon excitation was achieved with a Ti:Sa-laser (MIRA 900F; Coherent, Santa Clara, CA, USA) at 840 nm pumped by a solid-state laser (Verdi 5W; Coherent). A dichroic mirror (560DCXR) split the fluorescence signal into one detector (bandpass filter, HQ525/50M) for green fluorescence (Oregon Green 488 BAPTA-I, calcium signals), and another detector (bandpass filter, HQ610/75M) for red fluorescence (Alexa 594, fluorescent marker for dendritic morphology). The second system was a galvanometer scanning unit (TCSNT; Leica Microsystems, Mannheim, Germany) that was mounted on an upright microscope (Leica DMLF) equipped with a 40× objective (HCX APO W40×/0.8 NA), utilizing a Ti:Sa-Laser (Mira 900F; Coherent, Santa Clara, CA, USA) at 870 nm pumped by a solid-state laser (Verdi 8W; Coherent), and using similar filters for the detection of the green/red fluorescence signals. Calcium signals were recorded in line-scan mode with a temporal resolution of 2–2.2 ms/line, with a total scan time per trial in the range of 256 ms to 2.3 s. Recordings of membrane voltage and fluorescence were analyzed offline using pClamp 9 (Molecular Devices, Sunnyvale, CA, USA). Neurons were filled with a combination of the calcium-sensitive dye Oregon Green 488 BAPTA-I (200 μM) and the calcium-insensitive dye Alexa 594 (20–50 μM; both dyes from Invitrogen, Carlsbad, CA, USA) added to the intracellular recording solution to visualize the basal dendrites and to measure Ca²⁺ signals. To ensure a homogeneous dye distribution within the basal dendrites under investigation, dyes were allowed to fill the neuron for at least 20 min before Ca²⁺ signals were measured. Ca²⁺ transients are reported as relative changes in OGB-1 fluorescence and were calculated as dF/F (t) = (F(t) − F0)/(F0 − FB). Background fluorescence (FB) was measured in a neighboring indicator-free region. Resting fluorescence (F0) was calculated as the average value of the 50–100 ms period that preceded the stimulation. Calcium transients associated with physiological AP patterns (Figures 2A,C) were quantified by calculating the calcium-time integral during a 600 ms period starting from the 1st AP in the train (Figure 2E). The area rather than the peak amplitude was calculated since the goal was to measure the total calcium response at a given dendritic spot during the whole stimulation epoch. For simulated whisker-evoked calcium responses (Figure 4) the area was measured 145 ms before and 145 ms after the time point when, during the in vivo recordings (de Kock et al., 2007), the whisker was deflected. The peak amplitude of the calcium response evoked by the artificial 3AP patterns was quantified by a single exponential fit to the decay of the fluorescence transients, calculated as the averages of 3–5 trials. In the normalized plots (Figures 2H, 4B) the null hypothesis (tested using nonlinear regression analysis to fit a straight line) was that the calcium response did not change with distance, thus that the slope was zero. The comparison of the effect of the three different 3AP patterns (Figures 3C,D) was performed with repeated measures ANOVA, thus at each dendritic spot the three 3AP patterns were compared to each other. Data are presented as mean ± SD. GraphPad Prism 6 (GraphPad Software, San Diego, CA, USA) was used for statistical analysis. The diameter of the basal dendrites was obtained from the two-photon microscopy images.

Previously, we demonstrated that in L5A-st pyramidal neurons the temporal structure of physiological AP patterns is an important determinant for their backpropagation efficacy along apical but not basal dendrites (Grewe et al., 2010). We now addressed this issue comparing with the basal dendrites of L5B-tt neurons. In vivo recordings from L5A-st and L5B-tt pyramidal neurons of anesthetized rats have demonstrated significantly different spontaneous and evoked AP firing patterns in these neuron types (Manns et al., 2004; de Kock et al., 2007; de Kock and Sakmann, 2008, 2009). The randomly selected physiological AP firing patterns from L5A-st pyramidal neurons used in this study contained 1–5 APs (median: 1AP/500 ms; (Figures 2A,B), while the sequences recorded from L5B-tt pyramidal neurons consisted of 1–11 APs (median: 3APs/500 ms; Figures 2C,D). The bAP-evoked calcium responses were measured in the proximal region of the basal dendrite (5–10 μm from soma) and at different more distal locations along the same dendrite (40–278 μm). Even though on average more APs were evoked in L5B-tt neurons compared to L5A-st neurons, the average bAP-evoked calcium response evoked by a series of physiological AP patterns appeared to be similar in L5A-st and L5B-tt cells (Figure 2G) with regard to the absolute magnitude (proximal location: L5A 113 ± 49 (unit: time × dF/F), n = 7 dendrite spots (branches), 7 cells; L5B 131 ± 50, 13 dendrite branches, 10 cells, P = 0.4519, unpaired t-test; distal locations: L5A 106 ± 61, n = 10 dendritic spots, 5 cells; L5B 98 ± 61, 12 dendritic spots, 10 cells, P = 0.7646, unpaired t-test). Although averaging over all cells within the population can show general trends, the large variability between cells and dendrite branches from the same cells argues that a normalization for each branch (as in Figure 2H) is more representative of how the bAP evoked calcium response changes along the basal dendrite. Normalizing to each measured dendritic branch (Figure 2H), the calcium response measured at the proximal site for that branch, suggests that in L5B-tt (linear regression P = 0.0011), but not in L5A-st neurons (linear regression, P = 0.2332), there is a small decrease with distance along the dendrite (dendritic sites 0–125 μm). This difference in the calcium response along the dendrite (0–125 μm) between L5A-st and L5B-tt neurons is corroborated by a two-way ANOVA analysis of the average data (Figure 2G), which for the Interaction effect has a P value of 0.05.

Next, we asked whether the presence of a high frequency component would alter the efficacy with which the bAPs can evoke a calcium response. For this purpose, trains of three APs were used, where the interval between the first and third AP was fixed and the position of the second AP varied. One AP train had equally spaced APs (occurring at 0, 25 and 50 ms) at sub-critical frequency (40 Hz, mean-frequency pattern, F_mean; Kampa and Stuart, 2006), while the other two trains contained a supra-critical frequency (200 Hz) component positioned either at the beginning (0–5–50 ms; High-Low-frequency pattern, F_H–L) or the end (0–45–50 ms; Low-High-frequency pattern, F_L–H) of the train. Representative examples of the calcium transients evoked by these AP patterns at a single position in the basal dendrites of L5A-st and L5B-tt neurons are shown in Figures 3A,B, respectively. In L5A-st basal dendrites (range: 40–254 μm; median: 96 μm), back-propagation resulted in a similar calcium response for all three AP patterns (F_mean: 0.96 ± 0.58% dF/F; F_H–L: 1.04 ± 0.55% dF/F; F_L–H: 1.03 ± 0.67% dF/F; 8 cells, 15 dendritic positions P = 0.3253, repeated measure ANOVA; Figure 3C). In contrast, in L5B-tt pyramidal neurons (range: 57–168 μm; median: 101 μm) the amplitude of the calcium signal was significantly different for the three AP patterns (Figure 3D). The presence of a high-frequency component (F_H–L and F_L–H pattern) produced ~20% larger transients compared to the F_mean pattern (F_mean: 0.80 ± 0.41% dF/F; F_H–L: 0.95 ± 0.38% dF/F; F_L–H: 1.00 ± 0.56% dF/F; P = 0.0130 with repeated ANOVA and Holm-Sidak’s multiple comparisons test; F_mean vs. F_H–L, P < 0.05; F_mean vs. F_L–H P < 0.005, n = 7 cells, 11 dendritic positions). Taken together, these data suggest significant differences in the intrinsic properties of the basal dendrites of these two neuron populations, affecting the processing and propagation of information.

The physiological firing patterns derived from L5A-st neurons typically involve fewer APs than those derived from L5B-tt neurons (Figures 2A,B), thus the relatively larger calcium response in L5A-st suggest that L5A-st basal dendrites respond with a larger calcium transient per AP. To examine this point in greater detail, we used identical (“artificial”) 3AP patterns for both cell types to evoke calcium transients measured in a proximal location (5–10 μm from soma periphery). In agreement with the results using the physiological firing patterns, we found that when stimulated with identical 3AP patterns, the average calcium response (the average of all three different AP patterns), was significantly larger in L5A-st basal dendrites (1.29 ± 0.48% dF/F, unpaired t-test, P = 0.0083, n = 10 cells, data not shown) compared to L5B-tt basal dendrites (0.71 ± 0.13% dF/F, n = 7 cells). One possible explanation for these findings is related to the difference in diameter of basal dendrites at this proximal location for both neuron types. We found that the diameter of L5A-st proximal basal dendrites (≤10 μm from soma) was significantly smaller compared to L5B-tt basal dendrites (L5A-st: 1.71 ± 0.41 μm, n = 12 dendrites; L5B-tt: 2.32 ± 0.59 μm, n = 10 dendrites; unpaired t-test, P = 0.0104). A smaller diameter and therefore an increased surface-to-volume ratio may explain the larger calcium response in L5A-st compared to L5B-tt neurons (Holthoff et al., 2002; Anwar et al., 2014). At more distal dendritic locations there was no difference in diameter (L5A-st: 0.78 ± 0.15 μm, n = 18, median distance from soma 95 μm, range: 40–170 μm; L5B-tt: 0.80 ± 0.22 μm, n = 21, median distance from soma 100 μm, range: 20–168 μm; unpaired t-test, P = 0.7169).

To further investigate how calcium responses are evoked by natural stimulations, we specifically looked at the calcium responses elicited by the APs evoked by whisker stimulation. In the 500 ms long recordings used as a template for the firing patterns, whisker stimulation occurred at 145 ms (see Figures 2A,C). Under physiological conditions spontaneous spikes also back-propagate and are thus capable of eliciting calcium responses. Taking this into account, we calculated the relative increase in calcium before and after the simulated whisker-evoked AP pattern. The relative increase should better reflect the changes in calcium specifically induced by whisker stimulation. To evaluate the calcium response elicited by APs evoked by a “whisker-like” stimulation, we thus compared the calcium response elicited by the AP pattern during a time window of 145 ms before (spontaneous activity; “Ca-pre-whisk”) and after the simulated “whisker” stimulus (spontaneous plus evoked activity; “Ca-post-whisk”). Calcium responses were measured as previously described (see Figure 2 and “Materials and Methods” Section), with the difference that now only 145 ms was used to calculate the calcium response (the calcium-time integral). In the physiological AP pattern used in this part of the study the average number of APs during the “pre-whisk” stimulation period was for L5A-st 0.29 APs/145 ms and during “post-whisk” 0.64 APs/145 ms, and for L5B-tt 0.53 and 1.11 APs/145 ms, respectively. Calcium responses were normalized with respect to a “calcium index” calculated as 1 − (Ca-pre-whisk/Ca-post-whisk). The closer the calcium index is to 1 the larger the relative increase in calcium evoked by the simulated whisker-evoked AP pattern (Figure 4A). Measuring the relative change in calcium is important because, presumably, not only the absolute calcium amplitude but also the “whisker-evoked” change is crucial for dendritic computation. Normalizing the calcium index to the value at the dendrite spot close to the soma (5–10 μm), for each basal dendrite separately (L5A-st 7 cells, 7 basal dendrites; L5B-tt: 8 cells, 11 basal dendrites), indicates (Figure 4B) that with distance along the basal dendrite the calcium index decreases for L5B-tt cells (linear regression P = 0.0204) but not for L5A-st cells (linear regression P = 0.7804). A decreased index would thus suggest that the relative increase in calcium influx as a result of whisker stimulation is smaller in L5B-tt than in L5A-st. Presumably this is caused by the higher spontaneous discharge, (see Figure 2C) in the physiological patterns used to mimic L5B-tt cell activity, such that any additional APs evoked by the whisker stimulation has a relatively smaller effect. In comparison the L5A-st cells had a lower spontaneous firing rate in the physiological patterns used, even sometimes having no spikes preceding the whisker stimulus at 145 ms (see Figure 2A). In summary our data show that because whisker evoked spike trains differ between L5A-st and L5B-tt neurons the physiological effect on calcium dynamics is such that the basal dendrites are optimized to respond to different features of the spike trains. While L5A-st basal dendrites are more responsive to an increase in firing, the L5B-tt basal dendrites are in addition also responsive to the presence of high frequency spikes.

Previous studies have shown that bAP-evoked calcium responses differ in apical dendrites in L5A-st and L5B-tt pyramidal cells. For example, the calcium response to high-frequency AP patterns was more strongly attenuated in L5A than in L5B apical dendrites (Schiller et al., 1995; Grewe et al., 2010). Some aspects of the efficacy of bAP propagation has previously been reported for L5A pyramidal neurons (Grewe et al., 2010). In reference to that study the following interesting comparison can be made. It was shown that in the apical dendrite there is a developmental change between P14 and P25 such that in older animals the response to a 3AP pattern initially increases rather than decreases (see Figure 7B in Grewe et al., 2010). A similar developmental change is evident for the L5A basal dendrites where at P14 there is a marked decrease along the length of the dendrite (Figure 6C in Grewe et al., 2010) which is not the case at the older ages reported in the present study.

Using a variety of artificial 3AP patterns we show that L5A-st and L5B-tt pyramidal cells have different calcium responses depending on the exact temporal structure. These differences are likely to be important for synapse specific differences in synaptic plasticity. Interestingly, however, the AP pattern evoked by simulated whisker stimulation elicited rather similar calcium responses in both cell types even though the temporal structure and number of APs in the trains used differed. In our calculation we analyzed the relative increase in calcium evoked by APs occurring before and after whisker stimulation (thus providing a “calcium index”). The calculation depends not only on the whisker evoked AP patterns but also the spontaneous activity. Considering that L5B-tt neurons fire more APs it is somewhat surprising that the ratio post/pre is not larger in L5B-tt compared to L5A-st dendrites. This could be explained by the fact that L5A-st neurons fire few spontaneous spikes so whenever there is a spike there is a large influx in calcium.

The evidence presented in this in vitro study suggesting variability in the calcium responses evoked by AP patterns can be due to a variety of factors, including morphology and channel density. In vivo there is also likely a contribution of neuromodulatory states and activity-dependent plasticity (Hoffman and Johnston, 1998; Golding et al., 2001; Frick et al., 2004). In summary the data show that: (1) high frequency AP patterns are more efficiently propagated in L5B-tt compared to L5A-st neurons (Figures 3C,D); (2) the calcium response gradient is such that in vivo evoked AP patterns are more strongly attenuated in L5B-tt neurons (Figure 2H); (3) in distal locations the absolute calcium increase is similar in L5A-st and L5B-tt basal dendrites, even though the calcium responses are evoked by very different AP patterns. Considering that L5A-st and L5B-tt pyramidal cells receive input from different part of thalamus, and project to different sub-cortical structures (Groh et al., 2010) it is of interest to investigate how dendritic calcium dynamics is related to functional differences. Whisker evoked responses in ventral posterior medial thalamic nucleus (VPM) are larger (more spikes per stimulus) compared to responses in the posterior medial thalamic nucleus (POm; Diamond et al., 1992). Interestingly VPM axons target L5B and POm targets L5A, thus the higher sensitivity of L5B-tt neurons to high frequency AP firing could be related to the fact that high frequency bAP can be timed to higher frequency inputs and thus facilitate dendritic integration. The less active POm cells can, however, still activate a rather large calcium response in L5A-st cells because the L5A-st cells are “tuned” to integrating lower frequency inputs. Future experiments using physiological VPM/POm spike patterns to create dendritic stimulations in combination with physiological bAP patterns would be used to further investigate this question.