Experiments were performed in accordance with the Declaration of Helsinki and approved by the University of Szeged Ethical Committee. Before surgery, patients provided written consent for all tissue material. We used human cortical tissue that had to be surgically removed from patients (n = 14 female, n = 20 male) to access deep-brain target malformations (tumors, hydrocephalus, cysts, and aneurysms) for surgical treatment (Supplementary Table 1). Anesthesia was induced with intravenous midazolam (0.03 mg/kg) and fentanyl (1–2 μg/kg), followed by a bolus dose of propofol (1–2 mg/kg) administered intravenously. Patients received 0.5 mg/kg rocuronium to facilitate endotracheal intubation. The trachea was intubated to ventilate the patient with a mixture of O₂ and N₂O at a ratio of 1:2. Anesthesia was maintained with sevoflurane at a care volume of 1.2–1.5. During the surgical procedure tissue blocks were removed from various brain regions (parietal: n = 5, temporal: n = 14, frontal: n = 13, occipital: n = 2), and the resected tissue blocks were immediately immersed in ice-cold solution. Slices were cut perpendicular to the pia mater at a thickness of 320 μm with a vibrating blade microtome (Microm HM 650 V) in ice-cold solution containing (in mM): 75 sucrose, 84 NaCl, 2.5 KCl, 1 NaH₂PO₄, 25 NaHCO₃, 0.5 CaCl₂, 4 MgSO₄, and 25 D(+)-glucose, saturated with 95% O₂ and 5% CO₂. The slices were then incubated in the same solution for 30 min at 36 °C. After incubation the solution gradually changed to (in mM): 130 NaCl, 3.5 KCl, 1 NaH₂PO₄, 24 NaHCO₃, 1 CaCl₂, 3 MgSO₄ and 10 D(+)-glucose, saturated with 95% O₂ and 5% CO₂, and the slices were kept in this solution until use.

Somatic whole-cell current-clamp recordings were obtained at ~36 °C in artificial cerebrospinal fluid containing (in mM): 130 NaCl, 3.5 KCl, 1 NaH₂PO₄, 24 NaHCO₃, 3 CaCl₂, 1.5 MgSO₄ and 10 D(+)-glucose, from layer 2/3 pyramidal cells visualized by infrared differential interference contrast (DIC) video microscopy. Micropipettes (3–5 MΩ) were filled with intracellular solution containing (in mM): 126 potassium-gluconate, 4 KCl, 4 ATP-Mg, 0.3 GTP-Na₂, 10 HEPES, 10 phosphocreatine, 8 biocytin, 0.015 Alexa hydrazide 594 and 0.1 Oregon Green BAPTA-1 (pH 7.20; 300 mOsm). To investigate Ca²⁺ transients induced by the backpropagation of action potentials to the apical dendrite, one and three action potentials were elicited by brief (5 ms) current pulses injected into the cell body of the neurons.

We combined whole-cell recording and Ca²⁺ imaging for the optical recording of dendritic Ca²⁺ signals. The intracellular solution used for the experiments to characterize the Ca²⁺ signals elicited by the backpropagating action potential contained Oregon Green BAPTA-1 (OGB-1, 100 μM) high-affinity Ca²⁺ sensor. After the whole-cell configuration was acquired, the intracellular solution was allowed to diffuse into the cell protrusions for 15–20 min before the two-photon imaging was initiated. Fluorophores were excited by femtosecond Ti:sapphire laser pulses (Mai Tai Deep See; Spectra-physics, United States) at 800 nm wavelength to visualize the dendritic arborization of the pyramidal cells. Intensity changes of the emitted green fluorescence were collected with line scans across the dendrite.

Ca²⁺ signals were analyzed using custom MATLAB (The Math Works, Inc.) scripts. The amplitude of Ca²⁺ transients was calculated as the average of five consecutively recorded Ca²⁺ signals evoked at 2 s intervals and measured at each dendritic region. The Ca²⁺ transients were calculated as the relative change in fluorescence: ΔF/F₀ = (F − F₀)/F₀, where F₀ is the baseline prestimulus fluorescence, background fluorescence was subtracted from both F and F₀. Normalized ΔF/F₀ was calculated by normalizing each Ca²⁺ transient amplitude to the maximum Ca²⁺ signal amplitude measured on the same cell. After the recordings Z-stack images of the cells were acquired, and the distance of the studied dendritic sections from the cell body was measured by tracing the protrusions on the images (ImageJ, Simple Neurite Tracer plugin).

All ion channel blockers were administered extracellularly in the recording solution in our pharmacological experiments. Our experiments included the following pharmacones: SNX-482 (300 nM), NNC 55-0396 dihydrochloride (100 μM), nifedipine (20 μM), ω-conotoxin (1 μM), CdCl₂ (200 μM), and tetrodotoxin (1 μM).

Following the two-photon Ca²⁺ imaging experiments, brain slices were prepared for further morphological analysis of the biocytin-labeled cells. Slices were fixed in a fixative solution of 4% paraformaldehyde, 15% picric acid, and 1.25% glutaraldehyde in 0.1 M phosphate (PB, pH = 7.4) for at least 12 h. After multiple washes in 0.1 M PB, slices were immersed in 10% then 20% sucrose solution in 0.1 M PB for cryoprotection. The slices were frozen in liquid nitrogen and then thawed in PB. Slices were embedded in 10% gelatin and sectioned into 70 μm thick sections. Sections were incubated in a solution of conjugated avidin-biotin horseradish peroxidase (ABC; 1:100; Vector Labs) in tris-buffered saline (TBS, pH = 7.4) overnight at 4 °C. The enzyme reaction became visible by using 0.05% 3′3-diaminobenzidine tetrahydrochloride as a chromogen and 0.01% H₂O₂ as an oxidant. Sections were post-fixed with 1% OsO₄ in 0.1 M PB. Following several washes with distilled water, sections were stained in 1% uranyl acetate and dehydrated in an ascending series of ethanol. The sections were infiltrated with epoxy resin (Durcupan, Sigma-Aldrich) overnight and then embedded on glass slides.

Following the visualization of the recorded cells using DAB staining, 3D light microscopic reconstructions were performed using the Neurolucida system (MBF Bioscience, Williston, VT, United States) with a 100× objective. Analysis of morphological features was made by NeuroExplorer software (MBF Bioscience, Williston, VT, United States) and Origin 9 (OriginLab, Northampton, MA). The dendrite diameter was measured at the Ca²⁺ imaging site using 3D reconstructions of the examined cells. Diameters were obtained from intensity profiles using a custom-written script in Igor (Wavemetrics, Lake Oswego, United States) by measuring at 70% of the amplitude (Ozsvár et al., 2018) and were normalized to the diameter at the somatic-dendritic junction. This junction was identified by fitting an ellipsoidal boundary to the soma and selecting the midpoint of the apical stem’s origin. The diameters of the imaged dendritic locations were then averaged within 25 μm dendritic segments starting from the beginning of the apical dendrite. Dendritic spine density was calculated as spines/μm on 25 μm dendrite segments starting from the beginning of the apical dendrite.

Data presented as the mean ± S.D. Normality was tested with the Shapiro–Wilk test, for statistical analysis, ANOVA with post-hoc Bonferroni test, or Kurskal–Wallis with post-hoc Dunn test. For pairwise comparison two-sample t-test or Mann–Whitney test was used. Wilcoxon signed ranks or Friedman tests were used for repeated measures. Differences were accepted as significant if p < 0.05. The data are shown on boxplots, boxes indicate the 25th, 50th (median), and 75th percentiles, rectangle represents the mean value, and whiskers indicate S.D.

Somatically induced action potentials propagate through dendritic branches and activate voltage-gated Ca²⁺ channels throughout the dendrite, causing the influx of the Ca²⁺ into the dendrite. To investigate somatic AP-induced Ca²⁺ signals in the dendrites of human cortical pyramidal cells, we used acutely prepared slices of neurosurgically resected human neocortical tissue. The samples were predominantly obtained from the frontal and temporal lobes of patients who underwent tumor or hydrocephalus surgery. Data were collected from 34 patients aged from 2 to 82 years from cortical layer 2/3 (L2/3) pyramidal cells (n = 43, average somatic distance from the pia mater: 396.32 ± 88.57 μm). We examined apical and radial oblique dendrites of pyramidal cells using simultaneous somatic whole-cell patch-clamp stimulation and two-photon line scan imaging. Neurons were loaded with the high-affinity Ca²⁺ sensor Oregon Green BAPTA-1 (OGB-1100 μM) via the patch-clamp micropipette (Figure 1A). To characterize the spatial profile of Ca²⁺ transients, we measured the amplitude of Ca²⁺ signals from different locations along the apical dendrite (up to 263 μm from the soma) while evoking single AP or short bursts (three APs) through brief (5 ms) somatic current injections (Figure 1A). The amplitude of Ca²⁺ transients increased with distance from the soma, peaking between 50–100 μm, and subsequently declined in more distal regions. This spatial distribution pattern was consistent regardless of the number of APs elicited (Figure 1B). Ca²⁺ transients elicited by AP bursts were significantly larger at all dendritic locations compared to those evoked by a single backpropagating AP (Figure 1D). To assess whether differences in Ca²⁺ kinetics across dendritic locations might alter the spatial distribution of total Ca²⁺ entry, we also computed the area under the ΔF/F₀ curve (AUC) for bAP-evoked transients. The AUC exhibited a similar biphasic spatial profile to the peak amplitude, with a maximum in the 50–100 μm segment and reduced values proximally and distally (Figure 1C). Peak amplitude and AUC were highly correlated across all segments (Figure 1E; 1 AP: R² = 0.53, p = 1.7 × 10⁻⁴, 3 APs: R² = 0.82, p = 3.1 × 10⁻⁸), indicating that local variations in kinetics do not qualitatively change the spatial pattern of bAP-evoked Ca²⁺ influx.

We segmented the imaged dendrites into 50 μm-long sections starting from the soma and computed the average spatial calcium transient amplitudes of n = 36 human cortical supragranular pyramidal cells (Figure 2A). We observed a progressive increase in amplitude within the first and second segments followed by a gradual decline in the distal segments after reaching a maximum value (n = 756, p = 2.09 × 10⁻²⁷, Kruskal–Wallis test) (Figure 2B). To assess differences between dendritic compartments, we compared the magnitudes of Ca²⁺ transients in primary apical dendrites—defined as the thick dendrites emerging directly from the soma—with those in the thinner radial oblique branches. The amplitude of calcium transients was significantly higher in oblique dendrites compared to primary apical dendrites (primary: 60.36 ± 20.98%; oblique: 72.4 ± 24.58%, n = 547/203, p = 3.97 × 10⁻¹⁰, Mann–Whitney U test; Figure 2C). The spatial profiles of the calcium transient amplitudes were similar in both primary and oblique dendrites. In both cases, the distribution followed a nonlinear pattern, as described above, with significant variation across segments (primary apical dendrites: p = 4.22 × 10⁻², Kruskal–Wallis test; Figure 2D; oblique dendrites: p = 6.89 × 10⁻⁵, Kruskal–Wallis test; Figure 2E). When comparing the amplitude of transients between the two dendritic types across corresponding segments, significant differences were detected in the first segment (primary: 51.52 ± 19.56%; oblique: 66.92 ± 19.19%, n = 242/30, p = 1.1 × 10⁻⁴, Mann–Whitney U test,), and in the second segment (primary: 68.46 ± 19.54%; oblique: 78.51 ± 23.88%, n = 180/76, p = 1.56 × 10⁻³; Figure 2F). No significant differences were observed in the distal dendritic segments.

Kinetics and amplitude of intracellular calcium increase evoked by dendritic action potentials are highly dependent on several physical factors, including the distance from the soma, number of branch points, dendritic diameter, input resistance and number of spines. To investigate the influence of dendritic arborization patterns on the spatial distribution of Ca²⁺ transients induced by back-propagating action potentials, we performed three-dimensional reconstructions of the apical dendrites from the examined cells (Figure 3A; Supplementary Figure 1). Our analysis was limited to regions within 150 μm of the soma, and the dendrites were divided into 25 μm-long sections (Figure 3B). We performed correlation analysis on segments located proximal and distal to the location of maximum signal amplitude (Figure 3C). For spine density assessment, we first identified and counted spines on the apical dendrite of n = 10 pyramidal cells. The distribution of spine density showed a general increase with distance (average spine number in segments 1–6: 0.33 ± 0.65, 2.43 ± 3.59, 9.28 ± 7.64, 10.75 ± 7.04, 9.38 ± 6.27, 14.5 ± 10.29; p = 9.13 × 10⁻⁹, Kruskal–Wallis test; Figure 3D). In the initial half of the measured apical segment a positive correlation was observed between spine number and the average normalized amplitude of Ca²⁺ signals (before peak: R² = 0.09, p = 0.01, after peak: R² = −0.03, p = 0.94; Figures 3E,F). During the three-dimensional reconstructions (n = 10) of the biocytin labeled and DAB-stained neurons, we measured dendritic diameter at imaging locations using light microscopy. The normalized diameter showed a gradual decrease with the greatest thickness in the first and second segments (average dendrite diameter in segments 1–6: 2.32 ± 0.94, 1.28 ± 0.66, 1.27 ± 0.61, 1.67 ± 0.56, 1.18 ± 0.65, 1.11 ± 0.49 μm, p = 1.12 × 10⁻³, Kruskal–Wallis test; Figure 3G) and was negatively correlated with both the normalized amplitude of Ca²⁺ transients (before peak: R² = 0.22, p = 1.9 × 10⁻⁴, after peak: R² = 0.08, p = 0.06, Figure 3I) and the distance from the soma (before peak: R² = −0.02, p = 0.21, after peak: R² = 0.62, p = 2.6 × 10⁻⁴; Figure 3H). Additionally, the number of dendritic nodes quantified from the reconstructions (n = 10) showed no correlation with increasing distance (average number of nodes in segments 1–6: 1.5 ± 1.09, 0.81 ± 0.93, 0.9 ± 0.94, 0.69 ± 0.48, 0.77 ± 0.93, 0.17 ± 0.41, p = 0.09, Kruskal–Wallis test; Figure 3J), and a slight correlation was found between node number and the normalized amplitude of Ca²⁺ transients (before peak: R² = 0.1, p = 0.01, after node: R² = 0.05, p = 0.09; Figures 3K,L). Thus, the amplitude profile of the Ca²⁺ signal induced by bAP appears to correlate to some extent with dendritic geometry, specifically, spine density correlated positively, while diameter correlated negatively with the rise profile of the proximal calcium signal.

To clarify the contribution of different Ca²⁺ channel subtypes to Ca²⁺ signaling in apical dendrites, we performed experiments using VGCC blockers. First, we used the nonspecific calcium channel blocker CdCl₂ (200 μM, n = 4) and observed a significant reduction in the amplitude of Ca²⁺ transients compared to control at the same dendritic sites (control: 55.25 ± 28.18% CdCl₂: 16.69 ± 11.53%, p = 1.83 × 10⁻⁶, Wilcoxon signed ranks test). We then divided the Ca²⁺ signals into three 50 μm-long segments relative to the soma, 0–50 μm (proximal), 50–100 μm (medial) and 100–150 μm (distal) along the apical dendrite (Figure 4A). Nonspecific blockade of VGCCs resulted in a significant decrease in Ca²⁺ signal amplitude in the proximal (control: 54.95 ± 25.72%, CdCl₂: 15.88 ± 9.51%; p = 1.66 × 10⁻³, Wilcoxon signed ranks test) and medial segments (control: 54.56 ± 31.3%, CdCl₂: 19.02 ± 14.43%; p = 1.66 × 10⁻³, Wilcoxon signed ranks test) of the apical dendrite (Figure 4B). To evaluate changes in the different segments, we performed curve fitting on the calcium signal profiles. The distribution of Ca²⁺ transient amplitude was best fit with a second-order polynomial function (Figure 4B). We compared the quadratic (a) and linear (b) coefficients of the fits and found no significant difference in the slope (p = 0.312) and linearity (p = 0.316) coefficients.

Next, we focused on high-voltage-activated L-, N- and R-type calcium channels. The L-type VGCC antagonist nifedipine (20 μM, n = 5), the N-type VGCC antagonist ω-conotoxin (1 μM, n = 2) and the R-type channel antagonist SNX 482 (300 nM, n = 3) significantly decreased the amplitudes of Ca²⁺ signals (nifedipine: n = 26, p = 5.52 × 10⁻⁴; ω-conotoxin: n = 21, p = 1.75 × 10⁻⁴; SNX 482: n = 37, p = 1.49 × 10⁻⁶, Wilcoxon signed ranks test). Nifedipine reduced the amplitude of Ca²⁺ transients in dendritic segments located within 100 μm of the cell body (proximal: control: 44.57 ± 18.28%, nifedipine: 33.72 ± 10.7%, n = 17, p = 5.84 × 10⁻³, paired sample t-test) and in the medial segment (control: 67.62 ± 20.08%, nifedipine: 52.25 ± 15.32%, n = 7, p = 0.022, Wilcoxon signed ranks test; Figure 4C). Similarly, treatment with ω-conotoxin led to a decrease in the amplitude of fluorescence transients in both the proximal (control: 83.89 ± 15.54%, ω-conotoxin: 60.27 ± 11.04%, n = 11, p = 8.05 × 10⁻³, Wilcoxon signed ranks test) and medial (control: 98.56 ± 15.02%, ω-conotoxin: 62.96 ± 11.4%, n = 9, p = 0.01, Wilcoxon signed ranks test) regions of the apical dendrite (Figure 4D). Blocking R-type VGCCs by bath-applied SNX 482 resulted in a significant decrease in Ca²⁺ signals in the proximal (control: 47.22 ± 14.56%, SNX 482: 37.44 ± 14.05%, n = 11, p = 0.01, Wilcoxon signed ranks test) and medial segments (control: 55.14 ± 17.59%, SNX 482: 39.16 ± 17.37%, n = 16, p = 2.8 × 10⁻⁶, Wilcoxon signed ranks test), but not in the distal regions (control: 27.87 ± 12.44%, SNX 482: 24.02 ± 8.03%, n = 10, p = 0.15, Wilcoxon signed ranks test; Figure 4E). Comparison of curve fits of amplitude profiles revealed no significant differences between L- and R-type channel blocking, with no significant differences in slope (p = 0.835 and 0.132 for nifedipine and SNX 482, respectively) and linearity coefficients (p = 0.86 and 0.09 for nifedipine and SNX 482, respectively). However, for ω-conotoxin, the slope and linearity of the fitted curve were significantly different (p = 0.015, p = 0.014).

To test the contribution of the low-voltage-activated calcium channel we applied the T-type Ca²⁺ channel inhibitor NNC 55-0396 (100 μM, n = 4), which resulted in a significant decrease in Ca²⁺ signals (n = 51, p = 5.29 × 10⁻⁸, Wilcoxon signed ranks test). In different dendritic regions we also found in all regions tested (proximal: control: 57.43 ± 17.38%, NNC 55-0396: 44.94 ± 14.46%, n = 17, p = 1.78 × 10⁻³, Wilcoxon signed ranks test; medial: control: 77.74 ± 22.27%, NNC 55-0396: 62.28 ± 13%, n = 13, p = 3.3 × 10⁻³, Wilcoxon signed ranks test; distal: control: 63.13 ± 25.84%, NNC 55-0396: 47.21 ± 19.83%, n = 21, p = 8.79 × 10⁻⁴, paired sample t-test; Figure 4F). We found no significant change in the fitted curve coefficients representing initial slope (p = 0.392) and linearity (p = 0.415) (Figure 4F).

In summary, we found that cadmium significantly reduced but did not completely eliminate the voltage-activated calcium transients. By selectively blocking L-, N-, R-, and T-type calcium channels, we found that each contributed to the intracellular Ca²⁺ signaling of bAP in similar proportions overall. When the apical dendrites were divided into segments, the N-type calcium channel caused the greatest change in calcium signaling in the examined middle part of the dendrite segment (Figure 4G).