Electrophysiology and morphology of human cortical supragranular pyramidal cells in a wide age range

The study investigates how intrinsic electrophysiological and morphological properties of human cortical layer 2/3 (supragranular) pyramidal neurons vary across the human lifespan. While neuronal production and migration are largely complete at birth, substantial postnatal processes (myelination, synaptogenesis, pruning, and network refinement) continue through development and into aging, where declines in cognitive processing occur. Prior postmortem human studies documented dendritic and synaptic development, and animal studies showed pronounced early postnatal changes in neuronal biophysics, but a comprehensive cross-age, in vitro physiological study in human pyramidal neurons was lacking. The authors aim to define a morpho-electrical lifetime profile of human L2/3 pyramidal cells, testing the hypothesis that intrinsic biophysical and some structural features change systematically during early development and aging, with relative stability during adulthood.

Previous work in non-human models shows marked maturation of intrinsic membrane properties early postnatally, including changes in input resistance, membrane time constant, and action potential kinetics (e.g., rodent and macaque studies by Picken Bahrey and Moody 2003; Kroon et al. 2019; Elston and Fujita 2014; McCormick and Prince 1987). Postmortem human studies reported development of dendrites and synapses (Huttenlocher and Dabholkar 1997; Petanjek et al. 2008; Koenderink and Uylings 1995), synaptogenesis followed by prolonged pruning (Huttenlocher 1979; Petanjek et al. 2011), and age-related spine changes (Jacobs et al. 1997; Benavides-Piccione et al. 2013). Human in vitro studies revealed depth-related electrophysiological diversity in L2/3 pyramidal neurons (Kalmbach et al. 2018; Berg et al. 2021; Moradi Chameh et al. 2021) and age-related increases in sag (Guet-McCreight et al. 2023). Molecular studies show developmental and aging-related shifts in ion channel expression, including Na+ channels and KCNK leak channels (Huguenard et al. 1988; Erraji-Benchekroun et al. 2005; Aller and Wisden 2008). However, a cross-sectional, lifespan analysis of human L2/3 pyramidal neuron electrophysiology and morphology had not been performed.

Human neocortical tissue was obtained from neurosurgical resections (frontal, temporal, parietal, occipital lobes) performed for clinical indications (primarily tumors in adults/elderly and hydrocephalus in children) under approved protocols with informed consent. Slices (320 µm) were prepared in ice-cold protective solution and incubated before transfer to recording ACSF (~36°C). Whole-cell current-clamp recordings were made from visually identified L2/3 pyramidal cells using K-gluconate-based internal solution with 8 mM biocytin. Step current protocols (800 ms; steps of 20 pA; typically from −100 pA upward) were applied to extract subthreshold (resting Vm, input resistance, tau, sag ratio) and suprathreshold properties (rheobase, AP half-width, AP upstroke velocity, AP amplitude, F-I slope, first AP latency, adaptation, and additional AP-related features). Inclusion criteria: for general analyses, 457 recordings (Rs 24.93±11.18 MΩ, max 63.77 MΩ); for fast AP parameters, stricter quality (Rs ≤30 MΩ) yielded 331 cells (Rs 19.42±6.2 MΩ). Layer assignment was verified by measuring soma distance to the layer 1/2 border when possible (36% of cells recovered; mean 129.69±130.77 µm from L1). The dataset comprised 498 L2/3 pyramidal cells from 109 patients aged 1 month to 85 years. Analyses were grouped into seven age bins: infant (<1 y), early childhood (1–6 y), late childhood (7–12 y), adolescence (13–19 y), young adulthood (20–39 y), middle adulthood (40–59 y), late adulthood (≥60 y). Morphology: 3D reconstructions of 63 biocytin-filled neurons with intact apical tufts and no truncation (ages 0–73 y; infant n=7, early childhood n=8, late childhood n=11, adolescence n=11, young adulthood n=9, middle adulthood n=9, late adulthood n=8) were analyzed in NeuroExplorer to quantify total, apical and basal dendritic lengths, number of nodes, horizontal and vertical extents, and terminal segment lengths. Spine analysis: complete 3D spine annotation in n=6 cells (infant group: 83-day-old, n=3 cells from one patient; late adulthood group: ~64 years, n=3 cells from three patients). Spine density (spines/µm between bifurcations) and morphology categories (mushroom, thin, filopodium, branched, stubby) were assessed. Statistics: normality tested by Lilliefors; group comparisons via ANOVA with Bonferroni or Kruskal-Wallis with post-hoc Dunn; pairwise via Mann-Whitney or two-sample t-test; p<0.05 considered significant. Depth dependence, medical condition (tumor vs hydrocephalus), and sex effects were examined. UMAP embeddings were computed for multivariate electrophysiological features. Slice preparation, recording solutions, fixation, staining, and reconstruction protocols are described in detail (biocytin filling, DAB visualization, OsO4 postfixation, Neurolucida reconstructions).

• Subthreshold properties differed significantly across age groups, with the infant group most distinct: resting membrane potential (p=3.53×10⁻8), input resistance (p=1.29×10⁻16), membrane time constant tau (p=1.31×10⁻15), and sag ratio (p=5.2×10⁻4), Kruskal-Wallis. Infants had more depolarized resting Vm and markedly higher input resistance and tau; input resistance decreased sharply after infancy (post-hoc p=1.88×10⁻6). Sag ratio increased with age, peaking in late adulthood.
• Suprathreshold properties showed strong age dependence: rheobase (p=8.71×10⁻12), AP half-width (p=9.57×10⁻25), AP upstroke velocity (p=1.63×10⁻12), AP amplitude (p=2.24×10⁻11). Rheobase increased from infancy to adolescence (infant vs adolescence p=4.15×10⁻13) then declined (adolescence vs late adulthood p=5.23×10⁻8). APs in infants were broader (larger half-width), had slower upstrokes, and smaller amplitudes; AP amplitude declined again with aging.
• Firing patterns: F-I slope did not differ significantly across ages (p=0.055). First AP latency at rheobase was longest in infants (overall p=7.67×10⁻4; vs oldest p=8.41×10⁻6). Spike frequency adaptation differed modestly across ages (p=0.032), with lowest adaptation in early childhood.
• Depth dependence: Within-group comparisons by soma distance from L1 showed minimal effects, with a few exceptions (e.g., middle adulthood input resistance p=0.02; AP upstroke p=0.04; adolescence AP amplitude p=0.02; adaptation p=0.009).
• Morphology across ages: No significant differences in total dendritic length (p=0.37), apical length (p=0.6), basal length (p=0.28), total number of nodes (p=0.18), maximal horizontal (p=0.64) or vertical (p=0.51) extent. Average apical terminal segment length differed across ages (p=0.033), basal terminal length did not (p=0.85).
• Dendritic spines (infant vs late adulthood): Total spine density was higher in infants (p=7.57×10⁻40); both apical (p=2.02×10⁻31) and basal (p=3.8×10⁻12) dendrites showed higher densities in infants. Spine-type composition shifted with age: mushroom spines increased in late adulthood (apical p=4.4×10⁻9; basal p=9.04×10⁻8); thin spines and filopodia were enriched in infants (apical thin p=7.34×10⁻14; apical filopodia p=1.11×10⁻39; basal thin p=2.46×10⁻8; basal filopodia p=2.14×10⁻12). Branched spines were more frequent in infants (apical p=1.64×10⁻11; basal p=8.9×10⁻5). Stubby spines were more prevalent in older cells (apical p=7.19×10⁻5; basal p=6.97×10⁻9).
• Pathology and sex: Overall, electrophysiological and morphological properties were broadly similar between tumor and hydrocephalus groups, with sporadic differences (e.g., tau lower in hydrocephalus in young and late adulthood; AP half-width differences in young adulthood; rheobase differences in early childhood; some firing pattern differences in infants and late adulthood). Sex-related differences were sporadic in specific age bins for some electrophysiological and morphological metrics.
• Multivariate separation: UMAP embedding showed infant neurons clustered apart from older groups based on electrophysiological properties.

The data demonstrate pronounced maturational changes in intrinsic membrane and spike properties of human L2/3 pyramidal neurons primarily within the first year of life, relative stability across much of adulthood, and selective changes in aging. Early development features a depolarized resting Vm, high input resistance and tau, lower rheobase, broader and slower APs, and longer first-spike latencies, consistent with increasing cell size, changes in leak conductance (e.g., KCNK expression), and increasing voltage-gated Na+ channel expression. With aging, input resistance and tau rise again and sag increases, implicating altered leak and HCN channel function; AP amplitudes decrease and kinetics slow, in line with decreased Na+ channel expression. Together, these trajectories yield an inverted U-shaped pattern across the lifespan for several active properties and a monotonic increase of sag. Morphologically, dendritic arbor size and complexity remain relatively conserved across ages in these L2/3 neurons, while spine density decreases and spine-type composition shifts from thin/filopodia/branched (immature) toward mushroom/stubby (mature) with age, aligning with known synaptogenesis and pruning processes. These intrinsic and synaptic-structural changes likely contribute to age-dependent alterations in neuronal input-output transformations and network function, offering mechanistic links to developmental gains and aging-related declines in cognition. The results refine lifespan models of human cortical neuron function and provide empirically grounded parameters for biophysically realistic modeling of human cortical circuits.

This study establishes a comprehensive lifespan profile of electrophysiological and morphological features of human cortical L2/3 pyramidal neurons from birth to 85 years. The most substantial intrinsic changes occur during the first year of life, with neurons becoming less excitable and more temporally precise as they mature. Adulthood shows relative stability, followed by aging-associated increases in sag and passive resistive properties and reductions in AP amplitude. Dendritic arbor metrics are largely conserved across ages, while dendritic spine density declines and spine-type composition shifts toward mature forms in older neurons. These findings advance understanding of human cortical neuron development and aging and will aid the construction of more realistic models of human cortical function. Future work could expand spine analyses across more ages and brain regions, increase morphological sample sizes, systematically control for cortical depth and subtype differences, integrate transcriptomics to link ion channel expression to physiology, and assess synaptic and network-level consequences in disease contexts.

• Tissue source and clinical heterogeneity: Samples were obtained from patients undergoing neurosurgery for various pathologies (tumors, hydrocephalus, etc.), which may introduce biological variability despite similar handling across cases. Pathology subgroup analyses showed mostly minor differences but cannot eliminate confounding.
• Cross-sectional design: Age-related trends are inferred across different individuals rather than longitudinally within subjects.
• Sampling and regional variability: Most samples were from frontal and temporal lobes, with fewer from parietal and occipital cortex; regional and subtype differences may affect generalizability.
• Layer depth recovery: Soma position relative to L1 was confirmed in only 36% of cells; although depth analyses suggested minimal impact, incomplete depth information may mask depth-related effects.
• Electrophysiological quality constraints: Strict series resistance thresholds reduced the dataset for fast AP metrics, potentially biasing comparisons.
• Morphology and spine analyses sample size: Only 63 neurons underwent full 3D reconstruction and only 6 cells had exhaustive spine annotation, limiting statistical power and generalizability of morphological and spine findings.
• In vitro slice conditions: Slicing and recording conditions may alter physiological properties compared to in vivo function; anesthesia and perioperative factors could have indirect effects.