(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Adaptations of the axon initial segment in fast-spiking interneurons of the human neocortex support low action potential thresholds [1]

['Emoke Bakos', 'Hungarian Center Of Excellence For Molecular Medicine Research Group For Human Neuron Physiology', 'Therapy', 'Szeged', 'Department Of Physiology', 'Anatomy', 'Neuroscience', 'University Of Szeged', 'Ádám Tiszlavicz', 'Viktor Szegedi']

Date: 2025-12 

The mammalian brain exhibits notable interspecies variation. Microanatomical and molecular differences in homologous neurons, those with similar locations and developmental origins across species, are best characterized in the neocortical mantle, the center of complex brain functions; however, the purpose of these differences remains unclear. We performed whole-cell microelectrode recordings along with microanatomical and molecular analyses of human fast-spiking parvalbumin (pvalb)-expressing interneurons in neocortical tissue resected during brain surgery, comparing them with similar data obtained from the mouse neocortex. The action potential (AP) firing threshold was lower in human neurons than in mouse neurons. This was due to a deficiency in low-voltage–activated inhibitory Kv1.1 and Kv1.2 potassium channels in the axon initial segment (AIS), a specialized axonal region that determines AP threshold and initiation, in human cells. In contrast, Kv1 ion channels were prominent in mouse neurons. The AIS was also moderately elongated in humans. Computational simulations of fast-spiking interneurons revealed that the human-type AIS lowers the AP threshold and shortens the time lag for AP initiation. We found that the low membrane AP firing threshold in pvalb neurons is closely linked to slow membrane potential kinetics in the soma. Thus, the human AIS supports fast in–fast out circuit function in human pvalb neurons, compensating for electrically slow somatic membrane responses. When formulating therapeutic strategies that involve fast-spiking neurons, it is crucial to take into account the molecular and functional species differences.

Funding: Project no. TKP-2021-EGA-05 was implemented with support provided by the Ministry of Culture and Innovation of Hungary from the National Research, Development, and Innovation Fund ( https://nkfih.gov.hu ), financed under the TKP2021-EGA funding scheme (K.L). Project no. 2022-2.1.1-NL-2022-00005 was implemented with support provided by the Ministry of Culture and Innovation of Hungary from the National Research, Development, and Innovation Fund ( https://nkfih.gov.hu ), financed under the 2022-2.1.1-NL funding scheme (K.L). Additional funding was received from the EU’s Horizon 2020 research and innovation program ( https:/ec.europa.eu/programmes/horizon2020 ) under grant agreement no. 739593 (K.L.);Nemzeti Kutatási, Fejlesztési és Innovaciós Alap OTKA K 134279 and ADVANCED_150382 ( https://nkfih.gov.hu ) (K.L.); the Hungarian Scientific Research Foundation ( https://nkfih.gov.hu ) grant number ANN-135291 (K.L.); Eötvös Loránd Research Network ( https://hun-ren.hu ) grants ELKH-SZTE Agykérgi Neuronhálózatok Kutatócsoport and KÖ-36/2021 (G.T.); Ministry of Human Capacities Hungary (20391-3/2018/FEKUSTRAT (G.T.) and NKP 16-3-VIII-3) (G.T.); National Research, Development and Innovation Office ( https://nkfih.gov.hu ) grants GINOP 2.3.2-15-2016-00018 (G.T.); Élvonal KKP 133807 (G.T.), ÚNKP-20-5 - SZTE-681 (G.T.), 2019-2.1.7-ERA-NET-2022-00038 (G.T.), TKP2021-EGA-09 (G.T.), TKP-2021-EGA-28 (G.T.), and OTKA K128863 (G.T.). In addition, funding was received from National Institutes of Health ( https://www.nih.gov ) awards U01MH114812 (G.T.) and UM1MH130981 (G.T.); and National Research, Development, and Innovation Office ( https://nkfih.gov.hu ) grant no. ANN_135291 (A.S.). Grant No. 8110 was received from the University of Szeged for the open access publication of the manuscript ( https://szerzoknek.ek.szte.hu/tamogatott-oa-publikalas/ ). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Copyright: © 2025 Bakos et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

In the present study, whole-cell electrophysiological recordings in human and mouse ex vivo neocortical slices revealed an association between slow somatic V m changes and a lowered AP generation threshold in pvalb/PVALB interneurons. The reduced AP firing threshold in human cells was accompanied by molecular specialization of the axon initial segment (AIS), a speciated axonal structure that initiates APs [ 41 , 42 ], which is deficient in inhibitory Kv1 potassium channels in human neurons. Simulations of fast-spiking interneurons with a human-type AIS showed that a reduced AP threshold shortens the latency of AP initiation, enabling human pvalb neurons to function as fast in–fast out circuits despite slower somatic membrane responses [ 43 ].

Distinctive differences in homologous neurons between the rodent and human neocortex include electrical resistivity and the membrane potential level required to trigger action potentials (APs) [ 25 ]. Human neocortical excitatory pyramidal neurons exhibit lower input resistance (Rin) and require greater membrane depolarization (i.e., a higher AP firing threshold) to trigger APs compared with rodent neurons [ 24 , 29 , 35 ]. In contrast, inhibitory interneurons in the human neocortex display higher input resistance and a lower AP firing threshold relative to their rodent counterparts [ 26 , 27 ]. Prior studies suggest that these interspecies differences also apply to fast-spiking interneurons [ 28 , 31 , 36 ], the principal inhibitory neuron population in the neocortex, characterized by the expression of parvalbumin (PVALB), the gene encoding pvalb protein [ 23 ]. In humans, pvalb neurons exhibit elevated input resistance; therefore, they generate membrane potential (V m ) changes with reduced electric input [ 28 ]. For the same reason, V m changes, such as excitatory synaptic potentials that lead to APs, occur more slowly [ 28 , 31 , 36 – 39 ]. The delayed V m response in human fast-spiking neurons, compared with that in their rodent counterparts, appears counterintuitive, as the rapid transformation of excitatory input to AP output (i.e., input–output transformation) is a hallmark of pvalb interneurons, also known as “fast in–fast out” inhibitory circuits [ 38 , 40 ]. Pvalb interneurons play a crucial role in complex brain functions. They rapidly inhibit other neurons and orchestrate temporally coordinated network activities that underlie cognitive processes. Their ability to rapidly and reliably inhibit other neurons allows them to precisely control local circuits and shape brain network activity. We hypothesize that a low AP threshold coexists with slow membrane potential kinetics in human pvalb neurons to accelerate their operation in these processes.

All organs in the body evolved specialized functions to optimally support animal life and vitality. The mammalian brain exhibits interspecies differences not only in volume and cortical convolution but also in nerve cell microstructure and molecular composition [ 1 , 2 ]. Although the brain comprises homologous neurons, sharing similar locations and developmental origins across species, these cells show species-specific differentiation at the microscale level [ 3 – 6 ]. The purpose of this neuronal specialization across species remains poorly understood [ 7 , 8 ], although these adaptations likely optimize neural circuit function [ 9 ]. Neuron types and their species-specific features are best characterized in the neocortical mantle, the center of complex high-order brain functions [ 1 , 5 , 6 , 10 – 19 ]. Neocortical function has been intensively studied in relation to brain cognitive and behavioral processes in humans and animal models, the latter aiding in elucidating the roles of neuron types in these processes [ 20 – 22 ]. However, studies using resected human brain tissue have revealed various “human-specific” transcriptomic and microstructural features in neocortical neurons [ 7 , 10 , 11 , 23 – 31 ]. Many of these differences likely contribute to species-specific neuronal functions; however, the structure–function relationship remains unclear [ 12 , 16 , 17 , 19 , 30 , 32 – 34 ].

Results

Low AP threshold is associated with electrically slow somata in pvalb interneurons Compared with mouse pvalb neurons, human pvalb neurons exhibited higher electrical resistance, i.e., input resistance (Rin) in their somata, with no difference in capacitance (Cm). We measured average Rin and Cm using subthreshold V m steps of +5 to +10 mV (elicited by square-pulse currents at −70 mV; Fig 2A 1 ). Median Rin was 187.9 MΩ (IQR = 103.8–279.8) in human cells (n = 60) versus 125.7 MΩ (IQR = 89.5–167.5) in mouse cells (n = 50; p < 0.001, Mann–Whitney U test; Fig 2A 2 ). Median Cm in the soma was similar across species (humans: 32.4 pF, IQR = 24.3–53.7, n = 60; mice: 35.1 pF, IQR = 26.3–50.6, n = 50; p = 0.728, Mann–Whitney U test). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Low AP firing threshold in pvalb interneurons is associated with an electrically slow soma. (A) Human pvalb cells exhibited higher Rin compared with mouse cells at V m between −70 mV and the AP threshold. (A 1 ) Cell Rin was measured using positive current steps inducing 5–10-mV subthreshold depolarizations. Voltage shifts (ΔV m ) and currents steps (ΔI) used to calculate average Rin are shown in red. (A 2 ) Histograms and box plots show Rin in human (blue) and mouse (gray) pvalb neurons; bin size: 25 MΩ; box plots present the median, quartiles, and 5th and 95th percentiles (Mann–Whitney U test). (B) “Tau” was longer in human than in mouse pvalb cells for V m shifts in the soma. (B 1 ) “Tau” was measured from the ascending phase in +10-mV steps delivered at −70 mV (human: blue; mouse: gray). Left: Four superimposed +10-mV steps in human and mouse cells. Right: Ascending phase shows species differences in “tau” (human average: 8.9 ms; mouse average: 3.4 ms). (B 2 ) Histograms and box plots illustrate “tau” differences in human (blue) and mouse (gray) pvalb neurons; bin size: 0.5 ms (Mann–Whitney U test). (C) Soma “tau” and Rin were strongly associated in mouse (C 1 ) but not human (C 2 ) pvalb neurons (Spearman’s correlation). (D) Low AP threshold was linked with slow “tau” in mouse (D 1 ) and human (D 2 ) pvalb neurons. (D 3 ) Combined data from both species revealed a strong association between slower “tau” and lower AP threshold (Spearman’s correlation). The data underlying this Figure can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3003549.g002 The time course of V m changes (apparent tau, i.e., “tau”), measured from the ascending phase of +10 mV steps, was slower in human pvalb neurons (median “tau” of the V m step: 6.65 ms, IQR = 4.22–8.17, n = 60) than in mouse pvalb neurons (median: 4.17 ms, IQR = 3.27–6.12, n = 50; p < 0.001, Mann–Whitney U test; Fig 2B 1 and 2B 2 ). In mice, “tau” and Rin showed a significant positive correlation (r = 0.613, n = 50, p < 0.001; Fig 2C 1 ), whereas no such correlation was found in human pvalb neurons (r = 0.224, n = 60, p = 0.085) (Fig 2C 2 ). However, in both species, lower AP thresholds were associated with slower “tau” (human cells: r = −0.274, n = 60, p = 0.034; mouse cells: r = −0.535, n = 50, p < 0.001; Fig 2D 1 and 2D 2 ). Combined data from both species revealed a strong negative correlation between AP threshold and slow “tau” (r = −0.514, n = 110; p < 0.001, Spearman’s correlation; Fig 2D 3 ). Collectively, these results demonstrate that pvalb neurons with slower V m kinetics to depolarization in the soma tend to have lower AP firing thresholds. This association was particularly strong when analyzing human and mouse pvalb neurons together.

AIS is moderately elongated in human pvalb neurons and exhibits an axonal position associated with AP threshold To explore whether AIS morphology contributes to the lower AP threshold in human pvalb neurons, we visualized the AIS, a primary site of AP initiation, as its length and position can affect AP threshold [45]. We used immunohistochemistry against beta-IV-spectrin, a cytoskeletal protein enriched in the AIS [42,46,47], analyzing its immunofluorescence in three-dimensional (3D) confocal microscopy image stacks, with AIS length and distance from the soma quantified using a custom script. All cells were located in layer 2/3 of the neocortex and confirmed as pvalb-positive (Fig 3A 1 ). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. AIS is elongated in human pvalb neurons and axonally located according to AP threshold. (A) AIS in pvalb neurons was moderately longer in humans than in mice. AIS analysis was performed in triple-immunostained neurons from the human and mouse neocortex. (A 1 ) AIS was measured in 3D fluorescence confocal images of layer 2/3 pvalb-positive neurons. Left: AIS visualized via beta-IV-spectrin immunofluorescence. Arrows indicate AIS initiation and termination sites (Alexa 488) in a collapsed z-stack image of a human pvalb neuron. Right: Merged beta-IV-spectrin and pvalb (Cy5) immunoreactions. Horizontal dotted curves mark (i) the axon from the soma (s) and the proximal (ii) and (iii) distal AIS ends. (A 2 ) Histograms and box plots summarize AIS parameters from the upper schematic: a = distance from the soma to the AIS initiation site, b = AIS length, and c = distance from the soma to the AIS end. Top: AIS initiation distance from the soma did not differ between species. Middle: AIS was longer in human neurons than in mouse neurons. Bottom: AIS termination site was more distal in human neurons than in mouse neurons. Histogram bin: 1 µm, with kernel density; box plots show the median, quartiles, and 5th and 95th percentiles. (A 3 ) Cumulative AIS data show p-values (Student t test). (B, C) AIS measurements in electrophysiologically investigated pvalb neurons revealed that a low AP threshold is associated with a more distal AIS termination site. (B 1 ) AIS visualized via beta-IV-spectrin immunopositivity (Cy3) in human and mouse cells filled with biocytin (bcyt) and labeled with streptavidin–Alexa 488 conjugates. Immunofluorescence images show AIS features defined in A1. (B 2 ) Electrophysiologically studied human pvalb neurons displayed a longer AIS (b) with a more distal termination site (c) compared with mouse neurons, whereas the AIS start distance (a) from the soma did not differ between species (Mann–Whitney U test). Combined AP threshold data from human (blue) and mouse (gray) pvalb neurons revealed associations with AIS initiation site (C 1 ) and termination site (C 3 ) distance from the soma, but no association with AIS length (C 2 ) (Spearman’s correlation). The data underlying this Figure can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3003549.g003 Both human and mouse pvalb neurons exhibited cell-to-cell variation in AIS length and positioning along the axon (Fig 3A 2 ). The median AIS start point was similar between species (humans: 4.12 µm, IQR = 2.95–5.28, n = 100 cells in 17 patients; mice: 3.77 µm, IQR = 2.69–4.68, n = 78 cells in 15 mice; p = 0.265), but the AIS was longer in human cells (median: 16.68 µm, IQR = 14.08–19.78) than in mouse cells (median: 13.99 µm, IQR = 11.15–16.55; p < 0.001). Consequently, the AIS termination site was further from the soma in human neurons (21.01 µm, IQR = 17.60–24.66, n = 100 cells from 17 samples) than in mouse neurons (17.47 µm, IQR = 14.41–20.66 µm, n = 78 cells from 15 mice; p < 0.001, Mann–Whitney U test; Fig 3A 2 and 3A 3 ). We also investigated whether these morphological differences relate to the lower AP threshold observed in humans. In electrophysiologically studied and biocytin-filled neurons visualized via streptavidin–fluorophore conjugates, with confirmed immunopositivity for pvalb, we observed the AIS based on beta-IV-spectrin positivity and measured its length and distance from the soma along the axon, comparing these metrics against the AP firing threshold. Of the 28 human and 22 mouse cells recorded in the whole-cell clamp, intact somata and full-length AIS were successfully recovered in 13 human and 14 mouse pvalb immunopositive neurons. The total number of cells includes all cells that were electrophysiologically recorded and subsequently visualized via biocytin staining. Cells that failed include pvalb cells with ruptured or damaged somata, failure to confirm continuation of the axon after the beta-IV-spectrin-labeled AIS, and failure to clearly measure AIS parameters due to an unsuccessful immunoreaction in human tissue. The double immunofluorescence reaction had to be successful for the pvalb and beta-IV-spectrin antibodies, and the axon had to initiate from the soma with a fully visualized AIS. A complete AIS was confirmed if the biocytin-filled axon continued beyond the beta-IV-spectrin-labeled AIS (Fig 3B 1 ). The distance from the AIS’s initial point to the soma in human pvalb neurons was 6.10 µm (IQR = 2.55–8.50, n = 13) compared with 4.85 µm (IQR = 3.80–6.03, n = 14) in mouse neurons (p = 0.437; Fig 3B 2 ). AIS length was greater in human cells (17.60 µm, IQR = 14.65–20.20 µm) than in mouse cells (14.0 µm, IQR = 12.10–15.33; p = 0.004), and the AIS terminal site was farther from the soma in human neurons (23.60 µm, IQR = 20.85–25.75) than in mouse neurons (18.55 µm, IQR = 17.68–20.38; p < 0.001, Mann–Whitney U test). All AIS properties are presented in Fig 3B 2 . None of the AIS parameters were correlated with the AP threshold in either species. The correlation between AIS proximal distance and AP threshold was 0.536 (p = 0.054) in human cells (n = 13) and −0.297 (p = 0.93) in mouse cells (n = 14). AIS length was not significantly correlated with AP threshold in humans (r = 0.203, p = 0.493) or mice (r = 0.079, p = 0.773), nor was distance from the AIS termination site to the soma correlated with AP threshold in either species (humans: r = −0.239, p = 0.414; mice: r = 0.024, p = 0.988, Spearman’s correlation). However, when pooling data across species, the AP threshold was significantly correlated with AIS start distance (r = −0.478, p = 0.012) and AIS termination distance (r = −0.383, p = 0.0485) but not AIS length (r = −0.1340, p = 0.516, Spearman’s correlation; Fig 3C 1 –3C 3 ). Because the number of cells studied electrophysiologically and investigated anatomically is relatively small, and because the range of measured AP threshold (AP Thr) and AIS structure values is large, it is possible that our dataset is not large enough to reveal hidden associations between AIS geometry and AP threshold within each species. Nevertheless, these findings suggest that the difference in AP threshold between human and mouse pvalb neurons cannot be explained by AIS length or axonal location. However, they suggest that the position of the AIS may partly explain the large variability observed in the AP threshold between individual neurons. Specifically, a more distal AIS placement is associated with a lower AP threshold.

Axon diameter does not account for interspecies differences in AP threshold We examined axon diameter in biocytin-filled pvalb interneurons to assess its potential correlation with the AP firing threshold, particularly across species, as axon caliber can influence axon excitability. Thicker axons have lower longitudinal resistance (Ra) in the AIS. This allows for faster passive electrical conduction between the soma and the AIS. However, the lowered resistance in the axon simultaneously increases the AIS’s AP rheobase current. However, larger axons also have more ion channels, including voltage-gated sodium channels that initiate the AP, suggesting a complex relationship between axon diameter and AP Thr [48]. Axonal diameters were measured using high-magnification fluorescence confocal microscopy images of biocytin-filled neurons at five standardized positions along the axon: two pre-AIS sites and three sites within the AIS (Fig 4A 1 –4A 3 ). Reported data represent the mean values per cell for the pre-AIS and within-AIS portions of the axon. No significant differences were observed in axon diameter between human (n = 12) and mouse (n = 14) pvalb neurons in either in the pre-AIS [1.07 µm (IQR = 0.82–1.19) versus 0.88 µm (IQR = 0.71–1.14); p = 0.939] or within-AIS portion [0.83 µm (IQR = 0.70–0.96) versus 0.69 µm (IQR = 0.64–0.74); p = 0.260; one-way ANOVA on ranks; Fig 4B]. Additionally, in both AIS portions, axon diameter was not correlated with AP threshold in humans (pre-AIS: r = 0.245, p = 0.429; within-AIS: r = 0.392, p = 0.197; n = 12) or mice (pre-AIS: r = 0.091, p = 0.766; within-AIS: r = 0.056, p = 0.852; n = 12). Pooled data across species also showed no association (pre-AIS: r = 0.137, p = 0.518; within-AIS: r = −0.036, p = 0.866, Spearman’s correlation; Fig 4C 1 and 4C 2 ). Collectively, these results indicate that axon diameter does not contribute to the observed interspecies differences in AP threshold in pvalb neurons. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. Axon diameter fails to explain interspecies differences in pvalb interneuron AP firing threshold. (A) Axon cross-diameter was measured in 3D confocal images of biocytin-filled pvalb interneurons. (A 1 ) Schematic showing axon maximum width at five points. “Pre-AIS” measurements were taken at the axon’s start (0 µm from the soma) and 2 µm from the soma. “AIS” measurements were taken at the AIS start (shown as 0) and at 2 and 5 µm from the AIS initiation site. (A 2 ) Collapsed confocal image showing biocytin-filled human pvalb neurons (Alexa 488), with the AIS visualized via beta-IV-spectrin immunoreactivity (Cy3). (A 3 ) Enlarged image of the axon highlighting five measurement sites. (B) Axon thickness was similar between human (blue) and mouse (gray) pvalb neurons. Horizontal bars indicate medians and quartiles. (C) Axon diameter was not correlated with AP threshold in either the pre-AIS region (C 1 ) or the AIS (C 2 ). Data from both species were pooled (one-way ANOVA on ranks with Dunn’s pairwise post hoc test). The data underlying this Figure can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3003549.g004

Pvalb neurons in humans lack KCNA1 mRNA encoding Kv1.1 but possess KCNA2 mRNA encoding Kv1.2 We used patch-sequencing to measure the mRNA levels of low-threshold ion channels that regulate the AP firing threshold, namely Kv1.1 and Kv1.2 [44,51,53–55,58], in electrophysiologically studied fast-spiking pvalb neurons in human and mouse neocortex layer 2/3. Whole-cell recordings were followed by nucleus extraction and transcriptomic analysis [59,60]. We detected mRNA for 6,784 genes (IQR = 4,872–7,399) in humans (n = 21) and 3,617 genes (IQR = 3,173–4,305) in mice (n = 16; Fig 8A). Transcriptomes were uploaded to the Allen Institute neuron type identification server, which employs an algorithm categorizing neocortex layer 2/3 neuron types by gene expression (https://knowledge.brain-map.org/mapmycells/process/), enabling identification of pvalb neurons, even in cells with undetectable PVALB mRNA expression. Using this method, we identified 21 and 16 human and mouse pvalb neurons, respectively. These cells were analyzed for expression of KCNA1, KCNA2, KCNC1, KCNC2, KCND1, KCNQ1, KCNQ2, KCNQ3, KCNQ4, KCNQ5, KCNK2, KCNK4, and KCK10, i.e., genes encoding major potassium channels in axons that regulate AP threshold [54,55]. Additionally, we analyzed SCN8A expression and that of the housekeeping genes ACTB, GAPDH, and SLC8A1 (Fig 8A). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 8. Human pvalb neurons lack KCNA1 mRNA and express low KCNA2 mRNA levels. (A) Patch-sequencing analysis (illustrated in the upper schematic) of individual neurons revealed that human pvalb neurons do not express KCNA1, in contrast to its expression in mouse cells. (A 1 ) Heatmap showing transcript levels (TPM) of genes encoding low-threshold potassium channels in individual human (n = 21) and mouse (n = 16) pvalb neurons, as well as the housekeeping genes GAPDH and ACTB. KCNA1 expression differed quantitatively (TPM values) and categorically (positive or negative gene expression) in human and mouse cells. Expression of KCNA2, KCNC1, KCNC2, KCNK2, KCNK4, and KCNK10 was observed in both species (61), whereas KCND1 expression was nearly absent in both (62). 0-level indicates that the mRNA expression level is below the detection threshold. Human and mouse cells expressed similar levels of KCNQ1, KCNQ2, and KCQN4, whereas KCNQ3 and KCNQ5 expression levels were higher in humans. Additionally, SCN8A (encoding Nav1.6 channels) and SLC6A1 (encoding GABA transporter type-1) were highly expressed in both species. Significant differences in TPM values are marked in red (genes, ordinate). Bottom of the heatmap shows the pvalb cell subtype classification based on Allen Institute transcriptomic data. (B) Allen Institute database analysis of layer 2/3 pvalb neuron subtypes in the human and mouse neocortex confirmed KCNA1 mRNA deficiency in human cells. Heatmap shows median TPM values (scale on the right) across seven human and eight mouse pvalb neuron subtypes (abscissa). KCNA1 mRNA was absent in all human subtypes but robustly expressed in mouse cells. Statistical analyses (one-way MANOVA with Bonferroni post hoc test and chi-squared test with Benjamini–Hochberg adjustment) confirmed significant differences in both expression level and frequency (p = 0.0011 and p < 0.001, respectively). KCNQ3 and KCNQ5 expression levels were higher in humans. Significant differences in TPM values are marked in red (genes, ordinate). Abscissa shows pvalb cell subtype names. The data underlying this Figure can be found in S3 Data. https://doi.org/10.1371/journal.pbio.3003549.g008 All 21 human pvalb neurons lacked detectable KCNA1 mRNA expression, whereas 7 of 16 mouse neurons showed robust expression. The mRNA level heatmap in Fig 8A shows the gene expression (transcripts per million [(TPM]) of individual pvalb neurons in humans and mice. Comparing KCNA1 mRNA expression between human (n = 21) and mouse cells (n = 16) revealed a significant difference (p = 0.0016, one-way MANOVA with a post hoc Bonferroni test). Furthermore, categorical comparisons (positive or negative mRNA expression) confirmed that KCNA1 expression differed significantly between species (p < 0.001, chi-squared test following Benjamini–Hochberg adjustment at the 5% significance level; Fig 8A). Both species displayed KCNA2 mRNA expression without any species-specific difference (p = 0.970). We also tested whether pooled mRNA levels (TPM) of KCNA1 and KCNA2 were associated with AP threshold in patch-sequenced neurons. No association was observed for either gene in humans (r = 0.041, n = 20, p = 0.861) or mice (r = 0.236, n = 15, p = 0.388). When neurons from both species were pooled, the results remained nonsignificant (r = 0.175, n = 35; p = 0.312, Spearman’s correlation test). Both species exhibited similarly strong KCNC1 and KCNC2 mRNA expression, which respectively encode the Kv.3.1 and Kv3.2 channels characteristics of pvalb interneurons [43,61]. They also showed equally low or undetectable expression of KCND1 encoding Kv4.1 potassium channels [62]. No species differences were observed in KCNK2, KCNK4, or KCNK10 mRNA levels, and KCNQ1, KCNQ2, and KCNQ4 mRNAs were similarly expressed between species. However, human neurons displayed significantly higher mRNA levels of KCNQ3 (p < 0.001) and KCNQ5 (p < 0.001). Both human and mouse pvalb cells strongly expressed SCN8A mRNA, encoding Nav1.6 sodium channels. Additionally, all cells expressed high levels of ACTB and GAPDH. Notably, mRNA expression of SLC6A1, which encodes GABA transporter 1, was significantly lower in human neurons than in mouse neurons (p = 0.002). PVALB mRNA expression was undetectable in 10 of 21 human neurons and 6 of 16 mouse neurons, aligning with results from the Allen Institute open access database, which reports markedly low or absence of average PVALB mRNA expression levels in some layer 2/3 pvalb cells (Fig 8B). Natural fluctuations in different gene expression levels in individual cells might explain the occasional undetectable mRNA in our patch-sequenced cells, but it is also possible that our system occasionally loses mRNA or misses its conversion to cDNA for reasons we do not yet recognize. The patch-sequencing results closely aligned with the average mRNA profiles of pvalb neuron types in the Allen Institute database (Fig 8B). The database confirmed the absence of KCNA1 expression in humans but not in mice (p < 0.001), as well as expression of KCNA2 in both species. It also confirmed that among the potassium channel genes analyzed, only Kv1 channel genes were consistently expressed at lower levels in human pvalb neurons relative to mouse neurons (one-way MANOVA with a post hoc Bonferroni test). The heatmaps based on our patch-sequencing data and the Allen Institute data differ in that our dataset shows raw TPM results in individual neurons, whereas the Allen Institute data show the average TPM level for multiple cells of that cell type. This may explain why, unlike the Allen Institute dataset, ours occasionally shows undetectable mRNA levels for common genes, such as SLC6A1, in some cells. It is also noteworthy that the Allen Institute dataset shows an absence of mRNA levels for PVALB in some cell types, even though the dataset represents the average of many cells.

Kv1 channel blocker dendrotoxin lowers the AP firing threshold more prominently in mouse neurons than in human neurons We investigated whether Kv1 channel blockade through dendrotoxin-K (DTXK) affects AP generation differently in human and mouse pvalb neurons [53,58]. After 5 min of baseline whole-cell current–clamp recording (conditions similar to those shown in Fig 1), slices were exposed to 100 nM DTXK for 10 min, and AP threshold was remeasured (Fig 9A 1 and 9A 2 ). DTXK significantly altered the AP firing threshold in human neurons from −41.5 mV (IQR = −44.0 to −39.0 mV) to −41.0 mV (IQR = −45.3 to −41.0 mV; n = 11 cells; p = 0.0459; mixed-design ANOVA and post hoc Wilcoxon test with Holm’s correction). In mouse cells, the AP threshold was lowered from −32.0 mV (IQR = −35.5 to −28.5 mV) to −39.3 mV (IQR = −44.2 to −38.0 mV; n = 15 cells; p = 0.0007; mixed-design ANOVA and post hoc Wilcoxon test with Holm’s correction). The change in AP threshold between baseline and in DTXK differed significantly between the human and mouse cells (p = 0.0009 for species interaction by mixed-design ANOVA) (Fig 9B 1 –9B 3 ). The AP threshold was more negative in humans than in mice at baseline (p = 0,0005), but there was no statistically significant difference between the species at DTXK (p = 0.0862) (mixed-design ANOVA and post hoc Mann–Whitney U tests between species at each time point, with Holm-adjusted p-values). Table 2 show the full statistical analysis and its explanation. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 9. Kv1 channel blocker dendrotoxin lowers AP threshold in mouse but not in human pvalb neurons. APs were recorded under baseline conditions (black traces) in current–clamp mode at −70 mV (as shown in Fig 1) for 5 min, followed by application of 100 nM DTXK for 10 min, after which APs were evoked again (red traces). (A) APs and corresponding AP phase plots in baseline control and after DTXK application. (A 1 ) Left: Three superimposed APs from a human pvalb cell. Right: Corresponding phase plots under control and DTXK conditions. (A 2 ) Left: Superimposed APs recorded in a mouse pvalb neuron. Right: Corresponding AP phase plots. (B–C) The AP parameters that were affected differently by DTXK in human and mouse pvalb neurons (mixed-design ANOVA). (B) Effect of DTXK on the AP threshold in human (B 1 ) and mouse (B 2 ) neurons (Wilcoxon signed-rank tests within each species with Holm-adjusted p-values for multiple comparisons). (B 3 ) The mixed-design ANOVA indicates a significant interaction between species and time on the measured values, meaning the DTXK treatment differs between human and mouse neurons. Box plots present the median, quartiles, and 5th and 95th percentiles. (C) Application of DTXK did not affect the rheobase current in human pvalb neurons (C 1 ), but it significantly reduced the current in mouse cells (C 2 ) (Wilcoxon signed-rank test with Holm-adjusted p-values for multiple comparisons). (C 3 ) A significant interaction was found between species and time in the rheobase values, indicating that DTXK treatment differs between humans and mice (mixed-design ANOVA). The data underlying this Figure can be found in S4 Data. https://doi.org/10.1371/journal.pbio.3003549.g009 PPT PowerPoint slide

PNG larger image

TIFF original image Download: Table 2. AP parameters, Rin, and “tau” in pvalb neurons under baseline and DTXK exposure conditions. https://doi.org/10.1371/journal.pbio.3003549.t002 Comparisons of the effect of DTXK across species revealed that, in addition to an AP threshold shift, the rheobase current was affected differently in mice than in humans (mixed-design ANOVA interaction, p = 0.0082). The rheobase current was reduced in mice (p = 0.0015), but not in humans (p = 0.4648; post hoc Wilcoxon tests with Holm’s correction) (Fig 9C 1 –9C 3 ). Although the rest of the observed parameters failed to reveal a significant difference in the DTXK effect between humans and mice (mixed-design ANOVA interaction, p > 0.05), Kv1 channel blockade was associated with a reduction in the amplitude of the action potential afterhyperpolarization (AP AHP) in mice, but not in human cells. This difference reflects the stronger somatic Kv1 channel contribution to AP AHP in mice compared to human cells in control conditions. The AP half-width, which was shorter in mice under baseline conditions, did not remain significantly shorter after the wash-in of DTXK, reflecting the blockade of K-current components of the AP in mice. Additionally, DTXK was associated with a similar reduction in the maximum positive peak value of the AP in both species. This suggests that the effect of DTXK on the AP occurred through a mechanism other than direct Kv1.1 channel current. (A mixed-design ANOVA test with post hoc pairwise comparisons are shown in full in Table 2.) In addition, DTXK significantly increased input resistance in mouse cells but not in human cells. Kv1.1 channels are activated in the cell soma by V m depolarization, which reduces Rin under baseline control conditions. Blocking these channels with DTXK increased the input resistance. Consistent with this finding, DTXK reduced species differences in membrane “tau”. “Tau” differed more between species in control conditions than in the presence of DTXK (see Table 2). The results indicate stronger Kv1 channel activity in the AIS and soma of mouse neurons compared to human pvalb neurons.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3003549

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/