Sensory input drives rapid homeostatic scaling of the axon initial segment in mouse barrel cortex

Abstract

The axon initial segment (AIS) is a critical microdomain for action potential initiation and implicated in the regulation of neuronal excitability during activity-dependent plasticity. While structural AIS plasticity has been suggested to fine-tune neuronal activity when network states change, whether it acts in vivo as a homeostatic regulatory mechanism in behaviorally relevant contexts remains poorly understood. Using the mouse whisker-to-barrel pathway as a model system in combination with immunofluorescence, confocal analysis and electrophysiological recordings, we observed bidirectional AIS plasticity in cortical pyramidal neurons. Furthermore, we find that structural and functional AIS remodeling occurs in distinct temporal domains: Long-term sensory deprivation elicits an AIS length increase, accompanied with an increase in neuronal excitability, while sensory enrichment results in a rapid AIS shortening, accompanied by a decrease in action potential generation. Our findings highlight a central role of the AIS in the homeostatic regulation of neuronal input-output relations.

Fig. 1. AIS development in S1BF in vivo.

a Representative confocal images of AIS length maturation for one embryonal and three postnatal ages (E20, P1, P15, and P180) in cortical layers II/III and V. Immunostaining against ßIV-spectrin (gray), TOPRO (blue) or NeuN (blue) as indicated. Note the long AIS at P15 in both layers and the appearance of nodes of Ranvier in P180 animals. Arrowheads indicate start and end of AIS; orange lines indicate extent of a single AIS at P15. Scale bar 10 µm. b Population data of AIS lengths from E20 to P180 in layer II/III (top) and V (bottom). E20 data was pooled for supra- and infragranular layers. Initially, AIS length increases until P13–15, after which it decreases. Gray bars indicate the onset of active exploration and whisking (P12–14). Adult animals maintain an intermediate AIS length throughout life (for layer II/III and layer V, One-way ANOVA ***P < 0.001, Holm–Sidak’s post-hoc comparisons, n = 6 biologically independent experiments per age group, n = 5 biologically independent experiments for P7; >100 AIS per animal. For all comparisons, *P < 0.05, **P < 0.01, ***P < 0.001, summary of P values see Tables S1 and S2). Boxplots: median with 25–75% interval, error bars show minimum to maximum values. c Representative immunoblot of the major ankG isoforms (190, 270, 480 kilodalton, kDa) for postnatal developmental stages. Actin was used as a loading control. d Quantification of immunoblot data derived from three independent experiments run in parallel. Protein expression of the main ankG 480 kDa isoform peaks at P15 (Two-way ANOVA ***P < 0.0001 for time, *P = 0.067 for isoform, ***P = 0.0007 (time x isoform) for the interaction, n = 3 biologically independent experiments per age group, Tukey’s multiple comparisons tests *P < 0.05, **P < 0.01, ***P < 0.001, only a selection of statistical comparisons are depicted for visualization. Summary of all P values in Table S3). Data shown as mean ± SD.

Fig. 2. Long-term sensory deprivation elongates the AIS in layer II/III of young and adult mice.

a Experimental designs for whisker trimming. Group 1, daily bilateral trimming from P0 to P15, P21 or P45, respectively. Group 2, daily bilateral trimming from P10 to P15. Group 3, daily bilateral trimming until P21, followed by regrowth until P45. Group 4, daily bilateral trimming for 16 days in mice > P100. b Representative confocal images of layer II/III AIS stained for ßIV-spectrin (gray) and NeuN (blue) in control and deprived (Dep) mice. Arrowheads indicate start and end of AIS. Scale bar 10 µm. c Population data for Group 1 (control = gray; deprived = dark blue). Deprivation leads to longer AIS (Two-way ANOVA P < 0.0001 for deprivation and development, P = 0.417 (deprivation x development), >100 AIS/animal, n = 5 biologically independent experiments for P45 Dep, n = 6 biologically independent experiments for all other groups, Sidak’s multiple comparison tests *P = 0.042, **P = 0.002, ***P = 0.0002). d Population data for Group 2 (control = gray; deprived = red). Deprivation from P10 to P15 did not lead to significant length changes (unpaired two-sided t-test P = 0.53, >100 AIS/animal, n = 6 biologically independent experiments). e Length frequency histograms for P15 and P21 (control = gray; deprived = dark blue) Dep vs Ctrl, >600 AIS per condition, Kolmogorov–Smirnov test ***P < 0.0001, median: P15: Ctrl 30.8 µm, Dep 35.8 µm, P21 Ctrl 25.6 µm, Dep 31.0 µm). f Population data for Group 3 (deprived = light blue; age-matched controls = gray) at P45. Upon whisker regrowth, AIS length returned to control levels (unpaired two-sided t-test P = 0.22, >100 AIS/animal, n = 6 biologically independent experiments). g Population data for Group 4 (adult > P100; deprived = orange; age-matched controls = gray). Deprived animals showed significantly longer AIS after two weeks of whisker trimming (unpaired two-sided t-test **P = 0.0046, >100 AIS/animal, n = 6 biologically independent experiments). Boxplots: median with 25–75% interval, error bars show minimum to maximum values.

Fig. 3. Whisker trimming increases neuronal excitability in layer II/III pyramidal neurons.

a Left: Representative traces of AP trains elicited by current injection (500 ms, –50 pA, +100 pA). Note the increased firing frequency in the deprived neurons (blue trace). Right: Input/frequency relationship as determined by 500 ms injections of increasing currents. The deprivation group showed significantly increased firing frequencies (Two-way ANOVA ***P < 0.0001 for current injection and deprivation, Holm–Sidak’s multiple comparison test *P = 0.014, **P = 0.0037, n = 15 Ctrl cells, 20 Dep cells from at least 6 biologically independent experiments). Data are presented as mean ± SD. b Analysis of maximum slope f’(max) of the I-f curve and current at the maximum slope I at f’(max). Current at f’(max) was significantly lower in the deprivation group (Mann–Whitney test **P = 0.0028). Maximum slope was unchanged (unpaired two-sided t-test P = 0.15; n = 14 Ctrl cells, 18 Dep cells from at least 6 biologically independent experiments). c Representative traces of single APs elicited by 20 ms current injections; 10 pA increment injections to determine current and voltage threshold (inset). d Deprived neurons had significantly lower current threshold; voltage threshold was unchanged (unpaired two-sided t-test, current threshold *P = 0.015, voltage threshold P = 0.070 n = 15 Ctrl cells, 18 Dep cells from at least 6 biologically independent experiments). e Representative confocal image of a P15 Dep neuron filled with biocytin (blue) for post-hoc determination of AIS length (βIV-spectrin, gray). Arrowheads indicate start and end of AIS. Scale bar 10 µm. f Correlation analysis of the relationship between AIS length and current threshold. Results of linear regression analysis indicated in figure (n = 13 cells from at least 6 biologically independent experiments). Red dot indicates sample cell from image in e. g Left: Representative traces of spontaneous postsynaptic currents (PSCs) recorded at –90 mV. Top trace: 6 s of recording. Bottom trace: averaged PSCs across the entire recording session of 2 min for two sample cells. Right: Mean amplitude and frequency of PSCs. Deprived neurons received significantly more input, with on average a lower amplitude (unpaired two-sided t-test, amplitude: *P = 0.035, frequency: **P = 0.0042, n = 11 cells for Ctrl and Dep from at least 6 biologically independent experiments). Boxplots: median with 25–75% interval, error bars show minimum to maximum values.

Fig. 4. Rapid structural AIS remodeling after exposure to an enriched environment (EE).

a Experimental design: P28 mice (n = 5–6 biologically independent experiments/time point), exposed to EE conditions 12 h after unilateral whisker trimming. Mice remained in EE for 0, 1, 3, and 6 h, respectively, and were sacrificed immediately after. Layer II/III S1BF pyramidal neurons contralateral to the remaining whiskers represented stimulated side (EE); ipsilateral S1BF served as control (Ctrl). b Representative confocal images of EE and Ctrl after immunostaining of layer II/III neurons at 0, 1, 3, and 6 h EE exposure. AIS demarked by ankG (grey), with TOPRO (blue). Staining against c-Fos (red) served as indicator of neuronal activity. Inverted black & white panels highlight the increasing c-Fos signal over time. Scale bar 30 µm. c Quantification of c-Fos expression indicates rapid and significant upregulation after 1 h of EE (19.6% c-Fos⁺ neurons) with a peak expression after 3 h EE (38% c-Fos⁺ neurons), and downregulation after 6 h EE (1% c-Fos⁺ neurons). Two-way ANOVA ***P < 0.0001 for time, EE and EE x time, Sidak’s multiple comparisons test ***P < 0.0001, n = 3 biologically independent experiments. Data are presented as mean ± SD. d Length reduction in S1BF layer II/III neurons (1 h, 3 h EE). AIS length reduction is reversed after 6 h EE (Two-way RM ANOVA P = 0.021 for the factor EE, P = 0.16 for the factor time (ns), **P = 0.0046 for EE x time, Sidak’s multiple comparisons test *P = 0.043, **P = 0.0051, n = 5–6 biologically independent experiments; > 100 AIS/animal). Lines indicate matching data from the two hemispheres of individual mice. e Reversal of EE effects: After 3 h EE, AIS length normalized in home-cage (paired two-sided t-test P = 0.16, n = 5 biologically independent experiments, > 100 AIS/animal). f Exposure to a new EE induced AIS length shortening (paired two-sided t-test **P = 0.0035, n = 5 biologically independent experiments, >100 AIS/animal). g Population analysis (c-Fos⁺ and c-Fos⁻ cells) after EE reveals a significantly decreased AIS length in c-Fos⁺ neurons (One-way ANOVA P < 0.0001, Sidak’s multiple comparisons test ***P < 0.0001). Boxplots: median with 25–75% interval, error bars show minimum to maximum values.

Fig. 5. Decreased excitability of layer II/III pyramidal neurons after 3 h enriched environment (EE).

a Left: Representative traces of AP trains elicited by current injection (500 ms, −50 pA, +250pA). Note the decreased firing frequency in the EE neuron (green trace). Right: Input/frequency relationship as determined by 500 ms long injections of increasing currents. The EE group showed significantly decreased firing frequencies (Two-way ANOVA (current x EE) ***P < 0.0001 for current injection and EE, P = 0.38 for the interaction, Sidak’s multiple comparison test *P = 0.035, n = 13 Ctrl cells, 12 EE cells from 10 biologically independent experiments). Data are presented as mean ± SD. b Analysis of maximum slope f’(max) of the I-f curve and current at the maximum slope I at f’(max) for the respective neuron. f’(max) was significantly lower in the EE group (unpaired two-sided t-test *P = 0.032). I at f’(max) was unchanged (unpaired two-sided t-test P = 0.058 n = 13 Ctrl cells, 12 EE cells from 10 biologically independent experiments). c Representative traces of single APs elicited by 20 ms current injections; 10 pA increments of injected current used to determine AP current and voltage threshold (inset). d Analysis of current and voltage threshold for threshold APs. EE neurons had significantly higher current threshold (unpaired two-sided t-test *P = 0.033). Voltage threshold was unchanged (Mann–Whitney test P = 0.73, n = 13 Ctrl cells, 12 EE cells from 10 biologically independent experiments). e Representative confocal image of an EE neuron filled with biocytin (blue) for AIS length analysis via colabeling with βIV-spectrin (gray). Arrowheads indicate start and end of AIS. Scale bar 10 µm. f Correlation analysis of relationship between AIS length and current threshold. Results of linear regression analysis indicated in figure (n = 11 cells from at least 6 biologically independent experiments). Red dot indicates example cell from image in e. g Representative traces of postsynaptic currents (PSCs) recorded at –90 mV. Top trace shows 6 s of recording. Bottom trace shows the averaged detected PSCs across the entire recording session of 2 min. h Mean amplitude and frequency of PSCs. No significant difference was observed (unpaired two-sided t-test, amplitude: P = 0.058, frequency: P = 0.22, n = 8 Ctrl cells, n = 9 EE cells from at least six biologically independent experiments). Boxplots: median with 25–75% interval, error bars show minimum to maximum values.

Fig. 6. Rapid AIS length shortening in acute slices after 1 h of increased activity.

a Representative confocal images of layer II/III neurons in S1BF of acute slices that were subjected to 1 h of KCl (8 mM) or bicuculline (Bicu, 15 µM) treatment. Note the shorter AIS length in comparison to Ctrl conditions. Scale bar 10 µm. b Population analysis of AIS length in layer II/III S1BF after exposure to Bicu: 1 h leads to a shortening of AIS length (One-way ANOVA **P = 0.010, Dunnett’s multiple comparisons test **P = 0.0040). AIS length returns to baseline after 3 h and 6 h of Bicu treatment. N = 4 slices for Ctrl, n = 7 independent experiments for 1, 3 h; n = 8 independent experiments for 6 h, slices derived from 2 mice, > 200 AIS/ condition. c Population analysis of AIS length in layer II/III S1BF after exposure to elevated extracellular KCl: 1 h leads to shortening of AIS length (One-way ANOVA **P = 0.0044, Dunnett’s multiple comparisons test *P = 0.035). AIS length returns to baseline after 3 h and 6 h of KCl treatment. N = 4 biologically independent experiments for Ctrl, n = 8 biologically independent experiments for 1, 3, 6 h, slices derived from 2 mice, > 200 AIS/condition. Boxplots: median with 25–75% interval, error bars show minimum to maximum values.