Live imaging of excitable axonal microdomains in ankyrin-G-GFP mice

Abstract

The axon initial segment (AIS) constitutes not only the site of action potential initiation, but also a hub for activity-dependent modulation of output generation. Recent studies shedding light on AIS function used predominantly post-hoc approaches since no robust murine in vivo live reporters exist. Here, we introduce a reporter line in which the AIS is intrinsically labeled by an ankyrin-G-GFP fusion protein activated by Cre recombinase, tagging the native Ank3 gene. Using confocal, superresolution, and two-photon microscopy as well as whole-cell patch-clamp recordings in vitro, ex vivo, and in vivo, we confirm that the subcellular scaffold of the AIS and electrophysiological parameters of labeled cells remain unchanged. We further uncover rapid AIS remodeling following increased network activity in this model system, as well as highly reproducible in vivo labeling of AIS over weeks. This novel reporter line allows longitudinal studies of AIS modulation and plasticity in vivo in real-time and thus provides a unique approach to study subcellular plasticity in a broad range of applications.

Keywords: AIS plasticity; ankryin-G; axon initial segment; live imaging; mouse; neuroscience; node of Ranvier.

Figure 1.. Genetic strategy and creation of the ank-G-GFP line.

(A) Sketch of the intracellular scaffold underlying the axon initial segment (AIS). Ankyrin-G anchors ion channels to the axonal membrane. Ankyrin-G itself has spectrin binding domains, thus interacting with βIV-spectrin rings, which in turn bind to the axonal actin cytoskeleton. (B) Outline of transgenic modification before and after Cre-dependent recombination of ankyrin-G. The arrangement of the loxP and lox2272 sites ensures that recombination occurs only once. Both possible outcomes produce a functional Ank3 last exon with a coupled GFP sequence. (C) Sketch of the sequence of operations to create the ankyrin-G-GFP line. Detailed steps are outlined in the method section. The plasmid, targeting vector, and genomic target region are enlarged in the lower panel.

Figure 2.. Ank-G-GFP activation and expression in different experimental models intrinsically highlights the AIS in distinct neuron populations.

(A) Experimental design. Ank-G-GFP expression was triggered by (i) exposure of organotypic slice cultures (OTC) to an AAV-Cre, (ii) injection of AAV-Cre into ank-G-GFP mice, or (iii) breeding of ank-G-GFP mice to different Cre-lines. Anatomical regions are highlighted in green. Cartoons are used throughout the manuscript to indicate the brain regions investigated, the procedure of Cre-activation (heart = breeding with Cre-line; hexagon = AAV infection in vitro or injection in vivo), and tissue types (petri dish = dissociated cells; brain slice = OTC; mouse = cryosection, ex vivo acute slice, and in vivo). (B) Hippocampal OTC at days in vitro (DIV) 14 from ank-G-GFP mice exposed to AAV expressing Cre-recombinase under the synapsin promotor (AAV5-pmSyn1-EBFP-Cre). Ank-G-GFP+-AIS (left, green arrowheads) colabeled with βIV-spectrin (middle, magenta arrowheads) in pyramidal neurons; merged channels (right). Scale bar = 5 µm. (C) Cryosection prepared from an ank-G-GFP animal injected with a synapsin-Cre-tdTomato AAV 3 wk after injection. The ank-G-GFP+AIS (left, green arrowheads) is clearly discernible. Immunostaining against βIV-spectrin (middle, magenta arrowheads) indicates the same axon initial segment (AIS) (merged, right). Note the ank-G-GFP+ somatic envelop of hippocampal excitatory neurons (left, asterisks). Scale bar = 10 µm. (D) An ex vivo acute slice prepared from the hippocampus of an ank-G-GFP mouse injected with a combination of AAV1-hDlx-Flex-dTomato-Fishell_7 and AAV1-hSyn.Cre.WPRE.hGH. The tdTomato-positive interneuron (middle and right, magenta) is endowed with an ank-G-GFP+-AIS (left, green arrowhead). Scale bar = 10 µm. (E) Cryosection of neocortex from an ank-G-GFP × CaMKIIa-Cre mouse, highlighting ank-G-GFP+-AIS (left and middle, green arrowheads) and ank-G-GFP-, but βIV-spectrin+AIS (middle, magenta arrowheads) in layer II pyramidal neurons; merged channels (right). Scale bar = 10 µm. (F) Retinal whole mount preparation from an ank-G-GFP × CaMKIIa-Cre mouse. The image shows a peripheral aspect of the retina. The overlap of ank-G-GFP + AIS with the colabeling of ank-G is evident (green and magenta arrowheads in all panels), indicating that likely all RCG express CaMKII and consequently, all AIS are positive for both GFP and the intrinsic AIS marker ankyrin-G. Scale bar = 20 µm. (G) A cerebellar ex vivo acute slice prepared from ank-G-GFP × PV-Cre mice. This Purkinje cell was filled with biocytin via a patch pipette and stained with Streptavidin (right, magenta). The ank-G-GFP + AIS (left, green arrowhead, magnified from boxed region in right panel) is clearly discernible from surrounding ank-G-GFP-, βIV-spectrin + AIS (middle, cyan arrowheads). Scale bar (left and middle) = 10 µm, right = 20 µm.

Figure 3.. Ank-G-GFP activation and expression in nodes of Ranvier do not alter node morphology.

(A) Cryosection of hippocampal CA3 alveus from an ank-G-GFP × CaMKIIa-Cre mouse, highlighting ank-G-GFP+-node of Ranvier (noR, green circles) and ank-G-GFP–, but ankyrin-G + and Nav1.6+-noR (blue circles) in excitatory neurons. Magnification of the region demarked by a white box is shown in inverted black & white panels. All noR express ankyrin-G (middle) and Nav1.6 (left), but only those belonging to CaMKII + neurons express the ank-G-GFP construct (right). (B) Cryosection of neocortical white matter from an ank-G-GFP × CaMKIIa-Cre mouse, highlighting ank-G-GFP+-noR (green arrowhead). Ankyrin-G immunoreactivity (blue arrowheads) is seen in both nodes in the image. Caspr is expressed in paranodal regions of both noR (magenta arrowheads). (C) Quantification of the length of noR using the ankyrin-G signal in control (gray) and ank-G-GFP+ neurons (green) in cortical white matter of ank-G-GFP × CaMKIIa-Cre mice shows no difference between the groups (unpaired t-test, n=255 nodes in nine images from three animals). Scale bars A=20 µm, panels in A=2 µm; B=2 µm.

Figure 3—figure supplement 1.. Ank-G-GFP activation and expression in nodes of Ranvier do not alter node morphology.

(A) Top panel shows a cryosection of neocortical white matter from an ankyrin-G-GFP × CaMKIIa-Cre mouse, with GFP + and GFP- nodes of Ranvier (noR; see Figure 3A for details). The bottom panel shows an automated 3D reconstruction of noR using Imaris. Nodes were identified via the ankyrin-G channel, classified by the GFP channel, and properties were analyzed in Nav1.6 and ankyrin-G channels. All steps were automated within the Imaris software. Scale bar = 5 µm. (B) Nodes that are GFP + or GFP- show no differences regarding their length, ellipticity, and median fluorescence intensity of ankyrin-G or Nav1.6 signals (n=141 GFP- nodes, 91 GFP + nodes, one animal, Mann-Whitney U test, graphs use Tukey whiskers and 95% conficence intervals). (C) Top panel: The fluorescence intensity of the sodium channel Nav1.6 did not correlate with ankyrin-G-GFP fluorescence intensity, indicating unchanged levels of sodium channels. Bottom panel: The intensity of GFP fluorescence correlates positively with the levels of all ankyrin-G in GFP + but not GFP- nodes (n as in B; Pearson correlation details in the graph). This demonstrates that ankyrin-G-GFP does not change the Nav1.6 channel fluorescence intensity and provides a reliable predictor of ankyrin-G levels, though we do not exclude the possibility that native unlabeled ankyrin-G remains in the nodes to some degree.

Figure 4.. Ank-G-GFP expression does not affect the molecular axon initial segment (AIS) composition.

(A) Merged image of layer II/III pyramidal neurons in S1 of an ank-G-GFP × CaMKII-Cre mouse with intrinsic ank-G-GFP (green), co-labeled against βIV-spectrin (blue) and Nav1.6 (magenta). Bottom: Inverted black & white image of the ank-G-GFP (left) and Nav1.6 (right) signal in a single axon initial segment (AIS) (arrow in merged image). (B) Merged image of layer II/III pyramidal neurons in S1 of an ank-G-GFP x CaMKII-Cre mouse with intrinsic ank-G-GFP (green), co-labeled against Kv2.1 (magenta) and NeuN (blue). Bottom: Inverted black & white image of the ank-G-GFP (left) and Kv2.1 (right) signal in a single AIS (arrow in merged image). (C) Merged image of a layer V pyramidal neuron in S1 of an ank-G-GFP × CaMKII-Cre mouse with intrinsic ank-G-GFP (green), co-labeled against neurofascin-186 (magenta) and NeuN (blue). Bottom: Inverted black & white image of the ank-G-GFP (left) and NF-186 (right) signal in a single AIS (arrow in merged image). (D) Merged image of layer II/III pyramidal neurons in S1 of an ank-G-GFP × CaMKII-Cre mouse with intrinsic ank-G-GFP (green), co-labeled against TRIM46 (magenta) and ankyrin-G (blue). Bottom: Inverted black & white image of the ank-G-GFP (left) and TRIM46 (right) signals in a single AIS (arrow in merged image). (E) Merged image of layer II/III pyramidal neurons in S1 of an ank-G-GFP × CaMKII-Cre mouse with intrinsic ank-G-GFP (green), co-labeled against FGF14 (magenta) and NeuN (blue). Bottom: Inverted black & white image of the ank-G-GFP (left) and FGF14 (right) signal in a single AIS (arrow in merged image). All scale bars = 10 µm.

Figure 4—figure supplement 1.. Axon initial segment (AIS) length, position, and molecular composition remains intact after ankyrin-G-GFP expression.

(A) Measurements of length and position of AIS in CA1 pyramidal neurons that are either GFP positive (green) or negative (orange). The inclusion of GFP into the ankyrin-G gene does not significantly change AIS length (left panel) or distance to the soma (right panel). AIS measurements were based on the βIV-spectrin signal (n=3 animals). An independent Thy1-GFP control line (gray, n=3 animals) was used as an additional control. The number of individual AIS (white circles) and p-values are given within the graph (t-test for AIS length; Mann-Whitney U-test for AIS distance, graph uses Tukey whiskers). (B) Correlation of AIS lengths measured live via the ankyrin-G-GFP signal in a patch clamp chamber (gray) and post-fixation using antibodies against GFP, ankyrin-G, and βIV-spectrin in CA1 pyramidal neurons (sample preparation as in Figure 7, n=20 AIS, 1 animal). (C) Alternative visualization of data from panel B (post-fixation). Measurements for AIS signals using antibodies against the GFP, ankyrin-G, and βIV-spectrin show comparable lengths. (D) Immunosignals from Nav1.6 channels retain their fluorescence intensity across the AIS after the expression of ankyrin-G-GFP. AIS length was normalized from start to end of the ankyrin-G signal and fluorescence intensity from lowest to highest. Tissue preparation as described in Figure 4A (n=10 GFP +and 10 GFP- AIS). (E) Line plots of Kv2.1 and GFP fluorescence intensities along the AIS signal indicate no change in Kv2.1 expression. Normalization as in panel D. Shaded areas indicate 95% conficence intervals derived from bootstrapping. Tissue preparation as described in Figure 4B (n=10 GFP + and 10 GFP- AIS).

Figure 5.. Ank-G-GFP expression preserves axonal characteristics.

(A) Left: Representative image of a single layer V pyramidal neuron in M1 of an ank-G-GFP × CaMKII-Cre mouse with intrinsic ank-G-GFP (green), and colabeling against ankyrin-G (blue) and synaptopodin (synpo: magenta, arrows). Right: 3D-reconstruction of the axon initial segment (AIS) shown on the left. Three different rotations (0, 90, and 180°) of the same AIS indicate synpo clusters within the confinement of the axonal membrane of the AIS (magenta arrowheads). Scale bar left = 5 µm, right upper and middle panel = 5 µm, bottom panel = 2 µm. Quantification of synpo cluster number in AIS derived from ank-G-GFP × CaMKII-Cre mice showed no significant difference between GFP- and GFP +AIS (n=50 AIS from three animals, p=0.6444, unpaired t-test, whiskers from min to max values). (B) Left: Representative image of three AIS of pyramidal neurons in CA1 of an ank-G-GFP × CaMKII-Cre mouse with intrinsic ank-G-GFP (green), and colabeling against vGAT (blue) and βIV-spectrin (magenta). The only ank-G-GFP+AIS in this image is highlighted by a green arrowhead; ank-G-GFP- AIS are demarked by magenta arrowheads. Right: 3D-reconstruction of the GFP +AIS shown on the left. Three different rotations (0, 90, and 180°) indicate vGAT clusters (green arrowheads) along the AIS where GABAergic synapses innervate the axon. Note that both ank-G-GFP+, as well as ank-G-GFP- AIS are contacted by vGAT puncta (blue). Scale bar left = 10 µm, right upper and middle panel = 5 µm, bottom panel = 2 µm. Quantification of vGAT + puncta along AIS derived from ank-G-GFP × CaMKII-Cre mice showed no significant difference between GFP- and GFP +AIS (n=50 AIS from three animals, p=0.7917, unpaired t-test, whiskers from min to max values). (C) Left: Representative image of two AIS of layer II/III pyramidal neurons in S1 of an ank-G-GFP × CaMKII-Cre mouse with intrinsic ank-G-GFP (green), and colabeling against Iba1 (magenta), a marker for microglia. Right: 3D-reconstruction of the boxed region in the left panel. Two different rotations (0 and 45°) indicate that this AIS is contacted by at least one microglial process. Scale bar = 5 µm.

Figure 6.. The axon initial segment (AIS) nanostructure is maintained after GFP fusion to ankyrin-G.

(A) Left: Representative fluorescence image of dissociated hippocampal neurons in vitro, transduced either with lentivirus expressing recombination-deficient ΔCre fused to nuclear GFP (magenta arrowhead) or with Cre-recombinase fused to nGFP (green arrowhead). Right: STED images of AIS labeled with intrinsic GFP detected by the FluoTag-X4-GFP nanobody (green) and by an antibody against the C-terminus of ankyrin-G (magenta). Scale bars left and middle = 10 µm, right = 1 µm. Note, the nuclear GFP signal is derived from the virus and not the ank-G-GFP label. (B) Representative STED images of the nanoscale organization of the C-terminus of ankyrin-G (left), βVI-spectrin (middle), and Kv1.2 (right) in neurons infected with either ΔCre (top) or Cre virus (bottom panels). AIS shown in (B) (left bottom panel) is the same as the dual color panel A (right panel). Scale bars = 1 µm. (C) Autocorrelation analysis shows the characteristic ~190 nm periodic organization of the proteins imaged in (B). Note how the periodicity of the ank-G-GFP matches the periodicity of the C-terminus of ankyrin-G in all conditions. (D) Autocorrelation amplitude analysis of ankyrin-G and ankyrin-G-GFP (left panel), βIV-spectrin (middle panel), and Kv1.2 (right panel) is not statistically different in cultures treated with ΔCre, +Cre, or untreated (one-way ANOVA). Ank-G-GFP data (green) indicates the signal of the GFP label itself. For +Cre conditions, only GFP +AIS were used for analysis.

Figure 7.. Neuronal excitability is not affected by GFP fusion to ankyrin-G.

(A) Representative images of a ΔCre control neuron (upper panel) and a+Cre neuron (lower panel) from dissociated hippocampal cultures with the patch pipette attached. Scale bar = 25 µm. (B) Analysis of active and passive electrophysiological membrane properties, all indicating no significant changes between experimental and control groups (One-way ANOVA, Holm-Sidak’s multiple comparison test. p-values are indicated in each graph, n=25 ΔCre cells, 24+Cre cells). (C) Representative images of a biocytin-filled (magenta) hippocampal pyramidal neuron from an ank-G-GFP control (upper panel) and an ank-G-GFP × CaMKII-Cre animal (lower panel), colabeled for βIV-spectrin (blue) with intrinsic GFP signal (green). The black & white panels highlight the individual axon initial segment (AIS) in each cell. Note that in the control neuron (upper), no GFP signal for the AIS could be detected (white arrowheads). Scale bar = 20 µm. Representative traces of action potential (AP) trains elicited by current injection (250 ms, –200 to +250 pA) in control (upper panel) and ank-G-GFP × CamKII-Cre neurons (lower panel, green). (D) Analysis of RMP, Rs, AP rising phase, AP threshold, rheobase, and maximum AP number, all indicating no significant changes between experimental and control groups (One-way ANOVA and Tukey’s multiple comparisons test, or Kruskal-Wallis and Dunn’s multiple comparison test (non-parametric). P-values are indicated in each graph, n=28 Ctrl cells, 23 ank-G-GFP cells, 42 ank-G-GFP × CamKII-Cre cells. Whiskers in figure show min to max of values).

Figure 8.. Ank-G-GFP+AIS exhibit structural plasticity in conditions of altered network activity in vitro.

(A) Representative images of ank-G-GFP labeled axon initial segment (AIS) in organotypic cultures (OTCs). Lower left panel: AIS after exposure to 6 mM KCl. Lower right panel: AIS after exposure to 10 mM MgSO4, both for 20 DIV. Scale bar = 10 µm. Experimental procedures are outlined in the cartoon, indicating treatment duration and DIV. (B) Left in graph: AIS elongation after exposure of OTC to 10 mM MgSO4 for 20 d. Mean AIS length was increased by 4.2 µm compared to controls (n=6 OTC, 50 AIS per OTC). Right in graph: AIS remain unchanged after exposure of OTC to 6 mM KCl for 20 d (n=6 OTC, 50 AIS per OTC, one-way ANOVA and Tukey’s test). (C) Under conditions of decreased network activity (10 mM MgSO4 for 10 d), AIS are elongated (unpaired t-test, n=6 OTC, 50 AIS per OTC). Rescue conditions (normal growth medium) from DIV 10–20 resulted in a return to baseline AIS length (unpaired t-test, n=6 OTC, 50 AIS per OTC). (D) Under conditions of increased network activity (6 mM KCl for 10 d), no changes in AIS length were observed (unpaired t-test, n=6 OTC, 50 AIS per OTC). Consequently, no AIS length changes were seen after applying rescue conditions from DIV 10–20 (normal growth medium; unpaired t-test, n=6 OTC, 50 AIS per OTC). (E) Rapid AIS shortening in hippocampal acute slices. The GFP construct was activated by injection of an AAV5-CaMKIIa-Cre virus and acute slices were prepared and maintained in ACSF. KCl (15 mM) was applied to a selected cell body, which showed a rapid decline of GFP fluorescence along the AIS within 10 min of the application (green arrowhead also in inverted images). Neurons that were outside of the KCl application showed no significant length changes (magenta arrowheads, also in inverted images). Change in the relative length of individual AIS targeted by high potassium (green) and surrounding AIS (gray) is plotted against the duration of the application (four pairs of sample and control AIS, t-tests, scale bar = 10 µm. Whiskers in figure show min to max values).

Figure 9.. Live imaging of ank-G-GFP+-AIS in vivo.

Experimental setup: (A) cranial window was implanted in ank-G-GFP × CaMKII-Cre animals (two animals, 3–4-mo-old). Imaging supragranular neurons in S1 layer II/III allowed for stable GFP-signal visualization (right panel, inverted GFP signal). Scale bar = 10 µm. (B) AIS length analysis from the same individual axon initial segment (AIS) 1 wk apart showed no significant differences between the time points and ROI (paired t-test, p=0.1195; 30 AIS). Linear regression analysis indicated a strong correlation between individual AIS length in the same neurons from week 1 (magenta) and week 2 (green; r2=0.7892, p<0.0001, n=30 AIS from one ROI and animal). (C) 2PM revealed a robust and stable GFP signal during the duration of the imaging session. Individual AIS were easily detectable and maintained their overall geometry over the chosen time course (insets with 3D reconstruction). Scale bar overview = 20 µm, close up = 10 µm. Left ROI from week 1 imaging session. Magenta traces along individual AIS demark those subjected to length measurements. Right The same ROI as in the left panel a week later, with the same AIS marked for length analysis (green). Scale bar = 20 µm. (D) Experimental setup: Ank-G-GFP animals were injected contralateral to the imaging site with a retrograde Cre-recombinase and mCherry-expressing AAV resulting in GFP expression at the AIS of mCherry-positive neurons in M1 (caudal forelimb area). Imaging via a cranial window revealed robust GFP signal in infected cells. 3D stacks were produced 150–200 µm from the dura mater; stacks were imaged with 1 µm intervals and 50 images merged into a volume. Scale bar = 20 µm. (E) Experimental setup: Ank-G-GFP animals were injected with a retrograde Cre-recombinase and mCherry-expressing AAV in the medial prefrontal cortex resulting in GFP expression at the AIS of mCherry-positive neurons in the basolateral amygdala. Imaging was performed through a GRIN lens and allowed for visualization of individual mCherry-positive neurons and their respective AIS (green arrowhead). Scale bar = 10 µm.

Author response image 1.

Author response image 2.. Sample traces of action potentials triggered by current injections.