Visualising the cytoskeletal machinery in neuronal growth cones using cryo-electron tomography

Abstract

Neurons extend axons to form the complex circuitry of the mature brain. This depends on the coordinated response and continuous remodelling of the microtubule and F-actin networks in the axonal growth cone. Growth cone architecture remains poorly understood at nanoscales. We therefore investigated mouse hippocampal neuron growth cones using cryo-electron tomography to directly visualise their three-dimensional subcellular architecture with molecular detail. Our data showed that the hexagonal arrays of actin bundles that form filopodia penetrate and terminate deep within the growth cone interior. We directly observed the modulation of these and other growth cone actin bundles by alteration of individual F-actin helical structures. Microtubules with blunt, slightly flared or gently curved ends predominated in the growth cone, frequently contained lumenal particles and exhibited lattice defects. Investigation of the effect of absence of doublecortin, a neurodevelopmental cytoskeleton regulator, on growth cone cytoskeleton showed no major anomalies in overall growth cone organisation or in F-actin subpopulations. However, our data suggested that microtubules sustained more structural defects, highlighting the importance of microtubule integrity during growth cone migration.

Keywords: Actin; Cytoskeleton; Doublecortin; Growth cone; Microtubule; Neuron.

Fig. 1.

Analysis of growth cone component distribution. (A) Growth cone distribution of microtubules (MTs) and F-actin segmentation volumes. (B) Large vesicle and organelle (including mitochondria) segmentation volumes. (C) C-domain smooth ER, cyan false colour. (D) Large vesicles, false coloured in cyan. Arrowheads, embedded proteins. (E) C-domain endolysosomal components (left: early/sorting endosome; right: late endosome/multi-vesicular body), false coloured in green. Arrow indicates inward budding site. (F) Large vesicle and organelle segmentation volumes. (G) Small vesicles <100 nm diameter, false coloured in purple; clathrin coats, false coloured in orange. (H) Segmentation volume of ribosomes. Panels C–E and G show longitudinal views in tomograms (4× binned). Absolute segmentation volumes were measured in Chimera (Pettersen et al., 2004), and relative segmentation volume for each feature is expressed as a percentage of the total volume of all these features combined. Axon, n=2; C-domain, n=3; T-zone, n=3; P-domain, n=2; P-domain filopodia, n=2. Solid lines represent mean volumes, dashed lines indicate s.d. Scale bars: C, 100 nm; D,E,G, 60 nm.

Fig. 2.

Ultrastructure of filopodial F-actin arrays. (A) Central tomogram section (left, 4× binned), segmentation (right) of two filopodia. Left: large arrows labelled ‘P’ show position of cell front; smaller arrows show hole edges. Right: arrows indicate filaments that connect bundles; arrowheads indicate merging of bundles. Key indicates false colour scheme. (B) Tomogram sections (4× binned) of a large filopodium showing the (i) central region, (ii) cortex and (iii) cross-section (∼20 nm depth). Cyan dashed lines indicate relationship of panels. Dashed red hexagon indicates a hexagonal F-actin bundle unit. (C) Super-plot (Lord et al., 2020) of maximum width and height of filopodial bundles. Each data point represents a filopodial bundle; n=14 from 4 tomograms, each a different colour and shape; tomogram mean values are plotted in the same colour with larger shapes. Overall width, 11.6±1.7 nm; height, 6.4±1.2 nm; values represent the mean±s.d.; the line represents the overall median. (D) Top left: transverse view through a filopodial bundle (∼20 nm depth), showing hexagonal arrangement, equivalent but in a different bundle compared with panel B; red arrow shows missing-wedge direction. Bottom left: schematic of a hexagonal unit of 7 filaments; blue and green dashed lines indicate short and long inter-filament distances, respectively. Right: super-plot of inter-filament distances in filopodial bundles. Each data point represents a distinct, adjacent filament pair (short axes, n=60 from 4 tomograms; long axis, n=36 from 4 tomograms; each tomogram shown as a different colour; tomogram mean values are plotted in the same colour with larger shapes. Line indicates the overall mean (short axes, 12.1±1.2 nm; long axes, 20.9±1.2 nm; values represent the mean±s.d.). (E) Left: a single crosslinked actin sheet in a filopodial bundle (4× binned) showing ∼37 nm crosslink separation, indicated with arrows. Right: Super-plot of longitudinal crosslink separation in P-domain F-actin bundles. Each data point represents an individual pair of crosslinks along the F-actin longitudinal axis (n=44 from 4 tomograms, each tomogram shown as a different colour). Tomogram mean values are plotted in the same colour with larger shapes. Line indicates the overall mean (37.1±1.9 nm; mean±s.d.). Scale bars: A, 200 nm; B, 100 nm; D, 10 nm.

Fig. 3.

Complexity and modulation of F-actin arrays in growth cone P-domains. (A) Central tomogram section (4× binned, left), segmented rendering (right) of a P-domain region. Left: large black arrows containing a ‘P’, position of cell front; black arrows, carbon hole edges. Right: L, lamellipodial; black arrowheads, branched F-actin networks surrounding F-actin bundles. Key indicates false colouring. (B) (i) Transverse view (2× binned) through filopodial bundle (∼20 nm depth), showing the hexagonal arrangement; red arrow, missing-wedge direction. (ii) Same bundle and view; cyan dashed lines, longitudinal sections of panels. (iii–v) The centre of the transverse section in i and ii is indicated with a green dashed line in panels iii–v. A central filament is indicated, orange ‘0’ in panels ii and iv. Bottom: transverse schematic of filament arrangement, with the central ‘0’ filament in orange. Numbers indicate how longitudinal F-actin repeats shift by integers of 5.5 nm along the filament axis relative to the central filament. (C) (i) Longitudinal and (ii) transverse (∼20 nm depth) views of a large F-actin bundle in a 2x binned P-domain tomogram. Dashed cyan line in i illustrates position of transverse section in ii. ∼37 nm F-actin half helical repeat lengths indicated with yellow dashed arrows, rarer ∼27 nm F-actin half helical repeat lengths indicated with green dashed arrows. A short-repeat filament is false coloured in green and indicated in ii with a green arrowhead. (D) Super-plot of P-domain F-actin half helical repeat lengths. Individual points indicate single measurements along a filament axis, colours indicate different filaments. Mean values for each filament are shown in their respective colours with different larger shapes. Individual lines indicate overall median values. Repeat lengths for long-repeat filaments, 37.2±2.4 nm n=32, 4 filaments (3 tomograms); repeat lengths for short-repeat filaments; 27.3±1.6 nm, n=30, 4 filaments (3 tomograms). Values represent the mean±s.d. (E) Projections through subtomogram averages of long-repeat (left) and short-repeat (right) filaments showing half repeat distances. The sub-tomogram densities were calculated by averaging 1461 and 621 volumes for long-repeat and short-repeat filaments, respectively. Volumes were low-pass filtered to their estimated resolutions at Fourier shell correlation (FSC)=0.5 (Fig. S6). (F) Subtomogram averages of long-repeat filaments (left, mesh density) and short-repeat (right, mesh density) filaments with fitted F-actin (PDB: 7BT7) or F-actin–cofilin (PDB: 3J0S) models, respectively, noting that polarity of filament fitting is arbitrary and cannot be determined from the reconstructions. F-Actin and cofilin models are coloured in grey and green, respectively, with mesh density coloured to match the underlying models (within 10 Å). (G) P-domain tomogram sections (4× binned) illustrating F-actin branching from bundles (top left) and single filaments (top right), indicated with cyan false colouring. Bottom: Super-plot of P-domain branching angles. Each data point represents an individual branching structure (n=12 from 3 tomograms, shown as different colours). Mean values for each tomogram are shown in their respective colours with different larger shapes. Line indicates the overall mean (70.1±2.6°, mean±s.d.). (H) Longitudinal sections of a 2× binned P-domain tomogram through a central region showing radial F-actin bundles in false yellow colour (top) and the corresponding overlying cortical region (bottom), showing F-actin branching points at the cell cortex, indicated with cyan arrows. Scale bars: A, 200 nm; B, 20 nm; G, 40 nm; H, 50 nm.

Fig. 4.

Diverse F-actin arrays in neuronal growth cone T-zone. (A) Central tomogram section (4× binned, left), segmented rendering (right) of T-zone. Left: large black arrows containing a ‘P’, position of cell front; black arrows, carbon hole edges. Key indicates false colouring. (B) Transverse section through radial F-actin bundle in T-zone tomogram (4× binned), showing hexagonal filament organisation; yellow dashed lines indicate ∼12 nm inter-filament distances. (C) Longitudinal section through radial F-actin hexagonal bundle in T-zone tomogram (2× binned), showing 37 nm half helical repeat lengths indicated by arrows. (D) Longitudinal section through tomogram (4× binned) of radial F-actin bundle taper and derived circumferential F-actin bundle, white arrows indicating ∼37 nm F-actin crosslinks. (E) Longitudinal (left) and transverse (centre, ∼20 nm depth) sections, indicated with red dashed lines, of tapering radial F-actin bundle in a T-zone tomogram (4× binned). Traced representation (right) illustrates bundle tapering. White arrows indicate 37 nm inter-crosslink spacing. (F) (i) Longitudinal and (ii) transverse views of tapering radial F-actin bundle in T-zone tomogram (2× binned). In i, distal and proximal are top and bottom, respectively. Dashed cyan line in i indicates the position of the transverse section in ii. ∼37 nm and ∼27 nm half helical repeat lengths are indicated with yellow and green dashed arrows, respectively. Short-repeat filament lengths are false-coloured in green and their transverse positions within the bundle indicated in ii with green arrows. (G) Longitudinal tomogram sections (2× binned) of radial F-actin bundle taper (left) and F-actin bundle-derived dislocated single filament (right), dashed arrows indicate long F-actin half helical repeat lengths. (H) Longitudinal sections of T-zone tomogram (2× binned, left), central region showing radial and circumferential F-actin bundles in false yellow and orange colours, respectively; corresponding overlying cortical region (right), showing F-actin branching points at cell cortex. Branching points are illustrated with cyan arrows. Blue arrows in A, D, E indicate points where filaments bend away from tapering radial F-actin bundles. Scale bars: A, 200 nm; D,E, 50 nm, F–H, 40 nm.

Fig. 5.

Organisation and architectures of microtubules within growth cones. (A) Representative examples of MTs illustrating protofilament number and polarity. Each set of 3 images includes a ∼30 nm thick longitudinal section through MT volume (2× binned, left) with corresponding image Fourier filtered at the origin (right, red box) showing 13-protofilament moiré patterns. Below, corresponding MT rotational average of 30 nm thick section viewed towards the cell periphery, showing protofilament number and handedness (curved red arrow). When viewed from the minus end or plus end, rotational average images exhibit clockwise or anticlockwise slew, respectively. In longitudinal sections, growth cone periphery is towards top; plus and minus end directions are indicated (red ‘+’ and ‘–’); consensus protofilament (pf) architecture is indicated between dashed red lines. (B) MT polarity relative to neuron periphery in individual tomograms. MTs assigned ‘N/A’ were perpendicular to the axon axis or bent ∼180°. Axon, 59 total plus-end peripheral MTs, 0 minus-end peripheral from 3 tomograms (each a different cell); C-domain, 69 total plus-end peripheral MTs, 6 minus-end peripheral, 2 N/A from 4 tomograms (each a different cell); T-zone, 7 total plus-end peripheral MTs, 3 minus-end peripheral, 1 N/A from 3 tomograms (each a different cell); P-domain (PD), 2 total plus-end peripheral MTs, 0 minus-end peripheral, 0 N/A from 1 tomogram. (C) MT ends frequency per 1 µm MT length in individual tomograms. Each data point represents a separate tomogram; axon (n=4), C-domain (n=7), T-zone (n=4), P-domain (n=1). Line indicates mean from all tomograms for each region. 50 MT ends were found in a total of 318 μm MT length. (D) Number of short and tapered ends in individual tomograms (number of tomograms of axon, n=4; C-domain, n=7, 1 tomogram contained no ends; T-zone n=4; P-domain, n=1). (E) Longitudinal views of short MT ends. (F) Longitudinal views of tapered MT ends. Short ends – 10 nm thick slices through 4× binned tomograms; tapered ends – 30 nm thick slices through 2× binned tomograms, false coloured in magenta and orange, respectively. Right: traced representations. Plus and minus ends are indicated (red ‘+’ and ‘–’). (G) Four longitudinal sections through regions of a single MT associated in parallel with P-domain F-actin bundles. Black and white arrows indicate crosslinks between MT shaft or tip and F-actin bundles, respectively; cyan indicates ER. (H) Two 3D sections (left and right) through T-zone segmentations; black arrows indicate transverse F-actin bundles running perpendicular on dorsal and ventral surfaces of periphery-orientated MTs. Segmentation colouring is as in Fig. 4A. Scale bars: E–G, 50 nm; H, 100 nm.

Fig. 6.

Lumenal particles in neuronal growth cone MTs. (A) Longitudinal views (2× binned) of MT lumenal particles. Orange dashed boxes indicate regions where particles are every ∼8 nm. White arrows, larger ring-like particles ∼7–10 nm diameter, black arrows, smaller particles ∼3–5 nm in diameter. (B) Super-plot of lumenal particle frequency per 8.2 nm. Each small data point represents a separate MT coloured by tomogram, mean frequency for individual tomograms are indicated with large coloured shapes, and lines indicate the overall medians. Overall means: Axon, 0.43±0.07 nm, n=94 MTs from 4 tomograms; C-domain, 0.38±0.09 nm, n=134 MTs from 7 tomograms; T-zone, 0.26±0.16 nm, n=22 MTs from 4 tomograms; P-domain, 0.19±0.02 nm, n=2 MTs from 1 tomogram. Values represent the mean±s.d. (C) Representative 3D segmentations of C-domain MTs (magenta) and lumenal particles (LP, black) illustrating particle frequency variability. (D) MT as in panel A, with transverse sections (∼5 nm depth) at positions indicated by dashed coloured lines. White arrows, larger ring-like particles ∼7–10 nm diameter; black arrows, smaller particles ∼3–5 nm diameter. (E) Super-plot of lumenal particle diameters. Each small data point represents a separate lumenal particle coloured by tomogram. Different small data point shapes indicate different MTs within each tomogram. Large coloured shapes indicate mean particle diameters within individual tomograms, lines indicate overall medians. Overall means: Axon, 7.6±1.7 nm, n=122 from 4 tomograms; C-domain, 7.9±1.7 nm, n=217 from 7 tomograms; T-zone, 7.9±1.6 nm, n=97 from 4 tomograms; P-domain, 8.0±1.3 nm, n=21 from 1 tomogram. Values represent the mean±s.d. Scale bars: A,D, 50 nm; C, 100 nm.

Fig. 7.

MT lattice defects and discontinuities. (A,B) Longitudinal views of large (>16 nm long; 30 nm slice, 2× binned) (A) and small (<16 nm long; 10 nm slice, 4× binned) (B) MT lattice defects in growth cones, false coloured in magenta and orange, respectively; traced representation shown alongside for clarity. Blue boxed insert in panel B, 2× binned (5 nm thick) version of the blue dashed boxed region above, with magenta and orange arrows pointing to tubulin monomers in lattice and a defect, respectively. (C) Absolute number of small (<16 nm) and large (>16 nm) defects (number of tomograms of axon, 4; C-domain, 7; T-zone, 4; P-domain, 1). (D) Centre, raw ∼30 nm thick longitudinal MT slice including an MT defect (orange arrowhead). Right: Fourier filtered (at origin, FF) images of blue dashed boxed regions in centre panel showing 13-protofilament (pf) moiré patterns either side of defect. Left: raw image and rotational average (RAv) of ∼10 nm thick MT transverse section, indicated with a cyan dashed line, indicating 13-protofilaments in both cases. (E) Frequency of defects per 1 µm MT length. Each data point represents a separate tomogram; axon, 4, C-domain, 7, T-zone, 4, P-domain, 1. Scale bars: A,B,D, 50 nm.

Fig. 8.

Increased MT defects in doublecortin knockout neuronal growth cones. (A) Representative images of Dcx KO neuron MTs showing protofilament number and polarity, as described in Fig. 7A. The arrow labelled ‘P’ indicates position of the cell front. (B) Transverse views (viewed from plus end) of sub-tomogram averages of random 13-protofilament MT subsets in WT (top, 1200 MT segments, 13 MTs) and Dcx KO (bottom, 1190 MT segments, 20 MTs) neurons. Volumes were low-pass filtered to estimated resolutions (Fig. S7C). (C) MT polarity relative to neuron periphery in individual tomograms. MTs assigned ‘N/A’ were either perpendicular to the axon axis or bent ∼180° such that both minus and plus directions were orientated peripherally. Axon, 45 plus-end peripheral MTs, 0 minus-end from 3 tomograms (3 axons); C-domain, 23 plus-end peripheral MTs, 10 minus-end, 1 N/A from 3 tomograms (3 neurons). (D) Super-plot of lumenal particle frequency per 8.2 nm. Each small data point represents a separate MT coloured by tomogram, mean frequency for individual tomograms indicated with large coloured shapes, lines indicate the overall medians. Overall means: WT axon, 0.43±0.07 nm, n=94 MTs, 4 tomograms; C-domain, 0.38±0.09 nm, n=134 MTs, 7 tomograms. Dcx KO axon, 0.43±0.06 nm, n=78 MTs, 4 tomograms; Dcx KO C-domain, 0.37±0.07 nm, n=84 MTs, 5 tomograms. Values represent the mean±s.d. (E) Longitudinal views (4× binned) of short and tapered MT plus and minus ends in Dcx KO growth cones. (F) Absolute numbers of short, tapered and capped MT ends in Dcx KO grow cones. Dataset size, total individual MT number in multiple tomograms (5 Dcx KO axon tomograms, 2 had no ends; 4 Dcx KO C-domain tomograms). (G) Longitudinal views (4× binned) of example MT defects in Dcx KO growth cones. (H) Frequency of defects per 1 µm MT length in WT and Dcx KO neurons. Each data point represents a separate tomogram; WT axon, 4; WT C-domain, 7; Dcx KO axon, 3; Dcx KO C-domain, 4. Mann–Whitney tests; WT C-domain versus Dcx KO C-domain, *P<0.05; WT C-domain versus Dcx KO C-domain with outlier data point removed, P=0.1083 (not significant); WT axon versus Dcx KO axon, P=0.89 (ns, not significant). (I) Numbers of large and small MT lattice defects as a percentage of total MT lattice defects in WT (top) and Dcx KO (bottom) pre-cone axons and C-domains. Dataset sizes indicate the total number of defects. Panels E and G: MTs and lattice defects are false coloured in magenta and orange, respectively; traced representation (right and bottom) shown for clarity. Scale bars: E,G, 50 nm.