The Microtubule-Associated Protein Tau Mediates the Organization of Microtubules and Their Dynamic Exploration of Actin-Rich Lamellipodia and Filopodia of Cortical Growth Cones

Abstract

Proper organization and dynamics of the actin and microtubule (MT) cytoskeleton are essential for growth cone behaviors during axon growth and guidance. The MT-associated protein tau is known to mediate actin/MT interactions in cell-free systems but the role of tau in regulating cytoskeletal dynamics in living neurons is unknown. We used cultures of cortical neurons from postnatal day (P)0-P2 golden Syrian hamsters (Mesocricetus auratus) of either sex to study the role of tau in the organization and dynamics of the axonal growth cone cytoskeleton. Here, using super resolution microscopy of fixed growth cones, we found that tau colocalizes with MTs and actin filaments and is also located at the interface between actin filament bundles and dynamic MTs in filopodia, suggesting that tau links these two cytoskeletons. Live cell imaging in concert with shRNA tau knockdown revealed that reducing tau expression disrupts MT bundling in the growth cone central domain, misdirects trajectories of MTs in the transition region and prevents single dynamic MTs from extending into growth cone filopodia along actin filament bundles. Rescue experiments with human tau expression restored MT bundling, MT penetration into the growth cone periphery and close MT apposition to actin filaments in filopodia. Importantly, we found that tau knockdown reduced axon outgrowth and growth cone turning in Wnt5a gradients, likely due to disorganized MTs that failed to extend into the peripheral domain and enter filopodia. These results suggest an important role for tau in regulating cytoskeletal organization and dynamics during growth cone behaviors.SIGNIFICANCE STATEMENT Growth cones are the motile tips of growing axons whose guidance behaviors require interaction of the dynamic actin and microtubule cytoskeleton. Tau is a microtubule-associated protein that stabilizes microtubules in neurons and in cell-free systems regulates actin-microtubule interaction. Here, using super resolution microscopy, live-cell imaging, and tau knockdown, we show for the first time in living axonal growth cones that tau is important for microtubule bundling and microtubule exploration of the actin-rich growth cone periphery. Importantly tau knockdown reduced axon outgrowth and growth cone turning, due to disorganized microtubules that fail to enter filopodia and co-align with actin filaments. Understanding normal tau functions will be important for identifying mechanisms of tau in neurodegenerative diseases such as Alzheimer's.

Keywords: growth cone; microtubule; tau.

Figure 1.

Tau colocalizes with actin filaments as well as with stable and dynamic MTs in central, transition and peripheral regions of hamster cortical growth cones. STED microscopic image of a cortical neuron stained for tau (green), tubulin (magenta), and F-actin (blue) showing multiple dendrites and a single long axon tipped by a growth cone (A). Schematic of the growth cone indicates different domains and positions of the actin and MT cytoskeleton. Two examples of fixed cortical growth cones shown in inverted gray scale (B) and three-color overlay images (C). F-actin (blue) is heavily concentrated in filopodia as well as the transition region between the growth cone center and the periphery. Acetylated stable MTs (magenta) form a prominent bundled loop in the growth cone center. Tyrosinated dynamic MTs also form bundled loops from which individual MTs extend into the transition region and enter filopodia. Tau colocalizes with both F-actin (cyan) and stable and dynamic MTs (white) in all three regions of the growth cone. Three-color insets from boxed regions are higher-magnification images of individual filopodia and of the transition regions. In filopodia tau colocalizes with F-actin and is also positioned at the interface (arrowhead) between a single dynamic MT and an actin filament bundle. Note that acetylated MTs do not extend into filopodia. Insets of the transition region show colocalization of tau and F-actin (cyan) and extension of individual dynamic MTs into this region.

Figure 2.

Tau associates with dynamic MTs and actin filaments. Live cell imaging with TIRF shows association of tau with actin filaments and dynamic MT tips. In a growth cone (Movie 1) expressing GFP tau (green) and Tractin-td tomato (magenta) actin and tau colocalize in filopodia (A). Image sequence of the filopodium in boxed area shows consistent association of tau with dynamic actin filaments extending and retracting in the filopodium. A growth cone expressing GFP tau (green) and EB3 tdTomato (magenta; Movie 2; B) showing tau associating with EB3 comets. Image sequence of boxed region shows colocalization of tau and EB3. Scale bar, 5 μm.

Figure 3.

Endogenous hamster tau expression is knocked down with shRNA. Cultured cortical neurons (A) showing nucleofection with pSUPER-mTurquoise (control; left), hamster tau shRNA construct-1 (center), and hamster tau shRNA construct-2 (right). Arrows indicate neurons nucleofected with the control or tau knockdown constructs and stained with tubulin antibodies. Fluorescence intensity measurements of tau (B) showing reduced tau levels in the knockdown neurons. Western blots of hamster cortical neuronal lysates (C) comparing reduction of tau after knockdown with tau shRNA-1 and shRNA-2 constructs, quantified in (E). Western blots of cortical neuronal lysates showing reduced hamster tau after tau shRNA-2 knockdown and overexpression of human 3RS tau (D) and quantified in (E). Western blots of lysates of HEK cells nucleofected with human 3RS tau (F) showing that tau shRNA-1 and shRNA-2 do not affect levels of human tau. Western blots (G) showing specificity of the antibodies to 3RS tau (left), which recognizes endogenous hamster tau in control cell lysates. This antibody also recognizes specifically human GFP-3RS but not 4RL tau nucleofected into cortical neurons with shRNA-2 to reduce endogenous tau. In the Western blot (right) antibodies to 4RL tau recognize human GRP-4RL specifically but not GFP-3RS tau or endogenous hamster 4RL tau, which is not present in P0 cortical neurons. ***p < 0.001. Scale bar, 50 μm.

Figure 4.

Tau knockdown disrupts microtubule bundling, which is restored with human tau expression. P0 hamster cortical neurons were dissociated and nucleofected with a tau sh2 knockdown construct tagged with mTurquoise and replated after 3 DIV (A). In STED images of control growth cones (B, top row) tyrosinated MTs (magenta) are tightly bundled and form a characteristic loop in the central region of pausing growth cones. Antibodies to total tau (green) follow the organization of the bundled MTs. In the merged image tau colocalizes with the bundled MTs (white) and individual dynamic MTs extend into the growth cone periphery labeled with phalloidin to stain actin filaments (blue). In extending growth cones (C, top row) tyrosinated MTs (magenta) are also bundled but extend in straight trajectories and do not form loops. Tau (green) colocalizes (white) with bundled MTs. Following tau knockdown (B, C, middle row) tyrosinated MTs (magenta) in the growth cone center become unbundled and disorganized. Following rescue with tau expression (B, C, bottom row) MTs (magenta) are restored to bundles. Exogenous tau is shown in green. Fluorescence intensities across line scans of the growth cone center in control versus tau knockdown neurons measure the degree of MT bundling (D, E). Paused control growth cones (n = 20 growth cones) show two peaks of fluorescence (blue line) reflecting MT bundles in the left and right sides of the MT loop. In contrast following tau knockdown fluorescence is evenly distributed (magenta line) reflecting the lack of MT bundling (D). Fluorescence measurements of extending growth cones (n = 20) also show bundled MTs in controls versus splayed MTs following tau knockdown (E). In both graphs restoration of MT bundling with tau expression are indicated by the green line. ns - not significant, **p < 0.01, ***p < 0.001. Scale bar, 5 μm.

Figure 5.

Tau knockdown misorients dynamic MT trajectories, which are rescued by human tau expression. Neurons nucleofected with control shRNA or tau shRNA and cotransfected with EB3td tomato to label the dynamic plus ends of MTs at P0 (A) were replated and imaged live with TIRF microscopy (B). Image sequences at 3 s intervals from a control growth cone (B, top row; Movie 3) show relatively straight EB3 comet tracks in parallel arrays. This was confirmed in the maximum projection image (right) where EB3 comets were pseudocolored yellow and in measurements of EB3 comet track angular displacements (rose plot, far right). Following tau knockdown (B, middle row) EB3 comets were disorganized (Movie 3) as shown in the maximum projection image (right) and in rose plots (far right). Image sequence of a growth cone (Movie 3) after tau knockdown and transfection with human tau (B, bottom row) shows that human tau rescues straight EB3 comet trajectories similar to those in control growth cones as shown in the maximum projection image (right) and rose plot (far right). Scatter plots (C) of EB3 comets obtained from movies (10 min duration with 3 s acquisition intervals) of control, tau knockdown and tau rescue growth cones (n = 20 for each condition) show that tau knockdown increases EB3 comet velocity, displacement, and dynamicity (see Materials and Methods) but not comet lifetime. All EB3 comet parameters were rescued with human tau. ns - not significant, *p < 0.05, ***p < 0.001, ****p < 0.0001. Scale bar, 5 μm. Comparisons of dynamic parameters of EB3 comets under different conditions shown in Figure 5-1.

Figure 6.

Tau knockdown prevents MT protrusion into the growth cone periphery. To obtain images of dynamic MTs in relation to actin we labeled fixed growth cones with an antibody to tyrosinated tubulin and with phalloidin, respectively, and imaged growth cones with STED microscopy (A). In a representative control growth cone MTs form a prominent loop in the central region from which individual MTs extend into the transition region. MTs reorient to enter filopodia in apposition to actin filament bundles, as shown by overlap of tubulin (green) and actin (magenta) labeling in filopodia (arrows). Following tau knockdown the MT loop becomes disorganized and individual MTs (green) only penetrate partially into the growth cone transition region containing actin labeling (magenta). These MTs fail to enter filopodia or align with actin filament bundles. Rescue experiments with human tau expression show restoration of MT bundling, MT penetration into the growth cone peripheral domain, and close MT apposition (arrow) to actin filaments in filopodia. Growth cone area (B), density of filopodia in the peripheral domain (C), and MT tips in the transition region (D) are similar between tau knockdown and control growth cones. In tau knockdown growth cones a smaller percentage of MTs in the transition region invaded the peripheral domain (E). Quantification (F, G) of filopodia invaded by MTs showed that after tau knockdown fewer filopodia contained MTs and these MTs were significantly shorter than those entering filopodia in control growth cones. Rescue with human tau restored numbers and lengths of MTs entering filopodia to control levels. ns - not significant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar, 5 μm.

Figure 7.

Tau knockdown disrupts MT organization and dynamic exploration of the growth cone periphery without affecting retrograde flow rate. In growth cones nucleofected with the control vector, the actin marker mRuby-Lifecact (magenta) and EB3-GFP (green; n = 12 growth cones) live-cell imaging over 5 min at 3 s intervals with TIRF (Movie 4) showed that EB3 comets track one another in straight trajectories through all regions of the growth cone and are oriented in the direction of axon outgrowth. The maximum projection images of the movie frames for two controls emphasize the bundled MT arrays in the growth cone center and show EB3 comets extending into the growth cone periphery and invading individual filopodia (A). Image sequences of a single filopodium (boxed region) show an EB3 comet reaching the filopodial tip (arrows). Following tau knockdown (B; n = 12 growth cones), EB3 comets remained in the center of the two growth cones where MTs fail to form bundles. Only a few EB3 comets reached the periphery and in the single filopodium (boxed region) EB3 comets remained in the transition region at the base of the filopodium (arrows). In the tau knockdown growth cones the movements of comets were random; their tracks were disorganized and did not follow one another in the direction of outgrowth (Movie 4). Tau knockdown does not disrupt actin retrograde flow (C, D). Maximum projection images of the same growth cones shown above and in Movie 4 nucleofected with mRuby-Lifeact to label actin filaments were used to create kymographs of retrograde actin flow rates. In filopodia line scans were made from the base to the tip of filopodia (indicated by blue lines in controls and magenta lines after tau knockdown) in which dynamic MTs associate with actin filament bundles. Insets show kymographs of retrograde flow in individual filopodia (n = 45 filopodia from 15 control and 15 tau knockdown growth cones). Scatter plots show that rates of actin retrograde flow (E) are unchanged by tau knockdown. Plots of EB3 comet velocity in relation to retrograde actin flow rates (F) in 45 filopodia show that increased retrograde flow is accompanied by an increase in EB3 comet velocity. In contrast in filopodia after tau knockdown EB3 comet velocity is unrelated to retrograde actin flow rate. Scale bars: growth cones, 5 μm; image series in A and B, 2 μm.

Figure 8.

Tau knockdown reduces axon outgrowth, which is rescued with human tau Fluorescence images of fixed cultured cortical neurons (A) nucleofected with a control GFP-vector, hamster tau shRNA, hamster tau shRNA coexpressed with a single human tau isoform and hamster tau shRNA coexpressed with all 3R human tau isoforms. Neurons were nucleofected, replated, and allowed to extend axons for 24 h. Fixed cultures were labeled with antibodies to tubulin to show cortical axons. The arrow indicates a neuron that contains both the hamster shRNA tau knockdown construct and all 3R human tau isoforms. This neuron has an axon similar in length to a wild-type neuron in close proximity. The arrowhead indicates a neuron containing only the knockdown construct and has very little neurite outgrowth. As shown in the histograms (B), tau knockdown reduced axon length by ∼30%. Rescue with a single 3Rtau isoform only partially rescued axon outgrowth whereas all three human 3R tau isoforms restored axons to control lengths. *p < 0.05, ****p < 0.0001. Scale bar, 50 μm.

Figure 9.

Tau knockdown reduces axon outgrowth and growth cone turning in Wnt5a gradients. Schematic of a Dunn chamber (A) showing a cortical neuron growing in the Wnt5a gradient in the bridge region of the chamber. High to low Wnt5a gradient is indicated by the shading. The initial and final direction of axon outgrowth was defined by the direction of the 5 μm distal axon segment. The initial angle (α) is defined by the original direction of axon outgrowth and a line through the Wnt5a gradient. The final angle (β) was calculated as the angle between the final direction of axon outgrowth and a line through the Wnt5a gradient. Histograms (B) show rates of axon outgrowth of neurons in Dunn chamber Wnt5a gradients over 2 h. Wnt5a increases the rate of axon outgrowth by 20% but following tau knockdown Wnt5a fails to accelerate axon outgrowth. Rescue with a single 3RS human tau construct fails to rescue the rate of axon outgrowth. Histograms of mean turning angles (C) of control versus tau knockdown axons (N = 100 axons for each condition) in BSA and Wnt5a gradients during live-cell imaging for 2 h at 10 min intervals. Growth cones in BSA show equal amounts of attractive and repulsive turning, which is not significantly changed by tau knockdown. In Wnt5a gradients growth cones show slightly more repulsion than attraction to the gradient and tau knockdown significantly reduces turning angles in either direction. Rescue with a single 3RS human tau isoform is sufficient to restore turning angles to control levels. Scatter plots (D) of mean turning angles of individual growth cones measured in the histograms. The x-axis represents the initial angle of the growth cone (α) in relation to the Wnt5a gradient. Mean turning angles (the initial angle minus the final angle) are indicated by dots for each growth cone. Colored lines represent the best-fit linear regression of mean turning angles. *p < 0.05, ***p < 0.001.

Figure 10.

Schematic summary of results. In a control growth cone (left) tau associates with stable bundled MTs in the center and with dynamic MTs extending into the actin-rich periphery. Tau is also located at the interface between dynamic MTs and actin filaments in filopodia. MT organization and dynamic exploration of the periphery mediated by tau allows growth cones to extend and turn. In a growth cone following tau knockdown (right) MTs fail to form bundles in the center and penetrate only partially into the transition region in disorganized trajectories resulting in defects in axon outgrowth and turning behaviors.