The high molecular weight neurofilament subunit plays an essential role in axonal outgrowth and stabilization

Abstract

Neurofilaments (NFs) are thought to provide structural support to mature axons via crosslinking of cytoskeletal elements mediated by the C-terminal region of the high molecular weight NF subunit (NF-H). Herein, we inhibited NF-H expression in differentiating mouse NB2a/d1 cells with shRNA directed against murine NF-H without affecting other NF subunits, microtubules or actin. shRNA-mediated NF-H knockdown not only in compromised of late-stage axonal neurite stabilization but also compromised early stages of axonal neurite elongation. Expression of exogenous rat NF-H was able to compensate for knockdown of endogenous NF-H and restored the development and stabilization of axonal neurites. This rescue was prevented by simultaneous treatment with shRNA that inhibited both rat and murine NF-H, or by expression of exogenous rat NF-H lacking the C-terminal sidearm during knockdown of endogenous NF-H. Demonstration of a role for NF-H in the early stages of axonal elaboration suggests that axonal stabilization is not delayed until synaptogenesis, but rather that the developing axon undergoes sequential NF-H-mediated stabilization along its length in a proximal-distal manner, which supports continued pathfinding in distal, unstabilized regions.

Keywords: Axonal outgrowth; Axonal stability; Cytoskeleton; Nervous system development; Neurofilament.

Fig. 1.. Reduction of NF-H synthesis and steady-state levels by shRNA treatment.

(A) Autoradiographic and corresponding immunoblot analyses of material immunoprecipitated from Triton-soluble (“Sol”) and Triton-insoluble (“Insol”) fractions of undifferentiated cells incubated for 2 hr with ³⁵S-methionine 48 hr after transfection with a cocktail of CDS1-4. The accompanying graph presents densitometric analyses of the mean (± standard error) of 200 kDa+180 kDa material in duplicate autoradiographs. NF-H synthesis and steady-state levels were significantly reduced by treatment with CDS1-4 (p<0.001 and p<0.01 for synthesis and steady-state levels, respectively; ANOVA). Additional cells transfected with CDS1-4 48 hr prior to dbcAMP-mediated differentiation were subjected to immunoblot analysis with multiple antibodies (B) or double-label immunofluorescent analyses with anti-H and anti-M (C) as indicated. The accompanying graphs present the ratio of total NF-H/actin (B) and relative intensity ratio of NF-H/NF-M as indicated (C); steady-state levels of NF-H were significantly reduced by shRNA treatment as evidenced by both immunoblot and immunofluorescent analyses (p<0.01 for both; Student's t test; panels B and C, respectively). Bar: 25 µm.

Fig. 2.. shRNA-mediated NF-H knockdown compromises axonal outgrowth and stabilization.

(A) Phase-contrast images of representative fields of cells differentiated for 3 days without and with transfection with a cocktail of CDS1-4. The accompanying graphs present the mean (± standard error) the percentage of cells with axonal neurites and the length of these neurites as a function of respective somal diameters. shRNA treatment significantly reduced the percentage of cells with neurites and the length of existing neurites (p<0.01 for both versus untreated cultures; Student's t test). (B) Representative phase-contrast images of cells treated with CDS1-4 for 48 hr then differentiated for 7 days, after which alternate cultures received 1 µM colchicine for 2 hr. Colchicine treatment at this time reduces axonal neurite caliber but does not induce neurite retraction. By contrast, transfection with CDS1-4 during differentiation compromised the development of colchicine resistance. Bar: 25 µm. The accompanying graphs present the percentage of cells with axonal neurites, and the length of these neurites in respective somal diameters, following colchicine treatment without or with CDS1-4 transfection prior to differentiation (day 0) or for the final 24 hr of differentiation (day 6). Note that CDS1-4 reduced the percentage of cells with axonal neurites that resisted colchicine treatment, as well as the length of surviving neurites, when administered prior to (day 0) or following (day 6) differentiation (p<0.01 for both versus untreated cultures; Student's t test).

Fig. 3.. Co-transfection with GFP-H and shRNA.

(A) Replicate sets nitrocellulose replicas of homogenates of cells transfected with GFP-H with and without co-transfection with UTR-1 or CDS-4 and probed with anti-GFP, anti-NF-H antibody and anti-tubulin antibody DM1A (as a loading control) as indicated. The accompanying graphs present densitometric analyses of these replicas. Co-transfection with UTR-1 did not reduce GFP-H levels but reduced endogenous NF-H as evidenced by a reduction in total NF-H immunoreactivity. Co-transfection with CDS-4 reduced both GFP-H and total NF-H immunoreactivity (asterisks indicate p<0.01 versus cultures not receiving shRNA; ANOVA with Fischer's post-hoc analyses). (B) Representative fluorescent images of cells transfected as in panel A and reacted with anti-NF-H followed by Texas-Red conjugated secondary antibody. Bar: 25 µm. The accompanying graphs present densitometric analyses of 20–100 cells under each condition from multiple fields. As in immunoblot analyses, co-transfection with UTR-1 did not reduce GFP-H levels but reduced total NF-H immunoreactivity. Co-transfection with CDS-4 reduced both GFP-H and total NF-H immunoreactivity (asterisks indicate p<0.005 versus cultures not receiving shRNA; ANOVA with Fischer's post-hoc analyses).

Fig. 4.. Exogenous NF-H prevents the shRNA-mediated reduction in axonal length.

(A) Representative phase-contrast micrographs of cells were transfected with GFP-H without and with co-transfection with a shRNA corresponding to the untranslated region of mouse NF-H (UTR-1) or a coding sequence shared by both rat and mouse NF-H (CDS-4; Table 1). The accompanying graph presents quantification of neurite length as a function of respective somal diameters for nontransfected cells and cells transfected with GFP-H without and with co-transfection with UTR-1 or CDS-4 as indicated. Transfection with UTR-1 or CDS each reduced neurite length; cotransfection with GFP-H curtailed the reduction in neurite length mediated by UTR-1 but not by CDS-4 (asterisk indicate p<0.001, ANOVA with Fischer's post-hoc analyses). (B) Representative immunofluorescent images of total NF-H cells transfected as in panel A. The accompanying graph presents quantification of total NF-H within axonal neurites. A single asterisk indicates an increase (p<0.04) and the double asterisk indicates a decrease (p<0.04) in total axonal NF-H versus levels in axons of nontransfected cells; Student's t test). Bar: 25 µm.

Fig. 5.. The C-terminal sidearm mediates rescue of axonal neurites by exogenous NF-H.

(A) Representative epifluorescent images of differentiated cells and immunoblot analysis of Triton-soluble (S) and Triton-insoluble (I) fractions derived from cells transfected 48 hr previously with GFP-H or GFP-Htrunc as indicated. Inserts present a higher-mag of axonal NFs. Note that GFP-Htrunc is observed within filamentous profiles within axonal neurites and is recovered within the Triton-insoluble cytoskeleton. Nitrocellulose replicas were probed with anti-GFP and the NF C-terminal-specific antibody SMI-31. As anticipated, GFP-Htrunc lacks immunoreactivity with SMI-31 due to absence of the C-terminal sidearm. (B) Representative immunoflourescent images of NF-H in differentiated cells without transfection or co-transfected with UTR-1 and either GFP-H or GFP-Htrunc as indicated. (C) Quantification of the percentage of cells with axonal neurites without transfection or following transfection with UTR-1 without or with co-transfection with GFP-H or GFP-Htrunc. UTR-1 significantly reduced the number of cells with axonal neurites; as above (Fig. 4), co-transfection with GFP-H prevented this reduction, while GFP-Htrunc was unable to prevent this reduction (p<0.01 for both UTR-1 vs untransfected cells and UTR-1 + GFP-Htrunc vs untransfected cells; ANOVA with Fischer's post-hoc analyses). Bar: 25 µm.

Fig. 6.. Model for axonal pathfinding and stabilization.

In the conventional “elongation then stabilization” model, the axon stabilizes only after reaching its target. Based on our findings herein, we propose a “sequential stabilization” model, where more proximal regions of the growing axon progressively stabilize, and therefore physically support, continued distal pathfinding.