Mitogen-activated protein kinase (MAP) kinase plays a pivotal role in the development of the nervous system by mediating both neurogenesis and neuronal differentiation. Here we examined whether p42/44 MAP kinase plays a role in axonal transport and the organization of neurofilaments (NFs) in axonal neurites. Dominant-negative p42/44 MAP kinase, anti-MAP kinase antisense oligonucleotides and the MAP kinase inhibitor PD98059 all reduced NF phospho-epitopes and inhibited anterograde NF axonal transport of GFP-tagged NF subunits in differentiated NB2a/d1 neuroblastoma cells. Expression of constitutively active MAP kinase and intracellular delivery of active enzyme increased NF phospho-epitopes and increased NF axonal transport. Longer treatment with PD98059 shifted NF transport from anterograde to retrograde. PD98059 did not inhibit overall axonal transport nor compromise overall axonal architecture or composition. The p38 MAP kinase inhibitor SB202190 did not inhibit NF transport whereas the kinase inhibitor olomoucine inhibited both NF and mitochondrial transport. Axonal transport of NFs containing NF-H whose C-terminal region was mutated to mimic extensive phosphorylation was substantially less affected by PD98059 compared to a wild-type construct. These data suggest that p42/44 MAP kinase regulates NF anterograde transport by NF C-terminal phosphorylation. MAP kinase may therefore stabilize developing axons by promoting the accumulation of NFs within growing axonal neurites.
- Mitogen-activated protein kinase
- Axonal transport
- Axonal maturation
- Signal transduction
Mammalian neurofilaments (NFs), a major constituent of the axonal cytoskeleton, are 10 nm filaments composed of three polypeptide subunits, termed NF-H, NF-M and NF-L (for high, medium and low with respect to their molecular mass). NF subunits are among the most highly phosphorylated proteins in the nervous system (Hoffman and Lasek, 1975; Julien and Mushynski, 1983; Hoffman et al., 1984; Nixon et al., 1987). The C-terminal regions, or `sidearms' of the subunits differ markedly from each other. Extensive phosphorylation of NF-H and NF-M C-terminal sidearms (that extend away from the filament backbone) initiates within cell bodies and continues during axonal transport, resulting in segregation of extensively phosphorylated NFs within axons. In contrast, hypophosphorylated NFs are largely confined to perikarya (for review, see Pant and Veeranna, 1995; Miller et al., 2002). Phosphorylation has long been considered to regulate several aspects of NF dynamics, including rendering subunits more resistant to proteolysis (Pant, 1988) increasing NF spacing and axonal caliber (Hirokawa et al., 1984; Nixon et al., 1994; Sanchez et al., 2000), dissociating NFs from the anterograde motor kinesin (Yabe et al., 1999), slowing of axonal transport rate (Jung and Shea, 1999; Jung and Shea, 2004; Jung et al., 2000a; Jung et al., 2000b; Lewis and Nixon, 1998; Zhu et al., 1998) and promoting NF-NF interactions leading to bundling within axons (Gou et al., 1998; Leterrier et al., 1996; Yabe et al., 2001a; Yabe et al., 2001b). However, the extent to which C-terminal NF phosphorylation contributes to some of these phenomena continues to be debated (Ackerley et al., 2003; Rao et al., 2002; Shea et al., 2003).
Kinases phosphorylating the C-terminal regions of NF-H and/or NF-M include glycogen synthase kinase (GSK)-3α and 3β, cyclin-dependent kinase-5 (cdk5), mitogen-activated protein (MAP) kinase, casein kinase I and II and stress-activated protein (SAP) kinases (Brownlees et al., 2000; Giasson and Mushynski, 1996; Giasson and Mushynski, 1997; Guidato et al., 1996; Li et al., 1999; O'Ferrall et al., 2000; Sharma et al., 1999; Sun et al., 1996; Veeranna et al., 1998) (for reviews, see Pant and Veeranna, 1995; Julien and Mushynski, 1998). These kinases phosphorylate NFs at sites distinct from each other (Guidato et al., 1996; Sun et al., 1996; Veeranna et al., 1998), leaving open the possibility that they could mediate separate aspects of NF dynamics.
MAP kinases consist of a family of kinases with diverse effects on cellular proliferation, differentiation and degeneration (Cobb, 1999; Johnson and Lapadat, 2002). One member of this family, p42/44 MAP kinase, mediates multiple aspects of neuronal development including the elaboration of axons and dendrites (Abe et al., 2001; Obara et al., 2002; Valliant et al., 2002) and regulates axonal transport of some vesicular elements (Kholodenko, 2002). As the accumulation of phosphorylated NFs within axons is thought to represent a critical stage in axonal stabilization and maturation (Nixon and Shea, 1992), we examined here whether p42/44 MAP kinase played a role in axonal transport and organization of NFs in axonal neurites.
Materials and Methods
NB2a/d1 cells were cultured in DMEM (Sigma, St Louis, MO) containing 10% fetal bovine serum (Sigma) on dishes with a 1 mm hole drilled in the center and a coverslip sealed beneath the hole. To induce outgrowth of axonal neurites, cultures received 1 mM dibutyryl cyclic AMP (dbcAMP; Sigma) for 3 days (Jung et al., 1998; Yabe et al., 2001a; Yabe et al., 2001b).
Cells treated with dbcAMP for 2 days were transfected with a construct that encodes rat NF-M fused with green fluorescent protein (GFP-M) (Yabe et al., 1999) or with constructs (generous gifts of Dr Chris Miller, Inst. Psychiatry, King's College, London) encoding normal NF-H conjugated to GFP (GFP-Hwt) and the identical NF-H construct in which serines in selected KSP domains had been mutated to aspartates (`GFP-Hasp') to mimic a permanently phosphorylated state (Ackerley et al., 2003). Additional neurons were co-transfected with a mixture of GFP-M and constructs encoding constitutively active and dominant-negative p42/44 MAP kinase (Li et al., 1999), or with a construct expressing human tau 40 conjugated to enhanced cyan fluorescent protein (eCFP-htau40; generous gift of Dr E. Mandelkow) (Stamer et al., 2002). Transfection was carried out using lipofectamine according to the manufacturer's instructions (Sigma). Briefly, constructs (2 μg) were resuspended in 100 μl serum-free DMEM, then combined with 8 μl lipofectamine that had been resuspended in 100 μl serum-free DMEM and allowed to equilibrate for 45 minutes at room temperature. An additional 800 μl of serum-free DMEM was added and this mixture was then added to cultures. After incubating cultures for 3.5 hours an additional 1 ml DMEM containing 10% fetal bovine serum (FBS) was added (without removing the DNA:lipofectamine mixture) and incubation continued for 24 hours, after which the medium was replaced with 2 ml fresh DMEM containing 10% FBS. Cultures were first viewed at this time, because prior studies (Yabe et al., 1999; Yabe et al., 2001a; Yabe et al., 2001b) demonstrated that 18-24 hours were required for transfected cells to accumulate sufficient GFP-M for visualization. Accordingly, transfected cells had been treated with dbcAMP for 3 days prior to viewing, at which time they have elaborated axonal neurites containing phosphorylated NFs (Shea et al., 1988; Shea et al., 1990; Shea and Beermann, 1994). Successfully transfected cells were localized by GFP fluorescence. The effectiveness of MAP kinase variants expressed following transfection with either of the above MAP kinase constructs was confirmed by immunoblot analysis of phospho-NFs (see Results).
For co-transfection, 2 μl each of constructs encoding GFP-M and one of the MAP kinases were combined in the initial mixture described above. Transfection rates under these conditions routinely exceed 80%. In co-transfection experiments utilizing GFP-M and MAP kinase vectors, we did not attempt to confirm whether or not individual neurons were transfected with both vectors. Rather, we scored all neurons expressing GFP-M in cultures that had been transfected with GFP-M alone, as well as in cultures transfected with a mixture of GFP-M and MAP kinase. It remains possible that some of the neurons in cultures exposed to a mixture of vectors were in fact not transfected with, or did not express, MAP kinase. As we only scored neurons expressing GFP-M, we nevertheless remain confident of our results, as scoring neurons expressing GFP-M but not MAP kinase would cause us to underestimate the impact of alteration of MAP kinase activity on the distribution of GFP-M. Our transfection results presented here, although statistically significant, therefore are likely to underestimate the full effect of MAP kinase on NF axonal transport and organization. GFP-labeled structures exist in multiple forms in both NB2a/d1 cells and primary neurons, including those that are apparently filamentous, as well as punctate structures (e.g. Wang and Brown, 2001; Yabe et al., 1999; Yabe et al., 2001a; Yabe et al., 2001b), which in some instances may represent coiled filaments (Wang and Brown, 2001). In images where all or most of the GFP-labeled profiles are not clearly filamentous, we refer to them as `NF subunits' solely to avoid arbitrarily designating assembly into a filamentous form. This is not necessarily intended to indicate transport as non-filamentous oligomers or individual subunits (e.g. Yabe et al., 1999).
Antisense and sense oligonucleotide treatment
Cells treated with dbcAMP for 2 days were incubated for the final 24 hours of dbcAMP treatment with 10 μM oligonucleotides corresponding to the sequence for p42/44 MAP kinase in sense and antisense orientation. These oligonucleotides have previously been demonstrated to decrease MAP kinase levels and activity in cultured neuroblastoma (Ekinci et al., 1999). Oligonucleotides were added to GFP-M transfected cells immediately after transfection. Efficacy of antisense oligonucleotides was confirmed here by decreased MAP kinase levels in immunoblot analyses and alteration of NF phosphorylation.
Intracellular delivery of MAP kinase
Purified active MAP kinase (UBI, Lake Placid, NY) was delivered via the commercial Provectin delivery system (Imgenex, San Diego, CA) into cells that had been differentiated for 2 days with dbcAMP, and MAP kinase levels were examined 24 hours later (i.e., after a total of 3 days of dbcAMP treatment). For cells transfected with GFP-M, Provectin-mediated intracellular delivery of MAP kinase was carried out immediately after transfection. MAP kinase (final concentration 1 mg/ml) was mixed with the resuspended Provectin reagent in HEPES buffer for 5 minutes then added to cultures for 4 hours, after which medium was changed according to the manufacturer's instructions and cultures were viewed under phase-contrast and UV optics. In additional experiments, MAP kinase was heat-inactivated by boiling for 5 minutes prior to mixing with Provectin. The efficacy of Provectin-mediated intracellular delivery of protein was confirmed by intracellular delivery of β-galactosidase followed by histochemical analyses for β-galactosidase activity (provided as a positive control by the manufacturer). Efficacy of Provectin-mediated delivery of active MAP kinase was also confirmed by increased MAP kinase in immunoblot analyses and alteration of NF phosphorylation (see Results). As with double transfections, we did not ascertain whether individual cells expressed GFP-M and received detectable MAP kinase following sequential transfection and Provectin treatment; however, the high rates of transfection and intracellular delivery via Provectin suggest that most cells received both GFP-M and MAP kinase. Moreover, this is supported by our results (below), which demonstrate clear differences in transport of GFP-M following intracellular delivery of active MAP kinase.
Treatment with pharmacological inhibitors
Cultures were treated with 10 μM PD98059, an inhibitor of MEK, the upstream activator of p42/44 MAP kinase and 10 μM SB202190 (each from Calbiochem, La Jolla, CA), which inhibits p38 MAP kinase (Alessi et al., 1995; Cuenda et al., 1995; Manthey et al., 1998; Pang et al., 1995). As an additional control, some cultures were treated with 5 μM olomoucine (Sigma), which is reported to inhibit anterograde axonal transport (Ratner et al., 1998). Inhibitors were added for the final 2 hours of incubation following transfection.
Immunoblot analysis and immunofluorescence
For immunoblot analysis, cultures were rinsed with Tris-buffered saline (TBS) (pH 7.4), then homogenized in 1% Triton in 0.5M Tris (pH 6.8) containing 5 mM EDTA, 1 mM PMSF and 50 μg/ml leupeptin (Jung et al., 1998). Samples received an equal volume of 2× Laemmli treatment buffer, were boiled for 1 minute and electrophoresed on linear 7% polyacrylamide SDS gels. Separated proteins were transferred to nitrocellulose and were probed as described (Shea et al., 1997) with 1:1000 dilutions of monoclonal antibodies (SMI-31, RT97) directed against phospho-epitopes of NF-H and NF-M, and a polyclonal antibody (L3) generated against electrophoretically purified NF-L, followed by goat anti-mouse IgG conjugated to alkaline phosphatase as described previously (Jung et al., 1998). [SM1-31 was obtained from Sternberger Meyer Monoclonals, Lutherville, MD; RT97 was a generous gift of B. Anderton, Institute of Psychiatry, London, UK; anti-NF-L was raised in this laboratory (see Jung et al., 1998); anti-MAP kinase was obtained from UBI, Lake Placid, NY.]
For immunofluorescence, cultures were fixed with 4% paraformaldehyde in TBS for 5 minutes at room temperature. The cultures were then rinsed twice for 5 minutes in TBS, blocked for 30 minutes in TBS containing 1% bovine serum albumin (BSA) and 2% normal goat serum and incubated overnight at 4°C in TBS containing 1% BSA and 1:100 dilutions of RT97, L3, a polyclonal antibody directed against beta-tubulin (Sigma) and monoclonal antibodies directed against phosphorylated (PHF-1) (Greenburg and Davies, 1990), nonphosphorylated (Tau-1) (Kosik et al., 1989) and phospho-independent (5E2) (Joachim et al., 1987) epitopes of tau. The following morning, cultures were rinsed three times with TBS, re-blocked for 30 minutes, rinsed with TBS, then incubated for 30 minutes at 37°C in TBS containing 1% BSA and a 1:150 dilution of Texas Red conjugated goat anti-rabbit IgG. Cells were rinsed three times with TBS and stored at 4°C. Filamentous actin was visualized by staining with rhodamine-phalloidin (Molecular Probes, Inc.). Cultures were rinsed twice with PBS and fixed as described above. Samples were rinsed again with PBS, extracted for 5 minutes with -20°C acetone and air-dried. Rhodamine-phalloidin was dissolved in methanol, evaporated to dryness, then re-dissolved in 400 μl PBS according to the manufacturer's instructions. Cultures were incubated with this solution for 20 minutes at 25°C, rinsed twice with PBS and stored in PBS at 4°C in the dark until visualization. Additional transfected cultures and cultures treated with oligonucleotides and pharmacological inhibitors were stained without fixation with 40 mM Mitotracker (Molecular Probes) to monitor transport of vesicular elements (e.g. Trinzcek et al., 1999).
Image acquisition and densitometric analysis
Epifluorescent and corresponding phase-contrast images were captured with a DAGE CCL-72 camera via NIH Image software (available as freeware from http://rsb.info.nih.gov/nih-image) using a Scion LG-3 frame grabber or via a Photometrics Coolsnap camera operated by OpenLab software (Improvision). Images were stored as TIFF or PICT files and analyzed using NIH Image software as described (Jung et al., 1998). Briefly, densitometric analyses were carried out by encircling cells or axonal neurites with the freehand selection tool. Images were inverted and background was subtracted using the automated function. For calculation of relative density along axonal neurites, individual axonal neurites were divided into thirds. Axonal neurites were scored as possessing filamentous GFP within their distal-most third if both a signal greater than 10% above background was obtained and GFP was predominantly associated with the central bundle (Yabe et al., 2001a; Yabe et al., 2001b). Relative density of immunoreactive proteins in immunoblot analyses was also determined by encircling bands with the freehand tool on digitized images of immunoblots. Values obtained from multiple cells and blots were exported into Microsoft Excel for calculation of ratios and statistical analyses.
For real-time analysis of axonal transport of GFP-labeled structures, digital images of transfected cells were captured at intervals before and after treatment with pharmacological agents. Translocation of GFP-labeled structures was measured on digital images using an arbitrary vertical line drawn across the identical location in sequential images; particles were considered to have undergone net translocation provided that they translocated at least one full respective particle length (Yabe et al., 1999).
Manipulation of MAP kinase levels in cells
In order to probe the effects of MAP kinase activity on intracellular NF dynamics, we aimed to up- and down-regulate levels of MAP kinase in several ways, including incubation of cells with antisense- and sense-oriented oligonucleotides directed against MAP kinase (Ekinci et al., 1999), constructs expressing constitutively active and dominant-negative forms of MAP kinase (Li et al., 1999) and Provectin-mediated intracellular delivery of active and inactive MAP kinase. Immunoblot analysis of cytosolic extracts from cells treated with antisense oligonucleotides for 24 hours demonstrated a reduction in MAP kinase levels (Fig. 1). We also attempted to upregulate MAP kinase levels by intracellular delivery of active MAP kinase via the Provectin delivery system. The efficacy of Provectin-mediated intracellular delivery of protein into dbcAMP-treated NB2a/d1 cells was first examined by intracellular delivery of β-galactosidase followed by histochemical analysis for β-galactosidase activity. All cells (100%) expressed β-galactosidase activity following incubation with Provectin mixed with β-galactosidase, whereas none expressed activity following incubation with Provectin alone or with β-galactosidase in the absence of Provectin (Fig. 1A). As these data confirmed the efficacy of Provectin-mediated intracellular delivery, we next incubated cells under the same conditions with purified MAP kinase; intracellular delivery of MAP kinase via Provectin was confirmed by an approximate 35% increase in MAP kinase levels in immunoblot analysis (Fig. 1B). Immunoblot analyses also confirmed that treatment with antisense oligonucleotide reduced MAP kinase levels by nearly 40% (Fig. 1B).
MAP kinase alters intracellular NF phosphorylation
As MAP kinase has been reported to phosphorylate NF subunits (Veeranna et al., 1998), we next examined the effects of the above alterations in MAP kinase levels on NF phosphorylation in differentiated NB2a/d1 cells. Immunoblotting of Triton-insoluble cytoskeletons with antibodies to NF phospho-epitopes demonstrated that reduction of MAP kinase by each of several methods reduced NF phospho-epitopes, whereas increasing active MAP kinase also increased NF phospho-epitopes. Treatment of cultures with a pharmacological inhibitor (PD98059) that prevents MAP kinase activation by MEK1-mediated phosphorylation (Alessi et al., 1995; Pang et al., 1995) reduced NF phospho-epitopes (Fig. 2). Treatment with antisense but not sense oligonucleotides also reduced NF phospho-epitopes (Fig. 2). Transfection with the above construct expressing dominant-negative MAP kinase reduced NF phospho-epitopes, whereas transfection with the above construct expressing constitutively active MAP kinase increased NF phospho-epitopes (Fig. 2). Finally, intracellular delivery of active, but not inactive (boiled), MAP kinase increased NF phospho-epitopes (Fig. 2). Reprobing of the above immunoblots with a polyclonal antibody (L3) directed against NF-L revealed no change in NF-L levels following up- and down-regulation of MAP kinase activity, indicating that the above manipulations of MAP kinase activity induced alterations in NF phosphorylation state rather than total NF levels (Fig. 2).
MAP kinase affects NF dynamics
To examine whether alteration in MAP kinase activity affected NF dynamics, we transfected differentiated NB2a/d1 cells with a construct expressing NF-M conjugated to GFP (GFP-M) (Yabe et al., 1999; Yabe et al., 2001a; Yabe et al., 2001b) and monitored the distribution and organization of GFP-tagged NF subunits within axonal neurites following up- and down-regulation of MAP kinase activity by the above methods. Most neurites of cells differentiated and transfected as described in Materials and Methods exhibit a prominent NF bundle that has incorporated substantial GFP labeling along its length (Fig. 3A) (see also Yabe et al., 2001a; Yabe et al., 2001b). We therefore quantified the percentage of axonal neurites in which the NF bundle was labeled with GFP within the distal third of their length following up- and down-regulation of MAP kinase. This percentage was reduced following a 2-hour treatment with PD98059 (but not with its carrier, DMSO), treatment with antisense but not sense oligonucleotides corresponding to the sequence for MAP kinase and expression of dominant-negative MAP kinase (Fig. 3B,C). By contrast, this percentage was increased following Provectin-mediated intracellular delivery of active MAP kinase and by expression of constitutively active MAP kinase (Fig. 3C). Treatment with PD98059 did not prevent the increase in GFP-labeling of the central bundle induced by constitutively active MAP kinase (Fig. 3C). As the constitutively active kinase does not require phosphorylation for activity, this latter result therefore served as an additional control for the specificity of PD98059, as well as the constitutive activity of this construct (Li et al., 1999; Veeranna et al., 1998). Treatment with SB202190 did not alter the percentage of axonal neurites with GFP labeling within the distal axonal bundle (Fig. 3C), suggesting that, as observed for NF phosphorylation, p42/44 MAP kinase and not p38 MAP kinase, regulates NF dynamics within NB2a/d1 axonal neurites.
In addition to the large bundle that predominates along axonal neurites, endogenous NF subunits and subunits expressed following microinjection or transfection exist in multiple forms, including small filaments and non-filamentous punctate structures, both of which incorporate into the centrally situated NF bundle over time in a proximal to distal manner (Yabe et al., 2001a; Yabe et al., 2001b). As the possibility existed that manipulation of MAP kinase activity could have altered incorporation of subunits into the central bundle rather than overall transport, we also quantified total GFP fluorescence within the central and distal thirds of axonal neurites regardless of whether or not it was associated with the NF bundle. We observed that alteration of MAP kinase activity had the identical impact on distribution of total axonal GFP-M as it did on association of GFP-M within the bundle: a 2-hour treatment with PD98059 reduced the ratio of distal/central GFP-M by 29.7±5.9%, whereas co-transfection with constitutively active MAP kinase increased this ratio by 39.2±10.8%. These data suggest that MAP kinase influenced overall NF axonal transport rather than just the incorporation of subunits into the central bundle.
These phenomena are not likely to result from overall alterations in axonal transport as the levels and distribution along axonal neurites of mitotracker-labeled vesicular structures were unchanged compared to those observed in control cells following treatment with PD98059 or expression of either constitutively active or dominant-negative MAP kinase (Fig. 4). Similarly, PD98059 did not alter the distribution of newly transported (CFP-tagged) tau within axonal neurites; unlike GFP-M, identical levels of CFP-htau40 were detected in the distal-most region of axonal neurites with or without treatment with PD98059 (Fig. 5).
Alteration of MAP kinase activity under these conditions did not compromise overall axonal architecture. Immunofluorescence analyses indicated that PD98059 treatment altered neither steady-state levels of tau, nor levels of tau phosphorylated at the PHF-1 or Tau-1 epitopes, within the distal portion of axonal neurites (Fig. 5). In addition, no change was observed in steady state levels of filamentous actin or MTs within axonal neurites following incubation with PD98059 (Fig. 6). Axonal neurite length was not altered by PD98059 or by expression of either constitutively active or dominant-negative MAP kinase (Fig. 6). This indicates that changes in distribution of GFP-M within this relatively short incubation period were not derived from retraction or extension of neurites, or from overall disruption in steady state levels of cytoskeletal proteins. Consistent with immunoblot analyses however, NF phospho-epitopes were reduced within neurites and perikarya (Fig. 6).
To examine more closely the mechanism(s) by which inhibition of MAP kinase prevented accumulation of filamentous GFP-M within distal region of axonal neurites, we captured sequential images of neurites prior to and following treatment with PD98059 and monitored translocation of GFP-labeled structures over time. As described previously for these cells (Yabe et al., 2001b) and for cultured sympathetic neurons (e.g. Wang and Brown, 2001), the majority of GFP-labeled structures, which consist of both filamentous and punctate forms, normally underwent transport in an anterograde direction. We examined the behavior of these smaller NFs and punctate structures following up- and down-regulation of MAP kinase activity. Considerable heterogeneity in the number of these structures was observed among axonal neurites (e.g. Fig. 7) (see also Yabe et al., 2001a; Yabe et al., 2001b); nevertheless, alteration of MAP kinase activity affected axonal transport regardless of the number of structures or predominant form of GFP-M. Quantification of GFP-tagged small NFs and punctate structures in multiple axonal neurites revealed that 66±9% (mean±s.e.m.) underwent anterograde transport. Of the remaining GFP-labeled structures, 15±4% underwent retrograde transport, and 19±2% exhibited no apparent motion during our limited observation period. Within 15 minutes of PD98059 treatment however, only 5±2% of GFP-tagged structures underwent anterograde transport, indicating a reduction of approximately 90%. The percentage of subunits undergoing retrograde transport was not as severely affected and was reduced to 11±5% (a reduction of approximately 26%). The percentage of subunits displaying no net motion increased to 84±6% in accord with the reduction in moving particles (Fig. 7). Treatment with DMSO (the carrier for PD98059) at an identical concentration as utilized in PD98059 treatment did not alter transport of GFP-containing structures (Fig. 7). Transport of mitotracker-labeled structures was not altered by treatment with PD98059 under these conditions (Fig. 7), confirming that PD98059 did not exert overall toxicity. SB202190, which inhibits the activity of p38 MAP kinase, did not alter transport of GFP-labeled structures (Figs 7, 8). The kinase inhibitor olomoucine, statistically inhibited transport of both GFP- and mitotracker-labeled elements (Figs 7, 8), consistent with reports that olomoucine inhibits all anterograde fast axonal transport (Ratner et al., 1998).
Continued treatment (e.g. 2-4 hours) with PD8059 shifted the balance of transport of GFP-tagged structures to retrograde, such that approximately 65% of small filamentous and punctate structures exhibited retrograde transport (Fig. 9). Protracted treatment with PD90859 eventually completely depleted small filamentous and punctate structures from axonal neurites and induced overall retraction of the GFP-labeled filamentous bundle, resulting in a reduction of total axonal GFP fluorescence of approximately 45% within 6 hours (Fig. 9). This effect was not a reflection of an overall shift of total axonal transport to retrograde, as the total fluorescent signal within axonal neurites derived from mitotracker-labeled vesicular structures was reduced only by approximately 10% over this same interval (Fig. 9). This modest reduction in mitotracker represents an identical level of reduction in fluorescent bleaching resulting from repeated imaging of the same axonal neurite without treatment with PD98059: a 15.8±0.9% reduction in GFP fluorescence and a 9.6±4% reduction in mitotracker fluorescence were observed following sequential capturing of images of otherwise untreated cells intensity for second versus first image (n=3 untreated neurites). In some instances, long-term treatment with PD98059 induced retraction of both ends, or apparent `bunching' of NF bundles (i.e. retrograde retraction of the distal end along with anterograde coiling of both ends towards each other within the axonal hillock; Fig. 9).
The above data demonstrate that alteration in MAP kinase activity alters both NF C-terminal phosphorylation and NF axonal transport. However, it remained possible that these phenomena result from independent actions of MAP kinase. Alteration of NF axonal transport rate as a function of MAP kinase activity may actually be derived from changes in MAP kinase-mediated phosphorylation of NF motor proteins or linkers that maintain the association of NFs with their motor(s), rather than alteration in NF C-terminal phosphorylation. To address whether alteration in MAP kinase-mediated NF C-terminal phosphorylation directly impacted NF axonal transport, we transfected cells with constructs expressing wild-type (GFP-Hwt) NF-H conjugated to GFP, and the identical constructs in which serines in selected KSP domains known to be cdk5 phosphorylation sites have been mutated to aspartates (GFP-Hasp) to mimic a permanently phosphorylated state (Ackerley et al., 2003) (Fig. 10A). Consistent with the prior demonstration that cdk5 inhibits NF anterograde transport (Shea et al., 2004), this constitutively phosphorylated GFP-Hasp construct undergoes a slower transport rate than GFP-Hwt (Ackerley et al., 2003). It is unclear whether any or all of these mutated sites may be phosphorylated by MAP kinase. However, we reasoned that should MAP kinase promote anterograde NF transport by C-terminal sidearm phosphorylation of one or more of these mutated sites, anterograde transport of the GFP-Hasp construct may not be further inhibited by treatment with PD98059. Conversely, if the major impact of MAP kinase on NF transport is caused by something other than alterations in NF C-terminal phosphorylation, we anticipated that axonal transport of the GFP-Hasp and wild-type constructs would be similarly affected by PD98059. We monitored transport of these constructs as above both by quantifying the total level of GFP in the distal versus the central third of axonal neurites, as well as by real-time monitoring of the movement of GFP-tagged structures, before and after treatment with PD98059. A reduced distal/central ratio of total GFP was observed in GFP-HAsp-transfected versus GFP-Hwt-transfected cells. The day after transfection (when fluorescence is first consistently detectable within transfected cultures; see Materials and Methods), fluorescence intensity within the distal third of axonal neurites in cultures transfected with GFP-Hasp was only 64.2±0.9% of that in cultures transfected with GFP-Hwt. This correlates well with published results that GFP-Hasp undergoes transport at a rate approximately 65% of that of GFP-Hwt (Ackerley et al., 2003). Real-time monitoring of particle movement also yielded results similar to those reported previously (Ackerley et al., 2003): GFP-tagged structures in both GFP-Hwt- and GFP-Hasp-transfected cells displayed predominantly anterograde motion, with a minority of structures displaying retrograde motion and others pausing (defined as exhibiting no net motion over the observation period; Fig. 10D,E). We therefore conclude that GFP-H-asp and GFP-Hwt display similar transport kinetics in differentiated NB2a/d1 cells as previously demonstrated for cultured cortical neurons (Ackerley et al., 2003). We next compared the effect of inhibition of MAP kinase upon the transport of these constructs. The percentages of GFP-tagged particles in both GFP-Hwt- and GFP-HAsp-transfected NB2a/d1 cells undergoing anterograde or retrograde transport or not exhibiting any transport during the observation period (Fig. 10E) were similar to those observed for GFP-M (Fig. 7). The response of cells transfected with GFP-Hwt to PD98059 was also similar to that of cells transfected with GFP-M. However, cells transfected with GFP-Hasp were markedly less responsive to PD98059. As with GFP-M, the majority (71±10%) of anterogradely transporting GFP-tagged structures in GFP-Hwt-transfected cells ceased net translocation within 15 minutes of application of PD98059. By contrast, only ∼33±9% of anterogradely transporting GFP-tagged structures ceased net translocation in GFP-Hasp-transfected cells during this time, resulting in twice as many anterogradely transporting GFP-tagged structures maintaining anterograde transport following a 15-minute treatment with PD98059 in cells transfected with GFP-H-asp compared to those transfected with GFP-Hwt (P<0.05; Student's t-test; Fig. 10E). Failure of the constitutively phosphorylated construct to cease anterograde transport following treatment with PD98059 is consistent with the notion that MAP kinase-mediated phosphorylation of key sites within the C-terminal region of NF-H supports NF anterograde axonal transport. These constructs also yielded differing responses to longer (2-hour) treatment with PD98059, which increased retrograde transport of GFP-M (Fig. 9). GFP-Hwt fluorescence within distal axonal regions was reduced by 34.8±5.2% following 2-hour treatment with PD98059 (P<0.04, before versus after PD98059 treatment; Student's t-test). By contrast, GFP-Hasp fluorescence was not significantly reduced (12.4±14.7%; P<0.26, before versus after PD98059 treatment; Student's t-test; Fig. 10C). These data indicate that NF-H sidearm phosphorylation interferes with the increase in retrograde transport normally observed following long-term treatment with PD98059 (Fig. 9) and therefore support the conclusion that MAP kinase promotes anterograde NF axonal transport and/or inhibits retrograde NF transport and at least in part, phosphorylation of the C-terminus of NF-H.
We demonstrate that p42/44 MAP kinase promotes NF axonal transport and consistent with prior studies (Li et al., 1999; Veeranna et al., 1998), phosphorylates the C-terminus of NF-H and NF-M. Increased C-terminal NF phosphorylation has generally been considered to interfere with axonal transport, as accumulation of phospho-NFs within perikarya or proximal neurons accompanies certain neurodegenerative conditions (for reviews, see Grant and Pant, 2000; Julien, 1999). Similarly, hyperactivation of NF C-terminal kinases has been considered as one likely mechanism leading to disruption in normal NF distribution (Grant and Pant, 2000; Julien, 1999; Miller et al., 2002). Overexpression of the NF C-terminal kinase cdk5 (Shea et al., 2004), and treatment with glutamate (which activated SAP and MAP kinases) (Ackerley et al., 2000), each increased NF C-terminal phosphorylation and induced the accumulation of phospho-NFs within perikarya. We therefore anticipated that directly increasing MAP kinase activity would also impair NF axonal transport, especially as disregulation/over-amplification of the p42/44 MAP kinase can lead to neurodegeneration (e.g. Ekinci et al., 1999; Koistinaho and Koistinaho, 2002; Schroeter et al., 2002). However, despite increasing C-terminal NF phosphorylation within intact neuronal cells, we observed that MAP kinase promotes, rather than inhibits, NF transport. These data suggest that cdk5 and MAP kinase exert opposing influences on NF anterograde axonal transport, and confirm that the relationship of C-terminal NF phosphorylation to NF axonal transport and interaction is complex.
Hypophosphorylated NF subunits co-precipitate with kinesin and addition of MAP kinase to such preparations phosphorylates NF subunits and disrupts co-precipitation (Yabe et al., 2000). As the anterograde motor protein kinesin participates in transport of NF subunits into axonal neurites (Yabe et al., 1999), dissociation of NFs from an anterograde motor by MAP kinase under cell-free conditions seems at odds with the observed promotion of NF anterograde transport by MAP kinase within intact cells. One possible explanation is that MAP kinase-mediated dissociation of NF subunits from kinesin under cell-free conditions resulted from non-physiological hyperphosphorylation of NF subunits and/or kinesin linker protein(s). Alternatively, C-terminal NF phosphorylation derived from p42/44 MAP kinase may be essential to promote NF-NF associations leading to `bundling' within axons (Yabe et al., 2001a). MAP kinase may foster dissociation of NF subunits from their anterograde motor, perhaps via the same or additional phosphorylation events, and promote phospho-dependent NF bundling (Yabe et al., 2001a; Shea and Yabe, 2000). Compromise in the development of sufficient phospho-dependent NF-NF associations, coupled with dissociation of NFs from their anterograde transport system, might foster the observed net shift to retrograde transport. Notably, multiple NF kinases have been reported with distinct target sites; alterations in regulation of p42/44 MAP kinase may impair sequential phosphorylation by a distinct NF kinase, or place subunits at risk for inappropriate phosphorylation by a distinct kinase, leading to perturbations in NF transport and distribution. As the reduced impact of PD98059 on a constitutively phosphorylated NF-H construct (GFP-Hasp) supports the conclusion that p42/44 MAP kinase promotes anterograde NF transport by phosphorylation of NF-H sidearm phosphorylation, it also remains possible that cessation of anterograde NF transport following inhibition of p42/44 MAP kinase is due at least in part to decreased phosphorylation of other constituents, such as anterograde motors and/or linker proteins. In this regard, a second NF kinase, GSK-3B regulates the association of vesicular cargo with kinesin (Paglini et al., 2001); the influence of this kinase on transport of NFs has not been determined as yet. In addition, p42/44 MAP kinase has been proposed to regulate transport of vesicular elements, perhaps by associating with molecular motors (Kholodenko, 2002). Any putative influence of MAP kinase on motors and/or linkers would have to be confined to a subset, rather than all, linkers, as anterograde transport of mitotracker-labeled vesicular structures, as well as that of another axonal cytoskeletal protein, tau, was not altered following manipulation of MAP kinase activity.
The role of NF C-terminal phosphorylation in axonal transport has recently been questioned (Rao et al., 2003; Shea et al., 2003). However, our findings support the notion that NF-H C-terminal phosphorylation plays a role in NF axonal transport, as, consistent with the findings of Ackerley et al. (Ackerley et al., 2003), the relative level of GFP-Hasp within the distal third of neurites was reduced compared to that of GFP-Hwt. Despite this phosphorylation-mediated reduction in anterograde transport, this constitutively phosphorylated construct was less sensitive to inhibition of MAP kinase, supporting the interpretation that MAP kinase-mediated phosphorylation of NFs promotes their anterograde transport. These data support the possibility that phosphorylation at distinct sites, perhaps mediated by multiple kinases, may have additive and/or opposing effects on NF transport (see also Zhang et al., 2003). Although the data are consistent with the interpretation that MAP kinase promotes anterograde NF transport at least in part by C-terminal NF phosphorylation, this does not eliminate the possibility that MAP kinase also influences NF transport by phosphorylation of one or more motors and/or linker proteins. An additional possibility is that MAP kinase promotes anterograde NF transport somewhat indirectly by preventing the association of NFs with a competing retrograde motor; this could also be accomplished by either NF C-terminal phosphorylation and/or phosphorylation of retrograde motor(s) and/or their linkers (Shea and Flanagan, 2001).
The impact of alterations in activity of multiple kinases may be useful to discern the full relationship between C-terminal phosphorylation and NF transport. In this regard, p42/44 MAP kinase and p38 MAP kinase apparently exert differential effects on NF transport, as PD98059 (which inhibits p42/44 activity) inhibited transport of NFs whereas SB202190 (which inhibits p38 activity) did not. Increased or decreased p42/44 activity did not alter mitochondrial transport in our experiments, whereas p38 activity inhibits mitochondrial transport (De Vos et al., 2000), suggesting that these two members of the MAP kinase family may exert differential effects on transport of other elements as well. As inhibition of p38 activity increases p42/44 activity (Singh et al., 1999), it would be interesting to monitor the combined influence of alterations in these two kinases on NF distribution and/or overall axonal transport. Similarly, cdk5, which regulates fast anterograde transport of NFs (Shea et al., 2004) and vesicular cargo (Paglini et al., 2001; Ratner et al., 1998), inhibits p42/44 MAP activity (Sharma et al., 2002). Our findings leave open the possibility that cdk5, which promotes the accumulation of phospho-NFs within neuronal perikarya (Bajaj et al., 1998), does so in part by inhibiting MAP kinase, thereby indirectly inhibiting anterograde transport and increasing perikaryal NFs. This possibility is supported by the demonstration that both cdk5 and P42/44 MAP kinase are associated with cytoskeletal complexes that contain NF subunits in brain (Veeranna et al., 2000). Notably, olomoucine as utilized in the present study inhibited axonal transport of NFs, which is consistent with the prior report that olomoucine inhibited overall axonal transport (Ratner et al., 1998). Although olomoucine is active against cdk5, transfection-mediated increase in cdk5 activity diminished NF axonal transport, whereas another inhibitor active against cdk5 (roscovitine) promoted it. These data collectively suggest that olomoucine may exert its effect on NF axonal transport by inhibiting one or more kinases instead of or in addition to cdk5.
In prior studies demonstrating that glutamate impairs NF anterograde axonal transport, both p42/44 MAP kinase and SAP kinases were activated, leading the authors to suggest that either or both of these NF kinases may be responsible for glutamate-induced inhibition of anterograde NF transport (Ackerley et al., 2000; Brownlees et al., 2000). Our data support the possibility that increased SAP kinase activity, rather than increased activity of p42/44 MAP kinase, mediated the glutamate-induced inhibition of NF axonal transport. Although activation of either SAP (Brownlees et al., 2000; Giasson and Mushynski, 1996; O'Ferrall et al., 2000) or p42/44 MAP kinase (shown herein) increases NF phosphorylation our data demonstrates that activation of p42/44 MAP kinase increases rather than inhibits, NF anterograde axonal transport. Of interest therefore, would be to determine the influence of direct inhibition of SAP kinase on NF axonal transport. It should also be considered that the impact of alterations in kinase activities may not necessarily reflect normal kinase function, but may instead at least in part represent artifacts resulting from kinase overexpression or inappropriate levels of inhibition.
Our findings indicate that MAP kinase does not regulate axonal transport of tau at least to the degree to which it regulates transport of NFs. We did not carry out an exhaustive examination of tau, actin or MT dynamics and our limited data in this area should not be interpreted as indicating that MAP kinase does not regulate the dynamics of one or more of these proteins at some level. However, overexpression of tau in these cells affects bidirectional transport of other vesicles and organelles, apparently by interfering with their association with transport machinery. Tau overexpression eventually shifts the balance of NF transport in favor of net retrograde transport, and fosters the accumulation of NF clusters within perikarya (Stamer et al., 2002; Trinczek et al., 1999). As MAP kinase phosphorylates tau (Drewes et al., 1992; Lu et al., 1993; Roeder et al., 1993; Shea, 1997), there may be some as yet undisclosed interaction between tau and NFs that is modulated by MAP kinase.
Promotion of NF transport by p42/44 MAP kinase is consistent with the pivotal role of this kinase in development of the nervous system. The MAP kinase pathway mediates both neurogenesis and neuronal differentiation; stimulation of the MAP kinase pathway by epidermal growth factor induces neuronal proliferation, whereas stimulation by nerve growth factor induces differentiation (Vaudry et al., 2002). MAP kinase activity is furthermore required for multiple aspects of neuronal remodeling, including neurite outgrowth (Obara et al., 2002; Tang, 2003), axonal branching (Abe et al., 2001), dendrite formation (Valliant et al., 2002) and synaptic plasticity (Koh et al., 2002), as well as various non-neuronal cellular remodeling processes including oocyte and spermatocyte maturation (Kotani and Yamashita, 2002; Sette et al., 1999; Sun et al., 2002), spindle formation, chromatin condensation and nuclear reorganization following division (Lu et al., 2002; Stone et al., 2000) and cell migration (Nakamura et al., 2001). The data of the present study suggest that MAP kinase also participates in the stabilization of developing axons by promoting the accumulation of NFs within growing axonal neurites.
This research was supported by the National Science Foundation. We thank Dr Ron Liem for his NF-M cDNA, Dr E. Mandelkow for the tau construct and Dr C. Miller for his neurofilament constructs.
- Accepted February 2, 2004.
- © The Company of Biologists Limited 2004