Asymmetric clathrin-mediated endocytosis drives
repulsive growth cone guidance
33 ABSTRACT
Asymmetric Ca2+ elevations across the axonal growth cone mediate its turning responses to attractive and repulsive guidance cues. Here I show that clathrin-mediated endocytosis acts downstream of Ca2+ signals as driving machinery for growth cone turning. In dorsal root ganglion neurons, the formation of clathrin-coated pits is facilitated asymmetrically across the growth cone by directionally applied chemorepellent, Semaphorin 3A, or by Ca2+ signals that mediate repulsive guidance. In contrast, coated pit formation remains symmetric in the presence of attractive Ca2+
signals. Inhibition of clathrin-mediated endocytosis abolishes growth cone repulsion, but not attraction, induced by Ca2+ or extracellular physiological cues. Furthermore, asymmetric perturbation of the balance of endocytosis and exocytosis in the growth cone is sufficient to initiate its turning toward the side with less endocytosis or more exocytosis. With our previous finding that growth cone attraction involves asymmetric exocytosis, I propose that the balance between membrane addition and removal dictates bidirectional axon guidance.
34 INTRODUCTION
The formation of neuronal networks depends on the growth cone, a highly motile structure at the tip of elongating axons. Various environmental cues attract or repel axons by means of asymmetric cytosolic Ca2+ elevations across the growth cone (Henley et al., 2004; Togashi et al., 2008; Akiyama et al., 2009). The directional polarity of growth cone turning with regard to the localization of Ca2+
signals depends on the occurrence of ryanodine receptor (RyR)-mediated Ca2+-induced Ca2+ release (CICR): Ca2+ signals accompanied by CICR elicit growth cone turning toward the side with higher Ca2+ concentration (attraction) whereas Ca2+ signals without CICR elicit turning away from the side with higher Ca2+ concentration (repulsion) (Hong et al., 2000; Ooashi et al., 2005). Examples of downstream effectors of Ca2+ are Rho family GTPases (Jin et al., 2005) and actin-depolymerization factor/cofilin (Wen et al., 2007) that alter cytoskeletal organization for growth cone turning.
However, membrane trafficking may be an earlier mechanical event because attractive Ca2+ signals facilitate centrifugal vesicle transport and exocytosis within one minute, preceding any detectable changes in cytoskeletal dynamics in the growth cone (Tojima et al., 2007). This type of membrane trafficking is not involved in growth cone repulsion, suggesting that repulsive Ca2+ signals activate distinct but yet unidentified machinery.
One simple possibility is that growth cone repulsion is due to a local endocytosis that is triggered by repulsive Ca2+ signals. This hypothesis is consistent with previous findings that repulsive guidance cues enhance fluid-phase endocytosis in growth cones as assessed by dextran uptake (Fournier et al., 2000; Kolpak et al., 2009), although its functional significance remains obscure due to a lack of specific inhibitors of fluid-phase endocytosis. On the other hand, it is well known that clathrin- and dynamin-mediated endocytosis controls cell-surface expression of, and signal transduction by, receptors including those for axon guidance cues (Sorkin and Von Zastrow, 2002; Castellani et al., 2004; Piper et al., 2005; Bartoe et al., 2006), and growth cones do not respond to these cues when endocytosis mediated by clathrin and/or dynamin is pharmacologically inhibited (Piper et al., 2006; Kolpak et al., 2009).
In this chapter, I examine the hypothesis that asymmetric removal of growth cone plasma membrane via clathrin/dynamin-mediated endocytosis drives repulsive turning downstream of guidance signals, using direct local Ca2+ manipulation (Zheng, 2000; Ooashi et al., 2005) that can bypass receptor activation and any signaling events upstream of Ca2+. I show that the activity of clathrin/dynamin-mediated endocytosis becomes polarized in growth cones downstream of repulsive Ca2+ signals and that such asymmetry in endocytosis is required and sufficient for repulsive growth cone turning.
EXPERIMENTAL PROCEDURES Cell culture
As described previously (Ooashi et al., 2005), dorsal root ganglion (DRG) neurons from embryonic
35
day 9 - 10 chickens were dissociated and plated on a glass-based dish coated with laminin (approximately 10 g/ml; Invitrogen) or L1-Fc chimeric protein consisting of the extracellular domain of L1 and the Fc region of human IgG. The cultures were maintained in Leibovitz’s L-15 medium (Invitrogen) supplemented with N-2 (Invitrogen), 20 ng/ml nerve growth factor (Promega) and 750 g/ml bovine serum albumin (Invitrogen), in a humidified atmosphere of 100% air at 37°C.
Plasmid constructs
A plasmid encoding for EGFP fused to the light chain a of clathrin was a gift from J.H. Keen. To generate a plasmid for mCherry-clathrin fusion protein, the EGFP coding region in the EGFP-clathrin construct was replaced with mCherry (a gift from R.Y. Tsien) using NheI and HindIII.
A cDNA of wild type dynamin 1 (Kazusa DNA Research Institute) was subcloned into the XhoI-EcoRI site of the pEGFP-N1 vector (Clontech). EGFP-dynamin 1 K44A was a gift from M.A. McNiven. EGFP-AP180 C-terminus was a gift from H.T. McMahon.
Transfection
Dissociated neurons were transfected by electroporation (Nucleofector II; Lonza) according to the manufacture’s protocols and cultured for 1 - 2 days before microscopic analyses. The transfected cells were identified by their fluorescence.
Pharmacological agents
Unless otherwise noted, the following reagents were applied to some cultures at least 30 min before the experiments: 1 M monodansylcadaverine (MDC; Sigma), 20 M myristoylated dynamin inhibitory peptide (myr-P4; Tocris), 100 M tyrphostin A23 (Santa Cruz), 20 M Sp-cAMPS (Calbiochem), 20 M Rp-cAMPS (Calbiochem), 100 M ryanodine (Latoxan), 30 pM cyclosporin A (Santa Cruz), 1 nM deltamethrin (Calbiochem), 5 nM tetanus neurotoxin (TeNT; List) and 5 M of the active isomer (-) of blebbistatin (Calbiochem). Neurons were loaded with 1 M of an acetoxymethyl (AM) ester derivative of BAPTA (BAPTA-AM; Invitrogen) as described previously (Ooashi et al., 2005).
Dual-color total internal reflection fluorescence microscopy (TIRFM) imaging of clathrin and dynamin 1
Neurons were observed under an inverted microscope (IX81; Olympus) equipped with a total internal reflection illumination system (IX2-RFAEVA-2; Olympus) and a 100x objective (PlanApo TIRFM, oil, NA 1.45; Olympus). Unless otherwise noted, dual-color images of EGFP-dynamin 1 and mCherry-clathrin were obtained by simultaneous TIRFM as follows: EGFP-dynamin 1 and mCherry-clathrin immediately adjacent to the coverglass-cell interface were excited simultaneously with evanescent waves of a 488 nm solid-state laser (Melles Griot) and a 561 nm solid-state laser (LASOS Lasertechnik), respectively. The EGFP and mCherry emissions were split by a dichroic mirror (DM570; Olympus) and collected through a band-pass filter (FF01-514/30; Semrock) and a long-pass filter (BA610IF; Olympus), respectively. The fluorescence images of both channels were
36
acquired simultaneously with two CCD cameras (ORCA-AG with binning set at 2 x 2; Hamamatsu Photonics) mounted on different ports of the microscope. The image acquisition was controlled by MetaMorph version 7.5 software (Molecular Devices) and an electronic stimulator (Nihon Koden).
Fluorescent signals concentrated within 0.8 m in diameter were defined as clathrin or dynamin 1 puncta. Their lifetime was defined as the duration from appearance to disappearance of the puncta. Their migration rate was calculated as the whole distance of movement divided by their lifetime. Clathrin-coated pit (CCP) recruitment of dynamin 1 was defined as newly formed dynamin 1 puncta colocalizing with pre-existing clathrin puncta. This was termed ‘transient dynamin 1 recruitment’ in this chapter, because the lifetime of all dynamin 1 puncta under no-drug condition (Fig. 12C) was shorter than 10 s. The image analyses were performed using AquaCosmos 2.6 software (Hamamatsu Photonics).
TIRFM combined with UV-photolysis
I used a custom-built microscope comprised of an upright UV-photolysis unit (Olympus) and an inverted TIRFM system (Fig. 11A). The UV-photolysis unit consisted of a 75 W xenon light source, a shutter, a pinhole at the field-stop position, a band-pass filter (BP330-385; Olympus), and a 40x objective (LUMPlanFl, water dipping, NA 0.80; Olympus). The area that covered the central (C)-domain periphery to the leading edge on one side of the growth cone (~10 m in diameter) was exposed to UV light for photolysis. The border between the C- and peripheral (P)-domain was determined on a differential interference contrast (DIC) image of the growth cone (Tojima et al., 2007).
For UV-photolysis and TIRFM imaging of Ca2+ (Fig.11B, C), neurons were loaded with 2 M Oregon Green 488 BAPTA-1-AM (OGB-1-AM; Invitrogen) and 2 M o-nitrophenyl EGTA (NP-EGTA; Invitrogen) in the presence of 0.0025% Cremophor EL (Nacalai Tesque) for 30 min.
OGB-1 was excited with a 488 nm laser, and its emission was collected through a band-pass filter (FF01-514/30) and acquired with a CCD camera (ImagEM with binning set at 4 x 4; Hamamatsu Photonics) every 200 ms at an exposure of 30 ms. NP-EGTA was photolyzed by two UV pulses (pulse width, 100 ms; interpulse interval, 200 ms). The UV-shutter timing was controlled by an electronic stimulator such that a camera exposure was initiated 30 ms after the close of the UV-shutter. This interval was sufficiently long to prevent UV-induced artifacts from affecting Ca2+
imaging. For Ca2+ measurements, fluorescence intensities (F) of OGB-1 were averaged within regions of interest (ROIs). Relative fluorescence over the basal fluorescence (F/F0) was calculated, where F0 was the mean of 20 consecutive F values taken from -3.87 to -0.07 s (before the onset of UV photolysis). Changes in cytosolic Ca2+ levels were expressed as (F/F0), where (F/F0) = F/F0 - 1.
To examine the effect of Ca2+ signals on endocytosis (Fig. 11D-K, Table 2), NP-EGTA was photolyzed by repetitive UV irradiation for a duration of 100 ms every 3 s, and time-lapse TIRFM images of fluorescent proteins collected through an emission filter (FF01-514/30 or BA610IF) were acquired with a CCD camera (ORCA-AG; binning 2 x 2). For quantification of asymmetric endocytosis, the UV-irradiated area and the corresponding area on the opposite side of a growth cone
37
were defined as the near and far ROIs respectively. Far ROIs were positioned by an observer blind to the experimental conditions. The number of newly formed clathrin puncta within each ROI was normalized by the growth cone area covered by that ROI.
TIRFM combined with extracellular gradient application
In the experiments designed to analyze the effects of extracellular gradients on CCP formation (Figs.
14A-D, 15A-D), neurons expressing EGFP-clathrin were cultured on laminin and observed under the inverted TIRFM system described in the previous section. An extracellular gradient was generated with a glass pipette containing Semaphorin 3A (Sema3A; 100 g/ml in pipette; R&D Systems) or MDC (100 M in pipette), which was positioned at 50 m from the growth cone with 45-degree angle with respect to the original direction of axon elongation. A graded distribution of the reagent was formed across the growth cone within 1 min after the onset of pulsate positive pressure in the pipette (amplitude, 4 psi; pulse width, 20 ms; interpulse interval, 500 ms) (Akiyama et al., 2009).
EGFP was excited with a 488 nm laser, and its emission was collected through a band-pass filter (FF01-514/30) and acquired with a CCD camera (ORCA-AG; binning 2 x 2) every 3 s. To assess the asymmetry in endocytosis, two circular ROIs (~10 m in diameter) were positioned on the near and far side of the growth cone such that the ROI covered the C-domain periphery to the leading edge of the growth cone. All the ROIs were positioned by an observer blind to the experimental conditions. The number of newly formed clathrin puncta within each ROI was normalized by the growth cone area covered by that ROI.
Epifluorescence imaging of Ca2+ signals induced by physiological cues
Ca2+ signals evoked by physiological guidance cues, Sema3A and myelin-associated glycoprotein (MAG), were visualized by simultaneous and ratiometric imaging of two fluorescent Ca2+ indicators, OGB-1 and Fura Red (FR) (Figs.14E-K, 15A-G). Neurons cultured on laminin were loaded with 2
M OGB-1-AM and 0.5 M FR-AM (Invitrogen), and both indicators were excited simultaneously using an excitation filter (FF01-470/22; Semrock) and a 100x objective (UPLSAPO, oil, NA 1.40;
Olympus) on an inverted microscope (IX81). The OGB-1 and FR emissions were split by a dichroic mirror (T565lpxr; Chroma) equipped in an emission splitter (DV2; Photometrics). The split emissions were collected through a band pass filter (HQ530/30; Chroma) for OGB-1 and a long pass filter (BA610IF) for FR and acquired simultaneously with a CCD camera (ImagEM; binning 4 x 4) every 3 s. Extracellular gradients of Sema3A or MAG (150 g/ml in pipette; R&D Systems) were generated as described in the previous section. For Ca2+ measurements, ROI was defined as a circular zone that was maximal in size but remained inside the growth cone as it moved with time.
Fluorescence intensities (F) of OGB-1 and FR were averaged within the ROI. F/F0 was calculated individually for OGB-1 and FR, where F0 was the mean of 20 consecutive F values taken from -60 to -3 s (before the onset of Sema3A/MAG application). The F/F0 values for OGB-1 and FR were designated as ROGB-1 and RFR, respectively. Changes in cytosolic Ca2+ levels were expressed as
ROGB-1/RFR, where ROGB-1/RFR = ROGB-1/RFR - 1. In Figs. 14I-K, 15D-G, the peak Ca2+
amplitudes before and after the application of guidance cues were compared. Here, the peak
38
amplitude was defined as the mean of 5 consecutive ROGB-1/RFR values that included the maximal
ROGB-1/RFR in the middle.
Epifluorescence imaging of exocytosis induced by -latrotoxin (-LTX)
To monitor vesicle-associated membrane-protein 2 (VAMP2)-mediated exocytosis in growth cones (Fig. 16E-I), I transfected neurons on laminin with pH-sensitive version of Venus (pHVenus) fused to the luminal end of VAMP2 (pHVenus-VAMP2) (Tojima et al., 2007). To rule out the artifactual fluorescence changes due to alterations in growth cone shape, growth cone plasmalemma was also labeled with DsRed-Monomer-Mem (Clontech). pHVenus-VAMP2 and DsRed-Monomer-Mem were excited sequentially using excitation filters (FF01-470/22 and FF01-556/20, respectively;
Semrock) and a 100x objective (UPLSAPO) on an inverted microscope (IX81). The pHVenus and DsRed emissions were collected through band-pass filters (FF01-514/30 and S630/60m; Semrock and Chroma, respectively) and acquired with a CCD camera (ImagEM; binning 4 x 4) every 3 s.
Extracellular gradient of -LTX (10 nM in pipette, Sigma) was generated as described in the previous section. For quantitative analyses, the near and far ROIs were defined as described previously (Akiyama et al., 2009). Fluorescence intensities (F) of pHVenus and of DsRed were averaged within the ROI. F/F0 was calculated individually for pHVenus and DsRed, where F0 was the mean of 20 consecutive F values taken from -60 to -3 s (before the onset of -LTX application).
The F/F0 values for pHVenus and DsRed were designated as RpHVenus and RDsRed, respectively. The RpHVenus/RDsRed values, designated as R, were determined for near ROI (Rnear) and for far ROI (Rfar).
Changes in the asymmetry in exocytosis were expressed as Rnear /Rfar, where Rnear /Rfar = Rnear /Rfar - 1.
Growth cone turning assays
Growth cone turning induced by focal laser-induced photolysis of NP-EGTA was performed as described previously (Ooashi et al., 2005). Growth cone turning induced by an extracellular gradient of Sema3A, MAG, MDC, myr-P4 (20 mM in pipette) or -LTX was performed on a laminin substrate as described previously (Akiyama et al., 2009).
Statistics
Data are expressed as the mean ± SEM. Statistical analyses were performed using Prism version 4.01 software (GraphPad). P values < 0.05 were judged statistically significant.
RESULTS AND DISCUSSION
Visualization of clathrin-mediated endocytosis in growth cones
The growth cone, which consists of the organelle-rich C-domain and actin-rich P-domain, actively internalizes and recycles membrane components (Pfenninger, 2009). I analyzed the spatiotemporal dynamics of mCherry-tagged clathrin and EGFP-tagged dynamin 1 in growth cones of chicken
39
DRG neurons by TIRFM (Fig. 10). This method has been used extensively to visualize the formation of CCPs and clathrin-coated vesicles (CCVs) in non-neuronal cells, both processes involved in clathrin-mediated endocytosis (Merrifield et al., 2002; Rappoport and Simon, 2003;
Merrifield et al., 2005). CCPs, concentrated mCherry-clathrin visible as puncta, appeared in the P-domain and migrated toward the C-domain of growth cones. Then, EGFP-dynamin 1 was recruited transiently to CCPs followed by disappearance of mCherry-clathrin fluorescence, which should represent the transition from CCPs to CCVs. It is most likely that a series of these events corresponds to clathrin-mediated endocytosis, although some CCPs may be abortive and fail to complete endocytosis even if dynamin is recruited (Ehrlich et al., 2004). In my experiments, the rate of CCP migration in the P-domain was equivalent to the rate of F-actin retrograde flow (Tojima et al., 2007) and decreased significantly after treatment with blebbistatin, a specific myosin II inhibitor that slows F-actin flow (Medeiros et al., 2006): 4.18 ± 0.42 m/min before treatment versus 2.15 ± 0.35 m/min after treatment (n = 35 puncta in 7 growth cones; P < 0.01, paired t-test).
These data suggest that CCPs are coupled with F-actin flow.
CCP formation becomes asymmetric across the growth cone in response to repulsive Ca2+
signals
I tested whether the activity of clathrin-mediated endocytosis is controlled by Ca2+ signals generated by photolyzing a caged Ca2+ compound, NP-EGTA on one side of growth cones. One primary distinction between attractive and repulsive Ca2+ signals is the occurrence of CICR through RyRs that is regulated by extracellular substrates via cyclic AMP (cAMP) signaling (Ooashi et al., 2005):
photolysis-induced Ca2+ elevations cause growth cone attraction and repulsion when the cAMP pathway is activated and inactivated, respectively. As shown in Table 1, the turning direction can be switched by changing culture substrates (L1 or laminin) or by manipulating the activity of cAMP or RyR with the following drugs: Rp-cAMPS, a cAMP antagonist; Sp-cAMPS, a cAMP agonist;
ryanodine, a CICR inhibitor when used at a high concentration.
Using a custom-built microscope comprised of an upright UV-photolysis unit and an inverted TIRFM system (Fig. 11A-C), I evoked attractive or repulsive Ca2+ signals by photolyzing NP-EGTA on one side of a growth cone and compared endocytic activity in the UV-irradiated area (near) against that in the corresponding area on the opposite side (far) (Fig. 11D-K). I counted the number of newly formed clathrin puncta as a measure of endocytic activity, because almost all examined clathrin puncta (59/60, 98.3%) were accompanied transiently by dynamin 1 and presumed to participate in endocytosis. Furthermore, attractive and repulsive Ca2+ signals altered neither the percentage of clathrin puncta accompanied by dynamin 1 nor the lifetime of clathrin puncta (Table 2), supporting my method of assessing Ca2+-induced changes in endocytosis by measuring the appearance of clathrin puncta. Attractive Ca2+ signals (L1) had no detectable effect on the relative activity of endocytosis between both sides of the growth cone (Fig. 11F). In response to repulsive Ca2+ signals (L1 Rp-cAMPS; L1 ryanodine), however, endocytosis became asymmetric with higher activity in the near side (Fig. 11G, H). The asymmetry was abolished by MDC, an inhibitor of clathrin-mediated endocytosis (Ray and Samanta, 1996; Schutze et al., 1999) (Fig. 11I), while the
40
vast majority of clathrin puncta (48/50, 96.0%) were accompanied by dynamin 1 after MDC treatment. Also, the asymmetry in CCP formation induced by repulsive Ca2+ signals was abolished by cyclosporin A or deltamethrin inhibition of calcineurin, a Ca2+-dependent phosphatase (Fig. 11J, K), consistent with the fact that calcineurin mediates Ca2+-triggered endocytosis of synaptic vesicles (Cousin and Robinson, 2001). My results indicate that repulsive, but not attractive, Ca2+ signals cause asymmetric clathrin-mediated endocytosis across the growth cone via calcineurin.
Clathrin-mediated endocytosis is necessary for Ca2+-induced growth cone repulsion
I next tested whether Ca2+-induced growth cone turning requires clathrin/dynamin-mediated endocytosis using a variety of pharmacological and genetic inhibitors: MDC; tyrphostin A23, an inhibitor of cargo binding to AP-2 (Banbury et al., 2003); myr-P4, myristoylated dynamin inhibitory peptide that blocks binding of dynamin to amphiphysin (Marks and McMahon, 1998); dynamin 1 K44A, a dominant-negative dynamin 1 mutant that lacks GTPase activity (Damke et al., 1994); and the C-terminal fragment of AP180, an inhibitor of the accessory protein AP180 involved in clathrin-mediated endocytosis (Ford et al., 2001). The effects of these inhibitors on clathrin-mediated endocytosis in growth cones were characterized by TIRFM of clathrin and dynamin 1. MDC decreased the number of newly formed clathrin puncta but did not affect their lifetime (Fig. 12A, B). Tyrphostin A23 also attenuated the appearance of clathrin puncta (Fig. 12A).
In contrast, myr-P4 drastically prolonged the lifetime of both clathrin and dynamin 1 puncta without affecting the appearance of clathrin puncta (Fig. 12A-C). Similarly, dynamin 1 K44A associated with clathrin for extended periods of time and prolonged the lifetime of clathrin puncta (Fig. 12D, E).
These results are consistent with the fact that these inhibitors attenuate clathrin/dynamin-mediated endocytosis via distinct mechanisms.
Bidirectional growth cone turning was triggered by repetitive laser-induced photolysis of NP-EGTA on one side of the growth cone (Zheng, 2000; Ooashi et al., 2005; Tojima et al., 2009).
Ca2+-induced repulsion (L1 Rp-cAMPS; L1 ryanodine; laminin), but not attraction (L1; laminin Sp-cAMPS), was abolished by pharmacological endocytosis inhibitors (MDC, tyrphostin A23 and myr-P4) (Fig. 13 A-I). Similarly, dominant-negative mutants, dynamin 1 K44A and AP180 C-terminus, prevented repulsion but not attraction (Fig. 13J). Furthermore, calcineurin inhibitors, cyclosporin A and deltamethrin, blocked repulsion but not attraction (Fig. 13K). The involvement of calcineurin in Ca2+-induced growth cone repulsion is consistent with previous observation (Wen et al., 2004). My data strongly support the idea that Ca2+-induced repulsion requires calcineurin-mediated, clathrin- and dynamin-dependent endocytosis.
Extracellular physiological cues repel growth cones via asymmetric clathrin-mediated endocytosis
Ca2+ signals mediate growth cone turning responses to extracellular gradients of physiological cues such as Sema3A (Togashi et al., 2008) and MAG (Henley et al., 2004). I examined whether growth cone repulsion induced by these cues depends on clathrin-mediated endocytosis (Figs. 14, 15).
Directional application of Sema3A through a pipette caused asymmetric formation of CCPs with
41
more CCPs on the side of the growth cone facing the source of Sema3A (Fig. 14A-D). Because Sema3A causes ‘transient desensitization’ via clathrin-mediated removal of its receptor, neuropilin-1, from the growth cone surface (Piper et al., 2005), I examined the effect of MDC on intracellular signaling events downstream of the Sema3A receptor. Sema3A increased Ca2+ in growth cones regardless of the presence of MDC (Fig. 14E-K), indicating that Sema3A generated intracellular signals even when endocytosis was inhibited and that growth cones in my experiments showed subtle or negligible desensitization at the receptor level. These results supported my experimental design for studying the involvement of MDC-sensitive endocytosis downstream of Ca2+ signals.
Sema3A gradients repelled growth cones, and this repulsion was converted into attraction when endocytosis was inhibited by MDC (Fig. 14L, M, O). This attraction was negated by pretreating neurons with BAPTA-AM, a fast Ca2+ chelator, or with Rp-cAMPS (Fig.14O). Recently it has been shown that Ca2+-induced attraction requires VAMP2-mediated exocytosis and is sensitive to TeNT (Tojima et al., 2007). Therefore I tested the effect of TeNT and found that Sema3A-induced attraction in the presence of MDC depended on VAMP2-mediated exocytosis (Fig. 14N, O). These results suggest that Sema3A enhances both endocytosis and exocytosis via Ca2+ signals and can induce attractive turning if exocytosis predominates over endocytosis.
MAG also increased Ca2+ in growth cones regardless of the presence of MDC (Fig. 15A-G).
As reported previously (Tojima et al., 2007), MAG gradients attracted and repelled growth cones in the absence and presence of Rp-cAMPS, respectively (Fig. 15H, J, M). MDC did not affect MAG-induced attraction (Fig. 15I, M), indicating that clathrin-mediated endocytosis is dispensable for MAG-induced attraction. In contrast, MDC converted MAG-induced repulsion into attraction, and this attraction was blocked by TeNT (Fig. 15K-M). These results are in agreement with the idea that growth cone repulsion, but not attraction, requires clathrin-mediated endocytosis downstream of Ca2+ and imply that guidance cues evoke both endocytosis and exocytosis, the balance of which determines the turning direction.
Asymmetric endocytosis/exocytosis is sufficient for growth cone turning
Finally I tested whether direct manipulation of membrane trafficking causes growth cone turning (Fig. 16). I was able to produce asymmetric endocytosis by exposing growth cones to extracellular gradients of MDC, with less CCP formation on the side facing higher MDC concentration (Fig.
16A-D). Also, asymmetric VAMP2-mediated exocytosis was produced by directional application of -LTX, a stimulator of exocytosis (Ushkaryov et al., 2004) (Fig. 16E-I). Such gradients of MDC and -LTX induced growth cone attraction (Fig. 16J-L). Similarly, directionally applied myr-P4 attracted growth cones (Fig. 16L). These data indicate that asymmetric endocytosis/exocytosis across the growth cone is sufficient to initiate its turning toward the side with less endocytosis or more exocytosis.
To gain insight into mechanisms of how asymmetric endocytosis drives growth cone turning, I estimated the area of plasma membrane that was removed asymmetrically from the growth cone surface in response to Sema3A gradients (Fig. 16C). The number of newly formed CCPs in these growth cones after Sema3A application was 0.78 ± 0.11/µm2/min on near sides and 0.38 ±
42
0.06/µm2/min on far sides (n = 7 growth cones). Assuming the diameter of CCVs to be 120 nm (Conner and Schmid, 2003), the difference in the area of membrane removal between the near and far sides would be 0.018 µm2 per 1-µm2 growth cone area (1.8%) every minute, therefore the cumulative difference during the course of growth cone turning (40 min; Fig. 16L) corresponds to 72%. The actual difference should be larger, given that I detect only a subset of newly formed CCPs, being those that incorporated EGFP-clathrin. This estimation suggests that clathrin-mediated endocytosis can generate a substantial asymmetry in membrane removal across the growth cone. Therefore, in addition to the previously proposed role of fluid-phase endocytosis (Kolpak et al., 2009), a local regulation of the growth cone surface area by clathrin-mediated endocytosis may contribute to repulsive axon guidance.
Another possibility is that asymmetric endocytosis drives growth cone turning by redistributing specific functional molecules such as cell adhesion molecules. A similar mechanism exists in non-neuronal cells, in which clathrin mediates integrin endocytosis for focal adhesion disassembly in migrating cells (Ezratty et al., 2009). Clathrin-mediated endocytosis has been shown to retrieve cell adhesion molecules from the growth cone surface (Dequidt et al., 2007;
Diestel et al., 2007) and, therefore, could cause growth cone turning if perturbed asymmetrically.
In conclusion, together with our previous study (Tojima et al., 2007), I demonstrate that growth cone repulsion and attraction involve preferential removal and addition of plasma membrane components of the growth cone on the side with elevated Ca2+, respectively, and that the balance between these counteractive events may determine the directional polarity of axon guidance.
43
Table 1. Combinations of culture substrates and pharmacological drugs to generate attractive or repulsive Ca2+ signals
The growth cone turning direction with respect to asymmetric Ca2+ signals depends on the occurrence of CICR through RyRs. L1-mediated adhesion activates the cAMP pathway, thereby leading RyRs to the active state. In this situation, Ca2+ liberated from NP-EGTA upon photolysis triggers CICR through RyRs, which is sufficient to induce attractive turning. In contrast, laminin-mediated adhesion leads RyRs to the inactive state via inactivating the cAMP pathway. In this situation, photolysis-induced Ca2+ elevations are not accompanied by CICR and therefore trigger repulsion. In this chapter, I switched between attractive and repulsive Ca2+ signals by changing culture substrates (L1 or laminin) or by manipulating the activity of cAMP or RyR with the indicated pharmacological drugs. Sp-cAMPS, a cAMP agonist; Rp-cAMPS, a cAMP antagonist; Ryanodine, a CICR inhibitor when used at a high concetration (100 M).
44
Table 2. Effects of Ca2+ signals on dynamics of clathrin and dynamin 1 puncta
Attractive or repulsive Ca2+ signals were produced by repetitive UV-photolysis of NP-EGTA (pulse width, 100 ms; interpulse interval, 3 s) on one side of the growth cone (near ROIs). The percentage of clathrin puncta accompanied by dynamin 1 was determined using simultaneous dual-color TIRFM images acquired every 1 s. The lifetime of clathrin puncta was determined using TIRFM images of EGFP-clathrin acquired every 3 s.
45
Figure 10. Visualization of clathrin- and dynamin 1-mediated endocytosis in growth cones A, TIRFM image of mCherry-clathrin (red) and EGFP-dynamin 1 (green) expressed in a growth cone that has been cultured on a coverglass coated with the cell adhesion molecule L1. EGFP and mCherry were excited sequentially with 488/561 nm laser and imaged with a CCD camera
46
(ORCA-AG; binning 2 x 2). The broken blue line represents the margin of the growth cone.
Scale bar, 5 m. B, Magnified time-lapse images of mCherry-clathrin, EGFP-dynamin 1 and both proteins (merged) in the white boxed areas (ROIs 1 - 3) shown in (A). Newly formed clathrin puncta in the P-domain migrated toward the C-domain, colocalized transiently with dynamin 1 puncta, and then disappeared. Time (s) after the onset of time-lapse imaging is shown. Previous studies (Merrifield et al., 2002; Rappoport and Simon, 2003) suggested that this type of behavior of clathrin and dynamin 1 represents direct visualization of the formation of CCPs and their transition to CCVs: first, the appearance of clathrin puncta corresponds to the formation of CCPs; second, the transient recruitment of dynamin 1 should correspond to its assembly around the neck of CCPs and pinching the invaginated pits off the plasma membrane to form CCVs; and third, the disappearance of clathrin puncta should correspond to vertical movement of CCVs out of the evanescent field. C, Histogram of lag-time between dynamin 1 disappearance and clathrin disappearance. For precise measurements of the lag-time, TIRFM images of clathrin and dynamin 1 were aquired simultaneously with two CCD cameras every 500 ms. All the examined puncta showed positive lag-time values, indicating that dynamin 1 always disappeared prior to clathrin disappearance.
47
48
Figure 11. Repulsive, but not attractive, Ca2+ signals induce asymmetric clathrin-mediated endocytosis across the growth cone
A, Schematic representation of the microscope setup comprised of the upright UV-photolysis unit and the inverted TIRFM system. NP-EGTA was photolyzed with UV-light (330 - 385 nm) through a 40x water-dipping objective (NA 0.80). Fluorophores were excited with the evanescent wave of a 488 nm laser through a 100x oil-immersion objective (NA 1.45). The emitted fluorescence was collected by the 100x objective and acquired with a CCD camera. AOTF, acousto-optic tunable filter; BP, band-pass filter; Em, emission filter; DM, dichroic mirror. B, C, Generation and detection of asymmetric elevations of cytosolic Ca2+ concentrations near the substrate-facing plasma membrane of the growth cone using this microscope setup. B, TIRFM image of an Oregon Green 488 BAPTA-1 (OGB-1)-loaded growth cone on L1. Scale bar, 10 m. C, Time-course of
F/F0) on the side with UV irradiation (near ROI; red circle in B) and the opposite side (far ROI;
blue circle in B), where F/F0) represents changes in Ca2+ concentrations from the basal level.
Loaded NP-EGTA was photolyzed by two UV pulses. The first pulse started at 0 s (arrowhead).
Note that Ca2+ elevations were detected only in the near ROI. D-K, Effects of Ca2+ signals on CCP formation. D, TIRFM image of EGFP-clathrin in a growth cone cultured on L1 in the presence of Rp-cAMPS. NP-EGTA was photolyzed to generate repulsive Ca2+ signals. The UV-irradiated area and the corresponding area on the opposite side were used as near and far ROIs, respectively (blue circles). Scale bar, 10 m. E, Schematic representation of the growth cone shown in (D).
TIRFM images of EGFP-clathrin in this growth cone were acquired every 3 s before (pre) and after (UV) the onset of repetitive UV irradiation (time 0 s). Each red cross marks the position of a newly formed CCP within the ROIs during the indicated 30-s periods. Scale bar, 10 m. F-K, Near-to-far ratios of CCP formation. The y-axis indicates the number of newly formed CCPs per unit area within near ROI divided by that within far ROI, before (pre; -120 to 0 s) and after (UV; 0 to 120 s) the onset of repetitive photolysis. Each line represents a photolysis-induced change in a single growth cone. Elicited Ca2+ signals were attractive (L1) or repulsive (L1 Rp-cAMPS; L1 ryanodine). Also, the effects of Ca2+ signals were examined in the presence of MDC (I), cyclosporin A (J) or deltamethrin (K). *P < 0.05, ***P < 0.001, paired t-test.
49
50
Figure 12. Effects of endocytosis inhibitors on dynamics of clathrin and dynamin 1 puncta in growth cones
A, Effects of endocytosis inhibitors on CCP formation. TIRFM images were acquired every 3 s before (pre) and at least 5 min after bath application of the indicated drugs. The y-axis indicates the number of newly formed clathrin puncta per unit area and time in growth cone P-domains. *P <
0.05, paired t-test. B, Histograms of lifetime of clathrin puncta in the absence (no drug) or presence of the indicated drugs. The lifetime was determined using TIRFM images acquired every 3 s.
Note that the rightmost bins correspond to lifetime longer than 90 s. Numbers in parentheses indicate the number of clathrin puncta examined. C, Histograms of lifetime of dynamin 1 puncta in the absence (no drug) or presence of the indicated drugs. The lifetime was determined using TIRFM images acquired every 500 ms. Numbers in parentheses indicate the number of dynamin 1 puncta examined. D, E, Histograms of lifetime of clathrin (D) and dynamin 1 K44A (E) puncta.
Neurons were transfected with mCherry-clathrin and EGFP-dynamin 1 K44A, and their lifetime as puncta was determined using TIRFM images acquired every 3 s. Numbers in parentheses indicate the number of puncta examined.
51
52
Figure 13. Ca2+-induced repulsion but not attraction requires clathrin-mediated endocytosis A-H, Time-lapse DIC images of growth cones on L1 (A-D) or laminin (E-H) in the presence of the indicated drugs. Focal Ca2+ signals were generated by laser photolysis of NP-EGTA at red spots.
Time (min) after the onset of repetitive laser irradiation is shown. Scale bar, 10 m. I-K, Turning angles of growth cones, with positive and negative values indicating attraction and repulsion, respectively. The effects of pharmacological endocytosis inhibitors (I), dominant-negative mutants of endocytic machinery components (J) and calcineurin inhibitors (K) were tested. Elicited Ca2+
signals were attractive (L1; laminin Sp-cAMPS) or repulsive (L1 Rp-cAMPS; L1 ryanodine;
laminin). Numbers in parentheses indicate the number of growth cones examined. *P < 0.05,
**P < 0.01, ***P < 0.001, unpaired t-test for comparison between two groups; Dunnett’s test for comparison among three groups.
53
54
Figure 14. Sema3A-induced repulsion requires clathrin-mediated endocytosis downstream of Ca2+ signals
A-D, Sema3A gradient evokes asymmetric CCP formation. A, TIRFM image of EGFP-clathrin in a growth cone exposed to an extracellular gradient of Sema3A (arrow). The growth cone was on laminin. The blue circles indicate near and far ROIs for analyses. Scale bar, 10 m. B, Schematic representation of the growth cone shown in (A). Each red cross marks the position of a newly formed CCP within the ROIs during the indicated 30-s periods. Sema3A gradient was initiated at time 0 s. Scale bar, 10 m. C, D, Effects of Sema3A gradient (C) or control gradient (D) on near-to-far ratios of CCP formation. Control gradient was generated by ejecting PBS (vehicle). The y-axis indicates the number of newly formed CCPs per unit area within near ROI divided by that within far ROI, before (pre; -120 to 0 s) and after (Sema3A/PBS gradient; 120 to 240 s) the onset of gradient application. Each line represents a Sema3A/PBS-induced change in a single growth cone. **P < 0.01, paired t-test. E-K, Sema3A-generated Ca2+ signals in growth cones. E, A growth cone loaded with two Ca2+ indicators, OGB-1 and Fura Red (FR), was exposed to an extracellular gradient of Sema3A (arrow). The left panel shows FR fluorescence. The red circle indicates the ROI for Ca2+ measurements. The pseudo-color shows ROGB-1/RFR), a measure of changes in Ca2+ concentrations from the basal level. Time (s) after the onset of Sema3A application is shown. Scale bar, 10 m. F-H, Superimposed traces of time-course of
ROGB-1/RFR. Each colored line represents ROGB-1/RFR averaged within the ROI in a single growth cone that was exposed to Sema3A (F, G) or PBS (H) in the absence (F, H; no drug) or presence of bath-applied MDC (G). The x-axis represents time after the onset of gradient application. I-K, The peak Ca2+ amplitude before (pre) and after the application of Sema3A (I, J) or PBS (K) in the absence (I, K; no drug) or presence of MDC (J). ***P < 0.001, paired t-test.
L-N, Time-lapse DIC images of growth cones exposed to Sema3A gradients (arrows) in the absence (L; control) or presence of the indicated drugs (M, N). Time (min) after the onset of Sema3A application is shown. Scale bar, 20 m. O, Turning angles of growth cones, with positive and negative values indicating attraction and repulsion, respectively. Numbers in parentheses indicate the number of growth cones examined. ***P < 0.001, Bonferroni’s multiple comparison test.
55
56
Figure 15. MAG-induced repulsion but not attraction requires clathrin-mediated endocytosis downstream of Ca2+ signals
A-G, MAG-generated Ca2+ signals in growth cones. A, A growth cone loaded with OGB-1 and FR was exposed to a MAG gradient (arrow). The left panel shows FR fluorescence. The red circle indicates the ROI for Ca2+ measurements. The pseudo-color shows ROGB-1/RFR), a measure of changes in Ca2+ concentrations from the basal level. Time (s) after the onset of MAG application is shown. Scale bar, 10 m. B, C, Superimposed traces of time-course of ROGB-1/RFR. Each colored line represents ROGB-1/RFR averaged within the ROI in a single growth cone that was exposed to MAG in the absence (B; no drug) or presence of bath-applied MDC (C). The x-axis represents time after the onset of MAG application. D-G, The peak Ca2+ amplitude before (pre) and after the application of MAG in the absence (D; no drug) or presence of the indicated drugs (E-G). *P < 0.05, **P < 0.01, paired t-test. H-L, Time-lapse DIC images of growth cones exposed to MAG gradients (arrows) in the absence (H; control) or presence of the indicated drugs (I-L). Time (min) after the onset of MAG application is shown. Scale bar, 20 m. M, Turning angles of growth cones. Numbers in parentheses indicate the number of growth cones examined.
*P < 0.05, **P < 0.01, Bonferroni’s multiple comparison test.