Ni-NTA was from Qiagen (Valencia, CA). IPL-41 and pluronic F-68 for growing Sf9 cells were from Invitrogen (Carlsbad, CA). Mutagenesis primers were from Integrated DNA Technologies (Coralville, IA). Cloning and mutagenesis reagents, Protein A/G-Sepharose, and FastAP alkaline phosphatase were from Thermo Scientific. Anti-FLAG affinity resin (M8823), 3X FLAG peptide (F4799), and anti-FLAG antibody (F7425) were from Sigma. Anti-Src antibody was from Santa Cruz (sc-18). Anti-phosphotyrosine 4G10 Platinum antibody was from Millipore. Phosphatidylinositol 4,5-bisphosphate (PIP₂) and brominated phosphatidylcholine (PC), were from Avanti Polar Lipids. [γ-³²P]GTP and [γ-³²P]ATP were from PerkinElmer Life Sciences. Reagents for electrophoresis and immunoblotting were from Bio-Rad. Fluorescently labeled secondary antibodies for Infrared Imaging System were from LICOR. Other reagents, including anti-myc antibodies, buffers, charcoal-activated Norit A, GSH-Agarose, glutathione, ATP, GTP, protease inhibitors and PMSF were from Sigma (St. Louis MO). pCMV5 mouse Src was from Addgene (plasmid #13663, provided by Joan Brugge and Peter Howley).

Mutations were introduced into rat DNM2 isoform 1 (also known as isoform “ba” or “IIBA”; UniProtKB identifier P39052-1) using the QuikChange site-directed mutagenesis kit (Stratagene). cDNAs encoding wild-type DNM2 (DNM2^WT) and the DNM2^K562E and DNM2^A618T mutants containing C-terminal His₆ tags were subcloned into BacPAK9 plasmids (Clontech) and purified as described previously (Lin et al., 1997). The DNM2^{Δ DEE} mutant was generated with a C-terminal FLAG tag in a Fast-Bac vector. To purify this construct, infected Sf9 cells were homogenized in buffer A (20 mM HEPES (pH 7.5), 300 mM NaCl, 3 mM MgCl₂, 0.5 mM DTT, 1 mM EDTA, and 0.2 mM PMSF) supplemented with a protease inhibitor cocktail consisting of 10 μg/ml each of N-p-tosyl-L-lysine chloromethyl ester, N-p-tosyl-L-arginine methyl ester, N-p-tosyl-L-lysine chloromethyl ketone, leupeptin, and pepstatin A (without phosphatase inhibitors). The homogenate was centrifuged, and the supernatant was mixed with FLAG resin for 1.5 h, washed, and eluted with 3x-FLAG peptide. Purified dynamins were dialyzed against buffer A. Aliquots of purified proteins were frozen in liquid nitrogen and stored at −70°C. Immediately before use, the dynamin solutions were centrifuged at 213,000 × g for 15 min to remove aggregates.

Rat DNM2 PHD, comprising residues 510–620, was subcloned into a pGEX-KG vector using full length DNM2 as a template. GST-PHD was expressed in BL21 E. coli cultured for 4 h at 37°C in the presence of 0.5 mM IPTG. Cells were lysed in PHD-lysis buffer (20 mM HEPES (pH 7.5), 100 mM NaCl, 1 mM BME, 0.2 mM PMSF, and the protease inhibitor cocktail described above, and cell suspensions were centrifuged at 35,000 rpm in a Ti45 rotor for 30 min at 4°C. Supernatants were incubated for 1 h at 4°C with GSH-agarose beads, the beads were washed in PHD-lysis buffer containing 1 M NaCl, then eluted with PHD-lysis buffer containing 15 mM glutathione. Eluted protein was dialyzed overnight in lysis buffer. Mutations were introduced using the QuikChange mutagenesis kit using the WT PHD as template.

Constructs were generated by cloning dynamin cDNAs into vectors containing FLAG, myc, or EGFP tags for C-terminal positioning. cDNAs encoding wild-type DNM2 (DNM2^WT) and the DNM2^K562E, and DNM2^A618T mutants, all with C-terminal His₆ tags, were used as templates. The DNM2^{Δ DEE} mutant was generated by introducing the deletion mutation into FLAG, myc, or EGFP-tagged DNM2^WT templates using the QuikChange site-directed mutagenesis kit (Stratagene).

GTPase activities were measured by the release of ³²P_i from [γ-³²P]GTP following incubation at 37°C. Assay solutions contained 20 mM HEPES (pH 7.5), 2 mM MgCl₂, 1 mM GTP, and 100 mM NaCl. Reactions were initiated by addition of MgGTP and terminated by addition of 5% Norit-A activated charcoal in 50 mM NaH₂PO₄ at 4°C (Higashijima et al., 1987). All procedures involving radioactivity were carried out using protocols approved by the UT Southwestern Office of Safety and Business Continuity.

Wild-type (WT) and mutant forms of DNM2 in 300 mM NaCl were placed in quartz cuvettes and diluted with pre-warmed (37°C) 20 mM HEPES (pH 7.5) to obtain final concentrations of 1 μM dynamin and 50 mM NaCl. Self-assembly was monitored by measuring absorbance at 350 nm at 37°C and 15 s increments using a Beckman DU 650 spectrophotometer.

Liposome binding was performed by sedimentation assay according to Lee and Lemmon (Lee and Lemmon, 2001) using a brominated-PC:PIP₂ molar ratio of 97:3. Lipids were dissolved in chloroform, then dried under a stream of nitrogen followed by overnight drying under vacuum. Dried lipids were resuspended in 20 mM HEPES (pH 7.4) and 100 mM NaCl, followed by 10 freeze/thaw cycles in liquid nitrogen and bath sonication. To extract unilamellar vesicles and remove aggregated lipids, liposomes were extruded 10 times through 0.1 μm filters using a Mini–Extruder (Avanti Polar Lipids). Immediately prior to incubation with liposomes, solutions containing PHDs were centrifuged for 1 h at 265,000 × g at 25°C in a TL-100 tabletop ultracentrifuge. Binding assays were carried out by incubating 5 μM PHD with liposomes for 15 min at 25°C in 20 mM HEPES (pH 7.4) and 50 mM NaCl. Samples (80 μl) were then centrifuged at 214,000 × g for 1 h at 25°C. Supernatants were removed and pellets were resuspended in initial sample volumes. Equal volumes of samples before centrifugation and resuspended pellets were electrophoresed, proteins were visualized by Coomassie blue staining and quantified by LiCOR Odyssey system scanning.

Brightness analysis relies on the fact that mobile proteins with multiple attached fluorophores (e.g., tetramers) may be distinguished from those containing a single fluorophore (monomer) by analysis of the fluctuations of the average intensity. This principle is illustrated in Supplementary Figure 1. We note that the method requires comparison of the fluctuations, taken under identical instrument settings, from a monomeric brightness standard – in our case EGFP. Hence, by normalizing the measured brightness with the brightness of the monomer standard, the oligomerization state of our target protein can be quantified. The instrumentation for brightness analysis fluorescence fluctuation measurements was described previously (Chen et al., 2003). An excitation wavelength of 1,000 nm (for two-photon excitation) was used for all experiments. Brightness is calculated with Q-analysis (Sanchez-Andres et al., 2005). Monomer brightness of EGFP was obtained by averaging over 5 cells at various concentrations. The resulting normalized molecular brightness, b, is calculated by taking the brightness of an individual measurement divided by the brightness of monomeric EGFP.

For Raster Scan Image Correlation Spectroscopy (RICS) analysis, NIH3T3 cells plated on fibronectin-coated imaging dishes were transfected with plasmids encoding EGFP-tagged dynamins using Lipofectamine 3000. Cells were imaged 20–26 h after transfection at room temperature for a maximum duration of 90 min with an Olympus FV1000 confocal microscope set up for RICS. Fluorescence was excited with 488-nm light and green fluorescence was detected in a band of 505–525 nm with a 60×, NA 1.2 water immersion lens. For each RICS data set, 100 frames of 256 × 256 pixels were acquired with a pixel dwell time of 10 μs and a line time of 3.68 ms, pixel size was 50 nm, the waist (w₀) of the point spread function was 270 nm. Before raster image correlation, a moving average of 2 frames was subtracted to remove the immobile fraction. Data were analyzed using SimFCS software.

To monitor DNM2 phosphorylation in cells, HeLa cells transfected with Lipofectamine-2000 were incubated with 0.1 mM sodium orthovanadate for 1 h before lysis in RIPA buffer (50 mM Tris, pH 8.0; 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 5 mM EDTA, 0.1% SDS, 50 mM glycerophosphate) containing protease inhibitor cocktail (see above) and 1 tablet/10 ml buffer of Phosphatase Inhibitor Cocktail Tablets (Roche). Lysates were centrifuged at 1000 × g for 10 min at 4°C and resulting supernatants were incubated for 4 h with anti-FLAG antibody conjugated to agarose beads or anti-myc antibodies followed by protein A/G Sepharose to immunoprecipitate dynamins. Samples were washed 3 times with RIPA buffer supplemented with protease and phosphatase inhibitors (as above), then electrophoresed and blotted with anti-FLAG or anti-myc and anti-phosphotyrosine (4G10) antibodies. Immunoblots were scanned and quantified by LiCOR Odyssey. To phosphorylate DNM2 in vitro, FLAG tagged DNM (WT or mutant) were expressed in HEK293 cells and immunoprecipitated with anti-FLAG antibody conjugated to agarose beads in the absence of SDS or phosphatase inhibitors. To obtain non-phosphorylated dynamin, the beads were additionally incubated overnight at 4°C with 15 units/ml FastAP alkaline phosphatase (AP). The beads were then washed extensively with a solution containing 20 mM HEPES, pH 7.5, 150 mM NaCl, and 0.1% Triton to remove AP, and incubated with v-Src in the presence of 10 mM MgCl₂ and 0.5 mM ATP. Samples were electrophoresed, blotted with anti-FLAG and anti-phosphotyrosine antibodies, and scanned and quantified by LiCOR Odyssey.

Cells were grown in Dulbecco’s modified Eagle’s medium containing 10% fetal calf serum and antibiotics at 37°C in 5% CO₂. Protein concentration was determined using the modified Lowry method (Lowry et al., 1951) according to Peterson (Peterson, 1977) with BSA as a standard. SDS-PAGE was carried out according to Laemmli (Laemmli, 1970).

Lysines that contribute to the interaction of dynamins with PIP₂ are clustered in variable loop 2 (VL2) and β-strand 4 (β4) of the PHD (Figure 1A). The best studied CMT mutation in DNM2, K562E, is located in strand β4 and essentially eliminates PIP₂ binding and GTPase activation (Kenniston and Lemmon, 2010). The CMT-linked mutant examined in this study, DNM2^{Δ DEE}, contains a deletion of residues ⁵⁵⁵DEE⁵⁵⁷ in VL2. This deletion increases the electrostatic potential of VL2 (Figures 1B,C), potentially enhancing the association of the PHD with the negatively charged PIP₂ headgroup. However, we found that the ΔDEE mutation did not alter PHD binding to small unilamellar vesicles composed of phosphatidylcholine (PC) and PIP₂ (97:3 molar ratio) (Figure 1D). As expected, the K562E mutation prevented PHD binding to PC/PIP₂ vesicles, whereas the A618T CNM mutation had no effect.

At low ionic strength, dynamins self-assemble in vitro into rings and coils (Hinshaw and Schmid, 1995) resembling those that form around the necks of budding vesicles (Van der Bliek et al., 1993; Takei et al., 1995). Self-assembly stimulates dynamin GTPase activities from basal levels of ∼1–10 min^–1 to activated levels of ∼100–200 min^–1. Disassembly of in vitro-formed dynamin polymers occurs rapidly upon increase of ionic strength or addition of GTP (Warnock et al., 1996), presumably reflecting the GTPase-associated disassembly of dynamin rings that accompanies membrane scission in cells. In a prior study, we found that introduction of CNM-linked mutations R365W, E368K, and A618T stabilizes DNM2 polymers against GTP- and salt-induced disassembly (Wang et al., 2010). Structural studies suggest that the mutations weaken the PHD-stalk interface and, hence, would favor the open, assembly competent DNM2 conformation. Like most other residues mutated in CMT, the ⁵⁵⁵DEE⁵⁵⁷ motif is located on the opposite surface of the PHD to the C-terminal helix that contributes to the PHD-stalk interface (Figures 2A,B). Therefore, we were surprised to find that DNM2^{Δ DEE} showed almost identical resistance to GTP- and salt-induced disassembly as DNM2^A618T (Figure 2C). In contrast, the assembly properties of the CMT mutant DNM2^K562E were similar to those of DNM2^WT. Consistent with the enhanced self-assembly induced by the ΔDEE mutation, GTPase activation of DNM2^{Δ DEE} occurred at lower DNM2 concentrations and displayed higher maximal levels than DNM2^WT (Figure 2D). These results demonstrate that at least one CMT-linked mutation, ΔDEE, confers the gain-of-function in vitro properties that were hitherto observed only with CNM-linked mutations, and raise the possibility that the simple “stalk-occlusion” mechanism for auto-inhibition of dynamin polymerization by the PHD may be an oversimplification.

We next asked whether the ΔDEE mutation affects the self-assembly of DNM2 in living cells, as it does in vitro. We previously reported that DNM2^WT-EGFP, at concentrations of ∼2 μM or higher, is predominantly tetrameric in the cytosol (James et al., 2014), in agreement with its reported in vitro unassembled oligomeric state as measured by analytical ultracentrifugation (Muhlberg et al., 1997). However, the CNM-linked DNM2^R369W mutant forms higher-order cytosolic oligomers, up to 12-14-mers, at concentrations of ∼1 μM or higher (James et al., 2014). Using a refined version of the fluorescence fluctuation spectroscopy (FFS) approach known as brightness analysis (Qian and Elson, 1990; Sanchez-Andres et al., 2005) (see Materials and Methods and Supplementary Figure 1), as in our previous study, we found that EGFP-tagged DNM2^WT, DNM2^K562E, and DNM2^{Δ DEE} are predominantly present in an approximately tetrameric state in the cytosol of living U2OS cells (Figures 3A,B). In contrast, the CNM-linked mutant DNM2^A618T-EGFP assembled into species as high as 20–24-mers (Figure 3C). Using a distinct FFS approach known as raster scan image correlation spectroscopy (RICS) (Digman et al., 2005), which allows measurement of diffusion in every region of the cell, we obtained average diffusion coefficients (inversely proportional to particle radius) of 1.6 ± 0.4 μm²/s, 1.2 ± 0.3 μm²/s, and 0.5 ± 0.4 μm²/s (mean ± standard deviation) for DNM2^WT-EGFP, DNM2^{Δ DEE}-EGFP, and DNM2^A618T-EGFP, respectively (Figure 3D). Although DNM2^{Δ DEE}-EGFP moves somewhat more slowly than DNM2^WT-EGFP, these results are in general agreement with the N&B data, as a dynamin tetramer has a radius of ∼ 7–8 nm (Muhlberg et al., 1997), whereas dynamin rings (containing 26–28 monomers) are oblate ellipsoids with radii of ∼ 25 nm (Hinshaw and Schmid, 1995). Taken together, our results suggest that the ΔDEE mutation did not induce the formation of high-order DNM2 oligomers in the cytosol, despite its ability to stabilize DNM2 polymers in vitro.

To date, the most extensively characterized CMT mutant is DNM2^K562E, which is defective in PIP₂ binding. Having established that the ΔDEE mutation does not affect the PHD-PIP₂ interaction, we examined its role in DNM2 tyrosine phosphorylation, which also involves the PHD. Src-mediated tyrosine phosphorylation positively regulates dynamin function in receptor-mediated endocytosis (Ahn et al., 2002; Shajahan et al., 2004; Cao et al., 2010), Golgi budding (Weller et al., 2010), focal adhesion dynamics (Wang et al., 2011), and actin remodeling (Lin et al., 2020). Although the mechanism(s) underlying dynamin regulation by tyrosine phosphorylation are not yet understood, this modification was shown to enhance the self-assembly and GTPase activation of DNM1 in vitro (Ahn et al., 2002). We found that DNM2^{Δ DEE} underwent tyrosine phosphorylation to ∼ double the levels of DNM2^WT, DNM2^K562E, or DNM2^A618T in cells co-expressing c-Src (Figure 4A) and that DNM2^{Δ DEE} is phosphorylated to ∼ double the level of DNM2^WT by c-Src in vitro (Figure 4B). Thus, enhancement of tyrosine phosphorylation is apparently a direct effect of the ΔDEE mutation on the conformation of DNM2.

Two Src-dependent phosphorylation sites have been identified in dynamins, Y231 in the GTPase domain and Y597 in the PHD (Ahn et al., 2002) (see Figure 2B). To test whether increased phosphorylation of these two residues accounts for the hyperphosphorylation of DNM2^{Δ DEE}, we introduced Y231F and Y597F mutations, individually and in combination, into DNM2^WT and DNM2^{Δ DEE} and co-transfected these constructs with c-Src into HeLa cells. As expected, tyrosine phosphorylation of DNM2^WT was greatly reduced by introduction of Y231F/Y597F mutations (Figure 4C). Likewise, the Y231F/Y597F mutations diminished tyrosine phosphorylation of DNM2^{Δ DEE}, although to a somewhat lesser extent than occurred with DNM2^WT.

To determine if the ΔDEE mutation enhances the binding of DNM2 to Src, we compared Förster resonance energy transfer (FRET) between Src-mCherry and DNM2^WT-EGFP or DNM2^{Δ DEE}-EGFP in NIH-3T3 cells. Binding of Src to dynamin had previously been detected using gel overlay and co-immunoprecipitation assays (Foster-Barber and Bishop, 1998), but the interaction had never been examined in live cells. Using fluorescence lifetime imaging to monitor FRET, we observed that DNM2^WT binds to Src in cells (Supplementary Figure 2A) and demonstrated that the ΔDEE mutation had no discernible effect on the binding (Supplementary Figure 2B). As controls, FRET was not observed when DNM2^WT-EGFP and DNM2^{Δ DEE}-EGFP were expressed in the absence of an acceptor fluorophore (Supplementary Figure 3) or when EGFP and mCherry were co-expressed (Supplementary Figure 4). These results suggest that the increase in tyrosine phosphorylation of DNM2^{Δ DEE} is apparently not due to an increase in its affinity for the kinase.