Structural and functional characterization of the dominant negative P-loop lysine mutation in the dynamin superfamily protein Vps1

Abstract

Dynamin-superfamily proteins (DSPs) are large self-assembling mechanochemical GTPases that harness GTP hydrolysis to drive membrane remodeling events needed for many cellular processes. Mutation to alanine of a fully conserved lysine within the P-loop of the DSP GTPase domain results in abrogation of GTPase activity. This mutant has been widely used in the context of several DSPs as a dominant-negative to impair DSP-dependent processes. However, the precise deficit of the P-loop K to A mutation remains an open question. Here, we use biophysical, biochemical and structural approaches to characterize this mutant in the context of the endosomal DSP Vps1. We show that the Vps1 P-loop K to A mutant binds nucleotide with an affinity similar to wild type but exhibits defects in the organization of the GTPase active site that explain the lack of hydrolysis. In cells, Vps1 and Dnm1 bearing the P-loop K to A mutation are defective in disassembly. These mutants become trapped in assemblies at the typical site of action of the DSP. This work provides mechanistic insight into the widely-used DSP P-loop K to A mutation and the basis of its dominant-negative effects in the cell.

Figure 1

C. therm ^K56AVps1 GTPase‐BSE does not dimerize in the presence of GDP.AlFx. (a,b) Analytical size exclusion chromatography of (a) C. therm Vps1 GTPase‐BSE and (b) C. therm ^K56AVps1 GTPase‐BSE in the absence or presence of GDP.AlFx. The elution volume decreased in the presence of GDP.AlFx for wild type Vps1 GTPase‐BSE, but not ^K56AVps1 GTPase‐BSE, consistent with dimerization. (c) Determination of the absolute molecular mass of C. therm ^K56AVps1 GTPase‐BSE in the absence or presence of various nucleotides or nucleotide analogs by SEC‐MALS. The theoretical molecular masses of monomers and dimers are shown as dotted grey lines. The measured molecular masses in each case are shown on the trace. No changes in molecular mass were observed

Figure 2

C. therm ^K56AVps1 GTPase‐BSE binds to GMPPCP with a similar affinity as wild type Vps1 GTPase‐BSE, as determined by ITC. Binding of GMPPCP to C. therm Vps1 (a) GTPase‐BSE and (b) C. therm ^K56AVps1 GTPase‐BSE. (c) Binding of GDP to C. therm ^K56AVps1 GTPase‐BSE. In each case, upper panels show the thermograms for each titration. DP—differential power. Lower panels show the normalized binding isotherms obtained from the integrated peaks from the thermograms (filled circles), together with the fit obtained from a single site binding model (lines). Mean measured K _D values are shown on the isotherms

Figure 3

The crystal structure of C. therm ^K56AVps1 GTPase‐BSE in complex with GMPPCP. (a) Ribbon diagram of a C. therm ^K56AVps1 GTPase‐BSE dimer, in complex with GMPPCP. The BSEs are shown in lighter and darker purples and the GTPase domains are shown in shades of mint. GMPPCP is shown in stick representation, with the bound magnesium ion as a sphere. (b) 2mF_obs‐DF_calc density map of one of the GMPPCP molecules in the asymmetric unit, contoured at 1σ. The GMPPCP is shown in stick representation and the bound magnesium with its associated waters are shown as spheres. (c) Comparison of the nucleotide binding pocket of Vps1 GTPase‐BSE bound to GMPPCP (PDB 6DEF) and ^K56AVps1 GTPase‐BSE bound to GMPPCP. For wild type Vps1 GTPase‐BSE, the GTPase domain is shown in blue. For ^K56AVps1 GTPase‐BSE, the same color scheme is used as in A. (d) 2mF_obs‐DF_calc density maps observed around K56 in Vps1 GTPase‐BSE (PDB 6DEF, top) and K56A in ^K56AVps1 GTPase‐BSE (bottom). Maps were contoured at 1σ. Note the absence of H2O_cat and H2O_bridge in the K56A mutant. H2O_bridge was not visible in any of the molecules in the asymmetric unit, whereas weak density for H2O_cat could be built as a water in two of the ^K56AVps1 GTPase‐BSE molecules

Figure 4

S cer. ^K42AVps1 assembles into few, intense puncta on the perivacuolar compartment in wild type (W303A) and Δvps1 cells. (a) When expressed in W303A or Δvps1 cells, S cer. Vps1‐EGFP distributed between the cytosol and dynamic perivacuolar puncta. (b) S cer. ^K42AVps1‐EGFP concentrated to few intense puncta at the perivacuolar compartment. Note that in Δvps1 cells, very little residual cytosolic staining was observed. Vacuoles were stained with FM 4–64. Scale bar—5 μm

Figure 5

Overexpressed S cer. Vps1 GTPase‐BSE or S cer. ^K42AVps1 GTPase‐BSE is cytosolic. (a) S cer. Vps1 GTPase‐BSE‐EGFP expressed in W303A or Δvps1 cells was fully cytosolic. Note the Class F vacuolar phenotype in the Δvps1 cells. (b) S cer. ^K42AVps1 GTPase‐BSE‐EGFP expressed in W303A or Δvps1 cells was similarly cytosolic, in both W303A and Δvps1 cells. Scale bar—5 μm

Figure 6

S cer. ^K41ADnm1 assembles into few, intense puncta on the mitochondrial network in Δdnm1 cells. Δdnm1 cells were labeled with mitochondrial matrix‐targeted DsRed (Mito‐DsRed). S cer. Dnm1‐EGFP expressed in Δdnm1 cells restored the mitochondrial reticular morphology observed in W303A cells and largely localized to small puncta on the surface of the mitochondrial network. S cer. ^K41Dnm1‐EGFP failed to rescue the collapsed, netted, mitochondrial network and localized to intense puncta on the surface of the mitochondrial network. Scale bar—5 μm