Postprenylation CAAX processing is required for proper localization of Ras but not Rho GTPases

Abstract

The CAAX motif at the C terminus of most monomeric GTPases is required for membrane targeting because it signals for a series of three posttranslational modifications that include isoprenylation, endoproteolytic release of the C-terminal- AAX amino acids, and carboxyl methylation of the newly exposed isoprenylcysteine. The individual contributions of these modifications to protein trafficking and function are unknown. To address this issue, we performed a series of experiments with mouse embryonic fibroblasts (MEFs) lacking Rce1 (responsible for removal of the -AAX sequence) or Icmt (responsible for carboxyl methylation of the isoprenylcysteine). In MEFs lacking Rce1 or Icmt, farnesylated Ras proteins were mislocalized. In contrast, the intracellular localizations of geranylgeranylated Rho GTPases were not perturbed. Consistent with the latter finding, RhoGDI binding and actin remodeling were normal in Rce1- and Icmt-deficient cells. Swapping geranylgeranylation for farnesylation on Ras proteins or vice versa on Rho proteins reversed the differential sensitivities to Rce1 and Icmt deficiency. These results suggest that postprenylation CAAX processing is required for proper localization of farnesylated Ras but not geranygeranylated Rho proteins.

Figure 1.

Postprenylation CAAX processing is required for proper localization of all three Ras isoforms. (A) MEFs expressing (+/+) or deficient (–/–) in Icmt or Rce1 were transfected with GFP-tagged forms of H-, N- and K-Ras as indicated and imaged 24 h after transfection as described in Materials and Methods. (B) Constitutively active Ras mutants H-Ras61L and K-RasV12 were expressed and imaged as in A. Representative images are shown of >10 acquired on each of more than five independent experiments. Bars, 10 μm.

Figure 6.

Rac1 and Cdc42 do not require postprenylation processing to regulate actin remodeling. (A) MEFs expressing (+/+) or deficient (–/–) in Icmt or Rce1 were transfected with wild-type or constitutively active GFP-Rac1 or GFP-Cdc42. Twenty-four hours after transfection, cells were paraformaldehyde fixed, saponin permeabilized, and stained with Texas Red-phalloidin to image F-actin. Dual color images are shown (GFP in green and F-actin in red). Rac1-induced lamellipodia and Cdc42-induced fillopodia were equally evident in each cell type. (B) MEFs expressing (+/+) or deficient (–/–) in Icmt or Rce1 were transfected with wild-type GFP-Rac1 with or without oncogenic Dbl to induce ruffling. (C) Prevalence of ruffling shown in A and B among 100 randomly chosen cells transfected with the indicated constructs. (D) Prevalence of filopodia formation as shown in A among 50 randomly chosen cells transfected with the indicated constructs. Bars, 10 μm. Graphs show mean ± SEM, n = 3.

Figure 2.

GDI binding of Rac1 is not dependent on postprenylation CAAX processing. MEFs expressing (+/+) or deficient (–/–) in Icmt or Rce1 were transfected with GFP-Rac1. GFP-Rac1 was immunoprecipitated from cell lysates with anti-GFP antibodies and analyzed by SDS-PAGE and ¹²⁵I-protein A immunoblot for GFP and for RhoGDI. (A) Representative coimmunoprecipitation with recombinant RhoGDI (rRhoGDI) as a control run on the same gel (shown as separate panel because of a much lower PhosphorImager exposure time). The identity of the nonspecific band running slightly higher than RhoGDI is not known. (B) Cumulative data where the ability of RhoGDI to interact with GFP-Rac1 is quantified by coimmunoprecipitation as the PhosphorImager volumes of [GDI]/[GFP-Rac1], and the data are normalized across experiments by assigning a value of 1 to ratios from Icmt +/+ cells. Bars shown are mean ± SEM, n = 6.

Figure 3.

GFP-Rho proteins do not require postprenylation CAAX processing for membrane localization. MEFs expressing (+/+) or deficient (–/–) in Icmt or Rce1 were transfected with constitutively active forms of GFP-Rac1, GFP-Cdc42, GFP-RhoA, or wild-type GFP-RhoB (which does not interact with RhoGDI). Constitutively active forms of RhoGDI binding proteins were used to block binding to RhoGDI and thus reveal intrinsic membrane targeting. Cells were imaged 24 h after transfection as described in Materials and Methods. Representative images are shown of >10 acquired on each of more than five independent experiments. Bars, 10 μm.

Figure 4.

Subcellular fractionation confirms differential requirement for carboxyl methylation on membrane association of endogenous farnesylated versus geranylgeranylated proteins. (A) MEFs expressing (+/+) or deficient (–/–) in Icmt were metabolically labeled with [³⁵S]methionine, disrupted with a Dounce homogenizer, and separated into 100,000 × g supernatants (S) and pellets (P). Fractions were extracted with RIPA buffer, immunoprecipitated with a pan-ras antibody, analyzed by SDS-PAGE, and imaged by PhosphorImager. (B) The same cells without metabolic labeling were used to generate identical S100 and P100 fractions that were analyzed for endogenous RalA by ¹²⁵I-protein A immunoblots. RalA was chosen as a strictly geranylgeranylated Ras-related GTPase that lacks a cytosolic chaperon and is expressed at sufficient levels to be detected in MEFs. PhosphorImager analysis revealed 4.3 and 4.8% of total RalA and 2 and 30% of total Ras in the S100 fraction in +/+ versus –/– cells.

Figure 5.

Rho family proteins are substrates for carboxyl methylation. (A) In vitro methylation: endogenous Rac2, Cdc42, and RhoA, partially purified as 1:1 complexes with RhoGDI from human neutrophil cytosol by ion exchange chromatography and gel filtration, were incubated for 30 min with 10 μg of human neutrophil light membranes containing endogenous Icmt and the methyl donor [³H]AdoMet with or without GTPγS, and the reaction products were analyzed by SDS-PAGE and fluorography. (B) In vivo methylation: ECV304 cells stably expressing GFP-Cdc42, GFP-Rac1, or GFP-RhoA were transfected with pCMV or pCMV-Dbl (exchange factor for Rho proteins) and then metabolically labeled with [³H]methyl-methionine (precursor of AdoMet) and [³⁵S]methionine. GFP-tagged proteins were immunoprecipitated with anti-GFP antibodies, analyzed by SDS-PAGE, and visualized as [³⁵S]methionine-labeled proteins by PhosphorImager. Immunoprecipitated proteins were excised from the gel and analyzed for carboxyl methylation by alkaline hydrolysis. Numbers below the bands indicate cpm for base-releasable [³H]methanol (background 25 cpm) The experiment shown is representative of four independent experiments.

Figure 7.

Postprenylation processing is not required for integrin-induced actin remodeling. MEFs expressing (+/+) or deficient (–/–) in Icmt or Rce1 were incubated with BSA- or fibronectin-coated 1-μm latex beads and then paraformaldehyde fixed, saponin permeabilized, and stained with Texas Red-phalloidin. Subplasmalemmal F-actin formation was visualized by wide-field epifluorescence microscopy and scored as a ring of fluorescence surrounding fibronectin- but not BSA-coated beads. The positions of the beads were revealed by differential interference contrast microscopy.

Figure 8.

A prenyl swap reverses the differential sensitivity of Ras and Rho proteins to mislocalization in Icmt- and Rce1-deficient cells. MEFs expressing (+/+) or deficient (–/–) in Icmt or Rce1 were transfected with GFP-tagged Ras and Rho proteins with mutations in their CAAX sequences that direct alternative prenylation (i.e., geranylgeranylation of Ras proteins and farnesylation of Rho proteins). These constructs included GFP-N-Ras-CVLL, GFP-K-Ras-CVIL, GFP-RhoB-CAIM, and GFP-Rac1L61-CVLS.

Figure 9.

Farnesylated Rac1 requires complete CAAX processing to regulate actin remodeling. (A) MEFs expressing (+/+) or deficient (–/–) in Icmt or Rce1 were transfected with GFP-Rac161L-CVLS, a constitutively active, farnesylated form of the GTPase. Twenty-four hours after expression, the cells were analyzed for lamellipodia formation as in Figure 6. (B) MEFs expressing (+/+) or deficient (–/–) in Icmt or Rce1 were transfected with GFP-Rac1 with or without oncogenic Dbl to induce ruffling. (C) Prevalence of ruffling shown in A and B among 50 randomly chosen, transfected cells. Bars, 10 μm. Graphs show mean ± SEM, n = 3.