Activation of Rho GTPases in Smith–Lemli–Opitz syndrome: pathophysiological and clinical implications

Increased cofilin-1 phosphorylation in Dhcr7^Δ3-5/Δ3-5 brain tissue

In order to identify biological pathways that are altered in SLOS, we used proteomic techniques to identify differentially expressed proteins in Dhcr7^Δ3-5/Δ3-5 brain tissue. Total protein extracts were obtained from seven E18.5 Dhcr7^Δ3-5/Δ3-5 mutant embryos and seven Dhcr7^+/+ control embryos. Two-dimensional gel electrophoresis (2-DE) was used to separate proteins. Three replicate silver stained, two-dimensional gels were produced for both Dhcr7^+/+ and Dhcr7^Δ3-5/Δ3-5 total brain protein extracts. For comparison of total protein extracts from Dhcr7^+/+ and Dhcr7^Δ3-5/Δ3-5 brains, 1107 individual protein spots were quantified using PDQuest 8.0 software. Fifty-six (5.1%) of these protein spots were differentially expressed (R ≤ 0.83 or R ≥ 1.20, P < 0.05, unpublished data). Differentially expressed protein spots were excised from Coumassie blue stained preparative gels, and subjected to ‘in-gel’ digestion with trypsin. The resulting tryptic peptides were extracted and analyzed by a combination of high resolution MALDI tandem mass spectrometry and liquid chromatography/tandem mass spectrometry. MS/MS spectra from both instruments were analyzed and matched to protein sequences present in the Swiss-Prot database using the Mascot search program (22).

Figure 1B shows a portion of representative silver stained gels comparing proteins from Dhcr7^+/+ and Dhcr7^Δ3-5/Δ3-5 brains. Densitometric quantification of the indicated protein spot in three independent gels showed a 3.8-fold increase (P = 0.005) in mutant brain tissue compared with control brain tissue. Mass spectrometric analysis of this protein spot identified three peptides corresponding to cofilin-1 (Supplementary Material, Fig. S1). Western blot analysis was used to validate altered expression of cofilin-1 in Dhcr7 mutant brains. Whereas total cofilin-1 levels did not appear to differ between Dhcr7^+/+ and Dhcr7^Δ3-5/Δ3-5 brains, levels of P-cofilin-1 were significantly increased in E18.5 Dhcr7^Δ3-5/Δ3-5 brain tissue compared with control brain tissue (Fig. 1C). Analysis of brain tissue from five individual mutant and control embryos (Fig. 1D) showed a 5.8-fold increase in P-cofilin-1 levels (P = 0.02), and confirmed no significant differences in total cofilin (phosphorylated and non-phosphorylated, P = 0.30). Cofilin-1 levels are relatively constant from E12.5 through E18.5, but P-cofilin-1 levels vary during development. However, P-cofilin-1 levels are elevated in Dhcr7^Δ3-5/Δ3-5 brain tissue compared with control levels at all ages examined (Fig. 1E).

Activation of the Rho/Rac/Cdc42 GTPases and Lim kinase in Dhcr7^Δ3-5/Δ3-5 brain tissue

Two homologous proteins, Lim kinase-1 and Lim kinase-2 (Limk), regulate cofilin activity by phosphorylation of serine 3. Limk-2 is ubiquitously expressed; however, Limk-1 expression is restricted to neuronal tissue. Both Limk-1 and Limk-2 are activated by threonine phosphorylation at amino acid position 505 and 508, respectively. Protein levels of Limk-1 and Limk-2 are similar in Dhcr7^+/+ or Dhcr7^Δ3-5/Δ3-5 embryonic brain tissue (Fig. 2A). However, consistent with increased phosphorylation of cofilin, we found increased levels of phosphorylated Limk in E18.5 brains from Dhcr7^Δ3-5/Δ3-5 embryos compared with tissue from Dhcr7^+/+ embryos (Fig. 2A). Quantification of results showed no significant differences in total Limk-1 (P = 0.66) or Limk-2 (P = 0.74) protein levels; however, P-Limk-1/2 levels were significantly (P < 0.005) increased in SLOS brain tissue (Fig. 2B). The antibody used to detect phosphorylation of Limk does not differentiate between Limk-1 and Limk-2. The ratio of P-Limk 1/2 relative to either Limk-1 (P < 0.01) or Limk-2 (P < 0.05) was significantly elevated in mutant brain tissue.

Activation of Limk is mediated by Rho GTPases and their effector kinases (20). To understand further the etiology of increased P-cofilin-1 and P-Limk-1/2, we analyzed the activity state of the Rho-GTPase/Rho kinase, Rac-GTPase/p21 associated kinase and Cdc42/p21-activated kinase signaling cascades. Limk can be phosphorylated by p21 associated kinases (Pak), and Pak family members are activated when phosphorylated. Pak protein levels did not significantly differ (P = 0.45) in brain tissue from Dhcr7^+/+ or Dhcr7^Δ3-5/Δ3-5 embryos. However, we found a significant (P < 0.05) increase in the level of phosphorylated Pak in brain tissue from Dhcr7^Δ3-5/Δ3-5 embryos (Fig. 2C and D). Phosphorylation of Pak is mediated by Rac1 and Cdc42, which are small GTPases that cycle between an inactive state binding GDP and an active state binding GTP. Neither total levels of Rac1 (Fig. 3A and B) nor total levels of Cdc42 (Fig. 3C and D) immunoreactive material differed significantly (P = 0.30 and P = 0.69, respectively) between E18.5 brain tissue from Dhcr7^+/+ or Dhcr7^Δ3-5/Δ3-5 embryos. However, using Rac-GTP and Cdc42-GTP pull-down assays, we found that both activated Rac and Cdc42 levels were significantly increased (P < 0.001 and P < 0.01, respectively) in brain tissue from Dhcr7^Δ3-5/Δ3-5 embryos compared with control tissue (Fig. 3A–D). We were not able to directly demonstrate the activation of Rock; however, we were able to use Rho-GTP pull-down assays to demonstrate significantly (P < 0.05) increased activation of RhoA (Fig. 3E and F). Similar to what was observed for Rac1 and Cdc42, no significant difference (P = 0.64) in immunoreactive RhoA was observed (Fig. 3E and F). The activation of RhoA, Rac1 and Cdc42 appears to be specific and not due to non-specific activation of small GTPases. A Ras-GTP pull-down assay was used to evaluate the activation status of Ras in Dhcr7 mutant and control embryonic brain tissue. In contrast to what we observed with RhoA, Rac1 and Cdc42, we did not observe increased activation of Ras (Supplementary Material, Fig. S2). Supporting the idea that the activation of the Rho GTPase/LIM kinase regulatory cascade is not due to a non-specific mechanism, we did not observe increased phosphorylation of protein kinase A Cα, a protein that is regulated by G protein-coupled receptors (Supplementary Material, Fig. S3).

In addition to cofilin-1, the Rho/Rac/Cdc42 signaling cascades regulate other cytoskeletal proteins (23). Rock phosphorylates myosin light chain 2 (Mlc2) and regulates myosin/actin interactions. Mlc2 functions in both early and late stages of neuronal morphogenesis (24,25). Consistent with the increased activation of RhoA, analysis of Mlc2 expression showed similar Mlc2 protein levels in Dhcr7^+/+ and Dhcr7^Δ3-5/Δ3-5 E18.5 brain tissue, but significantly increased (3.4-fold, P < 0.005) levels of phosphorylated Mlc2 in Dhcr7^Δ3-5/Δ3-5 tissue (Fig. 4A and B). Pak phosphorylates and inactivates Op18 (Stathmin), and Op18 regulates microtubule assembly (26). Although Op18 protein levels did not differ between control and Dhcr7 mutant embryos, consistent with increased Pak activity, we observed increased phosphorylation (1.8-fold, P < 0.015) of Op18 in brain tissue from E18.5 Dhcr7^Δ3-5/Δ3-5 embryos (Fig. 4C and D).

Altered cofilin phosphorylation is due to cholesterol deficiency

In Dhcr7^Δ3-5/Δ3-5 brain tissue, there is both a deficiency of cholesterol and accumulation of 7DHC (12). Thus, observed abnormalities could result from sterol deficiency or a toxic effect of 7DHC. To separate these two potential mechanisms, we analyzed protein expression in brain tissue from Sc5d^+/+ control and Sc5d^−/− mutant embryos. Sc5d, or lathosterol 5-desaturase, catalyzes the enzymatic step immediately upstream of Dhcr7 and converts lathosterol to 7DHC (Fig. 1A). In Sc5d^−/− embryos, there is a cholesterol deficiency similar to that observed in Dhcr7^Δ3-5/Δ3-5 brains, but lathosterol, rather than 7DHC, is the accumulating sterol (27). Altered cofilin-1 expression was identified in a 2-DE proteomic comparison of brain tissue from E18.5 Sc5d^+/+ and Sc5d^−/− embryos (Fig. 5A and B). Analogous to what was observed in the SLOS mouse model, cofilin-1 protein levels were constant, but cofilin-1 phosphorylation was increased in embryonic brain tissue from the lathosterolosis mouse model (Fig. 5B and C). In addition, consistent with our findings in the Dhcr7 mutant brain tissue, Limk-1/2, Pak, Op18 and Mlc2 phosphorylation were all significantly increased in Sc5d mutant brain tissue (Supplementary Material, Fig. S4). The similar proteomic phenotype in Sc5d^−/− and Dhcr7^Δ3-5/Δ3-5 embryonic brain tissue strongly argues that this specific defect is due to a deficiency of cholesterol rather than a toxic effect of the accumulating intermediate.

In vivo disruption of actin filaments in embryonic Dhcr7^Δ3-5/Δ3-5 brain tissue

Cofilin-1 regulates actin filament formation by severing F-actin filaments and increasing barbed ends that can function as de novo filament nucleation sites. Thus increased phosphorylation of cofilin-1, as shown above, would be predicted to alter actin filament formation in vivo. To evaluate in vivo actin filament formation, we stained embryonic brain sections with rhodamine labeled phalloidin. Phalloidin is a toxin that binds to filamentous actin (F-actin). Thus, increased rhodamine-phalloidin staining is consistent with decreased cofilin-1 activity. Rhodamine-phalloidin stained midsagital brain sections from E18.5 Dhcr7^+/+ and Dhcr7^Δ3-5/Δ3-5 embryos are shown in Figure 6A and B. Quantification of fluorescence intensity in nine anatomical domains determined from analysis of nine midsagital sections from three control and three Dhcr7 mutant embryos showed that total phalloidin staining was increased an average of 1.26-fold (P = 0.005) in brains from Dhcr7^{Δ3-5/Δ3-53-5} embryos compared with controls. We noted regional difference in phalloidin staining, thus we analyzed staining intensity separately for each of the nine anatomical domains (Fig. 6C). After correction for multiple comparisons, we found significantly increased phalloidin staining in the caudate-putamen (1.44-fold, P < 0.05), optic area (1.73-fold, P < 0.001) and the fimbria (1.91-fold, P < 0.01). Although not significant after accounting for multiple comparisons, we noted trends toward increased staining in hippocampus and geniculate area and decreased staining in the cortical plate.

In vitro analysis of dendrite and axonal growth

Rho and Rac/Cdc42 differentially regulate dendritic and axonal growth in the developing central nervous system. Thus, we hypothesized that a perturbation of Rho/Rac/Cdc42 activity would functionally be reflected in dendritic number, dendritic length or axonal length. To test this, we prepared hippocampal neurons from E18.5 Dhcr7^+/+ and Dhcr7^Δ3-5/Δ3-5 and maintained neurons in cholesterol deficient medium. Biochemically, these neurons have markedly reduced cholesterol and increased 7DHC levels (Supplementary Material, Table S1). At 7 days in vitro (DIV7), neurons were transfected with a construct expressing green fluorescent protein. Three days after transfection, neurons were fixed and confocal images were obtained. Although no significant difference in the number of dendrites was present, we observed an increase in the total length of both dendrites and axons in neurons derived from Dhcr7^Δ3-5/Δ3-5 embryos compared with neurons derived from Dhcr7^+/+ embryos (dendrite: increased by 1.47-fold, P = 0.02; axon: increased by 1.53-fold, P = 0.004; Figure 6D and E).

Mouse models

Animal work was performed under an NICHD approved animal study protocol. Dhcr7 and Sc5d mutant mouse models and genotyping have previously been described (12,27). For timed matings, the identification of a copulatory plug was considered to be E0.5 and embryonic age was confirmed by inspection.

Protein sample preparation

Mouse brain tissue samples were rapidly dissected from E18.5 embryos and immediately washed with ice-cold 1× phosphate buffered saline (PBS). Tissue samples were then homogenized in 500 µl of lysis buffer [7 m urea, 2 m Thiourea, 4% CHAPS, 65 mm DTT, 40 mm Tris, protease inhibitor mix and phosphoprotease inhibitor (0.2 mm Na₃VO₄, 1 mm NaF)]. After homogenization, the sample was sonicated at 100 W for 30 s and centrifuged at 25 000g for 1 h. Protein samples from 7 or 8 individual embryos corresponding to each genotype (Dhcr7^+/+, Dhcr7^Δ3-5/Δ3-5, Sc5d^+/+ and Sc5d^−/−) were pooled. Pooled samples were used for 2-DE analysis and initial validation. For the temporal expression experiment, protein samples from 4 or 5 individual control or Dhcr7^Δ3-5/Δ3-5 embryos at E12.5, E14.5 and E16.5 were pooled. Protein concentration was determined using a Bradford assay. Samples were stored at −80°C until analysis.

Two-dimensional protein gel electrophoresis and identification of differentially expressed protein spots

Immobilized pH Gradient Isoelectric focusing (IPG-IEF) was performed using an Ettan IPGphor 3 IEF System (GE Healthcare). Each genotype was analyzed in triplicate. Eighty micrograms of protein were mixed with DeStreak™ Rehydration Solution (GE Healthcare) containing 1% IPG-IEF buffer, pH 3–10 non-linear, 0.2% DTT and a trace of bromophenol blue to give a total volume of 250 µl. This solution was then loaded onto precast IPG dry strips (pH 3–10; 130 mm × 3 mm × 0.5 mm; GE Healthcare) and rehydrated at 30 V for 12 h. IEF was performed at 20°C using stepped voltage changes of 500 V for 1 h, 1000 V for 1 h and 8000 V for 8 h for a total of 50–60 kV h.

Following IEF separation, the gel strips were equilibrated twice (15 min each) in an equilibration buffer composed of 50 mm Tris–HCl pH 8.8, 6 m urea, 30% glycerol, 2% SDS and a trace of bromophenol blue. One percent DTT was added to the first equilibration buffer, and 4.5% iodoacetamide was added to the second equilibration buffer. The equilibrated gel strips were then placed onto 1 mm 12% polyacrylamide vertical slab gels, and sealed with 0.5% agarose. SDS–PAGE was performed using a Hoeffer SE600 Ruby Electrophoresis Unit (GE Healthcare) for 15 min at constant current of 10 mA per gel, and then switched to 20 mA per gel until the Bromophenol blue reached the bottom of the gels.

After two-dimensional gel electrophoretic separation, gels were stained with ammoniacal silver nitrate (48). Stained gels were scanned at an optical resolution of 42.3 × 42.3 μm using a GS-800 Calibrated Densitometer (Bio-Rad) in transmissive mode. Spot detection, matching, quantization, normalization and statistical analysis for the triplicate silver-stained analysis gels for mouse brain samples of different genotype were performed using PDQuest 8.0 software (Bio-Rad) as previously described (48). The PDQuest software models protein spots mathematically as a 3D Gaussian distribution and determines maximum absorption after raw image correction and background subtraction. Spot intensities were then obtained by the integration of the Gaussian function with unit of intensity calculated as ‘Intensity*Area as parts per million’ (INT*Area PPM). The intensity of each protein spot was normalized to the total intensity of the entire gel. Quantitative analysis was performed using the Student's t-test between control and Dhcr7^Δ3-5/Δ3-5 or Sc5d^−/− mutant mouse brain gels. The confidence level was 95% (P < 0.05).

After identification of candidate protein spots, preparative 2-DE was performed with 2 mg of protein loaded on each gel. Protein spots were identified using Coomassie Brilliant Blue R-250 Staining Solutions (Bio-Rad) per manufacturer's protocol. Protein spots were excised from paired control and mutant preparative gels for protein identification.

Mass spectrometric protein identification

Excised protein spots were ‘in-gel’ digested with trypsin, and resulting peptides were extracted. Peptides were analyzed both by MALDI TOF/TOF using an ABI 4800 Proteomics Analyzer (Applied Biosystems) and by combined liquid chromatography/tandem mass spectrometry (LC/ESI/MS/MS) using an LCQ DECA ion trap mass spectrometer (Thermo Fischer).

For MALDI TOF/TOF analysis, tryptic peptides were applied as unseparated mixtures to sample plate wells. Typically 0.5 μl of extracted peptide, 5% of total sample in 1:1 acetonitrile: 0.1% TFA was applied with an equal volume of HCCA matrix (5 mg/ml in the same solvent as the sample); two internal standards were added to the matrix and along with tryptic autolysis peptides were used for mass scale calibration. Our typical procedure was to acquired a 400 laser shot spectrum of the full peptide mixture in reflector mode and then to fragment 4–6 peptides that are both intense and well separated from others within the limits of the timed ion selector operated at a resolution of 300; fragmentation spectra were acquired as unimolecular decompositions (collision gas off) using 1000 laser shots. There is no evidence of sample depletion from this approach.

For the LC/ESI/MS/MS measurements, tryptic peptides were separated by reversed phase chromatography and electrosprayed directly into the sampling orifice of the mass spectrometer. MS/MS spectra were collected in a data-dependent manner, with up to three of the most intense ions in each full MS scan being subjected to isolation and fragmentation. MS/MS spectra were extracted as dta files using default parameters with BioWorks v2.0 (Thermo Fisher). For each LC/MS/MS, run the dta files were merged into one database file prior to analysis.

MS/MS spectra from both the MALDI and LC/ESI measurements were analyzed to identify specific proteins. The Mascot v 2.2 (Matrix Science) search program was used to match both b and y ion series to protein sequences present in the Mus musculus subset of UniProt KB/Swiss-Prot/TrEMBL database (database version 56.0; 392 667 sequence entries, 22 July 2008). Mascot search parameters were as follows: enzyme, trypsin; one missed tryptic cleavage site permitted; fixed modification, carbamidomethylation of cysteine; variable modification, methionine oxidation; mass tolerance for precursor ions was ±1.2 Da (DECA) and ±0.15 Da (4800); mass tolerance for fragment ions, ±0.6 Da (DECA) and ±0.06 Da (4800). Only peptides with individual ions Mowse scores >32 indicating significant identity or extensive homology (P < 0.05) were used for protein identification. Peptide and protein identifications were validated using Scaffold v2.2 (Proteome Software, Portland, OR, USA). In Scaffold, the peptide identifications from Mascot were verified using the X! Tandem (www.thegpm.org v 2007.01.0.2) database search program (49). Probabilistic validations of the peptide identifications were performed using Peptide Prophet (50) and corresponding protein probabilities by Protein Prophet (51). The cutoff for peptide identification was set at greater than 95.0% probability and protein identification at greater than 99% probability with two or more identified peptides.

Western blot analysis

Twenty micrograms of protein were separated on 4–12% Nupage Tris-acetate gels (Invitrogen) and transferred to a nitrocellulose membrane using iBlot (Invitrogen) according to the manufacture's protocol. Chemiluminescent western blotting was performed using the Western Breeze kit from Invitrogen per manufacture's protocol. The following primary antibodies were used for western blot analysis: mouse monoclonal anti-actin (1:5000; Sigma), anti-cofilin-1 (1:2000; Cell Signaling Technology), anti-phospho-Cofilin (Ser3) (1:200; Cell Signaling Technology), rabbit polyclonal anti-Limk-1 (1:1000; Cell Signaling Technology), rabbit polyclonal anti-Limk-2 (1:1000; Cell Signaling Technology), rabbit polyclonal anti-phospho-Limk1 (Thr508)/Limk-2 (Thr505) (1:200; Cell Signaling Technology), rabbit polyclonal anti-Pak1/2/3 (1:1000; Cell Signaling Technology), rabbit polyclonal anti-phospho-Pak1 (Thr423)/Pak2 (Thr402) (1:200; Cell Signaling Technology), rabbit polyclonal anti-Mlc2 (1:1000; Cell Signaling Technology), rabbit polyclonal anti-phospho-Mlc2 (Ser19) (1:200; Cell Signaling Technology), mouse monoclonal anti-OP18 (1:2000; Santa Cruz Biotechnology), rabbit polyclonal anti-phospho-Op18 (Ser 16) (1:400; Santa Cruz Biotechnology). Band intensity on western blots was quantified and normalized to intensity obtained for actin. For quantitative analysis, five individual samples were analyzed. Data are expressed as the mean and standard deviation.

GTPase activation assay

RhoA, Rac1, Cdc42 and Ras activation assays were performed by using small GTPase activation assay kits (Cytoskeleton, Denver, CO, USA) per manufacture's protocol. Briefly, brains were isolated from five E18.5 embryos for each genotype and were lysed using the provided buffer. Total RhoA, Rac1 and Cdc42 levels were determined by western blot analysis of 20 mcg of total protein. For the determination of activation status, 300 mcg of the protein lysate was incubated with either RBD or PBD glutathione beads for 1 h at 4°C. After washing, the bound proteins were analyzed by western blotting with monoclonal antibody against RhoA, Rac1, Cdc42 or Ras to detect GTP-RhoA, GTP-Rac1 and GTP-Cdc42, respectively.

Histological analysis

For phalloidin staining heads from E18.5 Dhcr7 control and mutant embryos were fixed overnight at 4°C in PBS (pH 7.4) with 4% paraformaldehyde. After washing in PBS, the tissues were equilibrated in 30% sucrose/PBS (pH 7.4). Sagittal sections were cut at a thickness of 20 µm on a Leica Cryostat, model CM-3050S. Cryostat sections were washed for 30 min with 50 mm glycine-PBS containing 0.5% coldwater fish gelatin to block non-specific binding, and then incubated for 2 h in a PBS solution containing 1.5 U rhodamine phalloidin (Invitrogen), 0.5% cold water fish gelatin and 50 mm glycine. For a negative control, the rhodamine phalloidin was omitted. After washing in PBS, the stained sections were mounted using MOWIOL media. Fluorescent microscopy was performed on a Zeiss LSM 510 inverted microscope using 5× (NA 0.15) objective lens. With the same laser intensity, the fluorescent images for Dhcr7^+/+ and Dhcr7^−/− mouse brains were obtained. The fluorescent intensity was analyzed with MetaMorph software (Universal Imaging). Nine sections were analyzed from each brain and three brains were analyzed for each genotype. Whole brain section fluorescent intensity was compared using a two-tailed Student t-test. To account for multiple comparisons, fluorescent staining for specific anatomical regions was analyzed using one-way analysis of variance (ANOVA) with Bonferroni correction for multiple comparisons.

Hippocampal dendrite and axonal length analysis

Hippocampal neuron cultures were prepared and immunostained as previously described (52). Briefly, the hippocampus was dissected from Dhcr7^+/+ and Dhcr7^Δ3-5/Δ3-5 E18.5 embryos, digested with 0.125% trypsin, seeded onto poly-d-lysine (Sigma)-coated coverslips at a density of 2 × 10⁵ cells/cm², and cultured in Neurobasal medium (Invitrogen) supplemented with B27 (Invitrogen). Cells were grown at 37°C in a humidified atmosphere of 5% CO₂. The B27/Neurobasal medium has no detectable cholesterol by gas chromatographic/mass spectrometric analysis (data not shown). Neurons were transfected at DIV7 with a plasmid expressing Venus with Lipofectamine 2000 (Invitrogen) per manufacture's instructions. Three days after transfection, neurons were fixed in PBS containing 4% paraformaldehyde for 15 min at room temperature. Fixed neurons were stained with a rabbit anti-GFP antibody (MLB) and an Alexa 488-conjugated secondary antibody (Invitrogen). Images were acquired using an inverted Zeiss LSM510 confocal microscope with a 10× (NA 0.25) objective. Dendrites and axons from collapsed confocal images were manually traced and measured with MetaMorph software (Universal Imaging).