Targeting Isoprenylcysteine Methylation Ameliorates Disease in a Mouse Model of Progeria

See allHide authors and affiliations

Science  14 Jun 2013:
Vol. 340, Issue 6138, pp. 1330-1333
DOI: 10.1126/science.1238880

Methylation and Methuselah?

Hutchinson-Gilford progeria syndrome (HGPS) and other prelamin A–associated progeroid disorders arise when farnesylated and methylated forms of prelamin A accumulate at the nuclear envelope. Ibrahim et al. (p. 1330, published online 16 May; see the Perspective by Johnson) show that reducing the activity of the isoprenylcysteine carboxyl methyltransferase (ICMT) mislocalizes prelamin A, triggers prelamin A–dependent AKT-mTOR signaling, and eliminates disease phenotypes in 30-week-old progeria model mice. Reduced ICMT expression increased the proliferation and delayed the premature senescence of progeria model mouse fibroblasts and cells from children with HGPS.


Several progeroid disorders, including Hutchinson-Gilford progeria syndrome (HGPS) and restrictive dermopathy (ZMPSTE24 deficiency), arise when a farnesylated and methylated form of prelamin A accumulates at the nuclear envelope. Here, we found that a hypomorphic allele of isoprenylcysteine carboxyl methyltransferase (ICMT) increased body weight, normalized grip strength, and prevented bone fractures and death in Zmpste24-deficient mice. The reduced ICMT activity caused prelamin A mislocalization within the nucleus and triggered prelamin A–dependent activation of AKT-mammalian target of rapamycin (mTOR) signaling, which abolished the premature senescence of Zmpste24-deficient fibroblasts. ICMT inhibition increased AKT-mTOR signaling and proliferation and delayed senescence in human HGPS fibroblasts but did not reduce the levels of misshapen nuclei in mouse and human cells. Thus, targeting ICMT might be useful for treating prelamin A–associated progeroid disorders.

Children with Hutchinson-Gilford progeria syndrome (HGPS) exhibit premature aging phenotypes and often die during their teenage years. HGPS is caused by mutations in the gene encoding prelamin A and lamin C (LMNA) that result in an internally truncated form of prelamin A (called progerin) that accumulates at the nuclear rim and induces nuclear shape abnormalities (14). Progerin retains its carboxyl-terminal CAAX motif, which triggers farnesylation of the cysteine (i.e., the “C” in the CAAX motif) by protein farnesyltransferase (FTase) (fig. S1). Interfering with farnesylation with an FTase inhibitor (FTI) reduces the frequency of misshapen nuclei in Zmpste24-deficient fibroblasts and HGPS fibroblasts and ameliorates disease phenotypes in Zmpste24-deficient mice and other models of HGPS, although the benefits are modest (2, 510). A recent clinical trial of an FTI in children with HGPS showed a modest beneficial effect on disease phenotypes (11, 12).

After farnesylation, the last three amino acids (the “AAX” of the CAAX motif) of prelamin A and progerin are clipped off, and the farnesylcysteine is methylated by isoprenylcysteine carboxyl methyltransferase (ICMT) (fig. S1). Methylation of some CAAX proteins, including the RAS oncoprotein, is important for proper targeting to membranes within cells (13, 14), but the relevance of methylation to prelamin A–associated progeroid disorders is unknown. To address this issue, we took advantage of a hypomorphic Icmt allele (Icmthm), fortuitously created by the insertion of loxP sites flanking exon 1 of Icmt (14, 15). Icmthm/hm mice were leaner than Icmthm/+ littermates (fig. S2, A to D) but were healthy and lived for >2 years. The reduced ICMT expression in Icmthm/hm cells inhibited prelamin A processing to lamin A (fig. S2E).

We bred Zmpste24−/− mice harboring the Icmthm allele. ICMT expression and activity levels were 70 to 90% lower in Zmpste24−/−Icmthm/hm mice than in Zmpste24−/−Icmt+/+ littermates (fig. S3, A to D). Incubating Zmpste24−/−Icmthm/hm fibroblasts with a Cre-adenovirus (adCre) eliminated ICMT expression (fig. S3, B and D). The reduced ICMT activity in Zmpste24−/−Icmthm/hm livers resulted in a moderate accumulation of ICMT substrates (fig. S3E). To determine whether prelamin A existed in unmethylated form in Zmpste24−/−Icmthm/hm cells, we performed mass spectrometry of prelamin A in isolated fibroblast nuclei (fig. S3F). The relative levels of unmethylated prelamin A were higher in nuclei of Zmpste24−/−Icmthm/hm and Zmpste24−/−IcmtΔ/Δ cells than in Zmpste24−/−Icmt+/+ cells (fig. S3G).

As expected, Zmpste24−/−Icmt+/+ mice developed alopecia, a hobbling gait, growth retardation, and reduced grip strength; they began to die at ~10 weeks of age (1618) (Fig. 1, A to D, and fig. S4). By 30 weeks, all had died or been euthanized (Fig. 1D). At that time, Zmpste24−/−Icmt+/+ mice had multiple rib fractures adjacent to the costovertebral joints (Fig. 1, E and F). L2 vertebrae from Zmpste24−/−Icmt+/+ mice exhibited osteopenia and reduced osteoid, suggesting little or no ongoing bone formation (fig. S5A). In contrast, all Zmpste24−/−Icmthm/hm mice were alive at 30 weeks, and their gait, grip strength, fur, and body-weight curves were much better than in Zmpste24−/−Icmt+/+ mice (Fig. 1, A to D, and fig. S4).

Fig. 1 Targeting Icmt ameliorates disease phenotypes and prevents death in 30-week-old Zmpste24−/− mice.

(A) Photograph of 24-week-old littermate mice. (B) Body-weight curves of male Zmpste24−/−Icmt+/+ (n = 11) and Zmpste24−/−Icmthm/hm (n = 5) mice. (C) Kaplan-Meier plot showing the percentage of Zmpste24−/−Icmt+/+ (n = 21) and Zmpste24−/−Icmthm/hm (n = 9) mice with normal grip strength. (D) Kaplan-Meier plot showing survival of Zmpste24−/−Icmt+/+ (n = 21) and Zmpste24−/−Icmthm/hm (n = 9) mice. (E) Ventral view of spinal columns from an 18-week-old Zmpste24−/−Icmt+/+ mouse and a 30-week-old Zmpste24−/−Icmthm/hm mouse. Arrowheads indicate rib fractures at costovertebral joints. (F) Number of rib fractures in Zmpste24−/−Icmt+/+ (n = 21) and Zmpste24−/−Icmthm/hm (n = 9) mice. (G and H) Cellular parameters of L2 vertebrae (n = 6 per genotype). **P < 0.01; ***P < 0.001. Data are presented as mean ± SEM.

All Zmpste24−/−Icmthm/hm mice were sacrificed at 30 weeks of age, and none had rib fractures (Fig. 1, E and F). Osteoblasts in L2 vertebrae of Zmpste24−/−Icmthm/hm mice were six times as abundant as in Zmpste24−/−Icmt+/+ mice; osteoclast numbers were unchanged (Fig. 1, G and H, and fig. S5B). The numbers, thickness, and bone volume of trabeculae were higher in Zmpste24−/−Icmthm/hm mice than in Zmpste24−/−Icmt+/+ mice (fig. S5B). Moreover, at 13 to 15 weeks of age, Zmpste24−/−Icmthm/hm mice had increased mineral density and bone content (fig. S6A), as judged by dual x-ray absorptiometry (DXA). The DXA scans also revealed increased adipose tissue in Zmpste24−/−Icmthm/hm mice; however, total body weight at that time point did not differ (fig. S6B), consistent with the body-weight curves (Fig. 1B and fig. S4).

Misshapen nuclei are a hallmark of progeria cells. We initially suspected that the improved disease phenotypes of Zmpste24−/−Icmthm/hm mice would be accompanied by a lower frequency of misshapen nuclei in cultured fibroblasts, but this was not the case (Fig. 2, A and B). In control experiments, knockout of Fntb (encoding the FTase β-subunit) in Zmpste24−/− fibroblasts reduced the frequency of misshapen nuclei to wild-type levels (fig. S7A).

Fig. 2 Icmt deficiency mislocalizes prelamin A but does not reduce the frequency of misshapen nuclei in Zmpste24−/− fibroblasts.

(A) Confocal images of nuclei in primary mouse embryonic fibroblasts stained with a LAP2β antibody. Scale bar, 10 μm. (B) Frequency of misshapen nuclei in primary fibroblasts (n = 3 per genotype). Cre-adenovirus was used to produce Zmpste24−/−IcmtΔ/Δ cells from the parental Zmpste24−/−Icmthm/hm cells. βgal-adenovirus was used as control. (C) Super-resolution structured illumination microscopy (SR-SIM) immunofluorescence images of nuclei in liver sections stained with prelamin A– and LAP2β-specific antibodies and counterstained with 4′,6-diamidino-2-phenylindole (DAPI). Scale bar, 10 μm. ***P < 0.001.

We next determined whether reduced Icmt expression affects localization of prelamin A in Zmpste24−/− cells. Consistent with earlier studies (17, 18), prelamin A was mainly found at the nuclear rim in Zmpste24−/−Icmt+/+ hepatocytes, colocalizing with LAP2β (Fig. 2C). In contrast, prelamin A in Zmpste24−/−Icmthm/hm hepatocytes was abundant in the nucleoplasm (Fig. 2C and fig. S7B). Similar results were observed in skeletal muscle (fig. S7C). Prelamin A, which accumulated at moderate levels in Zmpste24+/+Icmthm/hm fibroblasts (fig. S2E), was entirely nucleoplasmic in liver sections (fig. S7D). The ratio of prelamin A and β-tubulin was similar in lysates of Zmpste24−/−Icmthm/hm and Zmpste24−/−Icmt+/+ tissues (fig. S7E). Thus, the hypomorphic Icmt allele partially mislocalizes prelamin A away from the nuclear rim but has no effect on the absolute levels of prelamin A.

Fibroblasts from Zmpste24−/− mice proliferate slowly and undergo premature senescence (19, 20). We defined the impact of Icmt deficiency on those phenotypes. As expected, Zmpste24−/−Icmt+/+ fibroblasts proliferated slowly and senesced prematurely (Fig. 3A). In contrast, Zmpste24−/−Icmthm/hm and Zmpste24−/−IcmtΔ/Δ cells proliferated at rates similar to those of wild-type cells (Fig. 3A). Inactivating Fntb abolished cell proliferation (fig. S8).

Fig. 3 Icmt deficiency prevents premature senescence of Zmpste24−/− fibroblasts by increasing AKT pathway signaling.

(A) Growth curves of primary fibroblasts (n = 4 per genotype) before (top) and after (bottom) incubation with Cre-adenovirus. (B) Immunoblots of fibroblast extracts show increased phosphorylation of AKT, S6, and 4E-BP1. Increased phosphorylation and inactivation of 4E-BP1 is evident by an increased ratio of γ and β/α isoforms. eIF4E and β-tubulin were the loading controls. (C) Immunoblots of fibroblast extracts with antibodies against prelamin A and cell cycle regulatory proteins. (D and E) Growth curves of primary fibroblasts incubated with inhibitors of (D) AKT (GSK690693, 10 μM) and (E) PTEN (VO-OHpic, 150 nM) (n = 3/genotype). (F) Immunoprecipitation (IP) and immunoblot (IB) analyses showing a methylation-dependent association between AKT and prelamin A. The lysates were also used directly for immunoblots for AKT (input). *P < 0.05; ***P < 0.001.

We next explored the impact of Icmt deficiency on intracellular signaling pathways. The AKT-mTOR signaling pathway affects cell growth and survival (21). We observed higher levels of phosphorylated AKT and greater mTOR activation in Zmpste24−/−Icmthm/hm cell lysates than in Zmpste24−/−Icmt+/+ lysates, evident by increased phosphorylation of its downstream targets ribosomal protein S6 and 4E-BP1 (Fig. 3B). Similar results were observed in comparisons of Zmpste24+/+Icmthm/hm and Zmpste24+/+Icmt+/+ cells (fig. S9A). Consistent with increased rates of proliferation, Zmpste24−/−Icmthm/hm cells had lower levels of the cyclin-dependent kinase inhibitors p27KIP1 and p21CIP1 and the tumor suppressors p16INK4A and phosphorylated retinoblastoma protein (Rb) (Fig. 3C). Levels of phospho-AKT and -S6 were lower in tissue lysates of Zmpste24−/−Icmt+/+ mice than in Zmpste24+/+Icmt+/+ mice; the levels were normalized in Zmpste24−/−Icmthm/hm tissues (fig. S9, B to D). Cytosolic phospho-AKT can translocate into the nucleus and trigger inactivation and degradation of p21CIP1 (21). Levels of phospho-AKT were high in nuclei of Zmpste24−/−Icmthm/hm hepatocytes and correlated with reduced levels of nuclear p21CIP1 (fig. S9E).

To determine whether increased AKT-mTOR signaling contributes to the increased proliferation of Zmpste24−/−Icmthm/hm cells, we performed cell proliferation assays with inhibitors of AKT (GSK690693) and mTOR (rapamycin). The AKT inhibitor blocked the proliferation of Zmpste24−/−Icmthm/hm cells, whereas the mTOR inhibitor had no effect (Fig. 3D and fig. S10, A to C). We also activated AKT in Zmpste24−/−Icmt+/+ cells with VO-OHpic, an inhibitor of phosphatase and tensin homolog (PTEN, a tumor suppressor upstream of AKT). VO-OHpic delayed the senescence of Zmpste24−/−Icmt+/+ cells (Fig. 3E and fig. S10D). Thus, Icmt deficiency overcomes senescence of Zmpste24−/− cells by activating AKT.

The increased AKT-mTOR signaling in Icmt-deficient cells was likely not caused by reduced methylation of RAS and RAS homologue enriched in the brain (RHEB) because their levels were unaffected by Icmt deficiency (fig. S11A). To test the possibility that the increased AKT-mTOR signaling was caused by unmethylated prelamin A, we analyzed fibroblasts from Icmthm/hm mice on a background of a mutant Lmna allele (LmnaLCO) that produces lamin C but no prelamin A (22). The levels of phospho-AKT and -S6 were lower in Icmthm/hmLmnaLCO/LCO lysates than in Icmthm/hmLmna+/+ lysates (fig. S11B). Thus, prelamin A is required for the increased AKT-mTOR signaling accompanying Icmt deficiency. Prelamin A and AKT were physically associated in Zmpste24−/−Icmt+/+ lysates; the association was reduced in Zmpste24−/−Icmthm/hm lysates (Fig. 3F). Perhaps an interaction between prelamin A and AKT in Zmpste24-deficient cells inhibits AKT-mTOR signaling (fig. S9, B to E).

To assess the effect of inhibiting ICMT in human fibroblasts, we suppressed ICMT expression in wild-type and HGPS fibroblasts with lentiviral short hairpin (sh) RNAs. The shRNAs reduced ICMT expression by 80% and had no discernible effect on the proliferation of fibroblasts from healthy subjects (Fig. 4A and fig. S12A). However, knockdown of ICMT in three HGPS cell lines delayed senescence and increased the mean proliferation rate (Fig. 4B and fig. S12B). The proliferation of HGPS cells also increased when ICMT activity was inhibited with N-acetyl-S-farnesyl-l-cysteine (AFC), a competitive ICMT inhibitor (fig. S12C). Rapamycin and an FTI dose-dependently reduced the proliferation of HGPS cells (fig. S12, D to G). The absolute levels of progerin were similar in shICMT-treated and control-HGPS cells, and phospho-AKT and -S6 levels were higher in shICMT-treated cells (Fig. 4C). Nuclear shape abnormalities in HGPS cells were unaffected by knockdown of ICMT (Fig. 4, D and E). Thus, cells from HGPS patients and Zmpste24-deficient mice respond similarly to ICMT inhibition.

Fig. 4 Icmt knockdown in human HGPS fibroblasts prevents premature senescence and increases AKT-mTOR pathway signaling.

(A) TaqMan analyses showing ICMT mRNA levels in skin fibroblasts from HGPS patients incubated with shRNA lentiviruses. Data are means of three independent cell lines incubated with two lentiviral clones expressing ICMT shRNAs or with a control clone containing a scrambled (SCR) sequence. (B) Growth curves of HGPS cells incubated with shICMT or shSCR lentiviruses (n = 3 cell lines per treatment). (C) Immunoblots of fibroblast extracts; β-tubulin was the loading control. (D) Confocal images of fibroblasts stained with an antibody against LAP2β. Scale bar, 10 μm. (E) Frequency of nuclear shape abnormalities in control and HGPS fibroblasts incubated with shICMT and shSCR lentiviruses (n = 3 cell lines per treatment).

Our findings raise the possibility that targeting ICMT could be an effective strategy for treating HGPS. Reduced Icmt expression was accompanied by improved disease phenotypes in Zmpste24−/− mice. Thus, the favorable effects in the setting of a hypomorphic Icmt allele suggest that the efficacy of ICMT inhibitors would not require complete inhibition of the enzyme.

Supplementary Materials

Materials and Methods

Figs. S1 to S12

References (2331)

References and Notes

  1. Acknowledgments: We thank the Proteomics and Imaging Core Facilities at the Sahlgrenska Academy; M. Dalin, P. Lindahl, and J. Nilsson for comments; C. Karlsson and P. Iranmanesh for technical assistance; and S. Ordway for editing the manuscript. This study was supported by a Starting Investigator Grant from the European Research Council; and by grants from the Swedish Cancer Society, the Swedish Research Council, the Swedish Children’s Cancer Fund, Västra Götalandsregionen, and the Ingabritt and Arne Lundberg’s Research Foundation (to M.O.B). Data are available in the supplementary materials.
View Abstract

Navigate This Article