A mutation in the c-Fos gene associated with congenital generalized lipodystrophy

Fibroblasts initiated from skin biopsies of patient and healthy caucasian volunteer controls were expanded for 4 cycles and stored in liquid nitrogen. Cells were recultured (DMEM, 10% FCS; Life Technologies, Darmstadt, Germany) and expanded for a maximum of 3 passages before harvesting. For exogenous stimulation fibroblasts were grown to 70% confluence and serum starved (1% FCS) for 40 h (quiescent fibroblasts indicated as basal in figure legends) prior to induction with 10^-7 M insulin, 10^-8 M IGF-1, 1.5×10^-8 EGF or 3.3×10^-9 M PDGF. Preadipocytes (3T3-L1) and muscle cells (A7r5) were cultured in DMEM and liver cells (HepG2) in RPMI 1640 (Life Technologies) supplemented with 10% FCS. Institutional research ethics approval (PV3641) in line with the Helsinki Declaration was obtained for this study.

Nuclear extracts from fibroblasts of the patient and controls, or HepG2 cells were prepared as described [11].

MAPK activity assays and western blotting with polyclonal anti-Akt or anti-phospho Akt antibody (New England Biolabs, Frankfurt, Germany) or polyclonal anti-MAPK antibody (BD Transduction; Heidelberg, Germany) were performed as described [11].

Total RNA was extracted with RNeasy Mini Spin Kit (Qiagen, Hilden, Germany). RT- PCR analyses were performed in triplicates with c-fos gene-specific probes and 18S RNA internal standard (Applied Biosystems, Darmstadt, Germany) as described [12]. Expression results were determined as relative RNA amounts of target.

C-fos promoter (nt −734 to +43; numbering based on TS (=1) according to K00650) was PCR amplified from genomic DNA of controls and patient using following primers: -734 to −712: 5’-GCGAGGAACAGTGCTAGTATTGCT-3’/ +43 to +12: 5’-CGGCTCAGTCTTGGCTTCTCAGTTGCTCGCT-3’. Wild type c-fos promoter fragment and the corresponding fragment of the patient were inserted in sense orientation into pGL3basic vector (Promega, Mannheim, Germany) to bring the reporter gene luciferase under control of wild type c-fos promoter (pc-fos-wt) or the mutated promoter (pc-fos -439T>A). The expression vector pFA-Elk-1 containing the regulative domains of transcription factor Elk-1 (aa 307 to 427) fused to the heterologous DNA binding domains of Gal4 (aa 1 to 147) under control of a MLV-promoter was used (Life Technologies). For monitoring transactivation the reporter plasmid pGal4-Luc5 containing the luciferase gene under control of 5× Gal4 binding elements was cotransfected in these experiments (Life Technologies). As reference of transfection efficiency the β-galactosidase expression vector pEF-ßGal was used. Independent plasmid preparations and cells were used for replicate experiments.

Cell suspensions (2×10⁵ cells/well) of 3T3-L1, A7r5, HepG2 cells or fibroblasts of patient and control were mixed with vectors as indicated in figure legends and pulsed for 18 msec (3T3-L1, A7r5, HepG2) or 9 msec (fibroblasts). For exogenous induction, cells were serum-starved on day one following transfection for 40 h and incubated with 10^-7M insulin or 3.3×10^-9M PDGF for 3 h before harvesting. Transfection, monitoring of transfection efficiency and luciferase assays were performed as described [11].

Genomic DNA was extracted from fibroblasts of patient or control using the Qiagen blood kit™. For reevaluation of the identified mutation only limited amounts of genomic DNA of the patient’s parents but no biopsies were available. The coding sequence of AGPAT2 (NM_006412.3), caveolin-1 (NM_001753.4; NM_001172895.1), seipin (NM_001122955.3), CAV/PTFR (NM_012232) and c-fos promoter (K00650) were analyzed by direct sequencing (ABI PRISM 3100, Applied Biosystems).

Genomic DNA of 319 unrelated caucasian subjects was isolated from PMBCs and c-fos promoter (−603 to +12) was amplified (−603 to −579: 5’ primer: 5’-AGGCTTAAGTCCTCGGGGTCCTGT-3’; +43 to +12: 3’ primer: 5’-CGGCTCAGTCTTGGCTTCTCAGTTGCTCGCT-3’). PCR products were reamplified with a mutated primer (−441: C>A) introducing a Tsp509I restriction site solely in wild type c-fos promoter (5’ primer: -470 to −440, c-fos mut −441: C>A: 5’-CATTGAACCAGGTGCGAATGTTCTCTCTA A-3‘ and 3‘ primer −337 to −307: 5’-AGATGTCCTAATATGGACATCCTGTGTAAG-3’). PCR products were Tsp509I digested and size fractionated. Genomic DNA of patient was always treated in parallel for control. 10% of samples were randomly chosen to confirm results by direct sequencing.

Footprinting analyses were performed using the sure track footprinting kit™ (GE Healthcare, Munich, Germany). Promoter fragments (nt −462 to −325) were isolated from pc-fos-wt or pc-fos-patient by EcoRI/EcoNI restriction and cohesive ends were labeled with 20 μCi [α³²P] dATP and Klenow fragment. 100.000 cpm of radiolabeled fragments were incubated with increasing concentrations of nuclear extracts (4 μg, 8 μg, 20 μg, 40 μg) from human liver cells (HepG2) and subsequently digested for 1 min with different concentrations of DNaseI (0.11U, 0.33U, 1.0U). Reactions were terminated and deproteinized by LiCl precipitation. DNA fragments were precipitated and separated on an 8% PAGE containing 7 M urea. Purine nucleotide sequence ladders from the EcoRI/EcoNI promoter fragments were loaded in parallel to confirm sequence of protected areas.

EMSA were performed according to [13] and reactions were analyzed on 5% non-denaturating PAGE. For analyses of the in vitro ternary complex formation with nuclear extracts of fibroblasts from control or the patient, 2 pmol sre-element promoter fragments (−331 to −280 5’CCCCTTACACGGATGTCCATATTAGGACATCTGCGTCAGCAGGTTTCCACG; 3’GGAATGTGCCTACAGGTAATAATCCTGTAGACGCAGTCGTCCAAAGGTGCCC) were labeled with 20 μCi [α32P]dGTP using 5U Klenow fragment, prior to use. For completion experiments 100- or 10-fold molar excess of unlabeled sre fragments or unspecific SP-1 promoter fragment (5’-GTTAGGGGCGGGATGGGCGGAGTT -3’) were used.

Analyses of the novel protein/DNA interaction at the c-fos promoter were performed with c-fos-wt (−451 to −431: 5’-TGTTCTCTCTCATTCTGCGCCG-3’) or c-fos-patient (−451 to −431, -439T>A: 5’-TGTTCTCTCTCAA TCTGCGCCG-3’) endlabeled with 5U PNK and 20 μCi [γ³²P]dATP, prior to use. For competition experiments 10× or 100× unlabeled c-fos-wt or pc-fos-patient fragment were used. Experiments were performed with nuclear extracts (5 μg) of human liver cells (HepG2).

For preparative EMSA a Cy3-labeled c-fos-patient fragment was used in the procedure. Gels were scanned using a Typhoon scanner (GE, Freiburg, Germany) and fluorescence marked bands were cut from gels. The gel slices were placed on a 10% SDS-PAGE for separation of complex proteins. Four independent EMSA replicates were performed and cutted protein/DNA complex were separated on four SDS-PAGE each. Of all protein bands three different punch samples were excised and subjected to mass spectrometry analyses according to [14] for identification. Acquired mass spectra (peptide mass fingerprint) were automatically calibrated and annotated using Compass 1.3 software and xml formatted peak lists were transferred to Proteinscape3.0 (Bruker Daltonik, Bremen Germany). MS peptide mass fingerprint were used to search a human sub-set of Swiss-Prot (Sprot_2011; 20249 20401 protein entries) non-redundant database using Mascot search engine (Version 2.2, Matrix Science Ltd, London, UK) Mass tolerance was set to 50 ppm for peptide spectra and a combined mascot score over 70 was taken significant (p < 0.01). For verifying the results each protein spot was picked and identified from at least three physically different gels.

Fibroblasts of patient and 6 individual controls were cultured to passage 6 each. Four replicate analyses of patient cells (initiated from two primary stored cell pools) and 6 individual controls (initiated from one primary stored cell pool each) were used. Equal amounts of total RNA were processed according to the GeneChip One-Cycle eukaryotic Target Labeling Assay (GeneChip Expression analysis technical Manual, http://​www.​affymetrix.​com/​support/​Technical/​manual/​expression_​manual.​affx) and used for expression analyses with Hu95A_v2 Arrays (Affymetrix UK Ltd). Syntheses of cRNA and fragmentation were quality controlled and monitored with a RNA 6000 nano kit (Agilent, Taufkirchen, Germany). Detection of probe sets was performed using a GeneChip scanner (GCOS 1.4 package, Affymetrix). The original CEL files were directly implemented into Genespring 12.0 (Agilent) for analyses. The Genespring 12.0 Volcano Plot analyses workflow with default settings (paired t-test, multiple testing correction: Benjamini-Hochberg) of the gene expression data sets were used to identify genes with statistic significant expression (p < 0.05) and a minimum 1.5-fold difference among conditions. Full data sets are available under accession number GSE39825 (http://​www.​ncbi.​nlm.​nih.​gov/​geo/​).

For functional annotation and conserved promoter element site search web based tools were used (http://​david.​abcc.​ncifcrf.​gov/​) [15, 16]. For functional annotation protein IDs or Affymetrix IDs, fold change, and t-test p-value of detection significance were imported to Ingenuity Pathway Analysis (IPA) System (http://​www.​ingenuity.​com). IPA was carried out with p < 0.002 as cutoff point. Pathways indicating altered transcriptional regulation were deduced from fold change differences observed.

Data are given as means ± S.D. Students t-test was used to determine statistical significance.

As the patient was the only known case of lipodystrophy in the family nothing remained but analyzing a possible defect in insulin signaling, we characterized known signaling pathways. For this purpose we utilized primary fibroblasts initiated from skin biopsies.

The in vitro insulin receptor binding capacity, auto- and substrate-phosphorylation was in normal range (data not shown). The insulin mediated signaling cascade including Akt abundance or phosphorylation (Figure 1A) and ERK1/2-MAPK activation or activity (Figure 1B) was comparable to controls in patient cells, indicating no disturbance in these pathways. A definite endpoint of MAPK cascades is the transcriptional activation of the immediate-early genes c-fos [17], whereas the c-fos gene activity is directly related to formation of a ternary transcription activation complex at the sre-element and the phosphorylation of the ternary complex factor Elk-1 by ERK1/2-MAPK [18]. Investigations showed that the formation of the ternary complex (Figure 1C) and Elk-1-activation dependent transcription was not altered in cells of the patient (Figure 1D). Nevertheless in patient cells induction of c-fos mRNA expression was nearly completely lost following insulin and IGF-1 induction and clearly reduced following PDGF and EGF induction in contrast to control cells (Figure 1E). These findings focused the defect to c-fos gene directly.

Sequence analyses of the patients’ c-fos gene identified next to a common heterozygous SNP in 5’UTR (rs7101; c.–60) a novel homozygous point mutation in the promoter at position c.–439 (Figure 2A) upstream of well characterized regulatory promoter elements. This mutation was not identified in 319 control subjects or in the confirmed parents of the patient, indicating a de novo mutation (Figure 2B).

One functional possibility of a point mutation in a promoter is the direct interference with protein/DNA interactions. Utilizing the c-fos promoter in nuclease protection assays revealed novel protein binding sites (P1-P3) only with DNA-fragments bearing the identified point mutation in P2 (Figure 3A). EMSA confirmed that this specific protein/DNA interaction only occurs if the point mutation was present (Figure 3B). A mass spectrometry approach revealed that the novel formed protein/DNA complex consisted of at least 13 proteins (Figure 3C). Database analyses showed that identified proteins were not classical transcription factors, but nucleases, helicases or structural proteins (Table 1).

To test the regulatory relevance of identified mutation in the c-fos promoter we performed promoter reporter gene analyses. Transfecting a wild type c-fos promoter into the patient’s cells revealed that the basal and inducible promoter activity is completely reconstituted (Figure 4A). Vice versa transfecting a c-fos promoter bearing the identified mutation into control cells revealed that the observed diminished expression and inducibility of c-fos was dependent from cellular environment (Figure 4A). These data exclude further proximal signaling defects and support a defect intrinsic to the c-fos promoter as the mutation identified. Further analyses demonstrated that this observation was not cell specific as in preadipocytes, muscle and liver cells c-fos promoter activity and activation was evenly inhibited by the point mutation (Figure 4B-D).

To test if the mutation and novel protein/DNA complex observed has an impact in cellular context, we performed a gene expression survey in the patient (Additional file 1: Table S1). Functional analyses of these results revealed that multiple metabolic or transcriptional processes and a large fraction of genes involved in cell cycle control and differentiation were affected (Additional file 2: Table S2). Data base analyses suggest that most of differentially regulated transcripts bear the c-fos responsive AP-1 promoter consensus sequence (Figure 5A). Further in silico analyses indicate that next to c-fos and further AP-1 complex factors, differentially expressed transcripts in the patient are also regulated by transcription factors necessary in adipocyte differentiation and maturation i.e. ATF4, PPARα, SREBP-1or-2 and CEBP family members (Figure 5B).

Congenital lipodystrophy is characterized by the complete loss of adipose tissue. In our patient with CGL we identified a homozygous de novo point mutation in the c-fos promoter. Although this is an unexpected finding, increasing genomic information indicated that approximately 50–100 germ line de novo mutations can occur in each individual [19]. Nevertheless, the mutation identified in this study has a clear functional relevance of c-fos promoter activity. Due to the identified mutation a novel mutation-specific protein/DNA interaction is formed. As consequence basal and inducible c-fos expression is ubiquitously lowered. The proteins identified in the complex represent no classical DNA interacting proteins but DNA modifying helicases, nuclear ribonucleoproteins (hnRP), structural proteins, and heat shock proteins. It was shown that the single strand binding ATP dependent helicases are involved in telomere maintenance, double strand breaking repair and DNA unwinding as needed to transcriptional activation, especially it has long been noted that UV initiated DNA damage interferes with inducibility of c-fos expression and UV induced DNA damage interferes with adipocyte differentiation [20, 21]. HnRPs localized to the nucleus are involved in transcriptional repression processes including c-fos mRNA transcription [22]. Furthermore nuclear actin or nucleolin are involved in chromatin-remodeling processes and actin can be associated to hnRNPs A/B type family as also shown here [23, 24]. Such modifications in chromatin structure are required for gene expression and adipocyte differentiation as local chromatin remodeling has been described in the regulation of the PPARγ transcription during adipocyte differentiation [25, 26].

All together the novel protein complex seems to act on c-fos promoter as ubiquitous transcriptional repressor. Mechanistically one can speculate that the complex alters DNA packaging, masks basal or activating promoter elements or interacts with transcription factors. The novel protein DNA complex identified might constitutively mimic a transcriptional repression process that usually is under tight regulation in cells [27].

The consequence of lowering c-fos expression is a differential gene expression pattern in our patient compared to controls. Further analyses shows that the reduced c-fos expression affects genes involved in multiple metabolic pathways, transcriptional control and a large fraction of genes involved in cell cycle control and differentiation processes most of them AP-1 dependent. Of special note are genes involved in adipocyte differentiation as Wnt5A, IL6, FGF-2, ODC-1, C/EBP, NR2F2, INHBB, SFRP1 or ITGA6. On the other hand gene functional annotation revealed that genes differentially expressed are also regulated by various adipose differentiation transcription factors as C/EBPs, PPARs or the SREBP family.

On cellular level the point mutation identified in our patient and the binding of the novel protein complex results in reduced basal as well as inducible c-fos expression but not complete loss of c-fos expression. This reduced c-fos expression influences expression on many other genes. This might be due to the fact that c-Fos doesn’t act as single transcription factor but is part of AP-1 complex. This protein complex consists of a combination of two proteins from various homologues proteins of Jun, ATF, MAF and Fos families, i.e. c-jun, JUNB, JUND, ATF2, ATF3/LRF1, B-ATF, JDP1, JDP2, c-Maf, MafB, MafA, MafG/F/K, Nrl, c-fos, Fra-1, Fra-2, FOSL or FosB. As consequence depending on dimer composition the transactivation activity of AP-1 complex varies from activation to repression of the target gene transactivation activity [9]. Furthermore, the occupancy of AP-1 sites by AP-1 transcription complex also influences transcription of overlapping or adjacent promoter elements. The direct competition of AP-1 to binding site CRE or ARE has been reported [27, 28]. Knockout mice deficient for c-Fos revealed phenotypes with severe osteopetrosis and altered hematopoiesis. They show reduced fetal and placental weight, reduced weight gain and reduced fat mass, but to our knowledge there are no studies assessing further metabolic parameters [29‐31]. However there is a reciprocal interaction between bone and energy metabolism [32]. Osteoblasts and adipocytes originate from a common mesenchymal progenitor and specific differentiation via BMPs and WNT pathways determine the cell fate to bone or adipose specific precursor cells [33]. This speculation is supported by mice overexpressing Fra-1 which develop lipodystrophy due to reduced adipocyte differentiation via C/EBPa inhibition and transcriptional repression [34]. Interestingly, patients with congenital lipodystrophy show increased bone age and density, enlarged epiphyses, sclerotic skeletons and alterations in dentition [35]. Furthermore the promoter activation of the immediately early gene c-fos is involved in various signaling cascades. One of those is IGF-1 signaling, that shares the signaling cascade and activation mechanisms of c-fos promoter as insulin [11]. As the role of IGF-1 and the GH/IGF-1 axis in various syndromes with growth restriction is well established [36] one can speculate that the growth alterations or skin and hair variations observed in the patient are most likely a consequence of interference with IGF-1 signaling.