A liver enhancer in the fibrinogen gene cluster

Fibrinogen is the major coagulation protein by mass and has a central role in platelet aggregation and thrombus formation. It is also an important determinant of blood viscosity and thrombophilia, and also plays a role in the proliferation of vascular endothelial and smooth muscle cells.^1-4 

The fibrinogen plasma level varies within the healthy population ranging from 1.5 to 3.5 g/L. An increase in fibrinogen concentration of 1 g/L above the normal range is associated with a near doubling of cardiovascular disease risk.⁵  Environmental factors such as smoking, body mass index, and age, or characteristic physiological states such as pregnancy and inflammation have been associated with higher or lower fibrinogen levels.⁶  The heritability of plasma fibrinogen level has been estimated between 20% and 50%.^7,8  Therefore, we speculated that common genetic variations localized in regulatory or coding regions of the fibrinogen cluster affect fibrinogen plasma levels that in turn influence formation of the fibrin network, its structure or the sensitivity of the fibrin clot to fibrinolysis. Thus, identifying new sequences contributing to the regulation of fibrinogen gene expression may add further insight into the genetic components involved in the heritability of plasma fibrinogen levels and could be of clinical relevance for the prediction of cardiovascular events.

Fibrinogen is a hexamer composed of 2 copies of 3 polypeptide chains Aα, Bβ and γ. These are encoded by 3 genes, FGB, FGA, and FGG (ordered from centromere to telomere) clustered in a 50-kb region on the long arm of human chromosome 4 (4q32). Fibrinogen biosynthesis occurs in the liver where the 3 genes undergo coordinate transcription. Functional transcription factor binding sites have been described within the 3 promoter regions.⁹  Two transcription factors, CCAAT-box/enhancer-binding protein (C/EBP) and the hepatocyte nuclear factor-1 (HNF-1), play an important role in the basal expression of FGA and FGB. The increased expression of the 3 genes during the acute phase is mediated by interleukin-6 responsive elements, identified in the promoter regions of all 3 genes.⁹ 

Further investigations into the transcriptional regulation of the fibrinogen gene cluster are required to explain the coordinate liver-specific transcription pattern of the fibrinogen genes.^10,11  The majority of studies on fibrinogen transcriptional regulation have focused on the proximal promoter regions. Surprisingly, the role of regulatory sequences situated at larger distances from the fibrinogen gene transcription start sites has not been addressed. Indeed, it is now generally assumed that highly tissue-specific or complex expression patterns of vertebrate genes are achieved by coordinate actions of promoter regions and regulatory elements.^12,13  These regulatory elements can be separated by several megabases from their target gene(s).¹⁴  Therefore the identification of these sequences remains challenging because it implicates short regulatory elements of a few hundred base pairs embedded in large genomic regions.

To identify regulatory elements driving tissue-specific transcription, comparative sequence analysis has been used successfully by starting with the underlying hypothesis that functional regions of the genome are less tolerant to nucleotide substitutions and thus portray higher evolutionary conservation than expected under neutral selection.^12,15-19  To this end, experimental strategies combining in vitro screens of numerous sequences followed by in vivo investigation of the spatio-temporal pattern of enhancer activity have been proposed.^20,21  Functional conserved noncoding sequences (CNCs) have been extensively associated with temporal, spatial, and quantitative regulation of gene expression,²²  playing roles in development²³  and disease.²⁴ 

Genome comparisons using the human, mouse, chicken, and frog sequences reveal remarkable conservation of the fibrinogen gene cluster structure as well as a conserved adjacent syntenic region.²⁵  It has been suggested that the need to keep long-range cis-regulatory elements in cis with their target genes during evolution has contributed to structured conserved blocks of synteny observed in vertebrate genomes.^25,26  This hypothesis implies that transcriptional regulatory mechanisms have been conserved for the fibrinogen genes and involve regulatory elements residing in the syntenic regions including the fibrinogen gene cluster.

To identify novel regulatory elements involved in the transcriptional regulation of the fibrinogen cluster, we assessed the in vitro regulatory potential of 14 CNCs from the syntenic genomic landscape of the fibrinogen cluster. We found 4 CNCs with enhancer activity and 1 potential silencer. Liver activity for 1 of the 4 enhancer CNCs was confirmed in vivo using EGFP reporter gene transgenic zebrafish and validated in LacZ reporter gene mouse embryos. These results demonstrate the presence of a novel liver enhancer within the fibrinogen cluster located between the FGB and FGA genes.

Human hepatoma HuH7 and human embryonic kidney (HEK-293T) cells were cultured in Dulbecco modified Eagle medium (Gibco) supplemented with 10% heat-inactivated fetal bovine serum, 100 U/mL penicillin and 100 μg/mL streptomycin (Invitrogen). Human hepatoma HepG2 cells were cultured in Eagle minimum essential medium (Sigma-Aldrich) also supplemented with 10% heat-inactivated fetal bovine serum 100 U/mL penicillin and 100 μg/mL streptomycin (Invitrogen).

Polymerase chain reaction (PCR) primers for the amplification of selected human CNCs (CNC1 to CNC14) are listed in supplemental Table 1 (available on the Blood Web site; see the Supplemental Materials link at the top of the online article). The amplified fragments, which include SalI restriction sites on both extremities, were inserted upstream of the minimal TATA-like promoter (pTAL) in the XhoI site of the pTAL firefly luciferase (pTAL-Luc) reporter plasmid (Clontech Laboratories).

HEK-293T, HuH7, and HepG2 cells were seeded at 10⁴, 1.5 × 10⁴, and 3 × 10⁴ cells per well, respectively, in opaque 96-well plates (PerkinElmer) 24 hours before transfection. Each well was transfected with 100 ng of pTAL-Luc firefly reporter construct and 8 ng of transfection control plasmid encoding the renilla luciferase gene (pRL-SV40; Promega), using FuGENE-HD (Roche) as transfection reagent, according to the manufacturer's instructions. These transient cotransfections were performed in technical triplicates and in 3 independent transfection experiments. Twenty-four hours after transfection, the activities of firefly and renilla luciferase were measured using the Dual-Luciferase Assay Kit (Promega). Luciferase activity ratio (firefly/renilla) for each transfection condition was normalized to the ratio obtained with control plasmid (pTAL-Luc without a CNC sequence) present on each plate. Enhancer activity of each CNC sequence was assessed by comparison with the basal transcriptional activity of the luciferase coding plasmids carrying no CNCs (normalized to 1.0). Normalization to the mean activity of the control condition gave the fold-change in luciferase activity values presented.

The human sequences of CNC4, CNC10, CNC12, and CNC13 were amplified from the pTAL-Luc reporter plasmid by PCR (Clontech Laboratories; supplemental Table 1), adding a HindIII site on their 5′ end and a BamHI site on their 3′ end, and cloned directionally into the homemade pTRAN plasmid. In this manner, the CNC sequences were inserted in front of a human β-globin/mouse Hspa1a minimal promoter and the enhanced green fluorescence protein (EGFP) gene, followed by a simian virus (SV)40 polyadenylation signal. The reporter cassettes (CNCs, β-globin promoter, EGFP, and polyadenylation signal) were amplified from the pTRAN vector, adding a SalI site upstream of CNCs, and subcloned via pCRII-TOPO (Invitrogen) between the XhoI and HpaI sites of pT2KXIGΔin, a Tol2 transposon system gene transfer plasmid.²⁷  A similar construct lacking a CNC was used as a negative control. As a positive control, we amplified by PCR 1.63 kb of genomic sequence from upstream of the zebrafish fgg gene transcription start site and cloned this region upstream of EGFP and the SV40 polyadenylation signal. Further details and oligonucleotides sequences used in this construction are available upon request. Tol2 transposase mRNA was in vitro transcribed with SP6 RNA polymerase from NotI-linearized pCS-TP,²⁸  using the mMessage mMachine Kit (Ambion).

AB strain zebrafish were raised and bred in standard conditions and embryos were obtained from natural crosses. A license for experimentation with zebrafish was obtained from the local veterinary authority. One- or two-cell embryos were microinjected with 1 nL of injection solutions composed of 35 ng/μL transposase mRNA, 25 ng/μL reporter gene plasmid, and 0.075% phenol red essentially as described by Fisher et al.²⁹  G₀ embryos were observed daily using a Leica MZ16FA fluorescence stereomicroscope and photographed at room temperature using a Leica DFC340FX digital camera (black and white) and Leica software LAF 2.1. Some G₀ embryos were raised to reproductive age and crossed with control fish. Expression of EGFP was then assessed in G₁ embryos up to 8 days postfertilization (dpf).

Two dpf G₁ transgenic or control embryos were fixed, depigmented, permeabilized, hybridized with DIG-labeled RNA probes, and developed for imaging essentially as described by Thisse and Thisse.³⁰  Digoxigenin (DIG)–labeled sense and anti-sense EGFP RNA probes were synthesized by in vitro transcription with SP6 and T7 RNA polymerases (both from Promega) from a linearized pCRII TOPO (Invitrogen) plasmid containing the complete EGFP open reading frame with promoter sequences flanking either side of the insert. Minor protocol changes, compared with Thisse and Thisse,³⁰  included probe hybridization at 60°C, and prehybridization, hybridization, and posthybridization washing steps made in 2-mL tubes rather than immersed nylon baskets. Before imaging, embryos were progressively washed in 25%, 50%, and 75% methanol in phosphate-buffered saline, 100% methanol, and then clarified in 100% glycerol. Images were obtained at room temperature using a Leica DFC420 digital camera and software LAS 2.8.1.

CNC12 was PCR amplified from human genomic DNA (Clontech), sequence-validated, and cloned into a Gateway compatible Hsp68-LacZ reporter vector³¹  (primers used for amplification: FgCNC12F: CCTCAATTTTCCATTAGCAA; FgCNC12R: AGGAAAGCAGTATCGTGAAG).

Transgenic mice were generated as previously described³²  in accordance with protocols approved by the Lawrence Berkeley National Laboratory.

A total of 66 embryos were harvested at embryonic day (E)14.5. Abdominal cavities were opened to facilitate reagent penetration of the liver. Embryos were fixed in 4% paraformaldehyde solution and subsequently incubated in staining solution (0.8 mg/mL X-gal [5-bromo-4-chloro-3-indolyl-beta-D-galacto-pyranoside], 4mM potassium ferrocyanide, 4mM potassium ferricyanide, and 20mM Tris [tris(hydroxymethyl)aminomethane] pH 7.5) for 24 hours at room temperature. Images were obtained at room temperature in phosphate-buffered saline using a Leica MZ16 light microscope at magnification 1.6× with a Leica DFC420 camera. Adobe Photoshop Elements was used as acquisition software.

The human fibrinogen gene cluster is located within a syntenic region of ≈215 kb (assembly NCBI36/hg18: Chr4: 155 675 587-155 891 061) conserved from human to frog (Xenopus tropicalis).²⁵  This conserved blocks of synteny (CBS) is constituted of the pleiotropic regulator 1 (PLRG1) gene at the centromeric end, the 50-kb fibrinogen cluster, a gene desert region of ≈131 kb and the lecithin retinol acyltransferase (LRAT) gene at the telomeric end. Using a multispecies alignment tool (rVISTA),³³  we searched for sequences defined by a minimum length of 100 bp and at least 70% homology with the mouse sequence. We also considered the presence of PhastCons conserved elements³⁴  as well as the evolutionary and sequence pattern extraction through reduced representations regulatory potential.³⁵  With these criteria, we selected 14 sequences (CNC1 to CNC14) localized within the syntenic landscape of the human fibrinogen locus (Figure 1A and supplemental Table 1) as candidates for regulatory elements of the fibrinogen locus. These sequences are all located in the CBS, except for CNC1 that is located 41-kb centromeric to FGB. We selected CNCs within the fibrinogen cluster (intronic, in promoter regions or intergenic) as well as CNCs outside the cluster. CNC2 and CNC4 were chosen due to their particular localizations (ie, partly overlapping the FGB and FGG proximal promoters, respectively). The lesser conserved CNC5 and CNC8 were selected due to their high evolutionary and sequence pattern extraction through reduced representations regulatory potential pattern. The 14 selected sequences do not show an homology greater than 78% with the mouse sequence, and are not deeply conserved throughout vertebrates; CNC13 and CNC14 are the most conserved with homologies greater than 70% in a human-chicken alignment.

The regulatory potential of the 14 selected CNC sequences was tested in vitro using a luciferase-based reporter assay. Their putative role in controlling hepatic-specific fibrinogen transcription was monitored by assaying them in 2 fibrinogen-expressing hepatoma cell lines (Huh7 and HepG2) versus a nonfibrinogen-expressing cell line (HEK-293T). The orientation-dependent regulatory function of each CNC was studied by cloning and assaying both orientations (see “In vitro enhancer assays using luciferase gene reporter plasmids”). For this purpose, we cloned short sequences (348 bp in average, SD 140 bp), without flanking genomic sequences. These constructs were transiently transfected in the 3 cell lines tested. The regulatory potential of each CNC was determined by normalization to the basal transcriptional activity of the backbone luciferase plasmid (Figure 1). Of the 2 sequences overlapping promoter regions, only CNC4 showed activity, interestingly in all 3 cell lines. CNC13 shows enhancer potential (relative luciferase activity above 2) in all 3 cell lines in both orientations, as might be expected for enhancer elements that are not orientation dependent.^13,36,37  CNC12 shows enhancer potential exclusively in hepatoma cell lines. CNC10 showed enhancer activity in HuH7 cells, but not in HepG2 cells where we observed inconsistent luciferase activities when comparing CNC10 orientations. Finally, taking into account a cutoff value of relative luciferase activity at or below 0.5 for both orientations for repressor activity, we found that CNC11 shows repressor activity in HepG2 cells, a lesser effect in HuH7 cells, and no activity in embryonic kidney cells.

In summary, the in vitro screen of 14 conserved sequences revealed 4 putative regulatory elements, of which 2 are potential hepatocyte enhancers (CNC10 and CNC12), and 1 is a potential hepatocyte repressor (CNC11).

To evaluate the regulatory potential in vivo and assess the spatiotemporal enhancer activity of CNC4, CNC10, CNC12, and CNC13 (Figure 2), we performed in vivo EGFP reporter assays in zebrafish embryos. If these 4 sequences are indeed implicated in the tissue-specific expression of the fibrinogen genes, they should drive liver expression of the reporter gene. The human sequences of CNC4, CNC10, CNC12, and CNC13 were cloned upstream of a minimal promoter and EGFP in a Tol2 transposase gene transfer plasmid. A negative control without a CNC sequence was used to evaluate the background expression of the EGFP encoding vectors. We also used as positive control a construct with EGFP downstream of a 1.63-kb zebrafish fgg promoter. These constructs were individually comicroinjected with capped Tol2 transposase mRNA in zebrafish embryos, resulting in mosaic transgenic animals (G₀). Embryos were regularly observed and at 5 dpf, 25 to 55 embryos per reporter construct were analyzed by fluorescence microscopy (Table 1). We observed reproducible EGFP expression patterns for each reporter construct (Figure 3A). The negative control construct (carrying no CNC) gave weak scattered EGFP-positive somite muscle cells. This weak muscle signal was sparsely observed for all 5 constructs. The majority of CNC4- and CNC10-injected embryos showed EGFP expression in the yolk sac (24/25 and 41/42, respectively), and a single individual per construct showed weak expression in the liver. Surprisingly, more than half of the CNC13 embryos did not show any GFP expression, and the other half showed weak expression in muscle cells or in the yolk. In contrast, CNC12 showed EGFP expression in the liver for 50% (18/36) of injected embryos. This rate is comparable with fish injected with a positive control plasmid, carrying EGFP under the control of the zebrafish fgg promoter (70% EGFP-positive livers).

To confirm that our observations made on G₀ embryos were not due to an insertion site bias or episomal expression of the CNC-EGFP transgene, we analyzed EGFP expression in transgenic progeny descending from 2 different CNC12 founders. For this purpose, a group of CNC12-injected embryos were raised to reproductive age and crossed with wild-type fish. EGFP expression in their transgenic progeny was analyzed during the first week of development. EGFP expression was not observed during the first day of transgenic G₁ embryo development. During the second day EGFP was seen in the presumptive gut and liver primordium. The liver EGFP signal became more evident as liver development progressed through 5 to 8 dpf (Figure 3B). The EGFP pattern observed at 8 dpf is very similar to that seen in transgenic fish carrying the EGFP gene under the control of the zebrafish fgg promoter (R.J.F., M.N.-A., unpublished data). Finally, to distinguish the EGFP expression driven by the CNC12 from the strong auto-fluorescence of the zebrafish embryo yolk sac, we performed in situ hybridizations with DIG-labeled EGFP RNA probes to clearly define the EGFP expression pattern in 2 dpf transgenic animals. Wild-type embryos and sense RNA probes were used as negative controls. EGFP mRNA was clearly and specifically detected in the developing gut, pronephros, and liver bud (Figure 4), a pattern of expression reminiscent of that of the zebrafish CCAAT/enhancer binding protein alpha (C/EBPα).³⁸ 

In summary, our in vivo reporter gene assay in transgenic zebrafish embryos demonstrates that CNC12 drives the robust liver expression of a transgene.

To confirm the in vivo liver activity of CNC12 in a mammalian model system, we performed a transgenic reporter gene assay in mouse embryos. Human CNC12 was cloned upstream of an Hsp68 minimal promoter driving the activity of a LacZ reporter gene. The linearized vector was used to generate transgenic embryos by pronuclear microinjection. A total of 66 embryos were collected at E14.5. Of the 66 embryos, 12 were transgenic as assessed by LacZ staining. Of these 12 embryos resulting from independent transgenic integration events, 10 showed staining in the liver. In most cases, liver staining was confined to individual scattered cells (8/10), whereas in 2 cases we observed strong staining in the median and left lobes that did not extend into other regions of the liver (Figure 5 and supplemental Figure 2). Of note, the mRNA expression patterns of the Fga, Fgb, and Fgg genes at E14.5 do not exhibit such a scattered or lobe-specific appearance (Visel et al³⁹  and www.eurexpress.org). There are at least 2 plausible explanations for the partial overlap in the pattern of expression of the fibrinogen genes and CNC12. First, additional enhancers may map within the locus and the coordinated action of all of these enhancers may be necessary to recapitulate the full fibrinogen gene expression pattern. Second, the activity of CNC12 in our in vivo assay may be weakened due to its integration site (positional effect) or by the absence of flanking sequences in the construct used for this experiment.

In this study, we provide both in vitro and in vivo evidence for the liver enhancer activity of a noncoding sequence, CNC12, located within the human fibrinogen cluster. Our data show that CNC12, a 340-bp element located between FGB (6.1 kb upstream) and FGA (7.7 kb downstream), is sufficient to drive liver expression of reporter genes in human hepatoma cells, in zebrafish embryos and in mouse embryos.

Identification of this novel enhancer sheds light on the regulation of the fibrinogen cluster because it is the first liver-active regulatory sequence to be identified in the 50-kb cluster outside of the well-known 5′-proximal promoters of FGA, FGB, and FGG. While additional experiments are necessary to formally prove the role of CNC12 in the regulation of fibrinogen, considering the specific liver activity of CNC12 and its position within the conserved block of synteny of the fibrinogen genes, it seems justified to suggest that CNC12 is involved in the gene expression control of the fibrinogen genes, both in zebrafish and mouse embryos, but also in humans. Due to its distinctive intergenic localization, CNC12 may be involved in the global coregulation of the 3 fibrinogen genes that is still not fully understood. Interestingly, a hyperfibrinogenemic mouse model was recently obtained by multiple insertions of a 80- to 100-kb transgene containing the complete mouse fibrinogen locus.⁴⁰  The expression of the exogenous fibrinogen genes imitated perfectly the endogenous liver-specific transcription pattern, suggesting that all the essential regulatory sequences mediating the tissue-specificity were present within the fibrinogen cluster.

Enhancers activate transcription, often in a temporally and tissue-restricted manner, mainly by acting on promoters.⁴¹  Some authors have suggested that regulatory sequences are kept in the vicinity of their target gene(s) due to evolutionary pressure restraining recombination events between regulatory elements and target genes, ultimately leading to the formation of conserved blocks of synteny.^25,26 

Of great interest to this study, CNC12 has binding sites for 2 important liver transcription factors.⁴²  Indeed, high-resolution chromatin immunoprecipitation (ChIP) assays, combining ChIP with high-throughput sequencing (ChIP-seq),⁴³  have defined all loci interacting with C/EBPα and HNF4α in liver samples from organisms representing 5 vertebrate orders (human, mouse, dog, short-tailed opossum, and chicken).⁴⁴  In absolute correlation with our data, C/EBPα and HNF4α were found to bind to sequences within human CNC12 (supplemental Figure 1). In addition, in situ hybridization performed for cebpa (Zebrafish Information Network identification: ZDB-IMAGE-040107-415)³⁸  and hnf4a (Zebrafish Information Network identification: ZDB-IMAGE-080403-438)⁴⁵  in zebrafish embryos depict very similar patterns to those we obtained for our CNC12-EGFP transgenic animals (Figure 4). Altogether, these ChIP-seq experiments, in situ hybridizations, and our data allow us to speculate that CNC12 liver activity is mediated through interactions with C/EBPα and HNF4α transcription factors.

Of the 14 sequences that we tested in vitro, CNC12 is not among the most conserved. Although it is commonly observed that deep phylogenetic conservation increases the likelihood of regulatory function,^31,46  moderately conserved regulatory sequences probably represent a group of more recently developed enhancers.⁴⁷  It has been proposed that comparisons of orthologous sequences from multiple closely related species may be a more sensitive and accurate approach for the identification of vertebrate regulatory sequences rather than focusing on deeply conserved regions.⁴⁸  Indeed, regulatory noncoding sequences conserved only within mammals have been shown experimentally to be functional.²¹  The conservation pattern of CNC12 shows 4 PhastCons elements³⁴  that are modestly conserved among 17 vertebrates (Figure 2), suggesting a recently developed mammalian-specific enhancer. This feature raises questions regarding the liver activity we observed in fish. Two hypotheses could explain the activity of a modestly conserved human element in the fish assay system: either the sequence important for transcription factor binding and enhancer activity is very short and although conserved between vertebrates is simply not identifiable; or the critical transcription factor binding sites show sequence plasticity and the fish transcription factors are able to recognize human binding sites.

The identification of an enhancer likely to be involved in the transcriptional regulation of fibrinogen gene expression reveals a further potential source of the large variability in circulating fibrinogen levels seen among healthy people. Indeed, in addition to posttranscriptional regulation by microRNAs,⁴⁹  environmental factors,⁶  modifier genes,⁵⁰  genetic variation within the 5′-proximal promoter or coding sequences, polymorphisms in more distant enhancers should be taken into consideration. As C/EBPα and/or HNF4α bind CNC12 in hepatocytes, and may therefore contribute to fibrinogen regulation, genetic variants affecting these transcription factors should also be taken into consideration. Subtle variations in upstream regulator expression may impact on fibrinogen production, affecting the interactions between promoters and distant enhancers. Therefore, variability in fibrinogen levels could be explained by perturbation at several levels of transcriptional regulation.

We thank Luciana Romano and Corinne Di Sanza for technical assistance and Professor Koichi Kawakami (National Institution of Genetics, Shizuoka, Japan) for providing Tol2 transposon system plasmids.

This work was supported by grants from the Dr Henri Dubois-Ferrière-Dinu Lipatti foundation and the Swiss National Science Foundation (grant number 31-A0119845). A.V. and C.A. were supported by grant HG003988 funded by the National Human Genome Research Institute and Department of Energy Contract DE-AC02-05CH11231, University of California, Energy Office Lawrence Berkeley National Laboratory. C.A. is also supported by an European Molecular Biology Organization long-term fellowship.

Contribution: A.F. designed and performed in silico analyses and in vitro experiments; R.J.F. designed and performed experiments with zebrafish; C.A. helped with the in silico analyses and designed and performed experiments with mice; R.D. provided expertise for the zebrafish experiments; A.V. designed the mouse experiments; M.N.-A. designed and directed the project; A.F., R.J.F., and M.N.-A. wrote the manuscript; and C.A., R.D., and A.V. edited the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Marguerite Neerman-Arbez, Department of Genetic Medicine and Development, 1 Rue Michel Servet, 1211 Geneva, Switzerland; e-mail: marguerite.neerman-arbez@unige.ch.