based candidate gene approach
Uitte de Willige, S.
Citation
Uitte de Willige, S. (2007, May 23). Novel genetic risk factors for venous
thrombosis; a haplotype-based candidate gene approach. Hemostasis and
Thrombosis Research Center, Department of Hematology, Faculty of Medicine,
Leiden University. Retrieved from https://hdl.handle.net/1887/11970
Version: Corrected Publisher’s Version
License: Licence agreement concerning inclusion of doctoral thesis
in the Institutional Repository of the University of Leiden
Downloaded from: https://hdl.handle.net/1887/11970
Note: To cite this publication please use the final published version (if
applicable).
Chapter 3.2
Polymorphism 10034C>T is located in a region
regulating polyadenylation of FGG transcripts and
influences the fibrinogen γ'/γA mRNA ratio
Shirley Uitte de Willige, Inge M. Rietveld, Marieke C.H. de Visser, Hans L. Vos and Rogier M. Bertina
Submitted for publication
Summary
Background Fibrinogen gamma haplotype 2 (FGG-H2) is associated with reduced fibrinogen γ' levels and fibrinogen γ'/total fibrinogen ratios and with an increased deep venous thrombosis (DVT) risk. Two FGG-H2 tagging SNPs, 9615C>T and 10034C>T, are located in the region of alternative FGG pre-mRNA processing.
10034C>T is located in a GT-rich Downstream Sequence Element (DSE) which comprises a putative Cleavage stimulation Factor (CstF) binding site.
Objectives To investigate the functionality of SNPs 9615C>T and 10034C>T and the importance of the DSE containing 10034C>T.
Methods Different mini-gene constructs containing FGG exon 9, intron 9, exon 10 and the 3' region were transiently transfected into HepG2 cells and quantitative real-time PCR was used to measure relative polyadenylation (pA) signal usage (pA1/pA2-ratio).
Results Compared to the reference construct CC (9615C-10034C; FGG-H1;
pA1/pA2-ratio set at 100%), the pA1/pA2-ratio of construct TT (9615T-10034T;
FGG-H2) was 1.4-fold decreased (71.5%, p=0.015). The pA1/pA2-ratio of construct CT (9615C-10034T) was almost 1.2-fold decreased (85.3%, p=0.001), whereas the pA1/pA2-ratio of construct TC (9615T-10034C) did not differ significantly from the reference construct (101.6%, p=0.890). Functionality of the putative CstF binding site was confirmed using constructs in which this site was deleted or its sequence altered by point-mutations.
Conclusions SNP 10034C>T is located in a GT-rich DSE involved in regulating the usage of the pA2-signal of FGG, which may represent a CstF binding site. We propose that the 10034C>T change is the functional variation in FGG-H2 which is responsible for the reduction in the fibrinogen γ'/total fibrinogen ratio and the increased DVT risk.
Introduction
Fibrinogen plays an important role in the haemostatic system, being the precursor of fibrin, the end product of blood coagulation.1,2 It is a 340 kD glycoprotein circulating in the blood at a concentration of approximately 9 μM (3 g/L). Fibrinogen molecules are composed of six polypeptide chains held together by disulfide bonds, (Aα, Bβ, γ)2. The different polypeptide chains are encoded by three separate genes;
fibrinogen alpha (FGA), fibrinogen beta (FGB) and fibrinogen gamma (FGG).3 Alternative transcripts of FGA and FGG do exist. The predominant Aα-chain contains 610 amino acids and is translated from the first five exons of FGA. The alternative Aα-chain (1-2% of Aα-chains) contains 846 amino acids and is translated from all six exons.4 The most abundant form of the γ-chain, γA, consists of 411 amino acids (Figure 1a). The variant γ'-chain (or γB in the nomenclature of Francis et al.5) contains 427 amino acids and 7-15% of all γ-chains are γ'.6-8 γA is formed by translation of all 10 FGG exons, while the γ'-chain is formed after alternative pre-
mRNA processing, resulting in translation of exons 1-9 and the first 60 nucleotides of intron 9.9,10 γA and γ' differ in their carboxyterminal sequence. The γA-chain ends with the unique AGDV sequence encoded by exon 10, which is involved in platelet- binding. In the γ'-chain this sequence has been replaced by 20 amino acids encoded by intron 9 (Figure 1a), which form binding sites for Factor XIII 11 and thrombin exosite II.12-14
Figure 1a Alternative FGG pre-mRNA processing. The γA-chain is translated from mRNA in which all 9 introns of the pre-mRNA have been removed and polyadenylation has occurred downstream from exon 10 at polyadenylation signal 2 (pA2). The γ'-chain arises from alternative FGG pre-mRNA processing. Intron 9 is not removed and polyadenylation occurs at an alternative site located in this intron (pA1), leading to the translationof a polypeptide with a unique 20-amino acid extension encoded by intron 9 substituted forthe carboxyterminal four amino acids of the γA-chain encoded by exon 10.9,10 The γ'-chain comprises approximately 7- 15% of the fibrinogen γ-chain found in plasma.8 Nearly all of the γ' protein occurs in vivo as a heterodimer with the γA variant in which one D region contains a γ' carboxyterminus and the other a γA carboxyl terminus (γA/γ' fibrinogen).7 Both SNPs 9615 C>T and 10034 C>T are specific for FGG-H2.
Recently, we found that haplotype 2 of FGG (FGG-H2) was associated with an increased deep venous thrombosis (DVT) risk and with reduced fibrinogen γ' levels and reduced fibrinogen γ'/total fibrinogen ratios.15 After inspection of the single nucleotide polymorphisms (SNPs) present in this haplotype, we observed that FGG- H2 tagging SNP 10034C>T [rs2066865] (numbering according to SeattleSNPs,16 GenBank Accession number AF350254) is located in a GT-rich region (GGTA[C/T]CTTTATTGACCAT at nucleotides 10030-10047) just downstream from the second polyadenylation (pA2) signal (AATAAA, pA2) at nucleotides 9997-10002 (Figure 1a). Actually, this region shows a 78% match with the Cleavage stimulation Factor (CstF) binding site consensus 2a sequence.17 CstF is a multi-subunit complex required for efficient cleavage and polyadenylation of pre-mRNAs.18 It binds via its CstF-64 subunit to a G/U-rich downstream sequence element (DSE) and stabilizes the binding of the cleavage and polyadenylation specificity factor (CPSF) to the pA-
signal.18,19 In FGG, the pA2-signal is used for the formation of γA mRNA in which intron 9 has been removed, whereas the pA1-signal, which is located in intron 9 at nucleotides 9558-9563, is used for the formation of γ' mRNA. In FGG-H2, which contains a T at nucleotide 10034 (underlined in the CstF binding site consensus 2a sequence in Figure 1b), the DSE is strengthened (more GT-rich) compared to the other common FGG haplotypes (H1, H3, and H4), which all have a C at position 10034. The affinity of the CstF-64/DSE interaction is an important determinant of the relative strength of competing poly(A)-sites. Therefore we hypothesized that the 10034C>T change will increase the affinity of the CstF-64/DSE interaction resulting in more frequent use of pA2 (more γA transcripts) at the expense of polyadenylation at pA1 (less γ' transcripts). FGG-H2 is therefore expected to produce relatively more γA transcripts and subsequently relatively less γ' transcripts. This would explain the reduced fibrinogen γ' levels and fibrinogen γ'/total fibrinogen ratios previously observed in homozygous carriers of FGG-H2.15 A second FGG-H2 specific polymorphism, 9615C>T [rs2066864], is located at a position downstream from pA1 in intron 9, two nucleotides downstream of a DSE homologous to CstF binding site consensus 2b,17 and might therefore also influence the relative use of the two pA- signals of the FGG transcript. The remaining three FGG-H2 specific SNPs are located in the promoter region (129A>T and 902A>G) and in intron 8 (7874G>A).
To investigate the functionality of SNPs 9615C>T and 10034C>T, and the importance of the DSE downstream of pA2 (containing 10034C>T) for the regulation of pA1/pA2 usage, we made different FGG mini-gene constructs containing FGG exon 9, intron 9, exon 10 and the 3' region. After transient transfection into liver- derived HepG2 cells, we measured the relative pA-signal usage (pA1/pA2-ratio) of FGG pre-mRNAs for each of the different constructs by quantitative real-time PCR.
Methods
FGG mini-gene constructs
Mini-gene constructs used were based on expression vector pcDNA3 (Invitrogen), containing a strong CMV promoter. A 1090 bp fragment containing FGG exon 9, intron 9, exon 10 and the 3' region was amplified by PCR with high fidelity polymerase (Taq/Tgo mixture, Roche) on genomic DNA samples homozygous for FGG-H1 (9615C and 10034C) or FGG-H2 (9615T and 10034T). The forward primer (5'-GTCGATCGGTCTAGACCACCATGGGTGGCACTTACTCAAAAGCATC-3') directed against the start of FGG exon 9 contained a Kozak sequence with a translation start site (italics) and an introduced restriction site for XbaI (underlined). The startcodon, which was in frame with the open reading frame of exon 9, was introduced to prevent potential problems with nonsense mediated decay of the spliced γA-specific mRNA. The reverse primer (5'-CAACTAGAATGCAAAGAGTTAGGCATAACATTTAGCA- 3'), directed against the sequence downstream of exon 10, contained an introduced restriction site for BsmI (underlined). PCR products and vector were double digested
with XbaI and BsmI and PCR products were cloned into the XbaI and BsmI sites of the vector. By double digestion with XbaI at nucleotide 983 and BsmI at nucleotide 3199, 2216 bp containing the Bovine Growth Hormone and SV40 polyadenylation sites were removed from the vector to prevent interference with our experiments.
The sequences of these constructs and all subsequent constructs were verified.
Sequencing reactions were performed using the ABI PRISM® BigDye Terminator Cycle Sequencing kit (Applied Biosystems) and run on an ABI PRISM® 3100 Genetic Analyzer (Applied Biosystems). Primer sequences for sequencing and mutagenesis (see below) are available on request. All mini-gene construct variants made are shown in Table 1.
FGG haplotype constructs
To study the effect of SNPs 9615C>T and 10034C>T on the relative use of pA1 and pA2, we made mini-gene constructs carrying the different alleles of these SNPs.
Construct CC (9615C-10034C) and construct TT (9615T-10034T) carried the H1 and H2 haplotypes of FGG respectively. Construct CT (9615C-10034T) and construct TC (9615T-10034C) were derived by exchange of fragments between constructs CC and TT, using a HindIII site upstream of the fragment in the vector and an internal HindIII site at nucleotides 9908-9913 of the insert.
Table 1 Mini-gene constructs Construct Remark
Construct CC 9615C and 10034C (FGG-H1), wild type construct Construct TT 9615T and 10034T (FGG-H2)
Construct CT 9615C and 10034T Construct TC 9615T and 10034C
DSE deletion DSE deleted (nucleotides 10013-10056) DSE mutant 1 DSE weakened by one nucleotide, 10031G>A
DSE mutant 2 DSE weakened by two nucleotides, 10031G>A and 10042G>A DSE mutant 3 DSE strengthened by one nucleotide, 10033A>G
DSE mutant 4 DSE strengthened by two nucleotides, 10033A>G and 10046A>G
DSE deletion construct
To study the functionality and importance of the GT-rich DSE downstream of pA2 (containing 10034C>T), we made a mini-gene construct in which the complete DSE (nucleotides 10030-10047) was deleted. To achieve this, site-directed mutagenesis was used to introduce two unique restriction sites near the DSE, using the QuikchangeTM site-directed mutagenesis kit (Stratagene) according to the manufacturer's protocol. A restriction site for BspEI was introduced at nucleotides 10012-10017 by mutation of nucleotides 10015T and 10016A into G. A restriction site for AgeI was introduced at nucleotides 10056-10061 by mutation of nucleotides 10059A and nucleotides 10060C into G. BspEI and AgeI are restriction enzymes that produce compatible restriction ends. After mutagenesis, the mutated construct was
restricted with BspEI and AgeI, deleting a fragment of 44 bp from nucleotide 10013 to nucleotide 10056 from the insert, but leaving pA2 at nucleotides 9997-10002 present. Subsequently, after purification from gel, the construct was self-ligated.
DSE mutant constructs
We used site-directed mutagenesis to make four DSE mutants (Table 1 and Figure 1b). In DSE mutant 1 (10031G>A) and mutant 2 (10034G>A, 10042G>A), the CstF-64/DSE affinity was weakened by decreasing its GT content and reducing the homology to the CstF consensus 2a sequence. In DSE mutant 3 (10033A>G) and mutant 4 (10033A>G, 10046A>G) the affinity of DSE/CstF-64 was improved by increasing its GT content and improving the homology to the CstF 2a consensus.
Figure 1b Schematic representation of the inserts of the mini-gene constructs. The insert contains FGG exon 9, intron 9, exon 10 and the 3' region. The positions of the SNPs at 9615 and 10034 are indicated. Enlarged, the deleted part (gray striped block) of the construct sequence containing SNP 10034C>T (bold and underlined) is depicted and aligned with the CstF binding site consensus 2a17 and the DSE mutants. The mutated nucleotides are bold and underlined. The vertical lines indicate the nucleotides between the construct sequence and the CstF binding site consensus sequence that align (14 out of 18).
Transfection conditions
Constructs were transfected into HepG2 cells, which express endogenous fibrinogen.
The human hepatocyte hepatoma cell line HepG2 (ECACC, #85011430) was cultured according to the instructions of the ECACC. Cells were cultured in 12-well plates and transfections were performed after 24 hours at 60-80% confluency using the Tfx-20 reagent (Promega), according to the manufacturer's protocol. One μg of each construct was transfected using 3 μl Tfx-20 reagent in a total volume of 400 μl growth medium (MEM supplemented with 10% (v/v) foetal calf serum, 60 U/ml penicillin/streptomycin and 0.1 mM non-essential amino acids). Since each mini- gene construct produced two transcripts from the same template (normal splicing of intron 9 by use of pA2, and no splicing by use of pA1) and since we expressed results as the ratio of the two transcript levels, the two transcripts themselves acted as internal controls to correct for differences in transfection efficiency. For each series of mini-gene constructs, three independent transfection experiments were
performed using two separate preparations of each construct.
Total RNA isolation
After harvesting the cells by trypsinisation, total RNA was isolated using the RNeasy mini kit (Qiagen), according to the manufacturer's protocol. Each RNA sample was incubated with 10 units DNase I (Roche) for 15 min at 37 °C, followed by 15 min of inactivation at 65 °C. The quality of each total RNA sample was checked by agarose gel electrophoresis.
cDNA synthesis
cDNA synthesis was performed using a first-strand cDNA synthesis kit for reverse transcriptase (RT) (SuperScriptTM II Reverse Transcriptase, Invitrogen) and 1 μg RNA isolated from the transfected HepG2 cells according to the manufacturer's protocol, except that a modified oligo d(T) primer (5'-AGCTGGTC
AGTCGTCAGCTGA(T)16-3') was used. With this primer, only polyadenylated RNA is used as template for cDNA synthesis.
Real-Time PCR analysis
For each mini-gene construct, the mean relative use of pA1 and pA2 (pA1/pA2- ratio) was measured by quantitative real-time PCR on a HT7900 instrument (Applied Biosystems) using two fluorescently labelled probes and 2 μl of 10-fold diluted cDNA as input. To prevent the formation of heteroduplexes, the relative use of pA1 and pA2 was analysed in two separate PCR reactions. In both reactions, the forward primer (5'-TGCAGATATCCATCACACTGG-3') was located on the vector. In this way, only mRNA transcripts from the constructs were amplified and amplification of endogenous fibrinogen mRNAs was prevented. In the pA1-reaction, the reverse primer (5'-TCATCCTCAGGGTAAAGTGAGTC-3') was located in the part of intron 9 that encodes the 20 additional amino acids of the γ' chain carboxyterminus. The pA1-specific probe (5'-TET-AGGTCAGACCAGAGCACCCT-BHQ-3') was located on the boundary of exon 9 and the 5' end of intron 9. In the pA2-reaction, the reverse primer (5'-GAAGTGAAGCTTTGCAAGTCC-3') was located in the 3' UTR of exon 10 that is specific for pA2-transcripts. The pA2-specific probe (5'-FAM- GACGTTTAAAAGACCGTTTCAAA-BHQ-3') was located on the boundary of the coding domain and the 3' UTR. The sizes of the products were 280 bp (pA1-reaction) and 340 bp (pA2-reaction).
Real-Time PCR efficiencies were calculated from the slopes of a series of cDNA dilutions. The corresponding real-time PCR efficiency rate (E) of one cycle in the exponential phase was calculated according to the equation: E=10[-1/slope].20 Both the pA1- and pA2-transcripts showed a real-time PCR efficiency rate of 2.14 in the investigated range from undiluted to 105 diluted cDNA input (n=6) with high linearity (Pearson correlation coefficient r>0.98). Quantitative values for individual
cDNA levels were obtained from the cycle number at which the fluorescence crossed the threshold (Ct value). In both reactions fixed thresholds were chosen at a point at which amplification was in the exponential phase (ΔRn was 0.10 for the pA1-reaction and 0.25 for the pA2-reaction). To determine the relative levels of the pA1- and the pA2-specific cDNAs, the function 2- Ct was used, where ΔCt is the difference in Ct
values between pA1 and pA2 of a construct (Ct pA1-Ct pA2). To confirm accuracy and reproducibility of the real-time PCR, each cDNA sample was analysed in triplo in three different runs. Ct values were approved when the intra-assay CV was <1%.
The inter-assay CV was <10%.
Statistical analysis
For all constructs, the differences in the mean relative use of the two pA-signals (pA1 and pA2) were analysed by calculating the mean ratio of pA-signal usage (pA1/pA2-ratio). For the FGG haplotype constructs and the DSE mutant constructs, the FGG-H1 carrying construct CC (9615C, 10034C) was used as reference construct. The reference construct used in the transfections with the DSE deletion construct contained 9615C, 10034C and the two newly introduced restriction sites, but with the DSE still present. In each series of constructs, the mean pA1/pA2-ratio of the reference construct was set at 100%. Results are expressed as mean pA1/pA2-ratio. Mean pA1/pA2-ratios of constructs were tested for differences with the pA1/pA2-ratio of the reference construct using two-sided Student's t-test. A difference was considered significant if p<0.05.
Results
FGG haplotype constructs
To study the effects of the FGG-H2 tagging SNPs 9615C>T and 10034C>T on the relative use of pA1 and pA2, we transfected HepG2 cells with constructs carrying the different alleles of these SNPs (Table 1). Mean pA1/pA2-ratios ± SD of the FGG haplotype constructs are shown in Figure 2 panel A. The pA1/pA2-ratio obtained with construct TT (9615T-10034T, FGG-H2) was 1.4-fold decreased (71.5%) compared to the pA1/pA2-ratio obtained with the reference construct CC (9615C- 10034C, FGG-H1) (p=0.015), indicating increased pA2-usage and/or decreased pA1-usage in cells transfected with construct TT.
Since FGG-H2 contains two SNPs in the region of alternative processing (9615C>T and 10034C>T), exchange constructs CT (9615C-10034T) and TC (9615T-10034C) were made, to determine which of these polymorphisms causes a reduction in the pA1/pA2-ratio. The pA1/pA2-ratio in cells transfected with construct CT was 1.2-fold decreased (85.3%) compared to the pA1/pA2-ratio obtained with the reference construct (p=0.001), while the pA1/pA2-ratio obtained with construct TC did not differ significantly from that obtained with the reference construct (101.6%, p=0.890). These results indicate that in our model system polymorphism
10034C>T, and not polymorphism 9615C>T, contributes to the reduction of the pA1/pA2-ratio observed for the FGG-H2 construct.
Figure 2 pA1/pA2-ratio ± SD for the different mini-gene constructs. Panel A: FGG haplotype constructs, panel B: DSE deletion construct, panel C: DSE mutant constructs. For each set of constructs the pA1/pA2-ratio of the reference construct was set at 100%. An asterisk indicates a significant difference (p<0.05) between the pA1/pA2-ratios obtained with the construct and the reference construct.
DSE deletion construct
To investigate the functionality and importance of the GT-rich DSE downstream of pA2, which contains the 10034C>T, we made a construct in which the complete DSE was deleted (Table 1, Figure 1b). We found that in cells transfected with this construct the pA1/pA2-ratio was 2.2-fold increased (222.3%, Figure 2 panel B;
p<0.001) compared to the pA1/pA2-ratio obtained in cells transfected with the reference construct, in which the DSE was still present. This indicates that the GT- rich DSE, containing SNP 10034C>T, plays a critical role in regulating the ratio between pA1 and pA2 derived mRNAs.
DSE mutant constructs
To further investigate the effect of mutations in the GT-rich DSE on the pA1/pA2- ratio, we made constructs in which this region was mutated at other positions than 10034. If SNP 10034C>T influences the pA1/pA2-ratio of FGG, strengthening or weakening the DSE by increasing or decreasing its GT-content and homology to the
CstF binding consensus 2a sequence should produce similar effects on the pA1/pA2- ratio as the 10034C>T change. Mean pA1/pA2-ratios ± SD of the DSE mutant constructs are shown in Figure 2 panel C. The pA1/pA2-ratio obtained with DSE mutant 1 (DSE weakened by one nucleotide) was 1.4-fold increased (142.4%) compared to the pA1/pA2-ratio obtained with the reference construct (p=0.048). In cells transfected with DSE mutant 2 (DSE weakened by two nucleotides) the pA1/pA2-ratio was 1.6-fold increased (160.7%) compared to the pA1/pA2-ratio in cells transfected with the reference construct (p=0.010). Although the pA1/pA2- ratio obtained with DSE mutant 2 was slightly higher than that obtained with DSE mutant 1, this difference was not statistically significant (p=0.412). Strengthening of the DSE by one or two nucleotides in DSE mutant 3 and DSE mutant 4 resulted in a 1.5-fold (66.2%) and 1.7-fold (59.7%) decrease in the pA1/pA2-ratio (p=0.001 and < 0.001, respectively). The difference between the pA1/pA2-ratios obtained with DSE mutants 3 and 4 was again not statistically significant (p=0.268). These results indicate that weakening or strengthening (or decreasing or increasing the GT-content) of the DSE influences the pA1/pA2-ratio of FGG accordingly.
Discussion
In the fibrinogen gamma gene, a GT-rich DSE is located in the 3' region at nucleotides 10030-10047, nine nucleotides downstream from the 3' end of the γA transcript (pA2), which shows a 78% match with the reported CstF binding site consensus 2a sequence (Figure 1b). The FGG-H2 tagging SNP 10034C>T is located in this putative CstF binding site. The T-allele strengthens this consensus to an 83%
match and increases the GT-content of the DSE. In this study we investigated the influence of the FGG-H2 tagging SNPs 9615C>T and 10034C>T on the use of pA1 and pA2 of FGG and, subsequently, the importance of the putative CstF binding site, containing SNP 10034C>T, for regulating pA2-usage. Our model system was based on the measurement of the relative pA signal usage (pA1/pA2-ratios) in HepG2 cells transfected with mini-gene constructs containing FGG exon 9, intron 9, exon 10 and the 3' region. We demonstrated that FGG-H2 tagging SNP 10034C>T, and not FGG- H2 tagging SNP 9615C>T, which is located in intron 9 near pA1, is the major contributor to the reduction of the pA1/pA2-ratio by the FGG-H2 construct. The rare T-alleles of these two SNPs are both present in FGG-H2 and recombination between these two SNPs in vivo appears very rare in the Caucasian population. It was not reported by SeattleSNPs16 and we did not find it once among the 942 subjects of the Leiden Thrombophilia Study (S. Uitte de Willige and F.R. Rosendaal, July 2006).21 This made it impossible to study the in vivo effects of the separate SNPs on DVT risk and the fibrinogen γ'/total fibrinogen ratio and made in vitro analysis of the recombinant haplotypes (9615C-10034T and 9615T-10034C) necessary.
We showed that the GT-rich DSE at nucleotides 10030-10047, containing 10034C>T, is important for the regulation of pA2 use by deleting and mutating this
site. We found a significant increase in the pA1/pA2-ratio after deleting this site.
Single nucleotide changes in this putative CstF binding site influenced the relative use of pA2 in the expected directions. However, there were no clear additive effects on the pA1/pA2-ratio when a second change was introduced. We observed that in strengthening the DSE sequence the effect of the 10033A>G change (DSE mutant 3) on the pA1/pA2-ratio (1.5-fold decrease) was stronger than that of the 10034C>T change (construct CT, 1.2-fold decrease), but similar to that of the FGG- H2 construct (construct TT, 1.4-fold decrease). These findings correspond to our previous finding that FGG-H2 is associated with approximately 40% reduction of the fibrinogen γ'/total fibrinogen ratio in vivo.15 Given the limitations of our in vitro model system, this is very similar to the 30% decrease in pA1/pA2-usage observed for the FGG-H2 construct when compared to the FGG-H1 construct. This finding will stimulate further in vitro studies of the alternative processing of the FGG pre-mRNA by using more subtle mini-gene constructs.
A potential weakness of our study is that we were unable to determine whether just the processing at pA2 is improved or that pA1-use also decreases. The reason is that we did not correct for transfection efficiency by using cotransfection with a reporter construct. Instead we focused on the ratio by which the two polyadenylation sites were used. This should not depend on the transfection efficiency, because for each construct two transcripts are produced from the same template (“γA” from pA2-use and “γ'” from pA1-use). In our mini-gene model we found that 10034T reduces the pA1/pA2-ratio, and since this allele strengthens the putative CstF binding site, we concluded that pA2 is more frequently used. We cannot distinguish whether this is at the expense of polyadenylation at pA1 or not.
Both scenarios would decrease the pA1/pA2-ratio. However, since we previously only found a correlation of FGG-H2 with reduced fibrinogen γ'/total fibrinogen ratios, and not with plasma total fibrinogen levels,15 we assumed that protein production and therefore total fibrinogen gamma mRNA synthesis does not change significantly, suggesting that the 10034T allele is responsible for a shift from pA1-usage to pA2- usage.
The term "alternative splicing" is commonly used to refer to the formation of the two alternative FGG transcripts. This view is largely based on the assumption that splicing precedes polyadenylation. However, these processes are now known to be tightly coupled in vivo.22 In the case of the FGG transcripts, it is clear that prior polyadenylation of a pre-mRNA at pA1 will prevent subsequent splicing of intron 9, whereas polyadenylation at pA2 might in fact stimulate removal of intron 9. Also, splicing of intron 9 will remove pA1 and make polyadenylation at that site impossible. Our results show an effect of a SNP in the binding site of a polyadenylation factor (CstF) on the ratio between the γ' and γA transcripts. This suggests that polyadenylation in fact plays a major role in determining the ratio
between the two transcripts. It might therefore be better to refer to "alternative polyadenylation" when describing the generation of the alternative FGG mRNAs.
Many other SNPs have been reported that are located in the 3' regulatory regions of human protein-coding genes (extensively reviewed in refs 23 and 24). Several of these are known to influence polyadenylation. A well known non-coding polymorphism, which is associated with an increased DVT risk, is the 20210G>A in the 3' UTR of the prothrombin gene. It has been demonstrated that the 20210A allele is more efficiently polyadenylated, leading to increased mRNA and protein expression.25-27 A second prothrombin gene polymorphism, 20221C>T, identified in a child with an acute vascular rejection and intrarenal segmental arterial thrombosis of an allogenic kidney transplant,28 in a 28-year old man with Budd-Chiari syndrome,29 and in a women with pregnancy complications,30 has also been shown to up-regulate prothrombin 3' end formation.31 Interestingly, this polymorphism is, like FGG 10034C>T, located in a putative CstF binding site and, like the FGG 10034C>T, the 20221C>T increases the number of uridines within the CstF binding site. This illustrates that the CstF sequence may contain clinically relevant gain-of- function mutations.
In conclusion, the results of this study support the idea that the DSE containing FGG SNP 10034C>T plays a role in the regulation of usage of the pA2-signal of FGG. By strengthening this DSE, the T-allele of SNP 10034C>T favors the formation of fibrinogen γA-chains and reduces the fibrinogen γ'/total fibrinogen ratio. The FGG-H2 haplotype predisposes to DVT and is associated with reduced fibrinogen γ' levels and reduced fibrinogen γ'/total fibrinogen ratios and these reduced ratios are also associated with an increased DVT risk.15 We propose that the 10034C>T change is the functional variation in FGG-H2 responsible for the reduction in the fibrinogen γ'/total fibrinogen ratio and the increased DVT risk.
Acknowledgements
This study was financially supported by grant 912-02-036 from the Netherlands Organization for Scientific Research (NWO). The authors would like to thank G. de Bruin and H.H.A.G.M. van der Putten for technical assistance.
References
1. Mosesson MW. The roles of fibrinogen and fibrin in hemostasis and thrombosis. Semin Hematol. 1992;29(3):177-188.
2. Herrick S, Blanc-Brude O, Gray A, Laurent G. Fibrinogen. Int J Biochem Cell Biol.
1999;31(7):741-746.
3. Kant JA, Fornace AJ, Jr., Saxe D, Simon MI, McBride OW, Crabtree GR. Evolution and organization of the fibrinogen locus on chromosome 4: gene duplication accompanied by transposition and inversion. Proc Natl Acad Sci U S A. 1985;82:2344-2348.
4. Fu Y, Grieninger G. Fib420: a normal human variant of fibrinogen with two extended alpha chains. Proc Natl Acad Sci U S A. 1994;91(7):2625-2628.
5. Francis CW, Marder VJ, Martin SE. Demonstration of a large molecular weight variant of the gamma chain of normal human plasma fibrinogen. J Biol Chem.
1980;255(12):5599-5604.
6. Mosesson MW, Finlayson JS, Umfleet RA. Human fibrinogen heterogeneities. 3.
Identification of chain variants. J Biol Chem. 1972;247(16):5223-5227.
7. Wolfenstein-Todel C, Mosesson MW. Human plasma fibrinogen heterogeneity: evidence for an extended carboxyl-terminal sequence in a normal gamma chain variant (gamma'). Proc Natl Acad Sci U S A. 1980;77(9):5069-5073.
8. Wolfenstein-Todel C, Mosesson MW. Carboxy-terminal amino acid sequence of a human fibrinogen gamma-chain variant (gamma'). Biochemistry. 1981;20(21):6146-6149.
9. Chung DW, Davie EW. Gamma and gamma' chains of human fibrinogen are produced by alternative mRNA processing. Biochemistry. 1984;23(18):4232-4236.
10. Fornace AJ, Jr., Cummings DE, Comeau CM, Kant JA, Crabtree GR. Structure of the human gamma-fibrinogen gene. Alternate mRNA splicing near the 3' end of the gene produces gamma A and gamma B forms of gamma-fibrinogen. J Biol Chem.
1984;259(20):12826-12830.
11. Siebenlist KR, Meh DA, Mosesson MW. Plasma factor XIII binds specifically to fibrinogen molecules containing gamma chains. Biochemistry. 1996;35(32):10448-10453.
12. Meh DA, Siebenlist KR, Mosesson MW. Identification and characterization of the thrombin binding sites on fibrin. J Biol Chem. 1996;271(38):23121-23125.
13. Lovely RS, Moaddel M, Farrell DH. Fibrinogen gamma' chain binds thrombin exosite II. J Thromb Haemost. 2003;1(1):124-131.
14. Pospisil CH, Stafford AR, Fredenburgh JC, Weitz JI. Evidence that both exosites on thrombin participate in its high affinity interaction with fibrin. J Biol Chem.
2003;278(24):21584-21591.
15. Uitte de Willige S, de Visser MCH, Houwing-Duistermaat JJ, Rosendaal FR, Vos HL, Bertina RM. Genetic variation in the fibrinogen gamma gene increases the risk of deep venous thrombosis by reducing plasma fibrinogen {gamma}' levels. Blood.
2005;106(13):4176-4183.
16. Nickerson, D. SeattleSNPs. NHLBI Program for Genomic Applications, UW-FHCRC, Seattle, WA. http://pga.gs.washington.edu. 2003.
17. Beyer K, Dandekar T, Keller W. RNA ligands selected by cleavage stimulation factor contain distinct sequence motifs that function as downstream elements in 3'-end processing of pre-mRNA. J Biol Chem. 1997;272(42):26769-26779.
18. MacDonald CC, Wilusz J, Shenk T. The 64-kilodalton subunit of the CstF polyadenylation factor binds to pre-mRNAs downstream of the cleavage site and influences cleavage site location. Mol Cell Biol. 1994;14(10):6647-6654.
19. Weiss EA, Gilmartin GM, Nevins JR. Poly(A) site efficiency reflects the stability of complex formation involving the downstream element. EMBO J. 1991;10(1):215-219.
20. Pfaffl MW. A new mathematical model for relative quantification in real-time RT-PCR.
Nucleic Acids Res. 2001;29(9):e45.
21. van der Meer FJM, Koster T, Vandenbroucke JP, Briët E, Rosendaal FR. The Leiden Thrombophilia Study (LETS). Thromb Haemost. 1997;78(1):631-635.
22. Kyburz A, Friedlein A, Langen H, Keller W. Direct Interactions between Subunits of CPSF and the U2 snRNP Contribute to the Coupling of Pre-mRNA 3' End Processing and Splicing. Mol Cell. 2006;23(2):195-205.
23. Chen JM, Ferec C, Cooper DN. A systematic analysis of disease-associated variants in the 3' regulatory regions of human protein-coding genes I: general principles and overview. Hum Genet. 2006;120(1):1-21.
24. Chen JM, Ferec C, Cooper DN. A systematic analysis of disease-associated variants in the 3' regulatory regions of human protein-coding genes II: the importance of mRNA secondary structure in assessing the functionality of 3' UTR variants. Hum Genet. 2006.
25. Poort SR, Rosendaal FR, Reitsma PH, Bertina RM. A common genetic variation in the 3'- untranslated region of the prothrombin gene is associated with elevated plasma prothrombin levels and an increase in venous thrombosis. Blood. 1996;88(10):3698- 3703.
26. Gehring NH, Frede U, Neu-Yilik G, Hundsdoerfer P, Vetter B, Hentze MW, Kulozik AE.
Increased efficiency of mRNA 3' end formation: a new genetic mechanism contributing to hereditary thrombophilia. Nat Genet. 2001;28(4):389-392.
27. Ceelie H, Spaargaren-van Riel CC, Bertina RM, Vos HL. G20210A is a functional mutation in the prothrombin gene; effect on protein levels and 3'-end formation. J
Thromb Haemost. 2004;2(1):119-127.
28. Wylenzek M, Geisen C, Stapenhorst L, Wielckens K, Klingler KR. A novel point mutation in the 3' region of the prothrombin gene at position 20221 in a Lebanese/Syrian family.
Thromb Haemost. 2001;85(5):943-944.
29. Balim Z, Kosova B, Falzon K, Bezzina WS, Colak Y. Budd-Chiari syndrome in a patient heterozygous for the point mutation C20221T of the prothrombin gene. J Thromb Haemost. 2003;1(4):852-853.
30. Schrijver I, Lenzi TJ, Jones CD, Lay MJ, Druzin ML, Zehnder JL. Prothrombin gene variants in non-Caucasians with fetal loss and intrauterine growth retardation. J Mol Diagn. 2003;5(4):250-253.
31. Danckwardt S, Gehring NH, Neu-Yilik G, Hundsdoerfer P, Pforsich M, Frede U, Hentze MW, Kulozik AE. The prothrombin 3'end formation signal reveals a unique architecture that is sensitive to thrombophilic gain-of-function mutations. Blood. 2004;104(2):428- 435.
