Variations in the prothrombin gene contribute to plasma prothrombin levels and the risk of venous thrombosis. The 20210G[RIGHTWARDS ARROW]A variation in the 3′-UTR region is associated with high plasma prothrombin levels (1.32 U mL−1 in AG- vs. 1.05 U mL−1 in GG-carriers) and with an almost 3-fold increased risk of thrombosis [1]. In addition, the G-allele of the common 19911G[RIGHTWARDS ARROW]A polymorphism in the last intron (M) of the prothrombin gene is associated with 4 U dL−1 higher prothrombin levels than the 19911A allele, leading to an 8 U dL−1 difference between AA- and GG-carriers. On its own, the 19911G allele is not a risk factor for thrombosis, but it seems to modulate the risk of the 20210A allele, if 20210A is present on the other chromosome [2,3]. More recently, additional rare variations in the 3′ end of the prothrombin gene have been reported at positions 20207, 20209, 20218, and 20221 [4–8], some of which have also been functionally tested [9]. All these variations were found by screening for the 20210A mutation by melting curve analysis using the Lightcycler technology. Two variations in the 3′ end of the prothrombin gene, 20207A[RIGHTWARDS ARROW]C and 20218A[RIGHTWARDS ARROW]G were first reported by Meadows and coworkers [5]. Here, we describe the functional analysis of these variants.

Heterozygous carriers of the 20207A[RIGHTWARDS ARROW]C or 20218A[RIGHTWARDS ARROW]G mutations were identified in the laboratory of E.L. Both individuals consented to their DNA being used for the present study. In an approach essentially identical to that used for the 20210G[RIGHTWARDS ARROW]A variation [10], DNA fragments corresponding to the 3′ untranslated region (UTR) of exon 14 and the downstream region were generated using polymerase chain reaction (PCR) and were cloned directly downstream of the luciferase cDNA in pGL3-basic, thereby replacing the SV40 late polyadenylation signal present in this plasmid. In addition, constructs were generated in which the fragments were present in tandem, which will allow a direct comparison of the effectiveness of the wild-type and mutant sites. All possible combinations of variant and wild-type sequences were generated (Fig. 1). The resulting constructs were transfected into human HepG2 cells and luciferase activity in the extracts was determined. In addition, cytoplasmic RNA was prepared from the transfected cells 24 h after transfection and reversely transcribed using an oligo-dT primer carrying a specific sequence at its 5′ end (5′-AGC TGG TCA GTC GTC AGC TGA (T)16-3′). The resulting DNAs were subjected to reverse transcriptase (RT)-PCR using a primer in the luciferase cDNA and a primer identical to the 5′ part of the oligo dT-primer [10]. The fragments were analyzed on agarose gel. In the case of the tandem constructs, the ratio between the products derived from the two polyadenylation sites was determined and expressed as the normalized polyA-ratio, in which the pA1/pA2 ratio of a construct was normalized by division by the pA1/pA2 ratio of the wt/wt construct (see also the legend of Fig. 1). To determine the location of the polyA-attachment site, the same RT-PCR fragments were ligated into pCR4-TOPO vector (Invitrogen, Carlsbad, USA) and sequenced.


Figure 1.  Result of reverse transcriptase-polymerase chain reaction for determination of normalized-poly(A) site ratios for the tandem constructs. Fragments were separated on an agarose gel. From left to right: M = size marker, 01 = construct with a single wild-type 3′-UTR element as a reference and control, 02 = 20207C/wt (pA1/pA2), 03 = wt/20207C, 04 = 20207C/20207C, 05 = 20218G/wt, 06 = wt/20218G, 07 = 20218G/20218G, 08 = wt/wt, 09 = untransfected control, Bl = no template control. A schematic representation of the constructs is shown to the left of the figure. The primers used are indicated by arrowheads. Note that the reverse primer is directed against a sequence present in the oligo-dT primer. This sequence is present only in cDNAs and is not present in the constructs themselves. The second polyadenylation unit is dotted in the lower construct, because it is absent in the single construct used in lane 2. Normalized poly(A) site ratios (n-poly(A) site ratio) and their 95% confidence intervals are shown below the photograph of the gel for each tandem construct. Each experiment (in which constructs were transfected in triplicate) was repeated four times (n = 12). Mean relative luciferase expression and 95% CIs after transfection in HepG2 cells of the constructs are shown in the next line. Results represent the means of two independent experiments in which two different DNA preparations were transfected in triplicate (n = 12).

Download figure to PowerPoint

First, we investigated the possibility that the 20207 and the 20218 mutations did alter the polyadenylation sites of the hybrid luciferase-prothrombin mRNAs. The G at position 20210 was the last non-A nucleotide in the majority of mRNAs (in 70% of cloned cDNAs for the 20207C construct and in 63% for 20218G). The heterogeneity seemed to be somewhat higher in the variant constructs than in the wild-type construct, in which 86% of the cDNAs ended at position 20210, but the major polyadenylation site remained at G20210 and/or A20211, as was the case with the mRNAs carrying the 20210G[RIGHTWARDS ARROW]A variation [10]. Next, the effects of the 20207A[RIGHTWARDS ARROW]C and the 20218A[RIGHTWARDS ARROW]G variations on polyadenylation efficiency was investigated, mainly using the tandem constructs. Cytoplasmic RNA was isolated from transfected HepG2 cells and reverse transcribed and the ratio between the pA1- and pA2-derived products was determined (Fig. 1). Ratios for constructs with two identical 3′ UTR elements were not significantly different: n-poly(A) site ratios were 1.00 [95% confidence interval (CI) 0.93, 1.07], 1.08 (95% CI 0.94, 1.21) and 1.16 (95% CI 1.00, 1.32) for wt/wt, 20207C/20207C and 20218G/20218G constructs, respectively. The data on the mixed wild-type variant constructs (Fig. 1) suggest that the wild-type poly(A) site is somewhat stronger than the variant sites, although the differences are small compared with those obtained with the corresponding 20210A constructs, in which the variant site was strongly preferred [n-poly(A) site ratios of 0.5 (95% CI 0.38, 0.62) and 2.15 (95% CI 1.88, 2.43) for wt/20210A and 20210A/wt constructs, respectively][10]. We conclude that the ratio between polyadenylation at site 1 or site 2 was not severely affected by mutations at positions 20207 or 20218. Mutations at these sites certainly do not lead to better polyadenylation sites, and may even decrease the efficiency of polyadenylation somewhat.

Lastly, we determined whether the different mutations affected protein expression. Luciferase expression levels for each construct were expressed relative to that of the single or tandem wild-type construct. Mean relative expression levels of the 20207C and 20218G were 89% (95% CI 84, 94) and 92% (95% CI 86, 98), respectively. The data on the tandem constructs are shown in Fig. 1. Levels were somewhat lower than those of the wild-type construct, which was set at 100% (95% CI 96, 104), and are in line with the mRNA data. We conclude that in our in vitro system mutations at positions 20207 and 20218 do not lead to significant changes of luciferase expression levels.

Five rare variations around the 20210-positions are known: 20207A[RIGHTWARDS ARROW]C, 20209C[RIGHTWARDS ARROW]A and 20218A[RIGHTWARDS ARROW]G (detected by Meadows et al. [5]), 20209C[RIGHTWARDS ARROW]T [6,8] and 20221C[RIGHTWARDS ARROW]T [4,7–9]. All variants are rare and four of them were not found in any of our own patient cohorts (20209C[RIGHTWARDS ARROW]A was not tested). At the level of the caucasian population, these variations will therefore not contribute significantly to the observed variation in prothrombin levels and to the development of venous thrombosis. However, this does not mean that such variations are not functional. Our results showed that the A to C change at position 20207 and the A to G change at position 20218 do not result in major changes in the position of the poly(A) attachment site, in the effectiveness of polyadenylation or in protein expression. Based on the results obtained with our model system, we would expect that these mutations do not contribute to the thrombotic risk of the patients carrying them. This is in contrast with our results on the 20210G[RIGHTWARDS ARROW]A mutation [10], using the same experimental set-up as used for the analysis of the present two variations, and with the results of Danckwardt et al. [9], who showed that the 20221C[RIGHTWARDS ARROW]T change also results in increased activity of the prothrombin polyadenylation region.


  1. Top of page
  2. References
  • 1
    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: 3698703.
  • 2
    Ceelie H, Bertina RM, Van Hylckama Vlieg A, Rosendaal FR, Vos HL. Polymorphisms in the prothrombin gene and their association with plasma prothrombin levels. Thromb Haemost 2001; 85: 106670.
  • 3
    Perez-Ceballos E, Corral J, Alberca I, Vaya A, Llamas P, Montes R, Gonzalez-Conejero R, Vicente V. Prothrombin A19911G and G20210A polymorphisms' role in thrombosis. Br J Haematol 2002; 118: 6104.
  • 4
    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: 9434.
  • 5
    Meadows CA, Warner D, Page S, Lyon E. Detection of novel mutation using fluorescent hybridization probes and melting temparature analysis. J Mol Diagn 2002; 3: 195.
  • 6
    Warshawsky I, Hren C, Sercia L, Shadrach B, Deitcher SR, Newton E, Kottke-Marchant K. Detection of a novel point mutation of the prothrombin gene at position 20209. Diagn Mol Pathol 2002; 11: 1526.
  • 7
    Balim Z, Kosova B, Falzon K, Bezzina Wettinger S, Colak Y. Budd–Chiari syndrome in a patient heterozygous for the point mutation C20221T of the prothrombin gene. J Thromb Haemost 2003; 1: 8523.
  • 8
    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: 2503.
  • 9
    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: 42835.
  • 10
    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: 11927.