Hemophilia A (HA), the most common inherited bleeding disorder in males, is caused by a deficiency of the human coagulation factor VIII protein, which catalyzes a key step in the clotting cascade [1,2]. In hemophiliacs, FVIII activity is impaired by a variety of different mutations in the FVIII gene (F8) on chromosome Xq28 [3,4]. The most common mutations are single-base substitutions in the coding region of F8 and a recurrent intron 22 inversion [5,6]. Furthermore, small and large deletions, duplications, insertions, splice site mutations and a further recurrent intron 1 inversion are rare causes of HA phenotypes [6–9]. The overall mutation detection rate is about 97% . For moderate and mild HA, mutation analysis failed in 4% and 12% cases, respectively .
In our laboratory, F8 diagnostic standard procedures, including analytical PCRs for intron 22 and intron 1 inversions, sequencing of the coding regions and flanking intronic sequences, and multiplex ligation-dependent probe amplification, did not reveal any causative mutation in 89 of > 2500 HA patients.
Recently, three different single-base substitutions in the promoter region of F8 were identified and suspected to account for HA [11,12]. For one of these, evidence for causality was demonstrated with a luciferase assay .
During the initial cloning of human F8 in 1984, the 5′-flanking sequence including the promoter was determined up to nucleotide position c.–1175 . About 10 years later, Figueiredo and Brownslee  narrowed down the sequence elements necessary for F8 expression in the main promoter region to ∼ 200 bp from positions c.–279 to c.–64.
Therefore, we sequenced the main promoter region of F8 (∼ 300 bp upstream of the F8 initiation codon) on an ABI 3130xl sequencer (Applied Biosystems, Life Technologies, Carlsbad, CA, USA) in 89 HA patients without mutations detected by standard methods. Two different sequence variants were identified: c.–219C>T and c.–255A>G. The first one was reported previously by Bogdanova et al. , and the second has not been described before. Both patients suffer from a mild form of HA (detailed FVIII plasma levels were not available). To date, only two further F8 promoter mutations have been reported, by Bogdanova et al.  (c.–112G>A) and Riccardi et al.  (c.–257T>G).
To investigate the causality of the four known promoter mutations, we performed luciferase assays and compared the activity of the mutated sequences with that of the wild type. For this purpose, the wild type and all four mutated F8 promoter sequences were cloned next to the 5′-end of the firefly luciferase gene in the promoterless pGL3-Basic vector (Promega, Mannheim, Germany). Each of the constructs was cotransfected into HEK293-EBNA and HepG2 cells, together with a pRL-SV40 vector (Promega) containing a Renilla luciferase gene (essential for normalizing) and a pGBKT7 vector (Clontech, Mountain View, CA, USA), which improves transfection efficiency. Two cell lines were used in order to validate our results in different cell types with high and low endogenous FVIII expression, one derived from human embryonic kidney (HEK293) and one from a human liver carcinoma (HEPG2). FVIII is primarily expressed in the liver, and results with HEPG2 cells are therefore expected to more closely represent the in vivo situation. Luciferase expression was measured with a luminometer, and the determined promoter activities are shown in Fig. 1.
The c.–219C>T mutation showed the greatest influence on promoter activity, which was reduced to ∼ 0.5% of that of the wild type after expression in HEK293 cells (Fig. 1A) and to ∼ 6% after expression in HEPG2 cells (Fig. 1B). McGlynn et al.  have characterized different protein-binding sites within the F8 promoter by DNase I protection assays in rat liver nuclear extracts. They identified a transcription factor-binding site for an as yet unidentified protein at positions c.–233 to c.–216 (binding site 5 from c.–62 to c.–45 according to McGlynn et al. ). The unknown transcription factor, which is assumed to be essential for full F8 promoter activity, may not be able to bind sufficiently to the corresponding binding site in the c.–219C>T promoter variant. If the unknown transcription factor acts in a tissue-specific manner, we propose that it has higher activity in liver-derived cells such as HEPG2 cells, and therefore enhances FVIII expression in these cells. However, Dai et al.  have reported a luciferase activity for the c.–219C>T variant of about 23%, which is in close agreement with the patient’s FVIII plasma level. We reinvestigated this mutation in order to validate the implementation of the luciferase assay in our laboratory. The discrepancy could be explained by the use of different vector constructs and different cell lines. Also, it is essential to remark that the luciferase assay is an indirect way of determining promoter activities, and the results cannot be compared directly with FVIII levels measured in patients’ plasma.
The patient carrying the novel promoter c.–255A>G variant suffers from mild HA, like his brother, who shows the same substitution in F8 inherited from the mother. The results of the luciferase assay in HEK293 cells (Fig. 1A) and HEPG2 cells (Fig. 1B) show an activity decrease to ∼ 5% of that of the wild type, supporting our assumption that this mutation is causative. Moreover, McGlynn et al.  identified a transcriptional binding site for D-site-binding protein (DBP) and CCAAT/enhancer-binding-protein (C/EBP) in the F8 promoter at positions c.–250 to c.–280. Both proteins are members of the basic region leucine zipper family, and bind to several cis-acting elements in F8 [16,17]. Transfection experiments with a luciferase reporter gene vector containing the main promoter region with C/EBP expression vectors revealed a strong increase in promoter activity, underlining the functional importance of these transcriptional factors . In addition, cotransfections with both C/EBP and DBP also generated strong trans-activation, indicating their transcriptional relevance in F8 promoter activation . Several positions of the binding sequence of C/EBP from positions c.–262 to c.–254 (TKNNGYAAK) are highly conserved, including position c.–255, which requires an A for optimal binding of C/EBP . Thus, the c.–255A>G mutation is assumed to prevent efficient binding of C/EBP, resulting in a decrease in F8 transcription and, consequently, in a mild HA phenotype, as observed in our patient.
Two nucleotides adjacent to the above-mentioned mutation, Riccardi et al.  detected a further nucleotide substitution at position c.–257. Although these two mutations are very close to each other, they seem to have different effects on the F8 promoter: c.–255A>G reduces promoter activity to ∼ 5% of that of the wild type, and c.–257T>G decreases it to ∼ 40%, in both cell lines (Fig. 1A,B). Position c.–257 is also part of the C/EBP-binding sequence, but it is less conserved than position c.–255, and is therefore more variable (C or T). In order to analyze the nucleotide tolerance at position c.–257, we performed mutagenesis experiments on this position: c.–257T>A and c.–257T>C. These substitutions showed the same effect on promoter activity as the suspected promoter mutation (Fig. 1A,B). Our results suggest that any nucleotide other than T at position c.–257 will lead to a decrease in promoter activities in luciferase assays. In addition, family analyses and the diagnostic findings of Riccardi et al.  support the conclusion that this F8 promoter mutation is indeed causative.
Finally, we analyzed the substitution at position c.–112G> A, which had been described earlier by Bogdanova et al. . In contrast to what was found for the three other mutations, luciferase assays showed only small decreases in promoter activity, the activities being up to ∼ 60% of that of the wild type after expression in HEK cells (Fig. 1A) and 80% after expression in HEPG2 cells (Fig. 1B). Position c.–112 is not located in any known transcription factor binding region. Therefore, this position and its flanking nucleotides appear to be of less importance for transcriptional regulation of the F8 promoter and FVIII expression. This is in accordance with the mild phenotype of the patient reported by Bogdanova et al. .
In conclusion, we were able to identify a new mutation in the F8 promoter (c.–255A>G) and to verify, with luciferase assays, the causality of all four F8 promoter mutations described in the literature so far. In our cohort, promoter variants account for < 0.1% of HA cases. Therefore, screening of the F8 main promoter region as a routine diagnostic method may help to decrease the number of unsolved HA cases.