A conditional knockout mouse model reveals endothelial cells as the principal and possibly exclusive source of plasma factor VIII

Abstract

The cellular source of coagulation factor VIII (FVIII) remains controversial. Like many coagulation proteins, FVIII is produced in the liver, and FVIII synthesis has long been associated with hepatocytes. But extrahepatic synthesis also occurs, and mounting evidence suggests that hepatocytes are not responsible for FVIII production. To determine the tissue that synthesizes FVIII, we developed a Cre/lox-dependent conditional knockout (KO) model in which exons 17 and 18 of the murine factor VIII gene (F8) are flanked by loxP sites, or floxed (F8(F)). In cells expressing Cre-recombinase, the floxed sequence is deleted, resulting in F8(F→KO) gene inactivation. When F8(F) mice were crossed with various tissue-specific Cre strains, we found that hepatocyte-specific F8-KO mice are indistinguishable from controls, whereas efficient endothelial-KO models display a severe hemophilic phenotype with no detectable plasma FVIII activity. A hematopoietic Cre model was more equivocal, so experimental bone marrow transplantation was used to examine hematopoietic FVIII synthesis. FVIII(null) mice that received bone marrow transplants from wild-type donors were still devoid of plasma FVIII activity after hematopoietic donor cell engraftment. Our results indicate that endothelial cells are the predominant, and possibly exclusive, source of plasma FVIII.

Figure 1

Conditional F8-KO alleles. (A) Targeted F8 gene. (B) F8^F allele. Excision of the neomycin resistance cassette by Flp recombinase produces the floxed (F8^F) allele, which is expressed normally. (C) F8^KO allele. Excision of exons 17/18 from the F8^F allele by Cre recombinase produces a F8^KO allele. (D) Alternative mRNA splicing of the F8^KO allele. The predicted exon 16/19 splice [F8^KO(16/19)] and an alternatively spliced transcript [F8^KO(Alt)] in which 46 bp of intron 16 is retained are produced in approximately equal amounts. (E) Domain structure of normal FVIII protein. The position of exons 17/18 within the A3 domain is noted. (F) Predicted FVIII polypeptide encoded by F8^KO(16/19) mRNA. (G) Predicted FVIII polypeptide encoded by F8^KO(Alt) mRNA. Flp recombinase target (FRT) sites are represented as open triangles, and Cre recombinase target (loxP) sites as gray triangles. The positions of duplex genotyping primers P1, P2, and P3 and RT-PCR primers P4 and P5 are shown. aa, amino acid; Ex, exon.

Figure 2

Genotyping PCRs. WBCs or tail DNA PCR was used for routine genotyping. A Cre-positive (experimental) and Cre-negative (control) male (F8 F/y) and female (F8 F/F) are shown for each tissue-specific Cre model. (A) Duplex PCR of WBC DNA using a common forward primer, P1, in combination with 2 alternative reverse primers, P2 and P3. As depicted in Figure 1B-C, a 712-bp product is amplified from the F8^F allele and a 462,bp product from the F8^KO allele. (B) Singleplex PCR of WBC DNA using primers P1 and P2 amplifies only the 712-bp F8^F allele-specific product. (C) Singleplex PCR of WBC DNA using primers P1 and P3 amplifies a 462 bp F8^KO allele-specific product but can also amplify a 1683-bp product from the F8^F allele. (D) A Cre-recombinase coding sequence–specific PCR of WBC DNA identifies Cre⁺ mice. No signal is seen for the Vav1-Cre transgene because it contains a codon-optimized iCre sequence, which is not recognized by the native Cre primers we used. (E) Duplex PCR analysis of tail DNA using primers P1, P2, and P3 detects both F8^F and F8^KO alleles, especially important for Vav1- and Tek-Cre mice, in which WBC DNA undergoes complete F8^F→KO gene conversion.

Figure 3

Plasma FVIII activity. A chromogenic assay was used to measure FVIII activity in plasma from tail bleeds. Males and females are shown separately, revealing an apparent sex difference. F8^F represents Cre^−/− control mice that result from breeding F8^F/y Cre^+/− males of the various Cre stains with F8^F/F females. Alb, Vav1, Cdh5^(Mlia), Cdh5^(Spe), and Tek represent experimental Cre^+/− littermates. F8^KO is the stably inherited exon 17/18-deleted strain. Hepatocyte-specific Alb-Cre has no effect on FVIII levels, whereas all endothelial Cre models result in reduced plasma FVIII, culminating with a severe hemophilic phenotype in the most efficient Tek-Cre model. *P ≤ .05, **P ≤ .001. n.s., nonsignificant.

Figure 4

Analysis of tissue gDNA and mRNA. Tissue samples were collected from males following whole-animal saline perfusion. DNA and total RNA were isolated from the same tissue sample and analyzed by allele-specific PCR. Except for the F8^KO/y mouse, all animals inherited a F8^F/y genotype, either without Cre or with coinheritance of a Cre transgene as indicated. (A) A total of 25 ng of gDNA was amplified using a duplex PCR that yields a 712-bp F8^F allele-specific product and a 462-bp F8^KO allele-specific product. Primer P1, P2, and P3 binding sites are shown in Figure 1B-C. (B) For RT-PCR, 100 ng of total RNA was reverse transcribed and then analyzed by PCR using primers P4 and P5, which span the exon 17/18 deletion cassette (as shown in Figure 1D). An 819-bp product is amplified from the F8^F allele, representing normally spliced F8 mRNA. For the F8^KO allele, in addition to the 407-bp product predicted for exon 16/19 mRNA splicing (F8^KO(16/19)), a 453-bp product is also present due to alternative splicing (F8^KO(Alt)). The “Standards” panels consist of PCR products amplified from defined mixtures of F8^F/y and F8^KO/y liver gDNA (panel A) or total RNA (panel B) at various ratios as indicated (F8^F:F8^KO). n.a. indicates a sample that was not analyzed.

Figure 5

Transplantation of WT bone marrow into FVIII^null mice. Unfractionated C57BL/6 BMNCs or BMMCs were transplanted into lethally irradiated FVIII^null recipients. (A) Plasma FVIII levels in BMT recipients. Tail blood samples were collected, and FVIII levels were determined by chromogenic assay. (B) Tail-clip survival test. The tail clipping test was performed at 15 weeks after BMT.