Liver Sinusoidal Endothelial Cells but Not Hepatocytes Express a Fluorescent Reporter Under the Endogenous F8 Promoter in a New Hemophilia a Mouse Model

Hemophilia A (HA) is an inherited bleeding disorder caused by mutations in the F8 gene leading to a deficiency in coagulation factor VIII (FVIII). Treatment of patients with HA has centered around replacement of FVIII protein. Gene therapy provides the possibility of permanently altering the HA phenotype. Gene therapy for HA induces FVIII expression in hepatocytes, despite the well supported belief that FVIII is not made within hepatocytes but rather in liver sinusoidal endothelial cells (LSECs). Ectopic expression of FVIII in hepatocytes has been implicated in mild toxicities and FVIII decline observed in adeno-associated viral (AAV) vector-based gene therapy, the causes of which remain unknown. Extrahepatic production of FVIII, e.g. in lymphoid tissues, has been controversial. A better understanding of where FVIII is produced is needed to improve gene therapy for HA and inform development of other therapeutic interventions.

We developed in collaboration with Jackson Laboratory a new strain of hemophilic mice on the C57BL/6J background expressing the fluorescent protein mScarlet under the endogenous F8 promoter. Using the CRISPR/Cas9 gene editing system, we inserted a transgene encoding mScarlet into exon 1 of the mouse F8 gene through homology-directed repair. We selected mScarlet for its relative brightness and stability, to maximize the chances of detecting its expression despite weak F8 promoter activity. The donor DNA consisted of two homology arms flanking the mScarlet-encoding transgene followed by the bovine growth hormone polyadenylation [poly(A)] signal. Insertion of the transgene at the first F8 codon in exon 1 and inclusion of poly(A) ensured knock-out of F8. Male animals showed a bleeding phenotype consistent with other HA mouse strains, such as C57BL/6J HA and B6/129 HA. We performed retro-orbital bleeding and tested mouse plasma samples using the activated partial thromboplastin time (aPTT) assay. Animals expressing mScarlet (n=4) and C57BL/6J HA mice (n=3) had similarly prolonged aPTT (mean 57 seconds) compared to non-hemophilic C57BL/6J mice (n=5, 41 seconds).

Production of mScarlet was first evaluated in the hepatocytes and LSECs of mScarlet mice (n=3) using flow cytometry. The livers were perfused in situ via the portal vein with a perfusion and collagenase medium. The livers were then processed into single cell suspensions, and hepatocytes and LSECs were separated by centrifugation. An anti-CD31 antibody was used to identify LSECs. Hepatocytes were identified by exclusion of CD45.2⁺ cells. CD31⁺cells from mScarlet mice but not from control C57BL/6J mice expressed mScarlet. Neither mouse strain had detectable mScarlet in hepatocytes.

Flow cytometry was then used to evaluate mScarlet in the splenic endothelial cells (ECs) and leukocytes. Using in situtranscardiac perfusion, mScarlet or control C57BL/6J mice received a perfusion and collagenase medium. Following processing into single cell suspensions, ECs were isolated from other splenic cells by magnetic separation using CD31 microbeads. Anti-CD31 and anti-CD45.2 antibodies were used to identify ECs and splenic leukocytes, respectively. Neither cell type from mScarlet or control mice had detectable mScarlet.

We next evaluated mScarlet expression via fluorescent microscopy of liver and spleen sections. Before sectioning, tissues were fixed using in situ transcardiac perfusion fixation. Since mScarlet fluorescence was not detectable in tissue sections, we employed an anti-mScarlet antibody. Anti-mScarlet binding was detectable in liver sections from mScarlet but not control C57BL/6J mice. The staining pattern was consistent with the sinusoidal arrangement of hepatic capillaries running between rows of hepatocytes toward central venules. We found the same staining pattern using an anti-CD31 antibody, supporting that the mScarlet-expressing cells were LSECs but not hepatocytes. No anti-mScarlet binding was detected in the spleens from either mScarlet or C57BL/6J mice.

We propose that FVIII is produced in LSECs but not in hepatocytes, splenic endothelial cells, splenic or hepatic leukocytes. Our novel mouse model expressing a well-defined and readily detectable reporter protein overcomes the limitations of immunohistochemistry using anti-FVIII antibodies, which complicated previous studies on the cellular origins of FVIII. Further investigation of extrahepatic FVIII production is ongoing.