Patterns of expression of factor VIII and von Willebrand factor by endothelial cell subsets in vivo

Abstract

Circulating factor VIII (FVIII) is derived from liver and from extrahepatic sources probably of endothelial origin, but the vascular sites of FVIII production remain unclear. Among organs profiled, only liver and lymph nodes (LNs) show abundant expression of F8 messenger RNA (mRNA). Transcriptomic profiling of subsets of stromal cells, including endothelial cells (ECs) from mouse LNs and other tissues, showed that F8 mRNA is expressed by lymphatic ECs (LECs) but not by capillary ECs (capECs), fibroblastic reticular cells, or hematopoietic cells. Among blood ECs profiled, F8 expression was seen only in fenestrated ECs (liver sinusoidal and renal glomerular ECs) and some high endothelial venules. In contrast, von Willebrand factor mRNA was expressed in capECs but not in LECs; it was coexpressed with F8 mRNA in postcapillary high endothelial venules. Purified LECs and liver sinusoidal ECs but not capECs from LNs secrete active FVIII in culture, and human and mouse lymph contained substantial

Fviii: C activity. Our results revealed localized vascular expression of FVIII and von Willebrand factor and identified LECs as a major cellular source of FVIII in extrahepatic tissues.

Figure 1

Expression of F8 mRNA in various mouse tissues. The data are from BioGPS and Geo Profiles and are presented as raw expression values.

Figure 2

Expression of F8 and vWF mRNA in fluorescence-activated cell sorted subtypes of ECs from various tissues. Lymphoid tissue ECs were released from the surrounding stroma by enzymatic digestion while preserving their viability, and the sorted EC subsets were then transcriptomically profiled. The transcriptomes of the sorted EC subtypes were combined with data from sorted ECs^, analyzed with the same microarray platform (Affymetrix Mouse Gene 1.0 ST arrays) and normalized on the basis of GeneSpring preprocessing and default normalization (RMA-16). The levels of F8 and vWF mRNA were plotted from these normalized transcriptomes. Raw Affymetrix expression values of at least 3 biological replicates (except n = 1 for PLN CAP; some values were excluded because of contamination with HEV-specific genes). Mean ± standard error of the mean. BM, bone marrow; cap, capillary; CNS, central nervous system; FRC, fibroblastic reticular cell; LP, lamina propria; MLN, mesenteric lymph node; PLN, peripheral lymph node; PP, Peyer’s patches SLN, subcutaneous lymph node.

Figure 3

Biologically active FVIII is present in cell lysates or culture supernatants of purified LECs and LSECs, but not in those of conventional capECs. FVIII activity was measured by using a Biophen FVIII:C Kit and was normalized to equal cell numbers. (A) Biological activity of FVIII in cell lysates of purified LECs and LSECs. Mean ± standard deviation of 3 biological replicates except lung LECs and lung capECs that were the average of 2 experiments. (B) FVIII activity in supernatants (24 hours of cell culture) was detected from primary cultured LSECs and human dermal LECs (HDLECs), but not from primary brain microvascular ECs (BMECs) and human umbilical vein ECs (HUVECs). The activity was normalized for cell number. The data are from 3 biological replicates.

Figure 4

Substantial amounts of biologically active FVIII are present in human prenodal and mouse thoracic or cyst lymph. Lymph was collected as described in the text, and FVIII:C activity was measured by using the Biophen FVIII:C Kit. The data represent FVIII:C activities of individual lymph samples relative to that of human or mouse pooled normal plasma level of FVIII (defined as 100%) as described in “Materials and methods.”