Digestion of the Murine Liver for a Flow Cytometric Analysis of Lymphatic Endothelial Cells

Studies analyzing liver lymphatics have primarily used immunohistochemistry to quantitate the frequency and diameter of lymphatic vessels in the liver. However, this method does not allow for the evaluation of LECs on a cell-by-cell basis or for expression of multiple markers, cytokines, chemokines, or transcription factors. Therefore, we asked whether liver LECs could be isolated from the liver and evaluated using flow cytometry. Previous work isolating lymph node LECs was performed using Liberase DL (collagenase I-II-and-dispase) and DNase 1 combined with the mechanical separation of the lymph node tissue with needles^14,16. Therefore, the same digestion protocol on liver tissue was used, or collagenase IV and DNase 1 was used as previously described, for immune cell isolation from the liver²² and followed the digestion with a density gradient separation (iodixanol) step to remove hepatocytes and increase the frequency of other cell types within the liver (Figure 1A). Following liver digestion and centrifugation with the density gradient, the cells were stained with CD45, CD31, PDPN, and CD16/32. CD16/32 has been described to be expressed by mature LSEC populations, but not LEC populations¹⁸. Thus, LECs were gated as CD45-, CD31+, CD16/32-, PDPN+ cells, while LSECs were gated as CD45-, CD31+, CD16/32+ and PDPN- (Figure 1A). Gates were set on live cells (ghost red negative) and based on isotype controls and FMO staining (Figure 1A). LYVE-1 APC on the LEC population was also confirmed (Figure 1B). Both methods allowed the visualization of the LEC population within the liver; however, the staining profile of the cells was visually better when using collagenase IV and DNase 1 than collagenase I-II-and-dispase (Figure 1C). Therefore, all future manipulations were performed using collagenase IV and DNase 1, as described in the protocol. On average, we obtained approximately 1,300 LECs per gram of liver tissue or 2,200 LECs per naive liver, using this digestion method.

To optimize the gating strategy and combination of fluorophores and to eliminate contamination of LSECs, both CD16/32 and CD146 were used. CD16/32 is Fc gamma receptors II and III and is expressed on mature LSECs but not on LECs¹⁸. CD16/32 could distinguish the PDPN+ CD16/32- cells from the PDPN- CD16/32+ cells, especially when using PDPN conjugated to APC or PE and CD16/32 conjugated to FITC (Figure 1A), but less well when PDPN was conjugated to PE-Cy7 (Figure 1C). The expression of CD16/32 is not found on LECs but is found on LSECs. Fc block uses the CD16/32 antibody to minimize the Fc receptor binding and nonantigen specific binding of immunoglobulins to the Fc receptors. Since CD16/32 was used to visualize LSECs, the non-specific binding of immunoglobulins was higher and CD146 was optimized as an alternative to measure LSECs. Therefore, we tested CD146 conjugated to V450, a vascular endothelial cell marker expressed highly by LSECs and other vascular endothelial cells²⁰ but with low to no expression by LECs²³. We first optimized the staining of CD146 using FMO and isotype controls for both CD16/32 and CD146 (Figure 2A). If we evaluated only the CD16/32+ cells or the CD16/32- cells, the CD16/32+ cells were all CD146+ and PDPN- while the CD16/32- cells were primarily CD146- and either PDPN- or PDPN+ (Figure 2B). To confirm this staining, FMO was used. By removing PDPN from the stain, the PDPN+ population disappeared (Figure 2C). Interestingly, there were no CD146+ PDPN+ cells, confirming that CD146 is not or very lowly expressed by PDPN+ LECs. Thus, we are confident that either of these markers can be used to gate out vascular endothelial cells in the liver.

To further validate that PDPN is an appropriate marker for LECs, liver sections from mice were stained with both PDPN (green) and F4/80 (brown) (Figure 3A). Neither the vascular endothelium nor macrophages expressed PDPN in the murine liver. We were also able to distinguish cholangiocytes, which can stain positive for PDPN, from lymphatic vessels, based on the distinct nuclear structures of the bile ducts and by cytokeratin 7 staining (Figure 3B; red: cytokeratin 7, green: PDPN). Since multiple markers are used for flow cytometry, such as CD31, which is not expressed by cholangiocytes²⁴, we are confident that we are removing these cells from the analysis, thereby confirming that cells from the liver that are CD45-, CD31+, CD146^lo/neg, CD16/32-, and PDPN+ are LECs. Finally, to provide evidence that the stained cells are LECs, we sorted this population of cells and used qualitative real-time polymerase chain reaction to evaluate the Vegfr3 and prox-1 expressions. Both Vegfr3 and prox-1 were evaluated, as these transcripts are expressed by other cells in the liver—prox1 by hepatocytes and Vegfr3 by LSECS (among others) — but no other cells besides LECs express both. The expression of both these markers was significantly higher in the sorted population than in the cultured murine macrophage cell line (RAW264.7) which does not normally express these markers but is similar in expression to cultured murine LECs (Figure 3C).

Figure 1: Representative flow cytometry analysis of collagenase I-II-dispase and collagenase IV-digested murine liver tissue. (A) This panel shows the gating strategy, fluorescence minus one staining, and isotype controls for the procedure. (B) This panel shows the LYVE-1 staining of CD16/32- PDPN+ cells. (C) This panel shows the final flow cytometry gate after digesting mouse livers with either collagenase I-II-dispase (e.g., Liberase DL) or collagenase IV and evaluating CD16/32 PerCP X PDPN PE-Cy7 where LECs are CD16/32- and PDPN+. Please click here to view a larger version of this figure.

Figure 2: Identification of CD146 and PDPN as appropriate markers for liver lymphatic endothelial cells. (A) This panel shows fluorescence minus one and isotype controls for CD16/32 (left) and CD146 (right). (B) This panel shows gated CD16/32 positive or negative cells determined from panel A. Shown is CD146XPDPN from both populations. (C) This panel shows fluorescence minus one for PDPN to demonstrate that the staining is absent when the antibody is not added. Please click here to view a larger version of this figure.

Figure 3: Identification of flow PDPN+ cells as lymphatic endothelial cells. (A) This panel shows representative immunohistochemistry from a mouse liver stained with PDPN (blue/green) and F4/80 (brown). The lymphatic vessel (LV), blood vessel (BV), and macrophage (MF) are labeled. Scale bar = 100 µm. Formalin-fixed paraffin-embedded tissue was deparaffinized for 20 min in xylene. Tissues were hydrated to water through a gradient of ethanol, and antigen retrieval was performed using pH 6 antigen retrieval buffer in a pressure cooker for 15 min. Tissue was blocked using 0.1% BSA and stained using anti-mouse PDPN and anti-mouse F4/80 for 1 h at room temperature. Anti-hamster IgG HRP and anti-rabbit IgG HRP were used as secondary antibodies. 3,3'-Diaminobenzidine (DAB)+ and Vina Green were used to detect the F4/80 and PDPN, respectively. Tissue was counterstained with hematoxylin and imaged on a Microscope. (B) This panel shows the same experiment performed as in panel A, except here, PDPN is shown in green, cytokeratin 7 in red, and 4′,6-diamidino-2-phenylindole (DAPI) in blue. Scale bar = 100 µm. Tissue was blocked using 5% donkey and 5% goat serum and stained using anti-mouse PDPN (8.1.1) 1:100 and anti-mouse cytokeratin 7 1:200 for 1 h at room temperature. Anti-hamster IgG AF647 and anti-rabbit IgG-PE were used as secondary antibodies. Tissue was counterstained with DAPI and imaged. Scale bar = 100 µm. The lymphatic vessel (LV), blood vessel (BV), and bile duct (BD) are labeled. (C) This panel shows a log fold change in Vegfr3 and prox-1 expression from sorted liver LECs based on the staining in the protocol (CD45-, CD31+, CD146^lo/neg, and PDPN+) compared to RAW cells or primary murine lymph node LECs. Sorted cells were passed through a biopolymer-shredding column, RNA was extracted using an RNA extraction kit, and cDNA was made using a reverse transcription kit. Transcript abundance was normalized to the housekeeping gene, Gapdh, for every sample. Please click here to view a larger version of this figure.