Hepatic targeting of transplanted liver sinusoidal endothelial cells in intact mice


  • Daniel Benten,

    1. Departments of Medicine and Pathology, Marion Bessin Liver Research Center, Cancer Research Center, Diabetes Research Center, Albert Einstein College of Medicine, Bronx, NY
    Search for more papers by this author
    • Daniel Benten and Antonia Follenzi contributed equally to the study.

  • Antonia Follenzi,

    1. Departments of Medicine and Pathology, Marion Bessin Liver Research Center, Cancer Research Center, Diabetes Research Center, Albert Einstein College of Medicine, Bronx, NY
    Search for more papers by this author
    • Daniel Benten and Antonia Follenzi contributed equally to the study.

  • Kuldeep K. Bhargava,

    1. Division of Nuclear Medicine, Long Island Jewish Medical Center, New Hyde Park, NY
    Search for more papers by this author
  • Vinay Kumaran,

    1. Departments of Medicine and Pathology, Marion Bessin Liver Research Center, Cancer Research Center, Diabetes Research Center, Albert Einstein College of Medicine, Bronx, NY
    Search for more papers by this author
  • Christopher J. Palestro,

    1. Division of Nuclear Medicine, Long Island Jewish Medical Center, New Hyde Park, NY
    Search for more papers by this author
  • Sanjeev Gupta

    Corresponding author
    1. Departments of Medicine and Pathology, Marion Bessin Liver Research Center, Cancer Research Center, Diabetes Research Center, Albert Einstein College of Medicine, Bronx, NY
    • Albert Einstein College of Medicine, Ullmann Building, Room 625, 1300 Morris Park Avenue, Bronx, NY 10461
    Search for more papers by this author
    • fax: 718-430-8975

  • Potential conflict of interest: Nothing to report.


Targeting of cells to specific tissues is critical for cell therapy. To study endothelial cell targeting, we isolated mouse liver sinusoidal endothelial cells (LSEC) and examined cell biodistributions in animals. To identify transplanted LSEC in tissues, we labeled cells metabolically with DiI-conjugated acetylated low density lipoprotein particles (DiI-Ac-LDL) or 111Indium-oxine, used LSEC from Rosa26 donors expressing β-galactosidase or Tie-2-GFP donors with green fluorescent protein (GFP) expression, and tranduced LSEC with a GFP-lentiviral vector. LSEC efficiently incorporated 111Indium and DiI-Ac-LDL and expressed GFP introduced by the lentiviral vector. Use of radiolabeled LSEC showed differences in cell biodistributions in relation to the cell transplantation route. After intraportal injection, LSEC were largely in the liver (60 ± 13%) and, after systemic intravenous injection, in lungs (67 ± 9%); however, after intrasplenic injection, only some LSEC remained in the spleen (29 ± 10%; P < .01), whereas most LSEC migrated to the liver or lungs. Transplanted LSEC were found in the liver, lungs, and spleen shortly after transplantation, whereas longer-term cell survival was observed only in the liver. Transplanted LSEC were distinct from Kupffer cells with expression of Tie-2 promoter-driven GFP and of CD31, without F4/80 reactivity. In further studies using radiolabeled LSEC, we established that the manipulation of receptor-mediated cell adhesion in liver sinusoids or the manipulation of blood flow–dependent cell exit from sinusoids improved intrahepatic retention of LSEC to 89 ± 7% and 89 ± 5%, respectively (P < .01). In conclusion, the targeting of LSEC to the liver and other organs is directed by vascular bed–specific mechanisms, including blood flow–related processes, and cell-specific factors. These findings may facilitate analysis of LSEC for cell and gene therapy applications. (HEPATOLOGY 2005.)

Endothelial cells (EC) play critical roles in organogenesis, tissue vascularization, and organ homeostasis.1–4 Liver sinusoidal endothelial cells (LSEC) exhibit unique properties, such as hyaluronic acid receptors5, 6 and production of coagulation factor VIII,7 and play a role in the modulation of immune responses.8–10 Angiogenesis requires proliferation of EC and mesenchymal cells as well as complex interactions between regulators of vasculogenesis and blood vessel maturation.11 Recent insights helped focus interest in the potential of EC in cell and gene therapies, ranging from blood vessel repair to organ reconstitution.12–16 Stem cells capable of generating EC are of considerable interest in defining mechanisms in lineage advancement and in cell therapy.17, 18 Studies using hematopoietic, mesenchymal, and epithelial cells indicate that analysis of cell engraftment and function is necessary to establish the fate of transplanted cells.19–21 This process involves cell targeting in specific organs (e.g., to establish suitable routes of cell administration or to assess cell trafficking in individual organs).

We were interested in transplanting LSEC, because this will be relevant for treating coagulation disorders such as factor VIII deficiency.7, 18 Moreover, hepatic endothelial damage is associated with conditions, such as sepsis, impaired liver function, veno-occlusive disease, and graft versus host disease, that could potentially benefit from endothelial reconstitution.22–25 In principle, transplanted LSEC could enter the liver through the systemic or portal circulation, because hepatic sinusoids receive blood from both the portal vein (70%) and the hepatic artery (30%).26 However, if LSEC escaped from the liver by exiting through central veins as a result of their small size, engraftment of LSEC could be affected.27 Therefore, we labeled LSEC with radionuclide, metabolic, or transgene markers and used donor cells natively expressing the green fluorescent protein (GFP) under control of endothelial-specific Tie-2 promoter28, 29 to establish cell-targeting mechanisms.


LSEC, liver sinusoidal endothelial cell; DiI-Ac-LDL, DiI-conjugated acetylated low density lipoprotein particles; GFP, green fluorescent protein; EC, endothelial cell; DMEM, Dulbecco's modified Eagle medium; MAA, macroaggregated albumin; MCT, monocrotaline; PBS, phosphate buffered saline; FBS, fetal bovine serum; ICAM-1, intercellular adhesion molecule 1; PECAM-1, platelet endothelial cell adhesion molecule 1; Ig, immunoglobulin; PFA, paraformaldehyde; In, indium; Tc, technetium; PBS-T, PBS containing 0.1% Triton X-100; PCR, polymerase chain reaction.

Materials and Methods

Animals and Animal Procedures.

F344 rats and C57BL/6 and FVB/N mice were obtained from the National Cancer Institute (Bethesda, MD). 129-Gt(ROSA)26Sor/J transgenic mice in C57BL/6J background and TgN(TIE2GFP)287Sato/J transgenic mice in FVB/N/J background (20–25 g) were obtained from the Jackson Laboratory (Bar Harbor, ME). The institutional Animal Care and Use Committees approved animal usage in conformity with the National Research Council's Guide for the Care and Use of Laboratory Animals (United States Public Health Service publication, revised 1996).

For cell transplantation, the spleen, portal vein, or inferior vena cava was exposed by laparotomy, and 1 to 2 × 106 LSEC were injected in 0.3 mL serum-free Dulbecco's modified Eagle medium (DMEM) (GIBCO, Grand Island, NY) using 27-gauge needles. Hemostasis was secured by pressure on injection sites or splenic ligature. Some animals received 53 mg/kg fibronectin-like protein polymer (Deepwater Chemicals Inc., Woodward, OK) or 1 × 106 macroaggregated albumin (MAA) particles (Pulmolite; CIS-US Inc., Bedford, MA) in saline intraportally immediately before cell transplantation. Other animals received 200 mg/kg monocrotaline (MCT) (Sigma, St. Louis, MO) in saline intraperitoneally 24 hours before intraportal cell transplantation.

Cell Isolation Procedures.

Mouse cells: The liver was perfused at 5 mL/min via the portal vein for 15 minutes with buffer at 37°C containing 1.9 mg/mL EGTA, for 2 minutes with buffer lacking EGTA, and for 5 to 8 minutes with buffer containing 0.03% (wt/vol) collagenase and 5 mmol/L CaCl2*2H2O. The perfusion buffer contained 10 mmol/L HEPES, 3 mmol/L KCl, 130 mmol/L NaCl, 1 mmol/L NaH2PO4/H2O, and 10 mmol/L D-glucose (pH 7.4) (all chemicals obtained from Sigma Chemical Co., St. Louis, MO; collagenase obtained from Worthington Biochemical Corp., Lakewood, NJ). The liver was dissociated in perfusion buffer, passed through dacron fabric with 80-μm pores, and centrifuged at 50g for 5 minutes to isolate hepatocytes, which were washed twice with DMEM at 4°C and kept on ice. Nonparenchymal cells in the supernatant were kept at room temperature as recommended by Braet et al.,30 washed at 50g for 2 minutes, pelleted at 350g for 7 minutes twice, and fractionated with Percoll (Sigma). Stock Percoll solution was prepared by mixing 9 parts Percoll with 1 part (v/v) ×10 phosphate-buffered saline (PBS) (pH 7.4). Percoll gradients were prepared in 15-mL tubes with 2 mL 75% stock Percoll solution (diluted with PBS) at the bottom and 1.5 mL of 35% stock Percoll solution at the top. Cells were resuspended in PBS and layered in 1 mL at the top of Percoll for centrifugation at 900g for 20 minutes without brakes. LSEC were in the lower of two bands and were recovered in 1 mL, mixed with an equal volume of PBS, and centrifuged at 900g for 7 minutes. The LSEC pellet was washed with DMEM at 350g for 7 minutes. For transplantation, LSEC were resuspended in serum-free DMEM.

Rat cells: F344 rat cells were isolated by perfusing liver at 10 mL/min as described,31 with buffer containing EGTA for 10 minutes, plain perfusion buffer for 2 minutes, and collagenase for 12–15 minutes. Rat LSEC were enriched with Percoll according to previously published protocols.30

Cell Viability and Characterization.

For sizing, wet cell preparations were microphotographed and compared with 15-μm-diameter latex microspheres (DuPont, NEN Company, Boston, MA) using Adobe Photoshop (Adobe Systems Inc., San Jose, CA). Cell viability was tested via trypan blue dye exclusion, uptake of DiI-Ac-LDL, and attachment to collagen-coated dishes in 199 medium with 15% fetal bovine serum (FBS) and antibiotics (GIBCO). For lipoprotein uptake, LSEC suspensions were incubated with 1:20 diluted DiI-Ac-LDL (Biomedical Technologies, Stoughton, MA) for 1 hour at 37°C. Cells were washed twice and observed under fluorescence microscopy.

To immunostain intracellular adhesion molecule 1 (ICAM-1, CD54) and platelet endothelial cell adhesion molecule 1 (PECAM-1, CD31), cell smears were air-dried and fixed in cold acetone for 10 minutes. For ICAM-1, cells were rinsed in PBS, blocked with 3% FBS and 0.05% avidin (Sigma) for 30 minutes each, and incubated with biotin-conjugated anti-mouse CD54 (1:500) (BD Biosciences Pharmingen, San Diego, CA) for 1 hour at room temperature. After washing with PBS, cells were incubated with avidin-FITC (Sigma) for 1 hour at room temperature and counterstained with DAPI-Antifade (Molecular Probes, Eugene, OR). PECAM-1 staining used rat anti-mouse CD31 (BD Biosciences Pharmingen) diluted 1:100 in 5% donkey serum followed by Cy2-conjugated donkey anti-rat immunoglobulin (Ig) G (1:100) (Jackson ImmunoResearch Labs., Inc., West Grove, PA). For the macrophage/Kupffer cell marker, F4/80, cell smears were fixed with 4% paraformaldehyde (PFA) in PBS for 10 minutes. After blocking, cells were incubated with rat anti-mouse F4/80 (1:500) (Serotec, Raleigh, NC) followed by anti-rat IgG labeled with Alexa Fluor 546 (1:500) (Molecular Probes) for 1 hour at room temperature.

Nuclear Medicine Procedures.

For radiolabeling, 3 to 6 × 106 LSEC or hepatocytes were pelleted at 350g and 50g, respectively, resuspended in normal saline, and incubated for 30 minutes at 37°C with 18.5 MBq indium-111(111In)-oxine (Mallinckrodt Inc., Hazelwood, MO), 370 MBq technetium-99m (99mTc) pertechnetate (UltraTag kit, Mallinckrodt Inc.), or 370 MBq 99mTc-exametazime (Ceretec; Amersham Biosciences, Piscataway, NJ). Cells were washed twice, and labeling efficiency was determined by measuring activity in cell pellets and supernatants with a dose calibrator. After labeling, cell viability was retested as described in Cell Viability and Characterization.

Cell biodistribution was analyzed in C57BL/6 mice 1 hour after transplanting 1 × 106111In-oxine–labeled LSEC with a gamma camera (Argus; ADAC Laboratories, Milpitas, CA). Energy discrimination used a 15% window centered on the 174- and 247-keV photopeaks of 111In. Dorsal images were acquired for 5 minutes on a 64 × 64 × 16 matrix. After imaging, 111In activity was counted in excised spleen, liver, lungs, kidneys, heart, and small intestine (Cobra II counter; Packard Instrument Co., Meridien, CT). Activity was expressed as a fraction of total radioactivity in each of these six organs.

Lentiviral Marking of LSEC With a GFP Transgene.

To obtain vesicular stomatitis virus (VSV)-pseudotyped lentiviral stocks, we cotransfected pCCL.PPT.hPGK.GFP.Wpre transfer construct, the third-generation packaging constructs pMDLg/pRRE and pRSV-REV, and the pMD2.G envelope into 293T cells, followed by ultracentrifugation of medium, as described.32 Titering of the vector on HeLa cells showed 1 to 2 × 109 transducing units per millliliter. For transduction, 4 × 106 LSEC were incubated in suspension with 4 × 108 transducing unit–concentrated lentivirus for 90 minutes at 37°C in a final volume of 1 mL. LSEC were washed twice with DMEM at 300g for 5 minutes before transplantation.

Polymerase Chain Reaction for Rosa26 LacZ Transgene.

Genomic DNA was extracted with DNeasy Tissue Kit (Qiagen Inc., Valencia, CA). LacZ primers were 5′TTCCGTCATAGCGATAACGAG3′ (forward) and 5′ACCGCATCAGCAAGTGTATCT3′ (reverse). Platinum PCR Supermix (Invitrogen, Carlsbad, CA) was used with 35 cycles at 94°C × 3 minutes, 94°C × 30 seconds, 55°C × 1 minute, 72°C × 1 minute, and final elongation at 72°C × 7 minutes. Primers for mouse GAPDH were 5′GGGTGGAGCCAAACGGGTC3′ (forward) and 5′GGAGTTGCTGTTGAAGTCGCA3′ (reverse) with 25 cycles at 94°C × 3 minutes, 94°C × 30 seconds, 56°C × 1 minute, 72°C × 1 minute, and 72°C × 7 minutes. The products were resolved in 1% agarose gels with expected sizes of 572 bp (LacZ) and 550 bp (GAPDH).

Transplanted Cells in Tissues.

To identify DiI-Ac-LDL–containing LSEC, spleen, liver, and lung samples were collected 1 hour after cell transplantation and frozen at −80°C in methylbutane. Seven-micrometer cryosections were fixed in PFA, counterstained with DAPI, and examined under fluorescence microscopy. To identify GFP-expressing LSEC, livers 1 and 2 weeks after cell transplantation were fixed in 4% PFA for 6 hours at 4°C, equilibrated for 48 hours in 20% sucrose, and frozen in methylbutane at −80°C. Five-micrometer cryosections were postfixed in PFA, washed in PBS, and examined under fluorescence microscopy. GFP immunostaining used sections blocked with 3% goat serum in PBS containing 0.1% Triton X-100 (PBS-T) for 1 hour at RT, followed by incubation with rabbit anti-GFP in PBS-T (1:300) (Molecular Probes) and peroxidase-conjugated anti-rabbit IgG (1:300) (Sigma) for 1 hour each at RT, with development using diaminobenzidine (DAB+ kit; DAKO Cytomation, Carpinteria, CA) and hematoxylin counterstaining. To distinguish between Kupffer cells and transplanted GFP-expressing cells, liver was fixed in 4% PFA, equilibrated in 20% sucrose, and frozen in methylbutane at −80°C as above. Five-micrometer-thick cryostat sections were postfixed in PFA; blocked in 5% goat serum, 1% bovine serum albumin, and 0.1% PBS-T; and incubated with rabbit anti-GFP (1:300) (Molecular Probes) and rat anti-mouse F4/80 (1:500) (Serotec). Sections were incubated with FITC-conjugated goat α-rabbit IgG and with Alexa Fluor 546–conjugated goat anti-rat IgG. Nuclei were stained with DAPI-Antifade.

Recovery and Analysis of Transplanted Cells.

Liver cells were isolated 1 week after cell transplantation and hepatocytes were removed by 50g centrifugations. The nonparenchymal cell fraction containing 2 million cells was incubated with 5 μL purified anti-mouse CD16/CD32 (Fcγ III/II receptor, BD Biosciences Pharmingen) for 15 minutes at 4°C. After washing with 4% FBS/PBS, cells were incubated with 2 μL allophycocyanin-labeled anti-mouse CD31 (BD Biosciences Pharmingen) for 20 minutes at 4°C, washed, and resuspended in 4% FBS/PBS. To identify nonviable cells, 1 μg/mL propidium iodide (Sigma) was added. Flow cytometry was performed using FACSCalibur, with 1 to 2 million events per sample, and CellQuest software (BD Biosciences Pharmingen).

Statistical Methods.

Data are expressed as the mean ± SD. The significance of differences was analyzed by Student t test, where applicable, using SigmaStat software (Jandel Scientific, San Rafael, CA). A P value of less than .05 was considered significant.


Characterization of Mouse Liver Cells.

Trypan blue was excluded by 91% to 100% LSEC. Compared with hepatocytes, LSEC were smaller (diameter 7.6 ± 1.5 vs. 28.0 ± 6.6 μm; P < .001 [t test]) (Fig. 1A–B). All cells in the LSEC fraction incorporated DiI-Ac-LDL (Fig. 1C), and virtually all expressed ICAM-1 (CD54), whereas ICAM-1 expression in hepatocytes was minimal (Fig. 1D–E). LSEC expressed PECAM-1 (CD31) in up to 70% (Fig. 1F), with F4/80 expression in 10.5 ± 1.8% cells (Fig. 1H), although hepatocytes were rare (<1%-2%). LSEC attached to culture dishes with characteristic cytoplasmic morphology (Fig. 1I).

Figure 1.

Cell characteristics. Hepatocytes were frequently (A) binucleated and (B) larger than LSEC. (B) The LSEC fraction contained rare hepatocytes (arrow). (C) DiI-Ac-LDL incorporation in a LSEC aggregate. (D) Hepatocytes expressed ICAM-1 minimally, whereas (E) LSEC were ICAM-1–positive with (F) CD31 expression in 70%. (G) The hepatocyte fraction lacked Kupffer cells, whereas (H) LSEC contained 10% Kupffer cells (arrows). (I) LSEC 3 days after culture with characteristic morphology. (Original magnification ×400.) H&E, hematoxylin–eosin; LSEC, liver sinusoidal endothelial cells; DiI-Ac-LDL, DiI-conjugated acetylated low density lipoprotein particles; ICAM-1, intercellular adhesion molecule 1; PECAM, platelet endothelial cell adhesion molecule.

Radiolabeling of Cells.

We initially studied rat liver cells (n = 3), because conditions for radiolabeling rat hepatocytes were available.31, 33 Rat hepatocytes incorporated all three labels with 50% to 80% efficiencies (Table 1). Although rat LSEC incorporated 111In-oxine, neither 99mTc-exametazime nor 99mTc-pertechnetate was incorporated. Moreover, 111In incorporation in LSEC was less than that observed in hepatocytes. Studies of mouse LSEC generally reproduced these findings with no 99mTc incorporation (Table 1). Further analysis of 111In-labeling showed that mouse LSEC (n = 7) incorporated 111In threefold more efficiently than rat LSEC (P < .001 [t test]), whereas 111In incorporation in rat and mouse hepatocytes was similar. The viability of 111In-labeled mouse LSEC was unaltered, as judged by trypan blue exclusion and attachment in dishes (n = 3).

Table 1. Efficiency of Liver Cell Radiolabeling
RadiolabelIncorporation of Activity (%)
  • *

    P < .001 (rat LSEC vs. rat hepatocytes [t test]).

  • P < .001 (mouse LSEC vs. rat LSEC [t test]).

  • Tested once.

111Indium-oxine15.6 ± 1.9*81.8 ± 9.152.9 ± 20.0*88.2 ± 1.2
99mTc-exametazime4.3 ± 0.5*72.2 ± 8.55.358.5
99mTc-pertechnetate1.3 ± 0.4*52.2 ±

Biodistribution of Radiolabeled LSEC in Intact Animals.

To identify differences in organ-specific LSEC targeting, we injected 111In-labeled LSEC into the spleen, portal vein, or inferior vena cava in C57BL/6 mice (n = 3 each). Imaging showed rapid translocation of LSEC in vascular beds (Fig. 2A–C). Intrasplenic injection deposited LSEC in the spleen, liver, and lungs, indicating extrasplenic cell translocations into liver and pulmonary capillaries. In contrast, LSEC were primarily found in the liver after intraportal injection. Similarly, after injection into the inferior vena cava, LSEC were largely observed in the lungs.

Figure 2.

Imaging of recipients transplanted with 111In LSEC. Animals are shown 1 hour after LSEC injection into the (A) spleen, (B) portal vein, or (C) inferior vena cava. The administration route affected cell distributions. (D) Transplanted cell distribution as a fraction of total 111In activity in all excised organs. *P < .001. Spl, spleen; LSEC, liver sinusoidal endothelial cells.

Analysis of 111In activity in organs indicated that 1 hour after intrasplenic injection, more LSEC were observed in the lungs (42 ± 7%) than in the liver (21 ± 4%) (P < .001 [t test]) (Fig. 2D). This result occurred despite drainage of splenic blood into the portal vein, which offers access initially to hepatic sinusoids. Retention of LSEC in the spleen itself was relatively limited (29 ± 10%), similar to the retention of transplanted hepatocytes.33 Little or no activity was found in the heart or small intestine (0.9 ± 0.1% and 0.4 ± 0.1%, respectively), although 7 ± 2% of 111In activity appeared in kidneys. Transplanted LSEC were best targeted to the liver after intraportal injection (60 ± 13%), although some cells migrated into the lungs (26 ± 13%) and spleen (5 ± 2%) (P < .001 [t test]), whereas activity in other organs was similar to that observed after intrasplenic transplantation (kidneys, 7 ± 2%; heart, 1 ± 0.1%; intestine, 0.7 ± 0.0%). After intravenous injection, the activity was mostly found in the lungs (70 ± 9%) or liver (22 ± 8%) but was limited in other organs (spleen, 2 ± 1%; kidneys, 7 ± 1%; heart, 1.6 ± 0.7%; intestine, 0.8 ± 0.4%).

Transplanted LSEC in Tissues.

To verify the targeting of LSEC in organs, we identified Rosa26 donor cells in C57BL/6 mice by LacZ polymerase chain reaction (PCR), which demonstrated transplanted LSEC in the spleen, liver, and lungs 1 hour after intrasplenic injection (Fig. 3A). The transgene was absent from the kidneys and small intestine, suggesting alternative explanations for 111In activity in these organs. After intravenous injection of cells, LacZ sequences were present in the lungs, spleen, liver, and small intestine, a result that was in agreement with the entry of LSEC into the systemic circulation (Fig. 3B). After intraportal injection of cells, LacZ PCR was positive for the liver, lungs, and possibly kidneys (Fig. 3C). To demonstrate transplanted cells in the longer term, we studied additional animals 1 week after intraportal injection of Rosa26 LSEC. PCR analysis of two such recipients showed transplanted cells in the liver in both mice, with only a faint PCR signal in the spleen of one mouse, whereas other organs, including lungs, showed no evidence for transplanted cell survival (Fig. 3D).

Figure 3.

PCR demonstration of LSEC. Rosa26 cells were transplanted into the (A) spleen, (B) inferior vena cava, or (C–D) portal vein, and tissues were obtained (A–C) 1 hour or (D) 7 days later. LacZ (gel on left) and GAPDH PCR products are shown. Lanes 1–6, individual cell recipients; lane 6, donor liver; lane 7, PCR mix alone; lane 8, 100-bp DNA ladder. MW, molecular weight.

To visualize transplanted LSEC in tissues with additional markers (e.g., within 1 hour), we transplanted syngeneic LSEC with DiI-Ac-LDL into two C57BL/6 mice per transplantation route. After intraportal injection, DiI-Ac-LDL–labeled LSEC were present in the liver (Fig. 4A), and lungs, a result that was in agreement with our radionuclide and PCR studies. Transplanted LSEC were observed in the periportal areas of the liver lobule, within hepatic sinusoids. Transplanted LSEC were seen in pulmonary capillaries after intravenous, intraportal, and intrasplenic injections (Fig. 4B) and in the splenic pulp after intrasplenic transplantation (Fig. 4C). Very few LSEC were observed in the liver after intravenous injection.

Figure 4.

Transplanted LSEC in tissues. (A–C) Recipients of DiI-Ac-LDL–containing LSEC 1 hour after injection into (A) portal vein, (B) inferior vena cava, or (C) spleen. Fluoresence microscopy showed DiI-labeled LSEC (arrows) (A) in sinusoids adjacent to portal areas, (B) pulmonary capillaries, and (C) splenic pulp. (D) GFP-expressing LSEC in liver sinusoids from a recipient of GFP-lentivirus–transduced cells (arrows). (E) GFP immunostaining in the same liver sample with more LSEC (arrows), including cells with elongated cytoplasmic processes lining sinusoids (arrowhead). (F) Negative control without GFP antibody. (G) GFP and F4/80 staining of Tie-2-GFP donor liver showing GFP in EC (green) and F4/80 in Kupffer cells (red, arrowhead). (H) Tie-2-GFP LSEC with GFP immunostaining (arrows) in congeneic FVB/N mouse liver 1 week after transplantation. (I) GFP and F4/80 costaining in liver in panel h demonstrated F480 in Kupffer cells (arrowhead) and its absence in Tie-2-GFP LSEC (arrows). (Original magnification ×400.) Panels A–B,D,G–H, DAPI stain; panels E–F, hematoxylin counterstain. Pa, portal area.

For long-term cell tracking, we marked freshly isolated LSEC with a GFP-lentiviral vector, which transduced 60% LSEC as indicated by GFP expression 72 hours after cell culture using microscopy or flow cytometry (data not shown). When LSEC were transduced with the GFP-lentivirus and transplanted immediately via the portal vein into syngeneic C57BL/6 mice (n = 3), we observed GFP-expressing cells with typical cytoplasmic processes in the liver after 1 week (Fig. 4D–E), as well as 2 weeks, further establishing that transplanted cells were targeted to the liver. The endothelial identity of these cells was verified by the analysis of FVB/N recipients 1 week after intraportal transplantation of Tie-2-GFP LSEC (Fig. 4G). GFP was expressed in Tie-2-GFP donors in the lining major vessels of EC, as well as sinusoids. GFP-positive LSEC were nonreactive for F4/80, which was observed in Kupffer cells throughout liver sinusoids. In Tie-2-GFP cell recipients, transplanted cells lined liver sinusoids (Fig. 4H), similar to Fig. 4D–E. Moreover, these cells were nonreactive for F4/80 (Fig. 4I), again confirming that the presence of Kupffer cells did not confound our results.

Reisolation of Transplanted LSEC.

To recover transplanted LSEC for analysis, we isolated liver cells with collagenase perfusion of recipient FVB/N mice (n = 4) 1 week after transplantation of 2 × 106 Tie-2-GFP LSEC. Flow cytometry of the nonparenchymal cell fraction from a Tie-2-GFP donor mouse showed GFP expression in 70% (Fig. 5A). In Tie-2-GFP cell recipients, LSEC were present within the nonparenchymal cell fraction (Fig. 5C), including with CD31 display, that further verified their endothelial identity (Fig. 5D). These transplanted GFP-expressing LSEC constituted 0.2 ± 0.1% of the nonparenchymal liver cell fraction.

Figure 5.

Flow cytometry for Tie-2-GFP cells. (A) Tie-2-GFP donor cells with GFP expression in 70% (R3 gate). (B) Nontransgenic FVB/N mouse liver cells showing absence of GFP-expressing cells. (C) GFP-expressing cells in the liver of a Tie-2-GFP LSEC recipient. These cells constituted 0.35% of the nonparenchymal cell fraction. (D) Analysis of cells shown in panel C with additional CD31 staining showed that virtually all GFP-expressing cells displayed CD31, confirming their endothelial identity. GFP, green fluorescent protein. R4 gate shown by the arrow indicates these cells.

Improving Intrahepatic Targeting of LSEC.

To determine whether LSEC targeting to the liver could be enhanced, we used MCT to induce endothelial injury25; infusion of an engineered fibronectin-like polymer (fibronectin), which helps in adhesion of EC34; or MAA particles for preventing egress of LSEC from hepatic sinusoids.35 Analysis of cell targeting 1 hour after transplantation showed that MCT did not improve intrahepatic retention of LSEC (Fig. 6). In contrast, the use of fibronectin or MAA particles significantly enhanced intrahepatic retention of LSEC (89 ± 5% and 89 ± 7%, respectively, vs. 60 ± 13% in controls; P < .001 [t test]) and decreased cell translocation in lungs (8 ± 4% and 4 ± 1%, respectively, vs. 26 ± 13% in controls; P < .001 [t test]). The goal of these studies was to establish the value of radiolabeled cells for tracking organ-specific retention and not increased cell engraftment in the longer term.

Figure 6.

Enhanced 111In-LSEC targeting. Fractional distribution of LSEC in mice pretreated with MCT, fibronectin-like polymer, and MAA particles. Fibronectin-like polymer and MAA improved intrahepatic retention of LSEC and decreased intrapulmonary cell translocation. *P < .001 (t test) versus untreated control animals. LSEC, liver sinusoidal endothelial cells; MCT, monocrotaline; MAA, macroaggregated albumin.


These studies indicated that suitable fractions of primary LSEC could be isolated, labeled with reporters, and targeted to specific organs. These findings will be relevant for studying the mechanisms in LSEC-mediated cell and gene therapy, as well as the trafficking and homing of EC in organs. Our cell isolation technique was suitable for obtaining highly enriched LSEC, although Kupffer cells were not eliminated. Previously, murine LSEC were isolated by centrifugal elutriation or magnetic sorting, which are more complex processes.7, 10 The simpler density gradients may be sufficient for some applications. We used Tie-2-GFP cells to exclude the possibility that Kupffer cells would be mistaken for LSEC. The Tie-2 gene encodes a receptor tyrosine kinase for angiopoietins that participate in angiogenesis.1 In turn, Tie-2 expression is governed by regulatory sequences that restrict gene expression to EC.28, 29

Previous studies have shown that rodent and human hepatocytes effectively incorporate 111In and 99mTc labels.31, 33, 36 Similarly, leukocytes and other cells incorporate 111In and 99mTc labels, and bone marrow–derived mesenchymal stem cells incorporate 99mTc-exametazime.37–39 In contrast, mouse and rat LSEC incorporated 111In-oxine but not 99mTc. Superior 111In incorporation in mouse (vs. rat) LSEC suggested regulation of 111In uptake in LSEC. This outcome may explain the variable efficiencies of 111In incorporation in LSEC preparations that have previously been recognized in other cell types.31, 40, 41

Optimal targeting of cells to organs is essential for addressing the therapeutic potential of the cells and mechanisms in transplanted cell and stem cell biology. Intravascular targeting of LSEC represents a challenge because of their small size and their potential for translocation into additional organs through vessels. Cell–vascular relationships play critical roles in the intrahepatic targeting of hepatocytes, which are larger than EC and are more readily restricted to the liver.27, 33 Similarly, myocardial targeting of mesenchymal stem cells depends on vessel and cell size differences, because intravenous injection entraps cells in lungs, and intracardiac injection is necessary for depositing cells in the myocardium.39

LSEC targeting after systemic intravenous, intraportal, or intrasplenic injections was accounted for in part by mechanical cell entrapment in vascular beds. Certainly, retention of LSEC in the liver and lungs after injection into the portal vein or vena cava, respectively, will be consistent with the size of hepatic sinusoids (5–7 μm) and pulmonary capillaries (3 μm). Additional mechanisms might regulate the distribution of transplanted LSEC. For instance, after injection of radiolabeled LSEC into the inferior vena cava, greater hepatic 111In activity compared with other organs suggested that cells were redirected to the liver. However, our studies using DiI-Ac-LDL–containing LSEC showed only occasional transplanted LSEC in the liver after vena caval injection. Therefore, the possibility remains that destruction of transplanted LSEC in the lungs produced secondary hepatic uptake of 111In, as observed previously for hepatocytes.35 Also, specific attachment or adherence factors might have contributed to LSEC targeting in the liver. Receptor interactions have been established between native liver cells and transplanted cells (e.g., bone marrow–derived stem cells, which require stromal cell–derived factor 1 for homing in the liver).42 Similarly, intravenous injection of pulmonary EC did not produce cell targeting in the liver or spleen,43 indicating both cell and organ type–specific differences. However, determining whether EC derived from extrahepatic organs will more easily survive and engraft in the liver requires prospective analysis. Notably, our studies showed that despite LSEC translocations in lungs, these cells were cleared within 1 week, indicating additional requirements for long-term cell survival.

Our findings concerning perturbations to enhance intrahepatic targeting of LSEC were in agreement with mechanovascular and receptor-mediated mechanisms. MAA particles are entrapped in proximal hepatic sinusoids, in view of their larger size, similar to hepatocytes.35, 44 Increased intrahepatic retention of LSEC in animals treated with MAA particles verified that LSEC exited from the liver via central veins rather than from intrahepatic portal systemic channels. Similarly, use of fibronectin-like polymer addressed the possibility of modulating LSEC adhesion in the liver, because this polymer promotes angiogenesis.34 It should be noted that LSEC express fibronectin receptors.45 Increased hepatic LSEC retention in animals treated with fibronectin-like polymer indicates that this approach merits further study. Finally, we examined whether sinusoidal endothelial disruption with MCT will alter LSEC retention in the liver. We established that MCT produces extensive endothelial damage in mouse liver (unpublished data, 2003). However, MCT did not improve intrahepatic retention of LSEC. Whether increased initial retention of transplanted LSEC will be associated with greater long-term cell survival is another issue. It may be that the mechanisms proposed (mechanical perturbation [MAA particles], receptor-mediated cell adhesion [fibronectin-like polymer], and superior access to the endothelial bed [MCT]) may affect cell engraftment differently.

We reisolated up to 0.35% of transplanted Tie-2-GFP LSEC in the nonparenchymal liver cell fraction. This likely represented a significant underestimation of the number of transplanted LSEC in the liver. For instance, not all EC in the Tie-2-GFP donor liver expressed GFP, and recovery of LSEC after collagenase perfusion was quite incomplete, because we isolated only 2 to 3 million LSEC from approximately 25 million total cells expected in the mouse liver. In our studies, transplantation of 2 million LSEC in the mouse liver was the equivalent of 8% of the LSEC mass (assuming 25 million LSEC total). Therefore, this very incomplete LSEC recovery could be consistent with engraftment of 5% to 10% of transplanted LSEC, and would have produced at most 0.4% to 0.8% Tie-2-GFP cells in the liver. Our recovery of up to 0.35% Tie-2-GFP LSEC in recipients should be consistent with these possibilities, as well as with engraftment of only 10% to 20% primary hepatocytes in the liver.27 However, mechanisms have been developed to enhance both engraftment and proliferation of transplanted hepatocytes in the liver, including liver repopulation.46 Therefore, our studies represent initial efforts that will require additional insight to begin understanding the therapeutic potential of liver reconstitution with LSEC.

In conclusion, the hepatic sinusoidal endothelium regulates the access of circulating molecules to hepatocytes, which are separated by the space of Disse from LSEC. Despite the physical separation, interactions between LSEC and hepatocytes are critical in organogenesis and remain important in the adult liver. For instance, the vascular endothelial growth factor stimulates LSEC to produce hepatocyte growth factor and other cytokines, which play cytoprotective, trophic, and regulatory roles in hepatocytes.3 Moreover, transplantation of LSEC offers therapeutic potential in coagulation disorders, including hemophilia A.18, 47 Besides the possibility of transducing cultured LSEC via retroviral and lentiviral vectors,48, 49 the feasibility of lentiviral gene transfer in freshly isolated LSEC, as shown here, will offer additional opportunities for cell and gene therapy.