Normal Lymphatic Development and Function in Mice Deficient for the Lymphatic Hyaluronan Receptor LYVE-1^{▿ †}

Generation of LYVE-1 gene-targeted mice.

Gene targeting of the LYVE-1 locus was accomplished using VelociGene technology and was initially reported in a summary table (29). Briefly, a bacterial artificial chromosome (BAC)-based targeting vector was derived using short regions of homology flanking exons 2 and 5 (termed 5′ and 3′ homology boxes, respectively) (Fig. 1A). These homology boxes were ligated to a reporter/selection cassette encoding transmembrane β-Gal (TM-LacZ) and a loxP-flanked neomycin resistance selection gene. Homologous recombination in bacteria through these homology boxes was used to replace the intervening sequences in a BAC containing the complete LYVE-1 gene. The replacement leads to the deletion of exons 2 to 5 and intervening introns and is designed, after splicing, to result in the fusion of the first 32 amino acids of LYVE-1 (containing the signal peptide) to TM-LacZ. The BAC employed in the construction was clone 343p22 (Incyte Genomics), which is approximately 150 kb in size. The cassette replacement results in a targeting vector with homology arms of 95 kb and 7 kb and a deletion of approximately 7 kb (Fig. 1A). This BAC-based targeting vector was used as described previously (29) to target the LYVE-1 locus in F1H4 hybrid embryonic stem (ES) cells. Several TaqMan probes, termed “loss-of-allele” probes, were designed to the deleted region and were used to establish that proper targeting had occurred in ES cells and subsequently in mice as described previously (29) (Fig. 1A). Two independent ES cell clones were derived and tested, and both clones behaved identically in their abilities to generate chimeric F₁ and F₂ mice. Knockout mice were backcrossed one additional generation onto the C57BL/6 background, resulting in an approximate 75% contribution of the C57BL/6 and 25% of the 129 backgrounds in the resulting N₂F₂ mice for studies reported here. Subsequent backcrossing onto the C57BL/6 background for more than five generations has produced no apparent change in phenotype. All procedures were carried out according to relevant U.S. federal or United Kingdom Home Office guidelines and institutional procedures as appropriate.

Whole-mount histochemistry and immunohistology.

Whole-mount staining for β-Gal in LYVE-1^+/− and LYVE-1^−/− mice and tumor cryostat sections was performed as described previously (2). For whole-mount preparations of dermal lymphatics, ear tissues were fixed overnight in 4% paraformaldehyde in phosphate-buffered saline (PBS), and the dorsal segment was blocked for 3 h in 30 mg/ml milk powder-10% fetal bovine serum (FBS) in PBS-0.1% (wt/vol) Triton X-100 prior to double staining overnight with rat anti-mouse CD31 (MEC 13.3)/Alexa594-conjugated goat anti-rat and hamster anti-mouse podoplanin (8.1.1)/Alexa488-conjugated goat anti-hamster antibodies. Preparation of intestinal whole mounts and immunostaining for lymphatics was performed essentially as described in reference 5 by use of antisera generated against mouse LYVE-1-human immunoglobulin G1 (IgG1)-Fc (1/10,000 dilution) or a directly biotinylated goat anti-mouse VEGFR-3 antiserum (R&D Systems). Fluorescent antibody-stained sections were viewed with a Bio-Rad Radiance 2000 laser scanning confocal microscope equipped with argon and green helium/neon lasers and analyzed using LaserSharp2000 software. Images were recorded in simultaneous scanning mode with lambda strobing at 4-μm intervals over a depth of 24 μm using a focusing motor and subsequently merged.

Immunohistology of cryostat sections.

Tissues were embedded in OCT mounting medium (Tissue-Tek) and frozen in liquid N₂ or in dry ice-ethanol and cut into 10-μm sections for fixation in acetone. Sections were stained with hamster anti-mouse podoplanin (clone 8.1.1; Developmental Studies Hybridoma Bank, IA), goat anti-mouse VEGFR-3 (R&D Systems), rat anti-peripheral lymph node addressin (PNAd), clone MECA79 (Becton Dickinson), or biotinylated hyaluronan binding protein (bHABP) (Seikagaku). After being washed, slides were incubated with the appropriate secondary antibodies: Alexa594-conjugated goat anti-hamster, Alexa594-conjugated donkey anti-goat, Alexa488-conjugated streptavidin (all from Molecular Probes), or fluorescein isothiocyanate (FITC)-conjugated goat anti-rat IgM (Southern Biotech). Then, viewing was performed with a Zeiss Axiophot microscope with epifluorescent illumination.

Electron microscopy.

Samples of formaldehyde-fixed skin and intestine were embedded in Spurr's epoxy resin, and thin sections were coated on Formvar-coated gold grids prior to staining with uranyl acetate and lead citrate. Preparations were examined with a JEOL 1200EX electron microscope.

FITC-dextran and Evans blue dye lymphangiography.

Fluorescence microlymphangiography of the tail was carried out essentially as described in reference 27. Mice were anesthetized with ketamine (80 μg/g body weight) and medetomidine (Domitol; 1 μg/g), placed on a heating pad, and injected with a solution of 10 mg/ml FITC-labeled dextran (2,000 kDa; Sigma) in PBS administered over a period of 20 min through a 29-gauge 1/2-in. needle inserted intradermally in the tail tip. Dye uptake was monitored continuously using a Zeiss inverted epifluorescence microscope.

For Evans blue dye lymphangiography, 40 μl of 1% (wt/vol) Evans blue was injected into hind-limb footpads of mice that were first anesthetized with isoflurane. After 20 min, mice were perfused with PBS and then photographed.

Measurement of tissue hyaluronan levels.

Levels of hyaluronan extracted from mouse tissues were estimated using a competitive enzyme-linked immunosorbent assay (ELISA). For the extraction step, tissue samples were digested with papain (250 μg/ml) in 5 mM cysteine-5 mM EDTA, pH 7.5, at 60°C for 24 h, followed by inactivation of the enzyme (100°C, 10 min) and centrifugation (900 × g, 5 min) to remove particulate matter. For the ELISA, 96-well Nunc Maxisorp plates were coated overnight (4°C) with 25 μg/ml rooster comb hyaluronan (Sigma) in coating buffer (15 mM sodium carbonate and 34 mM sodium bicarbonate, pH 9.4) followed by blocking in bovine serum albumin (10 mg/ml in PBS, pH 7.5) for 1 h at 37°C. Tissue samples (50 μl) or hyaluronan standards (12 to 1,600 ng/ml) were then mixed with biotinylated bHABP (50 μl, 1 mg/ml; Seikagaku) and incubated for 3 h at 37°C prior to washing and detection with peroxidase-conjugated streptavidin (1:500 DAKO) followed by O-phenylenediamine substrate (Sigma) and measurement of absorbance at 490 nm in a Bio-Rad plate reader.

Determination of serum Ig levels.

Levels of serum Ig classes were estimated by antigen capture ELISA. Briefly, ELISA plates (Maxisorp; Nunc) were coated overnight (2 μg/ml) with either rat anti-mouse IgA, rat anti-mouse IgG1, rat anti-mouse IgG2a, rat anti-mouse IgG2b, rat anti-mouse IgG3 (all from BD Pharmingen), rat anti-mouse IgM, or rat anti-mouse IgE (Southern Biotechnology) followed by rinsing and blocking (1 h) with PBS-0.1% Triton X-100 + 1% bovine serum albumin. Appropriately diluted sera were applied at room temperature for 1 h prior to detection with Ig-specific biotinylated secondary antibodies (0.25 μg/ml; Pharmingen) and poly-horseradish peroxidase-streptavidin (Endogen)/tetramethylbenzidine substrate (Pharmingen). Values were determined from measurements of absorption at 430 nm in a Bio-Rad ELISA plate reader.

Analysis of leukocyte populations by flow cytometry.

For quantitative estimation of tissue leukocyte populations, lymphoid organs were dissected, mechanically disrupted, and put through a 70-μm nylon mesh prior to staining (30 min on ice) with FITC-conjugated CD4 (GK1.5 clone), FITC-conjugated CD3, allophycocyanin-conjugated CD19, phycoerythrin (PE)-conjugated CD45R/B220-PE (clone RA3-6B2), NK1.1 (PK136) (all from BD Pharmingen), or allophycocyanin-conjugated rat anti-mouse CD8α (clones 5H10 and 53-6.7) as appropriate.

Granulocytes, NK cells, and monocytes/macrophages were assayed using rat anti-mouse Ly6G (Gr-1) or Ly6C-FITC (RB6-8C5 clone), FITC-conjugated rat anti-mouse CD49b or DX5-PE (DX5 clone), and rat anti-mouse CD11b-APC (M1/70 clone) or F4/80 (BM8), respectively. Cells were analyzed on a Becton Dickinson FACSCalibur instrument, and a total of 10⁶ cells per sample was collected.

Analysis of dendritic cell trafficking by FITC skin painting.

FITC skin painting was performed essentially as described previously (14). FITC (2.25 mg dissolved in 150 μl acetone dibutylphthalate) was applied to the shaved abdominal skin of wild-type, LYVE-1^−/−, and LYVE-1^+/− littermates. The following day, axillary and inguinal lymph nodes were dissected, the two lymph nodes on the left and right side were pooled, digested with collagenase D (0.5 mg/ml 30 min at 37°C), put through a 70-μm nylon mesh, stained with PE-labeled hamster anti-mouse CD11c (HL3) (BD Pharmingen), and analyzed by flow cytometry to quantitate dendritic cells dually labeled with FITC and CD11c.

Oxazolone-mediated contact hypersensitivity response.

Contact hypersensitivity was induced in appropriate mice by initial sensitization with 100 μl 3% (wt/vol) oxazolone solution (Sigma) in 4:1 acetone-olive oil applied to the shaved abdominal skin followed by sequential elicitation 5 and 12 days later by application of 10 μl 1% (wt/vol) oxazolone in acetone-olive oil on both dorsal and ventral aspects of the left ear. The control right ear was treated with acetone:olive oil only. Ear swelling was measured with a Mitutoyo loop handle dial thickness gauge (McMaster-Carr) immediately before and at various times after elicitation.

Growth of subcutaneous tumors.

B16F10 melanoma cells (kindly donated by Katja Simon, MRC Human Immunology Unit, Oxford) were resuspended in PBS, pH 7.5, and injected (0.5 × 10⁶ in 200 μl) subcutaneously into wild-type (n = 9) or LYVE-1^−/− (n = 11) mice. Tumor diameter was measured over a period of 5 weeks.

Lewis lung carcinoma cells (LL/2; American Type Culture Collection) were infected with retrovirus encoding VEGF-C in front of a green fluorescent protein-internal ribosome entry site construct or with green fluorescent protein-internal ribosome entry site constructs alone for controls. Infected cell populations were subjected to fluorescence-activated cell sorting twice on a DakoCytomation MoFlo high-speed cell sorter (Fort Collins, CO) prior to expansion and subcutaneous implantation in appropriate mice (11 to 12 weeks of age). After 14 days, tumors were harvested and processed for histological analysis. Tumor sizes were recorded after sacrifice by use of calipers, and tumor volumes were calculated using the formula length × width × height. Tumors were cut into 80-μm sections, stained with X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) reaction, and photographed with a Zeiss Axiophot microscope.

Generation of LYVE-1^−/− mice.

To generate mice lacking LYVE-1, we constructed a targeting vector that replaced the region from codon 33 (in exon 2) through exon 5 with a TM-LacZ reporter gene (Fig. 1A). The targeted gene thus encoded the N-terminal leader and the first eight amino acids from the mature LYVE-1 protein fused to a transmembrane version encoded by the TM-LacZ gene; when spliced, this gene encodes a β-Gal protein which is targeted to the cell membrane. Mice carrying the mutated allele were screened by quantitative genomic PCR using primers and probes as described for the wild-type or mutant allele ES cell screening in Materials and Methods (Fig. 1A). Mice heterozygous or homozygous for the mutated LYVE-1 allele were born at normal Mendelian ratios and displayed no obvious physical or behavioral abnormalities.

To confirm that expression of the LacZ gene was driven by the LYVE-1 promoter, we stained heterozygous LYVE-1^+/− mouse tissues for β-Gal activity and compared this with the pattern obtained using LYVE-1 antibody. As shown in Fig. 1B to E, both the β-Gal and the LYVE-1 antibody revealed the same extensive network of vessels characteristic of lymphatics on the surface of the intestine and associated Peyer's patches. Finally, the total absence of LYVE-1 antigen from LYVE-1^−/− mice was confirmed for whole-mount tissue sections by immunohistochemistry with cognate antibody (for example, see results for intestine shown in Fig. 1F).

Normal development, ultrastructure, and fluid drainage function of lymphatics in LYVE-1^−/− mice.

To investigate whether LYVE-1 deficiency affects the growth or patterning of the lymphatic vessels, we stained whole-mount sections of intestine with the lymphatic marker VEGFR-3. As shown in Fig. 2A and B, the lymphatic vessels formed similar networks of mostly small and often blind-ended capillaries in both wild-type and LYVE-1 gene-targeted mice. Similar results were found for dermal lymphatics within whole-mount sections of the ear stained with the lymphatic marker podoplanin and the panendothelial marker CD31, in which the morphology of the mostly small and often blind-ended capillaries (diameter, 50 to 100 μm) could be clearly seen (Fig. 2C and D). Moreover, the complex architecture of lymph node cortical and medullary sinuses—conduits for circulating lymphocytes and antigen-presenting cells—was also found to be conserved in LYVE-1 gene-targeted mice, as visualized by immunofluorescence staining for the lymphatic marker VEGFR-3 (Fig. 2F and G). Finally, transmission electron microscopy of initial lymphatic vessels in intestinal mucosa and skin (see Fig. S1 in the supplemental material) indicated a formation of overlapping interendothelial junctions in LYVE-1^−/− mice similar to that in LYVE-1^+/+ mice. Consistent with this apparently normal patterning of the lymphatic vasculature in LYVE-1 gene-targeted mice, such animals showed no obvious signs of lymphedema.

In order to uncover more-subtle defects induced by gene disruption, we visualized the uptake and drainage of the lymphatic tracking dye Evans blue after direct injection into the footpad. As shown in Fig. 3A and B, the dye was distributed equally well between lymphatic vessels and draining lymph nodes in wild-type and LYVE-1 gene-targeted mice. To corroborate these findings using a second, independent method, we monitored the uptake and clearance of the fluorescent tracker dye FITC-dextran by the dense lymphatic network present in the tail skin. To allow continuous monitoring of uptake, the dye solution was injected at a constant pressure of 45 cm H₂O for 20 min, and images were recorded using a fluorescence microscope. As shown in Fig. 3C and D, the FITC-dextran complex permeated the characteristic honeycomb network of dermal lymph vessels to similar depths (up to 20 μm) and at similar rates in both wild-type and LYVE-1^−/− mice. Overall, these results indicate that LYVE-1 integrity is not required for normal lymphatic fluid drainage function.

Hyaluronan homeostasis.

Aside from its role in tissue fluid homeostasis, the lymphatic system plays a significant role in the uptake and turnover of extracellular matrix components. Indeed, hyaluronan, the primary ligand for LYVE-1, is degraded in lymph nodes after release from its proteoglycan binding partners in the interstitial tissue matrix (4). Furthermore, LYVE-1 itself can function in vitro as an endocytic receptor for hyaluronan and is expressed abundantly in vivo within lymph node sinus endothelium (22). We therefore investigated the effect of LYVE-1 gene deletion on hyaluronan homeostasis by measuring the steady-state levels of the glycosaminoglycan in different tissues. The results (Fig. 4A) showed no significant difference in the concentrations of hyaluronan in skin, muscle, intestine, and lung between the two groups of animals, and only trace levels within liver. This was also confirmed by probing tissue sections from wild-type and knockout mice with a biotinylated bHABP, which revealed similar distributions of hyaluronan within, e.g., the dermis of the ear (Fig. 4B and C) as well as in liver, lymph node, and intestine (not shown). Finally, a comparison of the serum hyaluronan levels of normal (0.783 ± 0.081 μg/ml; n = 12) and knockout (0.762 ± 0.048 μg/ml; n = 12) mice indicated no difference due to LYVE-1 gene deletion.

Leukocyte populations are unaltered in LYVE-1^−/− mice.

Given the abundant expression of LYVE-1 in normal lymph node sinus endothelium and the role of secondary lymphoid tissue in regulating resident and circulating leukocyte populations, we considered the possibility that LYVE-1 deficiency might have an impact on leukocyte development and/or compartmentalization. Hence, we carried out a comprehensive analysis of the number and composition of major leukocyte populations, including the CD4⁺ and CD8⁺ (helper and cytotoxic) T-cell subsets, CD19⁺ B cells, NK1.1/Ly49b⁺ NK cells, Ly6G/Gr1⁺ neutrophils, and F4/80/CD11b⁺ monocytes/macrophages in peripheral blood, spleen, thymus, and lymph nodes by use of flow cytometry. The results (Fig. 5A and B; also, see Fig. S2 in the supplemental material) revealed no significant differences between any of these leukocyte populations within either wild-type mice or homozygous/heterozygous LYVE-1 gene-targeted mice. In addition, ELISA determination of individual serum immunoglobulin classes (IgG, IgM, and IgA) and subtypes (IgG1, IgG2a, IgG2b, and IgG3) indicated no significant differences between wild-type and knockout mice (see Fig. S3 in the supplemental material). Together, these results indicate that LYVE-1 function is not required for the development or compartmentalization of immune cells within primary or secondary lymphoid tissue.

Migration of dendritic cells in afferent lymph.

Based on homology with the leukocyte-homing receptor CD44, which supports rolling on blood vascular endothelium, we had speculated that LYVE-1 might mediate the adhesion or transmigration of cells entering the lymphatic vasculature (7, 10). A prominent example of such trafficking is the migration of antigen-presenting cells from tissue to draining lymph nodes during normal immune surveillance. To compare the process in normal mice with that in knockout mice, we measured the trafficking of epidermal dendritic cells to lymph node via afferent lymphatics using the technique of FITC skin painting. Dissection of the draining lymph nodes 18 h after dye application, followed by staining with the dendritic cell marker CD11c and flow cytometry, revealed that both the total CD11c-positive dendritic cell populations and the numbers of skin-derived CD11c/FITC double-positive cells were comparable in wild-type mice and both heterozygous LYVE-1^+/− and homozygous LYVE-1^−/− mice (Fig. 6A). Hence we conclude that LYVE-1 is not required for either entry or migration of dendritic cells through the afferent lymphatics.

Development and resolution of skin inflammation.

Each of the parameters described above was used to evaluate the role of LYVE-1 in normal tissue homeostasis. To determine whether LYVE-1 function is manifest under conditions of tissue stress, we subjected both wild-type and LYVE-1^−/− mice to allergen-induced contact hypersensitivity, a condition that promotes T-cell-mediated skin inflammation characterized by local tissue edema and the mobilization of epidermal Langerhans cell migration to draining lymph nodes via afferent lymphatics. To provoke inflammation, mice were sensitized by the consecutive topical application of oxazolone on the shaved abdominal skin and challenged 5 days later by the application of oxazolone on the ear. Measurement of edema as determined by ear thickness and microscopic examination of ear tissue after treatment showed no difference in the responses of the two groups of mice in either the magnitude or the rate of resolution of inflammation (Fig. 6B). These results suggest that the presence of LYVE-1 on lymphatic endothelial cells is required neither for T-cell-mediated hypersensitivity nor for clearance of the attendant inflammation/edema.

Growth and metastasis of subcutaneous tumors.

Finally, we considered the requirement for LYVE-1 integrity on tumor growth. The development of a lymphatic supply by either cooption or lymphangiogenesis is known to influence tumor fluid drainage and, consequently, tumor hydrostatic pressure. Hence, any defects in lymphatic vessel formation might be expected to limit tumor growth. To investigate these phenomena, we transplanted syngeneic (C57BL/6) B16F10 melanoma cells subcutaneously into wild-type and LYVE-1^−/− mice and monitored tumor diameter over a 5-week period. The results showed there was no obvious difference in tumor growth rates between the two groups of animals (Fig. 7A). We also investigated growth and lymph vessel development in a Lewis lung carcinoma that had been engineered to overexpress the lymphangiogenic growth factor VEGF-C by retroviral gene transfer. Again, these tumors showed similar growth rates in both wild-type and LYVE-1 gene-targeted mice (Fig. 7B), and there were no obvious differences in either the number of lymphatics, their patterns, or the depths of their penetration into tumor tissue, as assessed by β-Gal histochemistry (Fig. 7C). Together, these results indicate that LYVE-1 is not obligatory for either the growth or the apparent function of tumor-associated lymphatics.