CRSBP-1/LYVE-1-null Mice Exhibit Identifiable Morphological and Functional Alterations of Lymphatic Capillary Vessels

Abstract

CRSBP-1, a membrane glycoprotein, can mediate cell-surface retention of secreted growth factors containing CRS motifs such as PDGF-BB. CRSBP-1 has recently been found to be identical to LYVE-1, a specific marker for lymphatic capillary endothelial cells. The in vivo role of CRSBP-1/LYVE-1 is unknown. CRSBP-1-null mice are overtly normal and fertile but exhibit identifiable morphological and functional alterations of lymphatic capillary vessels in certain tissues, marked by the constitutively increased interstitial-lymphatic flow and lack of typical irregularly-shaped lumens. The CRSBP-1 ligands PDGF-BB and HA enhance interstitial-lymphatic flow in wild-type mice but not in CRSBP-1-null animals.

Introduction

Autocrine and paracrine regulation of cell growth are important mechanisms by which growth factors stimulate growth and migration of targeted cells in angiogenesis, organogenesis, embryonic development and tissue remodeling in response to injury or aging (1). In autocrine regulation, cell-surface retention provides an efficient mechanism for a secreted growth factor that is often secreted in relatively small amounts to interact with its receptor at the cell surface (1–7). Without cell-surface retention, a larger amount of growth factor would need to be secreted in order to accumulate in the extracellular compartment or the medium to reach a concentration that allows high-affinity binding to its receptor. In paracrine regulation, cell-surface retention orchestrates release of growth factor in a coordinated and directional manner to recruit target cells (via chemotaxis) to spatially orient cell growth and differentiation (1–7). All members of the PDGF superfamily, including platelet-derived growth factor-AA (PDGF-AA), PDGF-BB, placenta growth factor and vascular endothelial cell growth factor-A/C/D (VEGF-A/C/D), possess cell-surface retention sequence (CRS) motifs and exhibit cell-surface retention during secretion (3–6,8–10). These motifs contain clustered basic amino acid residues (Arg, Lys and His) and are evolutionarily conserved.

CRS binding protein-1 (CRSBP-1) is the only protein known to interact with peptides containing the CRS motifs of PDGF-BB and VEGF-A with high affinity (Kd = ~ 1 nM) (11). It has been shown to mediate cell-surface retention of the simian sarcoma virus (SSV) oncogene v-sis gene product (PDGF-BB) and insulin-like growth factor-binding protein-3 (a CRS-containing growth regulator) in SSV-transformed fibroblasts and H1299 cells co-expressing CRSBP-1 and IGFBP-3 (12). CRSBP-1 mediates ligand-dependent internalization and recycling at the cell surface, allowing efficient interaction of growth factor and receptor in cells expressing these three molecules (11–13). CRSBP-1 is a 120-kDa type I membrane disulfide-linked dimeric protein (11,12). Bovine CRSBP-1 cDNA encodes a 322-amino acid residue protein (12). The deduced amino acid sequences of human and murine CRSBP-1 cDNAs exhibit 61 and 56% identity with bovine CRSBP-1 cDNA, respectively (12). These two sequences were identified independently and named LYVE-1 (lymphatic vessel endothelial HA receptor-1) (14,15). LYVE-1, which was cloned using CD44 sequence homology cloning, is a hyaluronic acid (HA)-binding membrane glycoprotein (14,15). We found that expressed CRSBP-1 possesses distinct dual ligand (CRS motif and HA) binding activity and that CRSBP-1 overexpression enhances autocrine regulation of cell growth stimulated by the v-sis gene product and other growth regulators containing CRS (12).

The biological properties of CRSBP-1 have been studied mainly using cultured fibroblasts (11–15), but its role in intact animals has not yet been fully defined. The known ligands of CRSBP-1, including PDGF-BB and the VEGF family, are known to be involved in the development of the vascular and lymphatic systems (16,17). PDGF-BB is known to be a growth factor for either vascular or lymphatic capillary endothelial cells. The finding that CRSBP-1 and LYVE-1 are identical suggests a potential role of CRSBP-1 in the function of the lymphatic capillary vessels (12,14,16,17). LYVE-1 is localized to the luminal and abluminal faces of lymphatic vessels and has been used as a specific marker for lymphatic endothelium(14,15,18). However, the functional role of LYVE-1 in lymphatic vessels remains unknown. In this communication we demonstrate that CRSBP-1/LYVE-1 may mediate ligand-sensitive control of interstitial-lymphatic flow and its genetic ablation results in loss of such control and constitutively increased interstitial-lymphatic flow.

Materials and Methods

Construction of the Targeting Vector

The targeting vector (19) was constructed to remove parts of exon II, intron II, exon III and part of intron III of the CRSBP-1 gene and also to create a frame shift in the gene product, which contained 3 kb and 4.5 kb of 5' and 3' homology of the CRSBP-1 gene. The 5' 3-kb fragment, which covered parts of Exon I, Intron I, and a part of Exon II, was obtained by using an expanded high fidelity PCR kit from Roche with primer 1 (5'-AGGCACCCAGTCCAAGGTGCCGACCTC-3') in Exon I and primer 2 (5'-CACAAGGGCAACGCCCATGATTCTGCATG-3') in Exon II and SVJ 129 mouse genomic DNA as template. The PCR conditions were: 94°C 1 min; 50°C 1 min; 68°C 10 min for 35 cycles and extension at 72°C for 8 min. The PCR product of ~ 3 kb was cloned into PCR-XL Topo vector (Invitrogen). The Eco-RI fragment (~ 3kb) was ligated to the Eco-RI site of pPNT-Cass Lox A vector (9.3 kb). The sense orientation was determined by DNA sequencing. The 3' 7 kb fragment was obtained by using primer 3 (5' CAAAGCCTATTGCCACAACTCATCCG-3') in Exon III and primer 4 (5' GCCTCGTGTGCACCTTCTCCACTCTC-3') in Exon VI and cut with Xho and Not I. This 7 kb fragment was cloned into PCR-XL Topo vector. A 4.5 kb XhoI-Not I fragment was ligated to XhoI/NotI digested sites of the pPNT-Cass Lox A vector which already contained the 5' 3 kb fragment.

Generation of CRSBP-1-null (CRSBP-1^−/−) mice

The targeting vector (25 µg) was linearized with XhoI digestion and electroporated into the 129S1/SV-xP mouse-derived embryonic stem (ES) cells. The electroporation and generation of chimeric mice was done by the Yale University Gene Targeting Service. Heterozygote mice (a total of 12) were mated to produce homozygous knockout (CRSBP-1^−/−) mice. CRSBP-1^−/− mice were backcrossed for 10 generations onto the C57/BL/6 genetic background. Identical results were obtained on the mixed and inbred background. Genotyping was performed by PCR and Southern blot analysis of DNA obtained by tail biopsies at 10 days. Mice were genotyped by PCR using the upstream primer (in intron I near exon 2) with the downstream primer (in exon 3) for the wild-type allele and with the downstream primer (in intron III) for the disrupted allele, respectively. Mice were also genotyped by Southern blot analysis using a 200 bp sequence outside of the targeting vector probe. Tail DNA of wild-type and mutant mice were digested with Eco RI, transferred to membranes (Stratagene) and subjected to Southern blot analysis according to standard procedures. CRSBP-1-null mice (CRSBP-1^−/−) have been included in the Induecd Mutant Resources at the Jackson Laboratory, Bar Harbor, ME 04609 (Stock # 006221,B6. 129S1-X1kd1<tmJhua>/J)

Southern blot analysis of genomic DNA from wild-type (CRSBP-1^+/+), heterozygote (CRSBP-1^+/−) and homozygote (CRSBP-1^−/−) mice

EcoRI-digested genomic DNA isolated from tail tissues of wild-type (+/+), heterozygous (+/−), and homozygous (−/−) mice were blotted and probed with the ³²P-labeled PCR fragment (200 bp). A 5.4 kb restriction fragment was evident from the wild-type allele and a 3.9 kb fragment from the disrupted allele.

Germline transmission of the CRSBP-1^−/− mutation

The gene targeting event was identified by Southern blot analysis with EcoRI digested genomic DNA and the probe shown in Fig. 4A. Germline transmission was confirmed by Southern blot analysis of tail DNA of the F1 offspring from cross-breeding of chimera with wild-type C57BL/6 mice. Genotypes of wild-type (+/+) and homozygous (−/−) mice were identified by the presence of PCR product specific to either the wild-type or mutant allele.

Fig. 4. Morphological alterations of lymphatic vessels in the intestine of CRSBP-1^−/− mice.

Immunohistochemical stains of formalin-fixed, paraffin-embedded intestine tissue sections from wild-type (A) and CRSBP-1^−/− (B,C) mice, which were performed using antiserum to CRSBP-1 (A,B) or podoplanin C) and HRP + DAB (magnification, 200×). The lymphatic vessels in the intestine (duodenum) of wild-type (CRSBP-1^+/+) mice are positively stained (A). In the intestine of CRSBP-1^−/− mice, the lymphatic vessels in similar locations are identified by morphological analysis (B) and stained by antibodies to podoplanin (C). * indicates the location of lymphatic vessels. Lymphatic vessels are, in general, larger and have a distended shape in CRSBP-1^−/− mice (B,C). By contrast, they have various sizes and characteristic irregular shapes in wild-type mice (A). The arrow indicates the small-sized or irregularly-shaped vessels (A).

RNA preparation and RT-PCR

Wild-type and CRSBP-1^−/− mice were euthanized by CO₂ inhalation and liver tissues were removed immediately and frozen in liquid nitrogen. The liver RNA was extracted using RNAzole B (Tel-Test, Triendswood, TX). Briefly, 2 ml of RNAzole B solutions was added to 1 g of liver tissue and homogenized with a Polytron homogenizer; RNA was extracted with chloroform and precipitated with isopropyl alcohol. An aliquot of RNA was run on formaldahyde gel to check the integrity of the RNA. RT-PCR was performed with 2 µg of total RNA from wild-type and knockout mice using the RT-PCR kit with platinum Taq from Life Technology. The primers used to amplify the wild-type RNA were primers P5 (5'-AGAACTCTCCATCCAGCTTGGTG-3') and P8 (5'-GACACCTTTGCCATTGTTCCCACACC-3'). The primers used to amplify Exon I and Exon IV of the knockout mice RNA were primer P5 and primer P9 (5'-AACACGGGGTAAAATGTGGTAACG-3'). Primers specific for β-actin were used in control reactions.

Semi-quantitative Analysis of interstitial-lymphatic flow

To analyze the interstitial-lymphatic flow in the tissues of wild-type and CRSBP-1^−/− mice, 20 µl of 8 mg/ml fluorescein isothiocynate (FITC)-dextran (average M.W. ~ 2,000,000; Sigma) with and without PDGF-BB (1.25 µg/ml), HA (100 µg/ml) or both in phosphate-buffered saline (PBS) was injected intradermally by hand into the tails of mice (5 female mice for each group). The progressive diffusion of high molecular-weight FITC-dextran (which is taken up by lymphatic capillary vessels in the tissue near the injection site) was examined over time by monitoring fluorescence near the injection site with a UV light. The fluorescence intensity was determined by densitometry following photography. The relative fluorescence intensity in mice injected with PBS was taken as 100% (control). The interstitial-lymphatic flow analysis appeared to be semi-quantitative and specific. HA induced a decrease in the fluorescence near the injection site (10 min after injection) in a dose-dependent manner. HA at 10 and 100 µg/ml decreased the fluorescence intensity to 52±2 (n=5) and 33±10 (n=5), respectively. In the experiments to show the specificity of the analysis, both CRSBP-1^−/− and the wild-type mice exhibited similar diffusion rates of the low molecular-weight, hydrophobic and tissue-permeable fluorescent compounds fluorescein (M.W. 332) and hymecromone (M.W. 176) injected intradermally into the tails of these animal. These compounds diffused rapidly within the tissue. This suggests that increased egress of the flurorescent compound near the injection site is selective according to the M.W. of the agent chosen and consistent with a sieving process. Furthermore, PDGF-BB and HA each did not significantly affect the diffusion rates of these small M.W. compounds that were injected intradermally in wild-type mice

Immunohistochemistry

Tissues from wild-type and CRSBP-1^−/− mice were fixed in 10% neutral buffered formalin for 18 hr and then subjected to histologic processing. At the time of immunostaining, tissue sections were deparaffinized and hydrated. The sections were pretreated with H₂O₂ (0.3%) for 5 min and blocked with 10% goat serum plus 5% bovine serum albumin (BSA) in TBS (Tris-buffered saline) for 30 min. Sections were incubated with a rabbit serum against the C-terminal tail of bovine CRSBP-1 (12) (1:100 dilution) at room temperature for 30 min, rinsed with TBS, incubated with horseradish peroxidase (HRP)-labeled polymer (DAKO EnVision System) for 30 min, developed with diaminobenzidine (DAB), counterstained with hematoxylin, dehydrated sequentially with 70%, 95% and 100% ethanol and clear xylene, then mounted on coverslips.

For staining with antibodies to podoplanin, tissue sections were incubated with hamster anti-mouse podoplanin (AngioBio Co., Del Mar, CA). After washing, the sections were incubated with biotin-conjugated goat anti-hamster IgG (Southern Biotech, Birmingham, AL) (1:1000) for 30 min and washed 3 times with PBS. Sections were treated with HRP-streptavidin (1:1000) (Sigma) for 30 min. After washing 3 times with PBS, the sections were developed with DAB solution (1 mg/ml in 5 mM sodium phosphate buffer, pH 7.4) for 5 min.

Western blot analysis

Livers (0.7 g – 0.9 g) from wild-type and CRSBP-1^−/− mice were washed with buffer A (50 mM Tris-HCl, 1 mM EDTA, pH 7.4) twice to remove blood. The liver was homogenized in buffer A containing 1 mM phenylmethylsulfonyl fluoride (PMSF) (5 ml per gram tissue) with a tissue homogenizer. The homogenates were centrifuged at 2,000 rpm for 10 min to remove tissue debris and recentrifuged at 15,000 rpm for 30 min to obtain crude membranes. The crude membranes were washed with buffer A, solubilized in 1 ml buffer B (50 mM Tris-HCl, 3 mM EDTA, 1 mM PMSF and 1% Triton X-100) and centrifuged at 14,000 rpm for 10 min. The supernate was then rotated with wheat germ lectin (WHA)-agarose gel (volume = 40 µl) overnight. The WHA gel was washed three times with buffer B and eluted with 50 µl of buffer B containing 0.4 M N-acetylglucosamine. The eluents were subjected to 7.5% SDS polyacrylamide electrophoresis (SDS-PAGE), electrophoretic transfer to polyvinylidene fluoride (PVDF) membranes and immunoblot analysis with antiserum to CRSBP-1 (11).

Determination of the wet-to-dry weight ratios of tissues

Tissues (heart, liver, kidney and spleen) frozen and lyophilized with a Virtis lyophilizer (Freezemobile) for 48 hr. The dried tissues were then weighed (as dry weight). The wet-to-dry weight ratios of the tissues or tumors were determined.

Statistical Analysis

The values represented the mean±standard deviation (S.D.). Two-tailed unpaired Student's t test was used to determine the significance of differences between groups. p < 0.05 was considered significant.

Results

Generation of CRSBP-1^−/− mice

To investigate the role of CRSBP-1 in vivo, we determined the structure of the mouse CRSBP-1 gene and generated CRSBP-1^−/− mice. The entire genome structure of the mouse CRSBP-1 gene was elucidated based on mouse CRSBP-1 cDNA using the mouse genome sequence database. The CRSBP-1 gene resides on mouse chromosome 7. As shown in Fig. 1A, the mouse CRSBP-1 allele consists of 6 exons, which all contain coding sequences, and 5 introns. The signal sequence and the first four amino acid residues of the N-terminal polypeptide are encoded by exon 1. The transmembrane domain is encoded by exon 5 and the cytoplasmic domain is encoded by exon 6.

Fig. 1. Gene targeting of the murine CRSBP-1 locus.

The mouse CRSBP-1 allele (12.2 kb), which resides on chromosome 7, consists of 6 exons and 5 introns (A). A targeting vector (pPNT-Cass Lox A) was contsructed to delete part of exon 2, intron 2, exon 3 and part of intron 3 (B). The vector contains a cassette flanked by loxP elements, including sequences conferring neomycin resistance (under the direction of a Pol II promoter) and sequences encoding cre recombinase (Cre) [under the direction of a testis-specific angiotensin-converting enzyme (tACE) promoter]. Exons are symbolized as numbered rectangles. The recombinant allele in ES cells and the Neo-excised (disrupted) allele in knockout mice are shown in C and D, respectively. The black horizontal bar indicates the location of the probe used for Southern blotting. The expected sizes of the fragments for the wild-type (A) and disrupted (D) alleles after digestion with EcoR1 (R) are shown (5.4 and 3.9 kb, respectively).

To generate the CRSBP-1^−/− mutation in ES cells, we designed a replacement-type targeting vector containing a cassette flanked by loxP elements (18). The cassette includes sequences conferring neomycin resistance (Neo) as well as testes-specific angiotensin-converting enzyme promoter (tACE promoter) sequences that direct expression of cre recombinase (Cre) in the male germ line only (Fig. 1B). The targeting cassette vector was constructed to remove part of exon 2, intron 2, exon 3 and part of intron 3 and also to create a frame shift in the gene product (Fig. 1A and 1B). The targeting vector was linearized with XhoI digestion and electroporated into 129S1/Sv-p⁺ mouse-derived ES cells. After 24 hr, the cells were placed under positive/negative selection with 200 µg/ml G418 and 2 µM ganciclovir for 6 days. Clones resistant to this double selection were isolated and analyzed by Southern blot analysis using a probe for a sequence which is located in exon 1 but outside of the targeting vector (as indicated by the black horizontal bar in Fig. 1A) and a probe for a sequence which is located in exon 6 but outside of the targeting vector. Southern blot analysis was used to identify the homologous recombinant allele (Fig. 1C) ES line. Two independent, targeted ES clones were injected into C57BL/6J blastocysts and chimeric males were backcrossed for germline transmission to C57BL/6J females. The resulting Neo-excised allele (Fig. 1D) heterozygote mice (a total of 12) were mated to produce homozygous knockout (CRSBP-1^−/−) mice.

CRSBP-1^−/− mice were initially obtained on a mixed C57BL/6/129 background (n = 200) and backcrossed for 10 generations onto the C57BL/6 genetic background. Identical results were obtained on the mixed and inbred background. CRSBP-1^−/− mice appeared overtly normal and were fertile with normal litter sizes. No gross abnormalities (e.g., body weight, relative organ weight and wet-to-dry weight ratios of tissues, e.g., kidney, liver, lung, heart and spleen) were observed over their entire life span, with the oldest animal close to 2 years of age. The disruption of the CRSBP-1 gene was detected by allele-specific PCR (Fig. 2Aa). The wild-type specific band of 500 bp is only present in wild-type mice (Fig. 2Aa, lane 1) whereas the 400 bp band is specific for the disrupted allele (Fig. 2Aa, lane 2). The disruption of the CRSBP-1 gene was confirmed by Southern blot analysis, since the wild-type specific band at 5.4 kb is replaced by a band at 3.9 kb in CRSBP-1^−/− mice (Fig. 2Ab). The disruption of the CRSBP-1 gene was further confirmed by the absence of the wild-type specific RT-PCR band in CRSBP-1^−/− mice. As shown in Fig. 2B, the wild-type RT-PCR specific band at 419 bp (lane 1) is absent in CRSBP-1^−/− mice (lane 2). This suggests that exon 3 is deleted in CRSBP-1^−/− mice, as expected. In another set of RT-PCR experiments using primers P5 and P9, the products with the predicted sizes (268 bp and 534 bp) were detected in CRSBP-1^−/− and wild-type mice, respectively (data not shown). The absence of CRSBP-1 protein in CRSBP-1^−/− mice was proven by Western blot analysis with a polyclonal antibody directed against the cytoplasmic tail of CRSBP-1 (12). As shown in Fig. 2C, CRSBP-1 was detected in the liver of wild-type (CRSBP-1^+/+) mice (lane 1) whereas it was not detected in the liver of CRSBP-1^−/− mice (lane 2).

Fig. 2. Genotyping (A), RT-PCR (B) and Western blot (C) analyses of CRSBP-1 in wild-type and CRSBP-1^−/− mice.

(A) Genomic DNA was prepared from the tails of wild-type (CRSBP-1^+/+) and homozygous knockout (CRSBP-1^−/−) mice. The DNA was amplified by PCR with allele-specific primers to detect the wild-type CRSBP-1 allele (~ 500 bp) (a, lane 1) and the disrupted allele (~ 400 bp) (a, lane 2), respectively. Genomic DNA from the tails of wild-type (CRSBP-1^+/+) (b, lane 1), homozygote (CRSBP-1^−/−) (b, lane 2) and heterozygote (CRSBP-1^+/−) (b, lane 3) mice was digested with EcoRI and separated in a 0.8% agarose gel. Southern blot analysis was performed with a ³²P-labeled primer as indicated by the black horizontal bar (in Fig. 1A). The sizes of the band are shown (5.4 kb for the wild-type allele and 3.9 kb for the disrupted allele).

(B) Liver RNA from CRSBP-1^+/+ and CRSBP-1^−/− mice was isolated and subjected to RT-PCR using a sense primer in exon 1 and a reverse primer in exon 3. CRSBP-1^+/+ RNA produces a 419-bp product (lane 1) whereas CRSBP-1^−/− RNA does not yield any products (lane 2). In the control experiment, both RNAs yielded a typical product of β-actin using primers of β-actin. The arrow indicates the location of the 419-bp product.

(C) The Triton X-100 extracts of liver plasma membranes from CRSBP-1^+/+ (lane 1) and CRSBP-1^−/− (lane 2) mice were subjected to wheat germ lectin-Sepharose 4B affinity gel absorption as described (11). After extensive washing, the column was eluted with 0.4 M N-acetylglucosamine. The eluents were analyzed by 7.5% SDS-PAGE under reducing conditions followed by Western blot analysis using antiserum to CRSBP-1. The arrow indicates the location of the CRSBP-1 monomer (60 kDa).

Constitutively increased interstitial-lymphatic flow and lack of typical irregular-shaped lumens of lymphatic capillary vessels in CRSBP-1^−/− mice

Although CRSBP-1^−/− mice were overtly normal, in order to examine lymphatic capillary vessels, we performed comparative immunohistochemical analysis of tissues from age-matched wild-type, heterozygote (CRSBP-1^+/−) and homozygote (CRSBP-1^−/−) mice. Lymphatic vessels in wild-type and heterozygote mice were identified in all tissues examined using a specific antibody to CRSBP-1 (11,12). Lymphatic vessels in CRSBP-1^−/− mice were identified based on the morphology and specific location of lymphatic vessels in certain tissues (e.g., intestine and liver) and also by immunohistochemical analysis of all tissues examined using antibodies to podoplanin. Among tissues examined (liver, intestine, lung, ovary, uterus, vagina, cervix, salivary gland, skin, kidney, brain and foot pad), the liver and intestine exhibited differences in lymphatic morphology between CRSBP-1^−/− and wild-type mice most clearly. For tissues with abundant connective tissue (e.g., skin) and other tissues, no clear morphological differences were found between wild-type and CRSBP-1^−/− mice, perhaps due to limited distensibility of such tissues. However, the typical irregular shapes and/or smaller sizes of lymphatic vessels were detected only in a fraction of lymphatic vessels in the liver and intestine of wild-type and heterozygote mice. In the portal triad of the liver of CRSBP-1^−/− mice, lymphatic vessels had distended lumens marked by larger or expanded shapes and did not have collapsed irregular shapes and/or smaller sizes seen in wild-type mice (Fig. 3E/F versus 3A/B). The lymphatic vessels in the liver of CRSBP-1^−/− and wild-type mice were also identified by immunostaining using antibodies to CRSBP-1 and podoplanin (Fig. 3C/G and 3D/H, respectively). However, bile ducts in portal triads were also stained by antibody to podoplanin (Fig. 3D,H). In the intestine of CRSBP-1^−/− mice, lymphatic vessels had altered morphology, with distended lumens marked by extended or rounded shapes as compared to smaller or typically collapsed irregular shapes observed in wild-type mice (Fig. 4B and 4C versus 4A). In contrast to wild-type and heterozygote mice, no collapsed irregular shapes of lymphatic vessels were detected in the liver and intestine of any CRSBP-1^−/− mice examined (12 mice; female and male, 6–8 weeks old).

Fig. 3. Morphological alterations of lymphatic vessels in the liver of CRSBP-1^−/− mice.

Immunohistochemical stains of formalin-fixed, paraffin-embedded liver tissue sections from wild-type (CRSBP-1^+/+) (A/B and C/D from two different animals) and CRSBP-1^−/− mice (E/F and G/H from two different animals), were performed using antiserum to CRSBP-1 (A,B,C,E,F,G) or anti-podoplanin (D,H) and horseradish peroxidase + diaminobenzidine (DAB). In wild-type mice, the lymphatic vessels in the portal triad of the liver are positively stained with antiserum to CRSBP-1 (A,B,C; magnification, 100×) and with anti-podoplanin (D and H; magnification, 100×). In CRSBP-1^−/− mice, the lymphatic vessels in the portal triad are not stained but are identified by morphological analysis (E,F,G; magnification, 100×) and by immunostaining with anti-podoplanin (D and H; magnification 100×), lymphatic vessel: *. PV: portal vein; D: bile duct; A: artery. In CRSBP-1^−/− mice, the lymphatic vessels have distended lumen sizes and do not have the characteristic irregular shapes seen in wild-type mice (indicated by an arrow) (E,F,G,H versus A,B,C,D). Note: In the liver tissue sections shown above (A,B,C), the sinusoids were not stained using antiserum to CRSBP-1. However, moderate staining was observed in some sinusoids. The reason why other sinusoids were not stained remains to be elucidated.

The distended lumens and the lack of typical irregularly-shaped lumens of lymphatic vessels in CRSBP-1^−/− mice suggested that the lumens of lymphatic capillaries may be constitutively extended and permit rapid transit of macromolecules from the interstitium to lymphatic vessels in the whole animal system. To test this possibility, we intradermally injected FITC-dextran (average M.W. ~ 2,000,000) with and without PDGF-BB (1.25 µg/ml) and/or HA (100 µg/ml) into the tails (at a site ~ 2 cm from the tail tip) of CRSBP-1^−/− and wild-type (CRSBP-1^+/+) mice. We then compared the decrease in FITC-dextran fluorescence near the injection site over time in both groups. High molecular-weight and tissue-impermeable FITC-dextran enters lymphatic capillary vessels from the interstitial space after being injected into tissue and is a standard lymphatic imaging agent. As shown in Fig. 5A, FITC-dextran fluorescence at the injection site diminished more rapidly in CRSBP-1^−/− mice than in wild-type mice. By 10 min after injection, the fluorescence of FITC-dextran near the injection site was 30±5% in CRSBP-1^−/− mice as compared with wild-type mice (arbitrarily scored 100 %). In CRSBP-1^+/+ mice, PDGF-BB and HA caused diminished fluorescence near the injection site at 10 min, but PDGF-BB was more effective than HA at their optimal doses (Fig 5B and 5C). PDGF-BB decreased the fluorescence intensity to 20±5% (n=5) of control whereas HA decreased the fluorescence intensity to 40±10% (n=5) of control (Fig 5B). Administration of PDGF-BB and HA together did not decrease fluorescence as compared to controls (Fig. 5C). The mechanism for this unknown, but HA is known not to affect PDGF-BB binding to CRSBP-1 (12). However, PDGF-BB and HA, either alone or together, did not accelerate the egress of FITC dextran in CRSBP-1^−/− mice (Fig. 5D). These results suggest that CRSBP-1 may mediate ligand-sensitive control of interstitial-lymphatic flow and that loss of CRSBP-1 results in constitutively increased interstitial-lymphatic flow which is insensitive to CRSBP-1 ligand enhancement.

Fig. 5. Inability of CRSBP-1 ligands to enhance interstitial-lymphatic flow in CRSBP-1^−/− mice.

Twenty µl of FITC-dextran (8 mg/ml) were intradermally injected into the tails of wild-type (CRSBP-1^+/+) and CRSBP-1^−/− mice (A). CRSBP-1^+/+ and CRSBP-1^−/− mice were also intradermally injected with FITC-dextran±PDGF-BB (1.25 µg/ml) and/or HA (100 µg/ml) (B,C,D). After 10 min, photographs were taken in the presence of UV light. The arrows indicate the injection site of FITC-dextran. The bracket indicates the location of FITC-dextran fluorescence. The other bright areas are due to the light reflection of hair. The FITC-dextran fluorescence vanished more rapidly near the injection site in CRSBP-1^−/− mice than in CRSBP-1^+/+ mice (A). In CRSBP-1^+/+ mice, PDGF-BB and HA each (but not together) caused diminished fluorescence near the injection site. PDGF-BB was more effective than HA (B,C). In CRSBP-1^−/− mice, injection of PDGF-BB and HA (together or alone) was unable to diminish fluorescence (D).

Discussion

Under normal circumstances, lymphatic capillaries are generally collapsed and closed. An increase of the interstitial fluid pressure caused by an increased fluid volume in tissue is generally thought to stretch the fibers and expand the lymphatic lumen (20–22). This causes the intercellular junctions and blind ends of lymphatic capillary vessels to open, thus increasing interstitial-lymphatic flow (20). However, the exact molecular mechanisms involved in controlling the opening of lymph drainage and lymphatic flow in the interstitium remain unknown (21). Our finding of CRSBP-1 ligand-induced enhancement of interstitial-lymphatic flow in wild-type mice but not in null mice suggests that CRSBP-1 may mediate ligand-sensitive control of interstitial lymphatic flow and that its genetic ablation may result in loss of such control and constitutively increased interstitial-lymphatic flow. We propose a model (Fig. 6), which is modified from the published model (20), for the role of CRSBP-1 in controlling interstitial-lymphatic flow in wild-type mice and loss of such control in CRSBP-1^−/− mice. In lymphatic capillary vessels of wild-type mice, lymphatic endothelial cells have overlapping intercellular junctions that generate irregularly shaped lumens (partially closed or closed) (Fig. 6a). CRSBP-1 may be involved in the formation of the cellular extension and overlapping intercellular junctions. CRSBP-1 ligands PDGF-BB and HA each may cause contraction of lymphatic endothelial cells, resulting in opening of these junctions in the lymphatic capillaries, thus increasing interstitial-lymphatic flow (Fig. 6b). In CRSBP-1^−/− mice, lymphatic capillary endothelial cells appear to assume the contracted configuration and lose the ability to form cellular extension, pulling apart the endothelial cells, expanding the lumens of lymphatic capillary vessels, increasing interstitial-lymphatic flow and being insensitive to ligand-induced contraction (Fig. 6b).

Fig. 6. A model for CRSBP-1-mediated ligand-sensitive control of interstitial-lymphatic flow in wild-type mice and loss of such control in CRSBP-1^−/− mice.

Lymphatic capillary vessels, which have overlapping intercellular junctions that are formed by the extensive superimposition of adjacent lymphatic endothelial cells and generate collapsed irregularly-shaped lumens (closed or partially closed configuration), are firmly attached to the collagen and elastin fibers in adjacent tissues by anchoring filaments (20) (a). CRSBP-1, a membrane glycoprotein, may be involved in the formation of the cellular extension and overlapping intercellular junction of lymphatic endothelial cells. In wild-type mice, CRSBP-1 ligands induce contraction of lymphatic endothelial cells. In CRSBP-1-null mice, lymphatic endothelial cells assume the contracted configuration. These cause the intercellular junctions and blind ends of lymphatic capillary vessels to open, thus increasing interstitial-lymphatic flow (b). The thin arrows indicate the direction of cellular contraction in CRSBP-1^−/− mice or when CRSBP-1 ligands are present in CRSBP-1^+/+ mice (b).

CRSBP-1 ligands such as HA and PDGF-BB may function as mediators or sensors controlling opening of lymphatic capillary vessels. Since HA is known to be present in the interstitial space of tissues and in lymphatic fluid (23,24), we hypothesize that HA is a regulator of interstitial-lymphatic flow. CRSBP-1 is localized to both the luminal and abluminal faces of lymphatic capillary endothelial vessels. The difference between the HA concentrations in the interstitial space (HA_out) and in the lymphatic capillary lumen (HA_in) (23,24) may be an important factor controlling the opening of lymphatic capillary vessels. When HA_out is higher than HA_in, the lymphatic vessel dilates and becomes more porous as observed in lymphatic vessels of CRSBP-1^−/− mice (Fig. 6b). When the HA_out is lower than HA_in, the reverse occurs (Fig. 6a). PDGF-BB is believed to play a role in the control of tissue interstitial fluid pressure (IFP) (25). In animal models, PDGF-BB was shown to normalize dermal IFP, which had been lowered by inducing anaphylaxis or inhibiting the αIIβI integrin function (26). Treatment with a PDGF receptor kinase inhibitor or a PDGF-BB aptamer antagonist appeared to decrease interstitial hypertension and increase capillary-to-interstitium transport in implanted colonic carcinomas in rats (27). This led to the suggestion that PDGF-BB is involved in creating and/or sustaining a high tumoral IFP. However, the recent observation that blockade of PDGF β-type receptor activity leads to fluid accumulation or edema, implies the involvement of PDGF-BB in facilitating interstitial-lymphatic flow (28). Here we demonstrate that PDGF-BB can enhance interstitial-lymphatic flow in wild-type mice but not in CRSBP-1-null mice. This result suggests that CRSBP-1 may mediate PDGF-BB-enhanced interstitial-lymphatic flow. This result is also consistent with previous reports that mice deficient in PDGF-BB exhibit interstitial edema (29) and that blockage of PDGF-BB-induced signaling causes edema in human patients (28).

Two lines of evidence suggest that CRSBP-1 plays a role in regulating the function of lymphatic endothelium. These include: 1) CRSBP-1 ligands PDGF-BB and HA enhance interstitial-lymphatic flow in wild-type mice but not in CRSBP-1^−/− mice, and 2) PDGF-BB and VEGF-A peptides containing the CRS motifs of PDGF-BB and VEGF-A (11), like intact PDGF-BB and VEGF-A, enhance interstitial-lymphatic flow in wild-type mice (unpublished results). These peptides do not interact with the PDGF β-type and VEGF-A receptors, respectively (11). The importance of CRSBP-1 in mediating the ligand-sensitive control of interstitial-lymphatic flow is further supported by the observation of altered morphology of lymphatic capillary lumens in CRSBP-1^−/− mice. We observed distended lumens and the lack of typical irregularly-shaped lumens of lymphatic capillary vessels in certain tissues such as liver and intestine (in these null mice). This may be due to the fact that the lymphatic capillary vessels are present in the areas where connective tissues are not abundant (e.g., liver and intestine). The morphology of lymphatic capillary vessel lumens is strongly influenced by the surrounding connective tissues (which cause tissue compression) and staining and fixation procedures used in immunohistochemical staining. Even in the duodenum of wild-type mice, only a fraction of the lymphatic vessels exhibit typical irregular shapes of lumens. It is important to note that none of the lymphatic vessels examined in the liver and duodenum of CRSBP-1^−/− mice exhibited collapsed irregular shapes of lumens.

CRSBP-1 exhibits dual ligand (CRS motif and HA) binding activity. The CRS motif binding activity of CRSBP-1 is believed to be involved in cell-surface retention of growth factors in the PDGF superfamily and autocrine and paracrine growth stimulated by these growth factors (11–13). The physiological role of the CRS motif binding activity of CRSBP-1 in intact animals is unknown. CRSBP-1^−/− mice are grossly normal but exhibit identifiable microscopic morphological and functional alterations of lymphatic capillary vessels. This suggests that other proteins, such as membrane-bound proteoglycans (30,31) that can bind the CRS motifs can substitute for the function of the CRS motif-binding activity of CRSBP-1 to mediate autocrine and paracrine stimulation of cell growth during hemangiogenesis and lymphangiogenesis in the developing mouse, but no other known proteins can replace CRSBP-1 in mediating the interstitial-lymphatic flow-enhancing activity of PDGF-BB in CRSBP-1^−/− mice. The CRS motif-binding activity (or PDGF-BB binding activity) of CRSBP-1 appears to be involved in the control of lymphatic vessel opening. The CRSBP-1 ligands (e.g., PDGF-BB) may serve as mediators for distending the lumens of lymphatic capillary vessels and enhancing interstitial-lymphatic flow. Thus, the null mutation of CRSBP-1 only reveals a defect in the ligand-sensitive control, resulting in constitutively increased interstitial-lymphatic flow (Fig. 6b).