ngO is an albumin—binding glycoprotein expressed
`by continuous endothelium involved in albumin transcytosis
`
`JAN E. SCHNITZER
`Department of Medicine and Pathology, Division of Cellular and Molecular Medicine and Institute for
`Biomedical Engineering, University of California, San Diego, School of Medicine,
`La Jolla, California 92093-0651
`
`Schnitzer, Jan E. gpSO is an albumin-binding glycoprotein
`expressed by continuous endothelium involved in albumin
`transcytosis. Am. J. Physiol. 262 (Heart Circ. Physiol. 31):
`H246~H254, 1992.-——Albumin reduces capillary permeability
`and acts as a carrier for various small molecules. Recently, we
`identified a 60-kDa sialoglycoprotein (ngO) on the surface of
`cultured rat microvascular endothelial cells (MEG) that binds
`albumin and antiglycophorin serum (a—gp). We verified that 0(-
`gp recognizes the albumin-binding gp60 by affinity, purifying
`proteins from MEC extracts using immobilized albumin. ngO
`was immunoblotted with a-gp only when the MEC extract was
`reacted with albumin and not in controls. We immunoprecipi-
`tated ngO from biosynthetically radiolabeled MEC lysates and
`.from extracts containing endothelial surface proteins of iso-
`lated rat hearts that were radioiodinated in situ. ngO was
`immunoblotted selectively in rat tissue microvascular beds
`lined with continuous endothelium (heart, lung, diaphragm, fat,
`skeletal muscle, mesentery, and duodenal muscularis but not
`cortical brain) and not those exclusively lined with fenestrated
`or sinusoidal endothelium (adrenal, pancreas, liver, and small
`intestinal mucosa). MEC isolated from rat heart, lung, and
`epididymal fat pad expressed gp60 and bound albumin, whereas
`various nonendothelial cells and brain-derived MEC did not.
`gp60 is an albumin-binding glycoprotein expressed specifically
`on the surface of continuous endothelium that binds albumin
`apparently not only to initiate its transcytosis via plasmalem-
`mal vesicles but also to increase capillary permselectivity.
`rat; biological
`transport; capillary permeability; membrane
`receptors; membrane proteins; sialoglycoproteins; receptor-me-
`diated transport; vesicular transport; plasmalemmal vesicles;
`glycocalyx
`
`endothelium (10, 15, 19, 22), and its binding within
`transport pathways, such as plasmalemmal vesicles and
`the introit of intercellular junctions, apparently forms a
`molecular filter within these pathways (3) that can elec-
`trostatically (20) and sterically (3, 21) restrict the trans-
`port of water, small solutes, and macromolecules across
`the microvascular wall (6, 11—14, 19). In addition, albu-
`min is transcytosed across vascular endothelium via plas-
`malemmal vesicles (10, 15, 19) and acts as a carrier for
`small ligands bound to it, such as fatty acids (5). This
`apparent receptor-mediated transcytosis of albumin ap-
`pears to occur selectively in certain tissues with vascular
`beds lined with continuous endothelium (10, 28) and is
`influenced greatly by the ligands bound to albumin (5).
`The binding of albumin to cultured microvascular en-
`dothelium has been quantitated and immunolocalized to
`the surface of cultured rat microvascular endothelial
`monolayers (22). Specific albumin binding was shown to
`be saturable, reversible, compatible, dependent on cell
`type and cell number and to have a negative cooperativity
`in nature (22). Albumin binding was sensitive to pronase
`but not to trypsin digestion of the cell surface and was
`inhibited significantly by the presence of Limax flavus
`(LFA), Ricinus communis (RCA), and Triticum uulgare
`(wheat germ; WGA) agglutinins but not several other
`lectins (23). Recently, a group of rat endothelial plas-
`malemmal sialoglycoproteins has been identified both in
`situ and in culture (25). One of these sialoglycoproteins,
`called gp60, was identified as an albumin-binding protein
`because it 1) interacts with albumin conjugated to beads
`A PRIMARY FUNCTION of the attenuated layer of vascular
`endothelium is to act as a barrier to the transvascular
`(23); 2) binds RCA, LFA, and WGA but not'other lectins
`(23, 25, 26); and 3) is sensitive to pronase and sialidase
`transport of plasma molecules in many tissues. The
`but not to trypsin digestion (23, 25). Another laboratory
`selectivity of the endothelial barrier varies in different
`has also identified ngO as one of three major albumin-
`vascular beds and is strongly dependent on the structure
`binding proteins (8, 9). The structural and functional
`and type of endothelium lining the microvasculature
`relationship of these albumin-binding proteins to one
`(29). Several pathways exist for the transport of plasma
`another is unknown. Further characterization of gp60
`molecules across continuous endothelium: I) intercellu-
`showed that it apparently contains O—Iinked but not N-
`lar junctions are highly regulated structures that form
`linked glycans (25) and may also be antigenically related
`the paracellular pathway for the passive, pressure-driven
`to another sialoglycoprotein, namely glycophorin (26).
`filtration of water and small solutes; 2) plasmalemmal
`Polyclonal antiserum (a-gp)
`raised against purified
`vesicles transcytose plasma macromolecules, apparently
`by shuttling their contents adsorbed from blood from the
`mouse glycophorin gp3 recognized a 6Q—kDa WGA- and
`RCA-binding sialoglycoprotein on the sorfaée of cultured
`luminal to antiluminal aspect of the endothelium (16);
`rat microvascular endothelial cells (26). This apparent
`and 3) transendothelial channels, which may form tran-
`recognition of gp60 by a—gp was inhibited in the presence
`siently by the fusion of two or more plasmalemmal
`of murine glycophorins.
`.
`vesicles, may provide a direct conduit for the exchange
`In this study, the specific interaction of ngO with
`of both small and large plasma molecules (31). Recently,
`albumin and with a-gp serum is demonstrated. Then, a-
`it has become clear that capillary permeability in many
`vascular beds is also dependent on the interaction of
`gp is used as a probe ’for ngO. Because albumin binding
`these transendothelial transport pathways with the mul-
`and transcytosis Via plasmalemmal vesicles are not ob-
`served in the endothelium lining the vasculature of many
`tifunctional plasma protein, albumin (3, 6, 11—14, 19).
`Albumin binds to the luminal glycocalyx of continuous
`tissues, tissue-specific expression of ngO should be ex-
`H246
`0363—6135/92 $2.00 Copyright © 1992 the American Physiological Society
`
`Abraxis EX2013
`
`Cipla Ltd. v. Abraxis Bioscience, LLC
`
`IPR2018-00162; IPR2018-00163; IPR2018—00164
`
`Page 1 of 10
`
`

`

`ENDOTHELIAL ALBUMIN-BINDING GLYCOPROTEIN GFBO
`
`H247
`
`pected if gp60 does indeed function as a physiologically
`significant albumin-binding protein. Therefore various
`rat tissue and cell extracts were tested using a-gp for the
`presence of gp60. In addition, microvascular endothelial
`cells have been isolated from several organs and grown
`in culture to compare both gp60 expression and albumin
`surface binding. These experiments demonstrate I) the
`‘ specific expression of ngO only by those cultured endo-
`lhelial cells that bind albumin and not various other
`cells, 2) the presence of gp60 on the surface of vascular
`endothelium both in situ and in culture, and 3) the
`selective distribution of ngO in rat tissues with vascular
`beds lined with continuous endothelium that transcytose
`albumin via plasmalemmal vesicles.
`
`METHODS
`
`Isolation and growth of endothelial cells in culture. Male
`albino rats (Sprague-Dawley, 200—250 g) were anesthetized with
`ether. The‘ heart, lungs, and cortical brain were surgically
`removed from three to five rats, were submerged in cold Dul—
`becco’s modified Eagle’s medium (DMEM) supplemented with
`15—20% fetal calf serum (FCS)
`(DMEM+), and each was
`minced in a vial using small sterile surgical scissors. For the
`heart, only myocardium was used after careful excision of the
`epi- and endocardium. For the lung, only peripheral regions of
`the lobes were used. Both of these excision procedures are
`designed to eliminate obvious large arterial and venous seg—
`ments of the vasculature and to increase the probability for
`isolation of microvascular endothelium. After centrifugation
`for 5 min at 1,000 g at 4°C, the pellet was resuspended in 2 mg/
`ml of collagenase (from. Clostridium histolyticum,
`type II,
`Sigma) in DMEM+ at 37°C using five times the tissue volume.
`After a 1—h incubation, the suspension was poured over a sterile
`Nitex monoscreen cloth (no. 3—112—40xx; Tetko, Briar Cliff
`Manor, NY) to filter out tissue clumps. The filtrate was cen-
`trifuged for 5 min at 1,000 g, and the pellet was resuspended in
`DMEM+containing 2D rig/ml of heparin, 100 U/ml of penicil-
`lin G, 100 Mg/l’l’ll of streptomycin sulfate, and 10% bovine aortic
`endothelium-conditioned media. The cells were plated onto T~
`25 flasks (Corning, Corning, NY) for culture at 37°C with 5%
`C02 in air. After several days when confluency was reached,
`the cells were examined and sorted on a fluorescent-activated
`cell sorter using uptake of acetylated low—density lipoprotein
`labeled with 1,1’—dioctadecyl—3,3,3’,3'«tetramethylindocarbo-
`cyanine perchlorate (DiI-AcLDL) as in Refs. 22 and 80. Medin-
`Darby kidney cells (MDCK) cells were used as a negative
`control. Microvascular endothelial cells that exhibited uptake
`of DiI—AcLDL and the mixture of other cells in the heart and
`lung preparations that did not internalize DiI—AcLDL were
`replated separately and grown in culture using the above media
`(without heparin). After 24 h, this medium was replaced for all
`future culturing with DMEM+ for the nonendothelial cells and
`with DMEM+ with retinal—derived growth factor for the endo-
`thelial cells. Microvascular endothelial cells isolated from rat
`epididymal fat pads (RFC) were grown and plated as previously
`described (22).
`Immunoblotting of total cell lysate from cultured cells. After
`washing the confluent cell monolayers, their proteins were
`solubilized directly with cold solubilization buffer (SB) contain-
`ing 0.17 M tris(hydroxymethyl)aminomethane (Tris) — HCl (pH
`6.8), 3% (wt/vol) sodium dodecyl sulfate (SDS), 1.2% (vol/vol)
`fi-mercaptoethanol, 2 M urea, and 3 mM EDTA in double-
`distilled water as described previously (25). After incubation in
`boiling water for 10 min, a lysate volume equivalent of 106 cells
`was processed for SDS—polyacrylamide gel, electrophoresis
`(PAGE), and the separated proteins were electrotransferred
`onto Immobilon filters (Millipore, Bedford, MA) as in Refs. 25
`
`and 26. Strips of these filters were immunoblotted with rabbit
`serum, and any bound immunoglobulin G (IgG) was detected
`using anti-rabbit IgG antibodies conjugated to alkaline phos-
`phatase as in Ref. 26.
`Immunoprecipitation of gp60 radiolabeled biosynthetically
`using tritiatcd sugars. About 5 X 106 RFC cells were plated onto
`two T—75 flasks. After 1.5 h, 1.0 mCi of [2,6-8HImannose (45
`Ci/mmol; Amersham, Arlington Heights, IL) mixed in 3 ml of
`DMEM+ was added to one flask of cells or 1.0 mCi of [G-BH]
`galactose (20 Ci/mmol; ICN, Costa Mesa, CA) combined with
`1 mCi of [6-3H]glucosamine (27 Ci/mmol; ICN) mixed in 3 ml
`of DMEM+ was added to another flash of cells. After 3 days
`when the cell monolayer reached full confluency, the cells were
`washed with 10 ml of DMEM (3 times, 1 min) at 4“C, lysed
`with 5% Triton X-100 and 1% SDS in phosphate—buffered
`saline (PBS), and finally scraped from the flask. After a 10—
`min spin at 4°C (1,000 g), the supernatant was used for im—
`munoprecipitations overnight with a—gp as described in Ref.
`25. The precipitates were analyzed by SDS—PAGE and were
`visualized by fluorography as per Ref. 1.
`Immunoprecipitation of gde radioiodinated in aim. The
`endothelial luminal surface proteins of the heart vasculature
`were radioiodinated in situ using Nam‘z’l, and microspheres were
`covalently coated with lactoperoxidase and glucose oxidase as
`described previously (25). Proteins were extracted from 200 mg
`of heart tissue by m'mcing the tissue in 500 pl of 5% Triton X-
`100 and 1% SDS in PBS at 4“C. The lysate was centrifuged at
`13,000 g for 1 min at 4°C, and 100 pl of the supernatant (tissue
`extract) was mixed overnight at 4°C with 380 ul of PBS and 20
`,ul of rabbit antiserum. The ensuing immune complexes were
`then precipitated with protein A conjugated to Sepharose beads
`as in Ref. 25. The proteins bound to the beads were solubilized
`directly with 150 ,ul of cold SB, kept in boiling water for 10
`min, and analyzed by SDS-PAGE followed by autoradiography
`as described previously (‘25).
`.
`Immunoblotting of tissue extracts. Male albino rats (Sprague—
`Dawley, 200450 g) were anesthetized with an intraperitoneal
`injection of ketamine (100 ’mg/kg) and xylazine (33 rug/kg).
`The chest was opened through a median sternotomy. Through
`a needle inserted into the left ventricle and after the right
`atrium was cut for outflow purposes, the vasculature was per—
`fused with DMEM (Irvine Scientific, Irvine, CA) at a mean
`pressure of 60 mmHg first for 5 min at 37°C and then for 10
`min at 10°C. The heart, liver, cortical brain, epididymal fat
`pad, kidney, adrenals, pancreas, duodenum, diaphragm, and
`gastrocnemius skeletal muscle were excised, and 300 mg (weight
`wet) of each tissue was minced in 1 ml of SB at 4“C. For the
`duodenum, the muscosa of the intestinal wall was separated
`from the muscularis by scraping under visual
`inspection
`through a dissecting microscope. In some cases, only lung tissue
`was excised after perfusion through the right ventricle at a
`mean pressure of 20 mmHg with left atrial outflow. The lysates
`were centrifuged at 13,000 g for 1 min at 4’0, and the super-
`natants (tissue extracts) were processed equivalently for pre-
`parative SDS»PAGE and immunoblotting as described above.
`In the lung and fat pad preparations,;the centrifugation step
`did not pellet all of the tissue debris, and a top layer of fatty
`material was present and had to be carefully aspirated before
`the soluble material could be removed: In a few preparations,
`before tissue excision, the vasculature was perfused for 3 min
`directly with a protease inhibitor cocktail containing 400 ng/
`ml benzamidine, 10 pg/ml leupeptin, 10 ug/ml pepstatin A, 65
`rig/ml aprotinin, 10 ug/ml chymostatin, 100 ,ug/ml 0-phenan-
`throline, and 350 rig/ml phenylmethylsulfonyl fluoride. The
`presence of protease inhibitors did not alter the observed re-
`sults.
`Immunoblattlng of gpSO after interaction with immobilized
`albumin. A 1% Triton X<100-soluble fraction from the RFC
`cells was precipitated overnight in 90% ethanol at —20°C. The
`
`Abraxis EX2013
`
`Cipla Ltd. v. Abraxis Bioscience, LLC
`
`IPR2018-00162; IPR2018-00163; IPR2018—00164
`
`Page 2 of 10
`
`

`

`H248
`
`ENDOTHELIAL ALBUMIN—BINDING GLYCOPROTEIN GP6U
`
`albumin-binding glycoprotein gp60 and not just another
`surface glycoprotein of similar apparent molecular mass.
`Albumin was adsorbed to filters and then air dried to
`immobilize the protein to the filters. After blocking the
`filters, they were exposed to an RFC cell extract contain-
`ing gp60 and then washed. Proteins were eluted from the
`strips, separated by SDS-PAGE, electrotransferred to
`filters, and then immunoblotted with oz—gp. Controls
`included 1) exposing the cell extract to filters alone or
`to transferrin immobilized to filters and 2) testing the
`material desorbed from the transferrin and albumin fil-
`ters that had not been exposed to the cell extract. Figure
`1 shows that oz-gp detected a single 60-kDa protein only
`in eluates of the albumin~adsorbed strips interacted with
`the RFC cell extract. The eluates from all of the control
`strips were negative. These results indicate that 1) gp60
`interacts preferentially with albumin, 2) a-gp does in-
`deed recognize gp60 present in the RFC cell extracts,
`and 3) oz-gp can now be used with confidence as a probe
`for the albumin—binding glycoprotein gp60.
`Endothelial expression of gp60:
`immunoprccipitation
`with a—gp of endothelial glycoproteins radiolabeled biosyn-
`thetically in culture. Although it is clear that gp60 is
`located on the surface of vascular endothelium in culture
`(23, 25), the biosynthetic origin of gp60 is uncertain.
`Figure 2 shows that ngO can be immunoprecipitated
`specifically with or—g'p from lysates of RFC cells that had
`been biosynthetically radiolabeled with tritiated sugars.
`gp60 was radiolabeled successfully with a mixture of [6—
`3ngalactose and [6~"H]glucosamine but not with [2,6-
`3H]manncse alone. To ensure that gp60 was indeed ex—
`pressed during each radiolabeling procedure, both radi.
`olabeled cell lysates were subjected to SDS-PAGE, and
`the separated proteins were electrotransferred onto fil—
`ters that were immunoblottedWith a—gp. An equivalent
`signal for gpfiO was detected in both cases (data not
`shown). These results indicate that gp60 is indeed ex-
`
`Filters adsorbed with:
`+
`Albumin
`Transferrin —
`PBS alone
`-
`Fil’iers expOsed to:
`Cell extract
`
`+
`
`+
`_
`.
`
`-
`- +
`-
`
`+ j +
`
`-
`-
`+7
`
`+
`
`+
`-
`~
`
`-
`
`—
`+
`-
`
`-
`
`pellet was washed once using 70% ethanol and then resus—
`pended in PBS. The suspension was centrifuged at 13,000 g for
`5 min, and the supernatant was used as the cell extract. Strips
`of ImmobiIOn or nitrocellulose filters (1 x 0.25 in.) were incu-
`bated for 4 h at room temperature in 1) PBS alone, 2) 10 mg/
`ml of albumin in PBS, or 3) 10 mg/ml of transferrin in PBS.
`After air drying for at least 1 b, all strips were quenched for 1
`h with blocking solution (5% Blotto and 0.1% Nonidet P-40 in
`PBS) and then incubated overnight at room temperature in 0.5
`ml of a 1:50 dilution of the RFC cell extract in blocking solution.
`Some of the albumin strips and transferrin strips were also
`incubated in blocking solution without cell extract as additional
`controls. All strips were washed three times for 5 min with
`0.1% Nonidet P-40 in PBS and then cut into small pieces.
`Proteins were eluted using 100 ml of SB in a boiling water bath
`for 15 min, separated by SDS~PAGE on a mini-gel apparatus
`(Bio'Rad), and then electrotransferred onto lmmobilon filters
`at 40 V for 15 h. The filters were immunoblotted with a-gp as
`described above
`
`RESULTS
`
`gp60 interacts with both a—gp and albumin, Recently,
`we have shown that a-gp does interact with an RCA-
`and WGA—binding 60-kDa sialoglycoprotein (26) present
`on the surface of cultured microvascular endothelium
`derived from RFC. Other work (23) has implicated an
`RCA— and WGA-binding 60~kDa glycoprotein as an al~
`bumin-binding protein on the RFC cell surface. Because
`these data strongly suggest that a-gp is interacting with
`the albumin-binding protein gp60, we attempted to1n-
`hibit albumin binding to the surfaceof cultured RFC cell
`monolayers using an IgG fraction of a—gp isolated with
`immobilized protein A as in Ref. 26. The cell monolayers
`were first exposed for 10 min to therIgG fraction of a—gp
`(up to 100 pg/ml). Then 125I-albumin was added to
`achieve a final concentration of 2 [lg/ml, and the usual
`binding assay was performed (22, 23). Although this IgG
`fraction did interact with the 60-kDa protein by immu—
`noblotting of RFC cell lysates (26),
`it did not affect
`albumin binding to the RFC cell surface (data not
`shown), suggesting different binding sites for the anti~
`body and albumin. Therefore it became necessary first
`to ensure that a—gp is indeed interacting with the 60- kDa
`1 The observation that a-gp, even at high concentrations does not
`interfere with albumin binding to the cell surface suggests that albumin
`and or-gp may interact with ngO at different binding sites within the
`molecule. This postulation is supported by our recent investigation
`attempting to define the epitope recognized by a-gp that indicates that
`u ~gp reacts with a peptide region located apparently in the endodomain
`ofthe protein (unpublished observations). Conversely, albumin binding
`to the endothelial cell surface is expected to be via the ectodomain of
`gp60. An alternate explanation is supported by our immunofluorescence
`studies on permeabilized and nonpcrmeabilized RFC cells, which
`showed only mild labeling of the cell surface with a-gp (data not
`shown). Because the antibody was raised against an antigen denatured
`by SDS»PAGE, it is not surprising that a-gp apparently interacts
`poorly with the native form of ngO seen on the cell surface. The
`epitope may be masked in its native state and/or require some degree
`of gp60 denaturation before antibody recognition. Furthermore, since
`our results indicate that the epitope lies in the endodomain of ngO, it
`may, like glycophorin, interact with cytoskeletal proteins; this inter—
`action may interfere with antibody binding to the permeabilized cells.
`Unfortunately, the poor immunofluoresoence staining with u—gp pre-
`cludes more exact immunolocalization studies at the electron micro—
`scopic level on either cultured cells or tissue. These immunolocalization
`studies must be performed with a new antibody that recognizes 7gp60
`under more native conditions.
`
`GPGO—e
`
`1M Nl
`
`IM 1M 1M 1M
`
`Serum:
`
`Fig. 1. Specific interaction of 60»kDa glycoprctern (gp60) with albu-
`min immobilized on filters and subsequent detection by immunohlot-
`ting with antiserum against glycoprotein (a-gp). Filters were incubated
`with albumin in phosphate-buffered saline (PBS), transferrin in PBS,
`or PBS alone (as indicated for each lane above) and then air dried to
`immobilize the proteins onto filters. As indicated, strips were then
`either exposed or not exposed to rat epidyma] fat pad (RFC) cell extract
`in a blocking solution. Proteins were eluted from each strip, separated
`by SDS-PAGE, and electrotransferredonto lmmobilon filters. Strips
`of these filters containing eluted proteins are shown above and were
`immunoblotted either with nonimmune (NI) and/or immune a-gp (1M)
`serum. Bound immunoglobulin G (IgG) was detected using anti-rabbit
`IgG antibodies conjugated to alkaline phosphatase as described previ»
`ously (25). Gp60 was detected only after interaction of cell extract with
`albumin filters and not in various controls.
`
`Abraxis EX2013
`
`Cipla Ltd. v. Abraxis Bioscience, LLC
`
`IPR2018-00162; IPR2018-00163; IPR2018—00164
`
`Page 3 of 10
`
`

`

`ENDOTHELIAL ALBUMIN~BINDING GLYCOPROTEIN GPGO
`
`H249
`
`[MMUNOPRECIPITATION
`WITH ANTIvGLYCOPHORlN SERUM
`"‘GI’G--u--l“nl‘l-*r
`
`66 .
`SP“-
`
`45- NIMNIIM
`
`Fig. 2. Biosynthetically radiolabeled glycoproteins of RFC cells im—
`munoprecipitated with a-g‘p serum. RFC cells were radiolabeled with
`either [“H]mannose (Man) or both {3H]galactose and [“H]glucosamine
`(G/G) as descibed in METHODS. Lysates from these cells were subjected
`to immunoprecipitation with either nonimmune (N1) or immune (1M)
`wgp rabbit serum. Immunoprecipitates were separated by SDS<PAGE
`on a 5—15% gradient gel and visualized by fluorography. Only ngO was
`detected specifically with a-gp, The band found at 45 kDa is considered
`to be nonspecific since it was detected both in nonimmune serum and
`to a lesser extent in immune serum.
`
`presSed by vascular endothelium in culture. Furthermore,
`the lack of incorporation of [2,6-3H]mannose into gp60
`is consistent with our previous findings based on direct
`and sequential lectin affinity chromatography (25) that
`suggest that rat gp60 contains O-Iinked glycans but ap-
`parently not N—linked glycans.
`Endothelial cell isolation from rat tissues. Fluorescence—
`activated cell sorting Was used to isolate microvascular
`endothelial cells from a mixture of cells derived from a
`collagenase treatment of cortical brain, heart, and lung
`tissues (see METHODS). The specific ability of endothelial
`cells to internalize DiI-AcLDL was used to isolate endo-
`thelial cells from nonendothelial cells (22, 30). Those
`cells that lacked uptake of DiI-ACLDL were also saved
`from the heart and lung preparations. After sorting, all
`cells were replated, grown in culture for- several manages,
`and checked periodically for DiI-AcLDL uptake using
`either the fluorescence-activated cell sorter or simple
`fluoresencc microscopy of the cells grown on slides as
`described in Ref. 22. Figure 3 shows the fluorescence
`profile of DiI~AcLDL uptake of a typical cell preparation.
`The cells initially isolated from tissue have a bimodal
`cell distribution with considerable variation in cell size
`and fluorescence intensity. Most of the cells have fluo-
`rescence intensity far greater than the MDCK cells (neg—
`ative control) and appear to internalize DiI—AcLDL. A
`second group of cells has a fluoresceIICe intensity more
`comparable to the MDCK cells, especially when cell size
`is considered. The top 20% of the cells with the greatest
`fluorescence intensity and the bottom 20% of the cells
`with the lowest fluorescence intensity were collected
`separately using the cell sorter. These cells after growth
`in culture were examined for DiI-AcLDL uptake once
`again and, as shown in Fig. BB, the endothelial cells that
`were positive on the first sort are now unimodal in
`distribution with considerably greater fluorescence in-
`tensity than the negative control. Meanwhile, the other
`isolated cells shown in Fig. 30, which should be nonen-
`dothelial in origin, also appeared to have a unimodal cell
`distribution; however, in this case, their fluorescence was
`equivalent to the negative control. Many of the endothe-
`lial cell preparations were also checked periodically by
`immunofluorescence as described in Ref. 22 and in each
`case stained positively for other endothelial markers,
`such as angiotensin-converting enzyme or factor VIII
`(data not shown). This approach has allowed us to isolate
`
`NUMBER
`CELL
`
`MDCK
`
`MnoK
`
`Cell Size
`
`Fluorescence Intensity
`Fig. 3. Profile of fluorescenceactivated cell sort of cultured cells de-
`rived from a rat heart preparation. Cells were isolated from 4 rat hearts
`and grown in culture until Confluency (see METHODS). These cells were
`incubated with 1,1’ ~dioctadecylv3,3,3’ ,3’ -tetramethylindocarbocyanine
`perchlorate-labeled acetylated low-density lipoprotein (DiI-AcLDL)
`and processed for fluorescence-activated cell sorting as described pre-
`viously (20, 30). Cell size and fluorescence intensity profiles of cells are
`shown as a function of cell number. Profiles of initial population of
`heart-derived (IHD) cells are given in A. MDCK cells were also exam-
`ined as a negative control. Twenty percent of cells with highest and
`lowest fluorescence intensity were isolated separately using cell sorter
`and then grown in culture. After 3 passages, these 2 cell populations
`were reexamined for DiI-AcLDL uptake, and their profiles are shown
`in B [rat heart endothelial (RHE) cells are derived from those cells
`with highest fluorescence intensity of Dil—AcLDLpositive cells] and C
`[heart mixture (HM) of cells grown from bottom 20% of cells with
`lowest flourescence]. When total fluorescence of RHE cells was com-
`pared with that of MDCK cells, >98% of RHE cells internalized more
`DiI-Acl.DL than any control cells. On the other hand, HM cells
`exhibited same degree of fluorescence as MDCK cells.
`[To get a
`population of ceHs that lacked DiI-AcLDL uptake (i.e., were devoid of
`endothelial cells), we found it necessary to passage cells in culture at
`least twice]
`'
`
`and grow in culture endothelial cells from the rat heart,
`lung, and brain and nonendothelial cells from the heart
`and lung.
`.
`Immunoblotting lysates of cultured cells. The proteins
`from the organ—derived cells isolated and grown in cul-
`ture (as described above) Were solubilized, separated by
`SDS—PAGE, electrotransferred to filters, and then im-
`munoblotted with a-gp and nonimmune serum. Proteins
`from RFC cells acted as a positive control. Figure 4
`shows that gp60 was present only in the lysates of
`endothelial cells isolated from the heart and lung but not
`from the cortical brain.2 For the heart and lung prepa—
`rations, only the DiI-AcLDL—positive endothelial cells
`expressed gp60, whereas the mixture of nonendothelial
`cells (probably consisting of fibroblasts, pericytes, and/
`or smooth muscle cells) neith’er internalized DiI—ACLDL
`nor appeared to express gp60. These results indicate that
`under the conditions used here to isolate and grow these
`cells, microvascular endothelia from the heart and lung
`but not the brain express gp60 in culture, whereas non-
`endothelial cells from the heart “and lung do not. We
`have also checked by immunoblotting several specific
`2 gptiO can also be immunoblotted specifically with a—gp using lysates
`from other cultured endothelium derived from sheep pulmonary artery,
`bovine aorta, fetal bovine heart (ATCC), microvessels of pig atrium,
`and human umbilical vein (unpublished observations).
`
`Abraxis EX2013
`
`Cipla Ltd. v. Abraxis Bioscience, LLC
`
`IPR2018-00162; IPR2018-00163; IPR2018—00164
`
`Page 4 of 10
`
`

`

`H250
`
`ENDOTHELIAL ALBUMlN-BINDING GLYCOPROTEIN GP60
`
`Tissues - W
`Cells-
`RFC
`REE
`RHE (HM) RLE (LM)
`DiI‘AcLDL:
`+
`+
`—
`+
`+
`—
`
`61,60"
`
`
`
`
`
`.
`i
`’
`‘i
`.
`.
`,g
`H-
`. ..
`
`NI 1M NI 1M NIle IM NI
`IM 1M
`Serum:
`Fig. 4. Differential expression of ngO in rat microvascular endothelial
`cells isolated from different tissues. RFC cells along with microvascular
`endothelial cells isolated from rat brain (REE), heart (RHE), and lung
`(RLE) were processed for immunoblotting overnight with a 1:500
`dilution of either nonimrnune (NI) or immune (IM) ct-gp serum. In
`addition, a mixture of nonendothelial cells isolated from rat heart
`(HM) and lung (LM) were also immunoblottedwith u-gp. In each case,
`a lysate equivalent of ~106 cells was loaded onto a preparative gel. Both
`endothelial and nonendothelial cells were passaged 3 times in culture
`after their initial isolation and separation by fluorescence-activated
`cell sorting. Heart and lung endothelial cells both internalized DiI~
`AcLDL and expressed ngO, whereas nonendothelial cells did neither.
`_ (31160 was not detected in lysates of cultured brain endothelium.
`
`nonendothelial cell types grown in culture. Normal rat
`kidney fibroblasts [from American Type Culture Collec—
`tion (ATCC); NRK—49F] that bind albumin poorly (22)
`and rat aortic smooth muscle cells (from ATCC; A10)
`were negative for ngO expression (data not shown).
`Therefore, in the cultured cells tested so far, the expres—
`sion of ngO appears to be specific for certain endothelial
`cells.
`
`Selective albumin binding to rat cultured endothelial
`cells. Because the data given above indicate that ngO
`expression is limited to only certain endothelial cells [rat
`heart endothelial (RHE), RFC, and rat lung endothelial
`(RLE) but not rat brain endothelial (RBE) cells], the
`binding of albumin to these cells should vary in accord—
`ance with gp60 expression if ngO does indeed play a role
`in albumin binding. Rat serum albumin (BSA) was ra-
`dioiodinated, and the binding of 125LRSA to the surface
`of these cultured cells was compared using 0.1 rug/ml of
`125I-RSA as described previously (22). Because the RFC
`cells bind RSA and were used originally to characterize
`the kinetics of albumin binding to cultured microvascular
`endothelium (22), they acted as a positive control. As
`shown in Fig. 5, 125I—RSA binding was ~10—15 times
`greater for the RLE, RHE, and RFC cells than for the
`RBE cells. This small amount of binding of albumin to
`the RBE cells is comparable to that observed previously
`for rat fibroblasts (22). This selective albumin binding
`to certain cultured endothelial cells agrees well with the
`above observationthat the RLE, RHE, and RFC cells
`expressed gp60 but the RBE cells did not. In addition,
`the increase of 40—50% in 125I—RSA binding observed for
`the RLE and RHE cells over the RFC cells correlates
`well with the greater expression of ngO by these cells
`(see Fig. 4).
`L
`Presence of gp60 on the surface of microuascular endo—
`thelium in situ. Most of our previous work has focused
`on using in vitro systems to study molecular interactions
`at the surface of the vascular endothelium. Recently, we
`have examined the extent of phenotypic drift caused by
`cell culture by performing intravascular radioiodinations
`of endothelial surface proteins to compare those proteins
`identified in situ with those in culture (25). Here we
`
`30
`
` @220
`
`8 8
`m ‘5
`<15m.
`flaw
`3 5
`‘—
`
`0
`
`11.4%
`
`REE
`RLE
`RHE
`RFC
`Fig. 5. Differential albumin binding to cultured rat endothelial cells.
`RFC cells along with microvascular endothelial cells isolated from rat
`brain (REE). heart (RHE), and lung (RLE) were examined for their
`ability to bind radiciodinated rat serum albumin (BSA) using 0.1 mg/
`ml of 12’E'LRSA in an in vitro assay described previously (20). For each
`endothelial cell
`type, mean value of binding observed from 4 cell
`monolayers in dishes is expressed as rig/10“ cells, with error bars
`denoting calculated SD values. Data were also normalized relative to
`RFC cells and are expressed as a percentage, which is written above or
`within columns representing each cell type.
`
`205 -
`
`116
`
`97
`
`NI
`IM
`Fig. 6. Intravascular radioiodinated surface proteins of heart vascular
`endothelium immunoprecipitated with cr—g‘p ser