Isolation and culture of high endothelial cells from rat lymph nodes^*

The protective function of the immune system depends on the dissemination of immunocompetent cells throughout the body (Ford & Gowans, 1969; Ford, 1975; Sprent, 1977). The entire organism is surveyed by the continuous traffic of lymphocytes from the blood to the tissues. During this circuit lymphocytes adhere to capillary endothelium and subsequently migrate between endothelial cells to enter the extravascular space. Normally the rate of lymphocyte extravasation is highest in secondary lymphoid organs such as the spleen and lymph nodes (1N) (Gowans & Knight, 1964). In LN of rodents and humans lymphocytes migrate from specialized post-capillary venules (SPCV) that are lined with columnar or high endothelial cells.

The interaction of lymphocytes with this specialized endothelium has been investigated both in vivo, by following the LN localization of lymphocytes after intravenous administration, and in vitro by the adhesion of lymphocytes to the cut surfaces of SPCV in cryostat sections of peripheral LN. The effect of pretreatment of lymphocytes with a variety of reagents (enzymes, lectins, polysaccharides and antibodies) on their subsequent adhesion to endothelium has been studied using these techniques. However, selective pre-treatment of the endothelium has so far been precluded. Consequently our understanding of the role of this specialized endothelium in the lympho-cyte-SPCV interaction is limited (Ford et al. 1984).

The role of the endothelium can be studied more easily using an in vitro model of lymphocyte adhesion that uses cultured endothelial cells (EC) as the substrate, de Bono (1976) and Beesley et al. (1978), have used EC cultured from large blood vessels such as the aorta. However, lymphocytes do not normally migrate across the aortic wall. Therefore an attempt was made to isolate and culture endothelium from SPCV of normal LN.

High endothelial cells in rat LN synthesize a unique sulphated glycolipid. This macromolecule has been used to identify high endothelial cells in lymphoid tissue following the rapid, preferential uptake of ³⁵SC₄ by these cells (Andrews et al. 1982, 1983). This property was used to label high endothelial cells prior to isolation and to identify these cells after isolation. Primary lymph node cultures were obtained in which two different types of non-lymphoid cell were present in large numbers. The majority (>70%) were 20 μm diameter, round and ³⁵S-labelled, and were therefore high endothelial cells (HEC). The remaining unlabelled cells were 10–15 μm diameter and were identified as macrophages by phase-contrast microscopy. Isolated cells proliferated after 1–2 days and lymph node cultures were enriched for the former cell type as macrophages did not persist beyond 7–10 days. The identity of HEC was further substantiated using a polyclonal antiserum raised against lymph node cultures that stained high endothelium of SPCV in LN sections.

At confluence primary lymph node cultures bound lymphocytes as efficiently as high endothelium in lymphoid tissue and 40-fold more efficiently than rat aortic endothelial cells. It is concluded that lymph node cultures contain high endothelial cells (HEC) and that these cells continue to express surface determinants for lymphocytes in vitro.

Popliteal LN from 250 g F₁ hybrid rats, 96 h after footpad injections of 10⁷ parental lymphocytes (4 rats; 150–200 mg tissue) or cervical LN were washed twice with Dulbecco’s A+B (Oxoid Ltd, Basingstoke, Hants, UK) plus antibiotics (50unitsml^-1 penicillin + 50μgml^-1 streptomycin) (DAB), incubated in 10% Betadine (Napp Lab., Cambridge) in DAB for 10min and washed three times with 20ml DAB. Tissue was minced finely and allowed to settle out of 20 ml DAB for 1–2 min. The cell-rich supernatant was discarded and the tissue washed three times with 20 ml DAB, resuspended to 50 mg ml^-1 in RPMI 1640 containing 10mM-NaHCO₃, 20mM-Hepes, 2mM-glutamine, antibiotics (RPMI_inc) and 0·5% type II collagenase (Sigma Chemical Co., Poole, Dorset, UK). After shaking for 60 min at 37°C the tissue was dispersed with a Pasteur pipette, resuspended in 10 ml RPMI_inc and filtered through 100gm nylon mesh (Simon Special Products Ltd, Stockport, Cheshire, UK) supported in 23 mm Millipore Swinnex filter holder. Isolated cells were collected by centrifugation at 250g for 5 min and plated in RPMI 1640 containing 20mM-NaHCO₃, 2mM-glutamine, antibiotics (RPMIgro), and 20% heat-inactivated (56°C; 30 min) foetal calf serum (FCS) (Sera-Lab Ltd, Crawley Down, Sussex, UK) in a 55mm Petri dish (Nunc, Denmark). Cells were grown at 37 °C in a humidified atmosphere of 5% CO₂ in air. After 60 min, non-adherent cells were removed and fresh growth medium added.

Primary HEC cultures were routinely rinsed twice with DAB and RPMIgro plus 20% FCS added every 48–72h. Cells were subcultured using 01% trypsin (Difco 1:250; Detroit, Michigan)-0·025% EDTA in phosphate-buffered saline (PBS) and plated at 50% of confluent density.

In order to identify high endothelial cells isolated from LN these cells were labelled prior to enzyme digestion; 1–2 mm slices of popliteal LN (100 mg total) were incubated with 0·8 ml of 50μCiml^-1 sodium ³⁵SO₄ (>5mCiμg^-1S; Amersham International, UK) in RPMI_inc for 30 min at 37°C. Type II collagenase was added to a final concentration of 0·5% and HEC isolation continued as above. Isolated cells were plated on 13 mm Thermanox coverslips (Miles Laboratories Inc., Naperville, III) in 24-well tissue culture plates (1inbro Space Saver, Flow Laboratories, UK). After 60min, adherent cells were washed three times with DAB, fixed with neutral buffered formalin for 30 min, processed for autoradiography (Ford, 1978) and counter-stained with methyl green and pyronin. Non-lymphoid cells were scored for size and ³⁵S label. Cells with six or more silver grains were positive. In order to measure the yield of high endothelial cells following enzyme digestion, radioactivity associated with undigested tissue retained on 100 μm filters and with isolated cells was determined by scintillation counting after dissolution in NaOH (Ford, 1978).

Thoracic aorta from 250 g rat was cleaned and cut into 3 mm² pieces. The tissue was washed three times with 20 ml DAB and incubated in 5 ml RPMI_inc plus 0·2% type II collagenase for up to 120min at 37°C. Isolated cells and undigested tissue were plated in2×90mmNunc Petri dishes in RPMIgro plus 20% FCS. Non-adherent cells and tissue were removed after 24 h and cultures were rinsed and fed every 48–72 h.

A mixed culture containing morphologically distinct populations of endothelial cells, smooth muscle cells and fibroblasts was identified after 2 days. Areas of closely apposed 50–100 polygonal endothelial cells were selectively removed after 4–7 days using 0·1% trypsin/0·025% EDTA in PBS and an 8 mm diameter glass ring. Aortic endothelial cells (AEC) were vWF:Ag (von Willebrand factor antigen, formerly factor VIII related antigen) positive (see immunofluorescence) and grew to confluence with a doubling time of 2–3 days. AEC were routinely subcultured using 0–1% trypsin/0·025% EDTA in PBS and plated at 50% of confluent density. AEC isolated by this method were passaged seven times, after which their morphology changed significantly from polygonal to bipolar as described for ‘sprouting’ bovine AEC in culture (McAuslan & Reilly, 1979). AEC were discarded at this point. A similar method was used to select aortic fibroblasts from these mixed cultures after 4—7 days. Fibroblasts doubled every 24 h and were subcultured using trypsin/EDTA and plated at 10% of confluent density.

Primary cultures of HEC were plated at confluent density on 13 mm glass coverslips and used within 5 days. EC were rinsed once with DAB and preincubated in RPMI_inc plus 1% FCS for 60min at 37°C.

Thoracic duct lymphocytes (TDL) collected for 4h at room temperature via thoracic duct cannulae into 20 ml DAB plus 5 units ml^-1 heparin (Ford, 1978) were collected by centrifugation for 10 min at 250 g and resuspended in RPMI_inc plus 1% FCS at 10⁸ cells ml^-1; 3×10⁶ TDLml”¹ were suspended above EC and incubated at 37°C in static culture. Unadhered TDL were removed after agitation and EC washed once with RPMI_inc. This washing procedure was repeated five times and cultures were fixed with 2·5% glutaraldehyde in 0·1 M-cacodylate buffer for 30 min at 37°C. Cells were stained with Toluidine Blue, and TDL and EC nuclei were scored using high-power light microscopy; 100 fields of view at X1000 magnification were studied and the ratio of TDL:EC was calculated. This represented 1/50 of total coverslip area, and 1000 HEC were counted.

A sample of 10⁷ HEC from cervical LN (up to 6th passage) were detached from culture dishes using 0·025% EDTA in PBS for 5 min at 37°C, resuspended in 0·5 ml PBS and emulsified 1:1 with complete Freund’s adjuvant (Sigma Chemical Co.). The emulsion was administered by 4×0·25-ml subcutaneous injections into dorsal skin together with 6× 10¹⁰ heat-killed pertussis in PBS and followed by a boost of 5×10⁶ cells in incomplete Freund’s adjuvant 21 days later. Subsequent boosts of 5× 10⁶ cells were administered in PBS at intervals of at least 21 days. Blood was collected from the central ear artery 10 days after each of three boosts and anti-HEC antiserum was sequentially absorbed against rat serum proteins and foetal calf serum proteins coupled to Sepharose 4B..

Cryostat sections (5 Jim) of cervical and mesenteric LN, Peyer’s patch or aorta were extracted with dry acetone at –20°C for 10 min. EC were plated on 13 mm glass coverslips and after 24–48 h extracted with acetone or methanol at – 20°C for 10 min. Coverslips were mounted onto glass slides using U.V. adhesive (1octite, Welwyn Garden City, UK). Cultured cells and tissue sections were stained as follows. All incubations were performed at room temperature in a humidified atmosphere. Sera were diluted in PBS containing 0·1% bovine serum albumin (Fraction V, Sigma Chemical Co.) and centrifuged at 100000gfor 10min in an Airfuge (Beckman Ltd, UK) to remove aggregates. Nonspecific binding was blocked by incubating slides with 10% sheep serum for 30 min. Excess serum was removed by blotting and slides were incubated with anti-human von Willebrand factor antigen (vWF:Ag) antiserum (Sera-Lab Ltd) diluted 1: 20, anti-HEC antiserum diluted 1:50 or non-immune rabbit serum for 60 min and washed for 15 min in PBS with two changes. Slides were incubated with FITC-labelled sheep anti-rabbit immunoglobulin G (Sera-Lab Ltd) at 1:50 dilution for 60 min, washed for 15 min as above and mounted in glycerol containing 2·5% 1,4-diazobicyclo-(2,2,2)-octane (BDH Ltd, UK) to retard fading fluorescence (Johnson et al. 1982).

The production of sulphated glycolipid by HEC (Drayson et al. 1981) and the proliferation of HEC (Anderson et al. 1975) are both elevated in anti-genically stimulated LN. Popliteal LN undergoing a graft versus host response were therefore used as a source of ³⁵S-labelled and proliferating HEC. LN slices were labelled with ³⁵SO₄ and, because the half-life of sulphated glycolipid in HEC is about 60 min, the time taken to process isolated cells for autoradiography was kept to a maximum of 60 min in order to optimize localization of ³⁵S label in HEC.

Autoradiographs revealed three distinct populations of cells in collagenase digests of ³⁵S-labelled LN: (1) 20–30 μm diameter ³⁵S-labelled, pale-staining cells; (2) 10—15 μm diameter non-lymphoid cells, which were unlabelled and lightly stained; and (3) lymphocytes, which were unlabelled and darkly stained (Fig. 1A). Of the two populations of nonlymphoid cells (1 and 2) 71·7 ±4·9% (±S.D., n = 4) were large and labelled, and were therefore HEC. The majority (>95%) of the remaining unlabelled cells were medium sized. HEC in fixed and stained preparations were round cells and were found singly, occasionally with one or two lymphocytes attached.

In three experiments the yield of cell-associated radioactivity measured by scintillation counting after digestion of ³⁵S-labelled LN with 0·5% collagenase was 7·5, 25·7 and 3·5% (mean 12%). This was equivalent to 5×10³ of ³⁵S-labelled HEC from 100 mg LN measured by autoradiography. Assuming 10⁶ cells mg^-1 LN then HEC occupy 0·05% of lymphatic tissue. Because of the low yield of cell-associated radioactivity found by using collagenase digestion, a variety of other enzymes were compared. The yield of cell-associated ³⁵S label found by using 0-5% dispase (Boehringer-Mannheim) was 4%, a mixture of 0·5% collagenase and 0·5% testicular hyaluronidase (Sigma Chemical Co.) yielded 21·8%, and in three experiments a 0·5% collagenase/0·5% dispase mixture released 81·6, 49·2 and 63·8% (mean 65%) of LN-associated ³⁵S-label.

Although the yield of cell-associated ³⁵S label was increased fivefold, autoradiographs showed that a mixture of 0·5% collagenase and 0·5% dispase did not increase the number of ³⁵S-Iabelled HEC obtained. The difference in yield was accounted for by labelled cell debris resulting from the harsher enzyme treatment. Although small, the yield of viable HEC from 0·5% collagenase digests was adequate to establish primary cultures (see next section) so this enzyme concentration was used routinely.

Phase-contrast microscopy of cells isolated from LN by collagenase digestion showed three distinct populations of cells that could be directly identified with those in autoradiographs. HEC were large, round, single cells occasionally with one to three lymphocytes attached to their surface. Most HEC had a distinct nucleus with at least two nucleoli and HEC cytoplasm was fiat and vesiculated (Fig. IB). The medium-sized non-lymphoid cells identified by autoradiography clearly resembled macrophages by’ phase-contrast microscopy, having a phase-dark, condensed cytoplasm and an indistinct nucleus (Fig. IB). This population of cells did not proliferate but persisted in the cultures for up to 10 days. The lymphocytes in these cultures reflected incident light and were therefore spherical. During the first two days of culture the lymphocytes were washed away and HEC started to move around the culture dish. In so doing HEC adopted varying morphologies ranging from amoeboid, with at least one pseudopodium, to bipolar, with long cytoplasmic processes (Fig. 2A). After 2 days most HEC were no longer vesiculated and the cells started to proliferate. HEC remained motile such that divided cells became singly dispersed but occasionally clumps of 5–10 polygonal FIEC were seen by day 5 (Fig. 2B). At this stage of proliferation and at all later stages HEC displayed abundant granules dispersed around the nucleus (Fig. 2A-D). At confluent densities HEC underwent a dramatic morphological change to a bipolar one and HEC tended to align in parallel arrays (Fig. 2C).

To date, 54 primary HEC cultures have been established and maintained to first confluence taking an average of 10 days to double seven to eight times. At confluence HEC showed substantial cytoplasmic overlap but nuclear overlap was not obvious. HEC reached a density of 350 cells mm^-2 having been plated at 2 cells mm^-2 after isolation. Confluent HEC cultures contained a few heavily vesiculated cells (≈0·01%), which were morphologically similar to freshly isolated HEC. HEC morphology at all stages of proliferation was quite distinct from that of other vascular endothélia in culture, e.g. AEC (Fig. 2E), and from non-endothelial cells such as adventitial fibroblasts (Fig. 2F). Primary HEC cultures occasionally contained clumps of two to three microvascular EC (Folkman et al. 1979) and/or fibroblasts but neither cell type proliferated or persisted beyond 5 days after isolation.

Primary HEC cultures were passaged using trypsin-EDTA in PBS and plated at 50% of confluent density. HEC continued to proliferate but with an apparently reduced doubling time; 16 cultures have been maintained to second confluence achieving one cell doubling after 8 days and one culture has been maintained to eighth confluence achieving one cell doubling after an average of 6 days. Because of the significant reduction in proliferation rate after primary culture all subsequent HEC cultures have been termed ‘late-passage’ HEC. When plated at sub-confluent densities late-passage HEC flattened up to 100 μm diameter and exhibited ‘stress fibres’ along one axis of the cell (Fig. 2D). As HEC grew to confluence stress fibres disappeared and HEC aligned in a parallel array as seen at first confluence. Primary and late-passage HEC cultures have been frozen in RPMIgro containing 20% FCS and 10% dimethyl sulphoxide at 5×10⁵ to 10⁶ cells ml^-1, stored for up to 6 months and thawed successfully (Trypan Blue exclusion >95%).

In SPCV of peripheral LN lymphocytes adhere to high endothelium and then migrate between the endothelial cells to enter the LN parenchyma. In order to determine whether cultured HEC demonstrated this specific interaction with lymphocytes the adhesion of lymphocytes to cultured HEC was studied. Thoracic duct lymphocytes (TDL) were incubated for 45 min with primary cultures of HEC. After six washes to remove unadhered TDL phase contrast microscopy showed clumps of >10 TDL attached to single HEC (Fig. 3A). Following glutaraldehyde fixation these clumps of lymphocytes were no longer present in the culture but single TDL were attached to HEC in large numbers (Fig. 3B). The ratio of TDL:HEC nuclei in fixed and stained preparations was 0·6 ±0·2:1 (s.E.M., n = 12) (Fig. 3C). The percentage of TDL bound showed a wide variation from 1·5–7% of lymphocytes added.

Further examination of living cells by phasecontrast microscopy revealed that lymphocytes were either spherical, reflecting incident light or significantly flattened, appearing as phase-dark cells (Fig. 3A,B). High-power microscopy of fixed and stained cells showed that spherical TDL exhibited little intracellular detail and were attached to the exposed surface of HEC, whereas flattened TDL revealed a large nucleus with a low cytoplasm: nucleus ratio and were between HEC and the culture vessel. From a series of experiments the mean number of TDL underneath HEC was 22·8 ±3·4% (±S.E.M., n = 12). Cells other than lymphocytes were rarely observed in these preparations (<1%) but larger cells, which could be either blast cells or macrophages, have been seen on these occasions.

The adhesion of TDL to cultured aortic EC was compared since lymphocytes do not normally adhere to the aortic wall. Aortic EC showed a 40-fold lower affinity for TDL than HEC. After 45 min the ratio of TDL:EC nuclei in three experiments was 0·012, 0·016 and 0·022:1. All lymphocytes in these experiments were spherical and therefore were attached to the exposed surface of AEC (Fig. 3C,D). The adhesion of TDL to adventitial fibroblasts was similar with a ratio of 0·02: 1.

Preliminary observations on the kinetics of lymphocyte adhesion to cultured HEC showed that maximal binding occurred between 45 and 60 min. During longer incubations (60–120 min) the number of lymphocytes underneath the HEC layer increased in a linear fashion but there was a 50% reduction in the total number of cells remaining bound to HEC at 120min (Table 1). When the concentration of lymphocytes was varied from 0·1×10⁶ to 10×10⁶ cells ml^-1 the number of lymphocytes bound to cultured HEC after 60 min was directly proportional to the number added. Results from a single experiment showed that the percentage of lymphocytes bound was about ≈7% of total lymphocytes added over the range 0·3×10⁶ to 7·7× 10⁶ lymphocytes ml^-1 (Table 2).

One of the most widely used criteria for the positive identification of endothelial cells is immunohistochemical staining of vWF:Ag (Bloom et al. 1973; Hoyer et al. 1973; Jaffe et al. 1973). The distribution of vWF : Ag in LN endothelium was therefore determined using a rabbit anti-human vWF:Ag antiserum that cross-reacts with vWF:Ag in rat vascular endothelium (Bowman et al. 1981).

The luminal surfaces of the hilar artery and vein were positive for vWF : Ag as were arterioles, venules and capillaries lined with flat endothelium in the medulla of the LN (Fig. 4A,B). The staining pattern of capillary endothelium was distinctly punctate, which is the characteristic distribution of vWF: Ag in vascular endothelium. The luminal surface of the aorta was positive for vWF : Ag, showing a continuous line of staining as in the hilar artery and vein (Fig. 4E). Five days after isolation from aorta by collagenase digestion, morphologically distinct EC were vWF : Ag positive, showing a punctate pattern of staining as described by Bowman et al. (1981).

The numerous SPCV distributed throughout the paracortex of cervical LN did not stain with vWF : Ag antiserum. Occasionally some staining was restricted to the lumens of these vessels suggesting the presence of vWF : Ag in blood (Fig. 4C). Lymphatic endothelium of afferent vessels in the subcapsular sinus and of efferent vessels in the medulla was also negative for vWF:Ag. Primary HEC cultures 5–10 days after isolation did not stain for vWF:Ag.

The absence of vWF:Ag in SPCV and thereby a means of identification of isolated high endothelial cells led to the production of a polyclonal anti-HEC antiserum in order to determine if antigenic determinants were shared between cultured HEC and high endothelium of SPCV in peripheral LN.

Rabbit anti-rat HEC antiserum was absorbed against rat serum proteins and foetal calf serum proteins coupled to Sepharose 4B. This reduced the titre of the immune sera by≈20-fold but the pattern of immunofluorescent staining (see below) was not altered. Using this antiserum cultured HEC were stained uniformly throughout the cytoplasm (Fig. 5A). The cytoplasm of aortic EC was similarly stained in a uniform manner (Fig.5B).

In cryostat sections of LN and Peyer’s patches anti-HEC antiserum stained SPCV uniformly across the vessel wall. Lymphocytes either in the vessel lumen or throughout the LN parenchyma were not stained (Fig. 5C). This antiserum also showed a continuous line of staining associated with capillaries and lymphatic vessels. The continuous, flat endothelium of medullary sinuses (Fig. 5D) and the discontinuous, flat endothelium of the subcapsular sinuses (data not shown) were clearly outlined. The staining pattern of arterioles and venules in the LN and of the aorta revealed that anti-HEC antiserum stained throughout the walls of these vessels (Fig. 5E). The staining pattern of LN, apart from the obvious large vascular and lymphatic vessels, could probably be accounted for by the extensive vascular and lymphatic capillary networks which extend throughout this tissue (Fig. 5C-E) (Anderson et al. 1975).

An enriched population of cells has been cultured from rat LN that have several properties of high endothelial cells in lymphoid organs.

High endothelial cells in rat LN rapidly incorporate ³⁵SO₄ into a sulphated glycolipid-glycoprotein complex, which is rendered insoluble by fixation, and this macromolecule has been used to identify high endothelial cells from other vascular and stromal cells of lymphoid tissue (Andrews et al. 1980, 1982). Collagenase digestion of lymphoid tissue in which high endothelium was ³⁵S-labelled yielded primary cultures in which a majority of the cells (>70%) were ³⁵S-labelled and were therefore high endothelial cells (HEC). The only other non-lymphoid cells present in significant numbers (>25%) were identified as macrophages by phase-contrast microscopy. The demonstration that >90% of isolated cells belonged to only two different types of non-lymphoid cell was an unexpected finding. Primary lymph node cultures occasionally contained morphologically distinct micro-vascular EC and/or adventitial fibroblasts (>1% of isolated cells) but neither cell type survived for more than 5–6 days. Primary cultures were enriched for HEC as macrophages did not persist beyond 7–10 days. Isolated HEC proliferated to first confluence with an apparent doubling time of 1–2 days and underwent dramatic changes in morphology as a consequence of cell motility and cell division. HEC were subcultured and continued to proliferate in longterm culture for up to 2 months but with an apparently reduced doubling time of 6—8 days. Even though no attempt was made to purify these cultures, at no stage was there more than one morphologically distinct type of cell present.

Following antigenic challenge the proliferation of high endothelium in LN is increased 5-to 10-fold (Anderson et al. 1975). In order to increase the possibility of isolating high endothelial cells that might proliferate in vitro, popliteal LN undergoing a graft versus host response were used initially. When HEC cultures isolated from these ‘activated’ LN were compared with those isolated from cervical LN of rats kept under conventional animal-house conditions, similar degrees of HEC purity were obtained. HEC isolated from cervical LN proliferated in the same way as HEC from ‘activated’ popliteal LN. In practice, cervical LN proved to be a more convenient source of HEC cultures and were therefore used routinely.

The preferential labelling of HEC in lymphoid tissue by ³⁵SO₄ was crucial to the demonstration of HEC isolated from LN following enzyme digestion and it requires some discussion. Andrews et al. (1980) showed that under the labelling conditions employed in this study (maximum of 75 min) ³⁵SO₄ did not significantly label any other vascular or stromal cell in lymph node slices apart from HEC. However, following prolonged labelling in vivo, mast cell granules were significantly labelled 3 h after administration of radiolabel (presumably sulphated proteoglycans/glycosaminoglycans). Hence the time course of sulphate labelling of lymphoid tissue yielded the selectivity of HEC labelling. Extensive biochemical studies concluded that sulphate did not significantly label a pool of sulphated proteoglycan over this short time-course either in HEC or in any other vascular or stromal cell (Andrews et al. 1980, 1982).

The identity of HEC in lymph node cultures was further substantiated using a rabbit polyclonal antiserum raised against lymph node cultures, which stained high endothelium of SPCV in cervical and mesenteric LN and Peyer’s patches demonstrating that these cultures shared antigenic determinants with high endothelial cells in lymphoid tissue. No attempt was made to study the range of antigenic determinants recognized by this antiserum, therefore it is not possible to comment on the significance of the additional staining of vascular and lymphatic vessels that are not lined with high endothelium. This demonstration of shared antigenic determinants has suggested the production of mouse monoclonal antibodies to cultured HEC in an attempt to identify antigens specific to HEC. This approach will depend on at least two factors: first, that HEC display specific antigens in significant amounts; and second, that these antigens in the rat are significantly different from their counterpart in the mouse.

In order to determine whether cultured HEC continued to express surface determinants for lymphocytes in vitro, the adhesion of thoracic duct lymphocytes to these cells was studied. Primary cultures of HEC bound lymphocytes with a 40-fold higher affinity than either aortic EC or adventitial fibroblasts. Up to 25% of lymphocytes that adhered to these cultures were found to be between the cultured cells and the culture vessel following migration across the cell layer. The adhesion of lymphocytes to vascular endothelium cultured from large blood vessels is stimulated by monokines and/or lymphokines (Cavender et al. 1986; Yu et al. 1985). It seems unlikely that the adhesion of lymphocytes to cultured HEC reported here was dependent on such soluble factors in the cultures, since monocytes and/or lymphocytes had not been present in HEC cultures for at least 9 days before the adhesion assay was performed.

A preliminary study of the kinetics of lymphocyte adhesion to cultured HEC showed that the binding was saturated after 45–60 min incubation. An average ratio of lymphocytes : HEC of 0·6:1 was measured, which correlates well with ratios measured in histological preparations of normal lymph nodes of 0·75:1 (Anderson et al. 1975) and 1·62:1 (Hendriks et al. 1983). The number of lymphocytes that migrated across the HEC layer increased linearly beyond 60 min incubation. However, there was a 50% reduction in the total number of lymphocytes bound to HEC between 60 and 120 min. Before these observations can be explained fully further work is required to delineate the mechanisms involved in the attachment and migration of lymphocytes in this in vitro model. The adhesion of lymphocytes to cultured HEC was directly proportional to the concentration of lymphocytes added and lymphocyte adhesion was not saturated even at concentrations of 10⁷ lymphocytes ml^-1, which is two to three times that of normal rat blood.

High endothelial cells line post-capillary venules and are therefore a type of microvascular endothelium. The demonstrated lack of vWF : Ag localization in SPCV in comparison with its presence in other micro-vascular EC of rat LN suggests that the precise relationship between these two types of microvascular endothelia is not straightforward, at least with respect to their role in haemostasis. Since rat high endothelial cells do not contain vWF : Ag, which is the most commonly used marker of vascular endothelial cells, synthesis of the unique sulphated macromolecule by these cells proved to be the only method readily available for identifying isolated high endothelial cells.

There is a dearth of information about the role of high endothelium in normal lymphocyte extravasation. This demonstration of the reproducible isolation and culture of a population of high endothelial cells that continue to express surface determinants for lymphocytes in vitro merits further investigation.

I thank Dr Eric Bell and Dr Mark Drayson for reviewing the manuscript. I gratefully acknowledge the expert technical assistance of Tamar Aslan and Kathy Pinkney, the photographic skills of Jane Crosby and the secretarial assistance of Shirley John. This work was supported by Medical Research Council (UK) grant G972/455B.