EXPERIMENTAL
CELL
RESEARCH
192,
319-323
(19%)
SHORT NOTE Type I Collagen Fibrils Promote Rapid Vascular Tube Formation upon Contact with the Apical Side of Cultured Endothelium CHRISTOPHER
J. JACKSON'ANDKATHRYNL.JENKINS
Sutton Rheumatism Research Laboratory, Professorial Department Royal North Shore Hospital, St Leonards, New South Wales
Extracellular matrix (ECM) components phenotypitally modulate cultured endothelium. This paper examined the ability of ECM components to promote tube formation in uitro. When collagen type I was added to the culture medium of confluent neonatal foreskin or human umbilical vein endothelium at lo-100 Kg/ml tube-like structures formed rapidly. Tube formation did not occur with the addition of other ECM components at similar concentrations. Electron microscopy revealed that the lumen of the tubes consisted of collagen fibrils, with the surrounding cells having typical endothelial junctional complexes. These findings demonstrate that when collagen fibrils make contact with the apical side of endothelium they act as a stimulus and c 1991 provide a template for vascular tube formation. Academic
Press,
Inc.
INTRODUCTION Angiogenesis, or the formation of new blood vessels, occurs in normal conditions such as embryogenesis and wound healing as well as in diseases such as tumors, diabetic retinopathy, psoriasis, and rheumatoid arthritis [l]. Folkman has described the series of events that occur when a new blood vessel forms [a]. These include endothelial cell protease production, migration and proliferation, tubule formation, anastomoses, and basement membrane incorporation. However, the stimulus or mechanism of angiogenesis in either normal or pathological conditions is incompletely understood. The interaction of the endothelium and the extracellular matrix (ECM) is important in the process of angiogenesis. There have been many reports showing the effects of ECM components on the proliferation, migration, and phenotypic organization of endothelium [Xi]. ECM components can stimulate a confluent endothelial monolayer to form vessel-like tubes. Kubota et al. (61
i To whom
reprint,
requests
should
be addressed.
of Rheumatology, 2065, Australia
showed that Matrigel, a basement membrane preparation, stimulates endothelial cell tubule formation in uitro. Laminin, a major component of Matrigel, is at least partly responsible for this process [6, 71. Collagens also play an important role in controlling angiogenesis. In uivo, the inhibition of basement membrane collagen biosynthesis prevents angiogenesis [8]. In uitro, collagen promotes endothelial cell tube formation. Madri et al. [3] observed that when capillary endothelium was grown on interstitial collagens for 2-4 days it proliferated and formed a monolayer, but when grown on basement membrane collagens it formed tube-like structures. Montesano et al. [9] have shown that capillary endothelial cells reorganize into a network of branching and anastomosing capillary-like tubes when sandwiched between two layers of type I collagen gel. In this paper we show that acid-soluble type I collagen forms fibrils when added directly to culture medium and that these fibrils attach to the apical side of an endothelial cell monolayer and induce rapid tube formation. METHODS Cell culture. Human umbilical vein endothelium (HUVE) was kindly provided by T. To (Sutton Laboratory, RNSH). Neonatal foreskin (FS) microvascular endothelium was isolated by the method of Jackson et al. [lo]. HUVE was cultured in medium 199 containing 20% bovine F-N serum (Starrate, Bethrunga, Australia) plus 50 pg/ ml endothelial cell growth supplement (ECGS) and 50 @g/ml heparin (Sigma, St Louis, MO). ECGS, donated by T. To, was prepared by the method of Maciag et al. [ll]. FS endothelium was grown in BM 86Wissler medium (Boehringer Mannhein, GmbH, West Germany) supplemented with 40% bovine F-N serum, 100 pg/ml ECGS, and 100 pg/ml heparin. Cells were cultured in a humidified incubator at 37°C with 95% air&% CO,. All experiments were perCoating dishes with ECM components. formed in 35-mm tissue culture dishes (Lux. Flow Laboratories). Type I collagen was adsorbed to the surface by adding 1 ml of 100 pg/ml type I collagen (rat tail tendon, Flow Laboratories, 74-090-49) and incubating for 3 h at room temperature followed by three washes in Hanks’ balanced salt solution (HBSS, Sigma). Type I collagen gel was prepared by adding 1 ml of a mixture of 5X Minima1 essential medium (Flow Laboratories):O.l M sodium bicarbonate buffer:type I
319
0014.4827/91
Copyright All
rights
0
1991 of reproduction
by
Academic in any
form
$3.00
Press, Inc. reserved.
320
SHORT
collagen (2.5 mg/ml):distilled H,O (2:1:5:2) at 4°C and allowing to gel at 37°C for 1 min. One milliliteroftype IV collagen (5Opg/ml) (Collaborative Research Inc., Bedford, MA) was spread evenly over the dish, allowed to air dry, and washed twice in HBSS. One milliliter of laminin (10 pg/ml) (Flow Laboratories), 0.2% (w/v) gelatin, or fibronectin (20 fig/ml) was added to each dish for 45 min then washed twice with HBSS. Treatmuzt of endothelial monolayer with ECM components. HUVE and FS endothelial cells were plated in ECM-coated dishes (3.5-4.5 x lo5 cells/dish) and incubated overnight in growth media. In some experiments basal medium (medium 199 plus 5% bovine F-N serum) was used. The next day, the cells were confuent and the monolayer was refed with 1 ml fresh medium plus 40 ~1 soluble ECM components in the presence or absence of 10 rig/ml phorbol myristate acetate (PMA). The ECM components used were type I collagen (0.252.5 mg/ml in 17.5 mM acetic acid), laminin (1 mg/ml), type IV collagen (1 mg/ml in 17.5 mM acetic acid), fibronectin (1 mg/ml), or Matrigel (Flow Laboratories). Controls were performed by adding 40 ~1 solvent without ECM components present. Tube formation was assessed using phase microscopy (Nikon Diaphot) and transmission electron microscopy. Electron microscopy. Tissue culture dishes were rinsed twice in HBSS and fixed for 2 h in Karnovsky’s fixative [12] at 22°C. The samples were washed in 0.1 M cocodylate buffer and postfixed in 1% osmium tetroxide in 0.1 M cocodylate buffer for 1 h. Samples were dehydrated through an ethanol series and embedded in low viscosity epoxy resin. Thin sections were sequentially stained in uranyl acetate and lead citrate and examined in a Philips 400 transmission electron microscope. RESULTS
Confluent monolayers of HUVE and FS endothelial cells, passage 2-5, were grown in 35-mm dishes on various extracellular matrices in growth medium. Cells grown for 24 h on type I collagen gel, adsorbed type I collagen, type IV collagen, fibronectin, laminin, and gelatin all formed cobblestone-like monolayers (Fig. lA), in the presence or absence of PMA with or without the addition of the growth factors, ECGS and heparin. When acid-soluble type I collagen was added to the growth medium, in the presence of 10 rig/ml PMA, collagen fibrils formed immediately and within 5 min, the fibrils adhered to the endothelial cells. The cells, to which the collagen fibrils had bound, rapidly altered in morphology from cobblestone to spindle-shaped and elongated along the fibril (Fig. 1B). Within 30 min, endothelial cells had aligned along the fibrils to form tubelike structures, with translucent lumina, which varied in width from 20 to 100 ym. Tube formation was completed within 2-4 h. The degree of tube formation was observed to be dependent on the concentration of type I collagen. At low concentrations (lo-25 @g/ml) short discrete tubes were present, whereas at a concentration of 100 pg/ml, an extensive anastomosing network of tubes formed over the entire dish (Fig. 1C). Cells to which the collagen fibrils had not bound were not affected and remained as a monolayer. With refeeding every 2 days, the tubes and monolayer remained intact for at least 5 days. There was no morphological difference in the response to collagen between HUVE or FS endothelium.
NOTE
In the absence of PMA, collagen-induced tube formation was slower (6-8 h) and manv tubes were wider (>200 pm) and morphologically less distinct (Fig. 2A) than in the presence of PMA. In basal medium, without the growth factors ECGS and heparin, tubes did not form in response to type I collagen. Instead, the monolayer lost cell-cell contact (Fig. 2s) and lifted off the plate after 2-4 days in culture. Also, tubes did not form in response to type I collagen when 10 rig/ml PMA was added in the absence of ECGS and heparin. These results are summarized in Table 1. No tubes formed after the addition of type IV collagen, laminin, or fibronectin to the culture medium at similar concentrations to that used for type I collagen. On the addition of Matrigel to the culture medium, discrete refractile globular deposits of gel attached to the monolayer. However, tube formation did not occur (Fig. 1D). Tube formation was not dependent on the composition of the underlying matrix on which the cells were grown as the same response was observed using type I collagen gel, adsorbed type I collagen, type IV collagen, fibronectin, gelatin, or laminin as the basal matrix. Electron microscopy was performed on the cells 24 h after the addition of collagen. Sections cut perpendicular to the monolayer verified that endothelial cells formed intact tubes which enclosed a lumen. A notable feature was the presence of collagen fibrils within the lumen of all tubes examined (Fig. 3). The cells forming the tubes had typical endothelial cell junctional complexes (Fig. 3B) and were raised above and separate from the intact cell monolayer (Fig. 3A). In some instances, an intracellular lumen, a feature noted by others [6], was evident (not shown). DISCUSSION This paper demonstrates that fibrillar type I collagen induces an endothelial cell monolayer to rapidly form tubes. In other systems, endothelial cell tubes have taken from 1 day [6,13] to many weeks [14,15] to form. However, in this study tube formation begins within 15 min and is complete within 4 h after the addition of type I collagen. Ingber and Folkman [16] have suggested that capillary formation is dependent on two processes-the action of insoluble ECM components and soluble angiogenie factors. They have shown that alterations in ECM components switch fibroblast growth factor-stimulated endothelial cells between growth and differentiation [ 131. In agreement, this study demonstrates that growth medium containing the angiogenic factors heparin and ECGS was required to trigger collagen-induced tube formation and, furthermore, the addition of PMA amplified this response. The polarity of the endothelial cells plays an essential role in the formation of tubes, since type I collagen supports monolayer growth when cells attach via their ba-
SHORT
NOTE
of 10 rig/ml FIG. 1. FS endothelium was cultured in growth medium in the presence monolayer on a gelatin-coated dish. (B) Fifteen minutes after the addj Ition of type i collagen, collagen fibrils. (C) Three hours after type 1 collagen is added, the cell: 3 have aligned along the deposits of translucent lumina. (D) Three hours after Matrigel is added, refract ile globular formed. (Magnification x145)
sal surface but stimulates tube formation when it contacts the apical aspect of the endothelial cell. These findings are in concert with those of other workers. Collagen-coated discs stimulated tube formation when exposed to the apical side of endothelium [17]. Ingber and Folkman [13] observed tube formation l-2 days after the addition of collagen-coated microcarrier beads to the apical side of endothelium. Montesano et al. [9] have shown that endothelial cells will form a confluent monolayer when plated on type I collagen. However, when another layer of collagen was placed on top of the monolayer the cells formed tubes. They speculated that lumen formation in collagen gels was the result of the formation of a new nonthrombogenic apical surface segregated from the collagen matrix [9]. In contrast, this paper shows that endothelium forms a lumen as a result of the apical side of the cell making contact with, and surrounding, the collagen fibril. All other ECM components tested were unable to phenotypically alter the endothelial monolayer. The
321
PMA. (A) Endothelial cells form a confluent the cells (indicated by arrows) line up along the collagen fibrils to form a network of tubes with gel attach to the monolayer, but no tubes are
uniqueness of type I collagen to induce tube formation may be due to two physical properties-specific receptor domains on the collagen molecule and/or its fibrillar structure. Grant et al, [7] reported two domains on laminin, an RGD-containing sequence and a YIGSR sequence, that are responsible for endothelial cell attachment and tube formation. Collagen-induced tube formation might also involve RGD-dependent attachment, since type I collagen cont.ains several RGD sequences [18]. Although a receptor for type I collagen, a&, has been identified on endothelium [ 191, it is unknown whether differences exist between receptors on the apical versus basal side of endothelium. Endothelial cells have domains on the apical side which are distinct from those on the basal side [20, 211, including differences in cell membrane glycoprotein expression [22]. Whether the expression of such glycoproteins are important in the polar actions of collagen is unknown. Many reports [9, 13-15, 231 which describe endothelial cell tube formation, in vitro, have observed amor-
322
SHORT
NOTE
FIG. 2. FS endothelium shown 6 h after the addition of type I collagen. (A) In the absence of PMA, some tubes (indicated by arrows) are wider and less clearly defined than in the presence of PMA. (B) In the absence of ECGS and heparin, no tubes form and the monolayer loses cell-cell contact. (Magnification X145)
phous and fibrillar debris retained within the lumen. Although the origin and significance of this debris is unclear, the presence of such material led Feder [23] to suggest that the tubes were “inside-out,” depositing basement membrane proteins within the lumen. Folkman and Haundenschild [15] have suggested that the debris may act as a guide for the alignment of the tubes. Ingber and Folkman [13] observed that tendrils consisting of undefined ECM composition formed the scaffolding for tube formation on rigid substrata. As the tubes matured, the tendrils were broken down which resulted in the formation of a hollow lumen-containing tube [ 151. Results from this paper clearly demonstrate that collagen fibrils can act as a template for tube formation. In fact, it is likely that a three dimensional template is a prerequisite for tube formation. Kubota et al. [6] have shown that laminin promotes in vitro tube formation only when it is present as a component of Matrigel or when added to type I collagen gel. Grant et al. [7] have shown that the YIGSR peptide sequence promotes tube formation in the presence of Matrigel, but YIGSR alone could only stimulate endothelial cells to form “ring-like TABLE Morphological Response HUVE or FS Endothelium Type I Collagen to Various under Methods) Culture
Note. ments.
This
1
of a Confluent after the Addition Culture Media
media
Growth medium Growth medium + PMA Basal medium Basal medium + PMA table
represents
structures,” not intact tubes. We were also unable to show that laminin alone could induce tube formation. Thus, it seems that a three-dimensional gel is required as a template to provide support and direction for tube formation. It has been reported that “sprouting” bovine aortic endothelial cells involved in tube formation synthesize a
Monolayer of of 100 pglml (as Described
Response Tube formation (6-8 Rapid tube formation No tubes, monolayer No tubes the results
of over
h) (2-4 h) disrupted
20 separate
experi-
FIG. 3. Transmission electron micrograph of HUVE 24 h after the addition of type I collagen (100 fig/ml) and PMA (10 rig/ml) to the growth medium. (A) Cross section of a tube formed by endothelium surrounding collagen fibrils (C). Note that the monolayer remains intact (large arrows) (Magnification ~3,100). (B) Higher magnification of regions of cell-cell contact showing the typical endothelial junctional complexes (small arrows) (Magnification X12,500).
SHORT
predominance of type I procollagen, rather than the types III and IV procollagens synthesized by monolayers of endothelial cells [24]. It is possible that type I collagen secretion by endothelium plays a major role in formation of tubes that develop from monolayer cultures over extended time periods. This may explain the observations of Folkman and Haudenschild [15] who noted tube formation after neonatal foreskin endothelium was cultured in growth-enriched medium for 20 days. The relevance of in vitro tube formation to in uiuo angiogenesis is uncertain. However, it is possible that type I collagen could play a role in some physiological situations such as wound healing. We hypothesize that when a vessel is damaged, the apical side of the endothelial cell is exposed to the interstitium making contact with fibrillar collagen. Collagen may then play a dual role in promoting new vessel formation. First, in the presence of angiogenic factors it can bind to the exposed apical surface of endothelium and act as a stimulus for tube formation. Second, the collagen fibril acts as a template to provide the structure and support for vessel formation. Other mechanisms are likely to be involved in situations where vessel damage is not evident, such as tumor angiogenesis. The authors thank Dr. Peter Garbett for helpful comments, Dr. Les Schrieber and Dr. Andrew Nethery for critically reviewing this manuscript, and Dr. M. Vesk and P. Jameison for their expertise with electron microscopy. Chris Jackson is the recipient of the Spurway Fellowship from the Arthritis Foundation of Australia and has support from the Wenkart Foundation.
323
NOTE 3.
Madri,
J. A., and Williams,
4.
Madri, Amer.
J. A., Pratt, B. M., and J. Pathol. 132, 18-27.
5.
Herbst, (1988)
T. J., McCarthy, J. B., Tsilibary, J. Cell Biol. 106, 1365-1373.
6. Kubota, (1988) 7. 8.
S. K. (1983)
J. CellBiol.
Y., Kleinman, H. K., Martin, J. Cell Biol. 107, 1589-1598.
Grant, Martin, Ingber,
97,153-165.
Yannariello-Brown,
J. (1988)
E. C., and Furcht,
L. T.
G. R., and Lawley,
T. J.
D. S., Tashiro, Ken-Ichiro., Segu-Real, B., Yamada, G. R., and Kleinman, H. K. (1989) Cell 58,933-943. D., and Folkman, J. (1988) Lab. Inuest. 59, 44-51.
9. Montesano,
R., Orci,
L., and Vassalli,
P. (1983)
Y.,
J. Cell Biol.
97,
1648-1652. 10.
Jackson, C. J., Garbett, P. K., (1990) J. Cell Sci. 96, 257-262.
11. 12.
Maciag, T., Cerundolo, J., Ilsley, S., Kelley, P. R., and Forand, R. (1979) Proc. Natl. Acad. Sci. USA 76, 5674-5678. Karnovsky, M. J. (1965) J. Cell Biol. 27, 137A.
13.
Ingber,
14.
Maciag, T., Kadish, Weinstein, R. (1982)
15.
Folkman, 551-556.
16. 17.
Ingber, D., and Folkman, Tsuji, T., and Karasek,
18.
Bernard, M. P., Myers, J. C., Chu, M., Ramirez, F., Eikenberry, E. F., and Prockop, D. J. (1983) Biochemistry 22, 1139-1145.
19.
Giltay, Borne,
20.
Jaffe, S., Oliver, P. D., Farooqui, S. M., Novak, P. L., Sorgente, N., and Kalra, V. K. (1987) Biochim. Biophys. Acta 898.37-52.
21.
Rounds,
D. E., and Folkman,
J. (1989)
J., Wilkins, J. Cell Biol.
J., and Haudenschild,
Nissen,
B., and
J. Cell Biol.
Schrieber,
109,317-330.
L., Stemerman, 94, 511-520. C. (1980)
Nature
L.
M.
B., and
(London)
J. (1989) Cell 58, 803-805. M. A. (1985) J. Microsc. 137,
288,
75-83.
J. C., Brinkman, H. J., Modderman, P. W., von-demA. E., and vanMourik, J. A. (1989) Blood 73, 12351241.
S., and Vaccaro,
C. A. (1987)
Amer.
Rev.
Respir.
Dis.
J. Cell Biol.
103,
135,725-730. 22.
Muller,
W. A., and Gimbrone,
M. A., Jr. (1986)
2389-2402. REFERENCES
23.
Feder,
J., Marasa,
J. C., and Olander,
J. V. (1983)
J. Cell Physiol.
116,1-6. 1. 2.
Folkman, Folkman,
Received Revised
J., and Klagsburn, J. (1986) Cancer
July 16, 1990 version received
August
M. (1987) Science Res. 46, 467-473.
16, 1990
235,
442-447.
24.
Cotta-Pereira, G., Sage, H., Bornstein, P., Ross, Schwartz, S. (1980) J. Cell Physiol. 102, 1833191.
R.,
and
CELL
RESEARCH
192,
319-323
(19%)
SHORT NOTE Type I Collagen Fibrils Promote Rapid Vascular Tube Formation upon Contact with the Apical Side of Cultured Endothelium CHRISTOPHER
J. JACKSON'ANDKATHRYNL.JENKINS
Sutton Rheumatism Research Laboratory, Professorial Department Royal North Shore Hospital, St Leonards, New South Wales
Extracellular matrix (ECM) components phenotypitally modulate cultured endothelium. This paper examined the ability of ECM components to promote tube formation in uitro. When collagen type I was added to the culture medium of confluent neonatal foreskin or human umbilical vein endothelium at lo-100 Kg/ml tube-like structures formed rapidly. Tube formation did not occur with the addition of other ECM components at similar concentrations. Electron microscopy revealed that the lumen of the tubes consisted of collagen fibrils, with the surrounding cells having typical endothelial junctional complexes. These findings demonstrate that when collagen fibrils make contact with the apical side of endothelium they act as a stimulus and c 1991 provide a template for vascular tube formation. Academic
Press,
Inc.
INTRODUCTION Angiogenesis, or the formation of new blood vessels, occurs in normal conditions such as embryogenesis and wound healing as well as in diseases such as tumors, diabetic retinopathy, psoriasis, and rheumatoid arthritis [l]. Folkman has described the series of events that occur when a new blood vessel forms [a]. These include endothelial cell protease production, migration and proliferation, tubule formation, anastomoses, and basement membrane incorporation. However, the stimulus or mechanism of angiogenesis in either normal or pathological conditions is incompletely understood. The interaction of the endothelium and the extracellular matrix (ECM) is important in the process of angiogenesis. There have been many reports showing the effects of ECM components on the proliferation, migration, and phenotypic organization of endothelium [Xi]. ECM components can stimulate a confluent endothelial monolayer to form vessel-like tubes. Kubota et al. (61
i To whom
reprint,
requests
should
be addressed.
of Rheumatology, 2065, Australia
showed that Matrigel, a basement membrane preparation, stimulates endothelial cell tubule formation in uitro. Laminin, a major component of Matrigel, is at least partly responsible for this process [6, 71. Collagens also play an important role in controlling angiogenesis. In uivo, the inhibition of basement membrane collagen biosynthesis prevents angiogenesis [8]. In uitro, collagen promotes endothelial cell tube formation. Madri et al. [3] observed that when capillary endothelium was grown on interstitial collagens for 2-4 days it proliferated and formed a monolayer, but when grown on basement membrane collagens it formed tube-like structures. Montesano et al. [9] have shown that capillary endothelial cells reorganize into a network of branching and anastomosing capillary-like tubes when sandwiched between two layers of type I collagen gel. In this paper we show that acid-soluble type I collagen forms fibrils when added directly to culture medium and that these fibrils attach to the apical side of an endothelial cell monolayer and induce rapid tube formation. METHODS Cell culture. Human umbilical vein endothelium (HUVE) was kindly provided by T. To (Sutton Laboratory, RNSH). Neonatal foreskin (FS) microvascular endothelium was isolated by the method of Jackson et al. [lo]. HUVE was cultured in medium 199 containing 20% bovine F-N serum (Starrate, Bethrunga, Australia) plus 50 pg/ ml endothelial cell growth supplement (ECGS) and 50 @g/ml heparin (Sigma, St Louis, MO). ECGS, donated by T. To, was prepared by the method of Maciag et al. [ll]. FS endothelium was grown in BM 86Wissler medium (Boehringer Mannhein, GmbH, West Germany) supplemented with 40% bovine F-N serum, 100 pg/ml ECGS, and 100 pg/ml heparin. Cells were cultured in a humidified incubator at 37°C with 95% air&% CO,. All experiments were perCoating dishes with ECM components. formed in 35-mm tissue culture dishes (Lux. Flow Laboratories). Type I collagen was adsorbed to the surface by adding 1 ml of 100 pg/ml type I collagen (rat tail tendon, Flow Laboratories, 74-090-49) and incubating for 3 h at room temperature followed by three washes in Hanks’ balanced salt solution (HBSS, Sigma). Type I collagen gel was prepared by adding 1 ml of a mixture of 5X Minima1 essential medium (Flow Laboratories):O.l M sodium bicarbonate buffer:type I
319
0014.4827/91
Copyright All
rights
0
1991 of reproduction
by
Academic in any
form
$3.00
Press, Inc. reserved.
320
SHORT
collagen (2.5 mg/ml):distilled H,O (2:1:5:2) at 4°C and allowing to gel at 37°C for 1 min. One milliliteroftype IV collagen (5Opg/ml) (Collaborative Research Inc., Bedford, MA) was spread evenly over the dish, allowed to air dry, and washed twice in HBSS. One milliliter of laminin (10 pg/ml) (Flow Laboratories), 0.2% (w/v) gelatin, or fibronectin (20 fig/ml) was added to each dish for 45 min then washed twice with HBSS. Treatmuzt of endothelial monolayer with ECM components. HUVE and FS endothelial cells were plated in ECM-coated dishes (3.5-4.5 x lo5 cells/dish) and incubated overnight in growth media. In some experiments basal medium (medium 199 plus 5% bovine F-N serum) was used. The next day, the cells were confuent and the monolayer was refed with 1 ml fresh medium plus 40 ~1 soluble ECM components in the presence or absence of 10 rig/ml phorbol myristate acetate (PMA). The ECM components used were type I collagen (0.252.5 mg/ml in 17.5 mM acetic acid), laminin (1 mg/ml), type IV collagen (1 mg/ml in 17.5 mM acetic acid), fibronectin (1 mg/ml), or Matrigel (Flow Laboratories). Controls were performed by adding 40 ~1 solvent without ECM components present. Tube formation was assessed using phase microscopy (Nikon Diaphot) and transmission electron microscopy. Electron microscopy. Tissue culture dishes were rinsed twice in HBSS and fixed for 2 h in Karnovsky’s fixative [12] at 22°C. The samples were washed in 0.1 M cocodylate buffer and postfixed in 1% osmium tetroxide in 0.1 M cocodylate buffer for 1 h. Samples were dehydrated through an ethanol series and embedded in low viscosity epoxy resin. Thin sections were sequentially stained in uranyl acetate and lead citrate and examined in a Philips 400 transmission electron microscope. RESULTS
Confluent monolayers of HUVE and FS endothelial cells, passage 2-5, were grown in 35-mm dishes on various extracellular matrices in growth medium. Cells grown for 24 h on type I collagen gel, adsorbed type I collagen, type IV collagen, fibronectin, laminin, and gelatin all formed cobblestone-like monolayers (Fig. lA), in the presence or absence of PMA with or without the addition of the growth factors, ECGS and heparin. When acid-soluble type I collagen was added to the growth medium, in the presence of 10 rig/ml PMA, collagen fibrils formed immediately and within 5 min, the fibrils adhered to the endothelial cells. The cells, to which the collagen fibrils had bound, rapidly altered in morphology from cobblestone to spindle-shaped and elongated along the fibril (Fig. 1B). Within 30 min, endothelial cells had aligned along the fibrils to form tubelike structures, with translucent lumina, which varied in width from 20 to 100 ym. Tube formation was completed within 2-4 h. The degree of tube formation was observed to be dependent on the concentration of type I collagen. At low concentrations (lo-25 @g/ml) short discrete tubes were present, whereas at a concentration of 100 pg/ml, an extensive anastomosing network of tubes formed over the entire dish (Fig. 1C). Cells to which the collagen fibrils had not bound were not affected and remained as a monolayer. With refeeding every 2 days, the tubes and monolayer remained intact for at least 5 days. There was no morphological difference in the response to collagen between HUVE or FS endothelium.
NOTE
In the absence of PMA, collagen-induced tube formation was slower (6-8 h) and manv tubes were wider (>200 pm) and morphologically less distinct (Fig. 2A) than in the presence of PMA. In basal medium, without the growth factors ECGS and heparin, tubes did not form in response to type I collagen. Instead, the monolayer lost cell-cell contact (Fig. 2s) and lifted off the plate after 2-4 days in culture. Also, tubes did not form in response to type I collagen when 10 rig/ml PMA was added in the absence of ECGS and heparin. These results are summarized in Table 1. No tubes formed after the addition of type IV collagen, laminin, or fibronectin to the culture medium at similar concentrations to that used for type I collagen. On the addition of Matrigel to the culture medium, discrete refractile globular deposits of gel attached to the monolayer. However, tube formation did not occur (Fig. 1D). Tube formation was not dependent on the composition of the underlying matrix on which the cells were grown as the same response was observed using type I collagen gel, adsorbed type I collagen, type IV collagen, fibronectin, gelatin, or laminin as the basal matrix. Electron microscopy was performed on the cells 24 h after the addition of collagen. Sections cut perpendicular to the monolayer verified that endothelial cells formed intact tubes which enclosed a lumen. A notable feature was the presence of collagen fibrils within the lumen of all tubes examined (Fig. 3). The cells forming the tubes had typical endothelial cell junctional complexes (Fig. 3B) and were raised above and separate from the intact cell monolayer (Fig. 3A). In some instances, an intracellular lumen, a feature noted by others [6], was evident (not shown). DISCUSSION This paper demonstrates that fibrillar type I collagen induces an endothelial cell monolayer to rapidly form tubes. In other systems, endothelial cell tubes have taken from 1 day [6,13] to many weeks [14,15] to form. However, in this study tube formation begins within 15 min and is complete within 4 h after the addition of type I collagen. Ingber and Folkman [16] have suggested that capillary formation is dependent on two processes-the action of insoluble ECM components and soluble angiogenie factors. They have shown that alterations in ECM components switch fibroblast growth factor-stimulated endothelial cells between growth and differentiation [ 131. In agreement, this study demonstrates that growth medium containing the angiogenic factors heparin and ECGS was required to trigger collagen-induced tube formation and, furthermore, the addition of PMA amplified this response. The polarity of the endothelial cells plays an essential role in the formation of tubes, since type I collagen supports monolayer growth when cells attach via their ba-
SHORT
NOTE
of 10 rig/ml FIG. 1. FS endothelium was cultured in growth medium in the presence monolayer on a gelatin-coated dish. (B) Fifteen minutes after the addj Ition of type i collagen, collagen fibrils. (C) Three hours after type 1 collagen is added, the cell: 3 have aligned along the deposits of translucent lumina. (D) Three hours after Matrigel is added, refract ile globular formed. (Magnification x145)
sal surface but stimulates tube formation when it contacts the apical aspect of the endothelial cell. These findings are in concert with those of other workers. Collagen-coated discs stimulated tube formation when exposed to the apical side of endothelium [17]. Ingber and Folkman [13] observed tube formation l-2 days after the addition of collagen-coated microcarrier beads to the apical side of endothelium. Montesano et al. [9] have shown that endothelial cells will form a confluent monolayer when plated on type I collagen. However, when another layer of collagen was placed on top of the monolayer the cells formed tubes. They speculated that lumen formation in collagen gels was the result of the formation of a new nonthrombogenic apical surface segregated from the collagen matrix [9]. In contrast, this paper shows that endothelium forms a lumen as a result of the apical side of the cell making contact with, and surrounding, the collagen fibril. All other ECM components tested were unable to phenotypically alter the endothelial monolayer. The
321
PMA. (A) Endothelial cells form a confluent the cells (indicated by arrows) line up along the collagen fibrils to form a network of tubes with gel attach to the monolayer, but no tubes are
uniqueness of type I collagen to induce tube formation may be due to two physical properties-specific receptor domains on the collagen molecule and/or its fibrillar structure. Grant et al, [7] reported two domains on laminin, an RGD-containing sequence and a YIGSR sequence, that are responsible for endothelial cell attachment and tube formation. Collagen-induced tube formation might also involve RGD-dependent attachment, since type I collagen cont.ains several RGD sequences [18]. Although a receptor for type I collagen, a&, has been identified on endothelium [ 191, it is unknown whether differences exist between receptors on the apical versus basal side of endothelium. Endothelial cells have domains on the apical side which are distinct from those on the basal side [20, 211, including differences in cell membrane glycoprotein expression [22]. Whether the expression of such glycoproteins are important in the polar actions of collagen is unknown. Many reports [9, 13-15, 231 which describe endothelial cell tube formation, in vitro, have observed amor-
322
SHORT
NOTE
FIG. 2. FS endothelium shown 6 h after the addition of type I collagen. (A) In the absence of PMA, some tubes (indicated by arrows) are wider and less clearly defined than in the presence of PMA. (B) In the absence of ECGS and heparin, no tubes form and the monolayer loses cell-cell contact. (Magnification X145)
phous and fibrillar debris retained within the lumen. Although the origin and significance of this debris is unclear, the presence of such material led Feder [23] to suggest that the tubes were “inside-out,” depositing basement membrane proteins within the lumen. Folkman and Haundenschild [15] have suggested that the debris may act as a guide for the alignment of the tubes. Ingber and Folkman [13] observed that tendrils consisting of undefined ECM composition formed the scaffolding for tube formation on rigid substrata. As the tubes matured, the tendrils were broken down which resulted in the formation of a hollow lumen-containing tube [ 151. Results from this paper clearly demonstrate that collagen fibrils can act as a template for tube formation. In fact, it is likely that a three dimensional template is a prerequisite for tube formation. Kubota et al. [6] have shown that laminin promotes in vitro tube formation only when it is present as a component of Matrigel or when added to type I collagen gel. Grant et al. [7] have shown that the YIGSR peptide sequence promotes tube formation in the presence of Matrigel, but YIGSR alone could only stimulate endothelial cells to form “ring-like TABLE Morphological Response HUVE or FS Endothelium Type I Collagen to Various under Methods) Culture
Note. ments.
This
1
of a Confluent after the Addition Culture Media
media
Growth medium Growth medium + PMA Basal medium Basal medium + PMA table
represents
structures,” not intact tubes. We were also unable to show that laminin alone could induce tube formation. Thus, it seems that a three-dimensional gel is required as a template to provide support and direction for tube formation. It has been reported that “sprouting” bovine aortic endothelial cells involved in tube formation synthesize a
Monolayer of of 100 pglml (as Described
Response Tube formation (6-8 Rapid tube formation No tubes, monolayer No tubes the results
of over
h) (2-4 h) disrupted
20 separate
experi-
FIG. 3. Transmission electron micrograph of HUVE 24 h after the addition of type I collagen (100 fig/ml) and PMA (10 rig/ml) to the growth medium. (A) Cross section of a tube formed by endothelium surrounding collagen fibrils (C). Note that the monolayer remains intact (large arrows) (Magnification ~3,100). (B) Higher magnification of regions of cell-cell contact showing the typical endothelial junctional complexes (small arrows) (Magnification X12,500).
SHORT
predominance of type I procollagen, rather than the types III and IV procollagens synthesized by monolayers of endothelial cells [24]. It is possible that type I collagen secretion by endothelium plays a major role in formation of tubes that develop from monolayer cultures over extended time periods. This may explain the observations of Folkman and Haudenschild [15] who noted tube formation after neonatal foreskin endothelium was cultured in growth-enriched medium for 20 days. The relevance of in vitro tube formation to in uiuo angiogenesis is uncertain. However, it is possible that type I collagen could play a role in some physiological situations such as wound healing. We hypothesize that when a vessel is damaged, the apical side of the endothelial cell is exposed to the interstitium making contact with fibrillar collagen. Collagen may then play a dual role in promoting new vessel formation. First, in the presence of angiogenic factors it can bind to the exposed apical surface of endothelium and act as a stimulus for tube formation. Second, the collagen fibril acts as a template to provide the structure and support for vessel formation. Other mechanisms are likely to be involved in situations where vessel damage is not evident, such as tumor angiogenesis. The authors thank Dr. Peter Garbett for helpful comments, Dr. Les Schrieber and Dr. Andrew Nethery for critically reviewing this manuscript, and Dr. M. Vesk and P. Jameison for their expertise with electron microscopy. Chris Jackson is the recipient of the Spurway Fellowship from the Arthritis Foundation of Australia and has support from the Wenkart Foundation.
323
NOTE 3.
Madri,
J. A., and Williams,
4.
Madri, Amer.
J. A., Pratt, B. M., and J. Pathol. 132, 18-27.
5.
Herbst, (1988)
T. J., McCarthy, J. B., Tsilibary, J. Cell Biol. 106, 1365-1373.
6. Kubota, (1988) 7. 8.
S. K. (1983)
J. CellBiol.
Y., Kleinman, H. K., Martin, J. Cell Biol. 107, 1589-1598.
Grant, Martin, Ingber,
97,153-165.
Yannariello-Brown,
J. (1988)
E. C., and Furcht,
L. T.
G. R., and Lawley,
T. J.
D. S., Tashiro, Ken-Ichiro., Segu-Real, B., Yamada, G. R., and Kleinman, H. K. (1989) Cell 58,933-943. D., and Folkman, J. (1988) Lab. Inuest. 59, 44-51.
9. Montesano,
R., Orci,
L., and Vassalli,
P. (1983)
Y.,
J. Cell Biol.
97,
1648-1652. 10.
Jackson, C. J., Garbett, P. K., (1990) J. Cell Sci. 96, 257-262.
11. 12.
Maciag, T., Cerundolo, J., Ilsley, S., Kelley, P. R., and Forand, R. (1979) Proc. Natl. Acad. Sci. USA 76, 5674-5678. Karnovsky, M. J. (1965) J. Cell Biol. 27, 137A.
13.
Ingber,
14.
Maciag, T., Kadish, Weinstein, R. (1982)
15.
Folkman, 551-556.
16. 17.
Ingber, D., and Folkman, Tsuji, T., and Karasek,
18.
Bernard, M. P., Myers, J. C., Chu, M., Ramirez, F., Eikenberry, E. F., and Prockop, D. J. (1983) Biochemistry 22, 1139-1145.
19.
Giltay, Borne,
20.
Jaffe, S., Oliver, P. D., Farooqui, S. M., Novak, P. L., Sorgente, N., and Kalra, V. K. (1987) Biochim. Biophys. Acta 898.37-52.
21.
Rounds,
D. E., and Folkman,
J. (1989)
J., Wilkins, J. Cell Biol.
J., and Haudenschild,
Nissen,
B., and
J. Cell Biol.
Schrieber,
109,317-330.
L., Stemerman, 94, 511-520. C. (1980)
Nature
L.
M.
B., and
(London)
J. (1989) Cell 58, 803-805. M. A. (1985) J. Microsc. 137,
288,
75-83.
J. C., Brinkman, H. J., Modderman, P. W., von-demA. E., and vanMourik, J. A. (1989) Blood 73, 12351241.
S., and Vaccaro,
C. A. (1987)
Amer.
Rev.
Respir.
Dis.
J. Cell Biol.
103,
135,725-730. 22.
Muller,
W. A., and Gimbrone,
M. A., Jr. (1986)
2389-2402. REFERENCES
23.
Feder,
J., Marasa,
J. C., and Olander,
J. V. (1983)
J. Cell Physiol.
116,1-6. 1. 2.
Folkman, Folkman,
Received Revised
J., and Klagsburn, J. (1986) Cancer
July 16, 1990 version received
August
M. (1987) Science Res. 46, 467-473.
16, 1990
235,
442-447.
24.
Cotta-Pereira, G., Sage, H., Bornstein, P., Ross, Schwartz, S. (1980) J. Cell Physiol. 102, 1833191.
R.,
and
