Stimulation of corneal differentiation by interaction between cell surface and extracellular matrix: II. Further studies on the nature and site of transfilter “induction”

Corneal epithelial differentiation (primary stroma production) is dependent on the underlying extracellular matrix (ECM), for if the developing epithelium is enzymatically removed from the embryo, it fails to produce stroma in vitro unless it is cultured on collagenous ECM. We have previously shown that the stimulatory effect is mediated across Nucleopore filters in direct proportion to the surface area created by epithelial cell processes traversing the filter to contact ECM. Since collagenous ECM is insoluble under physiological conditions, transfilter stimulation of stroma production is probably due to an interaction of the epithelial cell surface with “inducer” ECM (killed lens capsule or purified collagen). We grew 5-day-old corneal epithelia on Nucleopore filters atop [³H]proline-labeled lens capsules and used both autoradiography and scintillation counting to show that radioactive collagen does not enter the epithelial cells in detectable amounts. We also show here that the stimulatory effect of collagen on collagen synthesis is not dependent on trapping of serum or binding of conditioned medium factors by ECM. Finally, we demonstrate that the stimulatory effect is reduced by removal of transfilter ECM after 6–12 hr in vitro. By 18–24 hr, however, cultured epithelium is less dependent on the substratum, probably because it has produced its own ECM. We conclude that: (1) the contact mediated collagen-cell surface interaction under study here requires the continuous presence of collagen in vivo and in vitro for maintenance of “stimulated” epithelial stroma synthesis; (2) the collagenous “inducer” interacts directly with epithelium rather than indirectly via trapped intermediates; (3) collagen acts at the epithelial cell surface without entering the cells.     

Control of corneal differentiation by extracellular materials. Collagen as a promoter and stabilizer of epithelial stroma production

The primary stroma of the cornea of the chick embryo consists of orthogonally arranged collagen fibrils embedded in glycosaminoglycan (GAG) produced by the epithelium under the early inductive influence of the lens. The experiments reported here were designed to test whether or not the collagen of the lens basement lamina is capable of stimulating corneal epithelium to produce primary stroma. Enzymatically isolated 5-day-old corneal epithelia were grown for 24 hr in vitro in the presence of ³⁵SO₄ or proline-³H on various substrata. Epithelia cultured on lens capsule synthesized 2.5 times as much GAG (as measured by incorporation of label into CPC precipitable material) and almost 3 times as much collagen (assayed by hot TCA extraction or collagenase sensitivity) as when cultured on Millipore filter or other noncollagenous substrata. A similar stimulatory response was observed when epithelium was combined with chemically pure chondrosarcoma collagen, NaOH-extracted lens capsule, vitreous humor, frozen-killed corneal stroma or cartilage, or tendon collagen gels; in the latter case, the magnitude of the effect can be shown to be related to concentration of the collagen in the gel. All of the collagenous substrata stimulate not only extracellular matrix production, but also polymerization of corneal-type matrix, as judged by ultrastructural criteria and by the association of more radioactivity with the tissue than the medium. Since purified chondrosarcoma collagen is as effective as lens capsule, the stimulatory effect on collagen and GAG synthesis by corneal epithelium is not specific for basal lamina (lens capsule) collagen.     

Response of basal epithelial cell surface and Cytoskeleton to solubilized extracellular matrix molecules

Corneal epithelium removed from underlying extracellular matrix (ECM) extends numerous cytoplasmic processes (blebs) from the formerly smooth basal surface. If blebbing epithelia are grown on collagen gels or lens capsules in vitro, the basal surface flattens and takes on the smooth contour typical of epithelium in contact with basal lamina in situ. This study examines the effect of soluble extracellular matrix components on the basal surface. Corneal epithelia from 9- to 11-d-old chick embryos were isolated with trypsin-collagenase or ethylenediamine tetraacetic acid, then placed on Millipore filters (Millipore Corp., Bedford, Mass.), and cultured at the medium-air interface. Media were prepared with no serum, with 10% of calf serum, or with serum from which plasma fibronectin was removed. Epithelia grown on filters in this medium continue to bleb for the duration of the experiments (12-14 h). If soluble collagen, laminin, or fibronectin is added to the medium, however, blebs are withdrawn and by 2-6 h the basal surface is flat. Epithelia grown on filters in the presence of albumin, IgG, or glycosaminoglycans continue to bleb. Epithelia cultured on solid substrata, such as glass, also continue to bleb if ECM is absent from the medium. The basal cell cortex in situ contains a compact cortical mat of filaments that decorate with S-1 myosin subfragments; some, if not all, of these filaments point away from the plasmalemma. The actin filaments disperse into the cytoplasmic processes during blebbing and now many appear to point toward the plasmalemma. In isolated epithelia that flatten in response to soluble collagens, laminin, and fibronectin, the actin filaments reform the basal cortical mat typical or epithelial in situ. Thus, extracellular macromolecules influence and organize not only the basal cell surface but also the actin-rich basal cell cortex of epithelial cells.     

Collagen synthesis by long-lived mRNA in embryonic chicken lens

Lens capsule collagen synthesis by epithelial and fiber cells was examined by immunoprecipitation and collagenase digestion in embryonic and posthatch chicken eye lens. Epithelial cells and lens fibers in the process of terminal differentiation produce alpha 1 and alpha 2 type IV collagen chains. At 6 days of embryonic development in addition to the alpha 1 (IV) and alpha 2 (IV) collagen chains, lens cells produce high molecular weight collagenase-sensitive proteins not immunologically related to type IV collagen. Lens capsule collagen components have been identified in central and outer fibers isolated from 18-day embryos and from 10-day posthatch chicken eyes. At these stages, fibers which have an increasing number of picnotic nuclei still show collagen synthesis due to long-lived mRNA. Analysis of collagen synthesis by lens cells incubated with actinomycin D suggests that stabilization of collagen mRNA occurs in lens fiber cells and to a lesser extent in epithelial cells as early as 6 days of embryonic development.     

Basement Membrane Formation in Transfilter Tooth Culture and Its Relation to Odontoblast Differentiation

The mesenchymal cells of the developing tooth differentiate into odontoblasts as a result of an epithelio-mesenchymal interaction. Odontoblast differentiation was studied in vitro by cultivating dental mesenchyme and epithelium with interposed filters. Separation of the two components by enzyme treatment resulted in removal of the basement membrane. When the epithelium was grown alone, or transfilter from killed lens capsule, the basement membrane was not restored. Transfilter cultivation with dental mesenchyme resulted in basement membrane formation, but only if the filter pores allowed penetration of cytoplasmic processes. Hence, a close association between the epithelial and the mesenchymal cells seems to be a prerequisite for the restoration of the basement membrane. Differentiation of odontoblasts took place only in explants in which a basement membrane was formed. Differentiation did not occur when contact of the mesenchymal cells with the basement membrane was prevented by small pore size filters. Further experiments demonstrating an intact basement membrane suggested that membrane contacts between the epithelial and the mesenchymal cells are not needed for odontoblast differentiation. Hence, we suggest that differentiation of odontoblasts is triggered via contact of the mesenchymal cells with the basement membrane.     

Synthesis of sulfated glycosaminoglycans by embryonic corneal epithelium

The primary corneal stroma is produced by the overlying epithelium. The endothelium appears between 4 and 5 days, fibroblasts at 6 days, and at 12 days the epithelium stratifies. We investigated the synthesis of glycosaminoglycan (GAG) by the epithelium during this developmentally significant period. The sulfated GAG synthesized by isolated 4–6-day-old corneal epithelia during the first 24 hr in vitro are entirely accountable for as chondroitin sulfates and heparan sulfates. Nearly 50% of the total sulfated GAG synthesized by epithelia on Millipore filters is lost to the medium, but only 30–40% is lost when frozen killed lens capsule or stroma is the substratum. Retention of isotope by the tissue is correlated with visible matrix polymerization. The relative amount of heparan sulfate synthesized by the developing epithelium 24 hr in vitro decreases from about 50% of the total sulfated GAG for 4-day-old epithelium to 12% for 12-day-old epithelium. A similar decrease in heparan sulfate synthesis occurs with time in culture. The relative amount of GAG identified as chondroitin sulfate and heparan sulfate is the same when ³H-glucosamine is used to label GAG as when ³⁵SO₄ is used. We conclude that the corneal epithelium produces only sulfated polysaccharides. Since hyaluronate is synthesized by whole 5-day-old corneas, it must be the product of the endothelium.     

Interaction of embryonic corneal epithelium with exogenous collagen,laminin, and fibronectin: Role of endogenous protein synthesis

Previously, we have shown that the embryonic corneal epithelium is capable of interacting with exogenous collagen, laminin, and fibronectin in soluble form, each of which causes isolated epithelium cultured on Millipore filter to stop blebbing, reorganize the basal cytoskeleton, and flatten. Here we examine the involvement of endogenously derived extracellular matrix (ECM) molecules in the interaction of the basal epithelial cell surface with the added ECM molecules. We demonstrate here that the isolated avian corneal epithelium cultured on Millipore filter is capable of synthesizing collagens and laminin, but not fibronectin. To examine whether the epithelium is capable of interacting directly with exogenous ECM components or if there is the necessity for production of a linker molecule, epithelial protein synthesis was inhibited with cycloheximide (CHX). The blebbing epithelium in the presence of CHX was then confronted with soluble ECM molecules added to the medium under the filter; such epithelia are able to interact with, and flatten in response to, both collagen and laminin. However, such inhibited epithelia continue to bled in the presence of fibronectin. We next used l-azetidine-4-carboxylic acid (LACA) to interfere with collagen secretion. Epithelia exposed to LACA are still capable of interacting with collagen and laminin, but not fibronectin, indicating a dependence on collagen secretion. These results suggest that fibronectin requires a linker protein, probably collagen, to interact with the basal epithelial surface, whereas both collagen and laminin may interact directly with the cell surface to transform the basal cytoskeleton into the cortical mat typical of differentiating corneal epithelium in situ.     

Development of the basement membrane and formation of collagen fibrils below the placodes in the head of anuran larvae

The development of the basement membrane and collagen fibrils below placodes, including the corneal region of the ectoderm, lens epithelium, nasal plate, and auditory vesicle in anuran larvae was observed by transmission electron microscopy and compared with that in nonplacodal regions such as the epidermis, neural tube, and optic vesicle. In the corneal region the lamina densa becomes thick concomitantly with the development of the connecting apparatuses such as hemidesmosomes and anchoring fibrils. The collagen fibrils increase in number and form a multilayered structure, showing similar morphology to the connective tissues below the epidermis. These two areas, i.e., the corneal region and epidermis, possess much collagenous connective tissue below them. On the other hand, the neural tube and ophthalmic vesicle that originated from the neural tube each have a thin lamina densa and a small number of underlying collagen fibrils. The lamina densa does not thicken and the number of collagen fibrils do not significantly increase during development. These two areas possess little extracellular matrix. The nasal plate and auditory vesicle show intermediate characteristics between the epidermis-type and the neural tube-type areas. In these areas, the lamina densa becomes thick and hemidesmosomes and anchoring fibrils develop. The number of collagen fibrils increases during development, but does not show an orderly arrangement; rather, they are randomly distributed. It is thought that the difference in the arrangement of collagen fibrils in different tissues is due to differences in the extracellular matrix around the collagen fibrils. Placodal epithelia have the same origin as epidermis, but during development their morphological characteristics differ and they are not associated with the pattern of extracellular matrix with characteristics of epidermal and corneal multilayered collagen fibril areas.     

Culture on basement membrane does not reverse the phenotype of lens derived mesenchyme-like cells

Definitive epithelia suspended within type I collagen gel give rise to individual, freely migrating cells that express the mesenchymal phenotype. They become elongate in shape, invade collagenous matrices and develop abundant RER. We investigated whether mesenchyme-like cells that derive from lens epithelia retain the mesenchymal phenotype or revert to epithelial phenotype when cultured on basement membrane (BM). Mesenchyme-like cells placed on top of BM gel or lens capsule BM retain the elongate, bipolar morphology of mesenchymal cells. They migrate individually along and into the BM matrix. Mesenchyme-like cells on or in BM ultrastructurally resemble true mesenchymal cells. They extend pseudopodia and filopodia, exhibit a circumferential actin cortex, and contain well developed RER. Mesenchymal products, such as type I collagen, continue to be expressed. We conclude that the phenotype of mesenchyme-like cells derived from definitive epithelia is stable even in or on matrix known to promote the epithelial genetic program. Their behavior, thus, is similar to that of true (secondary) mesenchymal cells in the embryo.     

Localisation of laminin and fibronectin during rat lens morphogenesis

Abstract. Immunofluorescence clearly localised laminin and fibronectin in the basement membranes of ocular epithelia through all stages of rat lens differentiation. Some fibronectin is also localised around the mesodermal cells associated with the epithelia. At 10 days of embryonic development, the presumptive lens ectoderm and optic veiscle are closely associated, and the "interspace" between the two tissues contains only a few mesodermal cells. Later, as the mesoderm is excluded and the lens palcode invaginates to form the lens pit, there is a marked increase in the concentration of both laminin and fibronectin in the interspace. At about 13 days, the interspace widens, and there is fluorescence for both glycoproteins in the basement membranes of the optic cup and lens vesicle; as the lens capsule thickens, the fluorescence for laminin increases in the latter. The unlabelled peroxidase anti-peroxidase (PAP) method shows that 'blebs' and 'blisters' of basement membranes, particularly from the optic vesicle, appear to give rise to cords of fibronectin- and laminin-positive material. These cords extend into the interspace and are associated with flocculent and fibrillar material. Therefore, the glycoproteins probably combine with other extracellular matrix (ECM) constituents, e.g. collagen, to form a network of fibrils in the interspace. This network must provide good adhesion between the lens placode and the optic vesicle so that invagination is co-ordinated to form the lens pit and the optic cup, respectively. It is suggested that, in addition to providing good adhesion between the tissues, this laminin- and fibronectin-rich ECM may stimulate the formation of basal extensions and cytoplasmic processes, particularly from the lens placode, and therefore, initiate the ectoderm to form lens placode.     

The identification of extracellular matrix (ECM) binding sites on the basal surface of embryonic corneal epithelium and the effect of ECM binding on epithelial collagen production

Previously, we have shown that embryonic corneal epithelia can interact with, and respond to, soluble extracellular matrices (ECM) (laminin, collagen, and fibronectin). The basal surface of epithelia isolated free of the underlying ECM can be seen to be disrupted by numerous blebs that sprout from this formerly smooth surface. Laminin, collagen, or fibronectin added to the culture medium cause the epithelium to reorganize its cytoskeleton and flatten its basal surface. We show here that ECM molecules at concentrations that reorganize epithelial cytoskeletal morphology also increase the amount of collagen produced by the epithelial cells. However, molecules that do not reorganize basal epithelial morphology (concanavalin A, heparin, bovine serum albumin) have no effect on collagen production. We also report that fluorescently labeled laminin, collagen, and fibronectin, when added to the medium surrounding isolated corneal epithelia, bind to and flatten the basal epithelial cell surface. The binding site on the basal surface is protease sensitive and is specific for each ECM molecule. These results are compatible with the idea that the basal epithelial plasmalemma possesses a diverse population of binding sites for ECM that link cell surface matrix to the cytoskeleton, causing a dramatic cytoskeletal reorganization which in turn results in enhanced production of collagen by the cells.     

Collagen modulates cell shape and cytoskeleton of embryonic corneal and fibroma fibroblasts: Distribution of actin, α-actinin,and myosin

In the embryo, fibroblasts migrating through extracellular matrices (ECM) are generally elongate in shape, exhibiting a leading pseudopodium with filopodial extensions, and a trailing cell process. Little is known about the mechanism of movement of embryonic cells in ECM, for studies of fibroblast locomotion in the past have been largely confined to observations of flattened cells grown on planar substrata. We confirm here that embryonic avian corneal fibroblasts migrating within hydrated collagen gels in vitro have the bipolar morphology of fibroblasts in vivo, and we show for the first time that highly flattened gerbil fibroma fibroblasts, grown as cell lines on planar substrata, can also respond to hydrated collagen gels by becoming elongate in shape. We demonstrate that the collagen-mediated change in cell shape is accompanied by dramatic rearrangement of the actin, α-actinin, and myosin components of the cytoskeleton. By immunofluorescence, the stress fibers of the flattened corneal fibroblasts grown on glass are seen to stain with antiactin, anti-α-actinin, and antimyosin, as has been reported for fibroma and other fibroblasts grown on glass. Stress fibers, adhesion plaques, and ruffles do not develop when the corneal or fibroma fibroblast is grown in ECM; these features seem to be a response to strong attachment of the cell underside to a planar substratum. When the fibroblasts are grown in ECM, antimyosin staining is distributed diffusely through the cytoplasm. Antiactin and anti-α-actinin stain the microfilamentous cell cortex strongly. We suggest that locomotion of the fibroblast in ECM is accompanied by adhesion of the cell to the collagen fibrils and may involve an interaction of the myosin-rich cytosol with the actin-rich filamentous cell cortex. Interestingly, the numerous filopodia that characterize the tips of motile pseudopodia of cells in ECM are very rich in actin and α-actinin, but seem to lack myosin; if filopodia use myosin to move, the interaction must be at a distance. Soluble collagen does not convert flattened fibroblasts on planar substrata to bipolar cells. Thus, the effect of collagen on the fibroblast cytoskeleton seems to depend on the presence of collagen fibrils in a gel surrounding the cell.     

Rate of basement membrane biosynthesis as an index to angiogenesis

A method was developed for assessing collagenous protein biosynthesis from [U-14C]proline in relation to angiogenesis in the chick chorioallantoic membrane (CAM). The rate of collagenous protein biosynthesis both in vitro and in vivo was maximum between days 8 and 11 of chick embryo development. This was the stage of maximum angiogenesis as shown by morphological evaluation of the vascular density. At day 10 the rate of collagenous protein biosynthesis was 11-fold higher than that of day 15, when angiogenesis had reached a plateau. The collagenous protein formed by CAM co-elutes on SDS-agarose chromatography with the collagenous component of [3H]-acetylated-basement membrane (BM) from bovine lens capsule. 8,9-dihydroxy-7-methyl-benzo[b]quinolizinium bromide (GPA1734), which was shown previously to be a specific inhibitor of BM collagen biosynthesis, caused about 80% reduction in collagenous protein synthesis by CAM. These results indicate that most of the collagenous protein synthesized by CAM was BM collagen and this can be used as a biochemical index of angiogenesis.     

Wound healing in the cornea of the chick embryo: III. The influence of pore size of millipore filters on the migration of isolated epithelial sheets in culture

The migratory activity of epithelia isolated from the cornea and the dorsal skin of chick embryos of different ages was examined in vitro. Five types of Millipore filters differing in pore size served as models to represent degrees of unevenness of the substrate instead of the natural wound beds of the corneal stroma and the dorsal dermis. Migration of the epithelium was rapid and extensive when the pore size was below 0.8 μm, but was inhibited or stopped when the pore size reached or exceeded 0.8 μm. The effect of surface properties of the substratum on the motility of the cell membrane and thus on the movement of the cells is discussed.     

Extracellular matrix production by embryonic epithelium cultured on type IV collagen Deposition of a primary corneal stroma-like structure containing large irregular type I fibrils without type II collagen

The corneal stroma of the chick embryo is deposited in two steps. The primary stroma is laid down by the corneal epithelium and it contains type I, type II and type IX collagens. Its formation is subsequent to the presumptive epithelial cells' migration onto the lens capsule (which is rich in type IV collagen). The secondary, ultimate stroma is synthesized by fibroblasts whcih, on day 5 of development, invade the swollen primary stroma. It is composed of a matrix of thin (25 nm), regular fibrils containing type I and type V collagens.We found that a chick corneal epithelium isolated from either a 6-day or a 14-day embryo was able to produce, in vitro, stroma-containing type I collagen fibrils. However, the amount of collagen deposited and its organization were highly dependent on the substratum used. Plastic or purified bovine type I collagen substrata led to the release of very few fibrils. Purified human type IV collagen induced the production of an abundant matrix made of large irregular collagen fibrils.When compared to native corneal stroma, there were two aspects in which this matrix differed: (1) it contained only type I collagen, as shown by indirect immunofluorescence, and (2) there were numerous large, irregular fibrils of about 100 to 130 nm in diameter.In conclusion, it is suggested that purified type IV collagen substitutes, in part, for the basement membrane and allows the production of a corneal stroma-like matrix by an embryonic corneal epithelium in culture. This production is possible even with a 14-day epithelium which, in vivo, is no more involved in the synthesis of the stroma collagens. Moreover, the regulatory effect of type II collagen, previously suggested by in vivo observations, may be confirmed in this in vitro system by the appearance of large fibrils in the newly deposited stroma that are made only by type I collagen.     

Matrix-cytoskeletal interactions in the developing eye

The embryonic avian corneal epithelium in vitro responds to extracellular matrix (ECM) molecules in either soluble or polymerized form by flattening its basal surface, organizing the basal cortical actin cytoskeleton, and stepping up its production of corneal stroma twofold. Embryonic corneal epithelia, like hepatocytes and mammary gland cells, seem to contain heparan sulfate proteoglycan (HSPG) in their plasmalemma, which may interact with actin on the one hand or underlying collagen on the other. Work on the corneal epithelium suggests that, in addition to HSPG, specific glycoprotein receptors for laminin and collagen exist in the basal plasmalemma and play the critical role in actually organizing the basal epithelial cytoskeleton. As yet, uncharacterized proteins may link such receptors to actin. We suggest that ECM-dependent organization of the cytoskeleton is responsible for ECM enhancement of corneal epithelial differentiation. Cell shape and exogenous ECM also affect mesenchymal cell differentiation. In the case of the corneal fibroblast migrating in collagen gels, an actin cortex present around the elongate cell seems to interact with myosin in the cytosol to bring about pseudopodial extension. Both microtubules and actin microfilaments are involved in fibroblast elongation in collagen gels. It follows from the rules presented in this review that the mesenchymal cell surface is quite different from the epithelial cell surface in its organization. Nevertheless, epithelial cell surface-ECM interaction can be modified in the embryo at particular times to permit predesignated epithelial-mesenchymal transformations, as for example at the primitive streak. Though basal surfaces of definitive, nonmalignant epithelia adhere rather strictly to the rules of epithelium-ECM interaction and do not invade underlying ECM, the environment can be manipulated in vitro to cause these epithelia to send out pseudopodia and give rise aberrantly to mesenchymal cells in collagen gels. Further study of this phenomenon should cast light on the manner in which epithelial and mesenchymal cells organize receptors for matrix molecules on their cell surfaces and develop appropriate cytoskeletal responses to the extracellular matrix.     

Three-dimensional structure of type IV collagen in the mammalian lens capsule.

The anterior lens capsule provides a thick, easily handled model system for the study of the organization of type IV collagen, the main component of basement membranes. We have used the technique of rapid freezing, deep-etch, and rotary replication to study the three-dimensional organization of the collagen skeleton in mammalian lens capsule after a variety of extraction procedures. In all cases the collagen appeared as a densely packed three-dimensional branching network of fine microfibrils. The organization of the microfibrils appears to show some regularity, with branch points approximately 40 nm apart. Most junctions are three-way and the network forms predominantly five-sided figures. This closely resembles the collagenous network described by Yurchenco and Ruben (1987, 1988) in human amniotic basement membrane and EHS tumor matrix, but extends their findings to another system for which X-ray diffraction data are available. The three-dimensional network is discussed in terms of molecular packing of type IV collagen in light of the information available from the diffraction data.     

Hemidesmosome formation by embryonic chick corneal epithelium in vitro

This study was undertaken in order to determine whether 15-day embryonic chick corneal epithelial cells can form hemidesmosomes when cultured on a variety of substrata. It was found that hemidesmosomes were formed on gelatin films, hydrated collagen gels, lens capsule, scraped corneal stroma, matrix produced by corneal endothelial cells and untreated tissue culture plastic. Hemidesmosomes were found after 5 days in cultures produced from either dissociated epithelial cells or whole epithelial explants. Hemidesmosomes occurred both singly and in groups and their morphology varied between well-defined structures with attachment plaques, sub-basal dense plates and connections to intracellular filamentous networks, and more rudimentary forms. The presence of extracellular material was often associated with the hemidesmosomes, although it was also possible to find hemidesmosomes where this material was absent. This work suggests that, in the embryonic chick cornea, extracellular structures such as anchoring filaments and anchoring fibres often associated with mature hemidesmosomes are not essential for hemidesmosome formation.     

Lymphocyte locomotion and attachment on two-dimensional surfaces and in three-dimensional matrices

The adhesion and locomotion of mouse peripheral lymph node lymphocytes on 2-D protein- coated substrata and in 3-D matrices were compared. Lymphocytes did not adhere to, or migrate on, 2-D substrata suck as serum- or fibronectin-coated glass. They did attach to and migrate in hydrated 3-D collagen lattices. When the collagen was dehydrated to form a 2-D surface, lymphocyte attachment to it was reduced. We propose that lymphocytes, which are poorly adhesive, are able to attach to and migrate in 3-D matrices by a nonadhesive mechanism such as the extension and expansion of pseudopodia through gaps in the matrix, which could provide purchase for movement in the absence of discrete intermolecular adhesions. This was supported by studies using serum-coated micropore filters, since lymphocytes attached to and migrated into filters with pore sizes large enough (3 or 8 mum) to allow pseudopod penetration but did not attach to filters made of an identical material (cellulose esters) but of narrow pore size (0.22 or 0.45 mum). Cinematographic studies of lymphocyte locomotion in collagen gels were also consistent with the above hypothesis, since lymphocytes showed a more variable morphology than is typically seen on plane surfaces, with formation of many small pseudopodia expanded to give a marked constriction between the cell and the pseudopod. These extensions often remained fixed with respect to the environment as the lymphocyte moved away from or past them. This suggests that the pseudopodia were inserted into gaps in the gel matrix and acted as anchorage points for locomotion.