Organization of Hyaluronan and Versican in the Extracellular Matrix of Human Fibroblasts Treated With the Viral Mimetic Poly I:C

Abstract

We have examined structural details of hyaluronan- and versican-rich pericellular matrices in human lung fibroblasts, as well as fixation effects after treatment with the viral mimetic, poly I:C. Lateral aggregation of hyaluronan chains was promoted by acid-ethanol-formalin fixation compared with a network appearance with formalin alone. However, hyaluronidase-sensitive cable structures were seen in live cells, suggesting that they are not a fixation artifact. With all fixatives, versican and hyaluronan probes bound alternately along strands extending from the plasma membrane. However, a yellow colocalization signal required aggregation/overlap of several hyaluronan/versican strands and was more pronounced after acid-ethanol-formalin fixation. In addition to the main cell surface, hyaluronan and versican were also associated with fine actin-positive membrane protrusions, retraction fibers, and surface blebs. After wounding plus treatment with poly I:C, cells displayed larger hyaluronan coats and cable-like structures, as well as more membrane protrusions. However, treated cells did not migrate and had increased stress fibers compared with control wounded cells. Deposition of hyaluronan into cable-like structures in response to poly I:C was diminished but still apparent following actin filament disruption with cytochalasin D, suggesting that the protrusions only partially facilitate cable formation. As seen by scanning electron microscopy, the membrane protrusions may participate in poly I:C–induced binding of monocytes to hyaluronan- and versican-rich matrices. These results suggest that poly I:C–induced hyaluronan- and versican-rich cable structures are not deposited during migration, and that cellular protrusions partially contribute to hyaluronan cable formation. This manuscript contains online supplemental material at http://www.jhc.org. Please visit this article online to view these materials. (J Histochem Cytochem 57:1041–1060, 2009)

Hyaluronan-dependent pericellular matrices, or cell coats, form around various cell types (Toole 2004; Evanko et al. 2007). A particle exclusion assay is widely used to demonstrate the presence of the cell coat (Clarris and Fraser 1968; Knudson et al. 1993; Evanko et al. 1999; Rilla et al. 2008). Recent studies using live-cell staining showed that plasma membrane protrusions, rather than hyaluronan alone, are responsible for an exclusion space wider than ∼20 μm (Rilla et al. 2005,2008; Kultti et al. 2006). Earlier reports showed that pericellular hyaluronan coats were associated with thin microspikes extending from smooth muscle cells during attachment and spreading (Evanko et al. 1999), suggesting that such fine protrusions may be important in cell locomotion. In epidermal cells and adenocarcinoma cells, overexpression of HAS3-GFP (hyaluronan synthase 3) induced the formation of thin microvillus-like protrusions that extend into the exclusion space (Rilla et al. 2005; Kultti et al. 2006). The enzyme was located on the smooth cell membrane and also concentrated along the protrusions. In HAS-overexpressing cells and in non-transfected cells, the existence of the membrane protrusions depended on hyaluronan, as evidenced by their rapid retraction following hyaluronidase digestion or failure to form when hyaluronan synthesis was inhibited (Rilla et al. 2008). In epithelial cells, the microvilli were up to 20 μm in length, but were highly variable in length in fibroblasts, and it was proposed that these protrusions may be an important “organelle” for the production and scaffolding of hyaluronan-dependent cell coats (Kultti et al. 2006). Other investigators had previously shown that the components of chondrocyte pericellular matrices, including CD44, were regularly spaced along small, protruding lamellipodia (Knudson et al. 1996), but in chondrocytes, the protrusions do not appear to reach the edge of the particle exclusion space (Rilla et al. 2008). Hyaluronan production is often associated with cell migration following wounding or treatment with growth factors (Turley and Torrance 1984; Heldin et al. 1989; Savani et al. 1995; Evanko et al. 1999,2001). However, the roles of the fine protrusions in matrix formation and remodeling, cell–cell interactions, and cell migration are still unclear.

Under inflammatory conditions, such as following treatment with the viral mimetic poly I:C, cells secrete large amounts of hyaluronan in the form of cable-like structures that avidly trap monocytes (de la Motte et al. 1999,2003; Majors et al. 2003; Selbi et al. 2006a,b; Jokela et al. 2008). The cable structures contain versican, TSG-6, and inter-α-trypsin inhibitor and can span many cells, and appear to interconnect cells and nuclei over long distances (de la Motte et al. 2003; Hascall et al. 2004). However, a single extended chain of high-molecular-weight hyaluronan would be ∼20 μm long (Rilla et al. 2008). Thus, the long cable structures and their properties may be partly dependent on cross-linking of hyaluronan through various mechanisms (Day and de la Motte 2005). The cable-like structure implies that hyaluronan may serve to transmit weak tensile forces in the pericellular matrix. However, the role of the aforementioned cellular protrusions in hyaluronan cable formation and matrix remodeling is not clear. Furthermore, increased hyaluronan synthesis has been associated with increased locomotory activity (Turley and Torrance 1984; Savani et al. 1995), and blockage of hyaluronan binding to cell surface receptors can diminish migration (Evanko et al. 1999), but it is not clear whether hyaluronan cable structures made in response to poly I:C facilitate migration or if the cables are deposited as a result of migratory activity.

In addition, one study showed that hyaluronan was prone to leaching from tissue sections after formalin fixation, and suggested that a fixative containing acetic acid-formalin-ethanol was preferable over formalin alone for preserving hyaluronan for histochemical localization in tissues (Lin et al. 1997). Reports about the hyaluronan-based cable structures have mostly employed methanol as the fixative (de la Motte et al. 1999; Majors et al. 2003), which would be expected to precipitate hyaluronan, and possibly contribute to a cable-like appearance of the matrix. Thus, a side-by-side comparison of these different fixatives and their effect on pericellular matrix morphology in cultured cells is warranted. In this study, we used various methods to examine the structural details of hyaluronan- and versican-rich matrix and its relationship to cell migration with a combination of poly I:C treatment and wounding of human fibroblasts. We found that the fixation method clearly impacts the morphology of cell-associated hyaluronan, but cable structures were seen in live cultures and with all the fixatives. We also found that the formation of hyaluronan matrix and cables in response to poly I:C is associated with increased actin stress fiber formation and is not coupled to migration. However, fine actin-positive cell surface protrusions clearly contribute to the formation, anchorage, and remodeling of hyaluronan-dependent matrix.

Materials and Methods

Reagents

Streptomyces hyaluronidase was purchased from ICN Biomedical, Inc. (Aurora, OH) and poly I:C from Invivogen (San Diego, CA). Cytochalasin D was from Sigma-Aldrich (St. Louis, MO). A biotinylated hyaluronan binding protein preparation (b-HABP) from cartilage was prepared as described (Underhill et al. 1993).

Cell Culture

Human lung fibroblasts (HLFs) were derived from explants of the lung, following removal of both the pleura and parenchyma, and were a generous gift from Professor Ganesh Raghu, Divison of Pulmonary and Critical Care Medicine, University of Washington, Seattle. The cells were isolated as described previously in accordance with approval from the institution's human subjects review committee (Raghu et al. 1988). HLFs were maintained in DMEM high-glucose medium supplemented with 10% FBS (HyClone; Logan, UT)), 1 mM sodium pyruvate, 0.1 mM non-essential amino acids, 0.43 mg/ml GlutMAX-1, and penicillin-streptomycin (penicillin G sodium, 100 U/ml, and streptomycin sulfate, 0.10 mg/ml; Invitrogen Life Technologies, Carlsbad, CA) at 37C in 5% CO₂. Cells were passaged with trypsin-EDTA (0.05% trypsin and 0.53 mM tetrasodium EDTA) and were used for experiments between passages 9 and 17 after initial isolation. The monocyte cell line U937 was obtained from the American Type Culture Collection (Rockville, MD) and maintained in RPMI 1640 (Invitrogen) with 10% FBS.

Cells were seeded on glass coverslips at 3.5 × 10⁵/well in 6-well plates in 10% FBS DMEM. After 24 hr, cells were growth arrested for 48 hr in 0.1% FBS DMEM, at which point the cells were ∼80% confluent. Following this period of serum deprivation, the cells tended to have very little hyaluronan on their surfaces and no cell coats by the particle exclusion assay (data not shown). Medium was then removed and cells were stimulated without or with poly I:C (20 μg/ml) in the presence of 10% FBS to stimulate the formation of hyaluronan-based cable structures (de la Motte et al. 1999) for 24 hr in fresh 10.0% FBS DMEM. Hyaluronan in the media and cell layers was measured by enzyme-linked sorbent assay as described previously (Wilkinson et al. 2004). In some experiments, cells were treated with and without 2 μg/ml cytochalasin D in the presence of poly I:C or 10% FBS alone for 24 hr prior to staining for hyaluronan and actin.

In other experiments, cells were scratch-wounded using a 200-μl pipette tip to create three approximately 1-mm-wide scratches per coverslip, just prior to stimulation with poly I:C. After 24 hr, the cells were fixed, and stained for matrix constituents or actin as described below and with 4,6-diamidino-2-phenyindole (DAPI) to visualize nuclei. Photographs were taken of 10 randomly selected 20× fields per coverslip that were centered between the wound edges. The number of cells that had migrated into the open area was measured by counting nuclei on calibrated photographs and expressed as cells/mm². The migration experiment was repeated at least three times with similar results. Significance was assessed using a Student's t-test.

Imaging of Hyaluronan, Versican, Membrane, and f-Actin

For comparing the effects of different fixation methods on the appearance of pericellular hyaluronan matrix, HLFs were fixed in either 100% methanol (de la Motte et al. 2003), 3.7% formaldehyde in PBS, pH 7.3, or acid-formalin-ethanol (3.7% formaldehyde-PBS, 70% ethanol, and 5% glacial acetic acid, all v/v) (Lin et al. 1997). Following rinsing in PBS, cells were stained for hyaluronan using a b-HABP followed by either streptavidin-Alexafluor 488 or streptavidin-Texas Red in PBS containing 1% bovine serum albumin as previously described (Evanko and Wight 1999). Versican was localized using monoclonal antibody 2B1 (Seikagaku Corp.; East Falmouth, MA), followed by Alexafluor 488–conjugated secondary antibody (Invitrogen). Phalloidin-Alexafluor 546 (Invitrogen) used to stain for f-actin was only compatible with formalin fixation, and was used at a dilution of 1:500 in PBS-1% BSA. The plasma membrane of live cells was labeled with the fixable membrane marker FM1-43FX (Molecular Probes; Invitrogen). Nuclei were stained with DAPI at 1 μg/ml in the mounting medium (Gel Mount; Biomeda, Inc., Foster City, CA). As controls, cells were predigested with hyaluronan-specific Streptomyces hyaluronidase, which abolished staining with the b-HABP (data not shown), or incubated with normal mouse IgG as a control for versican staining (not shown). Cells were examined using a Leica DMIRB microscope under epifluroescence optics using a 63 × 0.70 numerical aperture objective, and images were acquired using a Spot-cooled CCD camera and imaging program. Better resolution of structural details within the pericellular coats and the matrix deposited on the substrate was afforded by linear adjustment of brightness and contrast levels of the highlights and mid-tonal values of the entire original image using Adobe Photoshop Elements 2.0. Identical adjustments were made for each image when separate treatments were compared. Separate red/green/blue images were overlaid in Photoshop and were sharpened slightly using unsharp mask at 75%, with a radius of 5 pixels, and threshold equal to zero. Images presented are representative of at least three independent experiments.

Visualization of hyaluronan–versican matrices in live cells was done using a particle-exclusion assay (Knudson et al. 1993) as previously described (Evanko et al. 1999). In brief, cells were plated in 35-mm tissue culture dishes in DMEM containing 10% FBS. After 24 hr, 750 μl of a suspension of fixed and washed human erythrocytes (10⁸/ml) was added to the cells and allowed to settle for 15 min. Cells were examined using a Leica DMIRB microscope under phase contrast optics, and images were acquired using a Spot cooled CCD camera and imaging program. Degradation of the pericellular coats and matrix cables in live cells by digestion with Streptomyces hyaluronidase confirmed their structural dependence on hyaluronan.

Scanning Electron Microscopy

Scanning electron microscopy was done as previously described (Evanko et al. 1999). In brief, HLFs were plated on coverslips, serum-starved for 48 hr, and treated with poly I:C as described above for 24 hr. Then, to visualize monocyte binding to fibroblast pericellular matrices and cell protrusions, 100 μl of U937 cell suspension (3 × 10⁶/ml) was added to the HLF monolayers and allowed to adhere for 90 min at 4C (de la Motte et al. 1999). Plates were washed gently three times with PBS and fixed with Karnovsky's fixative in 0.1 M sodium cacodylate, 2 mmol/l CaCl, and 5% sucrose, containing 0.2% Ruthenium Red to precipitate proteoglycans, for 1 hr at 22C with or without prior digestion by Streptomyces hyaluronidase. Cells were rinsed with the cacodylate buffer containing 0.1% Ruthenium Red, and then postfixed in 1% OsO₄ in cacodylate buffer, 0.05% Ruthenium Red for 1 hr at 22C. Ruthenium Red typically causes the chondroitin sulate chains to collapse into a granule-like deposit. Coverslips were rinsed gently by dipping several times in PBS and then H₂O. Cells were air dried to maximize preservation of the hyaluronan network, and then lightly coated with gold for scanning electron microscopy.

Results

Fixation-dependent Morphology of Hyaluronan-dependent Matrix

To compare the effect of fixation on pericellular matrix morphology, normal HLFs grown in the presence of serum were fixed with either formalin in PBS, methanol, or acetic acid-formalin-ethanol, and then stained with b-HABP to localize hyaluronan. After fixation with formalin, the hyaluronan-based matrix generally tended to have a loose network-like form (Figures 1A and 1B). In digitally magnified images, individual hyaluronan strands could be distinguished primarily where they were anchored to the cell surface and in the immediately adjacent space (see also Figure 2B below). Due to the loose quality after formalin fixation, overall staining was slightly fainter, possibly owing to easier leaching of hyaluronan (Lin et al. 1997). Methanol fixation caused more precipitation and clumping of the hyaluronan matrix, which was especially noticeable in the thicker deposits (Figures 1C and 1D). Fixation with a solution containing acetic acid, ethanol and formalin resulted in a matrix that was clearly more “fibrous” in appearance and generally more intensely stained (Figures 1E and 1F). This was apparently due to a higher degree of precipitation and lateral aggregation of the hyaluronan strands. This resulted in a cobweb-like matrix in which the smaller fibers merged with larger fibers and connected several cells. This combing effect was particularly true of the hyaluronan that was tethered to the upper cell surface. In one experiment, we attempted to control the direction of flow of the acid-alcohol-formalin fixative, and many of the aggregated strands aligned in the direction of flow (data not shown). However, this was not rigorously pursued. These results suggest that the method of fixation can influence the final appearance of alignment and linearity of hyaluronan chains in the immediate pericellular matrix.

Figure 1.

Effect of fixation on pericellular matrix morphology. Hyaluronan staining of poly I:C–treated cells fixed with formalin in PBS (A,B), methanol (C,D), acetic acid-ethanol-formalin (E,F). Note the loose, wispy staining following formalin, slightly more precipitation and clumping following methanol, and pronounced lateral aggregation of hyaluronan with acid-ethanol-formalin. (B,D,F) Higher magnifications of the boxed areas of A,C,E.

Figure 2.

Versican and hyaluronan localization in the pericellular matrix and along protrusions. Cells were double stained for hyaluronan (red) and versican (green). (B,D,F) Higher magnifications of the boxed areas in A,C,E. (A,B) Control cell fixed with formalin. Note in the magnified images how the biotinylated hyaluronan binding protein preparation (b-HABP) and versican antibody bind in an alternating pattern (arrows) along individual hyaluronan strands of the pericellular matrix. Globular or random coil hyaluronan deposits (arrowheads) are also seen among the extended chains. (C,D) Poly I:C–treated cell fixed with acetic acid-ethanol-formalin. Note how a yellow colocalization signal requires lateral aggregation or overlap of several parallel hyaluronan strands, and is enhanced by this fixative. (E,F) Another poly I:C–treated cell showing a membrane protrusion (box) heavily decorated with versican. Note in F how a finer strand emanates from the tip of the protrusion (arrow). This specimen was fixed with acetic acid-ethanol-formalin.

Versican Localization in the Pericellular Matrix

Some specimens were double-labeled using a monoclonal antibody to versican (green) and b-HABP to localize hyaluronan (red) (Figure 2). The resolution of these images made it possible to see the arrangement and spacing of versican along the hyaluronan chains, as well as globular forms of hyaluronan among extended chains. Figures 2A and 2B show original and digital magnifications of the pericellular matrix in a formalin-fixed control cell. Typically, the green (versican antibody) and red (b-HABP) signals were seen in an alternating pattern along parallel strands that extended from the cell membrane 5 to 20 μm into the pericellular space. Individual red and green signals, as small as 100 nm, and as close to each other as ∼100–200 nm were resolvable after digital magnification. This size and arrangement suggests that many b-HABP moieties can bind along a single extended strand of hyaluronan, and that individual versican molecules are resolvable by fluorescence microscopy. Even though they were very closely spaced and on the same chain, the two probes did not necessarily produce a yellow signal. Larger globular deposits of hyaluronan with a diameter of ∼250 nm were often seen among the extended chains (Figure 2, arrowheads), and were seen regardless of the fixation method. However, it was not clear whether the globular deposits were part of a long chain or existed independently, or both. Versican did not appear to colocalize with these globular hyaluronan structures. These observations suggest that extended and globular forms of hyaluronan can exist within a few micrometers of each other along the edge of the cell.

Similar images of versican and hyaluronan in the pericellular matrix of a cell fixed with acid-formalin-ethanol are shown in Figures 2C and 2D. In this image, it is clear that a yellow signal that would be interpreted as “colocalization” of versican and hyaluronan only occurs when there is enough aggregation (requiring an estimated three to four coaligned chains) to allow adequate overlap of the red and green signals in the vertical plane. Therefore, acid-formalin-ethanol (Figure 2) and methanol (data not shown) tended to produce more yellow colocalization signal, whereas there tended to be less yellow signal with formalin, in spite of the extremely close proximity (100–200 nm) of the individual probes in the horizontal plane along apparent single chains.

Versican staining was also seen along cell protrusions. Figures 2E and 2F show a lateral protrusion that is heavily decorated with regular deposits of versican staining, and what appears to be a finer hyaluronan- and versican-positive strand emanating from its tip. We also noted that the spacing of individual signals for versican could vary within the matrix. In the looser networks, the spacing of green versican signals was between 400 and 800 nm. Occasionally more-compacted aggregates, where the spacing between versican signals was only 100–200 nm, were apparent as a substructure within looser networks of matrix. Although the functional significance of this is still unclear, it indicates that there can be microdomains within the matrix that potentially give rise to local differences in matrix material properties along the cell membrane, or alter the ability of the matrix to cross-link receptors.

Effect of Poly I:C on Pericellular Hyaluronan, Migration, and Filamentous Actin

Fibroblasts treated with poly I:C had more-prominent hyaluronan staining (Figures 3A and 3B) and much larger cell coats than control cells, as seen by particle exclusion assay (Figures 3C–3E). Under both conditions, the cell coats were prominent along the cell flanks, and along stretched tail portions (uropod). Fine microvillus-like protrusions, retraction fibers, membranous blebs, or lamellipodia could be seen in the exclusion space in both control and poly I:C–treated cultures, although the finest projections were difficult to see under phase contrast. Poly I:C treatment caused a slight but non-significant increase in total hyaluronan synthesis over this time (Figure 3F). However, there was a significant shift (p<0.01) in the proportion of hyaluronan in the cell layer in poly I:C–treated cultures (53% of total) compared with the control cells (5% of total).

Figure 3.

Effect of poly I:C on hyaluronan staining, pericellular coat formation, and hyaluronan synthesis. Hyaluronan staining of control (A) and poly I:C–treated (B) cells near the wound edge at 24 hr. Cells were fixed with acetic acid-ethanol-formalin. Particle exclusion assay of live control (C) and poly I:C–treated (D) cells. (E) Quantitation of coat and cell area ratios by point counting; *p<0.05 compared with control. (F) Hyaluronan synthesis measured by enzyme-linked sorbent assay in medium (black filled bars) and cell layers (open bars).

In control cultures, hyaluronan staining around migrating cells was usually a relatively thin and uniform coat, consistent with the particle exclusion assay. The control cells were characterized by a mostly long and motile morphology, and frequently had one or several long protrusions emanating from the leading edge or from the cell flanks that were coated with hyaluronan. In contrast, the poly I:C–treated cells at the wound edge had much larger amounts of hyaluronan, frequently in the form of long, cable-like deposits emanating from their surfaces. In poly I:C–treated cells, increased cable-like structures were apparent regardless of the fixation method (not shown). In fact, very long hyaluronidase-sensitive cable structures (approaching 1 mm or more in length) were occasionally visible above the cells in live cultures, due to trapped cell debris, suggesting that the cables themselves are not artifacts of fixation (Figure 4).

Figure 4.

Hyaluronan cable structure in live cells. (A) Occasional strands in poly I:C–treated cells were visible (arrows) above unfixed cells due to trapped particles of cell debris. (B) The same field as shown in (A). Five minutes after addition of Streptomyces hyaluronidase, the strand is destroyed.

In some experiments, poly I:C treatment was combined with scratch wounding, and we predicted that an increase in pericellular hyaluronan staining and cable-like hyaluronan deposits in response to poly I:C would be accompanied by increased migratory activity. However, cell migration following wounding in the presence of poly I:C was clearly inhibited, compared with cells in the presence of 10% FBS alone (Figure 5). Consistent with their decreased migratory capacity, the poly I:C–treated cells had increased actin stress fibers, compared with the control cells.

Figure 5.

Effect of poly I:C on f-actin and migration following wounding. (A,B) Phalloidin staining for f-actin of control (A) and poly I:C–treated (B) cells that were 50–100 μm in from the wound edge. Cells were fixed with formalin. (C) Number of cells that migrated into the open space; *p<0.05 compared with control.

Hyaluronan Is Associated With Actin-positive Membrane Protrusions and Smooth Cell Surfaces

Closer examination of control (Figure 6) and poly I:C–treated cells (Figure 7) revealed fine surface protrusions projecting into the pericellular space and surrounded by hyaluronan. The protrusions were labeled with the membrane dye FM1-43FX, indicating that they are surrounded by membrane (Figure 8). As noted above, the hyaluronan coating the cell surface and the fine protrusions of actively migrating control cells was usually fairly thin (2–10 μm). In many cases, hyaluronan staining along the protrusions appeared to be punctate and somewhat periodic. The membrane protrusions frequently were located on one side of a cell and scaffolded thick deposits of hyaluronan (Figures 6C, 7A, and 7B). Poly I:C–treated cells tended to have more and longer protrusions, which is consistent with increased coat size (Rilla et al. 2008). These extended from the lateral edges of the cell and also from the perinuclear region along the upper surface. Some hyaluronan-positive protrusions in poly I:C–treated cells were extremely long (up to 100 μm), whereas other protrusions were shorter (up to 20 μm) and thicker. In other cells, broad lamellipodia or spherical blebs with long, dense wads or cables of hyaluronan associated with them were seen along the stretched tail of the cells. Importantly, however, the long cable-like deposits of hyaluronan in poly I:C–treated cells also frequently arose from the surface of flattened cells that had very short or no apparent protrusions (Figures 7C and 7F), as well as from the protrusions themselves.

Figure 6.

Hyaluronan-positive membrane protrusions. Control cells were fixed with acetic acid-ethanol-formalin and stained for hyaluronan using b-HABP 24 hr after wounding. Nuclei were stained with DAPI (blue). (A) Migrating control cell with two long protrusions emanating from the leading edge and some shorter lateral protrusions with punctate hyaluronan staining (arrows). (B) Control cells with a thin coating of hyaluronan on the cell surface and surrounding the protrusions. (C) Numerous protrusions on one side of a control cell scaffold dense deposits of hyaluronan. Arrows in B,C indicate hyaluronan-positive protrusions.

Figure 7.

Poly I:C treatment induces cable structures and variably sized hyaluronan-positive protrusions. (A,B,E) Poly I:C–treated cells with numerous long, fine protrusions and associated hyaluronan staining (arrows). (C,D) Some poly I:C–treated cells had shorter, thicker microvillous protrusions (arrow). Arrowhead in C indicates a long hyaluronan cable arising from another cell. (F) Long parallel strands of hyaluronan matrix (arrowhead) can also arise from the upper surface of cells with small or no apparent protrusions. Cells were fixed with acetic acid-ethanol-formalin.

Figure 8.

Control (A,B) and poly I:C–treated (C,D) cells were incubated with the membrane marker FM1-43FX and then fixed with formalin.

Double staining for f-actin and hyaluronan in formalin-fixed cells showed that the hyaluronan-positive membrane protrusions were also actin-positive (Figure 9). Many of the protrusions were extremely slender, ranging in thickness from <100 nm–1 μm wide, and were much finer than the large stress fibers within the cell body. Digital enhancement of the brightness of the red channel was sometimes required to visualize the actin in the finest protrusions against the bright staining within the main part of the cell, suggesting that they contain very few actin filaments. Most of the individual projections were surrounded by a coat of hyaluronan 2–5 μm thick (Figure 8A, inset). The small size of the finest protrusions suggests that they may be extremely fragile and difficult to preserve by fixation unless they are adherent to the substrate, as has been noted previously (Kultti et al. 2006; Rilla et al. 2008).

Figure 9.

Membrane protrusions are actin-positive. Cells were fixed with formalin and double stained for f-actin (red) and hyaluronan (green). All panels show poly I:C–treated cells. (A) Arrows point to extremely fine actin-positive protrusions that extend from the lateral edge of a cell with a coat of hyaluronan 2–5 μm thick. Inset shows a higher magnification of boxed area and an actin-positive protrusion. (B) Fine actin- and hyaluronan-positive intercellular connections (arrow) bridge neighboring cells and are adjacent to shorter protrusions (arrowhead) that are not part of a cell–cell connection. (C) Two actin-positive protrusions (arrows) extend from a stretched cell tail or uropod. Note the dilated bulb at the tip of one of the protrusions. Bar equals ∼40 μm. (D) Bleb-like structures rich in f-actin have cable-like deposits of hyaluronan emanating from them. (E) A cable-like deposit of hyaluronan containing actin-positive protrusions and cell fragments (arrows). (F) Extremely long hyaluronan cables with no apparent actin were also seen above the cells (arrow).

In many instances, the hyaluronan- and actin-positive protrusions formed intercellular connections that bridged neighboring cells (Figure 9B). In addition to extending from the main cell body, some protrusions extended perpendicular to stretched uropods. In some cases, there was a distinct bulb or footpad at the end of a protrusion (Figure 9C). Round, actin-positive bleb-like structures also had long hyaluronan deposits (50 μm or more in length) associated with them (Figure 9D). Some of the actin-containing material associated with longer hyaluronan deposits in poly I:C–treated cultures were discernable as distinct protrusions, other times as cell fragments within the hyaluronan deposit, wherein the f-actin had formed clumps (Figure 9E). However, extremely long hyaluronan cables with no apparent actin or protrusions were also seen (Figure 9F). The longest cables spanned many cells, ranging in length from several hundred microns to a few millimeters long, and contained hyaluronan from several cells. Thus, the largest cable structures were typically much longer than the protrusions, but were occasionally anchored to the apical surface of some cells through the protrusions or blebs.

To address whether hyaluronan cable formation requires polymerized actin and membrane protrusions, control and poly I:C–treated cells were coincubated with cytochalasin D to disrupt the cytoskeleton. Cytochalasin D caused cell rounding within 1 to 2 hr. The fine protrusions were mostly retracted by this time, as well, and were only ∼10–20 μm long. Hyaluronan cable formation was generally diminished in cells treated with cytochalasin D. At 24 hr, hyaluronan staining was apparent around actin-rich remnants of the interconnections between the rounded cells in both control (data not shown) and poly I:C–treated cells (Figures 10A–10D). However, distinct cable-like deposits of hyaluronan, several hundred microns long and containing no apparent actin, were seen tethered to the upper surface of poly I:C plus cytochalasin D–treated cells. A single rounded cell with a long cable of hyaluronan is shown in Figure 10E. Actin was only present where the hyaluronan was anchored to the upper cell surface, and not in the cable itself. In contrast, control cells did not display any cables following cytoskeleton disruption. This suggests that poly I:C–induced hyaluronan cable structures can form in the absence of an intact cytoskeleton and long membrane protrusions.

Figure 10.

Actin cytoskeleton is not required for hyaluronan cable formation. Cells were treated with cytochalasin D and poly I:C together for 24 hr, fixed with formalin, and stained for actin (red) and hyaluronan (green). (A) Actin; (B) hyaluronan; (C) DAPI; (D) merge. The low-magnification image shows a group of rounded cells interconnected by a remnant network of actin- and hyaluronan-positive processes. The arrowhead points to a long hyaluronan cable with no apparent actin. No such cables were seen in control cells treated with cytochalasin D. (E) A single rounded cell with complete actin filament disruption following cytochalasin D treatment and a long hyaluronan cable emanating from its upper surface (arrowhead).

Scanning Electron Microscopy

Examination of control and poly I:C–treated cells by scanning electron microscopy revealed cell protrusions in the pericellular space with closely associated hyaluronan matrix (Figure 11). The matrix was usually in the form of small clumps or dense wads that collapsed along the protrusions and onto the coverslip. The protrusions were 40 nm wide or larger, and varied in length from a few micrometers to ∼100 μm. Some of the protrusions were smooth, whereas others were regularly studded with nodule-like structures. Some protrusions had one or several branches. Most of the protrusions emanated from the lateral edges of the cells, but others clearly projected from the upper cell surface and had dense matrix associated with them. As noted above, some of the protrusions had bulb-like dilations or footpads at their tips. Individual hyaluronan strands ∼10 nm wide emanated from nodule-like structures on the membrane of the main cell body and the slender protrusions. Ruthenium Red proteoglycan granules decorated the hyaluronan strands, and also aggregated into clumps along the protrusions (cf. Figure 11H and versican staining in Figure 2F). Digestion with Streptomyces hyaluronidase removed most of the hyaluronan strands and Ruthenium Red granules from the cell surface and protrusions, and revealed a filamentous core within the protrusions. Occasionally, several hyaluronan strands emanated from a single bleb-like structure on the cell surface. In contrast to control cells, which tended to have smoother protrusions and cell surfaces, poly I:C–treated cells tended to have predominantly studded protrusions and more of the nodular structures on the cell surface, as well as more-dense deposits of matrix.

Figure 11.

Scanning electron microscopy. Cells were stained with Ruthenium Red and then coated with gold. (A) Poly I:C–treated cell with a long (>100 μm) membrane protrusion emanating from the upper surface (arrow) and several shorter protrusions. Thick flocculent deposits of matrix are associated with protrusions and have collapsed onto the coverslip. (B) Higher magnification of the boxed region of (A) showing how the long apical protrusion joins with a lateral protrusion. Also note the nodules on the surface of the cell and studded appearance of the protrusions. (C) Boxed region from B (left), showing a few cell surface nodules, a lateral protrusion, and a bleb-like structure (arrow) with several putative hyaluronan strands emanating from it. (D) Boxed region from B (right), showing some matrix collapsed on the apical protrusion. A few individual hyaluronan strands and putative Ruthenium Red granules can be discerned (arrow). (E) Edge of another poly I:C–treated cell with studded and branched protrusions (arrow), overlying a finer protrusion (arrowhead) that connected to a distant part of the cell. (F) A control cell showing studded (large arrows) and smooth (small arrow) lateral protrusions extending into the pericellular space with collapsed flocculent matrix. (G) Cell protrusions from a control cell with dilated bulb-like structures on their tips (arrows). (H) Higher magnification of two protrusions showing individual hyaluronan strands (∼10 nm wide) emanating from dilated nodule-like structures (arrows). The strands are occasionally decorated with putative Ruthenium Red proteoglycan granules (arrowheads). (I) Digestion with Streptomyces hyaluronidase removes the fine strands and most of the granules and reveals that the protrusions have a filamentous core (arrows). Bars: A = 10 μm; B–I = 1 μm.

Monocytes Bind to Matrix Associated With Fibroblast Protrusions

Poly I:C treatment clearly increased U937 monocyte adhesion to the hyaluronan matrices and cable-like structures of fibroblasts, as has been reported previously for smooth muscle cells (de la Motte et al. 2003) (Figure 12B). However, monocytes were still capable of binding to the relatively thin hyaluronan coat of control fibroblasts, if washing procedures were not too rigorous (Figures 12A and 12D). The monocyte-adhesive matrix also stained positively for versican (Figure 12C). Scanning electron microscopy revealed that the matrix to which monocytes adhered often had one or several fine fibroblast protrusions associated with it (Figures 12D–12F). Monocytes were sometimes bound along several linearly aligned protrusions or retraction fibers that scaffolded the matrix (Figure 12E). Despite performing the assay at 4C, the monocytes had extended their own lamellipodia and filopodia, which appeared to be guided by the fibroblast protrusions and matrix (Figures 12D–12F). Monocytes also bound to the matrix associated with stretched fibroblast tails or uropods (Figures 12C and 12F).

Figure 12.

Monocyte binding to pericellular matrices of control (A,D) and poly I:C–treated (B,E,C,F) fibroblasts. (A,B) Hyaluronan is stained green and nuclei are stained blue (DAPI). These specimens were fixed with acetic acid-ethanol-formalin. (C) This specimen was fixed with formalin and stained for hyaluronan (red) and versican (green). Monocytes are bound in the matrix and along the uropod/tail of a partially rounded fibroblast. The inset in C shows a more-condensed aggregate of closely spaced punctate versican staining within a looser network (arrow). The size (∼100 nm) and spacing (∼100 nm) is consistent with each green signal representing antibody bound to one versican monomer in a compact aggregate. (D–F) Scanning electron micrographs reveal monocytes that have adhered to matrix associated with fibroblast protrusions or uropods (arrows). The monocytes have extended their own filopodia (arrowheads) into the matrix and are partially guided by the fibroblast protrusions.

Discussion

The observations presented here expand our view of the structural characteristics and fixation effects on hyaluronan- and versican-rich pericellular matrix of fibroblasts. Recent studies have shown that hyaluronan coats of various cell types have fine microvillus-like membrane protrusions of variable length that extend into the particle exclusion space (Rilla et al. 2005,2008; Kultti et al. 2006). The protrusions were proposed to be hyaluronan-synthesizing organelles. We add to this the finding that the membrane protrusions of human fibroblasts contain actin, and are also decorated with versican. As seen by scanning electron microscopy, the protrusions are studded with nodule-like structures, from which the hyaluronan strands emanate, and these protrusions are decorated with various densities of putative proteoglycan granules. An earlier study showed cell coats around filopodia that extended from smooth muscle cells during cell spreading and hyaluronan emanating perpendicular to the membrane of these processes (Evanko et al. 1999).

Previous studies have described the formation of hyaluronan cable structures under various conditions that appear somewhat distinct from cell coats (de la Motte et al. 1999,2003; Majors et al. 2003; Selbi et al. 2006a,b; Jokela et al. 2008). Others have shown possible hyaluronan structures in vivo, such as brain tissue (Baier et al. 2007). However, to date, other than by length, there is still no clear operational definition of cables compared with cell coats, inasmuch as both appear capable of binding to monocytes (Figure 12). Several types of cross-linking mechanisms that help to stabilize the hyaluronan matrix have been described (Day and de la Motte 2005), but it was not clear from earlier studies whether formation of hyaluronan cable structures involves cellular protrusions or cell migration.

Hyaluronan- and Actin-positive Membrane Protrusions

In addition to increased hyaluronan cables, poly I:C–treated fibroblasts tended to have more actin-containing membrane protrusions that stained positively for hyaluronan and versican and that emanated from the upper and lateral cell surfaces. Some protrusions were 100 μm or so in length and extremely slender (∼40 nm). Thus, they can be easily overlooked, especially by phase contrast, and for visualization require special live-cell techniques using GFP-HAS expression or membrane dyes (Rilla et al. 2008), actin staining with digital magnification, or scanning electron microscopy (present study). Owing to their length and increased surface area for hyaluronan synthesis, the protrusions may act as a hyaluronan “spinning” organelle. The protrusions, cell coats, and cables were often eccentrically located on one side of the cell, raising the possibility they have some role in regulating aspects of cell polarity. Spatially, however, the actin-positive protrusions showed only partial association with the longest hyaluronan deposits. In addition, we observed hyaluronan cables that contained no actin or that emanated from cells with very short or no apparent protrusions (Figure 7F). Combined with incomplete abrogation of cable formation after disruption of the actin cytoskeleton, this suggests that the long protrusions partly contribute to hyaluronan cable deposition and remodeling, but may not be required. Overexpression of HAS3 in epithelial cells resulted in the formation of microvillous protrusions (Kultti et al. 2006). The integrity of the protrusions depended on hyaluronan, because they rapidly retracted following hyaluronidase digestion, unless they were adherent to the substratum, and did not form in the presence of a hyaluronan synthase inhibitor (Rilla et al. 2008). Thus, increased hyaluronan synthesis appears to precede the formation of the protrusions.

In a recent study looking at “membrane tethers” pulled by touching cells with an atomic force microscope probe, the number and strength of the membrane tethers was diminished by hyaluronidase digestion (Sun et al. 2005), suggesting that the probe actually may have been pulling on the membrane via hyaluronan chains. It is not clear how or if such artificially produced membrane tethers relate to the fine protrusions described here, but they may be similar. In addition, one study showed how application of exogenous hyaluronan to the cell edge with a micropipette induced lamellipodial outgrowth in that location (Oliferenko et al. 2000). Perhaps a hyaluronan chain or chains and associated versican, extending from a surface nodule containing HAS enzyme, or a hyaluronan receptor linked to the cytoskeleton, provides an initiating locus for the growth of a protrusion. Extension and retraction of the protrusions probably contributes to remodeling of the matrix.

Migration and Poly I:C Treatment

After wounding, control fibroblasts tended to have relatively thin, uniform hyaluronan coats and migrated efficiently. In the presence of poly I:C, cell coats were larger, and cell surface hyaluronan staining and monocyte binding to the hyaluronan matrix were increased. This is consistent with previous studies showing hyaluronan cable formation and increased hyaluronan-dependent monocyte binding in smooth muscle cells treated with poly I:C (de la Motte et al. 2003). However, the poly I:C–treated fibroblasts did not migrate and displayed increased actin stress fibers, indicating that long hyaluronan deposits formed in response to poly I:C treatment do not form as a result of fibroblast migration, and may not facilitate migration. This is consistent with a recent study showing that poly I:C promotes myofibroblast differentiation and α–smooth muscle actin expression (Sugiura et al. 2008). The diminished migration was somewhat unexpected, inasmuch as increased hyaluronan has been associated with increased cell motility in other conditions (Turley and Torrance 1984; Savani et al. 1995; Evanko et al. 1999). Other studies have shown that increased cable formation in HAS2-overexpressing renal proximal tubular epithelial cells was accompanied by increased migration, as well as increased coat size, by particle exclusion (Selbi et al. 2006a). This suggests that there may be differences among cell types in how they use the cable structures, or that there is an independent effect of poly I:C on migration compared with HAS overexpression. Other factors, such as proteoglycan concentration and degree of cross-linking of the matrix, may affect its utility in promoting migration. Our results suggest that neither the actin-positive protrusions nor the hyaluronan cables were sufficient to overcome the inhibitory effects of poly I:C on cell migration. On the other hand, as shown in Figure 12, the membrane protrusions and associated matrix may serve to guide the extension of filopodia by bound monocytes, thus facilitating the establishment of cell–cell contacts. A “motor-clutch” mechanism was recently proposed that links actin to the extracellular matrix and influences the ability of cells to generate tension and extend filopodia (Chan and Odde 2008). By influencing viscoelastic properties, hyaluronan and versican along actin-containing protrusions and retraction fibers probably play an important role at the frictional slippage interface of such a clutch mechanism.

Pericellular Matrix Structure and Fixation Effects

Our morphological data are in general agreement with previously published models of the pericellular matrix in which an aggregating proteoglycan is required to maintain an extended conformation of the hyaluronan chains (Toole 2004). The predominant aggregating proteoglycan made by fibroblasts is versican (LeBaron et al. 1992). We have shown that versican is located along the strands of hyaluronan that extend from the cell surface and decorates many of the protrusions. In the simple cell coats (i.e., where no protrusions were evident), these strands were 5–20 μm in length, consistent with single extended hyaluronan chains. We also noted ∼250-nm globular or random coil-like deposits of hyaluronan among the extended chains with no apparent associated versican. The presence of versican on the extended chains apparently does not necessarily preclude binding of the b-HABP, even to sites that are within ∼100–200 nm of each other, because we typically saw alternating red and green signals very closely juxtaposed, and clearly along the same extended hyaluronan strand (Figure 2). This perhaps is not surprising, inasmuch as the HABP preparation contains link protein [in addition to aggrecan fragments (Underhill et al. 1993)], which would be expected to form a ternary complex with versican and hyaluronan. Studies with cartilage aggregates indicated that there were 100 saccharide units (23–49 nm) between aggrecan/link protein complexes along the hyaluronan strand, so there is potentially room for another link protein or aggrecan G1 domain to bind (Faltz et al. 1979a,b). In spite of this, our results demonstrate that individual red and green signals of hyaluronan and versican probes can be extremely close in proximity (within 200 nm) along the same chain in the horizontal plane, without necessarily producing a yellow signal. In other words, the lack of a yellow signal may sometimes fool us into thinking versican and hyaluronan are not colocalized. This observation indicates that caution should be used when interpreting such images. A strong yellow signal required the overlap of several hyaluronan- and versican-positive strands in the vertical plane, and depended on the fixation method, the degree of lateral aggregation of the chains, and the amount and thickness of the matrix.

Fixation-dependent differences in the fine structure of the pericellular matrix were apparent, and more precipitation of hyaluronan was seen with the alcoholic fixatives. With formalin alone, overall hyaluronan staining was slightly less intense; a looser amorphous network structure and generally less colocalization of versican and hyaluronan were seen. In contrast, acid-formalin-ethanol caused pronounced lateral aggregation of the hyaluronan chains, a cobweb-like morphology of the matrix, and more-pronounced yellow signals in the matrix. This is probably the basis for the better preservation and insolubility of hyaluronan reported in tissue sections using acid-formalin-ethanol, whereas hyaluronan was more easily leached from skin sections fixed with formalin alone (Lin et al. 1997). Thus, for overall preservation of hyaluronan in cell cultures, acid-formalin-ethanol would seem to be the fixative of choice. However, phalloidin was not compatible with the alcoholic fixatives. We speculate that the generally loose appearance of the pericellular matrix and cables following formalin fixation is a truer representation of the native structure, and that the lateral aggregation and combed appearance resulting from acid-formalin-ethanol may be more artifactual. Despite this, long cable-like deposits were also observed in formalin-fixed specimens (Figure 9). In addition, we observed hyaluronan cable structures in live cultures that are visible due to trapped cell debris and sensitive to hyaluronidase (Figure 4). Therefore, the cables themselves are not artifacts of fixation, but can be clearly influenced by fixation. Staining of cable structures in live cells might help shed more light on this issue. The hyaluronan emanating from the upper surface of cultured cells is passively subjected to swirling and eddy currents in the large volume of medium above, but it is not clear how much this can influence aggregation of long strands of hyaluronan compared with active cellular remodeling processes or cross-linking mechanisms (Day and de la Motte 2005).

Scanning electron microscopy revealed that the surface of fibroblasts and the protrusions were frequently studded with nodule-like structures, and that individual hyaluronan strands and clumps of putative proteoglycan granules were often attached at these locations. This is consistent with the frequently punctate and regularly spaced fluorescent staining with b-HABP and versican antibody along the protrusions. Poly I:C–treated cells tended to have studded protrusions and more nodules on their surfaces. Previous studies showed that overexpressed GFP-tagged HAS3 localized along and at the tip of the microvilli in epithelial cells (Kultti et al. 2006). It is possible, therefore, that the nodular structures seen here by scanning electron microscopy represent one or more of the hyaluronan synthases or hyaluronan receptors. More-careful studies using colloidal gold and antibodies to HAS isoforms, CD44, or RHAMM will answer this question more definitively. Clearly, more work will be required to fully define the functional significance of membrane protrusions and pericellular hyaluronan cables.