Hyaluronan is a major component of the epidermal extracellular matrix, is actively synthesized by keratinocytes and shows fast matrix turnover in the stratified epithelium. We probed the importance of hyaluronan synthesis in keratinocytes by establishing cell lines carrying the exogenous hyaluronan synthase 2 (Has2) gene in sense and antisense orientations to increase and decrease their hyaluronan synthesis, respectively. Compared with cell lines transfected with the vector only, most clones containing the Has2 sense gene migrated faster in an in vitro wounding assay, whereas Has2 antisense cells migrated more slowly. Has2 antisense clones showed delayed entry into the S phase of cell cycle following plating, smaller lamellipodia and less spreading on the substratum. The decrease of hyaluronan on the undersurface of Has2 antisense cells was associated with an increased area of adhesion plaques containing vinculin. Exogenous hyaluronan added to the keratinocyte cultures had a minor stimulatory effect on migration after wounding but did not restore the reduced migratory ability of Has2 antisense cells. Hyaluronan decasaccharides that displace receptor bound hyaluronan in keratinocytes, and Streptomyces hyaluronidase sufficient to remove most cell surface hyaluronan had little effect on cell migration. The results suggest that the dynamic synthesis of hyaluronan directed by Has2, rather than the abundance of pericellular hyaluronan, controls keratinocyte migration, a cell function vital for the repair of squamous epithelia following wounding.
Hyaluronan is a ubiquitous, high molecular weight glycosaminoglycan involved in various aspects of mammalian tissue physiology, including matrix space filling, angiogenesis, cell migration, differentiation, tumor invasion and metastasis and inflammation (for reviews, see Tammi et al., 2002; Toole et al., 2002). Hyaluronan is an important structural organizer of the extracellular matrix in connective tissues such as cartilage, but it also fills the extracellular space in the basal and spinous cell layers of human epidermis, where it has been suggested to facilitate metabolic exchange between the circulation in the dermis and the epidermis ( Tammi et al., 1988). The metabolic turnover of hyaluronan in the small extracellular space between adjacent keratinocytes is fast, with a half life of about one day in human skin organ culture ( Tammi et al., 1991) as well as in organotypic rat keratinocyte cultures ( Tammi et al., 2000). This suggests that hyaluronan also contributes to some dynamic cellular processes such as the migratory activity in moving upwards from the basal layer and in facilitating the shape change from columnar to squamous ( Tammi et al., 1988). Hyaluronan and its receptors are also involved in the response of epidermal keratinocytes to injury or irritation ( Kaya et al., 1997), including inflammation ( Tammi et al., 1994) and wound healing ( Oksala et al., 1995).
Hyaluronan is synthesized at the inner face of the plasma membrane by hyaluronan synthases (Has). The enzymes act by alternative addition of glucuronic acid and N-acetylglucosamine from their UDP-sugars to the growing hyaluronan chain, which is simultaneously extruded through the membrane into the extracellular space. Three isoenzymes have been identified in vertebrates, designated as Has1, Has2 and Has3 (for a review, see Weigel et al., 1997), each with distinct kinetic properties and product size ( Brinck and Heldin, 1999; Itano et al., 1999). The first reports on the roles of the different isoenzymes have shown the importance of Has2 in the maintenance of cartilage matrix ( Nishida et al., 1999) and ovulation ( Salustri et al., 1999). Furthermore, developing mouse embryos deficient in Has2 activity die during gestation in utero, whereas those lacking Has1 and Has3 show no major defects ( Camenisch et al., 2000). All three Has types are expressed in skin keratinocytes ( Pienimäki et al., 2001; Sugiyama et al., 1998).
Hyaluronan has long been associated with stimulated migration of cultured cells ( Schor et al., 1989), and extensive clinical data suggest that hyaluronan enhances spreading of epithelial cancers ( Anttila et al., 2000; Auvinen et al., 2000; Ropponen et al., 1998). Hyaluronan may contribute to cell migration as a structural component of the extracellular space, creating a highly hydrated, elastic matrix that may help cell movement by facilitating detachment and providing space for migration ( Tammi et al., 2002). In addition, hyaluronan probably controls the locomotion of many cell types by interacting with its receptors such as CD44 ( Bourguignon et al., 2000; Ladeda et al., 1998; Lewis et al., 2001; Thomas et al., 1992) and RHAMM/IHABP ( Akiyama et al., 2001; Assmann et al., 1999; Hofmann et al., 1998; Savani et al., 2001; Turley et al., 1991). A variety of signaling pathways have been reported to associate with RHAMM ( Hall et al., 1996; Zhang et al., 1998), and CD44 ( Bourguignon et al., 2000; Bourguignon et al., 2001; Lewis et al., 2001; Li et al., 2001; Ohta et al., 1997; Okamoto et al., 2001), which can account for the changes in migration. Still, the exact role of the hyaluronan ligand as a triggering or regulatory agent in the locomotion signaling has remained obscure. For instance, CD44 seems to have a relatively low affinity for hyaluronan, requiring clustering or oligomerization of CD44 for stable binding and implying the requirement for a size of hyaluronan sufficient to occupy multiple receptors ( Lesley et al., 2000). Signaling may thus depend on the size distribution of hyaluronan and the way hyaluronan is presented to the cell surface. Furthermore, the bulky hyaluronan may non-specifically mask or block other cell surface interactions when bound to its receptors or when it is being extruded through the plasma membrane during its synthesis.
The contribution of endogenous hyaluronan synthesis to migration has been confirmed by Has gene transfections. However, it turned out that overexpressed Has1 and Has2 enhance migration in melanoma cells ( Ichikawa et al., 1999) as does Has2 in mesothelioma cells ( Li and Heldin, 2001), whereas Has1, Has2 and particularly Has3 inhibit the migration of CHO cells ( Brinck and Heldin, 1999). Likewise, exogenous hyaluronan added in fibroblast cultures induced, inhibited or did not affect migration, depending on the tissue origin of the fibroblasts ( Andreutti et al., 1999). In keratinocytes, upregulated Has2 and hyaluronan synthesis by epidermal growth factor correlated with higher migratory activity ( Pienimäki et al., 2001). Obviously the influences on locomotion caused by increased hyaluronan and hyaluronan synthesis rate depend on the cellular background. Whether or not specific inhibition of endogenous Has expression is associated with changes in cell motility has not been studied. Further, whether or not soluble, exogenous hyaluronan surrounding the cell and that synthesized by the cell itself have similar effects on cell behaviour is also unknown.
Thus, although there is a wealth of evidence for the importance of hyaluronan synthesis in cell proliferation and migration, few details of the mode of action are available. The aim of this study was to modulate endogenous hyaluronan synthesis by upregulation and downregulation of Has2 in keratinocytes and to examine the consequences in terms of their proliferative and migratory activities, cell adhesion and morphology and to compare those to results obtained by addition of exogenous hyaluronan, removing endogenous cell surface hyaluronan and competing for hyaluronan binding to surface receptors. For this we used a non-transformed cell line that can differentiate in a manner closely resembling epidermal keratinocytes in vivo ( Tammi et al., 2000) and established clones stably transfected with constitutively active Has2 gene constructs in sense and antisense orientations. Our results demonstrate the importance of the rate of Has2-dependent hyaluronan synthesis for keratinocyte motility, spreading and adhesion in vitro.
Materials and Methods
Construction of the Has2 antisense and sense plasmids
The eukaryotic expression vector pCl-neo (5474 bp, Promega, Madison, WI) was linearized with SalI (MBI Fermentas, Vilnius, Lithuania), and a rat Has2 full-length cDNA (4172 bp, Gen Bank #AF008201) digested with SalI was ligated into the multiple cloning site of pCl-neo. After transformation, ampicillin-resistant JM109 bacterial clones were selected and their plasmids sequenced to confirm the presence of sense and antisense constructs of Has2. Plasmid DNA was prepared with the Qiagen plasmid midi preparation kit (Qiagen GmbH, Hilden, Germany).
A newborn rat epidermal keratinocyte (REK) cell line ( Baden and Kubilus, 1983) was cultured in minimum essential medium, (MEM, Life technologies Ltd, Paisley, Scotland) supplemented with 10% fetal bovine serum (FBS, HyClone, Logan, UT), 4 mM glutamine (Sigma, St. Louis, MO) and 50 μg/ml streptomycin sulfate and 50 U/ml penicillin (Sigma). Keratinocytes were passaged twice a week at a 1:5 split ratio using 0.05% trypsin (w/v), 0.02% EDTA (w/v) in phosphate-buffered saline (PBS, Reagena Ltd, Kuopio, Finland).
24 hours after plating on a 35 mm dish, ∼150,000 REKs were transfected according to the manufacturer's instructions with 3 μl FuGENE™ 6 transfection reagent (Boehringer-Mannheim, Mannheim, Germany) combined with 1μ g of plasmid DNA. The next day, the transfected cells were trypsinized, seeded on a 90 mm dish, and grown with 500 μg/ml G418 (Calbiochem-Novabiochem Corp., La Jolla, CA). The new selection medium was changed every 3-4 days until separate colonies about 0.5 cm in diameter were found. Individual colonies were trypsinized with sterile cloning cylinders, seeded into 24-well plates and grown to sufficient numbers for the experiments. The transfected genes were maintained by keeping G418 continuously in the culture medium at 250 μg/ml except during the experiments. The cell lines were designated as follows: W, wildtype; M, mock-transfection (pCl-neo without insert); A, antisense; and S, sense cell lines. The cell lines were verified to be mycoplasma negative.
Confluent cultures were trypsinized and lysed in 10 mM Tris, pH 8.0, 0.1 mM EDTA, 0.3 M Na-acetate, 1% SDS, then digested with proteinase K (Sigma) and RNAse A. DNA was extracted with phenol and chloroform, precipitated with ethanol, digested with EcoRI (MBI), electrophoresed on a 0.9% agarose gel and transferred by capillary blotting to a nylon membrane ( Sambrook et al., 1989). The membrane was probed with a radiolabeled Has2-specific cDNA probe (1200 bp), which was amplified from the human chondrosarcoma cell line (HCS 2/8) ( Takigawa et al., 1989) using human Has2-specific primers 5′-GAA ACA GCC CCA GCC AAA GAC-3′ and 5′-CTC CCC CAA CAC CTC CAA CC-3′. The intensities of the bands were measured with a CCD camera and analyzed by NIH Image® software. The number of Has2 gene construct copies in each analysis was estimated by comparing the intensity of the construct band to that of the endogenous Has2 gene.
Hyaluronan disaccharide analysis
Media (400 μl) from keratinocytes grown on 35 mm plates with 1% FBS was analyzed for secreted hyaluronan levels. Healon® (Amersham Pharmacia Biotech, Uppsala, Sweden) was used as a standard. The samples were boiled for 10 minutes to denature proteins and digested with 40 μl of proteinase K (Sigma, 600 μg/ml in 100 mM ammonium acetate, pH 6.5) for 1.5 hours at 60°C. After proteinase K inactivation by boiling for 10 minutes, 50 μl of 50% trichloroacetic acid was added and the samples centrifuged for 15 minutes at 13,000 g. Each supernatant was dialyzed overnight against water and evaporated to dryness after the addition of 0.5 nmol mannose as an internal standard. Each sample was dissolved in 100 mM ammonium acetate, pH 6.5 and digested for 3 hours at 37°C with 2 mU of Streptococcus hyaluronidase (Seikagaku Kogyo Corp., Tokyo, Japan). The samples were dried under vacuum centrifugation, and 5 μl of 0.1 M 2-aminoacridone (AMAC, Lambda Fluoreszenztechnologie GmbH, Graz, Austria) in 3:17 (v:v) acetic acid:dimethylsulfoxide, and 5 μl of 1 M NaBH3CN was added followed by incubation overnight at 37°C. The AMAC-derivatized disaccharides were stored at -20°C until electrophoresis as described previously ( Calabro et al., 2000), with the following modification: 30% PAGE gels were cast in the laboratory in 100 mM Tris-borate buffer, pH 8.9, and the same buffer was used as the running buffer. The intensities of hyaluronan disaccharide bands derived from the hyaluronan standards, internal standards and samples were digitized on a UV-light box using a CCD camera. Quantitative image processing was done with NIH-Image®.
RT-PCR with Has2 and GAPDH primers
For RT-PCR, keratinocyte RNA was isolated with the RNeasy® Mini kit (Qiagen GmbH, Hilden, Germany) and treated with DNAse. Equal quantities of the RNA, measured with a spectrophotometer, were subjected to RT-PCR reactions with the RNA PCR Core Kit (Perkin Elmer, Branchburg, NJ). Rat Has2 and GAPDH specific primers, 5′-TCG GAA CCA CAC TGT TTG GAG TG-3′ and 5′-CCA GAT GTA AGT GAC TGA TTT GTC CC-3′, and 5′-TGA TGC TGG TGC TGA GTA TG-3′ and 5′-GGT GGA AGA ATG GGA GTT GC-3′, respectively, were designed from GenBank sequences AF008201 and M17701, respectively. In the assay for Has2 expression, the primer specific for Has2 (sense) mRNA was used for reverse transcription to avoid amplification of the possible Has2 antisense transcripts. The resulting products were run on an agarose-gel and visualized by ethidium bromide fluorescence.
bHABC-staining and image analysis
Keratinocytes were seeded at ∼20,000 cells/well on eight-well chamber slides precoated for 30 minutes at 37°C with FBS (Nalge Nunc, Naperville, IL) and grown at 37°C for 48 hours. The slides were washed with 0.1 M sodium phosphate buffer, pH 7.4 (PB), fixed at room temperature for 30 minutes with 2% paraformaldehyde (v/v) and 0.5% glutaraldehyde (v/v) and washed 5×2 minutes with PB. Cells were permeabilized at room temperature with 0.3% Triton X-100 in 3% BSA and probed with 3 μg/ml of bHABC in 3% BSA overnight at 4°C. After washing with PB, the slides were incubated with avidinbiotin peroxidase (ABC standard kit, Vector Laboratories Inc., Burlingame, CA) for 1 hour, and the color was developed with 3,3′ diaminobenzidine (DAB) and H2O2 and mounted in Supermount (BioGenex, San Ramon, CA), as described previously ( Tammi et al., 2001). The specificity of the staining for hyaluronan was controlled by removing hyaluronan with Streptomyces hyaluronidase (Seikagaku Kogyo Corp., Tokyo, Japan), and the specificity of the bHABC probe was verified by pretreating it with hyaluronan oligosaccharides (average size 20 monosaccharides).
The optical density measurements were done as described before ( Tammi et al., 1998). A Leitz BK II microscope with a 16× objective with 0.45 numerical aperture (Leitz, Wetzlar, Germany) was connected to a 12-bit digital camera (Photometrics CH 200, Tucson, AZ) equipped with a KAF 1400 CCD detector (Eastman Kodak Corp., New York, NY). Camera control and image analysis were done with IPLab software (Signal Analytics Vienna, VA). Ten fields (731×841 μm) beginning from a randomly selected corner were systematically sampled along a line across each well, and area-integrated mean optical density (OD) values, including both DAB-positive and background intensities, were calculated for each whole digitized area. In addition, DAB-positive areas were estimated from binary images with a cut-off at an OD value of 0.13. On the basis of the positive area data and the sum of the pixel values that fulfilled the positivity criteria, the mean area-integrated OD values for the DAB-positive material were calculated.
For confocal analysis of hyaluronan localization, cells were fixed with 2% paraformaldehyde (v/v), permeabilized and treated with bHABC as described above, but instead of the ABC reagent, FITC-labeled avidin (1:500 dilution, 1 hour, Vector) was used as a reporter. After washing, cells were mounted in Vectashield (Vector).
For double staining of vinculin and hyaluronan, the anti-vinculin mAb (1:1000, Sigma) was added to the bHABC solution (5 μg/ml), and in the secondary step, Texas red-labeled anti-mouse secondary antibody (Vector, 1:50) and fluorescein isothiocyanate-labeled avidin (1:500) were used together. Micrographs were obtained with an Ultraview® confocal scanner (Perkin Elmer Life Sciences, Wallac-LSR, Oxford, UK) on a Nikon Eclipse TE300 microscope using a 100× oil immersion objective.
Measurement of adhesion plaques
Cells were seeded at ∼10,000 cells per well on eight-well chamber slides, fixed after 24 hours and stained for vinculin as above. Using the confocal microscope, the area of vinculin staining in a plane just above the substratum was measured to estimate the number and size of adhesion plaques. 20 randomly selected fields per cell line were recorded using a 60× oil immersion objective. The 12-bit greyscale images were linearly scaled to eight-bit and filtered with an unsharp mask (radius 6, amount 170, threshold 60) using Adobe Photoshop 5 software (Adobe Systems, San Jose, CA). Further processing was done with IPLab software (Scanalytics Inc, Fairfield, VA). Each image was duplicated, and plaques without overlying diffuse fluorescence were directly thresholded using a constant threshold value (image a). Since some of the adhesion plaques did not have a constant intensity ratio with the background owing to fluorescent structures above the focal plane, they were separated from the background with impulse filtering (matrix 5×5 pixels; each kernel has the value -1 except the central pixel, which has a value of +24; division coefficient 5; post-filter offset 140) and thresholded using a constant value (image b). Images a and b were combined digitally, and the count and areas of individual plaques were measured automatically. Structures smaller than eight pixels were excluded from analysis. Finally, the cell area was segmented with the aid of a colored overlay superimposed on the original image, and the cumulative areas of the plaques were related to cell numbers in each field.
For the immunocytochemical localization, cultures on chamber slides were fixed with 2% paraformaldehyde in PBS for 20 minutes at room temperature, washed with PBS and incubated with the OX-50 antibody at 1:100 dilution overnight at 4°C. After washes with PBS, the signal was visualized with 1:50 diluted Texas Red-labeled antimouse antibody (Vector) for 1 hour at room temperature.
For FACS analysis, cells were detached with 0.02% EDTA in PBS, blocked with 1% BSA in PBS for 10 minutes and then sequentially incubated with OX-50 (1:50 dilution), biotinylated antimouse antibody (1:200) and FITC-avidin (1:1000) for 30 minutes. Cells were fixed with 1% paraformaldehyde for 20 minutes and analyzed in a fluorescence-activated cell sorter.
Cells were seeded in 24-well culture plates at ∼60,000 cells/well. Fresh culture medium was added every day to ensure optimal growth conditions for every cell line. Cells from duplicate wells were trypsinized and counted with a hemocytometer after 4 hours to determine plating efficiency, and after 1, 2, 3, 4 and 5 days to determine the proliferation rate. The number of detached cells in media was also counted following concentration by low-speed centrifugation. Doubling times of the cells were determined at days 0-1, 1-2 and 2-3 ( Darbre and King, 1984) by calculating log2/m, in which m represents the slope of a straight line determined by two successive time points in the growth curve [the plot of log(cell number) against time].
The proliferation rates in the wounded cultures were studied using bromodeoxyuridine (BrdU) labeling and detection kit I from Roche (Roche Diagnostics, Mannheim, Germany). Cells were labeled with 10 μM BrdU for 2 hours, fixed in 70% ethanol in 50 mM glycine-HCl buffer, pH 2.0 for 20 hours ( Dorsch and Goff, 1996) and immunostained with anti-BrdU antibody and FITC-labeled secondary antibody according to the manufacturer's instructions. To visualize all nuclei, propidium iodide (1 ng/ml) was included in the primary antibody solution. The labelings were done 2, 6, 10, 16 and 22 hours after the wounding. The specimens were photographed with a 20× objective on the confocal microscope at 10 consecutive fields from the wound edge at 488 and 560 nm wavelengths. The number of BrdU-positive cells and propidium-iodide-positive nuclei were counted using the NIH Image® software.
Determination of cell cycle phase by FACS analysis
An equal number of wild-type and Has2 antisense (A22) cells were plated on a 90 mm dish. After 16 hours, cells were trypsinized, fixed with 70% ethanol for 24 hours at 4°C and treated with RNAase (0.15 mg/ml, Sigma) for 3 hours at 37°C. Cells were incubated with propidium iodide (10 μg/ml, Sigma) for 2 hours at 37°C, and DNA contents of individual cells were analysed with a fluorescence-activated cell sorter.
The transfected and control cells were seeded at ∼500,000 cells/35 mm plates and grown until confluence. A cell-free area was introduced by scraping the monolayer crosswise with a sterile 1 ml pipette tip, which cleared cells from ∼1000 μm wide lanes. The cultures were then washed with Hank's balanced salt solution (Euroclone Ltd, Pero, Italy), and fresh medium with 10% FBS was added. The effects of exogenous high molecular mass hyaluronan (Healon®, Pharmacia, Uppsala, Sweden) and of purified hyaluronan decasaccharides ( Tammi et al., 1998) on migration were studied in medium without FBS. Streptomyces hyaluronidase was present at 1 U/ml in serum-free medium during the migration experiments on cultures pretreated with 5 U/ml of the same enzyme before wounding. The areas covered by the cells were measured immediately after scraping and 24 hours later using an Olympus CK 2 inverted phase contrast microscope, a Panasonic Wv-CD 130-L video camera and NIH Image® software. The average distance the outermost cells had migrated was calculated using the formula: (√b-√a)/2, where a is the area covered by the cells at 0 hours and b is the area covered by the cells after 24 hours. The results (in pixels) were converted to micrometers.
Keratinocytes were seeded at ∼20,000 cells/well on eight-well chamber slides (Nalge), grown for 24 hours, washed with cold PB and stained with the Annexin V-FITC Apoptosis Detection Kit (R&D Systems, Minneapolis, MN) according to the manufacturer's instructions. The percentage of positive cells was counted.
Spreading analysis of individual cells
Spreading rates were determined by measuring the areas occupied by individual cells 3, 6, 9, and 24 hours after seeding at ∼20,000 cells/well on eight-well chamber slides. Five images of each cell line were captured at every time point using a CoolSNAP CCD camera (Photometrics) mounted on a Nikon Eclipse TE300 inverted microscope with a 20× DIC objective. The areas of each individual cell were measured using NIH Image® software. The number of cells measured per image was 3-20, and the total cell number measured at each time point was 21-96. For the presentation of general cell morphology, images were obtained from individual cells 24 hours after seeding at∼ 100,000 cells/well on two-well chambered cover glasses (Nunc), using a 60× immersion oil DIC objective.
The data were subjected to a one-way analysis of variance or paired t-test (StatView 512+ Software, Abacus Concepts, Berkeley, CA). Comparisons between the means of different cell lines were done using Fisher's PLSD test.
Following transfection and transfer into the selection medium, separate G418-resistant colonies were usually found within 1-2 weeks, whereas all non-transfected cells were totally killed in 5-7 days. A number of colonies (n=8-12) were randomly isolated from each plate. The presence of the Has2 gene constructs was confirmed by Southern blotting. The Has2 probe recognized the antisense gene in a fragment of about 6.9 kb and the sense gene in a 1.4 kb fragment after EcoRI digestion of the genomic DNA. The endogenous gene was found in an ∼8 kb fragment in all cell lines (data not shown). Cell lines not exhibiting the transfected gene were excluded from further analyses. The transfected Has2 gene constructs were present in copy numbers ranging from 1 to 10, as estimated by the band intensities relative to the endogenous gene. Three mock-transfected clones, six antisense clones and five sense clones containing the expected genes were saved for further analysis. All antisense cell lines showed lower mean band densities in Has2 RT-PCR whereas the sense cells lines had higher levels of Has2 mRNA compared with those of mock cells with an empty vector ( Fig. 1). The transfections thus accentuated or reduced the expression level of Has2, as expected. The REK cells also express the other two hyaluronan synthases, Has1 and Has3 ( Pienimäki et al., 2001).
Synthesis of hyaluronan
Cytochemical assays of hyaluronan attached to the REK cell layers confirmed our previous findings ( Tammi et al., 1998) of a patchy distribution on cell surfaces and variation in quantity between individual cells ( Fig. 2A). The pattern of hyaluronan distribution on cells was similar in the wild-type, mock, sense and antisense Has2 transfected cells, but the general staining intensity of the sense and antisense cell lines appeared higher and lower, respectively, than that of wild-type or mock transfected cells ( Fig. 2A). Interestingly, the cultures of Has2 sense cells contained a greater proportion of motile looking, spindle-shaped cells that were also strongly hyaluronan positive ( Fig. 2A). A set of cultures from sense and antisense cell lines were assayed for optical density using image analysis, which confirmed the difference in cell-associated hyaluronan ( Fig. 2B).
The REKs grown in monolayer cultures synthesize hyaluronan at a rate that depends on cell density, with lowdensity cultures producing more hyaluronan per cell ( Tammi et al., 2001). Fig. 2C shows an experiment where all antisense and sense cell lines were analysed at the same time and hyaluronan secretion plotted against cell density. This and other similar experiments indicated that in most sense cell lines the secretion of hyaluronan into growth medium exceeded that in the antisense cell lines. The average production of hyaluronan was very similar between the mock-transfected lines M1, M2, and M3 (15% maximum difference) and 12-25% lower than in the parental wild-type cells (data not shown). Taken together, there were two clear differences: (1) cell-associated hyaluronan ( Fig. 2A) is much greater in sense than antisense cells with Mock and Wt at an intermediate level, (2) a linear correlation of hyaluronan in medium relative to cell density ( Fig. 2C), with sense and antisense curves clearly displaced; again M2 at an intermediate level.
Cell size, morphology and spreading rate
Examination of the cell lines by inverted phase contrast microscopy suggested distinct differences in cell size, with antisense cells appearing smaller than the sense cells. However, there were no significant differences in cell volume between Has2 sense, antisense and mock cell lines as measured with FACS (data not shown), suggesting that the established cell lines most probably differ in their ability to extend on the substratum. Antisense cells produced several smaller lamellae, as compared with control cells that had a single, wide lamellipodium ( Fig. 3A). These findings prompted us to determine the spreading of the different cell lines by measuring the areas covered by individual cells 6 hours after plating ( Fig. 3B). The area covered by the mocktransfected lines was close to that of the wildtype. The average areas occupied by the sense cell lines were generally similar to those of the mock-transfected and parental controls ( Fig. 3b). In contrast, four out of the six antisense cell lines showed less spreading than any of the controls. In three of these the difference was statistically significant ( Fig. 3B) and remained so at least for 24 hours ( Fig. 3C). The result suggests that Has2 is involved and necessary in the spreading of keratinocytes. However, addition of Streptomyces hyaluronidase to the plating medium at 1 TRU/ml, a concentration that eventually removes pericellular hyaluronan ( Tammi et al., 2001), did not reduce the spreading rate in the wild type, mock (M1), and the sense (S27 and S29) cell lines (data not shown). This indicates that high molecular weight hyaluronan can be removed shortly after its synthesis without an effect on spreading, or that the cellular compartment where hyaluronan exerts its influence is shielded from the hyaluronidase in the growth medium.
Confocal analysis showed that both in mocktransfected and sense cells, hyaluronan was relatively abundant at the undersurface of the spreading cells, whereas the signal for hyaluronan was low under the antisense cell lines ( Fig. 4A). By contrast, the staining for vinculin, representing the adhesion plaques to the substratum, appeared more prominent in the antisense cell lines compared with those of the mock and especially of sense cells ( Fig. 4A). Hyaluronan was excluded from clusters of vinculin-positive adhesion plaques and vice versa; hyaluronan-rich areas were devoid of large adhesion plaques ( Fig. 4A). In general, focal adhesions were less abundant in polarized cells with wide lamellapodia and high hyaluronan expression. The total area of focal adhesions per cell area as indicated by vinculin positivity in the undersurface of the cells was significantly higher in four of the Has2 antisense cells compared with the mock-transfected cell lines, whereas none of the sense cell lines differed significantly from the mock lines ( Fig. 4B). Blocking Has2 thus allows an apparently tighter cell attachment through the vinculin-containing adhesion plaques.
The effect of the Has2 gene transfections on cell proliferation was studied in three randomly selected representatives of the sense and antisense cell lines and compared to one mock cell and the wild-type cells. Equal numbers of the cells from each line were plated and counted after different time periods in culture. The number of Has2 antisense cells did not increase at all during the first 24 hours after plating ( Fig. 5A). Consequently, their cell numbers lagged behind but did reach the same level as control and wild-type cells at confluency ( Fig. 5A).
Two of the three Has2 sense cell lines grew at the same rate as the controls, but one (S27) appeared to remain at a lower density when reaching confluency ( Fig. 5B). These observations were documented in several repeated experiments. Analysis of variance applied to the data set with all cell lines and time points indicated that the antisense lines differed significantly from all other cell types on day 1. The calculated doubling times of the antisense cell lines on day 0-1 were on average 41 hours, compared with 21-25 hours in the other cell lines. However, there were no differences between the cell lines on days 1-2 and 2-3 in this parameter, with all showing an average of 17 hours.
Cell counts after 4 hours (plating efficiency) indicated that a similar number of transfected and parental cells adhered (data not shown). The number of apoptotic cells was also similar in all cell lines, which excludes cell death as a cause of the lag in the early growth of the antisense cells. Likewise, the numbers of floating cells were not different between the cells lines (data not shown). Flow cytometric analysis was performed on one of the antisense cell lines to determine the relative DNA content shortly after plating. As deduced from the DNA content at 16 hours, the percentage of the A22 antisense cells in the cell cycle phases G0/G1, S and G2/M were 52.8%, 34.0% and 11.0%, respectively, whearas the corresponding figures in the wild-type cells were 44.4% 35.7% and 18.9%. The difference persisted but was reduced at 24 hours. These results indicated that after plating, the reduced Has2 expression in antisense cells caused a transient delay in entering the S-phase of the cell cycle.
Wounding induced the wild-type REKs to migrate from the edge of the cleared area at a rate of 7 μm/hour on average. The migration of the mock-transfected cells (M1, M2 and M2) did not markedly differ from that of the wild-type cells ( Fig. 6). Three of the Has2 overexpressing cell lines migrated significantly faster compared with mock-transfected cells, whereas migration of four cell lines with the Has2 antisense gene was significantly reduced ( Fig. 6). These differences in motility were also seen in non-wounded, non-confluent monolayer cultures where Has2 overexpressing cells appeared to have wider lamellipodia and fill the empty areas more efficiently than Has2 antisense cells.
The role of hyaluronan synthesis in the present migration model was further examined by localization of newly synthesized hyaluronan in the wounded cultures. Confluent cultures were digested with Streptomyces hyaluronidase to remove existing hyaluronan, washed, wounded and stained 8 hours later for the new hyaluronan chains emerging on the cells. In mock-transfected cultures, numerous hyaluronan-positive spots were found in cells close to the wound edge, whereas cells in non-wounded areas showed less newly synthesized hyaluronan ( Fig. 7). By contrast, antisense cells rarely showed this wound-edge induction of hyaluronan expression whereas the cell lines carrying the Has2 gene in sense orientation not only showed intensely stained cells in the wound edge but also an elevated hyaluronan staining in non-wounded areas ( Fig. 7). This suggests that hyaluronan synthesis was specifically induced in keratinocytes lining the cleared area and that this hyaluronan synthesis contributed to the migratory activity of the cells.
To exclude the chance that the observed differences in migration were related to different proliferation rates of the cells, cultures were labeled with bromodeoxyuridine 14-16 hours after wounding. The number of labeled nuclei representing cells in the cell cycle S phase were counted from a 250-300 μm wide field close to the wound edge. In wild-type cultures, 32% of the cells in this field were labeled, whereas the transfected cell lines showed slightly higher labeling indices (38.4±2.6% in three mock lines, 39.5±2.5% in five sense lines, and 40.0±2.8% in six antisense lines, means±s.e.). This indicates that cell proliferation differences in the wound edge did not contribute to the differences of migration between the sense and antisense cell lines.
The addition of highly purified hyaluronan of the same size as that newly synthesized by REKs did not significantly increase the migration rate in mock, sense and antisense cell lines ( Fig. 8A). Importantly, the reduced migration of antisense cells could not be restored by this exogenous hyaluronan. Moreover, when cell surface hyaluronan was first cleared with Streptomyces hyaluronidase treatment, and 1 U/ml of the enzyme was added to the medium for the following 24 hour migration experiments, there was no marked change in the migration of mock-transfected, Has2 antisense and Has2 sense cells ( Fig. 8B). Experiments were also done with wild-type cells in the presence of 450 μg/ml of purified hyaluronan decasaccharides. The decasaccharide-treated cultures showed 97±13% (mean±s.d., six separate experiments) of the migration in control cultures ( Fig. 8C), even though this oligosaccharide size and concentration are known to displace receptor-bound hyaluronan and about half of the total cell surface hyaluronan in confluent REK cultures ( Tammi et al., 1998). Taken together, these experiments suggest that exogenous hyaluronan has a limited effect on the migration of the REK cells and that continuous, gradual cleavage of endogenous, newly synthesized hyaluronan or blocking its binding to cell surface receptors has little influence on the Has2-modulated migratory response.
Immunocytochemical stainings for CD44 of all the cell lines showed a strong signal on plasma membranes and no apparent changes in the localization of the signal (data not shown). Flow cytometric analysis of CD44-associated fluorescence intensities in mock transfected (75±3, mean±s.e., three cell lines), antisense (56±5, six cells lines), and sense (53±4, five cell lines) cells indicated no correlation between Has2 expression level and cell surface CD44, suggesting that Has2-associated phenotypes were not caused by changes in the quantity of cell surface CD44.
The Has2 transfections in REKs indicate that even a moderate shift from the endogenously regulated expression level of this gene strongly modulates the behavior of epidermal keratinocytes, in particular, migration after wounding. The antisense experiments showed, further, that even the basal migratory activity of keratinocytes requires a sufficient level of Has2. These experiments confirm the idea emerging from other experiments with the same keratinocytes that some of the phenotypic alterations induced by epidermal growth factor (EGF), such as the lamellipodial extension and high migratory activity, are directly associated with hyaluronan synthesis ( Pienimäki et al., 2001).
Hyaluronan and keratinocyte migration
The strong control that Has2 has on keratinocyte motility in an in vitro wound healing assay shows that hyaluronan synthesis is not just coincident with the migratory activity but is required for efficient motility of these cells. It was recently shown that transfection of the Has genes into fibroblasts enhances cell motility ( Itano et al., 2002). Furthermore, hyaluronan synthesis was specifically induced at the wound edge, suggesting that Has upregulation, quite probably of Has2, is an inherent feature of the proper wound healing response. Wounding of mesothelial cell monolayers in a similar manner to the present cultures induces Has2 expression and hyaluronan synthesis in the cells that begin migration from the wound edges ( Yung et al., 2000). Human keratinocytes migrating to cover a gingival wound show elevated hyaluronan staining ( Oksala et al., 1995), and hyaluronan accumulates in wounded mouse epidermis (R.T., unpublished), indicating that this hyaluronan synthesis response is also a part of wound healing in vivo.
Hyaluronan is thought to act as a signaling ligand that influences the activity of the intracellular locomotory system through cell surface receptors. However, the present experiments show that adding soluble, purified hyaluronan of the same high molecular weight as that synthesized by the cells themselves did not markedly increase the reduced migration rate caused by the Has2 antisense gene. This indicates that exogenous hyaluronan either does not have access to the receptors used by endogenous hyaluronan or that the mechanism of Has2 action is not simple hyaluronan receptor activation by ligand binding.
Another way of probing the role of receptor-mediated signaling in migration was to add hyaluronan decasaccharides to displace endogenous high molecular weight hyaluronan bound to cell surface receptors. The oligosaccharide treatment had no influence on migration, suggesting that continuous occupation of cell surface receptors by intact high molecular weight hyaluronan or the formation of a hyaluronan coat is not necessary for keratinocyte migration in the present assay. This conclusion was supported by the fact that continuous fragmentation of the cellular hyaluronan coat by Streptomyces hyaluronidase did not inhibit the migration of keratinocytes.
Hyaluronan synthesis and cell adhesion
The adhesive properties of the keratinocytes were clearly changed in Has2 antisense cells as indicated by the area of adhesion plaques. While just a small proportion of hyaluronan lies under the cell, this may still be important in the dynamic turnover of the focal adhesions that occurs in migration. If hyaluronan synthase is inserted or activated in the plasma membrane domain under a cell, the rapidly extruded chain can create pressure that may make existing adhesions more labile or reduce the number or size of new contacts. Hyaluronidase added in the culture medium may not have free access to this domain. Electron microscope analysis shows plasma membrane distension away from the substratum at hyaluronan deposits under the cell ( Pienimäki et al., 2001), and the present data indicated that adhesion plaques are excluded in the cell underside domains occupied by hyaluronan. The view that Has2 controls cell adhesion is supported by the finding that its overexpression reduces contact inhibition of cell growth and the organization of cadherin and filamentous actin ( Itano et al., 2002).
Further, the strong influence of hyaluronan synthesized endogenously as compared with that added in the medium is probably due to the correct positioning of the endogenous molecule, whether it is involved in signaling and/or associated with cell adhesion or plasma membrane dynamics ( Oliferenko et al., 2000).
Impairment of proliferation of the antisense cells
The delay in the initiation of the proliferation cycle in antisense cells was not completely unexpected, since active hyaluronan synthesis is often correlated with rapid proliferation. However, there was no evidence in the antisense cells for a block in the G2 phase of the cell cycle, as when the receptor RHAMM is inhibited ( Mohapatra et al., 1996), or during cytokinesis, as occurs when hyaluronan synthesis is inhibited by periodate oxidized UDP-glucuronic acid ( Brecht et al., 1986). On the contrary, the cellular DNA content indicated a reduced proportion of cells with duplicated DNA, suggesting that there is a regulatory point in the G1 stage of the cell cycle that is influenced by hyaluronan synthesis. The present finding of delayed entry into the S-phase of the cell cycle in the Has2 antisense cells fits well with a recent report of increased proportion of cells in the S and G2/M phases associated with Has2 overexpression ( Itano et al., 2002). The delay in the initiation of proliferation was correlated with reduced spreading, but further experiments are needed to reveal more of the molecular mechanisms that may involve signaling through the PI-kinase ( Itano et al., 2002).
Lamellipodia extension and hyaluronan synthesis
Extension of plasma membrane into lamellipodia is the prerequisite for cell spreading and migration and involves vectorial growth bursts of actin filament bundles and arrays ( Svitkina and Borisy, 1999). The formation of lamellipodia is triggered, for instance, by the GTPase Rac, which can be activated by the interaction of the hyaluronan receptor CD44 and the guanidine nucleotide exchange factor Tiam1 ( Bourguignon et al., 2000). The expression level of CD44 correlates with migratory activity ( Peck and Isacke, 1996; Thomas et al., 1992), and blocking of CD44 reduces cell spreading and migration in some cell types ( Ichikawa et al., 1999; Ladeda et al., 1998; Oliferenko et al., 2000). We found no significant difference in CD44 levels between Has2 sense and antisense cells, suggesting that the reduced spreading in antisense cells is not due to alterations in CD44 expression. Thus, although CD44 may be essential, it apparently cannot support lamellipodia formation in keratinocytes without sufficient Has2 expression and hyaluronan synthesis. Rapidly fluctuating hyaluronan synthesis, controlled, for example, by growth factors ( Pienimäki et al., 2001), seems a good regulator of the dynamic changes of keratinocyte migration.
Expression of the transfected genes in keratinocytes
Although the sets of mock, sense and antisense clones significantly differed from each other in their behaviour, in the individual clones the magnitude of these functional properties was not correlated with the level of Has2 expression. The mechanisms of this clonal variation are not known, but may involve concurrent changes in gene regulation, in some cases preventing the penetration of the phenotype. This is likely with genes vital for survival, such as Has2 ( Camenisch et al., 2000). In the future, experiments with conditional expression of the sense and antisense regulators of Has2 might be used to avoid these problems.
The present findings suggest that keratinocytes represent a cell type in which hyaluronan synthesis is strictly regulated and that this regulated synthesis level influences major cell functions. We collected a number of individual transfected clones of keratinocytes, some with multiple sense or antisense gene copies, and all displayed a moderate change in the level of hyaluronan secretion consistent with their level of Has2 expression. This is unlike the marked increase previously noted in some of the clones from other Has2-transfected cells types ( Brinck and Heldin, 1999; Itano et al., 2002; Kosaki et al., 1999; Watanabe and Yamaguchi, 1996). Although the upper limit of hyaluronan synthesis is probably specific for keratinocytes, we believe that the lower limit was determined by cell viability. The cells carrying the antisense gene(s) grew very slowly in the clonal densities following transfection (data not shown), and it is conceivable that transfectants with more efficient antisense Has2 expression might have been excluded owing to a growth disadvantage. No previous studies on stable antisense Has transfections are currently available for comparison.
Finally, the continuous rat keratinocyte parental cell line used in this study is unique among available keratinocyte models. The cells can be cloned while maintaining their ability to undergo epidermal differentiation in the absence of a feeder layer ( Tammi et al., 2000). This property allowed selection of the stably transfected clones used in this study and demonstrates the potential of these cells for transfection with other genes that may influence keratinocyte biology. The ability of the sense and antisense Has2 clones to undergo epidermal differentiation is currently under investigation.
Special thanks for Riikka Tiihonen for technical assistance. We also thank Mikko Mättö for the expertise with FACS analysis and Csaba Fülop for helpful discussions. This work was supported by Academy of Finland, grant #40807 and #54062 (M.T.), Finnish Cancer Foundation and Finnish Cancer Institute Foundation (RT), EVO funds of Kuopio University Hospital (M.T.) and Kuopio University Biotechnology Funds (R.T., M.T.).
- Accepted June 26, 2002.
- © The Company of Biologists Limited 2002