American Journal of Pathology, Vol. 142, No. 4, April 1993Copyright American Societyfor Investigative Pathology

CD44 and Hyaluronan Expression in HumanCutaneous Scar Fibroblasts

Diana V. Messadi* and Charles N. BertolamitFrom the Section of Oral and Maxillofacial Surgery* UCLASchool of Dentistty, and Shriners Hospitalfor CrippledChildrent, Los Angeles Unit, Los Angeles, California

Fibrotic disorders of skin and other organs aretypicaly associated with an abnormal accumu-lation ofextracelular matrix. This studyfocuseson a matrix constituent, hyaluronan-which isknown to be altered infibrotic disorders ofskin-and on CD44, a ceU adhesion molecule andputa-tive receptor for hyaluronanL Tissue sampleswere obtained from biopsies of human normalskin; normal cutaneous scar, and hypertrophiccutaneous scar. After culturing, ceUs were studiedby single- and double-labeling immunohistochem-istry using the two anti-CD44 monoclonalantibod-ies, BU-52 andJI 73, and a biotinylated hyaluro-nan binding complex probe, b-HABR Certaincultures were pretreated witb Streptomyces hy-aluronidase to assess the dependency ofCD44 ex-pression on the presence of endogenous hyalu-ronan. CD44 expression, both in thepresence andthe absence ofexogenous hyaluronan, was quan-titated by radioimmunobinding assay. OveraUglycosaminoglycan synthesis and identificationof hyaluronan were accomplished by precursorincorporation assays and by quantitative celu-lose acetate electrophoresis. CD44 wasfound tobe a normal human adult fibroblastic antigenwhose expression is markedly increasedfor by-pertrophic scarfibroblasts compared with nor-mal skin fibroblasts. Although hyaluronan wasfound to be the predominant glycosaminoglycanconstituent ofthepericelular matrixfor thesefi-broblasts, CD44 attachment to the ceU surface isneither mediated by hyaluronan nor is the pres-ence of hyaluronan a prerequisitefor CD44 ex-pression. Exogenous hyaluronan induced a de-cline in measurable CD44 expressionfor normalskinfibroblasts but notfor hypertrophic scarfi-broblasts. These observations are compatiblewith current understanding ofthe way ceUs man-

age the hyaluronan economy ofthe extracelularmatrix and emphasize phenotypic heterogene-ities between fibroblasts derived from normalversus scar tissues. (Am J Pathol 1993, 142:1041-1049)

Fibrotic disorders of skin and other organs are typi-cally associated with an abnormal accumulation ofextracellular matrix.1 Increasingly, it appears that cell-ular receptors for specific matrix components can beidentified on the surface of connective tissue cellsand may participate in the recognition, binding, in-ternalization, and catabolism of certain matrix con-stituents.2 Comparing cells derived from normal ver-sus fibrotic tissues would seem a logical step inestablishing whether the existence, properties, andfunction of such receptors are implicated in matrixaccumulation disorders and whether such receptorsserve as cellular markers for different phenotypic be-haviors among fibroblastic substrains. In pursuingthis line of reasoning, we have focused on a matrixconstituent, hyaluronan, which is known to be alteredin fibrotic disorders of skin3 and on CD44, a cell ad-hesion molecule and putative receptor for hyaluro-nan.4

In this study, we establish the presence of CD44 inhuman cutaneous scar tissues and demonstrate in-creased levels in scar fibroblasts compared with nor-mal skin fibroblasts. This is the first study to demon-strate expression of CD44 on human cutaneous scarfibroblasts and to explore the relationship of CD44and hyaluronan in such cells. A marked, time-dependent decline in CD44 expression was associ-ated with preincubation of normal skin fibroblasts withhyaluronan. This may reflect a hyaluronan-induceddown-regulation of CD44 for normal skin fibroblasts

Supported by PHS grant DE 09178 (National Institutes of Health,National Institute of Dental Research) and Shriners Hospitals forCrippled Children grant 15957.Accepted for publication October 2, 1992.Address reprint requests to Dr. Diana V. Messadi, Section of

Oral and Maxillofacial Surgery, UCLA School of Dentistry, Los An-geles, CA 90024-1668.

1041

1042 CD44, and Fibroblasts,Messadi et alAJP April 1993, Vol. 142, No. 4

but not for hypertrophic scar fibroblasts-possibly in-dicating the predominance of different fibroblasticsubstrains in skin versus scar.

Materials and MethodsTissue Samples and Cell CulturingTissue samples were obtained as biopsies of nor-mal human skin, normal cutaneous scar, and hyper-trophic cutaneous scar. In all cases, tissues wereremoved surgically as a part of treatment proce-dures. The freshly biopsied tissues were washedseveral times with Hanks' buffer and then cut intosmall pieces. Portions were taken for standard his-tological examination (stained with hematoxylin andeosin), whereas others were frozen in isopentaneand cryosectioned for immunohistochemical stud-ies. Remaining tissues were used to initiate fibro-blast cultures by the methods of either Diegelmannet a15 or Botstein et al.6 Cells were subsequentlycounted in a Coulter counter and inoculated ata density of 5 x 105 cells/75 cm2 tissue flask(Corningware, Corning, NY) in Dulbecco's modifiedeagle medium supplemented with 25 mmol/L N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acidbuffer, penicillin (100 pg/ml), streptomycin (100 pg/ml, Sigma Chemical Company, St. Louis, MO) and10% fetal bovine serum. Cultures were fed twiceweekly with the same medium (i.e., Dulbecco's me-dium plus fetal bovine serum) and subcultivatedusing 0.05% trypsin + 0.53 mol/L ethylenediamine-tetraacetic acid. Previous experiments in our labo-ratory and elsewhere confirm that fibroblasts are theonly cell type present after subcultivation; neverthe-less, cultures were routinely examined with anti-vimentin and anti-fibronectin antibodies that arepositive controls for fibroblasts and assure a mono-cellular fibroblastic phenotype. Cells were utilizedduring fourth passage. Three normal skin, three nor-mal scar, and three hypertrophic scar cell lineswere used.

ImmunohistochemistryCultured fibroblasts were plated in Lab-Tek cham-ber slides (Miles Scientific, Naperville, IL) coatedwith poly-D-lysine (Sigma) to prevent cell detach-ment and loss. Each slide contains 8 distinct culturechambers, allowing staining to be performed for thevariously treated fibroblasts derived from normaland scar tissues. While still subconfluent, mediawere removed, slides were washed with Hanks'buffer, and the slides were stained using the variousmonoclonal antibodies (MAbs) and probes.

Monoclonal Antibodies and ProbeTwo anti-CD44 MAbs were used: BU-52 (The Bind-ing Site, San Diego, CA) and J173 ascites, (a gift ofDr. John Pesando, The Biomembrane Institute, Se-attle, WA). Both are mouse anti-human antibodies ofthe IgGl isotype, and both were used at a concen-tration of 1:100. Two positive control MAbs wereroutinely used: anti-vimentin (at a concentration of1:40) and anti-fibronectin (at a concentration of1:100). Both were of the IgGl isotype (Sigma). As anegative control, mouse IgG was used: To localizehyaluronan, a biotinylated form of the hyaluronanbinding complex (b-HABR) was used at a concen-tration of 2 pg/ml in 10% calf serum, 90%phosphate-buffered saline (PBS). This complex, de-rived from cartilage proteoglycan, consists of hyalu-ronan binding region protein (HABR) + link and wasa gift of Dr. Charles Underhill (Georgetown Univer-sity Medical Center, Washington, DC).7 A MAb wasalso used for the detection of a non-CD44 hyaluro-nan binding protein (HABP) possessing propertiesof a putative hyaluronan receptor molecule. This an-tibody, designated MAb IVd4, was a gift of Dr. BrianToole (Tufts University, Boston, MA)8 and was of theIgM isotype.

Single LabelingThe avidin-biotin horseradish peroxidase com-

plex procedure of Hsu et al,9 was used withmodifications for each antibody tested. The primaryantibody was incubated at 4 C overnight in a hu-midifying dark box (Accurate Chemicals, Wesbury,NY). After washing with buffer for 10 minutes withshaking, sections were incubated with biotinylatedsecondary antibody (Vector Laboratories, Burlin-game, CA), diluted 1:200 in PBS, 0.1% sodiumazide for 1 hour at room temperature, and thenwashed in buffer. Avidin-biotin-horseradish peroxi-dase complex (Vectastain ABC Kit Standard, VectorLaboratories) was placed on the slides for 30 min-utes at room temperature and then washed. Sites ofantibody reactivity were localized with the chro-mogens 3,3-diaminobenzidine (Sigma) or 3-amino-9-ethylcarbazole (Dakopatts, Carpenteria, CA). Pos-itivity appeared as dark brown staining in the caseof the former or bright red staining in the case ofAEC. The reaction was terminated by dipping theslides in water and then counterstaining with May-er's hematoxylin.

Comparisons in positivity for the different antibod-ies were made semiquantitatively as described pre-viously10 in which labeled cells within each high-power field were quantified per mm2 using an

Hyaluronan 1043AJP April 1993, Vol. 142, No. 4

ocular grid micrometer. The proportion of labeledcells was then graded as: 0, 1+ (up to 25% la-beled), 2+ (25-50% labeled), 3+ (50-75% la-beled), or 4+ (75-100% labeled).

Double LabelingThe goal of the double labeling procedure was to

simultaneously disclose both CD44 and endoge-nous hyaluronan on the cell surface. Anti-CD44MAb and b-HABR probe were used. An original se-quential double immunenzymatic labeling methodmodified after Hancock et all1 was performed,again using chamber slides. CD44 was first local-ized by a four-layer peroxidase-anti-peroxidasemethod, consisting of incubations with MAb BU-52(anti-CD44) at a concentration of 1:100. After incu-bation for one hour, the bridging antibodies, rabbitanti-mouse IgG and swine anti-rabbit IgG, wereeach incubated for 30 minutes, followed by the rab-bit peroxidase-anti-peroxidase incubated for 30minutes. Each layer was followed with severalwashes. For the chromogenic substrate reaction,AEC was again used to give a red color. Cells werethen labeled with b-HABR probe using a 2-layermouse alkaline phosphatase antialkaline phos-phatase technique. After incubating overnight at 4C and washing, slides were incubated with avidin-biotin-complex-alkaline phosphatase following themanufacturer's directions (Vector Laboratories). Thesubstrate reaction was achieved using the alkalinephosphatase kit Ill (Vector Laboratories) to give ablue color. The reaction was terminated by washingin buffer; no counterstain was used. Red indicatedCD44, whereas blue indicated hyaluronan. Color-mixed cells (red and blue substrates) expressingboth antigens (co-labeled with both primaryantibody/probe) appeared dark violet. Sectionswere washed and mounted in a 1:1 (v/v) mixture ofglycerol and PBS. Control slides were obtained byomitting the primary antibody/probe in either label-ing to show a lack of cross reaction between thefirst and second antibody sequences.

Enzymatic PretreatmentCells were seeded in chamber slides and allowedto grow for 24 hours. Media were then changed andreplaced with media that contained either 0 or 20turbidity reducing unit TRU/ml Streptomyces hyalu-ronidase (Calbiochem-Behring Corp., La Jolla, CA).Cells were incubated for 3 hours at 37 C. The en-zyme reaction was stopped and slides were pre-pared for immunohistochemistry as described

above to detect expression of CD44 and hyaluro-nan in hyaluronidase-treated and hyaluronidase-untreated cultures.

Radioimmunobinding AssayA Dot Microfold apparatus (V&P Scientific Inc., SanDiego, CA) was used to measure the expression ofCD44 and HABP in normal skin, normal scar, andhypertrophic scar cell lines and to determine the ef-fect of hyaluronidase pretreatment on CD44 expres-sion. Cells either treated or untreated with Strepto-myces hyaluronidase, as described above, werefixed with 2% glutaraldehyde for 30 minutes thenseeded at 5 x 104 cells/well in the manifold. Themanifold has the configuration of a 96-well platewhose bottom surface consists of glass microfilterpaper (Whatman, Clifton, NJ). Five replicates ofeach condition and each cell line were used. Eachwell was incubated with 100 p1 of anti-CD44 (BU-52)at a postsaturation concentration of 1:10 for onehour. This concentration was based on previous ti-tration assays at concentrations of 1:100, 1:50, and1:25 for each of the different cell lines. A negativecontrol, in this case mouse IgG antibody, was alsoused. Wells were then washed with PBS, and thesecondary antibody, a 35S-conjugated sheep anti-mouse Ig, was used at 0.1 pCi/well. After 30 min-utes, the wells were washed several times to re-move nonspecific binding. Well sites were cut outseparately from the filter, and each was counted forimmune reactivity by liquid scintillation analysis. Inanother set of experiments, MAb IVd4 at a concen-tration of 1:50 was incubated for one hour with thedifferent cell lines (each having been pretreatedwith Streptomyces hyaluronidase), and expressionwas quantified as mentioned above using 35S-conjugated mouse 1g. Results were expressed asdisintegrations per minute (DPM) for each cellularcondition and each antibody tested.

Effects of Extrinsic Hyaluronan on CD44ExpressionThe effect of cold (nonradioactive) hyaluronan onCD44 expression by normal skin and hypertrophicscar fibroblasts was examined: normal skin fibro-blasts were seeded at 5 x 104 into 24-well plates.After 24 hours, the cultures were incubated with hy-aluronan at two different concentrations (2 nmol/Land 20 nmol/L) at 37 C for 0 to 60 minutes. Thecells were then washed, fixed with 2% gluteralde-hyde, and incubated with BU-52 at 1:20 for 2 hours

1044 CD44, and Fibroblasts,Messadi et alAJPApril 1993, Vol. 142, No. 4

at room temperature. A mouse IgG was used as anegative control. Binding of CD44 was revealed byincubation with 35S-conjugated sheep anti-mouseIg.

Similarly, hypertrophic scar fibroblasts wereseeded at 5 x 104 into 24-well plates. After 24hours, cells were incubated with 20 nmol/L of coldhyaluronan at 37 C for 0 to 360 minutes. Cells werethen washed, fixed with 2% gluteraldehyde, and in-cubated with BU-52 at 1:20 for 2 hours at room tem-perature. Mouse IgG was again used as a nega-tive control. The cells were then washed and incu-bated with 1251-conjugated goat anti-mouse Ig(ICN Biochem, Costa Mesa, CA) for 1 hour, washedagain, solubilized with 2N NaOH, and radioactivitycounted.

GAG SynthesisIn fourth passage, three fibroblast cell lines derivedfrom human normal skin, normal scar, and hyper-trophic scar were seeded into three 75-cm2 flasks(5 x 104 cells/10 ml) and cultured. After 48 hours,media were removed and cells were washed with

Hanks' buffer, then incubated with 3H-glucosamine(221 mCi/mmol, New England Nuclear, Boston, MA)at a concentration of 5 pCi/ml in serumfree medium.After 48 hours in culture, cells were harvested.

Labeled GAG from medium, pericellular, and cell-ular fractions were isolated as described previ-ously.12 Radioactivity for each fraction was mea-sured by liquid scintillation analysis, and the resultswere normalized. Individual GAG constituents werethen quantitated by electrophoresis on cellulose ac-etate plates as described by Cappelletti et al,13,14as modified by Bronson et al.15

Results

CD44 ExpressionQualitatively, simple visual inspection and scoringrevealed that CD44 is expressed in cutaneous fibro-blast cultures (Figure 1A). Cells from both uninjuredskin and from normal and hypertrophic scar tissuesexhibited cell surface reactivity for CD44 as dis-closed by the CD44 MAbs, BU-52 and J173. Both

I

-

",,;

. :..*. . 5

Figure 1. Single immunocytochemical labeling offibroblastsfor CD44 (A) using AEC as chromogen andfor hyaluronate (B) using alkaline phos-phatase blue substrate kit as chromogen. Note numerous CD44-positive cells as opposed to relativelyfew b-HABR-positive cells. Using double label-ing techniques (C), CD44+/IHA+ cells seem to be restricted to a subpopulation ofCD44+ fibroblasts. (D): co-expression of both CD44 and hyalu-ronate on the same cells. (A, B) x200, (C) x 100 and (D) x 400.

Hyaluronan 1045AJP April 1993, Vol. 142, No. 4

MAbs were diffusely expressed in 75% of the cul-tures with no difference in the pattern of distributionbetween them. When CD44 expression was quanti-tated by radioimmunobinding assay using 35S-conjugated secondary antibody, a distinct differ-ence in CD44 expression was seen. Hypertrophicscar fibroblasts showed the highest expression,whereas, normal (i.e., uninjured) skin fibroblastsshowed the lowest (P < 0.005). CD44 expressionfor normal scar fell between these two extremes(Figure 2). The fibroblastic lines tested were not ge-nerically positive for hyaluronan-binding proteins

a

.

0*

-TT

H..

A Tissm of Origin

a

0a

0

B Anti-CD44 Concentration (nM)Figure 2. A: Comparison ofCD44 expression, for normal skin (NSk),normal scar (NSc), and bypertropbic scar (HSc)fibroblasts. Data areexpressed as disintegration per minute (DPM) based on 35S-conjugated mouse secondary antibody. A significant difference be-tween normal skin and scarfibroblasts (P < 0.005) is observed usingone-way analysis of variance. Values are expressed as mean S.E.B: Fibroblastic CD44 for cultures derived from normal skin (NSk),normal scar (NSc), and bypertropbic scar (HSc). A negative controlcell line derivedfrom invasive human bladder cell carcinoma (HCV)is also shown. Cells were seeded in complete media at a density of3X 104 cells/well in 24-well plates. Cultures were subsequently fixedwith 296 glutaraldebyde for 30 minutes, washed and incubated withincreasing concentrations (0 to 50 nmolkL) of mouse anti-CD44 IgG(MAb BU-52). After 1 hour of incubation, cultures were washed andthen incubatedfor 1 hour with 35S-conjugated sheep anti-mouse an-tibody (Ig). Each data point represents the average of values fromtnplicate wells. Bound DPM shown along the y-axis reflects theamount ofcell surface CD44. Again, thefibroblasts derivedfrom scartissues are seen to be markedly higher in CD44 than are those fromnormal skin.

(HABPs) inasmuch as they exhibited little reactivityto MAb lVd4 (developed from chicken embryo braintissues), an anti-HABP antibody reactive for certainearly embryonic tissues.

Hyaluronan ExpressionUsing biotinylated b-HABR probe, hyaluronan washistochemically localized in cell cultures (Figure1 B). Some fibroblasts showed little or no re-activity, whereas others were strongly positive forhyaluronan-making it clear that hyaluronan expres-sion is more variable than CD44 expression. Thiswas true for all three categories of fibroblasts andwas also evidenced by double-labeling immunocy-tochemistry (Figure 1D). Whereas 75% of the cellsare CD44 positive (red color; Figure 1C), not all ex-press hyaluronan. Only a subset of CD44-positivecells for each cell line were hyaluronan positive. Theblue and red (violet) chromogen indicates those fi-broblasts that co-express CD44 and hyaluronan.Among all the GAGs produced by fibroblasts de-

rived from normal skin, normal scar, and hyper-trophic scar, hyaluronan was the most prominent(Figure 3). 3H-glucosamine incorporated into nondi-alyzable (macromolecular) CPC-precipitable com-ponents (i.e., GAG) was detected in all three culturefractions (medium, pericellular, and cellular) for thenine cell lines tested. The majority (60%) of this in-corporated label was secreted into the medium; therest was associated with the pericellular matrix andintracellular components. For all three culture frac-tions, similar electrophoretic GAG profiles were ob-tained, and in all cases the predominant GAG elec-trophoresed in the position of hyaluronan. Asecondary, lesser peak was seen in the position ofdermatan sulfate. Hyaluronan is thus established asa prominent GAG found in the intra- and extracellu-lar culture fractions of fibroblasts derived from bothnormal and hypertrophic scar tissues.

Relationship between CD44 andHyaluronanNo direct relationship was seen between expressionof CD44 and expression of hyaluronan. Nearly allthe cells expressed CD44, whereas only a subset offibroblasts expressed hyaluronan. Whereas cleardifferences were seen in CD44 expression betweencells from the different tissues of origin (hyper-trophic scar > normal scar > normal skin), no com-parable relationship was seen for hyaluronan ex-pression.

1046 CD44, and Fibroblasts,Messadi et alAJP April 1993, Vol. 142, No. 4

treated with hvaluronidase and those untreated with

U

Eaz

1 2 3 4 5 6 7 8 9

A

hyaluronidase (P > 0.1) (Figure 5).A relationship between CD44 expression and ex-

posure of cultures to exogenous hyaluronan wasobserved (Figure 6). When normal skin fibroblastswere incubated with nonradioactive (cold) hyaluro-nan at two different concentrations (2 nmol/L and 20nmol/L), a marked, time-dependent decline (down-regulation) of CD44 expression occurred. Hyper-trophic scar fibroblasts cultivated under the sameconditions did not exhibit a similar decline (Figure7).

Slice Number

s0o

60'

40-

40

30

20 -

C4fs

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

Slice Number

1 2 3 4 5 6 7 9 10 11 12 13 14 15 16 17 18 19

C Slice Number

Figure 3. Quantitative cellulose acetate electrophoretic profile of me-dium fraction (A), pericellularfraction (B), and cellularfraction (C)of subconfluent cells from human normal skin, normal scar, andhypertrophic scar, labeled for 48 hours with 3H-glucosamine. 7hemigration positions ofknown GAG standards are shown: HP = hep-arin, DS = dermatan sulfate, HS = heparan sulfate, HA = hyaluro-nan, C4/6S = chondroitin-4-sulfate and chondroitin-6-sulfate. Hy-aluronan is shown to be the predominant GAG present in themedium, penicellular, and cellularfractions for all the cell lines.

Streptomyces hyaluronidase pretreatment of cellsdid not abolish their CD44 reactivity to MAb BU-52but did abolish staining for hyaluronan by b-HABRprobe (Figure 4). Quantitatively, there was no signif-icant difference in CD44 expression between cells

DiscussionFamilies of cell adhesion molecules are expressedin many different tissues and probably participate indiverse cellular interactions at distinct anatomicsites. CD44 is one such molecule, having numerousputative functions in a variety of cell types. Previ-ously known as Pgp-1, ln(Lu)-related p80, Hermes,ECM-111, and H-CAM, the co-identity of all these in-dependently discovered molecules has become es-tablished within the past year. The contribution ofthe CD44 molecule to lymphocyte activation, matrixadhesion, and attachment of lymphocytes to lymphnode high endothelial venules is now widely recog-nized.16A relationship between CD44 and hyaluronan

was initially advanced by Goldstein et al17 and Sta-menkovic et aI18 who independently reported thatthe extracellular domain of CD44 is homologous tothe hyaluronan binding region of cartilage link pro-tein. This finding suggested that CD44 might bindhyaluronan and, thus, that it might act as a cell sur-face receptor for hyaluronan.19 In fact, a murine re-ceptor for hyaluronan has now been described thatresembles Pgp-1/CD44 in size, cellular presenta-tion, and interaction with the cytoskeleton.20,21 Mi-yake et a120 and Aruffo et aI21 both confirmed abinding interaction between hyaluronan and CD44and established CD44 as the principal cell surfacereceptor for hyaluronan in lymph node high endo-thelial cells. Aruffo et a121 established hyaluronan asa ligand for CD44 based on the abolition of reactiv-ity of hyaluronidase-treated lymph node tissue to asoluble CD44 receptor immunoglobulin fusion pro-tein used as a probe for hyaluronan.

Such interactions between CD44 and hyaluronanhave evoked special interest in view of hyaluronan'srole in various biological processes, including cell-to-cell adhesion, cell migration during morphogene-sis, and modulating the behavior of endothelial and

Ez

20'

B

60

50

aa

a

n. I . . . . . . . . . . . -

la -

4P DS HS r e fl&1

Hyaluronan 1047AJP April 1993, Vol. 142, No. 4

Figure 4. Striking loss of (b-HABR) reactivity was observed when fibroblasts were treated with Streptomyces hyaluronidase. Immunohistochemicaltechnique using 3,3-diaminobenzidine as chromogen was applied giving a dark brown coloration ofthe positive cells (A); note complete loss ofby-aluronate expression (B) after hyaluronidase treatment. (A, B) X200.

inflammatory cells,22 particularly in the hyaluronan-enriched extracellular environment of early tissuerepair and remodeling.23 The present study exam-ined the relationship between hyaluronan and CD44in human adult skin and scar fibroblasts. It is thefirst investigation to report CD44 as a normal humanadult fibroblastic antigen in culture and to assesshuman fibrotic tissues for the presence of CD44.Further, hyaluronan, as a known ligand for CD44,was shown to be the predominant GAG constituentof the pericellular matrix of fibroblasts derived fromnormal human and hypertrophic scar tissues. Thisis an important distinction between human and ani-mal fibroblasts inasmuch as heparan sulfate pre-dominates in the latter.24

Our previous work has indicated that hyaluronanbinding to the surface of cutaneous scar fibroblastsseems to be a receptor-mediated phenomenon andis probably a prerequisite for hyaluronan endocyto-sis and degradation.25 The accumulation of hyaluro-nan within the extracellular matrix depends on bio-synthesis, transport, secretion, and degradation-all processes that may differ between fibroblastsfrom normal versus hypertrophic scar (or other fi-brotic tissues). The presence and properties of hy-aluronan receptors on such cells could be impor-tant in understanding mechanisms of hyaluronanbinding to cell surfaces, subsequent internalization,

* NSk* NSc

T * HSc2ILa

11ima

21.11oc

- HAase + HAaseTreatment

Figure 5. CD44 expression in fibroblasts derived from normal skin(NSk), normal scar (NSc), and hypertrophic scar (HSc) tissues, un-treated and treated with Streptomyces hyaluronidase (mean SE).Data are expressed as DPM of35S-conjugated mouse secondary anti-body. These results show no significant reduction in the expression ofCD44 in the different cell lines when treated with the hyaluronidaseenzyme (P > 0.1) (t-testforpaired samples).

--*- 20 nM HA-0-- 2nM HA

Cold HA incubation In Minutes

Figure 6. Effect of incubation with nonradioactive (cold) hyaluro-nan on expression ofCD44 by normal skin fibroblasts. Fibroblasts (5X 104) were seeded into 24-well plates. After 24 hours, the cultureswere incubated with hyaluronan at two different concentrations (2nmol/L and 20 nmol/L) at 3 7 Cfor the periods shown (in minutes)along the x-axis. The cells were then incubated with BU-52 (anti-CD44) (1:20) for one hour. Binding ofBU-52 was revealed by incu-bation with 35S-conjugated sheep anti-mouse Ig and is expressed inDPM. A decline in CD44 is seen as the duration of incubation withhyaluronan increases.

15000 -

X. 10000*

5000

0-

I

-qr-

1048 CD44, and Fibroblasts,Messadi et alAJP Aprfl 1993, Vol. 142, No. 4

40000

30000-

a

;1

10000*

I 100 200

Cold HA Incubation In Minut

Figure 7. CD44 expression for fibroblasts derivedfiscar as afunction ofduration of incubation with naluronan at a concentration of 10 nmol/L. BindinUCD44) was revealed unth 125I-conjugated goat antiing is expressed in DPM along the y-axis; incubatioin minutes is expressed along the x-axis. For hypedecline is seen as duration of incubation with hyalu

and degradation. In fact, Culty et al rfirmed that CD44 participates in thedegradation of hyaluronan by transf(blasts (SV-3T3 cells) and alveolar macr

In the present study, Streptomyces hdid not abolish CD44 reactivity butstaining for hyaluronan with b-HABR prsult establishes that CD44 attachmensurface is not mediated by hyaluroncports the concept of CD44 as an intenent of the cell membrane. If CD44 weaccretion to the hyaluronan-enrichedmatrix (which, in turn, is bound to theby some other means), then hyalurorment would have abolished both hyaCD44 expression. Thus, the CD44 rrpears to be both distinct from hyalurodependent on the presence of hyaluronpression.We have demonstrated that norma

blasts express significantly less CDhypertrophic scar fibroblasts and that aenous hyaluronan to cultures of normeblasts causes a further decline in CD44A similar effect was not seen for hypercells. This prompt down-regulation of Cmal skin fibroblasts but not by hypertrcbroblasts in response to external hyalexplain why hypertrophic scar fibroblkhigher levels of CD44 in the first plac(Further, it supports the view that fibroblfrom normal skin and hypertrophic scai

long to different fibroblastic substrains exhibitingdifferent phenotypic behaviors.

In principle, a decline in CD44 expression by nor-mal skin fibroblasts could have several explana-tions: 1) the anti-CD44 monoclonal antibody and hy-aluronan may be competing for the same bindingsite on the CD44 molecule; 2) the antibody and hy-aluronan may be competing for different epitopes,but exposure to hyaluronan effectively conceals theantibody's recognition site or induces a conforma-tional change that alters its recognition site; or 3)CD44 may participate in both the binding and the

300 * 400 internalization of hyaluronan, with binding involvingthe internalization of the entire receptor-ligand com-

em hypertrophic plex as has been shown in both endothelial and inonradioactive by- 3T3 fibroblastic systems.26'27 Among these three al-g of BU-52 (anti- ternatives, the first and second have been excludedi-mouse 1g. Bind-n of hyaluronan because incubating normal skin fibroblasts with in-?rtrophic scar, no creasing concentrations of nonradioactive hyaluro-ironan increases.

nan over the range of 0 to 20 nmol/L did not causea decrease in binding by the anti-CD44 antibody aswould be expected if hyaluronan were either com-

recently con- peting for the same recognition site as the anti-uptake and CD44 antibody or if it were concealing the recogni-ormed fibro- tion site (data not shown). These findings arerophages.26 compatible with the interpretation that normal skinyaluronidase fibroblasts internalize hyaluronan in the form of adid abolish receptor-ligand complex (i.e., CD44-hyaluronan),obe. This re- leading to less cell surface expression of CD44it to the cell when these cells are exposed to extrinsic hyaluro-an and sup- nan. In contrast, hypertrophic scar fibroblasts do)gral compo- not down-regulate CD44 in response to hyaluro-)re simply an nan-possibly because they do not internalize

pericellular receptor-ligand or because they recycle receptorscell surface to the cell surface very quickly.nidase treat- In conclusion, this study observed evidence foriluronan and phenotypic heterogeneities between fibroblasts de-nolecule ap- rived from normal and scar tissues with regard tonan and not CD44 expression. Also, relationships were identifiedian for its ex- between expression of CD44 and the presence of

its known ligand, hyaluronan. Specifically, CD44skin fibro- was identified as a normal human adult fibroblastic

44 than do antigen whose expression is increased for fibro-idding exog- blasts derived from hypertrophic scar comparedal skin fibro- with those from normal skin. Unlike for animal fibro-[ expression. blasts, hyaluronan was shown to be the predomi-rtrophic scar nant GAG constituent of the pericellular matrix of,D44 by nor- these fibroblasts. However, CD44 attachment to thephic scar fi- cell surface was not mediated by hyaluronan, andluronan may the presence of hyaluronan was not a prerequisiteasts express for CD44 expression. Moreover, exogenous hyaluro-e (Figure 2). nan induced a decline in measurable CD44 expres-lasts derived sion at the cell surface, which is not seen for hyper-r tissues be- trophic scar fibroblasts. These observations are

n -- -

Hyaluronan 1049AJP Apil 1993, Vol. 142, No. 4

consistent with current understanding of the waycells manage the hyaluronan economy of the extra-cellular matrix in other systems.

AcknowledgmentsWe thank Drs. John Pesando, Charles Underhill,and Brian Toole for their generous gift of mono-clonal antibodies.

References1. Rockwell WB, Cohen IK, Ehrlich HP. Keloids and hy-

pertrophic scars: a comprehensive review. Plast Re-constr Surg 1989, 84:827-834

2. Underhill CB. The interaction of hyaluronate with thecell surface: the hyaluronate receptor and the coreprotein. The Biology of Hyaluronan. Edited by Wiley C,Ciba Foundation Symposium 1989, 143 pp87-106

3. Laurent TC, Fraser RE. Hyaluronan. FASEB J 1992,6:2397-2404

4. Culty M, Miyake K, Kincade PW, Silorski E, ButcherEC, Underhill CB. The hyaluronate receptor is a mem-ber of the CD44 (H-CAM) family of cell surface glyco-proteins. J Cell Biol 1990, 111:2765-2774

5. Diegelmann RF, Cohen I, McCoy B. Growth kineticsand collagen synthesis of normal skin, normal scarand hypertrophic scar fibroblasts. Cell Physiol 1979,98:341-346

6. Botstein GR, Sherer GK, LeRoy EC. Fibroblast selec-tion in scleroderma. Arthritis Rheumat 1982, 25:189-195

7. Green SJ, Tarone G, Underhill CB. Distribution of hy-aluronate and hyaluronate receptors in the adult lung.J Cell Sci 1989, 89:145-156

8. Banarjee SD, Toole BP. Monoclonal antibody to chickembryo hyaluronate binding protein: changes in distri-bution of binding protein during early brain develop-ment. Dev Biol 1991, 146:186-197

9. Hsu SM, Raine L, Fanger H. A comparative study ofthe peroxidase antiperoxidase method and an avidin-biotin complex method for studying polypeptide hor-mones with radioimmunoassay antibodies. Am J ClinPathol 1981, 75:734-738

10. Messadi DV, Pober JS, Murphy GF Effects of recombi-nant gamma interferon on HLA-DR and DQ expres-sion by skin cells in short-term organ culture. Lab In-vest 1988, 58:61-67

11. Hancock WW, Lord RH, Colby AJ, Diamantstein T,Rickles FR, Dijkstra C, Hogg N, Tilney NL. Identifica-tion of IL-2R+ T cells and macrophages within reject-ing rat cardiac allografts, and comparison of the ef-fects of treatment with anti-IL 2R monoclonal antibody.J Immunol 1987, 138:164-170

12. Bertolami CN, Berg S, Messadi DV. Binding and inter-nalization of hyaluronate by human cutaneous fibro-

blasts. Matrix 1992, 11:11-2113. Cappelletti R, Del Rosso M, Chiarugi VP. Rapid multi-

sample separation of the five most widespread animalglycosaminoglycans. Anal Biochem 1979, 93:37-40

14. Cappelletti R, Del Rosso M, Chiarugi VP. A new elec-trophoretic method for the complete separation of allknown glycosaminoglycans in a monodimensionalrun. Anal Biochem 1979, 99:311-315

15. Bronson RE, Bertolami CN, Siebert EP. Modulation offibroblast growth and glycosaminoglycan synthesis byinterleukin-1. Coll Relat Res 1987, 7:323-332

16. Haynes BF, Telen MJ, Hale LP, Denning SM. CD44-Amolecule involved in leukocyte adherence and T-cellactivation. Immunol Today 1989, 10:423-428

17. Goldstein LA, Zhou DFH, Picker LJ, Minty CN, Bar-gatze RF, Ding JF, Butcher CB. A human lymphocytehoming receptor, the hermes antigen, is related to car-tilage proteoglycan core and link proteins. Cell 1989,56:1063-1072

18. Stamenkovic I, Amiot M, Pesando JM, Seed B. A lym-phocyte molecule implicated in lymph node homing isa member of the cartilage link protein family. Cell1989, 56:1057-1062

19. Lesley J, Schulte RA, Hyman R. Binding of hyaluronicacid to lymphoid cell lines is inhibited by monoclonalantibodies against Pgp-1. Exp Cell Res 1990, 187:224-233

20. Miyake K, Underhill CB, Lesley J, Kincade PW. Hyalu-ronate can function as a cell adhesion molecule andCD44 participates in hyaluronate recognition. J ExpMed 1990, 172:69-75

21. Aruffo A, Stamenkovic I, Melnick M, Underhill CB,Seed B. CD44 is the principal cell surface receptor forhyaluronate. Cell 1990, 61:1303-1313

22. Buckingham RB. Interactions between cells of the im-mune system and hyaluronate synthesis by humandermal fibroblasts. The Biology of Hyaluronan. Editedby Wiley C, Ciba Foundation Symposium 1989, 143,pp170-186

23. Toole BP, Gross J. The extracellular matrix of the re-generating new limb. Synthesis and removal of hyalu-ronate prior to differentiation. Dev Biol 1971, 25:57-61

24. Bronson RE, Argenta JG, Siebert EP, Bertolami CN.Distinctive fibroblastic subpopulations in skin and oralmucosa demonstrated by differences in glycosami-noglycan content. In Vitro Cell Dev Biol 1988, 24:1121-1126

25. Berg S, Ellis DG, Bertolami CN. Affinity of dermal peri-cellular matrix for hyaluronic acid. J Dent Res 1990,69:1946 (abstr)

26. Culty M, Nguyen HA, Underhill CB. The hyaluronatereceptor (CD44) participates in the uptake and degra-dation of hyaluronan. J Cell Biol 1992, 116:1055-1062

27. McGary CT, Raja RH, Weigel PH. Endocytosis of hy-aluronic acid by rat liver endothelial cells: evidence forrecycling. Biochem J 1989, 257:875-884