CD44 expression on murine tissues

CD44 is the generic name for a complex set of cell surface glycoproteins which are involved in macrophage and lymphocyte interaction with endothelia (Goldstein et al., 1989; Stamenkovic et al., 1989; Gallatin et al., 1989; Idzerda et al., 1989; Picker et al., 1989; Butcher, 1990). Most studies of CD44 have focused on expression in cell lines, where it has been shown that there are a large number of molecular variants. In addition to differences in post-translational modification (Idzerda et al., 1989; Lokeshwar and Bourguignon, 1991; Camp et al., 1991), the core protein of CD44 may display different variants indicated by the detection of several alternatively spliced mRNAs (Goldstein and Butcher, 1990; Stamenkovic et al., 1991; Brown et al., 1991; Dougherty et al., 1991; Günthert et al., 1991; Hofmann et al., 1991; Jackson et al., 1992). Recent analyses of mRNA from several cell lines indicate that as many as six core protein variants may exist. These can be formed by the insertion of different combinations of “cassette” sequences at a specific site in the extracellular domain (Jackson et al., 1992). Analyses of protein expression in human cells indicate that two molecular sizes are predominant: (1) low M_r CD44, 80-90 × 103, is expressed in lymphoid tissue (mostly macrophages and subsets of T cells) and (2) high M_r CD44, 130–160 × 103, is expressed on tumor cells and keratinocytes. Higher M_r forms (>200 × 103) have also been noted on tumor cells, and probably contain chondroitin sulfate (Brown et al., 1991). Evidence shows that the CD44 form found on lymphocytes functions as a hyaluronate receptor (Stamenkovic et al., 1991; Lesley et al., 1990; Culty et al., 1990; Aruffo et al., 1990; Lesley et al., 1992) mediating the binding of lymphocytes and macrophages in tissues. This function is not performed by large CD44(E) (Stamenkovic et al., 1991), nor by in vitro generated variants of the low M_r form lacking most of the cytoplasmic domain (Lesley et al., 1992). CD44 binds to the heparin-binding domain of fibronectin through chondroitin sulfate (Jalkanen and Jalkanen, 1992) and it also interacts with vascular addressin (Picker et al., 1989).

In the mouse, CD44 was originally identified on cultured cells as Pgp-1 (Hughes et al., 1983, 1981; Hughes and August, 1981) and on lymphocytes as Ly24 by monoclonal antibody (mAb) IM7 (Trowbridge et al., 1982; Lesley and Trowbridge, 1982). It has also been identified as a thymocyte differentiation antigen (Lesley et al., 1988). In addition, other rat mAbs to the mouse protein exist (Lesley and Trowbridge, 1982; Rao et al., 1991; Zhou et al., 1989; Sy et al., 1991), and these have been used to study CD44 expression, mostly in lymphoid tissue, and the structure and function of different molecular forms produced by cells transfected with modified cDNA (Lesley et al., 1992). It is clear that mRNAs coding for CD44 variants are expressed in cultured cell lines, both of normal tissue and tumor origin; however, studies of CD44 proteins in normal mouse tissues have not included evaluation of the cell types in non-lymphoid tissue that express CD44, nor the molecular forms found at these sites.

During previous work on tumor cells grown in culture (Kennel et al., 1981), we generated a mAb, 133–13A, which precipitates a 90-100 × 103 M_r glycoprotein (P100) from a variety of cultured cells, but which has a more restricted distribution in animals (Kennel et al., 1987). We have used this mAb to quantitate expression of P100 in several mouse tissues, and have shown that the amount of P100 in mouse lungs correlates with the macrophage content (Kennel et al., 1989). Microdistribution studies of 125I-labeled mAb 133–13A have identified cellular sites of expression of the antigen in mouse tissues (Kennel et al., 1991; Kennel, 1992). In this paper we show that P100 is a form of CD44. Since most previous work on CD44 has been done on molecular forms expressed in tumor cell lines, we have used mAb 133-13A to evaluate the molecular forms of CD44 expressed in different normal tissues of mice. This work shows that large forms of CD44 are expressed in basal epithelia of esophagus and trachea, and that large amounts of low M_r CD44 are found in non-lymphoid tissue, including the eye, uterus, adrenal, intestine, stomach and salivary gland. The low M_r CD44 form accounts for more than 90% of the total CD44 detected in the mouse.

Isolation and purification of mAb 133-13A, a rat IgG_2b, has been described previously (Kennel et al., 1981, 1987, 1989). All cell lines are of BALB/c origin. Line 1 is an alveolar type II cell lung carcinoma (Kennel et al., 1981). EMT-6 is a mammary carcinoma (Kennel, 1992) obtained from Dr Leaf Huang, University of Tennessee, Knoxville. NMLC is an uncloned, immortalized culture generated from dissociated normal lung cells (Kennel et al., 1987). Line 1 is propagated in McCoy’s 5A medium with 10% fetal bovine serum (Kennel et al., 1981) and EMT-6 and NMLC are grown in DME medium with 10% fetal bovine serum (Kennel, 1992). BALB/c BD female mice were from the Oak Ridge Biology Division specific pathogen-free colony and were used between the ages of two and four months.

Microdistribution of 125I-mAb 133-13A, specific activity of approximately 10,000 cts per min, was determined as previously described (Kennel et al., 1991). Briefly, mice were injected in the tail vein with 0.2 mls of 125I-mAb (10-1000 mg of 125I-mAb in 0.01 M sodium phosphate buffer, pH 7.6, 0.15 M NaCl (PBS) and 5 μg/ml bovine serum albumin). 24 hours later, animals were killed, and tissues were collected and fixed in buffered formalin for embedding in paraffin, sectioning and autoradiography. Three doses of mAb were tested to insure penetration to epithelial sites (Kennel, 1992). Slides with 5 mm sections were exposed for one month before emulsion development.

Immunohistochemistry was performed on paraffin sections of normal mouse tissue which had been fixed in buffered formalin. Sections were deparaffinized, blocked with 10 μg/ml bovine serum albumin in PBS and incubated for 2 hours with 500 ng/ml 125I-mAb which had been radioiodinated as described above. Sections were washed and processed for autoradiography (Kennel et al., 1992).

Tissue samples were homogenized with a Polytron homogenizer at 10% wet weight/volume suspension in PBS containing 0.5% NP40, 10 μg/ml leupeptin, 100 μg/ml phenylmethylsulfonyl fluoride and 5 mM EDTA (lysis buffer). The suspensions were clarified by centrifugation for 10 minutes at 15,000 g. 1 ml samples were electrophoresed on 8% acrylamide-SDS gels (Kennel et al., 1989) and proteins were transferred to Immobilon P (Millipore) membranes in a Hoefer mini transfer unit at 200 mA for 1 hour. Membranes were washed overnight in PBS plus 0.1% Tween-20 and blocked with a mixture of 5% non-fat dry milk proteins and 20% fetal bovine serum in PBS before incubation for 3 hours with detection 125I-mAb 133–13A (500 ng/ml at 5,500 cts per min/ ng-1). Membranes were washed and subjected to autoradiography at -80oC on Cronex intensifying screens (Dupont) for 1-3 days. Autoradiograms were scanned using a Model 300A Computing Densitometer and associated software (Molecular Dynamics). Digital values for band intensity were converted to amounts of 125ImAb 133-13A bound using the known values of specific activity of the mAb and a standard curve of digital values versus cts per min of 125I from the same preparation. Replicate evaluations were within 10% of each other. Scans of different exposure times of the same western blots gave nearly identical values when corrected for exposure time.

P100 protein was purified from a detergent lysate of cultured Line 1 tumor cells using immunoaffinity chromatography as previously described (Kennel et al., 1989). Approximately 10 mg of protein was precipitated with trichloroacetic acid to remove residual detergent before application to the sequencing chamber. Alternatively, up to 50 mg of purified P100 was digested with cyanogen bromide (Matsudaira, 1989) in 70% formic acid. The resultant peptides were separated on 10% acrylamide SDS-PAGE, transferred to Immobilon P in CAPS (3-[cyclohexylamino]-1-propane-sulfonic acid; Sigma) buffer and the major stained band, of ∼40,000 M_r, was cut from the membrane for direct automated sequencing (Matsudaira, 1987). Amino acid sequence determination was conducted on an Applied Biosystems 470A gas phase sequenator, as previously described (Kennel et al., 1989).

Spleen cells suspended in PBS supplemented with 0.1% BSA and 0.1% NaN3 (PBS+) were incubated on ice in the presence of phycoerythrin-conjugated IM7 mAb (PharMengen) and fluoresceinconjugated mAb 133–13A for 30 minutes. After 3 washes with PBS+ and the addition of 20 mg propidium iodide (PI), 4 parameter data from 10,000 events were collected using a FACStar Plus (Becton Dickenson) equipped with an argon laser emitting a beam with a wavelength of 488 nm. The data collected were analyzed using the Multi2D software (Phoenix Flow Systems). The red blood cells and dead cells (PI+) were eliminated from the analysis by gating on the size and propidium iodide uptake (PI-).

Line 1 cells were surface radioiodinated using lactoperoxidase, washed, solubilized in lysis buffer and subjected to immunoprecipitation with mAb 133–13A, mAb IM7 or control rat mAb 13514, as described previously (Kennel et al., 1981). Immunoprecipitates were processed for treatment with N-glycanase (genzyme, Cambridge, MA) as recommended by the supplier, and analyzed by SDS-PAGE followed by autoradiography.

Immunodepletion was done on the solubilized 125I-labeled Line 1 surface proteins using mAb 133–13A or control mAb 135–14 coupled to Sepharose beads (Kennel et al., 1989). Labeled proteins (200 ml) were incubated with 10 ml immunobeads for 3 hours at RT. The supernates from the beads (13A cleared or 14 cleared) were subjected to double antibody solution immunoprecipitation and the samples were analyzed by SDS-PAGE.

The molecular target of mAb 133–13A has been characterized by previous work (Kennel et al., 1989). The antigen termed P100 is found chiefly on macrophage-like cells in spleen, lung and peritoneal exudate cells. The protein isolated from Line 1 tumor cells has a M_r of about 92 × 103 and a somewhat smaller apparent M_r (81–86 × 103) when isolated from 3T3 cells or macrophages. This distribution and size difference has also been observed with the Pgp-1 antigen (Hughes et al., 1981, 1983; Hughes and August, 1981).

P100 purified from Line 1 cells was subjected to amino acid sequence analyses. The partial sequence obtained, HQQIDLXVTCRYAGVFHVEKNGRY, is identical to that deduced from cDNA analysis of murine CD44 (positions 23-44) with the X position being tentatively identified as N and the C identified by lack of a detectible amino acid at this position. Cysteine would not be detected unless the protein was carboxymethylated before analysis. This sequence from purified mature protein contains two amino acids (HQ) of the putative leader sequence (Nottenburg et al., 1989). Cyanogen bromide cleavage of P100 followed by gel purification and transfer of peptides to polyvinylidene difluoride membranes (Matsudaira, 1989) was used to generate the internal amino acid sequence. The sequence obtained, SLALSKH (G/K/F)ET, matched 8 out of 10 positions of CD44 and was adjacent to an M at position 70 consistent with CNBr cleavage chemistry. These data indicate that P100 is indistinguishable from the CD44 protein in two areas of amino acid sequence and in behavior on SDSPAGE.

Surface proteins immunoprecipitated from radioiodinated Line 1 cells were analyzed by SDS-PAGE. As shown in Fig. 1C, mAb 13A and mAb IM7 (CD44 standard) precipitate proteins having the same mobility on SDS-PAGE, both before and after treatment with N-glycanase. Furthermore, pretreatment of the labeled proteins with immobilized mAb 13A (13A cleared) removes nearly all of the labeled protein reaction with mAb 13A on mAb IM7, while treatment with control beads (14 cleared) does not remove the CD44 (Fig. 1D). Positive control immunoprecipitates with a mAb to an unrelated protein (b₄ integrin) show that neither sample was significantly degraded by protease during the procedure (data not shown).

Finally, two-color flow cytometry was conducted on mouse spleen cells with fluorescein labeled mAb 133-13A and phycoerythrin labeled mAb IM7 (Fig. 1A,B). Quadrants were set on gated data to indicate the cells that were negative (3), IM7 positive (1), 133–13A positive (4) and positive for both IM7 and 133-13A (2). Although there appear to be a few IM7 positive cells that are 133-13A negative, this is probably due to the low number of fluorescein molecules conjugated to 133-13A. Data for positive cells fall on a diagonal indicating that cells with different levels of CD44 (IM7 signal) have the corresponding levels of positivity for 133-13A. Thus proteins detected by mAb IM7, the CD44 standards and mAb133-13A on two different cell types are indistinguishable.

One method of determination of expression of cell surface proteins in the whole animal is to study the microdistribution of radiolabeled mAb injected into live animals (Kennel et al., 1991, 1992; Kennel, 1992). This method can overestimate expression in the vascular space even when different mAb doses are used (Kennel, 1992; Kennel et al., 1992) and so, like immunohistochemistry (IHC), it cannot be used to quantitate expression. However, in vivo mAb deposition is very specific, with less background than standard IHC methods, and detection can be more sensitive. Some of the results of the in vivo localization of 125I-mAb 133-13A in BALB/c mice are shown in Fig. 2. Heavy mAb deposition is noted in lymphoid tissue such as spleen (Fig. 2C) and thymus (not shown). In spleen the mAb localizes in the red pulp and around the edges of follicular areas, consistent with distribution for macrophages (Hughes et al., 1983). In the liver, mAb 133-13A is found in the sinusoids, probably associated with Kupfer cells of the reticulo-endothelial system. This mAb has also been used to quantitate alveolar macrophages in the lung and in peritoneal exudate cells (Kennel et al., 1989).

In addition to these sites, mAb is deposited at some locations where lymphoid cells are not found in high numbers. Basal cells of the esophagus (Fig. 2A and D) and trachea (Fig. 2G) are labeled. The zona fasciculata and zona reticularis of the adrenal gland (Fig. 2I) and the vascular choroid layer of the eye (Fig. 2F) are also labeled with mAb 133–13A. These results are mostly consistent with the IHC analyses of expression in the mouse and in the human (Picker et al., 1989). IHC studies in the mouse done with direct binding of 125I-mAb 133–13A followed by autoradiography confirm these results (Fig. 3) although the staining is lighter in most cases. Labeling of the basal cells of the esophagus (Fig. 3A) and trachea (Fig. 3G) is relatively strong (compare with unstained corresponding samples, Fig. 3B and H). The red pulp of the spleen and edges of the follicular areas are also stained (Fig. 3C) in the same pattern as for antibody microdistribution. Immunohistochemistry of the adrenal (Fig. 3I) is very light, but again shows a similar distribution to that seen with the in vivo technique. Finally, Fig. 3F shows strong staining of macrophages in the lung, and little or no staining of alveolar epithelium. In addition to the sites noted above, macrophages in several organs, the ductal epithelium of the salivary gland and the perifollicular regions of the thymus are CD44 positive (data not shown). Results with negative control 125I-mAb have only a few silver grains located chiefly in lumenal spaces of blood vessels (data not shown).

Western blots of proteins from various cells and organs are shown in Fig. 4. Proteins from detergent extracts of 1 mg wet weight of tissue were separated on 8% SDS-polyacrylamide gels without the addition of reducing agent, since reducing agents destroy reactivity with this mAb. Proteins transferred to membranes were detected with 125I-mAb 13313A followed by autoradiography. This method is very sensitive and has a low background staining. Samples were divided into panels according to the length of exposure to X-ray film necessary to give dark bands. Several cell lines (top panel) have the highest concentration of CD44. Two epithelial tumor cell lines, Line 1, a lung alveolar carcinoma and EMT-6, a mammary carcinoma, display only the low M_r form of CD44, as do lymphoid organs such as spleen and thymus. Sites of expression in basal epithelia (esophagus and trachea) have a large fraction of the high M_r CD44 forms appearing as doublet bands in the 165–205 103 region. In addition, the salivary gland has a large fraction of high M_r CD44. Other non-lymphoid sites such as the eye, uterus, adrenal (not shown), stomach and intestine display predominantly the low M_r form. These data are summarized in Table 1. Significant heterogeneity exists in the SDS-PAGE mobility of both the high and low M_r forms (Fig. 4). The data do not distinguish between altered primary amino acid structure and post-translational modification to account for these differences. No forms with M_r > 205 × 103 were found, unless samples were first exposed to mAb and frozen and thawed before analyses. These procedures generated multiple artifactual bands ranging from 90 × 103 to > 300 × 103 M_r (data not shown). mAb binding to western blots was quantitated by densitometry scanning (Table 1). The concentration of CD44 is the highest in cell lines. For Line 1 cells, 1 mg wet weight or about 5 × 105 cells results in binding of 65.8 ng of mAb. Assuming an M_r for mAb of 150 × 103, this translates to 600,000 sites per cell, a value which agrees well with that previously published of 520,000 sites, determined by classical radioimmunoassay methods (Kennel et al., 1989). On the average, Line 1 tumor grown intramuscularly has about half the concentration of CD44 as do Line 1 cells grown in tissue culture. This is probably because of the presence of CD44 negative cells in the tumor.

Of the normal tissues, the spleen has by far the highest concentration of CD44 epitopes per mg of tissue as determined by this method. Eye, thymus, uterus, adrenal, stomach, liver and salivary glands show heavy deposition of injected 125I-mAb and also show relatively large amounts of CD44 in the western blot data. In esophagus, trachea and salivary gland, about 50% of the total CD44 is in the high M_r form, while in the lung, about 20% is the high M_r form (Table 1). Altogether, the high M_r forms represent less than 10% of the total CD44 detected in the mouse tissues tested.

The quantitation data and the in vivo 125I-mAb deposition data are generally consistent with published values obtained by radioimmunoassay (RIA) (Hughes et al., 1983; Kennel et al., 1989; Kennel, 1992). Exceptions are the liver, lung and intestine, which have smaller values relative to spleen than the RIA values indicated.

The CD44 family of glycoproteins has been identified by several mAbs to lymphoid and other tissues. The fact that CD44-like proteins have been detected in at least five different experimental settings (Goldstein et al., 1989; Stamenkovic et al., 1989; Gallatin et al., 1989; Hughes et al., 1981; Trowbridge et al., 1982; Rao et al., 1991; Zhou et al., 1989; Quackenbush et al., 1990) attests to the immunogenicity of the molecule as well as its broad distribution. The function of CD44 was originally shown to involve phagocytosis (Hughes et al., 1983; Hughes and August, 1981) and binding of cells to extracellular matrix (Gallatin et al., 1989), as well as lymphocyte trafficking (Goldstein et al., 1989; Stamenkovic et al., 1989; Butcher, 1990). The demonstration that CD44 functions as a hyaluronate receptor (Culty et al., 1990; Aruffo et al., 1990) and recent studies indicating that different molecular forms of CD44 have different abilities to bind hyaluronate (Stamenkovic et al., 1991; Lesley et al., 1992) add another dimension of functional complexity. The importance of alternate forms is evidenced by their expression on tumor cells (Dougherty et al., 1991; Günthert et al., 1991; Sy et al., 1991; Birch et al., 1991) and their correlation with metastases.

Up to seven molecular forms of CD44 have been implicated by sequencing of alternatively spliced mRNAs (Goldstein and Butcher, 1990; Stamenkovic et al., 1991; Brown et al., 1991; Dougherty et al., 1991; Günthert et al., 1991; Hofmann et al., 1991; Jackson et al., 1992), while only three major size classes of CD44 have been defined by protein analyses (Picker et al., 1989; Omary et al., 1988): very high M_r (> 220 × 103; high M_r CD44 (130–160 × 103) and the most common low M_r CD44 (81–90 × 103). Two forms have been identified by amino acid sequence differences (Quackenbush et al., 1990). Only one mAb (Günthert et al., 1991) has been shown to distinguish between two molecular forms, although many mAbs to different epitopes have been identified. It is not known what functional role these forms may play in vivo since, without exception, alternate protein forms of CD44 have been identified on cell lines and not normal tissues. The possibility that low M_r CD44 and CD44(E), one of the high M_r forms, may have different functions (Stamenkovic et al., 1991; Lesley et al., 1990) makes it important to study not only the distribution of CD44 antigens but also to identify the molecular forms expressed in animal tissues.

In early work on mouse lung tumors, we developed a mAb, 133–13A, with broad reactivity for cells in culture (Kennel et al., 1981). It became clear that this mAb reacted with macrophages in lung and other sites (Kennel et al., 1987; 1989). In studies of microdistribution of 125I-mAb in animals (Kennel et al., 1991; Kennel, 1992), information about the target antigen, P100, was obtained. Protein sequencing of the glycoprotein purified from mouse lung tumor identifies this protein, P100, with the CD44 family. Immunoprecipitation of the same surface protein from these cells with mAb 133–13A and mAb IM7 (a standard antibody to murine CD44) indicates that P100 and CD44 are the same. In addition, mAb 133–13A stains the same cell population from spleen as mAb IM7. The epitope identified by mAb 133–13A has not been defined molecularly, but it must be extracellular since the mAb binds to viable cells. It is also likely that the epitope is a protein sequence, since the mAb binds to western blots of “native” samples, but does not recognize the target molecule when it is reduced with mercaptoethanol (data not shown). This mAb does not cross react with human cells (data not shown), indicating that it may possibly react at the site on murine CD44 (positions 187–259), which has the least homology with human CD44 (Zhou et al., 1989; Nottenburg et al., 1989).

Previous work on the distribution of CD44 in normal murine tissue is limited and expression in human tissues has been studied only with conventional IHC (Picker et al., 1989). The purpose of this work is to identify cellular sites where CD44 is found in normal tissue, and determine what M_r forms are expressed. Tissue distribution of CD44 expression has been identified by deposition of 125I-mAb 133–13A after intravenous injection. While this method has some drawbacks (Kennel, 1992), it is the most specific and sensitive technique for cellular localization of antigens in tissue due to the low level of background. IHC of fixed sections with 125I-mAb 133–13A supports the localization data. Distribution of 125I-mAb 133–13A in mice compares well with IHC studies of human tissue (Picker et al., 1989). One difference between the IHC data on human tissue and the data for mice presented here is that CD44 in murine lung is expressed predominantly on macrophages (Kennel et al., 1989) and only weakly, if at all, on Type 1 cell membranes as indicated for humans. Accumulation of 125I-mAb is high in the spleen, liver, eye and intestines (Kennel, 1992, and Fig. 2), but is also particularly striking in basal esophageal and tracheal epithelium as well as adrenal cortex (Figs 2 and 3). Previous analyses of mouse tissue had not implicated epithelium as a major site of expression (Hughes et al., 1983).

Few data are available about the forms of CD44 expressed in either human or murine normal epithelia. Protein sequencing has confirmed the presence of two different C-terminal domains in peripheral blood cells from a chronic lymphocytic leukemia patient (Quackenbush et al., 1990), and western blot analyses demonstrate that the large M_r form is expressed in a human squamous epithelial cell line (Picker et al., 1989). Thus, nearly all studies on molecular forms of CD44 have been conducted on cell lines.

To test the hypothesis that the large M_r form of CD44 is expressed in epithelium while small CD44 is predominant in lymphoid tissue, western blot experiments from isolated mouse tissues were done. Both high and low M_r forms of CD44 (Fig. 4) were detected. No forms of > 205 × 103 M_r were noted. Chondroitin sulfate modification is present only on very high M_r forms (Jalkanen and Jalkanen, 1992). Although three consensus sites for glucosaminoglycan (gag) addition are present in both mouse and human CD44, they are apparently not derivatized in the small form. It is likely that splice variant of CD44 containing exon 4 which has another site for gag addition is the active site. Since this site is not the epitope recognized by mAb 133–13A, it is likely that the derivatized form, if present, would have been detected. Since chondroitin sulfated CD44 has been found only in tumor cell lines (Brown et al., 1991) and as a minor component on lymphocytes, it is not surprising that it was not detected in normal murine tissue.

The low M_r form is expressed in lymphoid tissue such as spleen and thymus as previously reported; however, it is also expressed in non-lymphoid tissue where the cellular site cannot be easily identified at the resolution of these experiments (eye, liver, uterus, intestine, stomach and adrenal). In the lung, major expression is on alveolar macrophages (Kennel et al., 1989) and 80% of the CD44 is of the low M_r form. Expression in lung alveolar epithelium was not detected (Fig. 3). Expression in bronchiolar epithelium could account for the high M_r form in this organ. The two sites, esophagus and trachea, where expression can be localized to the basal epithelium, have about 50% of their CD44 in high and 50% in low M_r forms. The high M_r forms appear as doublet bands of about equal intensity and may be the result of alternative splicing, differential glycosylation or modification with chondroitin sulfate; however, treatment with hyaluronidase or chondroitinase does not affect the mobility of these two bands (data not shown). The small CD44 expressed in these organs may be due to transient macrophages in the muscle layer, or to expression in the mucous glands. Overall, these data are consistent with large CD44 forms such as CD44(E) being expressed in normal basal epithelia; however, only the low M_r forms are detected in skin and intestine.

Quantitation of CD44 by densitometry scanning of western blot autoradiograms is in general agreement with radioimmunoassay (RIA) data (Hughes et al., 1983; Kennel et al., 1989; Kennel, 1992). Determination of the number of sites of mAb 133–13A binding per Line 1 cell gives similar results with the two methods. Data sets are also consistent for CD44 expression in spleen, lung, uterus and skin, while RIA estimates a higher concentration in liver and intestines than does western blot quantitation. These values may be low due to proteolysis not being inhibited by leupeptin and PMSF, or CD44 may be lost due to binding to insoluble ligands such as basement membrane-bound hyaluronic acid (Jalkanen and Jalkanen, 1992). Treatment of tissue with hyaluronidase did not release significant amounts of CD44 from liver or intestines (data not shown). The data show that, in vivo, the small M_r form is by far the predominant one expressed, accounting for more than 90% of the total CD44 detected.

Finally, the question of the function of CD44 in these tissues is still at issue. The small M_r form of CD44 has been identified as the hyaluronate receptor; however, it is clear that the C-terminal domain is necessary for binding (Lesley et al., 1992). The western blot analyses used here may not differentiate between a deletion of the C-terminal 70 amino acids and altered glycosylation. It has been shown that nearly all lymphoid CD44 has a M_r of about 81 × 103, as do most other sites of expression of the low M_r form. CD44(E) expression in the basal epithelium may have some other function (Stamenkovic et al., 1991; Jalkanen and Jalkanen, 1992), as may the small M_r forms of CD44 in non-lymphoid tissue, adrenals, eye, etc. Attempts to bind CD44 from tissue lysates to immobilized hyaluronic acid have failed (data not shown). This may be due to low affinity of the solubilized CD44, which must function as a monomer molecule. It is likely that stable attachment of CD44 to polymers of hyaluronic acid requires multivalent binding. This hypothesis is consistent with weak interactions of cells with no anchoring C-terminal domain (Lesley et al., 1992). Thus CD44 may function as a weak interaction molecule in the same manner as do the selectins in white blood cell trafficking.

Jim Wesley did the work on the tissue autoradiography. Dr Michael Fry interpreted the mAb distribution and IHC slides. Joel Harp helped with scanning densitometry. Nette Crowe, Jim Hickey and Sylvia Allen provided assistance in manuscript preparation and Dr Peggy Terzaghi-Howe and Diana Popp provided useful comments on presentation of the data and manuscript content.

The research was sponsored by the Office of Health and Environmental Research, US Department of Energy, under contract DE-AC05-84OR21400 with the Martin Marietta Energy Systems, Inc.