Mesenchymal Stem Cell Surface Antigen SB-10 Corresponds to Activated Leukocyte Cell Adhesion Molecule and Is Involved in Osteogenic Differentiation

Authors


  • Parts of this work were presented at the 19th Annual Meeting of the American Society for Bone and Mineral Research in Cincinnati, OH, U.S.A., 1997.

Abstract

Bone marrow contains a rare population of mesenchymal stem cells (MSCs) capable of giving rise to multiple mesodermal tissues including bone, cartilage, tendon, muscle, and fat. The cell surface antigen recognized by monoclonal antibody SB-10 is expressed on human MSCs but is lost during their developmental progression into differentiated phenotypes. Here we report on the immunopurification of the SB-10 antigen and its identification as activated leukocyte-cell adhesion molecule (ALCAM). Mass spectrometry establishes that the molecular mass of ALCAM is 80,303 ± 193 Da and that it possesses 17,763 ± 237 Da of N-linked oligosaccharide substituents. Molecular cloning of a full-length cDNA from a MSC expression library demonstrates nucleotide sequence identity with ALCAM. We also identified ALCAM homologs in rat, rabbit, and canine MSCs, each of which is over 90% identical to human ALCAM in their peptide sequence. The addition of antibody SB-10 Fab fragments to human MSCs undergoing osteogenic differentiation in vitro accelerated the process, thereby implicating a role for ALCAM during bone morphogenesis and adding ALCAM to the group of cell adhesion molecules involved in osteogenesis. Together, these results provide evidence that ALCAM plays a critical role in the differentiation of mesenchymal tissues in multiple species across the phylogenetic tree.

INTRODUCTION

Cells of the osteoblastic lineage are derived from precursors originating in the mesodermal layer of the trilaminar embryo. These cells give rise to the appendicular skeleton and contribute to elements of the axial skeleton as well. Such precursors, which also exist in the postnatal organism, are referred to as mesenchymal stem cells (MSCs)(1,2) since they have the capacity to develop into a variety of tissue types including bone, cartilage, tendon, muscle, fat, and hematopoietic-supportive stroma. Recently, techniques for the isolation of adult periosteum- or bone marrow-derived human and animal MSCs have been described,(3–9) as well as techniques for directing their differentiation into the osteogenic,(8,10) chondrogenic,(5,11) tenogenic,(12,13) myogenic,(14) adipogenic,(15) and mature stromal lineages.(16) These cells have also been shown to retain their developmental potential following extensive subcultivation in vitro,(17) supporting their characterization as stem cells.

Identification of these cells in situ has remained elusive, in part, due to the relative paucity of specific molecular markers. Although several monoclonal antibodies (MAbs) that react with the surface of marrow-derived progenitor cells in culture have been described,(18–21) none of these investigations have reported on the molecular identity of their respective antigens. Recently, a new series of MAbs directed against the surface of human MSCs undergoing osteoblastic differentiation was reported.(22) One of these MAbs, referred to as SB-10, reacts with the surface of human MSCs during fetal long bone and calvarial development, as well as the surface of marrow-derived MSCs in culture prior to differentiation. Immunoreactivity is lost during lineage progression when the osteoblast marker alkaline phosphatase (ALP) is expressed. The SB-10 antigen was observed not only in the outer periosteum of developing bones, but also in the developing brain, lung, and esophagus.(22)

In this report, we present the identification, characterization, molecular cloning, and potential function of the SB-10 antigen during osteogenesis. Using immunoprecipitation, we purified the SB-10 antigen for mass spectrometry and amino acid sequence analysis. Peptides arising from the protease digestion corresponded to the recently reported activated leukocyte-cell adhesion molecule (ALCAM), a CD6 ligand.(23) We then obtained a full-length cDNA clone for ALCAM from a MSC expression library and subsequently identified ALCAM homologs in MSC cultures derived from rat, rabbit, and canine bone marrow. Finally, we evaluated the role of ALCAM during the process of osteogenic differentiation in vitro. These results provide the first evidence for a developmentally regulated cell surface molecule on MSCs that is conserved across multiple species.

MATERIALS AND METHODS

Cell culture and monoclonal antibodies

Mesenchymal stem cells were isolated from fresh bone marrow aspirates of rat, rabbit, canine, and human sources as previously described in detail(3–5,7,8) and cultivated in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Cells were grown in 185 cm2 flasks and serially passaged when they approached confluence. Human myelomonocytic HL60 cells were obtained from American Type Culture Collection (Rockville, MD, U.S.A.) and maintained in Iscove's modified Dulbecco's medium plus 10% FBS. Monoclonal antibodies used in this study were SB-10,(22) SB-1,(24) and J4–81,(25) which was a gift from J. Pesando. MAbs were purified by Protein-G agarose (Boehringer Mannheim, Indianapolis, IN, U.S.A.) affinity chromatography. Antibody Fab fragments were prepared from purified MAbs by digestion with papain followed by affinity purification (Pierce, Rockford, IL, U.S.A.).

Immunoblotting analyses

MSCs cultured under standard conditions were scraped in a lysis buffer composed of 150 mM NaCl (Sigma, St. Louis, MO, U.S.A.), 50 mM Tris (Sigma), pH 7.4 with protease inhibitors of 10 mM EDTA (Sigma), 1 mM phenylmethylsulfonylfluoride (Sigma), 1 μg/ml Aprotinin (Boehringer Mannheim), and 1 μg/ml Leupeptin (Boehringer Mannheim). Cells were sonicated for 30 s on ice, the nuclei and debris were centrifuged to a pellet at 1000g for 10 minutes, and the supernatant was then centrifuged at 48,000g for 2 h at 4°C. The pelleted plasma membranes were resuspended in 0.3% (w/v) 3-[(3-Cholamidopropyl)dimethylammonio]-1-propane sulfate (CHAPS) (Sigma) in phosphate buffered saline (PBS), sonicated, and then centrifuged at 13,000g for 30 minutes to remove insoluble material. Aliquots of the extract were prepared for electrophoresis in sodium dodecyl sulfate (SDS) sample buffer. Some samples were further prepared by boiling, or by boiling and reduction with fresh β-mercaptoethanol. A 4% SDS-polyacrylamide (Sigma) stacking gel, and a continuous 10% separating gel were used in all cases. Proteins separated by polyacrylamide gel electrophoresis (PAGE) were electrophoretically transferred to a polyvinylidene difluoride (PVDF) membrane (BioRad, Hercules, CA, U.S.A.) using standard techniques. The membranes were cut into strips, blocked overnight with 5% nonfat dry milk-PBS-Tween, and then incubated with hybridoma culture supernatant diluted 1:100 in 5% (w/v) nonfat dry milk in PBS-Tween for 1 h on a shaker plate at 25°C. Control experiments used irrelevant primary antibody SB-1, which specifically binds chicken ALP,(24) or no MAb at all. Following thorough washing of PVDF membranes with PBS, detection of bound primary antibody was performed according to the VectaStain Elite ABC Kit instructions (Vector Labs, Burlingame, CA, U.S.A.).

Immunoprecipitation, antigen purification, and peptide sequence analysis

CHAPS-soluble MSC membrane extract (∼27 mg) from ∼1.5 × 108 cells was incubated with 200 μg of purified SB-10 antibody for 1 h at 4°C. A 200 μl slurry of Protein-G agarose was added, and the mixture was incubated in rotation for 16 h at 4°C. The resin was washed exhaustively with fresh lysis buffer containing 0.3% CHAPS, and the affinity-bound material was eluted by boiling for 5 minutes in SDS sample buffer. Following unreduced SDS-PAGE of the eluted material, gels were stained briefly with 0.05% Coomassie-G (Sigma) in 5% acetic acid/10% methanol, destained in 5% acetic acid/10% methanol, and the ∼99 kDa bands from five identical preparations were excised. In-gel digestion was carried out using lysine C-endoproteinase (Wako Chemicals, Richmond, VA, U.S.A.).(26) High pressure liquid chromatography (HPLC) separations were performed on a Hewlett-Packard 1090 LC system (Hewlett-Packard, San Fernando, CA, U.S.A.) using a Vydac C18 column (2.1 × 250 mm). The separated peptides were then subjected to N-terminal amino acid sequence analysis using a Hewlett-Packard G1005A Sequencing System.

Mass spectrometry

Matrix-assisted laser desorption/ionization time of flight (MALDI-TOF) mass spectrometry was performed using a Hewlett Packard G2025A instrument with version A.03.00 software. Immunoprecipitate samples were prepared for mass spectrometry using thin polycrystalline matrix films.(27) A 0.8 μl volume of sinapinic acid solution (Hewlett Packard) was dried onto the G2025A mesa using a Hewlett Packard G2024A sample preparation station, and then gently crushed against a sheet of polypropylene plastic. A saturated solution of sinapinic acid (Aldrich, Milwaukee, WI, U.S.A.) was also prepared in 3:2:1 isopropanol:water:formic acid (isopropanol, HPLC grade from VWR, Inc., Bridgeport, NJ, U.S.A.; formic acid from Aldrich), and the insoluble material was removed by two rounds of centrifugation at 13,000g. A 1 μl volume of the antigen sample was then diluted with 1 μl of this sinapinic acid solution. A 0.8 μl volume of the resulting solution was applied to the crushed matrix on the mesa and allowed to evaporate in air until it reduced to approximately one-third of its original volume. The mesa was then immersed for 3 s in 30 ml of 0.05% (v/v) trifluoroacetic acid (Pierce). The mesa was dried using the G2024A sample preparation station and loaded into the instrument. Spectra were acquired using a laser power of 11.2–11.4 μJ, with the number of shots accumulated indicated in the text.

Deglycosylation of immunoprecipitates

Antibody SB-10 and irrelevant control immunoprecipitates were prepared from MSCs as described above. The cell extracts were pretreated with Protein-G agarose to remove any components which bound nonspecifically. The affinity-bound material was eluted by boiling in 25 μl of 0.5% SDS solution for 5 minutes. The solution was briefly centrifuged and the supernatant was transferred to a new tube for alkylation and deglycosylation of the immunoprecipitates. Bovine fetuin (Sigma) was digested in parallel to confirm appropriate activity of the glycosidases. A 2.5 μl volume of 1 M Tris, pH 8.5, was added to each sample followed by 1 μl of 1% (v/v) triethylphosphine (Aldrich) in methanol and 1 μl of 1% (v/v) 4-vinylpyridine (Aldrich) in methanol. The reactions proceeded for 4 h at 37°C. The volume of the samples was reduced by half in vacuo, after which water was added to return to the original volume. A 10 μl volume of 7.5% (v/v) NP-40 (Sigma) was added to the solutions followed by a 15 h digestion with 0.1 U N-glycosidase F (Boehringer Mannheim) at 37°C. The pH of the samples was adjusted to 7 by the addition of 1% (v/v) acetic acid (HPLC grade, VWR), and 2 h digestion with 3.1 mU neuraminidase (Genzyme, Cambridge, MA, U.S.A.) was then performed. A final 15-h digestion with 0.075 mU of O-glycanase (Genzyme) was performed at 37°C. Aliquots were removed after digestion with N-glycanase and O-glycanase for MALDI-TOF mass spectrometry.

Molecular cloning of ALCAM from human MSCs

A human MSC cDNA expression library was constructed using the λZAP vector system (Stratagene, La Jolla, CA, U.S.A.). Polymerase chain reaction (PCR) amplification of a 372 base pair (bp) fragment specific for ALCAM was performed with the following 5′ primer (5′-ATTATCAT- TTCCCCTGAAGAGAAT-3′) and 3′ primer (5′-GCCTAAGAGAGAAACTGTCCTAGTT-3′). The resulting PCR product was cloned using the TA Cloning Kit (Invitrogen, San Diego, CA, U.S.A.) and sequenced using Sequenase (U.S. Biochemical, Cleveland, OH, U.S.A.) to confirm the nucleotide sequence. The PCR product was labeled with32P-dCTP and used to probe the human MSC cDNA library. Following secondary screening with the same probe, selected plaques were excised and reinfected into the SOLR strain of Escherichia coli, restriction enzyme mapped, and selected clones were sequenced. Two clones with a common, unique internal restriction site were ligated together using T4 DNA Ligase (GIBCO-BRL, Gaithersburg, MD, U.S.A.) to produce a full-length cDNA for ALCAM. PCR oligonucleotides were designed with an NcoI site on the 5′ primer (5′-CATGCCATGGCTCCGTCAGTGGCCCACC-3′) and a BamHI site on the 3′ primer (5′-CATGGATCCAACTAGGACAGTTTCTCTCTTAGG-3′) to enable cloning into the pET-3d vector (Stratagene). PCR was performed with the addition of 5% glycerol and 0.1% bovine serum albumin, with annealing at 60°C. Vent DNA Polymerase (New England Biolabs, Beverly, MA, U.S.A.) was used to amplify the full-length ALCAM clone to reduce the chance of mutations due to PCR error. The sequence was confirmed using the Prism 377 DNA Sequencer (Applied Biosystems, Foster City, CA, U.S.A.).

Identification of ALCAM homologs in canine, rabbit, and rat MSC cultures

Total cellular RNA from canine, rabbit, and rat MSC cultures was prepared with Tri-Reagent (Molecular Research Center, Cincinnati, OH, U.S.A.) and chloroform, and then precipitated with isopropanol. First strand cDNA synthesis was completed using SuperScript II Reverse Transcriptase (GIBCO-BRL), and these cDNAs were employed as templates for PCR using oligonucleotide primers for the 372 bp fragment noted previously, and a 1593 bp fragment specific to ALCAM. The 5′ primer for this 1593 bp fragment was (5′-GGCTCCCCAGTATTTATT-3′), and the 3′ primer was the same as that used for the 372 bp fragment. PCR products corresponding to ALCAM peptide fragments were generated, and then sequenced on the Prism 377 DNA Sequencer to determine the extent of homology with human ALCAM.

Northern blot analysis

Total cellular RNA from human MSCs, selected fetal tissues, and HL60 cells was prepared as described above for the canine, rabbit, and rat MSCs. Twenty micrograms of total RNA was fractionated on a 1% agarose gel containing 6.7% formaldehyde and transferred to a Hybond-N+ nylon membrane (Amersham, Arlington Heights, IL, U.S.A.). RNA was cross-linked to the membrane by baking at 80°C in a vacuum oven and probed with a random-primed, double-labeled (32P-dCTP and32P-dATP) ALCAM 372 bp PCR fragment. Membranes were exposed to phosphoimaging screens and then visualized on a STORM phosphoimaging unit (Molecular Dynamics, Sunnyvale, CA, U.S.A.). The blots were reprobed with a β-actin PCR product (778 bp) after stripping the membrane with 0.1% SDS, 50% formamide in 0.5× SSC at 65°C for 30 minutes.

Osteogenic differentiation of human MSCs

Human MSCs were directed into the osteogenic lineage in vitro by the addition of osteogenic supplements (OS) (100 nM dexamethasone, 10 mM β-glycerophosphate, and 50 μM ascorbic acid-2-phosphate), as previously described.(10) Cultures were assayed for cell number, ALP activity and mineral deposition according to established techniques.(10) Cultures grown under control or OS conditions were also incubated with Fab fragments of SB-10 and SB-1 at a concentration of 10 μg/ml. Fresh Fab fragments were added with each routine medium change every 3 days, for a total of 12 days.

RESULTS

Immunoblotting of the SB-10 antigen

Immunoblots of human MSC plasma membranes, extracted with 0.3% CHAPS and probed with antibody SB-10, showed a specific immunoreactive band at an apparent molecular weight of ∼99 kDa (Fig. 1). The position and intensity of this band did not change when the extract was denatured by boiling; however, immunoreactivity was completely abolished when the extract was reduced with β-mercaptoethanol prior to electrophoresis. To compare the antigen present on cultured MSCs with the antigen recognized in cryosections of limb and lung, extracts of these human fetal tissues were prepared and similarly probed with SB-10. A specific immunoreactive band migrating at an apparent molecular weight of ∼99 kDa was also present in both the lung and limb extracts (data not shown).

Figure FIG. 1.

Immunoblotting of the SB-10 antigen on cultured human MSCs. Plasma membrane protein extracts were prepared, separated, and electrotransferred to PVDF as described in the Materials and Methods and probed with SB-10 (lanes 1–3) or control antibody SB-1 (lanes 4–6). Five micrograms of protein extract were added to each lane. Samples were loaded under native conditions (lanes 1 and 4), following boiling (lanes 2 and 5), and after boiling and reduction (lanes 3 and 6). The electrophoretic mobility of reduced molecular mass standards in kilodaltons is shown to the left in this and all subsequent immunoblots.

Purification and characterization of the SB-10 antigen

Following precipitation of the immune complexes in MSC extracts incubated with antibody SB-10, the eluted proteins were separated by SDS-PAGE, stained with Coomassie-G, and the ∼99 kDa fragment was excised from each gel (Fig. 2A). This process was repeated five times, and the samples pooled, in order to recover enough material for internal protein sequence analysis. Peptides generated by in-gel digestion with Lysine C endoproteinase were separated by reverse-phase HPLC (Fig. 2B). Eluted peaks were hand collected as individual fractions and subjected to N-terminal sequence analysis. The sequenced peptides were referred to as K1 through K8 (Table 1). The peptide sequence found in fraction K7 was identical to a fragment of the peptide in K5 beginning at amino acid residue 164, and as such is not included in Table 1. A search of the Entrez database at the National Center for Biotechnology Information using the BLAST(28) program revealed sequence identity with human ALCAM, whose GenBank accession number is L38608.(23)

Figure FIG. 2.

Immunopurification of the SB-10 antigen, and HPLC separation of digested peptides. (A) CHAPS-soluble membrane proteins from human MSCs were incubated with purified SB-10, the complexes immunoprecipitated, and the affinity-bound material was eluted as described in the Materials and Methods. Following unreduced SDS-PAGE, the gel was stained with Coomassie-G. (B) The ∼99 kDa bands were excised and the protein digested as described in the Materials and Methods. The recovered peptides were separated by reverse-phase HPLC. Collected peaks were referred to as K1 to K8 and were subjected to N-terminal sequence analysis. The results are shown in Table 1. (C) Control digest of a blank piece of polyacrylamide excised from the same gel.

Table Table 1. Sequence Analysis of Peptides Derived from Lysine-C Endoproteinase Digestion of the SB-10 Antigen
original image

MALDI-TOF mass spectrometry was performed on immunoprecipitated material. Figure 3A shows that purified ALCAM, which migrates in SDS-polyacrylamide gels at a position corresponding to ∼99 kDa, has a measured mass by MALDI of 80,303 ± 195 Da. Reduction, alkylation, and removal of N-linked oligosaccharides reduces the mass to 63,590 ± 44 Da (Fig. 3B). No further reduction in mass was observed following treatment with neuraminidase and O-glycanase, which remove sialic acid residues and O-linked oligosaccharides, respectively (Fig. 3C).

Figure FIG. 3.

Deglycosylation of the SB-10 antigen, ALCAM, monitored by MALDI-TOF mass spectrometry. (A) SB-10 antigen, 500 shots accumulated. The average mass and standard deviation for reduced SB-10 antigen from five measurements is shown to be 80,303 Da. The ion corresponding to a molecular weight of 42,300 Da is an artifact observed in MALDI-TOF mass spectra of protein G immunoprecipitates. (B) Reduced SB-10 antigen, alkylated with 4-vinylpyridine and digested with N-glycosidase F, 244 shots accumulated. Ions produced by polypeptides of molecular weight 24,000 and 53,500 Da correspond to reduced and 4-vinylpyridine alkylated SB-10 antibody light and heavy chains, respectively. The artifact observed in (A) is also present. The average mass and standard deviation for de-N-glycosylated SB-10 antigen from four measurements is shown. (C) Reduced, alkylated, de-N-glycosylated SB-10 antigen treated with neuraminidase and O-glycanase, 153 shots accumulated. Peak assignments are the same as in (B) with the average mass and standard deviation for fully deglycosylated SB-10 from four measurements shown.

Molecular cloning of a full-length ALCAM cDNA

Using cDNA from a human MSC expression library as a template, PCR was performed with a 5′ primer corresponding to peptide K6 which begins at residue 446 in human ALCAM (Table 1), and with a 3′ primer corresponding to the carboxy-terminal residue and a portion of the untranslated region of ALCAM. The resulting 372 bp PCR product was cloned, sequenced, and found to match the published sequence of ALCAM exactly. Following screening of the cDNA library, reinfection into the SOLR strain of E. coli, and restriction enzyme mapping of selected clones, two clones with a common, unique internal restriction site were ligated together to produce a full-length cDNA for human ALCAM. Complete sequence analysis of the full-length clone confirmed identity with an ALCAM variant that contains a serine for asparagine substitution at residue 231 of the mature protein.

To verify that antibody SB-10 recognizes ALCAM, we performed a series of immunoblots with J4–81, an antibody known to react with ALCAM.(23,25) A standard CHAPS-soluble human MSC extract was subjected to immunoblotting, and a single band of reactivity at ∼99 kDa was observed in samples incubated with either SB-10 or J4–81 (Fig. 4A). An additional set of experiments was carried out by first performing an immunoprecipitation with antibody SB-10 and then subjecting the eluted material to SDS-PAGE and immunoblotting with SB-10 and J4–81. Figure 4B confirms that a single ∼99 kDa band is specifically reactive with both SB-10 and J4–81. Additional bands present in the lanes exposed to these antibodies represent nonspecific reactivity of secondary antibody which occurs in the presence of an irrelevant primary control antibody. A further comparison of the immunoreactivity of these two antibodies confirmed that both recognize the surface of HL60 cells and activated leukocytes, two populations known to express ALCAM. Finally, immunostaining of fetal limb sections with J4–81 reveal a pattern that is identical to that observed with SB-10 (data not shown).

Figure FIG. 4.

Anti-ALCAM immunoblotting of human MSC extracts and SB-10 immunoprecipitates. (A) CHAPS-soluble membrane proteins from human MSCs were separated, electrotransferred to PVDF, and probed with SB-10 (lane 1), J4–81 (lane 2), and irrelevant SB-1 control (lane 3). (B) Immunoblot of a SB-10 immunoprecipitation from a CHAPS-soluble membrane extract of human MSCs. Following elution, separation, and electrotransfer to a PVDF membrane, individual lanes were probed with SB-10 (lane 1), J4–81 (lane 2), or SB-1 (lane 3).

ALCAM mRNA expression in human fetal tissues and animal MSC cultures

Expression of ALCAM mRNA in selected human fetal tissues was confirmed by performing reverse transcribed (RT)-PCR using oligonucleotide primers coding for the 372 bp fragment described above. Figure 5A demonstrates amplification of this cDNA fragment from cultured human MSCs and from samples of fetal limb periosteum, fetal lung, and fetal central nervous system (CNS). These results correspond to the immunohistochemical staining that demonstrate ALCAM protein expression in these tissues.(22)

Figure FIG. 5.

PCR amplification of ALCAM fragments in various human and animal cells, and Northern blot analysis of ALCAM mRNA expression. (A) Total RNA was extracted from human MSCs, and human fetal limb periosteum, fetal lung, and fetal CNS. Samples were reverse transcribed into cDNA, and amplified using an oligonucleotide primer pair coding for a 372 bp fragment specific for ALCAM. Reaction products were visualized on ethidium bromide-stained agarose gels. (B) Total RNA was extracted from human, rat, rabbit, and canine MSC cultures and subjected to RT-PCR as described above. PCR amplification of the 372 bp fragment was performed, along with amplification of a specific 1593 bp fragment. (C) Northern blot analysis. Twenty micrograms of total RNA from human MSCs, HL60 cells, fetal lung, fetal limb, fetal CNS, control bacteria, and MSCs from canine, rabbit, and rat was loaded in each lane. The 372 bp ALCAM fragment was double labeled and used to probe the membrane. The size of the reactive mRNA was determined using an RNA ladder and probing it with a fragment of lambda DNA (data not shown).

MSC cultures from rabbit, rat, and canine marrow were also evaluated for the expression of ALCAM, or a homologous protein. Using oligonucleotide primers coding for ALCAM-specific fragments of 372 and 1593 bp, RT-PCR of mRNA from these cultures produced fragments of the appropriate size (Fig. 5B). Following PCR amplification of homologous ALCAM cDNAs in animal MSC cultures, sequence analysis was performed using a series of human ALCAM-based oligonucleotide primers. Primers corresponding to the amino-terminal region of the human protein were not capable of supporting PCR amplification from animal mRNA. Hence, nucleotide sequence for the amino-terminal residues corresponding to positions 1–63 of human ALCAM could not be determined in the animal samples. Nevertheless, we obtained reliable sequence data for rabbit and canine ALCAM homologs that map to the remaining 90% of the full-length human cDNA. The nucleotide sequences translate to a protein sequence for rabbit ALCAM which is 95% identical and 97% similar to human ALCAM, and the canine sequence has 94% identity and 95% similarity to human ALCAM. ALCAM from rat MSCs showed similar homology, although a central region of the protein's nucleotide sequence could not be amplified by PCR using primers that were successful in human, rabbit, and canine samples. Of the 155 residues that were deduced for the rat, 89% were identical and 92% were similar to the corresponding region of human ALCAM. The final sequence determination for each species was verified by three independent nucleotide sequencing experiments. The nucleotide and protein sequence data reported in this paper appear in the EMBL, GenBank, and DDBJ Nucleotide Sequence Databases under the accession numbers of Y13240 and Y13241 (rat), Y13242 (canine), and Y13243 (rabbit). Finally, Northern blot analysis of total RNA from human MSCs and fetal tissues demonstrates a single band migrating at an apparent size of ∼6.1 kb (Fig. 5C). HL60 cells were also included as a positive control, since these cells have been shown to express ALCAM mRNA.(23) By contrast, the reactive mRNA observed in canine, rabbit, and rat MSCs is slightly smaller, migrating at ∼5.8 kb.

SB-10 Fab fragments accelerate in vitro osteogenesis of human MSCs

Since previous data indicates that ALCAM is also capable of homophilic interactions,(23,29–31) and ALCAM expression is lost during the in vivo transition from osteoprogenitor to ALP-positive preosteoblast,(22) we evaluated the possibility that blocking ALCAM-mediated interactions would influence osteogenic differentiation. Preliminary studies in our laboratory also indicate that undifferentiated MSCs express low levels of CD6 on their surface. Therefore, cell-to-cell binding of MSCs may be mediated, in part, through this heterotypic ALCAM-CD6 interaction as well. Using a model for in vitro differentiation of human MSCs,(10) we discovered that the addition of SB-10 Fab fragments significantly elevated the level and accelerated the onset of ALP protein expression and mineral deposition during osteogenesis (Table 2). By day 8 of culture, MSCs grown in OS medium containing SB-10 Fab fragments demonstrated nearly three times the ALP activity compared with those cultures exposed to OS alone. By day 12, ALP activity in samples receiving the SB-10 Fabs was still elevated to nearly twice that present in cultures containing only OS. The percentage of cells in the dish that were stained for ALP also increased in the presence of SB-10 Fabs, but not when SB-1 (control) Fabs were added. Furthermore, SB-10 Fab fragments accelerated the terminal differentiation events, as indicated by the early deposition of mineralized extracellular matrix. Cultures grown under control conditions did not contain substantial calcium by day 12, although samples cultivated in OS or OS plus SB-1 Fabs possessed equally modest amounts of mineral. By contrast, cultures incubated in OS plus SB-10 Fabs contained over five times the quantity of calcium than that present in either OS or OS plus SB-1 Fab-treated cultures. The addition of SB-10 or SB-1 Fabs to control cultures had no effect on ALP activity or mineral deposition (data not shown).

Table Table 2. Alkaline Phosphatase Activity in MSC Cultures Grown in the Presence of Anti-ALCAM Fab Fragments
original image

DISCUSSION

We have identified a developmentally regulated cell surface antigen present on human mesenchymal stem cells that is conserved across multiple species. The monoclonal antibody SB-10, which recognizes a surface antigen on human MSCs in vivo and in vitro,(22) was shown to specifically react with the type I membrane glycoprotein known as ALCAM. First discovered on thymic epithelial cells,(32) activated T cells, B cells, and monocytes,(23) ALCAM is a member of the immunoglobulin superfamily of cell adhesion molecules (CAMs), and has been established as a ligand for CD6,(23) a member of the scavenger receptor cysteine-rich family of proteins.(33) ALCAM has significant peptide sequence homology with BEN (SC-1/DM-GRASP), a neuroneal protein in chick,(29,30,34) neurolin, its fish homolog,(35) and MUC18, a cell surface marker of human melanoma tumor progression.(36)

MALDI-TOF mass spectrometry demonstrates that the intact ALCAM molecule has a mass of 80,303 ± 193 Da, although the apparent molecular weight under native or reducing conditions is ∼99 kDa when analyzed by SDS-PAGE.(32) Anomalous mobility of glycosylated proteins is common, e.g., COMP,(37) and the SB-10 antigen proved to have N-linked oligosaccharides. The reduction in mass following treatment of the sample with N-glycosidase accounts for a carbohydrate mass of 17,763 ± 237 Da, after correcting for the mass of the alkylating groups. The mass observed after deglycosylation corresponds well to the mass of the ALCAM core protein, calculated from the cDNA, following modification of all 10 cysteine residues with 4-vinylpyridine (63,343 Da). If all 10 potential sites for N-linked carbohydrate moieties are occupied, the average mass of each of the carbohydrate chains would be 1780 Da, a number within the commonly observed size range for these substituents. Immunoblotting results also indicate that the antigenicity of the SB-10 epitope is dependent on disulfide linkages. Since ALCAM is known to have five immunoglobulin-like disulfide linkages in the extracellular domain, it is likely that SB-10 recognizes one of these conformation-dependent regions.

Molecular cloning of a full-length cDNA from our human MSC expression library confirmed the nucleotide sequence identity with ALCAM. The isolated clone represents a variant containing a substitution of serine for asparagine at residue 231; however, it is possible that additional forms may exist in the MSC cDNA library that correspond to the polymorphisms previously noted.(23) We determined that the mRNA coding for human ALCAM has an apparent size of 6.1 kb, whether derived from fetal tissue, cultured MSCs, or the HL60 cell line, a finding in contrast to that of Bowen and colleagues,(23) who estimated the size to be 5.2 kb. Using a PCR-based approach, we identified ALCAM homologs in rat, rabbit, and canine MSC cultures, each of which contain a single mRNA species of ∼5.8 kb. One may therefore infer, from the highly conserved peptide sequence in different species, that ALCAM plays an important role in the development of a variety of mesenchymal tissues.

The family of CAMs described in this report demonstrate a temporal expression pattern during development that is strictly regulated. Neurolin is highly abundant in embryonic zebrafish brain, but much less so in adult.(35) In chick embryogenesis, BEN is expressed not only in the CNS but on preosteogenic cells of the cranium. Unlike ALCAM, which we have shown to reside on the surface of osteoprogenitor cells of the head, spine, and limbs,(22) BEN is absent from the developing limb bones of aves.(34) Although we and others have detected ALCAM in sections of developing human brain, skin, esophagus, and lung,(22,32) this is the first report describing the expression of ALCAM on human mesenchymal stem cells. Furthermore, since studies of lymphocyte ontogeny reveal that neither T cells nor B cells are present in human tissues prior to 10 weeks of gestation,(38,39) lymphocytic contamination of our samples used for immunohistochemical or molecular analyses of ALCAM expression is not a consideration.

The morphogenesis of bone results from the differentiation of ALCAM-positive MSCs that reside in the outer layer of the embryonic periosteum. MSCs are spindle-shaped and tightly packed in this region, but become cuboidal and less densely arranged as they down-regulate ALCAM expression, emerge from the periosteum to up-regulate ALP, and secrete osteoid matrix as mature osteoblasts. We explored the hypothesis that a decrease in the cellular adhesion of MSCs in the periosteum would foster their migration into, and differentiation within, this inner cambium periosteal layer. Our results confirm that the addition of SB-10 (anti-ALCAM) Fab fragments to human MSCs undergoing osteogenesis in vitro stimulates the differentiation process. Expression of cell surface ALP is more abundant and occurs sooner in cultures exposed to SB-10 Fabs than cultures treated with control Fabs. The data also show that mineralization is accelerated upon the addition of SB-10 Fabs. Since anti-ALCAM antibodies do not induce differentiation in the absence of osteoinductive supplements, it is unlikely that the disruption of cell-to-cell binding alone is capable of initiating the differentiation cascade. Furthermore, the combined homotypic ALCAM and heterotypic ALCAM-CD6 binding are just two of the mechanisms responsible for undifferentiated MSC cell adhesion. Recent studies of other CAMs during osteogenesis establish that they too are exquisitely controlled in a specific developmental sequence. NCAM is present on proliferative preosteoblastic cells, but is lost as they become osteocytic.(40) OB-Cadherin, another homotypic binding protein, is scarce in osteoprogenitor cells but is up-regulated in concert with ALP, an observation that may indicate a coordinate expression mechanism.(41) Furthermore, mature human osteoblasts possess ICAM-1 and VCAM-1 on their surface, serving as potential anchors for T cells that mediate homeostatic processes in bone metabolism.(42) Finally, MSCs in the bone marrow may interact with hematopoietic cells through either the CD6 receptor or homotypic ALCAM binding, thereby establishing a molecular mechanism for interaction between these unique cell types.

Acknowledgements

We thank Earl Lawrence, Jian Zhang, and Fred Shemer for expert technical assistance, Dr. Leslie Kerrigan for helpful discussions regarding this work, and Dr. Daniel R. Marshak for critical reading of the manuscript. This work was supported by Osiris Therapeutics, Inc.

Ancillary