Molecular Characterization of a Novel Intracellular Hyaluronan-binding Protein*

Hyaluronan has well defined functions in extracellular matrices and at the surface of cells. However, several studies have now shown that significant pools of hyaluronan are also present intracellularly, but its function therein is unknown. One avenue of investigation that may assist in defining the function of intracellular hyaluronan is to identify intracellular hyaluronan-binding proteins. In previous studies we identified CDC37, a cell cycle regulatory protein, using a monoclonal antibody that recognizes a novel group of hyaluronan-binding proteins. In this study, we have identified a second hyaluronan-binding protein with this antibody and characterized its properties. This protein, which we have termed IHABP4, was also found to be an intracellular and a specific hyaluronan-binding protein, containing several hyaluronan-binding motifs: (R/K)X 7(R/K) (where R/K denotes arginine or lysine and X denotes non-acidic amino acids). Furthermore, we have determined the gene organization ofIHABP4 and cloned cDNAs for the chick, mouse, and human homologs. Comparison of the deduced chick, mouse, and human protein sequences showed that the hyaluronan-binding motifs, (R/K)X 7(R/K), in these sequences are conserved; both chick and mouse IHABP4 were shown directly to bind hyaluronan. Biochemical fractionation and immunofluorescent localization of epitope-tagged IHABP4 indicated that it is mainly present in the cytoplasm. These data support the possibility that intracellular hyaluronan and its binding proteins may play important roles in cell behavior.

Hyaluronan is a ubiquitous glycosaminoglycan (GAG) 1 that is present in extracellular and pericellular matrices. On the basis of its unique physical and chemical properties, it has been well documented that hyaluronan plays a critical role in dynamic structural changes within extracellular matrices during development and tissue remodeling, as well as in maintenance of mechanical properties and homeostasis of many tissues (1,2). Interactions of hyaluronan with hyaluronan-binding proteins (HABPs) are involved in establishing and modifying the structural properties of extracellular matrices. Furthermore, it is increasingly appreciated that hyaluronan is a crucial pericellular and cell surface component that actively participates in regulating cell behavior through its interactions with cell surface HABPs such as CD44 (1)(2)(3)(4). A surprising new development, however, has been the discovery of several intracellular HABPS (IHABPs), but the interactions and functions of intracellular hyaluronan and IHABPs are not yet understood.
Convincing studies showing the existence of intracellular pools of hyaluronan in vivo and in vitro (5)(6)(7)(8) and the discovery of several IHABPs, including CDC37 (9), RHAMM/IHABP (10,11), and P32 (12), have raised an interesting question. Does intracellular hyaluronan, like extracellular hyaluronan, play an important role in regulating cellular behavior through interactions with IHABPs? Recently, it was reported that rapid uptake of Texas red-labeled hyaluronan and its subsequent accumulation in the cell processes, perinuclear area, and nucleus of transformed cells were associated with enhanced cell motility (13). Furthermore, careful examination confirmed the intracellular localization of endogenous hyaluronan as well as hyaluronan-binding sites and demonstrated dynamic changes in their pattern of distribution during proliferation of smooth muscle cells and fibroblasts (8). These data suggest strongly that intracellular hyaluronan has important roles in cellular processes.
Using the monoclonal antibody (mAb), IVd4, which recognizes a novel group of chick HABPs (14), we have previously identified and characterized CDC37, a cell cycle regulatory protein (9,15). In the present study, we have characterized a cDNA, LH21, that encodes a second novel chick HABP recognized by mAb IVd4, using similar strategies as described previously (9,15). This HABP, like CDC37, is also an intracellular protein containing multiple hyaluronan binding motifs, (R/ K)X 7 (R/K), in which B denotes arginine or lysine and X denotes any amino acid except glutamic or aspartic acid (16). Furthermore, we have cloned the mouse and human homologs of this gene; shown that the hyaluronan binding motifs are conserved among the deduced chick, mouse, and human protein sequences; and demonstrated that the chick and mouse proteins bind hyaluronan specifically. These results imply that the hyaluronan-binding properties of this protein have functional importance. We have termed this protein IHABP4, because three other IHABPs have so far been characterized: CDC37 (9), RHAMM/IHABP (10,11), and P32 (12).

EXPERIMENTAL PROCEDURES
Expression Library Screening with mAb IVd4 -A chick embryo heart cDNA library, supplied by Drs. R. Markwald and E. Krug, University of S. Carolina Medical School, was screened with mAb IVd4 as described previously (9). One of the positive clones, which was repeatedly selected * This work was supported by National Institutes of Health Grants CA73839 and CA82867 (to B. P. T.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF227683, AF227684, and AF241831.
3Ј-and 5Ј-RACE Reactions and Cloning of Mammalian Homologs-For cloning the full-length cDNA of chick LH21, a library of adaptorligated double strand cDNA was made with poly(A) ϩ RNA from day 6 chick limbs using the Marathon cDNA amplification kit according to the manufacturer's instructions (CLONTECH, Palo Alto, CA). The library was used as a template for performing both 5Ј-and 3Ј-RACE reactions based on the partial cDNA sequence of the initial clone LH21. The primer, F1, used for 3Ј-RACE was 5Ј-TTCTACCAGCTGCTGGACGAC-GAGT-3Ј, corresponding to bases 243-267 of the cDNA sequence; and the primer, R1, used for 5Ј-RACE was 5Ј-GCCGTGCAACCACTTTGCT-GAGGT-3Ј, complementary to bases 345-322 of the sequence. Unique bands obtained from 3Ј-and 5Ј-RACE were purified and ligated into the PCRII vector (Invitrogen) for sequence analysis. The full-length cDNA was then obtained by fusing the 3Ј-and 5Ј-RACE products by PCR using the Marathon kit according to the manufacturer's instructions, followed by sequencing. The sequence was verified by direct amplification of the whole cDNA sequence by reverse transcription-PCR using a primer pair at the extreme 5Ј-and 3Ј-ends of the extended cDNA sequence with the double strand cDNA library as template.
For cloning the mammalian homologs of LH21, mouse and human EST as well as GenBank data bases were searched using the chick sequence with BLAST (Basic Local Alignment Sequence Tool) search programs. A partial human cDNA sequence encoding a protein called Ki-1/57 antigen (GenBank accession number u77327) and a mouse EST sequence (accession number w29443) were found to be the most likely candidates for the mammalian homologs of chick LH21. Based on these two partial mammalian cDNA sequences, a similar strategy to that used for cloning full-length chick cDNA was used to clone fulllength cDNAs of the mammalian homologs of LH21, with some modifications. Briefly, mouse and human heart poly(A) ϩ RNAs were purchased from CLONTECH Laboratories, Inc. The cDNA templates for RACE reactions were made using the SMART RACE cDNA amplification kit (CLONTECH). For 5Ј-RACE reactions, a GC-rich PCR system was used (CLONTECH).
Sequencing and Sequence Analysis-After RACE products were cloned into the PCRII vector, the nucleotide sequences of these products were determined by the double stranded DNA/dideoxy chain termination method using a Sequenase 2.0 kit (U.S. Biomedical Corp.).
BLAST programs on the web site established by the National Center of Biological Information were used for similarity searches in the Gen-Bank and EST data bases. The SignalP program (17) was used to predict signal peptide sequences; the PSORT program was used to find possible signals for subcellular localization. For prediction of possible motifs, simple modules, and domains in the protein sequences, SMART (Simple Modular Architecture Research Tool) (18,19) and MOTIF programs were used. For multiple sequence alignment, the PRETTYBOX program of the GCG software package (version 10) was used.
Northern Blot Analysis-For chick RNA preparation, whole chick embryos as well as chick embryonic tissues (limbs and brain) were used. Total RNA and poly(A) ϩ RNA were prepared as described previously (15). Samples of the poly(A) ϩ RNA preparations were subjected to electrophoresis in a formaldehyde gel and then transferred to a nitrocellulose membrane. The membrane was probed with chick clone LH21 cDNA labeled with [ 32 P]dCTP using a random priming DNA labeling kit (Roche Molecular Biochemicals). Hybridization and washes were done under standard high stringency conditions.
For Northern blot analysis of mouse mRNAs, a nitrocellulose membrane preblotted with multiple poly(A) ϩ RNAs prepared from different mouse tissues was purchased from CLONTECH. Probe preparation, hybridization, and controls were conducted according to the manufacturer's instructions.
Determination of Gene Organizaton by PCR-The genomic region encompassing the chick LH21 cDNA sequence was obtained by amplifying several overlapping genomic fragments, because the size of this genomic region is too large for direct amplification (Ͼ20 kb). Cloning of PCR products and identification of boundary sequences between introns and exons were determined by the same strategy as previously used (15).
Construction and Expression of Epitope-tagged Recombinant Proteins-The complete coding region sequences of chick LH21 and three truncated mutant cDNAs (M1, M2, and M3; see Fig. 7A) were amplified by PCR reactions using different combinations of reverse and forward primer pairs (see below) with the cloned full-length chick LH21 cDNA as template. The resulting PCR products were first cloned into the PCR cloning vector PCRII to confirm their sequence authenticity through sequence analysis from both directions of the insert. After sequence confirmation, all inserts were cut out of the PCRII vector and subcloned into a mammalian expression vector, PCI-neo (Promega). Two reverse primers were designed to produce a common 27-bp sequence encoding a 9-amino acid hemagglutinin tag, followed by the specific chick LH21 cDNA sequences. Thus all resulting constructs could be expressed as recombinant proteins with the hemagglutinin tag at their carboxyl termini. Four constructs were made: wtChLH21-HAtag,M1ChLH21-HAtag, M2ChLH21-HAtag, and M3ChLH21-HAtag. Primers used for making these chick LH21 constructs were as follows: ChF1 primer was 5Ј-CTCGAGACCATGATGAAGGGGATGGGCTG-3Ј; ChF2, 5Ј-ATGAATCGATTCTACCAGCTG-3Ј; ChR1, 5Ј-TCAGGCG-TAATCTGGCACATCGTAAGTTAAAGCAGGAAAATCCTC-3Ј; ChR2, 5Ј-TCAGGCGTAATCTGGCACATCGTATGCTCCTCTTGATCCACA-3Ј (underlined regions in the two reverse primers indicate the sequences encoding the hemagglutinin tag). ChF1 and ChR1 were used to amplify the complete coding region sequence of chick LH21; ChF1 and ChR2 for the M1 deletion; ChF2 and ChR1 for the M2 deletion; and ChF2 and ChR2 for the M3 deletion (see Fig. 7A).
COS-1 cells or CHO-K1 cells were cultured in Dulbecco's modified Eagle's medium or F-12 medium (Life Technologies, Inc.), respectively, supplemented with 10% fetal bovine serum (Hyclone). The day before transfection, cells were subcultured in a 6-well plate to 60 -80% confluency. These subconfluent cells were transfected with highly purified plasmid DNA (Qiagen, Chatsworth, CA) and SuperFect transfection reagent (Qiagen) according to the manufacturer's instructions. At 24 -48 h post-transfection, cells were lysed with M-Per mammalian extraction reagent (Pierce, Rockford, IL) then subjected to SDS-PAGE and Western blot analysis using high affinity rat monoclonal antibody (clone 3F10) against hemagglutinin tag (Roche Molecular Biochemicals), followed by horseradish peroxidase-conjugated second antibody (goat anti-rat IgG; Pierce) and development with Supersignal chemiluminescence reagent (Pierce).
For stable transfection, transiently transfected cells prepared as above were transferred into fresh growth media containing 800 g/ml neomycin G418 (Life Technologies, Inc.) selection drug at 24-h posttransfection and were continuously grown under drug selection for about 2 weeks. Individual cell clones were picked with cloning rings and expanded with the same growth media containing 500 g/ml G418. The expression level of tagged recombinant protein in these selected stable transfectants was analyzed by SDS-PAGE and Western blotting as above.
Glycosaminoglycan-binding Assays-We used two assays to assess hyaluronan binding activity. The first was a [ 3 H]hyaluronan overlay assay as described previously (9,14). Briefly, after separation by SDS-PAGE, proteins were transferred to a nitrocellulose membrane. The blot was incubated in a solution of 1.2 ϫ 10 7 cpm [ 3 H]hyaluronan/ml of PBS, in the presence or absence of 500 g/ml hyaluronan hexasaccharide, for 16 h at room temperature, then washed thoroughly with cold PBS and dried at room temperature. The blot was then sliced into 2-mm segments, suspended in acetone, and counted.
In the second assay, binding of GAGs by epitope-tagged mouse recombinant protein in cell lysates was analyzed as described previously (20). In brief, CHO-K1 or COS-1 stable transfectants were cultured continuously under G418 selection pressure in 100-mm dishes. When confluent, the cells were washed with cold PBS and lysed with M-Per lysis buffer, 1 ml/dish (Pierce). Insoluble materials in the lysates were removed by centrifugation at 12,000 ϫ g at 4°C for 10 min. Aliquots of 200 l were incubated with 100 g of GAG (hyaluronan from rooster comb, American Seikagaku; chondroitin sulfate C from porcine rib cartilage, Sigma; or heparan sulfate from bovine intestine, Sigma), hyaluronan hexamer (14), or RNA (yeast; Roche Molecular Biochemicals) in 100 l of H 2 O. As a negative control, water was added instead of GAG or RNA solution. After 1-h incubation at room temperature, cetylpyridinium chloride (CPC) was added to a final concentration of 1% (w/v), and the solution was mixed and incubated for another hour at room temperature with slow shaking. Precipitated complexes containing GAG or RNA and bound proteins were collected by centrifugation at 12,000 ϫ g for 10 min. Pellets were washed three times with 1 ml of solution containing 1% CPC and 30 mM NaCl for each wash, and finally suspended in 50 l of 2ϫ sample buffer (2% SDS, 200 mM dithiothreitol, 120 mM Tris-HCl, pH 6.8). Samples were then subjected to SDS-PAGE and Western blot analysis with rat monoclonal antibody (3F10 clone, Roche Molecular Biochemicals) against hemagglutinin tag as described above. Untreated lysate was loaded as a positive control. The relative densities of the bands obtained in the Western blots were determined using a BioImage whole-band analysis package (Millipore).
Immunofluorescent Staining-NIH3T3 cells, cultured in 4-well chamber slides (Falcon) in Dulbecco's modified Eagle's medium plus 10% fetal bovine serum (Hyclone), were transfected with pure plasmid DNA for either amino-or carboxyl-hemagglutinin-tagged cDNA constructs. At 48 h post-transfection, cells were fixed with 4% formaldehyde for 10 min at room temperature then permeabilized with 0.3% Triton X-100 in PBS for 10 min. The expressed protein was detected by rat monoclonal antibody (3F10 clone, Roche Molecular Biochemicals) against hemagglutinin tag followed by secondary antibody conjugated with rhodamine (Pierce). Controls for protein staining were non-permeabilized cells, preincubation of primary antibody with tag peptide, or incubation with rhodamine-conjugated secondary antibody alone.

RESULTS
Cloning of Chick LH21 cDNA-In previous studies (9, 15), we reported the molecular characterization of a chick IHABP, CDC37. The cDNA for CDC37 was isolated from a chick embryo heart cDNA library by immunoscreening with mAb IVd4. This antibody was selected by screening hybridomas raised against a partially purified preparation of chick embryo HABPs in the presence and absence of polymeric and oligomeric hyaluronan. It recognizes a novel group of HABPs present in a variety of tissue and cell types, especially in the embryo (14). Here we report characterization of a cDNA, LH21, for a second IHABP recognized by mAb IVd4, which we have termed IHABP4.
We chose to investigate the cDNA clone LH21, because, like the original clone for CDC37, it displayed consistently strong immunoreactivity with mAb IVd4 during successive rounds of immunoscreening of the chick heart cDNA expression library. The LH21 clone was isolated and sequenced by the same strategy as described previously for CDC37 (9). The size of the cDNA insert was about 1.2 kb, but initial examination of the sequence suggested that the insert might be a partial cDNA sequence. Northern blot analysis was performed using labeled LH21 cDNA to hybridize with mRNA preparations from chick embryonic brain and limb, and from the whole embryo. The result (Fig. 1B) showed a prominent band slightly larger in size than 18 S ribosomal RNA, and a weak band similar in size to 27 S ribosomal RNA. Because mRNA was used rather than total RNA, the signals were not due to ribosomal RNAs. Furthermore, in a separate Northern blot analysis, total RNAs prepared from chick whole embryo, brain, and limb were loaded in the same gel together with the mRNA preparations. Strong signals were detectable only from the mRNA samples, not from the total RNA samples (data not shown), indicating that the transcription level of LH21 in the embryo is relatively low, and confirming that the signals from the mRNA preparations in Fig. 1B were specific. By comparing the size of the LH21 transcript (ϳ2.0 kb as estimated from Fig. 1B) with that of the LH21 cDNA we had cloned (ϳ1.2 kb), it was clear that clone LH21 was incomplete.
To extend the sequence of cDNA LH21, we designed primers F1 and R1 (Fig. 1A) for 3Ј-and 5Ј-RACE reactions, respectively, according to the known partial sequence of the clone. At least 10 independent clones for each PCR product from 3Ј-and 5Ј-RACE reactions were sequenced. By fusing the 3Ј-and 5Ј-RACE products through a PCR reaction, we now obtained an extended LH21 cDNA of ϳ1.8 kb. To eliminate any possible artifact during PCR and cloning, the cloned cDNA sequence was further confirmed by direct amplification of the whole cDNA by reverse transcription-PCR, using a chick embryo limb cDNA library as template. The size and sequence of this di- rectly amplified PCR product were the same as that of the fused cDNA. The size of the extended LH21 cDNA (ϳ1.8 kb) plus a poly(A) tail is consistent with that of the major band from Northern blot analysis (ϳ2.0 kb, in Fig. 1B). The putative translational start codon conforms to the Kozak consensus sequence around a translation initiation site; furthermore, the presence of an in-frame stop codon, upstream of the start codon in the 5Ј-untranslated region, supports the prediction of an open reading frame. At the end of the 3Ј-UTR is a typical polyadenylation signal, AATAAA. Thus, we believe that this cDNA represents the full-length LH21 cDNA for chick IHABP4. The GenBank accession number for the chick sequence is AF227683.
The extended cDNA sequence described above contains an open reading frame of 1071 bases, encoding a putative 357amino acid polypeptide with an expected molecular mass of ϳ42 kDa (Fig. 1A). Lysates of COS-1 cells transfected with epitope-tagged LH21 were analyzed by Western blotting with antibody against the hemagglutinin tag. The molecular mass of the chick LH21-encoded protein, IHABP4, was found to be ϳ45 kDa (see Fig. 7B, lane 1), consistent with the size of cDNA.
Characterization of Chick IHABP4 Gene-To analyze the organization of the IHABP4 gene, we used a long PCR strategy by which we determined CDC37 gene organization (15). Because the chick genomic region covering the LH21 cDNA could not be directly amplified by a single long PCR, numerous primers in both forward and reverse directions were designed along the cDNA sequence. Several genomic DNA fragments (Ͻ10 kb) were amplified by long PCR using different sets of these primers, and cloned into the PCR2.1 vector for sequencing. The boundaries at both ends of a given exon were determined by comparing partially sequenced genomic sequences and cDNA sequences, and by PCR with primers that were either complementary to the cDNA sequence immediately upstream from the 5Ј-end of an exon or identical to the cDNA sequence immedi- ately downstream from 3Ј-end of an exon. The organization of the IHABP4 gene is shown in Fig. 2. The size of the genomic region covering the LH21 cDNA is about 20 kb. There are at least seven exons and six introns in the IHABP4 gene, and all boundaries between introns and exons comply with the GT (5Ј-end of the intron) and AG (3Ј-end of the intron) rules for splicing sites (Fig. 2B). The size of each intron was determined by PCR as described previously and found to range from ϳ1.0 to ϳ6.0 kb.
Cloning of Mouse and Human IHABP4 Homologs-To determine whether any possible homologs of chick IHABP4 were present in the data bases, a similarity search was performed using BLAST programs. Significant similarity to the deduced LH21-encoded protein sequence was obtained for a partially characterized human protein of unknown function, called Ki-1/57 (21). Ki-1/57 is an intracellular protein recognized by the Ki-1 monoclonal antibody originally developed against an apparently unrelated cell surface protein, CD30 (22). However, the available cDNA sequence encodes only a partial protein sequence; the sequence encoding the amino-terminal part of the protein is missing due to difficulties encountered in cloning the full-length cDNA (21). In addition, we found several newly released mouse EST sequences, which have very high similarity to human Ki-1/57 cDNA. Based on the above sequence information, we assumed that the cDNA sequence of human Ki-1/57 and the mouse EST sequences could represent portions of cDNAs for mammalian homologs of chick IHABP4, and we began attempts to clone the full-length cDNA sequences for both human and mouse IHABP4.
We first assembled selected mouse EST sequences to obtain a maximum partial sequence, and then confirmed the assembled sequence by direct reverse transcription-PCR using mouse heart mRNA as template. Based on this resulting mouse cDNA sequence, primers for 5Ј-and 3Ј-RACE reactions were designed to extend the cDNA using the same approach as that used in chick LH21 cloning. As a result, the 3Ј part of mouse cDNA was successfully obtained from 3Ј-RACE reactions, but the 5Ј-RACE reactions did not give any distinct band even after numerous trials with different primers, high temperature RT reactions, and hot-start Taq DNA polymerase enzymes. Similarly, the authors who partially cloned human Ki-1/57 cDNA could not obtain the 5Ј part of the human cDNA and suggested that a rigid secondary structure at the 5Ј-end of the transcript might prevent reverse transcription by conventional methods (21). Because the nature of the 5Ј part of the mouse cDNA might be very similar to that of the human cDNA and because rigid GC-rich regions are often found at the 5Ј region of mRNAs, we introduced a commercially available GC-rich PCR system into our 5Ј-RACE reactions. This strategy enabled us to obtain the 5Ј part of mouse cDNA (Fig. 3A). As we predicted, the cloned 5Ј region (ϳ400 bp from the 5Ј-end) of mouse cDNA shows very high GC content (Ͼ75%, see Fig. 3A). Interestingly, multiple deletions reducing the GC content and overall size of this region were found in the corresponding part of chick LH21 cDNA (Figs. 1A and 4), thus explaining the lack of problems encountered during chick LH21 cDNA cloning. The cloned mouse cDNA is about ϳ2.5 kb in size and contains a 1233-base open reading frame region encoding 411 amino acids (Fig. 3A). The GenBank accession number for the mouse sequence is AF227684. To confirm that the deduced mouse IHABP4 protein is encoded by the cloned cDNA, we made two epitope-tagged mouse IHABP4 constructs. One of these is tagged with hemagglutinin at the amino terminus and the other at the carboxyl terminus so as to eliminate possible interference with expression, hyaluronan binding, or immunostaining by the 9-amino acid hemagglutinin tag at either end of the protein. Transient expression of mouse IHABP4 in CHO-K1 cells showed that the recombinant protein was equally expressed from either the N-tagged or C-tagged construct, and the size of the protein in each case was estimated to be ϳ55 kDa (Fig. 3B). Thus the mouse IHABP4 protein is indeed encoded by the cloned cDNA. The molecular mass of mouse IHABP4 is in good agreement with the molecular mass of human Ki-1/57 (57 kDa) reported previously (21).
We also completed the missing 5Ј part of the partial cDNA sequence of human Ki-1/57 by the same strategy used for cloning mouse cDNA. The deduced peptide sequence was extended from an original 299 amino acids to 413 (Fig. 4). Because extremely high similarity (ϳ90% identity) was found between mouse IHABP4 protein and the completed human Ki-1/57 antigen (Fig. 4), we regard the Ki-1/57 as the human IHABP4 homolog. The GenBank accession number for the human sequence is AF241831.
In the deduced mouse IHABP4 protein, as well as in the extended human Ki-1/57 protein, the hyaluronan-binding motifs, -(R/K)X 7 (R/K)-, were found to be well conserved in corresponding regions to those in chick (Fig. 4). This supported the possibility that IHABP4 may be a novel hyaluronan-binding protein. Comparison of chick IHABP4 with its mammalian homologs (Fig. 4) reveals ϳ48% identity and ϳ60% similarity between chick and mouse or chick and human proteins. The carboxyl-terminal region (ϳ130 amino acids) is more conserved among all three species (ϳ80% identity) than the rest of the protein and also contains a virtually identical -(R/K)X 7 (R/K)motif, KAVVIHKSK(R), suggesting that this region might be a critical region for the function of this protein.
We conducted further BLAST searches using the mouse and human sequences. In addition to Ki-1/57, another human pro-tein, CGI-55 (GenBank accession numbers AAD34050 and NP056455), was found to have a high degree of homology to IHABP4 but is of unknown function. Partial homology was also FIG. 5. Tissue distribution of IHABP4 expression. Northern blot of poly(A) ϩ RNA isolated from several mouse adult tissues (CLON-TECH) was hybridized with a mouse cDNA probe. The blot was washed under high stringency and exposed for 48 h at Ϫ70°C with two intensifying screens. A major transcript of mouse IHABP4, ϳ2.5 kb, and a minor transcript of ϳ5.5 kb were observed in mouse heart, brain, liver, and kidney; no signal was detectable in mouse spleen, lung, or skeletal muscle. A much smaller major transcript, ϳ1.5 kb, and two minor transcripts, ϳ2.5 and 4.0 kb, were observed in mouse testis. The relative positions of RNA molecular weight markers are indicated at the left of the blot.

FIG. 4. Comparison of chick IHABP4 with mammalian homologs.
Identical amino acid residues at the same positions in the sequences are marked with black boxes; conserved residues are marked with shadowed boxes. Dots indicate gaps or deletions in the sequences. Hyaluronanbinding motifs are underlined. Note: the hyaluronan-binding motif near the amino termini of the three species is marked with a dashed line. In this case, the chick motif (residues Lys 49 to Arg 58 ) is not strictly aligned with the human and mouse motifs (residues Arg 49 to Arg 58 ). GenBank accession numbers for these sequences are AF227683 (chick), AF227684 (mouse), and AF241831 (human). noted with regions of some putative nucleic acid-binding proteins from invertebrates, e.g. the Drosophila gene products vig (accession number AAF44918) and CG11844 (accession number AAF56404). Although these proteins have some relationship to RNA-binding proteins, they do not contain RNA recognition domains as assessed using the SMART program. We also used the SMART program to search for RNA recognition domains in IHABP4 but found none. We used several programs to search for other possible motifs and simple modules in IHABP4 to obtain clues as to its subcellular localization and its structural relationship to other proteins, but no such motifs were revealed.
Expression of IHABP4 mRNA in Mouse Tissues-To investigate mouse IHABP4 expression pattern in mouse tissues, and to examine whether the cloned mouse cDNA is close in size to its transcript, Northern blot analysis was conducted. As shown in Fig. 5, mouse heart, brain, liver, and kidney tissues express a prominent transcript of ϳ2.5 kb, consistent with the size of cDNA we have cloned. The origin of a faint band at Ͼ5.0 kb is unclear. However, spleen, lung, and skeletal muscle tissues do not express IHABP4 mRNA, or express very low levels, indi-cating that IHABP4 is unlikely to be a general housekeeping gene. To our surprise, the major transcript of IHABP4 in testis is much more abundant and smaller (ϳ1.35 kb) than that in other tissues. Two minor transcripts around 2.5 and 4.0 kb are also expressed in the testis. The reasons for these differences are not known, but alternative splicing, promoter usage, or poly(A) usage are possibilities.
Hyaluronan-binding Properties of IHABP4 -As discussed above, the amino acid sequence of IHABP4 includes several conserved hyaluronan-binding motifs. However, because it is not yet known whether these motifs are in fact hyaluronan-binding sites within IHABP4, we further examined whether IHABP4 protein actually binds hyaluronan, using two approaches.
For the first procedure, we expressed the recombinant chick IHABP4 as a fusion protein in bacteria (9). The bacterial proteins were then separated by SDS-PAGE, transferred to nitrocellulose membrane, and cut into strips. Separate strips were overlaid with [ 3 H]hyaluronan or the antibody IVd4, which recognizes IHABP4 (9, 14). A single band was obtained in the IVd4 Western blot, and this band corresponded with the position of [ 3 H]hyaluronan binding (Fig. 6A). Addition of hyaluronan oli-FIG. 6. Hyaluronan-binding activity of IHABP4. A, bacterial lysate containing chick LH21 fusion protein was prepared as described previously (9). An aliquot of the lysate was electrophoresed, transblotted, and overlaid with [ 3 H]hyaluronan (solid bars), whereas another blot prepared from an identical aliquot of the lysate was incubated with [ 3 H]hyaluronan in the presence of hyaluronan hexasaccharides (hatched bars). Each blot was cut into 2-mm segments for measurement of radioactivity, which was then plotted against the distance of migration during electrophoresis. The inset shows a Western blot (W) of a third blot after incubation with mAb IVd4, and the total bacterial lysate (T) after electrophoresis and staining with Coomassie Blue. Note that the peak of [ 3 H]hyaluronan binding corresponds to the position of the IVd4-positive band in the Western blot. B, cell lysate aliquots from CHO-K1 cells stably transfected with N-tagged mouse IHABP4 construct were incubated with different GAGs, RNA, or hyaluronan oligosaccharides and subjected to CPC precipitation. Coprecipitated proteins were subjected to Western blot analysis with antibody against the hemagglutinin tag. The extreme right lane contains an aliquot of cell lysate without addition of CPC or GAG. Densitometry was performed on these blots, and the relative amounts of IHABP4 binding, expressed as a fraction of that obtained with hyaluronan, are given at the bottom of each lane. HA, hyaluronan; HS, heparan sulfate; CS, chondroitin sulfate; CPC, cetyl pyridinium chloride.
gosaccharides inhibited binding of [ 3 H]hyaluronan, indicating that this binding was specific. In addition, extracts of bacteria lacking IHABP were analyzed and found to exhibit no detectable binding of IVd4 or [ 3 H]hyaluronan (data not shown).
In the second procedure, we produced stable transfectants expressing N-and C-tagged IHABP4. We then tested binding of both N-tagged and C-tagged mouse IHABP4 to various GAGs, using a solution assay in which cell lysates containing tagged IHABP4 are mixed with GAG, and then co-precipitation of IHABP4 with GAG by CPC is measured (20). As can be seen in Fig. 6B, N-tagged mouse IHABP4 was efficiently co-precipitated after preincubation with exogenous hyaluronan and addition of CPC. However, very small amounts of IHABP4 were co-precipitated with heparan sulfate or chondroitin sulfate. The amounts of co-precipitated IHABP4 obtained in the presence of the various reagents was measured by densitometry and expressed relative to the amount co-precipitated with hyaluronan. The percentages relative to hyaluronan were 0.9% with heparan sulfate, 4.0% with chondroitin sulfate, and zero with hyaluronan oligomers, indicating that the interaction between mouse IHABP4 and hyaluronan is specific and not sim-ply due to charge interactions. We also performed the assay using C-tagged mouse IHABP4 and obtained the same result as above (data not shown). These results demonstrate that IHABP4 is a specific HABP.
Because IHABP4 may be related to putative nucleic acidbinding proteins, we also determined whether IHABP4 binds RNA. To do this we used the second procedure above, because RNA binds to CPC in a similar fashion to GAGs. Some binding of IHABP4 was obtained but to a much smaller degree (6.7%) than with hyaluronan (Fig. 6B).
Intracellular Localization of IHABP4 -A 22-amino acid hydrophobic stretch that complies with the Von Heijne (23) Ϫ3, Ϫ1 rule for signal peptide cleavage is present at the amino terminus of the deduced IHABP4 protein. However, analysis by the more rigorous SignalP program (17) indicates that the amino terminus is unlikely to contain a signal sequence. We also suspected that a phosphatidylinositol glycan (GPI) anchorage sequence might lie at the carboxyl terminus of IHABP4 based on the , ϩ 2 rules of Gerber et al. (24). Because of the uncertainty with respect to signal peptide and GPI anchorage sequences, we made epitope-tagged wild type (wt) chick IHABP4 as well as mutated forms with the putative signal and/or GPI sequences deleted (Fig. 7A). We then transiently expressed the epitope-tagged proteins in COS-1 cells and analyzed their distribution in secreted and cell lysate fractions. We expected that, if wt IHABP4 is an intracellular protein, the expressed proteins would be found exclusively in the cell-associated fraction whether or not deletions were made. If it is a secreted protein, wt IHABP4 would appear in the medium but the M2 and M3 mutants would be cell-associated. If it is bound to the plasma membrane via a GPI anchor, wt IHABP4 would be cell-associated but the M1 mutant would be in the medium. The recombinant wt IHABP4 protein and all three mutated proteins were detected only in the cell lysate fractions, not in the media (Fig. 7B), suggesting that IHABP4 is most likely an intracellular protein.
To further investigate the subcellular localization of chick IHABP4 in COS-1 cells, stable transfectants of epitope-tagged wt IHABP4 and its mutant forms were generated. These transfectants were lysed and then fractionated by differential centrifugation. Western blot analysis again showed that the majority of expressed protein (ϳ70%) was in the cytosolic fraction; about 30% was found in membrane and other organelle fractions. We conclude that chick IHABP4 is an intracellular protein, mainly present in the cytoplasm. Stable transfectants of N-and C-tagged mouse constructs in CHO-K1 cells were also used for subcellular fractionation. Subcellular fractionation gave a similar distribution as found for chick IHABP4, i.e. the majority (ϳ70%) of the IHABP4 protein was found in the cytosolic fraction and ϳ30% in other membrane and organelle fractions (data not shown).
We used immunocytochemistry to examine the distribution of IHABP4 in cells transiently transfected with hemagglutinintagged IHABP4. IHABP4 was visualized with antibody to hemagglutinin tag. IHABP4 staining was seen in the cytoplasm as a diffuse, network-like pattern, especially in the perinuclear region (Fig. 8). IHABP4 staining was completely eliminated in the presence of the peptide tag or by omitting primary antibody; also, no IHABP4 staining was observed in non-permeabilized cells (data not shown). DISCUSSION In this study, we have characterized a novel and specific IHABP. Until the function of this protein is better understood, we propose to call it IHABP4, because three other IHABPs have been convincingly characterized, namely CDC37 (9), RHAMM/IHABP (10,11), and P32 (12). IHABP4 contains multiple hyaluronan-binding motifs throughout its sequence, notably one stretch with three overlapping motifs in the middle region of the protein sequence. These motifs are conserved in the chick, human, and mouse sequences, suggesting that IHABP4 may have a hyaluronan-related function. This conclusion is supported by our data showing that both chick and mouse IHABP4 bind hyaluronan in a specific fashion. However, regions within IHABP4 may be related to putative nucleic acid-binding proteins, e.g. the Drosophila vig gene product. Although IHABP4 does not contain an established RNA recognition domain, we tested IHABP4 for RNA-binding activity and found it to exhibit weak binding compared with hyaluronan. Nevertheless, until further functional work is carried out, the question of whether hyaluronan is the natural ligand for IHABP4 must remain open. The homologies found between IHABP4, the human protein CGI-55, and the above-mentioned Drosophila proteins suggest that IHABP4 belongs to a family of related genes.
Unlike the other known IHABPs referred to above, IHABP4 mRNA is not ubiquitously expressed in adult tissues. IHABP4 mRNA is found at significant levels in adult heart, brain, liver, kidney, and testis, as well as in embryonic tissues, but not in adult spleen, lung, or skeletal muscle. A very prominent transcript that is smaller than that found elsewhere was seen only in testis. It is not yet known whether a different protein product from that found in this study results from this smaller transcript. The human homolog, also known as Ki-1/57, is expressed in activated, but not resting, human lymphocytes and is enriched in some tumor cells (21). Obviously, to understand the function of this novel HABP, further investigation will be needed, including examination of the expression levels of its mRNA and protein in various normal cells versus corresponding tumor cells, as well as its interaction with other regulatory or structural protein(s). In reference to the latter, previous evidence suggests that Ki-1/57 is associated with intracellular kinase activity (21).
A considerable amount of evidence has shown that sulfated GAG chains and proteoglycans are present in the cytoplasm and nucleus of a variety of cell types and that in some cases these intracellular GAG populations are likely to be involved in regulation of cell behavior. For example, it has been shown that heparan sulfate can act as a trans-repressor that interferes with the action of c-Fos and c-Jun on transcription events in vitro, and preliminary evidence has suggested that endogenous nuclear heparan sulfate has such a regulatory function in vivo (25). In support of this possibility, specific inhibitory subpopulations of heparan sulfate are targeted to the nucleus during cell proliferation (26,27). Also, dynamic changes occur in the levels of biglycan and glypican proteoglycans in the nucleus during the cell cycle, and the core proteins of these proteoglycans have been shown to contain nuclear localization motifs (28). There is also definitive evidence that heparin is essential for retention of several proteases present within secretory granules of mast cells (29).
It has become increasingly evident that hyaluronan is also FIG. 8. Subcellular localization of transiently expressed IHABP4. NIH 3T3 cells were transfected with N-tagged mouse IHABP4 construct and grown for 48 h. A, fixed, permeabilized cells were stained with rat monoclonal antibody against hemagglutinin tag followed by goat secondary antibody conjugated with rhodamine. B, phase contrast image of the same field. Note that only some of the cells express IHABP4 (two in this field), because the cells are transiently transfected and not cloned. The remaining cells provide an internal control for background staining. present in the cytoplasm, nucleus, and other organelles in various types of tissues and cells (5)(6)(7)(8). Recent studies provide evidence that exogenous hyaluronan added to transformed 10T1/2 cells accumulates in multiple subcellular compartments and directly affects cell motility (13). Moreover, it has been shown that there is a dramatic increase in endogenous intracellular hyaluronan and redistribution from nucleolus to cytoplasm (mainly the perinuclear area) and interchromosomal regions during mitosis of mitogen-stimulated smooth muscle cells and fibroblasts (8). There is also an increase in pericellular hyaluronan during cell division (30,31); however, extracellular hyaluronan that becomes internalized after addition to dividing cells is distributed in a different pattern to that of endogenous intracellular hyaluronan (8). Similar changes in the amounts and distribution of intracellular hyaluronan-binding sites also occur in dividing cells (8). Thus, it is of interest that Ki-1/57 is localized in the cytoplasm, nucleus, and nucleolus of Hodgkin's and myeloma cell lines (32), indicating that hyaluronan and Ki-1/57, i.e. IHABP4, may have overlapping distributions. Future studies will establish whether hyaluronan and IHABP4 interact during cell division.
The data discussed above suggest a possible role for intracellular hyaluronan and IHABPs in cell division. However, it is not clear the extent to which hyaluronan is directly targeted to various intracellular sites after synthesis rather than being first secreted and then re-internalized. Synthesis and interactions of pericellular hyaluronan have been shown to promote cell detachment and rounding during mitosis (30,31), anchorage-independent growth in culture (33,34), and tumor growth and progression in vivo (33)(34)(35)(36)(37). Yet hyaluronan internalization has also been implicated in some of these processes (36,38,39). Possibly, a dynamic balance between pericellular hyaluronan and intracellular hyaluronan exists, and hyaluronan reinternalization might contribute to this balance. This dynamic balance could in turn be involved in regulating various cellular behaviors. In this regard it is noteworthy that internalization into endothelial cells of a complex of heparan sulfate-proteoglycan and a heparin-binding protein targets the latter to mitochondria, consequently preventing apoptosis (40). Also, as mentioned above, exogenously added hyaluronan can be targeted to various intracellular compartments and results in increased cell motility (13). Interestingly, targeting of hyaluronan to the nucleus was blocked by (R/K)X 7 (R/K)-containing peptide in the latter study, suggesting the possible involvement of an IHABP. Although re-internalization of hyaluronan, and other GAGs, is likely to contribute to intracellular pools, the question of how these polymers reach sites such as the cytosol, mitochondria, or nucleus remains unclear. One possibility is that they are transported in a retrograde manner through the Golgi, endoplasmic reticulum, and cytosol in a similar manner to some protein toxins or by other less understood, "non-classical" routes (41,42).
Although re-internalization subsequent to secretion is one likely pathway whereby GAGs become localized to intracellular sites, recent evidence (8) indicates that it is unlikely to be the sole pathway for distribution of intracellular hyaluronan. Another possible mechanism for intracellular targeting of hyaluronan arises from its unique mechanism of synthesis. Hyaluronan is synthesized at the cytoplasmic face of the plasma membrane, and secretion takes place by extrusion during polymer elongation (43). It is thus conceivable that a subpopulation of hyaluronan polymer is directly deposited in the cytosol rather than crossing the plasma membrane; this in turn may be regulated by interaction with IHABPs.
The most important, and puzzling, question arising from the various studies discussed above is the exact mechanism whereby intracellular hyaluronan might influence cellular events. It is possible that the high level of hydration associated with hyaluronan (2) also plays a role in structural changes in the cytoskeleton or nuclear matrix during cell division or motility, e.g. in regulating cell shape or volume changes. Interestingly, recent evidence suggests that hyaluronan-associated changes in hydration play an important role in growth plate expansion during bone development (44) but that most of the hyaluronan at this site is intracellular (45). Irrespective of the role of intracellular hyaluronan, it is to be expected that its functions will be regulated and/or mediated by IHABPs. However, direct evidence for intracellular interaction in situ and for the functional consequence of these interactions is needed to elucidate this possibility.