HYAL2, a Human Gene Expressed in Many Cells, Encodes a Lysosomal Hyaluronidase with a Novel Type of Specificity*

Using Expressed Sequence Tags (ESTs) deposited in the data banks, a cDNA has been assembled that encodes a protein related to the hyaluronidases from bee venom and mammalian sperm. Expression of this cDNA yielded a polypeptide termed HYAL2, which is located in lysosomes. The HYAL2 protein was shown to have hyaluronidase activity below pH 4. However, it only hydrolyzed hyaluronan of high molecular mass from umbilical cord, rooster comb, and a Streptococcus strain. The reaction product was a polysaccharide of about 20 kDa, which was further hydrolyzed to small oligosaccharides by the sperm hyaluronidase. Conversely, hyaluronan fragments from vitreous humor, which had a molecular mass of about 20 kDa, were not cleaved by the HYAL2 enzyme to any detectable extent. These results provide evidence for the existence of structural domains in hyaluronan, which are resistant to the action of this enzyme. The structural and functional implications of these findings are discussed.

Hyaluronidases have been isolated from many different sources such as mammalian testes and serum, snake and insect venoms, salivary glands of leeches, and pathogenic streptococci (1)(2)(3). These enzymes degrade hyaluronan (hyaluronic acid (HA) 1 ), a glycosaminoglycan present in the extracellular matrix of vertebrates to oligosaccharides (4). The sequences of two related hyaluronidases from animal cells have recently be elucidated via cDNA cloning. These are the enzymes from honeybee venom (5) and from mammalian testis (6,7). The testicular enzyme originally termed PH-20 (8) is located at the head of the sperm; upon contact with the egg, it hydrolyzes the HA present in its outermost cumulus layer (9). The PH-20 protein is normally expressed only in mammalian testis. However, it has recently been shown that this enzyme is present in some tumor cells (10). A related hyaluronidase termed HYAL1, which is present in human serum, has recently been characterized (11).
In recent years, several partial cDNA sequences have been deposited in the data banks that are derived from genes that potentially code for polypeptides related to the PH-20 hyaluronidase. Three of these genes are located close to each other on chromosome region 3p21.3 (12). This region is deleted in some cell lines from small cell carcinomas of the lung (13). The genes were thus provisionally termed LuCa-1, -2, and -3. The serum enzyme HYAL1 is the product of the LuCa-1 gene (11). Starting from commercially available ESTs we have now assembled the complete LuCa-2 cDNA. Expression of this cDNA yields a hyaluronidase with a rather unusual substrate specificity, which is located in lysosomes. It is proposed to replace the term LuCa-2 by HYAL2 for both the gene and its product.

EXPERIMENTAL PROCEDURES
Materials-HA from human umbilical cord, human vitreous humor, rooster comb, and Streptococcus zooepidemicus as well as hyaluronidase from bovine testis, i.e. PH-20 (14), were purchased from Sigma. Radioactive HA was prepared using recombinant DG42 HA synthase and [ 14 C]UDP-glucuronic acid (NEN Life Science Products) as described previously (15).
Assembly of the HYAL2 cDNA-DNA and protein sequences of human PH-20 (GenBank S67798) were compared with the EST subset of the GenBank library using Smith-Waterman algorithm. Clones of interest were purchased from Research Genetics, Inc. EST 38838 (Gen-Bank R51257) overlapped at the 5Ј-end with yet another EST cloned by The Genexpress cDNA program (GenBank F11962). In addition, they all matched a cDNA deposited under the name of LuCa-2 (Gen-Bank U09577). The full-length HYAL2 cDNA could be assembled from the EST sequence R51257 by PCR amplification with 5Ј-specific primers. DNA was sequenced using the chain termination method with T7 DNA polymerase (Amersham Pharmacia Biotech). For multiple tissue Northern blots (CLONTECH), DNA was labeled with [␣-32 P]dATP and hybridized to poly(A) ϩ RNA on Nylon filters at 68°C using standard procedures. Filters were washed twice with 2 ϫ saline/ sodium phosphate/EDTA at 65°C and exposed to x-ray films at Ϫ70°C.
Expression of the HYAL2 Protein in Escherichia coli-Using two synthetic oligonucelotides (GAG AGG ATC CAT GCC CCA AGG CTT TAG G and GAG AGA AAG CTT CAA GGT CCA GGG TAA AGG CCA GG) and the clone EST 38838, a DNA fragment was amplified by PCR that coded for the HYAL2 protein without the putative signal sequence. The amplified DNA fragment and the plasmid pMW172 were digested with BamHI and HinDIII, ligated with T4 ligase, and transformed into E. coli BL21(D3) as described (16). After a 3-h induction with isopropylthio-␤-galactoside, bacteria were lysed with lysozyme in 10 mM Tris, pH 7.6, 10 mM NaCl. The insoluble fraction was washed twice with 4 M NaCl, followed by washes with distilled water. The resulting pellet was enriched in insoluble inclusion bodies, which were resuspended in water and stored at Ϫ20°C.
Antibody Production-An antiserum against HYAL2 was produced in rabbits by standard techniques as described previously (16). The serum was purified by affinity chromatography as follows. Insoluble protein from inclusion bodies was dissolved in 8 M urea, 10 mM Tris, pH 8.8, and dialyzed against buffer overnight. Soluble protein was immobilized on Sepharose Q (Amersham Pharmacia Biotech), rinsed with 1 M NaCl and equilibrated with PBS. Serum was diluted in PBS and mixed with protein beads overnight at 4°C. The loaded beads were washed extensively with PBS. Antibodies were eluted with 100 mM glycine at pH 2.5. The neutralized antibody solution was stored at 4°C.
Western Blots-Various tissues and cells were homogenated in PBS. Aliquots containing two micrograms of protein were separated on 10% polyacrylamide gels in presence of SDS, blotted onto nitrocellulose filters, and incubated with diluted antiserum in phosphate-buffered saline, 200 g/ml bovine serum albumin, 0.2% Tween 20 overnight at 4°C. Filters were washed twice with phosphate-buffered saline, 0.2% Tween 20) for 20 min and then incubated with diluted goat-anti-rabbit IgG horseradish peroxidase conjugate (Bio-Rad) for 2 h and washed * 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) AJ000099.
again as before. Bound secondary antibodies were detected using the ECL Western blotting analysis system (Amersham Pharmacia Biotech).
Preparation of Recombinant Vaccinia Virus (vv) Expressing HYAL2-Using synthetic oligonucleotides (Hy-1, AGG CCC AGG CCC CAC CGT TAC ATT GGC CCT GGT GCT GGC GGT GGC ATG GGC C; Hy-2, GAG AGA GAA TTC AGC AGC CAA TCA TGA TGC GGG CAG GCC CAG GCC CCA CCG T; Hy-3, GAG AGA GAA TTC CAA GGT CCA GGG TAA AGG CCA GG) and the clone EST 38838, a cDNA fragment was amplified by PCR that contained the complete coding sequence of the human HYAL2 gene. Hy-1 and Hy-3 primers were used in the first 10 PCR cycles, and a diluted portion of the reaction product was then amplified with Hy-2 and Hy-3. The PCR fragment and the vector pgpt-ATA-18 stop3 (17) were digested with EcoRI and ligated to yield the plasmid pATA-H2. Recombinant vv was selected and prepared as described before (6,16). Cellular lysates were fractionated by differential centrifugation. The supernatant of a 10,000 ϫ g spin was used for the activity measurements. Fractions were extracted with Triton-X114, and the detergent phase was discarded. Triton X-100 was added to the aqueous phase to a final concentration of 1%, and samples were stored at Ϫ20°C. Cells were pulse-labeled with [ 35 S]methionine (American Radiochemicals), and protein products were immunoprecipitated with anti-HYAL2 serum and Staphylococcus aureus extract (Life Technologies, Inc.). Precipitated proteins were fractionated by SDS-polyacrylamide gel electrophoresis, and dried gels were exposed for 24 h at Ϫ70°C.
Assays for Hyaluronidase Activity-Initial tests for hyaluronidase activity were carried out as described (6). Hyaluronidase activity was also monitored by agarose gel electrophoresis. HA samples from different sources (50 g) were separated on a 0.5% agarose, Tris/acetate/ EDTA gel and stained with Stains-All (Sigma). DNA prepared from bacteriophage was digested with StyI and used as a size marker (18). This separation was calibrated using HA preparations of which the average molecular weight had been determined by low angle laser light scattering. We also confirmed the molecular mass of these preparations by size exclusion chromatography on Sephacryl 400HR (Amersham Pharmacia Biotech), which had been calibrated by dextran standards prepared from Leuconostoc mesenteroides (Fluka).
Expression of GFP Fusion Protein in Cell Culture-The HYAL2 cDNA was subcloned into the EcoRI site of pEGFPN2 (CLONTECH). This yielded the plasmid pHYAL2-EGFP, which expressed the fusion protein under the control of the cytomegalovirus promotor. Superfect (Quiagen) was used to stably transfect C6 cells with pHYAL2-EGFP. Transfected cells were selected in the presence of G418. Acidic compartments were labeled using Lysotracker Red DND-99 (Molecular Probes). Green and red fluoresence was distinguished using a Bio-Rad MRC600 confocal microscope equipped with an Argon laser and 100 ϫ lenses. Fluorescence was monitored by BHS and A1/A2 filtersets, respectively.

RESULTS
Assembly of the HYAL2 cDNA-From ESTs deposited in the data banks and a partial cDNA assigned to the human LuCa-2 gene on chromosome 3p21.3 (GenBank U09577) the complete coding region was assembled as described under "Experimental Procedures." The polypeptide encoded by this cDNA is 35.2% identical to the human PH-20 hyaluronidase (see Fig. 1). We suggest calling this the HYAL2 gene and hyaluronidase.
Tissue Distribution of HYAL2 mRNA and Protein-Using the HYAL2 cDNA, we could show by Northern blot analysis that mRNAs hybridizing with this probe are present in all human tissues tested, the sole exception being adult brain (see Fig. 2). Using an antiserum generated against the human HYAL2 polypeptide, extracts from different mouse tissues were tested on Western blots. A protein with the molecular mass of about 60 kDa was detected in all tissues, adult brain again being a conspicuous exception (Fig. 3). These results demonstrated both at the mRNA and the protein level that the gene encoding HYAL2 is widely expressed.
Expression of HYAL2 by Recombinant Vaccinia Virus-Recombinant vv containing the HYAL2 cDNA after the late 11,000-dalton promotor was prepared. Expression of the HYAL2 protein was then tested in HeLa and C6 glioma (see "Experimental Procedures").
In the first set of experiments, HeLa cells were infected with recombinant vv as well as with wild type virus as a control. After 18 h, cells were harvested and lysed by freezing and thawing. HYAL2 protein was detected by Western blotting, and hyaluronidase activity was assayed by turbidity measurements (21) and agarose gel electrophoresis (18).
As shown in Fig. 4A, hyaluronidase activity at pH 3.8 was about 2-3 times higher in HeLa cells infected with recombinant vv than in control cells infected with wild type vv. The enzymatic activity in the latter probably corresponds to the lysosomal hyaluronidase present in HeLa cells (22). Indeed, on a Western blot a protein with about the same molecular mass was detected in HeLa cells infected with wild type or recombinant vv (see Fig. 4D). The endogenous and the HYAL2 enzyme both have an acidic pH optimum (see also Fig. 4C). Moreover,

FIG. 2. Northern blot analysis.
A, radioactively labeled HYAL2 cDNA was hybridized to a Northern blot containing 2 g/lane of poly(A) ϩ RNA of various human tissues. 1, heart; 2, brain; 3, placenta; 4, lung; 5, liver; 6, skeletal muscle; 7, kidney; 8, pancreas. The blot was exposed to x-ray films with intensifying screens for 72 h at Ϫ70°C. B, for loading control, the same blot was stripped, hybridized with a human ␤-actin probe, and exposed to x-ray films for 90 min. after centrifugation, hyaluronidase activity could be detected in the 1,000 ϫ g as well as in the high speed pellet. Upon extraction with Triton-X114 (23), the enzymatic activity remained in the aqueous layer. This indicates that the endogenous and the HYAL2 enzymes are soluble proteins with a similar subcellular distribution.
Since adult brain does not contain detectable amounts of the HYAL2 mRNA and protein, we next used C6 glioma cells. As shown in Fig. 4B, only C6 cells expressing the HYAL2 cDNA contained hyaluronidase activity, whereas control cells or cells infected with wild type vv were negative. This was confirmed by a Western blot (see Fig. 4D, lanes 3 and 4). The enzyme activity had a pH optimum of about 3.8 in this cell line (Fig. 4C).
Hyaluronidase activity was also tested using separation of substrates and products by gel electrophoresis and subsequent staining with Stains-All. The electrophoresis was calibrated using HA preparations of known molecular mass (see Fig. 5, A  and B). Surprisingly, HA from umbilical cord was only digested to a fragment with a molecular mass of about 20,000 Da (Fig.  5C, lanes 2, 5, and 8). The same results were obtained with HA from S. zooepidemicus (see Fig. 5C, lane 9) and from rooster comb (data not shown). Upon subsequent addition of testicular PH-20 hyaluronidase, this intermediate size HA was rapidly degraded to small oligosaccharides (Fig. 5, lane 7).
Subsequently, we tested a preparation of HA from vitreous humor (Sigma) that already had a size of about 20 kDa (Fig. 5C,  lane 12). This substrate was not cleaved to any detectable extent by the HYAL2 enzyme (Fig. 5C, lane 13). We also prepared radioactively labeled HA using the recombinant DG42 HA synthase. This yields an intermediate size product comparable with the HA fragments prepared from vitreous humor (15). Labeled substrate (1500 dpm) was used in a standard reaction mixture. After incubation with the HYAL2 enzyme, the product was separated by agarose gel electrophoresis. As compared with a control incubated in the absence of enzyme, 96.4% of the radioactivity was recovered in the region corresponding to the original substrate.
To confirm this result, radioactively labeled HA was also immobilized on microtiter plates. The radioactivity released after incubation with either HYAL2 or PH-20 was measured in a liquid scintillation counter. In a typical experiment, incubation with HYAL2 or the PH-20 hyaluronidase resulted in the release of 150 and 1650 cpm, respectively (the control was 137 cpm).
We also checked the extent of hydrolysis by measuring the increase of reducing sugar termini (20). After hydrolysis of high molecular weight HA with HYAL2, the increase of reactive termini was below the detection limit of this reaction. The subsequent addition of PH-20 hyaluronidase yielded significant amounts of reducing ends. The optical density at 585 nm increased from zero after incubation with HYAL2 to 0.32 after degradation with PH-20.
Several control experiments were performed in an effort to find an explanation for this unexpected observation. Thus, at the end of the incubation with HYAL2, an aliquot was removed, boiled for 5 min, cooled rapidly and then incubated further in the presence of freshly added enzyme. No further hydrolysis of FIG. 3. Western blot analysis. A, protein extracts of different mouse tissues were separated by 10% polyacrylamide gel electrophoresis, blotted on nitrocellulose, and incubated with polyclonal antibody raised against bacterially expressed human HYAL2. Bound antibodies were detected by ECL-technology (Amersham Pharmacia Biotech). Samples used were brain (1), whole blood (2), heart (3), kidney (4), lung (5), liver (6), muscle (7), testis (8). B, protein gel stained with Coomassie Brilliant Blue containing the same samples as used in A; in addition protein length standard is shown.

FIG. 4. Enzymatic activity of HYAL2 expressed by recombinant vv.
A, hyaluronidase activity in lysates from HeLa cells infected with recombinant vv (circles) and wild type virus (triangles). B, activity in C6 glioma cells. C, enzymatic activity in lysates from C6 glioma cells at different pH values. Percent change in transmission at 600 nm is plotted against the time or pH value (C). Samples in A and B were incubated for 5 h at pH 3.8. Aliquots were used for Western blots (D). Lanes 1 and 2, HeLa cells infected with wild type and recombinant vv, respectively; lanes 3 and 4, same for C6 glioma cells the reaction product could be observed after this denaturation step. To test for possible product inhibition, we also added to the reaction mixture HA oligomers (up to 10 mg/ml) generated by digestion of HA with the PH-20 enzyme. This did not alter the activity of the HYAL2 enzyme.
As tested by gel electrophoresis, HYAL2 does not hydrolyze commercial preparations (Sigma) of chondroitin sulfates A, B, and C, heparan sulfate, and heparin (data not shown).
Biosynthesis and Subcellular Localization of HYAL2-C6 glioma and other cells (RK13,  were infected with recombinant vv. After 12 h, cells were incubated for 1 h in the presence of [ 35 S]methionine and then chased with an excess of unlabeled methionine for 3 h. Cells and media were harvested, immunoprecipitated with anti-HYAL2 serum, and subjected to gel electrophoresis. In these experiments, neither immunopre-cipitated HYAL2 protein nor hyaluronidase activity could be detected in the media (data not presented).
To learn more about the subcellular localization of this protein, an in-frame fusion of cDNAs coding for HYAL2 and the green fluorescent protein (EGFP) was constructed. C6 cells were stably transfected with this plasmid and then selected in the presence of G418. As shown in Fig. 6A, the HYAL2-EGFP fusion protein gave green signals in vesicular structures. This fluorescence colocalized with the Lysotracker Red (see Fig. 6B). Control C6 cells transfected with pEGFP showed only a diffuse green fluorescence in the cytoplasm of cells (data not shown). DISCUSSION The human HYAL2 cDNA encodes a mature polypeptide of 452 amino acids that shows 36.5% identity with the PH-20 HA of known sizes such as 2 MDa (triangles), 480 kDa (squares), and 240 kDa (circles) were compared with a preparation from bovine vitreous humor (crosses). C, hydrolysis of HA by HYAL2 was characterized by agarose gel electrophoresis. Substrate was incubated with extracts of C6 glioma cells containing recombinant HYAL2 (final volume 50 l) at 37°C for different time periods. Samples were boiled for 5 min, and one-half was subsequently separated by agarose gel electrophoresis and stained with Stains-All. To the samples indicated, 1 g of bovine testicular PH-20 hyaluronidase was added. HA from umbilical cord was incubated for 2 or 15 h, respectively, with extracts from cells infected with wild type vv (controls, 1 and 4), recombinant HYAL2 (2 and 5), and PH-20 enzyme (3 and 6). Part of the sample shown in 5 was removed after 13 h and incubated for 2 h with PH-20 (7) or boiled for 5 min and then incubated for another 2 h in the presence of fresh extract (8). HA from S. zooepidemicus (10) was digested with Hyal2 (9) or PH-20 (11). HA from vitreous humor (12) was incubated with HYAL2 (13) or PH-20 (14). Phage DNA digested with StyI (M) (fragment lengths in kbp are to the right) was coelectrophoresed and used for HA size estimation. hyaluronidase present on the head of spermatozoa. At the nucleotide level, the identity is 43.1%. The HYAL2 gene is located on chromosome 3p21.3 in a region that is deleted in most small cell carcinomas of the lung as well as other tumor cells (12,13). This gene was tentatively termed LuCa-2; in view of the results presented in this communication, we propose to now call it HYAL2. The hyaluronidase present in mouse (24) and human (11) serum was termed HYAL1 earlier. This enzyme is apparently secreted by a variety of cells.
The human HYAL2 gene is expressed in all cells tested, the sole exception being adult brain. It is noteworthy that several of the ESTs encoding part of the HYAL2 enzyme have been isolated from infant human brain. The expression of the HYAL2 gene may thus be developmentally regulated. Indeed, the mouse HYAL2 mRNA is present in brains from embryos but disappears for unknown reasons soon after birth. 2 Using recombinant vv, the HYAL2 protein has been expressed in several cell lines; here we present the results obtained with HeLa and C6 glioma cell. The HYAL2 cDNA codes for a preprotein with an amino-terminal signal peptide, yet the protein is not secreted into the media to a measurable extent. The product of this gene is a soluble hyaluronidase that resembles the lysosomal enzyme present in HeLa cells. Both have a pH optimum below 4 and react with the same antiserum. A fusion protein of HYAL2 and EGFP could be shown to be localized in lysosomes of C6 glioma cells. We conclude from these results that HYAL2 encodes a lysosomal hyaluronidase present in many cell types.
It was apparent from initial turbidity measurements that the activity of HYAL2 was rather low compared with that of the testicular PH-20 enzyme. Subsequent experiments with a gel electrophoresis assay yielded a totally unexpected result. The HYAL2 enzyme hydrolyzed only HA of high molecular mass, as is present in umbilical cord, rooster comb, and the coat of a Streptococcus strain. The reaction product was a polysaccharide of about 20 kDa, which corresponds to 50 -60 disaccharide units. After heating and rapid cooling, no further hydrolysis by freshly added HYAL2 enzyme could be observed. However, upon subsequent addition of the PH-20 enzyme, this intermediate was quantitatively degraded to small oligosaccharides. HA of smaller size as well as a synthetic product made in vitro with the DG42 HA synthase were not degraded to a detectable extent by the HYAL2 enzyme. A similar finding has been reported by Sampson et al. (24), who showed that human fibroblasts contain a hyaluronidase that degrades high molecular weight HA to products with a size of 10 -40 kDa. Interestingly, in these experiments, binding of HA to the cells appeared to be rate-limiting for degradation.
The results obtained with the HYAL2 enzyme are of some interest from both a structural and a functional point of view.
First, it appears that ordered domains are present in HA, which are resistant to the action of the HYAL2 enzyme. HA is a linear polymer that behaves in solution as a random coil containing large amounts of solvent (25). However, the flexibility of this coil is locally reduced by the presence of hydrogen bonds (26). Moreover, several groups have presented evidence that HA chains can form a complex network containing numerous helical structures (27,28). Antiparallel helices have indeed been observed in x-ray diffraction studies with HA films (29). Evidence from physico-chemical studies suggests that similar structures exist in HA solutions (25,30). Our results on the activity of the HYAL2 enzyme provide the first biochemical data in support of defined domains in HA.
Second, HA fragments generated by the HYAL2 enzyme may have distinct biological functions. One could be the stimulation of angiogenesis, which has been observed in different experimental situations (10,31). Moreover, it is of some interest that HA fragments can induce the expression of enzymes such as nitric oxide synthase via a NF-B/I-B␣ autoregulatory loop in murine macrophages (32,33). The participation of HYAL2 in this signaling process can now readily be tested.