Expression and Characterization of a Soluble, Active Form of the Jaagsiekte Sheep Retrovirus Receptor, Hyal2

Retrovirus entry into cells is mediated by specific interactionsbetween virus envelope glycoproteins and cell surface receptors.Many of these receptors contain multiple membrane-spanning regions,making their purification and study difficult. The jaagsiektesheep retrovirus (JSRV) receptor, hyaluronidase 2 (Hyal2), isa glycosylphosphatidylinositol (GPI)-anchored molecule containingno peptide transmembrane regions, making it an attractive candidatefor study of retrovirus entry. Further, the hyaluronidase activityreported for human Hyal2, combined with its broad expressionpattern, may point to a critical function of Hyal2 in the turnoverof hyaluronan, a major extracellular matrix component. Herewe describe the properties of a soluble form of human Hyal2(sHyal2) purified from a baculoviral expression system. sHyal2is a 54-kDa monomer with weak hyaluronidase activity comparedto that of the known hyaluronidase Spam1. In contrast to a previousreport indicating that Hyal2 cleaved hyaluronan to a limit productof 20 kDa and was active only at acidic pH, we find that sHyal2is capable of further degradation of hyaluronan and is activeover a broad pH range, consistent with Hyal2 being active atthe cell surface where it is normally localized. Interactionof sHyal2 with the JSRV envelope glycoprotein was analyzed byviral inhibition assays, showing >90% inhibition of transductionat 28 nM sHyal2, and by surface plasmon resonance, revealinga remarkably tight specific interaction with a dissociationconstant (K_D) of 32 ± 1 pM. In contrast to results obtainedwith avian retroviruses, purified receptor was not capable ofpromoting transduction of cells that do not express the virusreceptor.

INTRODUCTION

Top

Introduction

Retroviruses enter host cells by executing a complex molecularprogram resulting in the fusion of viral envelope and the hostcell plasma membrane. Key players in this process are the retroviralenvelope glycoprotein (Env) and the corresponding target cellreceptor proteins. Although protocols for production and purificationof solubilized recombinant Env fragments have been developedfor several retroviruses, their receptor counterparts have generallyproven difficult to produce and purify because they are classIV (multi-pass) integral membrane proteins. For example, thegammaretrovirus receptors Pit1, Pit2, Xpr, Flvcr, and Cat-1function or are predicted to function as transporters for smallmolecules and contain from 8 to 13 predicted membrane-spanningregions (44).

The current model for fusogenic virus entry is based on studiesdone with the influenza virus (8). Endocytosis, followed bythe acidification of the endosomal compartment, causes the envelopeglycoprotein to undergo profound structural reorganization,bringing the viral membrane into very close proximity to theendolysosomal membrane (reviewed in references 10 and 21). Althoughsome retroviruses require this acidification step for productiveinfection (42, 49), many are able to infect cells directly atthe plasma membrane at neutral pH (25, 42, 60). In these cases,it has been proposed that the interaction between the retroviralenvelope glycoprotein and its entry receptor, displayed on thetarget cell surface, results in a conformational change therebyserving as an initial step in the fusogenic program (4). Receptor-inducedconformational changes in retroviral envelope glycoproteinshave been observed in a number of systems (4, 24, 28, 51, 57),and in each case the conformational change is believed to leadto activation of envelope glycoprotein fusogenicity. Thus, understandingthe nature of the interaction between retroviral envelope glycoproteinsand their respective receptors is essential to understandingretroviral entry.

Most previous work aimed at the characterization of Env-receptorinteractions has been carried out by studying binding of envelopeglycoprotein fragments to receptor-expressing cells. Env/receptorcomplex formation has typically been detected by either directlabeling of the envelope construct with ¹²⁵I (5, 16, 31, 32)or by using fluorescently labeled secondary antibodies againstEnv (18, 36). While these strategies circumvent the difficultiesassociated with receptor protein purification, the binding measurementsare likely complicated by the presence of other proteins andmacromolecules on the cell surface. A major advance in the fieldwas recently made by Hoffman et al. with their report of directkinetic measurements of the human immunodeficiency virus type1 (HIV-1) gp120 interaction with chemokine receptors by surfaceplasmon resonance (SPR) (27). However, even in this case, CCR5and CXCR4 had to be incorporated into a cell-derived membranecontaining other surface molecules that could potentially complicatethe analysis.

Ideally, detailed analysis of Env-receptor binding would becarried out in a system containing only the two interactingspecies, and there are a few characterized retroviral systemsthat involve receptors with one or no membrane-spanning regionsthat can be made in soluble form. For instance, studies aimedat examining the fusogenic properties of subgroup A avian sarcomaand leukosis virus (ASLV-A) Env (13, 20, 26, 54) have becomepossible since the development of a functional soluble Tva receptor(3). The Tvb receptor, used by ASLV groups B, D, and E (7),as well as the recently identified mouse mammary tumor virusreceptor (49), each possess a single membrane-spanning region.Lastly, the jaagsiekte sheep retrovirus (JSRV) receptor hyaluronidase2 (Hyal2) is linked to the cell surface solely through a glycosylphosphatidylinositol(GPI) anchor (47).

JSRV is a simple retrovirus that causes lung cancer in sheep.Several lines of evidence indicate that Hyal2 is the JSRV receptor,including radiation hybrid mapping results (46), the fact thatresistant cells can be rendered susceptible to transductionby expression of human or ovine Hyal2 orthologs (19, 47), andthe detection of specific binding of a hybrid JSRV Env SU subunit/humanimmunoglobulin G (IgG) constant region (JSU-IgG) to cells expressingHyal2 receptors (36).

JSRV can cause acute multifocal cancer in infected sheep (53),but unlike most other acutely transforming retroviruses, itdoes not contain a host-derived oncogene. Expression of JSRVEnv can transform cells in culture (2, 37, 45, 47), indicatingthat Env is the oncogene responsible for JSRV pathogenesis invivo. In the epithelium-derived BEAS cell line, expression ofJSRV Env causes transformation that is dependent on its interactionwith Hyal2 and the subsequent degradation of the JSRV Env/Hyal2complex (15). It is interesting to note that the Hyal2 geneis contained within the 3p21.3 locus, which is deleted in themajority of small-cell lung cancers and has been proposed tobe a tumor suppressor (35). In contrast, fibroblast-derivedcell lines are transformed in a Hyal2-independent manner, whichinvolves the activation of the phosphatidylinositol 3-kinase(PI3K)/Akt signaling pathway by the cytoplasmic tail of JSRVEnv (1, 45). It is not known which mechanism predominates indiseased animals.

The normal function of Hyal2 is unclear. Hyal2 is a member ofthe hyaluronidase family of proteins, which includes the spermhyaluronidase Spam1 and the serum hyaluronidase Hyal1 (12).Hyaluronidase-catalyzed degradation of hyaluronan, a linearrepeating disaccharide, is thought to be important for the remodelingof the extracellular matrix as well as for maintaining homeostasisin various connective tissues in vertebrates. In addition, multipleroles for hyaluronan and its degradation products in cancerbiology have been proposed (for review, see reference 56). Hyaluronidaseactivity associated with expression of human Hyal2 in culturedcells has been reported (34). Unlike other hyaluronidases, itappeared that Hyal2 was unable to digest hyaluronan to completionand left a 20-kDa-limit product, which might have unique biologicalroles (34). However, we were unable to detect hyaluronidaseactivity in cells following expression of human Hyal2 usinga retrovirus expression system, while we were easily able todetect the activity of Hyal1 using the same expression system,and we estimated that the hyaluronidase activity of Hyal2 wasat least 50-fold lower than that of human Hyal1 (47). The differencein results from these previous studies may be due to a higherlevel of Hyal2 production from the vaccinia virus expressionsystem used in the first study, but it is puzzling that Hyal2has such low activity if its normal function is cleavage ofhyaluronan.

Here we describe the production and purification of a solubleform of human Hyal2 (sHyal2), characterize its hyaluronidaseactivity, and examine its interaction with JSRV Env. We findthat sHyal2 is a weak hyaluronidase, active over a wide pH rangeand capable of complete degradation of hyaluronan. PurifiedsHyal2 specifically interacts with retroviral particles pseudotypedwith JSRV Env, as evidenced by its ability to inhibit transductionof susceptible cells. However, purified sHyal2 was not capableof facilitating transduction of non-receptor-expressing cells.SPR measurements of the binding of sHyal2 to the JSU-IgG showan unusually tight interaction due primarily to an extremelyslow dissociation rate.

MATERIALS AND METHODS

Top

Materials and Methods

Cell culture. Mammalian cell lines were maintained in Dulbecco's modifiedEagle medium with high glucose (4.5 g per liter) (Gibco) and10% fetal bovine serum (HyClone) at 37°C in a 10% CO₂-airatmosphere unless otherwise noted. Insect cell lines were maintainedat 27°C in air. Sf9 cells were grown in SF-900 II serum-freemedium (SFM), and High5 cells were grown in Express Five SFM(Gibco).

Generation of recombinant baculovirus encoding sHyal2. Mutagenic PCR was used to modify a cDNA encoding human Hyal2to contain an SphI restriction site upstream of the start codonand a NotI site in place of the codon just after the GPI additionsite in the protein. The resulting fragment was ligated to aNotI/HindIII linker encoding His₆ and was then cloned into thepFastbac1 (Gibco) transfer vector. In this procedure, the Hyal2leader peptide was preserved, while the C-terminal hydrophobicregion, predicted to be removed in the mature polypeptide, wasreplaced by a His₆ tag (Fig. 1A). The resulting cDNA encodedresidues 1 to 447 of human Hyal2, separated from the His₆ tagby Ala₃ (encoded by the NotI restriction site). Subsequent stepsfor generation of the starting stock of recombinant baculovirusin Sf9 cells were carried out according to the manufacturer'sinstructions. Expression of recombinant protein in infectedcultures was monitored by immunoblotting with anti-Penta-Hisantibody (QIAGEN). Final virus stock expansion was carried outby Custom Baculovirus Services (BD Biosciences).

View larger version (26K):
[in this window]
[in a new window]
FIG. 1. Structure and properties of purified sHyal2. (A) The structures of native Hyal2 and sHyal2 are shown (both drawn to scale). Amino acids 1 to 21 constitute the endoplasmic reticulum signal peptide for both proteins. A GPI anchor is predicted to replace all residues following amino acid 447 in Hyal2, which also contains a C-terminal hydrophobic tail at residues 456 to 473 that localizes the protein in the membrane prior to GPI anchor addition. In sHyal2, a C-terminal His₆ tag (attached with an Ala₃ linker) replaces the GPI anchor attachment site and the hydrophobic tail of the native sequence. (B) Size exclusion chromatography using a Superdex 200 HR 10/30 column shows that sHyal2 is monomeric in solution and migrates with an apparent molecular mass of $~$ 40 kDa. Overlaid onto the trace of sHyal2 is a trace of Bio-Rad gel filtration chromatography standards with sizes in kilodaltons indicated. (C) SDS-PAGE of sHyal2 and protein standards under reducing conditions indicates a molecular mass for sHyal2 of $~$ 50 kDa.

Expression and purification of recombinant sHyal2. Recombinant baculovirus stock ( $~$ 10⁸ PFU/ml) was used to inoculatespinner cultures of High5 cells for protein production at theoptimal multiplicity of infection of 5. The cultures were infectedby resuspending log-phase cells in virus stock, incubating thesuspension at room temperature for 10 to 15 min, and dilutingin fresh medium to a density of 10⁶ cells/ml. The infected cultureswere grown in $≤$ 500 ml per 1-liter spinner flask.

When cell viability in infected cultures dropped to $~$ 50% (approximately48 to 72 h postinfection), cells and cell debris were removedby centrifugation at 3,000 x g and the supernatant was furtherclarified by centrifugation at 7,000 x g. The resulting materialeither was dialyzed immediately against equilibration buffer(150 mM NaCl, 50 mM Tris-HCl, pH 8) or was first precipitatedby addition of ammonium sulfate to 30% (wt/vol). The precipitatewas dissolved in phosphate-buffered saline (PBS), and the solutionwas then dialyzed. The treated baculovirus supernatant was nextpassed over a column containing Talon resin (Clontech), thecolumn was washed with 10 column volumes of wash buffer (300mM NaCl, 50 mM Tris-HCl, pH 7.4, 10% glycerol, 20 mM imidazole)and the bound protein was eluted with 3 column volumes of elutionbuffer (wash buffer with 200 mM imidazole). The eluted proteinwas concentrated by ultrafiltration (Amicon Ultra-15 modules;Millipore) and fractionated on a Superdex 200 HR 10/30 sizingcolumn (Amersham Biosciences AB) using PBS or PNEA, which iscomposed of 20 mM PIPES [piperazine-N,N'-bis(2-ethanesulfonicacid), pH 7.0], 150 mM NaCl, 1 mM EDTA, and 0.02% NaN₃.

Expression and purification of the JSU-IgG. The generation of a plasmid encoding JSU-IgG has been described(36). This plasmid encodes a hybrid protein consisting of theJSRV Env SU fused to a human IgG Fc domain. The plasmid wastransfected into 293 human embryonic kidney cells, and the cellswere grown for 24 to 36 h in Dulbecco's modified Eagle's mediumwith 10% Ultra-Low IgG fetal bovine serum (FBS) (Gibco). Conditionedmedium was harvested and added to a column containing proteinA Sepharose resin (Amersham Biosciences AB), the column waswashed with 10 column volumes of 100 mM sodium citrate-0.02%NaN₃ buffer, pH 4.5, and the JSU-IgG protein was eluted with2 column volumes of the same buffer adjusted to pH 3.5. Theeluate was washed into PNEA by ultrafiltration (Amicon Ultra-15modules; Millipore).

Hyaluronidase assay. Hyaluronidase activity was assayed by incubation of proteinsamples with hyaluronan and analysis of digestion products byagarose gel electrophoresis, as previously described (47). Briefly,50-µg samples of hyaluronan derived from human umbilicalcord (Calbiochem) were treated with either purified sHyal2 orSpam1 (bovine testes; Calbiochem) for 12 to 16 h at 37°Cin a reaction buffer containing 100 mM formate (for pH 3 to5.1), 100 mM MES (morpholineethanesulfonic acid) (for pH 5.5to 7.3), or 100 mM Tris (for pH 7.8 to 11). The samples weremixed with 1/6 volume of 7x loading buffer (7x TAE [40 mM Tris-acetate,1 mM EDTA, pH 8.0], 85% glycerol) and electrophoresed in 0.5%TAE-agarose gels at 50 V for 8 to 10 h. A 1-kb DNA ladder (Invitrogen)was used as a molecular weight standard. The gels were stainedin 0.005% Stains-All (Sigma) in 50% ethanol overnight and thenphotographed on a fluorescent light transilluminator.

Biosensor analysis of the interaction between JSU-IgG and sHyal2. SPR measurements were carried out in HBS-EP buffer (10 mM HEPES[pH 7.4], 150 mM NaCl, 3 mM EDTA, 0.005% P-20 surfactant [BiacoreAB, Uppsala, Sweden]) using a Biacore 3000 system (Biacore AB).JSU-IgG was immobilized on a CM5 research-grade sensor chip(Biacore AB) by amine coupling chemistry by using the manufacturer'sprotocols. Immobilization of 3,700 response units (RU) resultedin optimal responses for subsequent analyses.

The kinetics of the interaction between JSU-IgG and sHyal2 weremeasured by injecting five concentrations of sHyal2 (5 to 74nM) in randomized duplicate runs. Optimal regeneration was achievedby injection of 3 M KSCN at a flow rate of 100 µl/minfor 30 s over the biosensor surface followed by injection ofHBS-EP buffer at 20 µl/min for 3 h. Injections of sHyal2and control solutions were carried out at 20 µl/min fora period of 200 s followed by an hour-long dissociation phase.

The data were analyzed by the method described by Myszka (43).First, the sHyal2 binding signal measured on the biosensor referencesurface (capped by amine coupling reagents) was subtracted fromthe sHyal2 binding signal obtained on the surface derivatizedby the JSU-IgG. Next, a data set was constructed by averagingdata obtained from several HBS-EP injections over the biosensorsurface. This averaged data set was then subtracted from thereferenced signal, resulting in doubly referenced data whichwere then analyzed with BIAevaluation 3.0 software (BiacoreAB) to globally fit data and derive kinetic and equilibriumvalues describing the intermolecular interactions.

RESULTS

Top

Results

Expression and purification of sHyal2. During maturation, the carboxy end of Hyal2 is removed and replacedwith a GPI anchor that tethers Hyal2 to the cell surface. Tomake a soluble form of the protein (sHyal2), we made an expressionvector that encodes Hyal2 with a His₆ tag replacing the GPIanchor (Fig. 1A). The native signal peptide was retained inthe construct to allow proper posttranslational processing andsecretion from cells. sHyal2 was expressed in insect cells byusing a recombinant baculovirus expression system and was purifiedwith a combination of immobilized metal affinity and size exclusionchromatography. The identity of the protein was verified byN-terminal sequencing (data not shown). sHyal2 is a monomericprotein in solution (Fig. 1B). A size consistent with the predictedmolecular mass of 50 kDa was apparent by sodium dodecyl sulfate-polyacrylamidegel electrophoresis (SDS-PAGE) analysis (Fig. 1C), indicatinglow utilization of the four predicted N-linked glycosylationsites in this construct. Matrix-assisted laser desorption ionization-time-of-flight(MALDI-TOF) mass spectrometry analysis showed a single 54-kDaspecies in the purified protein samples. This value was usedfor subsequent concentration calculations.

Hyaluronidase activity of sHyal2. We found that sHyal2 and the known hyaluronidase Spam1 couldconvert a mixed high-molecular-mass population of hyaluronanto a population with a narrow size range comigrating with the2-kb DNA marker (Fig. 2A), previously determined to be around20 kDa (33, 34). Comparison of hyaluronan degradation by 1.6and 1.0 µg of sHyal2 (Fig. 2A and B, lane correspondingto pH 3.8, respectively) to that by known amounts of Spam1 ledto the conclusion that approximately 200-fold more sHyal2 thanSpam1 is required to digest 50 µg of hyaluronan to the20-kDa stage in the same period of time. Therefore sHyal2 is $~$ 200-fold less active than Spam1 at this pH (pH 3.8). The pHoptimum of the sHyal2 hyaluronidase activity was broad, withan optimum centered around pH 5.5 (Fig. 2B). In a control reaction,incubation of hyaluronan at pH 3.8 or 11.0 without added proteindid not result in degradation of the polymer (Fig. 2B and datanot shown, respectively).

View larger version (67K):
[in this window]
[in a new window]
FIG. 2. sHyal2 is a weak neutral-pH-active hyaluronidase. (A) Fifty-microgram samples of HA were incubated at 37°C and pH 3.8 for 14 h in a total volume of 40 µl with no added protein, with various amounts of sperm hyaluronidase Spam1, or with 1.6 µg of sHyal2. The samples were separated in 0.5% agarose and visualized with Stains-All. (B) Fifty-microgram samples of HA were incubated at 37°C for 14 h in a total volume of 40 µl with no added protein at pH 3.8 or with sHyal2 (1.0 µg per sample) at the indicated pH values. Samples were analyzed as in panel A.

In contrast to previous results which indicated that Hyal2 cleavedhyaluronan to a limit product of 20 kDa (34), we find that degradationof hyaluronan by sHyal2 can proceed beyond this intermediate.Digestion of hyaluronan with increasing concentrations of sHyal2at the optimal pH of 5.5 resulted in increasing degradationof the 20-kDa intermediate (Fig. 3, left panel), similar toresults obtained with Spam1 (47; data not shown). Further degradationof the 20-kDa hyaluronan intermediate was also observed afteran additional treatment with sHyal2 after the initial incubationperiod (Fig. 3, right panel). Degradation of this 20-kDa populationwas a much slower reaction step than the initial cleavage ofhigh-molecular-mass hyaluronan to the 20-kDa form for both sHyal2(Fig. 3) and Spam1 (47).

View larger version (83K):
[in this window]
[in a new window]
FIG. 3. sHyal2 is capable of complete digestion of hyaluronan. (Left panel) Fifty-microgram samples of HA were incubated with no added protein or with various amounts of sHyal2 at 37°C and pH 5.5 for 14 h in a total volume of 40 µl. Samples were analyzed by agarose gel electrophoresis as in Fig. 2. (Right panel) To examine the effect of repeated sHyal2 treatment on HA degradation, a 50-µg sample of HA was incubated with 1.6 µg of sHyal2 for 12 h, 1.6 µg of sHyal2 in PNEA buffer was added to the sample, and the sample was incubated for an additional 12 h (lane labeled "1.6 µg sHYAL2 x 2"). In parallel, 50-µg samples of HA were incubated with or without 1.6 µg of sHYAL2 for 12 h, PNEA buffer without sHyal2 was added, and the samples were incubated for an additional 12 h (lanes labeled "1.6 µg sHyal2" and "No added protein", respectively). Samples were analyzed by agarose gel electrophoresis as in Fig. 2.

Inhibition of JSRV vector transduction by sHyal2. To test whether purified sHyal2 could specifically interactwith JSRV-pseudotype retroviral particles, JSRV-pseudotype andamphotropic murine leukemia virus (MLV)-pseudotype vectors wereused to infect NIH 3T3/LL2SN cells that express receptors forboth viruses, in the presence or absence of sHyal2 (Fig. 4).The presence of 9 nM or more sHyal2 caused a dramatic decreasein titer for the JSRV-pseudotype vector, with 28 nM sHyal2 inhibitingtransduction by 93%. In comparison, sHyal2 had no effect onthe titer of the control amphotropic MLV-pseudotype vector atany concentration tested. These results demonstrate a specificinteraction of sHyal2 with the JSRV Env protein on virions thatresults in inhibition of virus entry into cells.

View larger version (19K):
[in this window]
[in a new window]
FIG. 4. sHyal2 specifically inhibits transduction by a JSRV-pseudotype retroviral vector. PJ4/LAPSN (similar to PJ14/LAPSN) (46) and PA317/LAPSN (39) vector-producing cell lines were used to generate JSRV-pseudotype and amphotropic MLV-pseudotype vectors. The LAPSN vector encodes human placental alkaline phosphatase (AP) and bacterial neomycin phosphotransferase. These vector stocks were used to transduce NIH 3T3 TK^– cells expressing human Hyal2 and that also express the endogenous mouse amphotropic retrovirus receptor Pit2. Experiments were carried out in the presence of various concentrations of purified sHyal2 that was added to the cells just prior to virus addition. Two days after virus exposure, the cells were stained for AP and AP⁺ foci were counted. Results are expressed as a percentage of the transduction rate observed without sHyal2 addition for each vector. Each data point represents the average of three experiments, and standard deviations are shown.

sHyal2 does not stimulate JSRV vector transduction of receptor-deficient cells. Others have shown that soluble receptor protein can induce infectionof receptor-deficient cells by avian retroviral vectors (13,30), and we tested whether sHyal2 could induce JSRV vector infectionof receptor-deficient NIH 3T3 cells in a similar manner. A JSRV-pseudotypevector was incubated with or without sHyal2 at concentrationsfrom 2 to 1,000 nM for 30 min on ice and cells were exposedto the vector in the presence of Polybrene with or without centrifugationas described previously (13). While the titer of the untreatedvirus on NIH 3T3 cells expressing human Hyal2 was 7 x10⁴ perml, the titer of treated or untreated vector on receptor-deficientNIH 3T3 cells was $≤$ 50 per ml, indicating that sHyal2 cannot efficientlyinduce infection of receptor-deficient cells.

Analysis of purified JSU-IgG. JSU-IgG prepared as described in Materials and Methods was analyzedby size exclusion chromatography on a Superdex 200 HR 10/30column and was found to be highly multimeric (apparent molecularmass of $~$ 400 kDa) (data not shown). However, nonreducing SDS-PAGEanalyses showed that the multimers were likely "daisy chains"of interchain disulfide-bonded protein (data not shown), consistentwith the large number of cysteines (15 residues) per JSU-IgGmonomer that could provide free thiols for multimerization.However, the purified JSU-IgG multimers appeared to be properlyfolded and not the result of nonspecific aggregation, becausethe material was stable in solution and did not precipitate.Flow cytometry experiments also show specific binding of JSU-IgGto cells (36; data not shown), and control SPR experiments demonstratedthat JSU-IgG did not interact with other proteins nonspecifically(data not shown).

Binding kinetics of recombinant Hyal2 to the JSU-IgG. The interaction between Hyal2 and the JSRV Env was further studiedby SPR using JSU-IgG covalently bound to the sensor surface(ligand) and sHyal2 in solution flowing over the surface (analyte)(Fig. 5, top panel). The analysis of the SPR responses witha 1:1 Langmuir binding model (Fig. 5) resulted in a good globalfit ( ${chi}$ ² = 0.097), with a dissociation constant (K_D) value of32 ± 1 pM. Such tight binding is explained by the extremelyslow dissociation of the complex (k_off = [5.33 ± 0.07]x 10^–6 s^–1; k_on = [1.68 ± 0.01] x 10⁵ M^–1s^–1), corresponding to a complex half-life of $~$ 36 h. Bindingof either Spam1 or soluble JSU-IgG to JSU-IgG coupled to thesensor surface was undetectable, demonstrating the specificityof the interaction of Hyal2 with bound JSU-IgG (data not shown).

View larger version (38K):
[in this window]
[in a new window]
FIG. 5. Measurement of the kinetics of the interaction between sHyal2 and JSU-IgG. (A) Reference-subtracted data (gray) illustrating the interaction between the immobilized JSU-IgG and sHyal2 are shown. Following the initial phase of buffer flow, solutions of indicated concentrations of sHyal2 in buffer were injected during the time interval of 0 to 200 s at a rate of 20 µl/min. This was followed by a return to initial buffer flow to allow for measurements of dissociation of the complexes formed. BIAevaluation software was used to build a curve fit (black line) to the data. (B) Deviations of data from the fit in panel A are shown.

DISCUSSION

Top

Discussion

In this report we describe the production of a soluble formof the JSRV receptor Hyal2 using a baculovirus expression system.Since Hyal2 is normally tethered to the cell membrane by a GPIanchor, we generated sHyal2 by replacing the same amino acidsdeleted during GPI anchor addition with a His₆ tag. The resultingsHyal2 protein was purified as a 54-kDa monomer with perhapsone N-linked glycosylation.

We used SwissModel (52) to generate a model of the structureof Hyal2 (data not shown), based on the bee venom hyaluronidasestructure (38; PDB accession code 1FCQ), encompassing humanHyal2 residues 32 to 355. This region includes two of the fourpotential N-linked glycosylation sites, one of which is predictedto be fully buried in the model, while the other is partiallyburied. Thus, a likely explanation for the underutilizationof N-linked glycosylation sites observed for sHyal2 is thatthese sites are insufficiently exposed on the molecular surfacefor proper modification.

Our modeling efforts also revealed that the pattern of primarysequence conservation between Hyal2 and bee venom hyaluronidaseis consistent with conservation of the overall fold. In particular,the predicted catalytic residues of Hyal2 were appropriatelymapped to the active site of the bee venom hyaluronidase structure.Residues 355 to 447 fall outside of the model, due to the absenceof corresponding sequences in bee venom hyaluronidase, and mayrepresent a separate, cysteine-rich C-terminal domain.

Hyal2 was previously reported to be a functional lysosomal hyaluronidasewith an acidic pH optimum (34). A subsequent study found thatHyal2 was actually a cell surface protein, but hyaluronidaseactivity could not be detected in cells expressing Hyal2 followingretrovirus-mediated gene transduction (47). Here we find thatsHyal2 is indeed able to degrade hyaluronan, but the rate ofdegradation is about 200-fold lower that that of Spam1. Thisresult is consistent with the previous finding that hyaluronidaseactivity is below the limit of detection in lysates of cellstransduced with a Hyal2-expressing retroviral vector. Further,we found that sHyal2 is active over the pH range of 4.5 to 8.6(Fig. 2B), but unlike the neutral-pH-active Spam1 (23; datanot shown), it shows no significant activity minimum at aroundpH 5.5. The broad pH range of the hyaluronidase activity ofsHyal2 including normal physiological conditions is consistentwith Hyal2 being active at the cell surface where it is normallylocalized.

sHyal2-mediated hyaluronan degradation proceeds in a mannerlike that previously observed for Spam1 and Hyal1 (47; unpublisheddata). The reaction first transforms a population of hyaluronanof highly heterogeneous molecular mass to a population previouslyestimated to be $~$ 20 kDa (33, 34). This initial rapid degradationis followed by a slower degradation step in which the 20-kDapolysaccharide population is further digested (Fig. 3), likelyproducing end-stage tetramer and hexamer products, as is thecase for Spam1 (11, 50). Although it was previously reportedthat Hyal2-mediated hyaluronan degradation stops at the 20-kDapolysaccharide fragment size (34), it is likely that the amountof enzyme or time of digestion was insufficient to detect digestionof hemagglutinin (HA) past the 20-kDa HA intermediate.

We tested the ability of sHyal2 to specifically inhibit transductionby JSRV-pseudotype MLV particles as a measure of sHyal2 biologicalactivity. sHyal2 is capable of significantly inhibiting JSRVEnv-mediated transduction events, while having no effect onan amphotropic vector. This result indicates that sHyal2 canspecifically interact with JSRV Env and gives further confirmationthat sHyal2 is a biologically active, properly folded protein.Inhibition of viral entry by soluble receptor-derived Env ligandshas been demonstrated for at least two other retroviral systems.Soluble CD4 has been shown to inhibit infection by certain strainsof HIV, in part by causing shedding of the gp120 surface subunitof HIV Env (40, 41). Similarly, a soluble version of the ASLV-Areceptor Tva is capable of inhibiting ASLV-A infection (3),and in this case the mechanism appears to involve triggeringof the Env-mediated fusion (14, 26).

Although sHyal2 inhibits transduction of receptor-expressingcells, it does not facilitate transduction of receptor-deficientcells, as has been observed for soluble receptor-like moleculesin the ASLV-A and ASLV-B systems (13, 30). It has recently beenreported that initiation of the fusogenic program of ASLV canoccur at neutral pH (20). It is possible that, unlike ASLV,JSRV requires acidification of its environment for initiationof fusion, which might explain its inability to transduce receptor-deficientcells even in the presence of soluble receptor. Alternatively,the binding of the sHyal2 to the JSRV-pseudotype particles maydestabilize the metastable Env glycoprotein to an extent greaterthan is observed for the ASLV systems, resulting in rapid JSRVEnv inactivation.

Using SPR, we found very tight binding of sHyal2 to JSU-IgG,with a 32-pM K_D, primarily due to an extremely slow dissociationrate. This K_D is significantly different from the previouslyreported value for binding of JSU-IgG to cells expressing humanHyal2 of 9 ± 6 nM (36), as determined by flow cytometry.This discrepancy may be explained by several factors. Sincethe state of the JSU-IgG, as discussed in this report, was multimeric,the cytometry experiments formally examined non-1:1 interactions,which were not accounted for in the earlier analysis. The prioranalysis also did not take into account the amount of inactiveJSU-IgG in such preparations, which would lead to an overestimateof the actual K_D. Furthermore, the macroscopic recognition eventsdescribed in that report took place in the context of the complexenvironment of the cell surface (glycocalyx), perhaps leadingto reduced affinity between JSU-IgG and the cell surface-boundHyal2. In addition, since Hyal2 is a GPI-anchored protein (47),it is likely to be associated with lipid rafts (6), which wouldaffect the context in which it is presented on the cell surfaceand may bias its ability to bind ligands. The SPR analysis performedhere directly measured the association between two biochemicallypure interactors, giving the best estimate of the microscopicequilibrium and rate constants describing the interaction ofJSRV Env and Hyal2 in the absence of other factors.

Previously reported affinities for other retroviral Env/receptorpairs span a wide range of values and include 500 nM for theHIV-1 Env/CXCR4 (determined by SPR) (27), a range of 11.6 to0.83 nM for the HIV-1-CD4 interaction (for citations, see reference3), values of 17 and 1.5 nM (determined by flow cytometry; references17 and 62, respectively) and 0.3 nM (determined by enzyme-linkedimmunosorbent assay [ELISA]) (3) for the ASLV-A Env/solublerecombinant Tva, and 92 pM for the amphotropic MLV Env immunoadhesin/Pit2(determined by cell-associated radiolabel detection) (32). Thus,the affinity value we are reporting is the strongest Env-receptorinteraction observed to date, although it is within an orderof magnitude of its closest neighbor.

It has been previously noted that viral interactions with targetcell receptors typically display higher affinities than thenormal biological interactions in which these cell surface moleculesare implicated (59). In our case, the extreme stability of thecomplex suggests that its formation is essentially irreversible,particularly considering the additional effects of avidity andrebinding due to the higher local concentrations of two membrane-anchoredinteractors. Such slow dissociation kinetics have been describedfor multiple types of interactions for which complex dissociationis undesirable. For the interaction between major histocompatibilitycomplex molecules and bound antigenic peptides, complex stabilityis essential for proper antigen presentation at the cell surface,where complex K_off rates have been estimated at 10^–4 s^–1and slower (29, 48). High-affinity antibodies with similar K_offrates (10^–5 to 10^–4 s^–1) have also been described(61), and additional examples exist among the more classicalprotein-protein interactions such as the VEGF/VEGFR-1 complex(10^–5 s^–1) (58) and the pepsin/rSQAPI complex (10^–4s^–1) (22).

The high affinity of JSU-IgG for sHyal2 and the resultant longhalf-life of the complex indicates that JSRV Env is highly proficientin its ability to anchor incoming virus particles onto targetcells expressing human Hyal2, and presumably onto cells expressingsheep Hyal2 as well. One reason this might be important forthe biology of JSRV is that this enhanced attachment functionmay reduce the impact of environmental stresses that JSRV normallyhas to confront during infection of the lung. In fact, it hasbeen reported that JSRV-pseudotype particles demonstrate greaterresistance to the antimicrobial properties of lung fluid (9),which contains compounds with surfactant- and detergent-likeactivities. If the exposure of virus to lung fluid causes damageto most of the exposed Env glycoprotein molecules, the remaindermay still successfully anchor the particle to the host cellby virtue of their strong interaction with the Hyal2 receptorand allow fusion to occur.

Another way in which the high-affinity interaction between JSRVEnv and Hyal2 may be important for JSRV is the ability to utilizelow levels of available surface Hyal2. The relative abundanceof surface Hyal2 in cells of the lung epithelium or establishedcell lines has not been reported to date. However, it has beenrecently demonstrated that overexpression of endogenous JSRV-derivedenvelope glycoprotein can interfere with exogenous JSRV infectionin sheep cells (55). Since this study also found that endogenousJSRV is expressed in ovine tissues, it was argued that thismay offer a degree of resistance to natural exogenous JSRV infectionby lowering the surface levels of Hyal2. Thus, the stabilityof the JSU-IgG/sHyal2 complex may offer JSRV the ability tocompete for surface receptor "occupied" by the endogenous JSRVEnv or may suggest that few interactions with its receptor arerequired for stable JSRV adhesion to host cell surface.

Finally, the extremely high stability of the JSU-IgG/sHyal2complex may help explain the oncogenic properties that JSRVEnv displays in epithelial cells. It has been reported thatJSRV Env expression can disrupt the repressed state of the RONtyrosine kinase by causing the degradation of its natural repressor,Hyal2 (15). In its active state, RON is capable of initiatinga ligand-independent signaling cascade which leads to cell transformation.It is proposed that the degradation of Hyal2 results from theformation of a stable complex between it and JSRV Env. Datapresented here show that this complex is indeed very stable.

ACKNOWLEDGMENTS

We thank Deb McMillen (Genomics & Proteomics Facility, Universityof Oregon) for performing the N-terminal sequencing of sHyal2and Phil Gafken (Proteomics, Fred Hutchinson Cancer ResearchCenter) for performing the MALDI-TOF analysis.

This work was supported by National Cancer Institute traininggrant CA77116 (V.V.), grant AI48675 (R.K.S.), and grants DK47754and HL66947 (A.D.M.) from the National Institutes of Health.

FOOTNOTES

* Corresponding author. Mailing address: Fred Hutchinson Cancer Research Center, 1100 Fairview Ave. N., Room C2-105, P.O. Box 19024, Seattle, WA 98109-1024. Phone: (206) 667-2890. Fax: (206) 667-6523. E-mail: dmiller{at}fhcrc.org.

REFERENCES

Top