A Novel Function of Heparan Sulfate in the Regulation of Cell-Cell Fusion*

Despite the important contribution of cell-cell fusion in the development and physiology of eukaryotes, little is known about the mechanisms that regulate this process. Our study shows that glycosaminoglycans and more specifically heparan sulfate (HS) expressed on the cell surface and extracellular matrix may act as negative regulator of cell-cell fusion. Using herpes simplex virus type-1 as a tool to enhance cell-cell fusion, we demonstrate that the absence of HS expression on the cell surface results in a significant increase in cell-cell fusion. An identical phenomenon was observed when other viruses or polyethylene glycol was used as fusion enhancer. Cells deficient in HS biosynthesis showed increased activity of two Rho GTPases, RhoA and Cdc42, both of which showed a correlation between increased activity and increased cell-cell fusion. This could serve as a possible explanation as to why HS-deficient cells showed significantly enhanced cell-cell fusion and suggests that HS could regulate fusion via fine tuning of RhoA and Cdc42 activities.

Despite the important contribution of cell-cell fusion in the development and physiology of eukaryotes, little is known about the mechanisms that regulate this process. Our study shows that glycosaminoglycans and more specifically heparan sulfate (HS) expressed on the cell surface and extracellular matrix may act as negative regulator of cell-cell fusion. Using herpes simplex virus type-1 as a tool to enhance cell-cell fusion, we demonstrate that the absence of HS expression on the cell surface results in a significant increase in cell-cell fusion. An identical phenomenon was observed when other viruses or polyethylene glycol was used as fusion enhancer. Cells deficient in HS biosynthesis showed increased activity of two Rho GTPases, RhoA and

Cdc42, both of which showed a correlation between increased activity and increased cell-cell fusion. This could serve as a possible explanation as to why HS-deficient cells showed significantly enhanced cell-cell fusion and suggests that HS could regulate fusion via fine tuning of RhoA and Cdc42 activities.
Cell-cell fusion is an important physiological process widespread in organisms ranging from yeast to humans (1). It is critical for several biological phenomena including fertilization, placenta formation, skeletal muscle and bone development, tumorigenesis, immune response, and stem cell differentiation (1)(2)(3)(4)(5)(6)(7)(8)(9). Defects in cell-cell fusion can lead to serious diseases, such as myotonic dystrophy, centronuclear myopathy, preeclampsia, and osteopetrosis (10 -13). Defects in sperm-egg fusion are a major cause of infertility (5). Cell-cell fusion has also been utilized for therapeutic applications, including the generation of monoclonal antibody-producing hybridomas (14) as well as new agents for cancer immunotherapy (15)(16)(17).
Because of its critical nature, many studies have looked at the mechanism by which cell-cell fusion occurs. Although it can occur in a variety of different biological processes, many of the fusion events share common characteristics (8). For example, tetraspanin proteins function in gamete-, myoblast-, macrophage-, and virus-mediated fusion events (18 -21). Although many mediators of cell-cell fusion are known, little is known about the fine-tuning mechanisms that may regulate the membrane fusion process.
Viruses have been a useful tool for studying cell-cell fusion since the discovery that they could induce the fusion of somatic cells in vitro (22). Enveloped viruses, like herpes simplex virus type-1 (HSV-1), 2 use transmembrane viral proteins to mediate fusion with the host cell during entry and spread (23)(24)(25). For HSV-1, fusion occurs after the virus has attached to host cells by binding to heparan sulfate (HS) using glycoproteins gB and gC (26). Fusion of the virus envelope with the plasma membrane requires that an additional glycoprotein, gD, binds to one of its receptors, a process that also requires HSV-1 gB, gH, and gL (27)(28)(29). During HSV-1-mediated cell-cell fusion, gB, gD, gH, and gL are expressed on the surface of infected cells, allowing them to bind and fuse with surrounding uninfected cells, forming syncytia.
Heparan sulfate proteoglycans are ubiquitously expressed cell surface molecules composed of a protein core, commonly syndecan, covalently attached to one or more HS glycosaminoglycan (GAG) side chains via a linker region (30). HS polysaccharide chains are composed of alternating hexuronic acid and D-glucosamine units (30,31). HS chains undergo extensive modifications during their biosynthesis, including sulfation and epimerization, resulting in a variety of structurally diverse HS chains (30,(32)(33). This diversity allows HS to interact with an array of functionally unrelated proteins and participate in various processes, such as the regulation of embryonic development, angiogenesis, blood coagulation, growth factor/ cytokine interactions, cell adhesion, and lipid metabolism (30).
Much remains to be learned about the cell-cell fusion mechanism and regulation of this phenomenon. The purpose of our study was to examine the effect of HS on cell-cell fusion and how it may function in the fusion mechanism. Using HSV-1 as a tool, we discovered that the absence of HS from the cell surface significantly enhanced the ability of cells to fuse with each other. This effect was also seen independently of HSV-1 in cells that neither expressed HSV-1 glycoproteins nor their receptors. This suggests a novel role for HS as a negative regulator and a fine-tuner of cell-cell fusion events.
(Northwestern University). HeLa cells were provided B. P. Prabhakar (University of Illinois at Chicago). All CHO cell lines were grown in Ham's F-12 medium (Invitrogen) supplemented with 10% fetal bovine serum and penicillin and streptomycin (P/S) (Invitrogen/Invitrogen). HeLa and MRC5 cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% fetal bovine serum (FBS) and penicillin and streptomycin (P/S), whereas Vero cells were grown in DMEM with 10% FBS and P/S. The ␤-galactosidase-expressing recombinant HSV-1 (KOS) gL86 virus strain was provided by P. G. Spear.
HSV-1 Entry Assay-Standard entry assay was used as described previously (35). CHO cell lines grown to subconfluent levels in 6-well plates were transfected with 1 g of nectin-1 or pcDNA3.1 using Lipofectamine (Invitrogen). After 16 h, cells were replated into 96-well dishes, and recombinant HSV-1 gL86 was added in a 2-fold serial dilution. Immediately after virus addition, CHO-677 and CHO-745 cells were spinoculated at 1200 ϫ g for 2 h at 37°C, whereas CHO-K1 cells were incubated at 37°C without spinning. After 2 h, cells were washed with phosphate-buffered saline (PBS) to remove exogenous virus and incubated at 37°C for an additional 4 h. A ␤-galactosidase assay was performed using soluble substrate o-nitrophenyl-␤-d galactopyranoside. Enzymatic activity was measured at 410 nm with a microplate reader (Spectra Max 190, Molecular Devices, Sunnyvale, CA).
For polyethylene glycol (PEG, Roche Applied Science)-induced fusion, target and effectors cells were only transfected with plasmids expressing the luciferase gene and T7 RNA polymerase, respectively. PEG treatment was performed similarly to the protocol provided by the manufacturer with some deviations. Target and effector cells were disassociated from 6-well plates, centrifuged, and mixed in a prewarmed solution of 50% PEG diluted in PBS for 1 min at 37°C. Cells were washed multiple times with prewarmed serum-free media to remove all traces of PEG and replated in 24-well dishes in fresh media.
When treating with cytochalasin D or latrunculin A, target and effector cells were pretreated for 1 h at various concentrations. After treatment, cells were washed and mixed together in fresh media, and fusion was measured after 16 h.
Syncytia Assay-Target cells were additionally transfected with 0.5 g of a plasmid expressing cyan fluorescent protein fused to a nuclear localization signal (Clontech, Mountain View, CA). Effector cells were also transfected with a red fluorescent protein-expressing plasmid fused to a nuclear export signal (NES) (39). Target and effector cells were mixed in a 1:1 ratio, and syncytia size and number were compared after 72 h using microscopy at the 40ϫ objective (Nikon Eclipse TE2000-E).
Heparinase/Chondroitinase and Soluble HS/Chondroitin Sulfate (CS) Treatment-Target, effector, or both cell populations were pretreated with a 1:1 mixture of either heparinase II and III (4 units/ml), chondroitinase ABC (4 units/ml), or a combination of both heparinase and chondroitinase (Sigma). Cells were pretreated with heparinase and/or chondroitinase ABC for 2 h before mixing. After pretreatment, cells were centrifuged at 10,000 ϫ g for 2 min, washed, and mixed together in fresh media. For a control, some cells were mock-treated. For soluble HS and soluble CS assays, either 1 g/l concentrations of soluble HS or 1 g/l concentrations each of soluble chondroitin sulfate A, chondroitin sulfate B, and chondroitin sulfate C (Sigma) was added to target and effector cells immediately after they were mixed together.
Cell Aggregation-Assay was performed as previously described (40,41). Cells cultured in 6-well dishes were dissociated using Hanks'-based enzyme free dissociation buffer (Invitrogen). After washing twice in PBS, cells were resuspended in an aggregation buffer composed of Hanks' balanced salt solution containing 1 mM calcium chloride, 1 mM magnesium sulfate, 1% bovine serum albumin, and 10 mM HEPES. Pipetting was used to disperse any cell aggregates to create single cell suspensions. Next, 0.5 ml of cell suspension was added to 24-well plates previously incubated with 0.5 ml of aggregation buffer overnight at 37°C. Culture plates were rotated on a shaker (100 rpm) at room temperature for 60 min. Cells were removed by pipetting, and wells were washed with 200 l of aggregation buffer. To separate weakly aggregated cells, cells and wash were combined and pipetted several times. An aliquot of cell suspension was placed on a microscopic slide, and aggregate size was compared at the 40ϫ objective (Zeiss Axiovert 200).
Pulldown Assay and Western Blot Analysis-A previously described protocol was followed (42). Target and effector cells were mixed together for a given time period, centrifuged at 10,000 ϫ g for 2 min, and lysed with cell lysis buffer supplemented with protease inhibitor. Cell lysate was centrifuged at 12,000 ϫ g for 15 min at 4°C. Supernatant was transferred to new tubes, and equal amounts of aliquot were removed to measure total Cdc42/RhoA expression as a control. Samples were incubated with either glutathione S-transferase (GST)-Rhotekin-RBD (Pierce) or GST-PAK-PBD beads (Cytoskeleton) for 1 h at 4°C to pull down RhoA GTP and Cdc42 GTP, respectively. Bound proteins were resolved on SDS-PAGE gels and transferred to nitrocellulose membranes. After blocking at room temp for 2 h in Tris-buffered saline with 3% bovine serum albumin, anti-RhoA or anti-Cdc42 antibody (1:500) was incubated overnight at 4°C. After multiple washings, horseradish peroxidase-conjugated anti-mouse IgG (RhoA) or anti-rabbit IgG (Cdc42) (Jackson ImmunoResearch Laboratories) was added for 1 h at room temperature. Immunoreactive bands were developed with Super Signal West Femto Maximum Sensitivity substrate (Pierce) and imaged on Eastman Kodak Biomax MR film. Protein bands were quantitatively analyzed using NIH ImageJ Version 1.41.
Cell Enzyme-linked Immunosorbent Assay-Protocol was similar to that described elsewhere (35,43). Target and effector cells grown in 6-well plates were replated in 96-well dishes. Primary antibodies for gB, gD, gH-gL, nectin-1, or the four syndecan isoforms diluted in PBS with 3% bovine serum albumin (1:500) were incubated with cells for 1 h at room temperature. After washing, cells were subsequently incubated with either biotinylated anti-rabbit IgG or anti-mouse IgG (1:5,000) diluted in PBS with 3% bovine serum albumin and streptavidin-conjugated horseradish peroxidase (1:10,000), each for 1 h at room temperature (Amersham Biosciences) with washing in between. The slow kinetic form of 3,3Ј, 5,5Ј-tetramethylbenzidine substrate (Sigma) was added, and optical density was read at 650 nm with a microplate reader.
Statistics-The data shown are the means Ϯ S.D. of triplicate measures and are representative of three independent experiments. Error bars represent Ϯ 1 S.D.). Statistical significance was calculated using Student's t test. A p value that was Ͻ0.05 was considered statistically significant.

The Absence of HS Increases Cell-Cell Fusion Mediated by
HSV-1 Glycoproteins-Uninfected cells expressing HSV-1 glycoproteins (gB, gD, gH, gL) fuse with cells expressing gD receptors (29). Based on the same principle, cell-cell fusion was com-pared for HS-expressing wild type Chinese hamster ovary (CHO-K1) cells with two HS-deficient CHO cells lines (CHO-677, CHO-745). CHO-677 cells are deficient for the enzymes N-acetylglucosaminyl/glucuronosyltransferase, which are necessary for HS chain synthesis (44). However, these enzymes are not required for the synthesis of other GAGs; thus, although CHO-677 cells are deficient for HS, they do express other GAGs. In CHO-745 cells, the enzyme xylosyltransferase, which is required for the transfer of xylose to the GAG core protein and is the key enzyme of GAG chain initiation, is inactive (44). This prevents the synthesis of all GAGs. All three CHO cell lines were split into two populations: target and effector cells. Target cells were transfected with plasmids expressing a HSV-1 gD receptor, in this case nectin-1, and the luciferase reporter gene under the control of the T7 promoter. Effector cells were transfected with plasmids expressing T7 polymerase and the four HSV-1 glycoproteins essential for cell-cell fusion: gB, gD, gH, gL (29). Fusion of target and effector cells allows T7 polymerase to bind to the T7 promoter, activating expression of the luciferase reporter gene. This allows us to measure cell-cell fusion by measuring the luciferase reporter gene activity.
Luciferase activity was measured at several time points up to 16 h post-mixing of target and effector cells. A significant increase (p Ͻ 0.05) in cell-cell fusion was seen in both HSdeficient CHO lines compared with the HS-expressing CHO-K1 cells (Fig. 1A). For a negative control, target cells were mixed with effector cells lacking gD, which is required for cellcell fusion. The same control was also used for subsequent experiments as well, unless otherwise stated. Necessary controls were tested to verify that changes in cell-cell fusion were specifically because of HS expression and not differences in properties among the CHO cell types, including transfection efficiency, total and surface protein expression for nectin-1 and the four HSV-1 glycoproteins, and surface expression of all four syndecan isoforms. No significant changes were seen among the three CHO cell lines in any of these controls (data not shown), validating the idea that changes in cell-cell fusion was specifically due to HS expression. Not only did CHO-677 and CHO-745 cells show higher cellcell fusion in transiently transfected cells, but the same phenomenon was seen in HSV-1-infected effector cells as well (Fig.  1B). Although target cells were still transfected as discussed earlier, effector cells were transfected with nectin-1 and T7 polymerase. Effector cells were incubated with HSV-1 (KOS) gL86 for 1 h before mixing with target cells. Because HS is required for efficient HSV-1 entry, CHO-677 and CHO-745 cells were spinoculated with HSV-1 (KOS) gL86. Spinoculation is a low speed centrifugation technique that allows HSV-1 to enter HS-deficient cells at levels comparable with HS-expressing cells (45). Spinoculation of CHO-677 and CHO-745 cells resulted in levels of entry similar to entry in unspinoculated CHO-K1 cells (data not shown). As with the transiently transfected cells, an increase (p Ͻ 0.05) in cell-cell fusion was seen in CHO-677 and CHO-745 cells compared with wild type CHO-K1 cells (Fig. 1B).
Syncytia are large, multinucleated cells that form as a result of cell-cell fusion. To compare the number and size of syncytia among each CHO cell line, target cells were additionally transfected with a plasmid expressing a cyan fluorescent protein attached to a nuclear localization signal, limiting expression of cyan fluorescent protein to the nuclei. Effector cells were additionally transfected with a red fluorescent protein attached to a NES, limiting expression to the cytoplasm (39). Syncytia could then be identified as cells expressing red cytoplasm and at least two blue nuclei (Fig. 2). The top panels show representative syncytia formed in HSV-1 mediated cell-cell fusion in each CHO cell type, showing this fluorescent technique can be used to specifically identify syncytia formed by HSV-1 induced cell-cell fusion. The bottom panels, which consist of target cells mixed with effector cells missing gD, show no syncytia formation. Table 1 lists the average number of syncytia formed in each CHO cell type and the average number of nuclei in those syn-cytia. Syncytia from three wells of a 24-well plate were counted, and the average number of syncytia per well was determined. Table 1 shows that CHO-677 and CHO-745 cells formed a significantly greater number of syncytia (p Ͻ 0.05) than CHO-K1 cells, and these syncytia also contained a slightly higher average nuclei count. The specific impact of HS expression was further examined by looking for changes in cell-cell fusion when HS was added to HS-deficient cell lines and when HS was removed from HS-expressing cell lines.
The Removal or Addition of HS Affects Cell-Cell Fusion-To remove HS from the surface of CHO-K1 cells, these cells were treated with identical concentrations of the enzymes heparinase II and heparinase III to cleave HS from the cell surface. Cell-cell fusion in heparinase-treated CHO-K1 cells was compared with fusion seen in cells that were mock-treated. According to the results, removal of HS from the cell surface by heparinase II/III resulted in a significant enhancement (p Ͻ 0.001) of cell-cell fusion in CHO-K1 cells compared with mock-treated cells (Fig. 3A). Heparinase-treated CHO-K1 cells showed levels of cell-cell fusion statistically similar to levels of cell-cell fusion seen in mock-treated CHO-677 (p Ͼ 0.05) but not CHO-745 (p Ͻ 0.05), suggesting CHO-745 cells possess additional properties that allow them to fuse more efficiently. However, based on the CHO-K1 heparinase results, it appears the absence of HS is still primarily responsible for enhanced fusion in CHO-745 cells. CHO-677 and CHO-745 cells showed no significant changes in cell-cell fusion (p Ͼ 0.05) when treated with heparinase, showing the heparinase itself had no impact on cell-cell fusion.
In addition to heparinase treatment alone, CHO cells were also treated with chondroitinase ABC alone or a mixture of heparinase and chondroitinase ABC. Chondroitinase ABC cleaves another GAG from the cell surface, CS. Unlike heparinase treatment alone, treatment with chondroitinase ABC alone did not result in an increase in cell-cell fusion (Fig. 3A). Instead, a slight decrease was seen instead in cell-cell fusion (p Ͼ 0.05). No significant differences in cell-cell fusion were seen (p Ͼ 0.05) in cells treated with heparinase alone compared with those treated with a mixture of heparinase and chondroitinase ABC (Fig. 3A). These results further suggest that it is specifically the removal of HS from the cell surface and not the removal of GAGs in general that is primarily responsible for the increase in cell-cell fusion.
Vero and HeLa cells, two cell lines naturally susceptible to HSV-1 infection, treated with heparinase also showed significantly enhanced fusion in both HeLa (p Ͻ 0.05) and Vero cells (p Ͻ 0.001) compared with mock-treated cells (Fig. 3B),   OCTOBER 23, 2009 • VOLUME 284 • NUMBER 43

Heparan Sulfate in the Regulation of Cell-Cell Fusion
suggesting the effect of HS on cell-cell fusion is not cell type-specific. If the removal of HS can cause an increase in cell-cell fusion in CHO-K1 cells, it was tempting to examine whether the addition of soluble HS to CHO-677 and CHO-745 cells could also affect cell-cell fusion. Soluble HS was added to each CHO cell type immediately after target, and effector cells were mixed. Treatment with soluble HS resulted in a significant decrease in cell-cell fusion (p Ͻ 0.05) compared with mock-treated cells in each CHO cell type (Fig. 3C). A similar decrease in cell-cell fusion was seen in each CHO cell type using concentrations of soluble HS as low as 1 g/ml (data not shown). This decrease raises the possibility that soluble HS may function by competitively binding to a viral glycoprotein (such as gB) or a cellular receptor.
Treatment with soluble chondroitin sulfate A, B, and C also resulted in a decrease in cell-cell fusion; however, unlike soluble HS, this decrease was not statistically significant (p Ͼ 0.05) (Fig.  3C). Similar results were seen using concentrations of soluble CS as low as 1 g/ml (data not shown). Additionally, there was a significant difference (p Ͻ 0.05) between the level of cell-cell fusion after soluble HS treatment and the level of fusion after soluble CS treatment in each CHO cell type. Although the addition of GAGs in general seems to be partly responsible for the decrease in cell-cell fusion, the results suggest that the addition of HS specifically results in a decrease in cell-cell fusion. Next, we further examined the specificity of this impact of HS expression on cell-cell fusion.

The Function of HS during Cell-Cell Fusion Is an Intrinsic
Property and Not HSV-1-specific-After receiving data that suggested the effect on cell-cell fusion was not cell type-specific, we next determined if the effect of HS was (i) gD receptorspecific and (ii) HSV-1 specific. To determine gD receptor specificity, cell-cell fusion was measured with target cells expressing the gD receptor, herpesvirus entry mediator, in addition to those expressing nectin-1. The same increase in cell-cell fusion was seen (p Ͻ 0.01) in CHO-677 and CHO-745 cells when herpesvirus entry mediator was the gD receptor instead of nectin-1 (Fig. 4A), suggesting the effect of HS on cell-cell fusion is not gD receptor-specific.
Other herpesviruses, like HSV-2 and CMV, can also induce cell-cell fusion. To see if HS expression had a similar impact on HSV-2-mediated cell-cell fusion, effector cells for each CHO cell types were transiently transfected with the glycoproteins necessary to induce HSV-2-mediated cell-cell fusion (gB-2, gD-2, gH-2, gL-2), and target cells were transfected with nectin-1, which also serves as a HSV-2 gD receptor. Like HSV-1, expression of HSV-2 gB, gD, gH, and gL in various cell lines results in polykaryocyte formation (37). Significantly higher cell-cell fusion (p Ͻ 0.001) was seen in CHO-745 and CHO-677 cells compared with CHO-K1 cells during HSV-2 mediated cell-cell fusion (Fig. 4B).
For CMV-mediated cell-cell fusion, MRC5 cells were used instead because of their susceptibility to CMV-induced cell-cell fusion (38). Effector cells were transfected with several CMV glycoproteins involved in cell-cell fusion (gB, gH, gL, and gO (38)), whereas target cells were transfected with only the luciferase gene expressing plasmid because MRC5 cells are naturally susceptible to CMV infection. Effector and target cells were pretreated with heparinase II/III before mixing. Heparinasetreated MRC5 cells showed significantly enhanced cell-cell fusion (p Ͻ 0.05) compared with those that were mock-treated (Fig. 4C). Together, the previous few results suggest the effect of HS on cell-cell fusion is neither gD receptor-, cell type-, nor HSV-1-specific.
Considering these results, it is possible that although HS is involved in HSV-1 infection, its role in cell-cell fusion is not limited to HSV-1-mediated cell-cell fusion. These results raise the possibility that HS may be a global regulator of cell-cell fusion. To test this idea, cell-cell fusion was compared in all three CHO cell types in the absence of both HSV-1 glycoprotein and gD receptor protein expression.
PEG was used to induce fusion between two CHO cell populations; that is, those transfected with only the luciferase gene and those transfected with only T7 polymerase. Normally, cellcell fusion cannot efficiently occur between these cells, but in the presence of PEG fusion can be enhanced. HS-deficient CHO-677 and CHO-745 cells showed significantly higher cellcell fusion (p Ͻ 0.01) compared with CHO-K1 cells when induced by PEG (Fig. 4D). Even mock-treated CHO-677 and CHO-745 controls demonstrated higher background fusion than corresponding CHO-K1 cells. These results show the effect of HS on cell-cell fusion is independent of viral protein and host cell receptor interactions, strongly implicating HS in the global regulation of cell-cell fusion.
Negative Charge Repulsion Is Not Responsible for Changes Seen in Cell-Cell Fusion-Removal of sialic acid, a negatively charged cell surface molecule, from the cell surface can result in a significant increase in syncytia formation mediated by human immunodeficiency virus and respiratory syncytial virus (46,47). Like sialic acid, HS is also negatively charged. To test whether the negative charges on HS on the cell surface interfered with the ability of CHO-K1 cells to come into contact with one another compared with HS deficient CHO-677 and CHO-745 cells where the negative charges of HS are absent, we initially performed a cell aggregation assay. No significant changes in aggregate size were seen among the three CHO lines used in our study (Fig. 5A). Thus, negative charges on HS do not likely prevent cells from coming into close contact with each other, and the effect of HS is not at the level of the cell aggregation step. To prove it further, we resorted to the spinoculation procedure (45). Spinoculation may be able to overcome any repulsion created from the net negative charge of the two cell populations (46). It was clear spinoculation of CHO-K1 and, likewise, CHO-677 and CHO-745 cells did not significantly enhance fusion (p Ͼ 0.05) (Fig. 5B), suggesting again that the effect of HS was likely not because of its negative charged nature.  OCTOBER 23, 2009 • VOLUME 284 • NUMBER 43

JOURNAL OF BIOLOGICAL CHEMISTRY 29659
To further dwell on the significance of HS in cell-cell fusion, we asked whether its presence on only target or effector cells had any differential effects on cell-cell fusion. Thus, either the target cells were treated with heparinase II/III and cocultivated with untreated effector cells or vice versa. Noticeably, there was only a slight increase in cell-cell fusion when only the effector cells were treated with heparinase, whereas a significantly higher level of cell-cell fusion was seen when target cells alone were treated (p Ͻ 0.01) (Fig. 5C). If the reduction of the net negative charge of HS was solely responsible, it is assumed heparinase treatment of either population would have resulted in a similar increase in cell-cell fusion as a similar reduction in the net negative charge should have occurred. Hence, it is likely that HS regulates fusion via a mechanism that may not be charge-charge repulsion.
Polymerization of the Actin Cytoskeleton Is Critical for HSV-1-mediated Cell-Cell Fusion-The actin cytoskeleton has been closely associated with cell-cell fusion, and it is believed reorganization of the actin cytoskeleton is a general requirement for this process (48). These rearrangements can be triggered by specific ligand-receptor interactions on the cell surface. Members of the small Rho GTPase family play a critical role in this pathway, specifically Cdc42 and RhoA, which we will focus on in this study. Activation of Cdc42 and RhoA can lead to rearrangements of the actin cytoskeleton required for cell-cell fusion (49).
Because HSV-1 was used as a tool to study the effect of HS on cell-cell fusion, we first wanted to determine whether actin polymerization and reorganization was indeed critical for HSV-1 mediated cell-cell fusion. Each CHO cell type was treated with one of two different actin depolymerizing agents, latrunculin A (Fig. 6A) and cytochalasin D (Fig. 6B), to inhibit actin cytoskeleton reorganization. Cells treated with either actin depolymerizing agent showed a significant decrease (p Ͻ 0.05) in cell-cell fusion in a dose-dependent manner (Fig. 6, A  and B). Therefore, as expected, cell-cell fusion mediated by HSV-1 also requires actin rearrangement.
Rho GTPase Signaling during Cell-Cell Fusion Is Affected by HS Expression-Next, we compared Cdc42 and RhoA activity during cell-cell fusion in HS-expressing CHO-K1 and HS-deficient CHO-745 cells. Although these two Rho GTPases have been shown to be specifically activated during HSV-1 entry (42,50), it is currently unknown if they are also activated during HSV-1-mediated cell-cell fusion. To determine this, pulldown and Western blot assays were used to detect activation of RhoA (Fig. 7A) and Cdc42 (Fig. 7B) during cell-cell fusion. According to the results, both Rho GTPases are activated during HSV-1 mediated cell-cell fusion in both CHO-K1 and CHO-745 cells. RhoA was activated at 20 min after mixing of target and effector cells in both CHO-K1 (Fig. 7A, fourth lane) and CHO-745 (Fig.  7A, sixth lane) cells. Cdc42 was activated earlier than RhoA with activation occurring as soon as 1 min after mixing of target and effector cells for both CHO-745 (left panel) and CHO-K1 (right panel) cells (Fig. 7B, fourth lane for both panels).
To determine the direct effect Cdc42 and RhoA activity has on HSV-1-mediated cell-cell fusion, target and effector cells FIGURE 5. Ability of HS to regulate cell-cell fusion may not solely be because of its negative charge density. A, cell aggregate size was compared in all three CHO cell types. Cells were incubated in an aggregation buffer in 24-well plates previously incubated with 0.5 ml of aggregation buffer overnight at 37°C. Cells were rotated on a shaker (100 rpm) at 37°C for 60 min. Cells were removed from wells by gentle pipetting, and an aliquot was placed on a microscopic slide and viewed (scale bar ϭ 45 m). B, cell-cell fusion was compared in cell populations spinoculated at 1200 ϫ g for 2 h at 37°C after mixing with cells that were not spun. Cells were incubated for another 14 h. C, either both, one, or neither cell populations of CHO-K1 cells were pretreated with equal amounts of heparinase II and III for 2 h before mixing. Fusion was measured 16 h later. Error bars represent Ϯ 1 S.D. wt, wild type.
were transiently transfected with plasmids either expressing a dominant negative mutant of Cdc42 (Cdc42(T17N)), whose mutation abolishes its affinity for GTP, or with a constitutively active form of either Cdc42 (Cdc42(Q61L)) or RhoA (RhoA(Q63L)), whose mutations eliminate its intrinsic GTPase activity. A cell-permeable RhoA inhibitor (Cytoskeleton Inc.) was used to inhibit RhoA activation. When the constitutively active form of RhoA was expressed, a significant increase in cell-cell fusion was seen (p Ͻ 0.05), whereas treatment of target and effector cells with 2.0 g/ml concentrations of the cell permeable Rho inhibitor caused a significant decrease (p Ͻ 0.05) in cell-cell fusion (Fig. 7C).  Expression was measured at 0, 1, 5, and 10 min after mixing (third through sixth lanes, respectively, for both CHO-745 and CHO-K1). Similar controls were used as in A (first and second lanes for both CHO-745 and CHO-K1). Total RhoA (A) or total Cdc42 (B) protein expression was determined for a control. C, target and effector cells were transfected with either a constitutively active form of RhoA or empty vector or were pretreated with 2 g/ml of a Rho cell-permeable inhibitor for 4 h before mixing. D, target and effector cells were transfected with a dominant negative or constitutively active form of Cdc42 or empty vector. For C and D, fusion was measured after 16 h. Error bars represent Ϯ 1 S.D. Western blots are representative of three independent experiments. wt, wild type.
Based on these results, it appears there may be a direct correlation between RhoA activity and cell-cell fusion. Expression of either the dominant negative form of Cdc42 or the constitutively active form of Cdc42 resulted in a significant increase (p Ͻ 0.05) in cell-cell fusion (Fig. 7D). This may suggest that Cdc42 activity is tightly regulated during cellcell fusion, and when disrupted it may enhance cell-cell fusion.
Cotransfection of both constitutively active mutants (Cdc42ϩ and RhoAϩ) resulted in a further enhancement of cell-cell fusion that was significantly higher than the level of fusion seen when Cdc42ϩ or RhoAϩ were transfected alone (p Ͻ 0.05) (supplemental Fig. 1). The level of cell-cell fusion seen in CHO-K1 cells cotransfected with the constitutively active forms of both Rho GTPases was similar to CHO-677 cells (p Ͼ 0.05) and significantly higher than CHO-745 cells (p Ͻ 0.05) transfected with vector alone. This further suggests that Rho GTPase activity may play a critical role in the differences seen in cell-cell fusion in HS-deficient cells.
The results suggest a higher level of Rho GTPase activity in CHO-745 cells compared with CHO-K1. This was confirmed by quantification of the protein bands, which showed a significant increase in the intensity of RhoA bands in CHO-745 cells relative to CHO-K1 cells at both 0 and 20 min. A similar increase was also seen with Cdc42 as well. RhoA and Cdc42 activity was quantified by calculating the relative intensity of each active RhoA/Cdc42 protein band in relation to the corresponding band intensity of its total RhoA/Cdc42 (NIH ImageJ Version 1.41). These data along with the correlation seen between increased levels of RhoA/Cdc42 activity with increased cell-cell fusion (Fig. 7, C and D) suggest enhanced Rho GTPase activity could explain why increased fusion is observed in HS deficient cell lines.
To determine whether HS expression has a specific effect on RhoA and Cdc42 activity, Rho GTPase activity was measured and compared in CHO-K1 and CHO-745 cells either treated with heparinase II/III or mock-treated (Fig. 8). CHO-K1 cells treated with heparinase showed an increase in both Cdc42 (Fig. 8A, fifth lane) and RhoA (Fig. 8C, fifth lane) activity at 5 and 20 min, respectively, compared with those mocktreated at the same times (Fig. 8, A and C, fourth lane). Once again, Cdc42/RhoA activity was quantitatively determined by calculating the relative intensity of each active Cdc42/RhoA protein band relative to its corresponding total RhoA/ Cdc42 protein band intensity (NIH ImageJ Version 1.41). CHO-745 cells showed no difference in Cdc42 (Fig. 8B) and RhoA (Fig. 8D) activity when cells were treated with heparinase (Fig. 8, B and D, fifth lane) compared with those mock-treated (Fig.  8, B and D, fourth lane). This suggests the heparinase itself had no effect on Rho GTPase activity. This helps reinforce the idea that HS is specifically responsible for the difference seen in Rho GTPase activity. However, further work must be done to determine how HS regulates Rho GTPase signaling.

DISCUSSION
Our study has shown a novel role for heparan sulfate in the regulation of cell-cell fusion. Using HSV-1 as a tool, we observed that the absence of HS expression on the cell surface resulted in a significant increase in the fusion event. This same enhancement in fusion was seen independently of HSV-1 induction, suggesting HS function during cell-cell fusion may be a shared characteristic among multiple fusion events. According to the results, HS may be involved in the fine tuning of certain Rho GTPases, which may in turn regulate cell-cell fusion events.
Clearly, there may be additional ways by which HS regulates cell-cell fusion. Based on our results, it does not appear the general removal of negative charge from the surface of the effector and target cells is mainly responsible for enhanced fusion. No significant differences were seen in aggregate size, and spinoculation had no effect either. Because it appears CHO-K1 cells can bind to one another just as well as CHO-677 and CHO-745, this suggests that HS most likely affects the actual fusion phase of cell-cell fusion. In addition, no significant enhancement in cell-cell fusion was seen when another negatively charged GAG, CS, was also removed from the cell surface, further suggesting the general removal of negative charge was not mainly responsible for cell-cell fusion.
Although a significant enhancement in fusion occurred when target cells alone were treated with heparinase, effector cell treatment showed only a minor increase. This finding is similar to the respiratory syncytial virus study mentioned previously, except the authors saw increased cell-cell fusion when the effector cells alone, and not target, had their sialic acid removed (46). The human immunodeficiency virus study instead saw increased syncytia formation when either the virus or target cells had sialic acid removed (47). This led the authors of the human immunodeficiency virus study to conclude that reduced negative charge was mainly responsible for their results, whereas the respiratory syncytial virus study believed this explanation may have only played a minor part in their results (46,47). In our case, because the highest level of fusion was seen when both cell populations were treated, it is possible the reduction of negative charge may play a minor role in enhanced cell fusion.
Our results showed that although heparinase-treated CHO-K1 cells showed similar levels of cell-cell fusion as CHO-677 mock-treated cells, CHO-745 cells still showed significantly higher fusion. Although CHO-745 cells differ from CHO-677 in that CHO-677 cells express other GAGs besides HS, removal of CS from the cell surface did not result in a further increase in cell-cell fusion. This suggests the higher level of fusion in CHO-745 cells is not because of the absence of other GAGs on their cell surface, at least not CS. Further studies must be done to see if other GAGs beside HS and CS have an effect on cell-cell fusion or if CHO-745 cells possess an additional feature allowing them to fuse more efficiently. However, based on the significant increase in cell-cell fusion seen in heparinase-treated CHO-K1 cells, it appears the absence of HS is still primarily responsible for the enhanced fusion seen in CHO-745 cells.
Although a large number of viruses have been shown to bind HS in vitro, only a limited number of studies have examined the importance of HS on viral infection in vivo. Several viruses, such as foot-and-mouth disease virus, Venezuelan equine encephalitis virus, Sindbis virus, Murray Valley encephalitis virus, and classic swine fever virus bind HS as an adaptation to growth in cell culture. However, the acquired ability to bind HS leads to attenuation of infection in vivo, possibly because of a decreased ability to spread via cell-cell fusion (51)(52)(53)(54)(55)(56).
Although only limited information is available on the use of HS by viruses in vivo, the significance of HS in cell-cell fusion is suggested by the following clinically important observations. (i) Expression of syndecan-1, a cell-surface Heparan sulfate proteoglycan, is down-regulated during myoblast fusion (57,58), (ii) during placenta formation heparanase expression is up-regulated when trophoblasts fuse to form syncytiotrophoblast (59), (iii) heparin, which is closely related to HS, increases the risk for osteopetrosis, possibly by negatively regulating cell-cell fusion of macrophages (60), and (iv) fusion prone metastatic tumor cells demonstrate low heparan sulfate proteoglycan expression, suggesting heparan sulfate proteoglycans play a critical role in the regulation of tumor cell metastasis (61)(62)(63)(64)(65)(66)(67).
Involvement of HS in membrane fusion may directly or indirectly influence many other important HS functions as well. Members of the EXT gene family encode HS polymerase and transfer glucuronic acid and N-acetylglucosamine (68). Humans who carry an EXT mutation can suffer from hereditary multiple exostoses, the formation of multiple benign bone tumors, and short stature (69,70). Mice that carry a mutation show axon guidance errors and severe defects during embryonic development, such as failure to gastrulate because of a lack of an organized mesoderm and extraembryonic tissues (69). These defects can lead to early embryonic death (69). Similarly, defects in xylosyltransferase, which initiates glycosylation of proteoglycans, can lead to development of diseases such as diabetes, atherosclerosis, cancer, diabetes, Graves disease, and osteoarthritis (71,72).
Additional defects in other enzymes involved in HS biosynthesis can lead to abnormal mast cell formation, bilateral renal agenesis, decreased insulin secretion, and various developmental abnormalities in the brain, lung, and kidneys (69,(73)(74)(75). These defects as well can often lead to early embryonic or neonatal lethality (69). Although premature death along with the variety of other severe defects caused by HS deficiency make it difficult to specifically look at the effect of HS knock-out on cell-cell fusion in vivo, it would still be important to design innovative ways, such as conditional knockouts, to study how these deficiencies may connect with HS involvement in membrane fusion and the associated cytoskeleton changes.
The reorganization of actin cytoskeleton appears to be a general requirement for the cell-cell fusion mechanism (48). For example, several intracellular proteins involved in Drosophila myoblast fusion have roles in actin cytoskeleton regulation (76 -78). Use of actin depolymerizing agents in our study resulted in a significant decrease in cell-cell fusion, further demonstrating the need for actin cytoskeleton rearrangements for cell-cell fusion and specifically demonstrating its requirement for HSV-1-mediated cell-cell fusion. It is believed Rho GTPases, such as Cdc42 and RhoA, are activated during cellcell fusion to induce these actin cytoskeleton changes.
Our study showed Cdc42 and RhoA were activated during HSV-1-mediated cell-cell fusion. These results correlate with previous studies showing both Cdc42 and RhoA are activated during HSV-1 entry (42,50). However, the current study found a significant increase in active GTP-bound Cdc42 and RhoA protein expression when HS was absent from the cell surface. This was further confirmed by measuring their expression after HS was removed from CHO-K1 cells using heparinase. Expression of constitutively active forms of both Cdc42 and RhoA significantly enhanced cell-cell fusion. This implies enhanced expression of Cdc42 and RhoA in HS-deficient or heparinasetreated cells should result in significantly higher levels of cellcell fusion, possibly explaining why HS-deficient cells show higher levels of cell-cell fusion. Cotransfection of constitutively active forms of both Rho GTPases significantly enhanced fusion in CHO-K1 cells to levels similar to or higher than levels of cell-cell fusion in HS-deficient cells transfected with empty vector, suggesting that activity of these Rho GTPases could play a crucial role in the enhanced cell-cell fusion seen in HS-deficient cells. The results also suggest Cdc42 and RhoA activity must be tightly regulated during cell-cell fusion and HS may be involved in their regulation.
Our study also determined the relative cell surface expression levels of syndecans on wild type and the mutant CHO-K1 cells. Syndecans are a family of differentially expressed proteoglycans commonly attached to HS or CS side chains (79) with multiple diverse functions including important regulatory roles in cytoskeleton rearrangement, cell adhesion, tumor metastasis, and microbial pathogenesis (79 -86). Our results show that each CHO cell type expressed similar levels of syndecan on their surfaces (data not shown). This suggests that differences in Rho GTPase activity seen in our study were not because of differences in syndecan expression, which further implies that HS expression may regulate Rho GTPase activity leading to the differences seen in cell-cell fusion.
Further studies must be done to identify if increased Rho GTPase activity is the cause of the increase in cell-cell fusion or if it is rather an effect of increased cell-cell fusion. If it is the cause, then further work must be done to determine the connection between HS and the regulation of Rho GTPases.