Opposing Roles of Syndecan-1 and Syndecan-2 in Polyethyleneimine-mediated Gene Delivery*

Polyethyleneimines (PEIs) are efficient non-viral vectors for gene transfer. Heparan sulfate proteoglycans have been proposed to be the cell-surface receptors for PEI·DNA complexes (polyplexes). Here, we investigated if syndecan-1 (SDC1) and syndecan-2 (SDC2) are involved in PEI-mediated transfection. Following addition of polyplexes to HEK293 cells, green fluorescent protein-tagged SDCs rapidly formed clusters with PEI that were dependent of lipid raft integrity. However, although SDC1 overexpression slightly enhanced PEI-mediated gene expression, SDC2 dramatically inhibited it. Confocal microscopy analysis showed that SDC1·polyplex endocytosis occurred within minutes after addition of polyplexes, whereas SDC2·polyplex endocytosis took hours. Expression of SDC1 cytoplasmic deletion mutants revealed that the SDC1 cytoplasmic tail is required for gene expression, but not for clustering or endocytosis, whereas overexpression of SDC1/SDC2 chimeras showed that the SDC2 ectodomain is responsible for the inhibitory effect on gene transfer. This study provides evidence that SDCs may have opposing effects on PEI-mediated transfection.

cations, enabling the rapid production of recombinant proteins and viral vectors (5). The linear and branched PEI (BPEI) 25-kDa forms have been shown to be among the most efficient non-viral vectors for in vitro (6 -8) and in vivo (7)(8)(9)(10)(11)(12) gene delivery. A key characteristic of PEIs resides in their intrinsic buffering capacity, a feature also known as the proton sponge effect (1). According to one hypothesis, the high buffering capacity of PEIs leads to osmotic swelling and rupture of endosomes, resulting in the release of polyplexes into the cytoplasm and passive nuclear entry during mitosis when the nuclear membrane temporarily breaks down (13).
Although it is generally accepted that membrane-associated heparan sulfate proteoglycans (HSPGs) play an important role in the uptake of polyplexes, their involvement as receptors for PEI⅐DNA polyplexes remains to be clearly demonstrated. The membrane-associated HSPGs encompass a wide variety of molecules, including betaglycan, also known as TGFR-3 (14); CD44v3 (15); syndecans (SDCs) (16); and glypicans (17,18). The SDC family is composed of closely related proteins (SDC1-4) encoded by four different genes. These proteins constitute the most abundant forms of membrane HSPGs and play a central role in several aspects of cell physiology (19 -22) such as cell adhesion (23), modulation of growth factor activity (24), and organization of the microfilament cytoskeleton (25). The extracellular domain of SDCs, which show little primary sequence homology, possesses three to five glycosaminoglycan attachment sites and has a putative protease cleavage site in its extracellular juxtamembrane domain. SDCs have a highly homologous transmembrane domain, which is essential for their homodimerization and oligomerization (26). The cytoplasmic domain possesses two highly homologous regions (C1 and C2) surrounding a variable region. The C2 region contains a PDZ-binding motif capable of interacting with numerous PDZ domain-containing proteins such as syntenin (27), CASK/Lin-2 (28), synbindin (29), and synectin (26). Notwithstanding their predominance as cell-surface HSPGs that display a high degree of structural and functional heterogeneity, no study has been undertaken to examine the potential role of syndecans in PEI-mediated transfection.
In this study, we address the question of whether SDC1 and SDC2 could act as carriers of PEI⅐DNA polyplexes. We determined the impact of their overexpression on gene transfer and gene expression efficiency and analyzed their fate following addition of polyplexes. We found that although SDC1 enhances polyplex-mediated transgene expression when overexpressed in HEK293 cells, SDC2 significantly reduces it. The endocytosis, kinetics, and intracellular fate of SDC⅐polyplex complexes differ significantly between the two SDCs, and this may be related to their transfec-tion competency. Using cytoplasmic domain deletion mutants of SDC1 and SDC1/SDC2 chimeras, we evaluated the role of their various domains in polyplex-mediated endocytosis and transgene expression. Taken together, these results bring a new level of complexity to PEI-mediated transfection mechanisms and provide evidences that SDC1 may act as a receptor for polyplexes.
PEI Labeling-BPEI was labeled with RITC by modification of a protocol described previously (31). BPEI (10 mg/ml) was mixed with 2 mg/ml RITC in 0.5 M bicarbonate buffer (pH 9.0) and incubated at room temperature under constant agitation for 2 h. The unbound RITC was removed by applying the reaction mixture to a Sephadex G-25 column (Bio-Rad).
Cell Culture and Transfection-HEK293-EBNA1 cells (clone 6E) were cultured in FreeStyle TM 293 expression medium supplemented with 0.1% Pluronic F-68, 50 g/ml Geneticin (G418), and 10 mM HEPES. For protein expression, 1 ϫ 10 6 cells were transfected using 293fectin according to the manufacturer's instructions. Twenty-four hours post-transfection, 1 ϫ 10 6 cells were transfected using PEI⅐DNA complexes (polyplexes) made as follow. Two micrograms of BPEI (for fluorescence microscopy and flow cytometry analysis) or RITC-labeled BPEI (for confocal microscopy analysis) was added to 1 g of plasmid DNA (pTT5 for fluorescence and confocal microscopy analysis or pTT5-DsRed for flow cytometry analysis); both had been previously diluted in 100 l of hybridoma serum-free medium (Invitrogen). Solutions were mixed and incubated at room temperature for 15 min before their addition to the cell culture.
Flow Cytometry Assay-The percentages of both GFP-and Discosoma sp. red fluorescent protein (DsRed)-expressing cells were determined by flow cytometry using an EPICS Profile II (Beckman Coulter) equipped with a 15-milliwatt argon ion laser. Viable transfected cells were quantified by using appropriate gating to exclude dead cells, debris, and aggregates in a forward-scatter-against-side-scatter plot.
Colocalization Experiments and Microscopy-Colocalization of fluorescent proteins and BPEI-RITC was performed using a confocal microscope. Cells were plated for 1 h onto poly-Dlysine-coated glass-bottom 35-mm culture dishes (No. 0, Mat-Tek Corp.). After washing cells with fresh medium, plated cells were transfected with BPEI-RITC for the indicated periods of time. Cells were then fixed in phosphate-buffered saline con-taining 4% formaldehyde for 10 min. After washing the cells twice with phosphate-buffered saline, nuclei were stained with 2 g/ml Hoechst 33258 for 10 min.
Fluorescent images were analyzed either on a confocal microscope (Zeiss Axiovert 200M inverted microscope) using a Plan-Apochromat 63ϫ (numerical aperture 1.4, differential interference contrast) objective or on a conventional inverted fluorescence microscope (Leica DMIL) using c-Plan 40ϫ (numerical aperture 0.5) objective. The confocal microscope was equipped with three lasers. A laser diode (excitation, 405 nm) and a band-pass filter (420 -480 nm) were used to capture the signal recorded as blue; an argon laser (excitation, 488 nm) and a bandpass filter (505-530 nm) were used to capture the signal recorded as green; and finally, a helium/neon laser (excitation, 543 nm) and a band-pass filter (550 -625 nm) were used to capture the signal recorded as red. Zeiss LSM 510 META confocal system version 3.2 software and IrfanView version 3.85 were used for image acquisition by confocal microscopy and conventional microscopy, respectively.

Polyplexes Induce Clustering of SDC1 and SDC2 and Require
Lipid Raft Integrity-Previous reports suggest that HEK293 cells express low or undetectable levels of SDC1, SDC2, or SDC4 (32)(33)(34). Accordingly, we were not able to detect SDCs by Western blotting using specific anti-SDC1-4 antibodies. However, we were able to show the presence of SDC2, SDC3, and SDC4 (but not SDC1) mRNAs by reverse transcription-PCR (data not shown).
To investigate the impact of SDC1 and SDC2 overexpression on PEI-mediated gene expression and to be able to follow their fates following polyplex addition, we tagged the C-terminal end of both proteins with GFP. The GFP-tagged truncated form of CD4 (⌬CD4), which lacks the helix in the cytoplasmic domain, preventing its internalization by Nef (35), was used as a non-HSPG control transmembrane protein. Cells were transiently transfected with each construct using the lipid-based 293fectin reagent. Twenty-four hours post-transfection (hpt), the percentage of GFP-positive cells was determined by flow cytometry. For the whole study, the average transfection efficiency using this method was 77 Ϯ 10% (n ϭ 37). Confocal microscopy revealed that all GFP-tagged proteins were localized mainly to the cell membrane (supplemental Fig. S1), indicating that the GFP tag does not perturb folding or targeting of these proteins.
It was reported previously that SDC4 clustering is induced in the presence of basic fibroblast growth factor-2 (36). Moreover, it has been shown that polyarginine induces clustering of SDC1 (37) and that clustering of SDC2 is crucial for spine formation in neurons (38), underlying the important role of syndecan clustering for some biological activities. To determine whether PEI can induce SDC1 and SDC2 clustering, polyplexes were added to GFP-tagged SDC1-or SDC2-expressing cells, and SDCs were followed by fluorescence microscopy (Fig. 1). Within 10 min of polyplex addition, formation of GFP clusters could be easily monitored for both SDCs, but not for ⌬CD4-GFP. Clustering is indeed very rapid, as it was detectable within 1 min after polyplex addition to cells. 3 Interestingly, most of the cell surface-expressed SDCs were found concentrated in these clusters, indicating that recruitment of SDC1 and SDC2 proteins by polyplexes is very efficient. Clustering of SDC1 and SDC2 was also observed following addition of DNA-free BPEI 3 and linear 25-kDa PEI as well as after addition of polyplexes made with cross-linked PEIs (supplemen-tal Fig. S2) (39). In contrast, no clustering could be detected after addition of the cationic liposomes 293fectin and Lipofectin (supplemental Fig. S2).
It was reported previously that clustering and subsequent endocytosis of SDC1 and SDC4 require the integrity of plasma membrane lipid rafts (40,41). To determine whether lipid raft integrity is also required for polyplex-induced clustering of these receptors, SDC1-and SDC2-expressing cells were treated for 1 h with 10 M methyl-␤-cyclodextrin (M␤CD), a cyclic oligosaccharide known to deplete cholesterol from the plasma membrane and to disrupt lipid raft integrity. Polyplexes were added to cells, and SDC1 and SDC2 clustering was followed by fluorescence microscopy. After 10 min (Fig. 1) and even after longer periods (1 and 3 h), 3 clustering of SDC1 and SDC2 was completely inhibited by M␤CD.
Gene Expression Is Enhanced by SDC1 but Inhibited by SDC2-To determine the impact of SDC1 and SDC2 overexpression on PEI-mediated transgene expression, cells expressing SDC1, SDC2, or ⌬CD4 were transfected with polyplexes containing a DsRed-expressing vector. At 24 hpt, the percentages of double-fluorescent cells (green for SDC and ⌬CD4 expression and red for DsRed expression) were determined by flow cytometry and expressed relative to ⌬CD4 (control), taken as 100%. PEI-mediated transfection of SDC1-expressing cells resulted in an average increase of 12 Ϯ 4% (n ϭ 4) in double-labeled cells compared with control cells. Surprisingly, transfection efficiencies in SDC2-expressing cells were strongly decreased by 80 Ϯ 3% (n ϭ 4) compared with the ⌬CD4 control.
SDC1 but Not SDC2 Rapidly Internalizes Polyplexes-To address this striking difference between SDC1 and SDC2, we examined the fate of polyplexes by confocal microscopy using RITC-labeled PEI (Fig. 2). In good agreement with the above results, SDC1 and SDC2 clustering could be easily monitored at the cell surface 10 min after polyplex addition (Fig. 2, a and b).
In addition, these clusters colocalized with polyplexes, in accordance with the expected capacity of SDC1 and SDC2 to take up PEI⅐DNA complexes. At 1 hpt, most SDC1 clusters were endocytosed along with polyplexes, both appearing concentrated in an area close to the nucleus ( Fig. 2a and the supplemental video). In contrast, SDC2 clusters were still located at the cell surface in tight association with polyplexes ( Fig. 2b). At 6 hpt, several large patches of endocytosed polyplexes were found free of SDC1 clusters (Fig. 2a). The endocytosed polyplexes appeared to be concentrated in large particles as a result of the putative fusion of smaller ones. In stark contrast, only a few polyplexes were found endocytosed at 6 hpt in SDC2-expressing cells (Fig. 2b), and these were still associated with this Redistribution of these proteins was followed by fluorescence microscopy. Formation of clusters could be easily monitored for SDC1 and SDC2 10 min after polyplex addition, whereas no clusters could be observed for ⌬CD4. To explore the role of lipid rafts in polyplex-induced clustering, ⌬CD4-, SDC1-, or SDC2-expressing cells were treated with 10 M M␤CD for 1 hpt. Polyplexes were then added to cells, and cluster formations were followed by fluorescence microscopy. Compared with polyplex addition in untreated cells, no clusters could be observed in M␤CD-treated cells. Scale bar ϭ 10 m. receptor. In both cases, polyplexes completely free of SDCs could be found in the cell cytoplasm at 24 hpt (Fig. 2, a and b). In agreement with a recent report showing that entry of polyplexes into the nucleus occurs during mitosis when the nuclear membrane is temporarily disrupted (13), some polyplexes were also found close to the chromatin in early post-mitotic cells (supplemental Fig. S3a), whereas some cells displayed polyplexes in their nuclei after mitosis (supplemental Fig. S3b).

The Cytoplasmic Tail of SDC1 Is Not Required for Polyplex Endocytosis but Is Necessary for Gene
Expression-Because gene expression was not prevented but was slightly enhanced by overexpression of SDC1, we next wanted to evaluate the importance of its various domains with regard to PEI-mediated transfection. We therefore generated a series of C-terminally GFP-tagged truncated forms of SDC1 (Fig. 3) termed SDC1⌬C2 (without the C2 domain), SDC1⌬VC2 (without the variable and C2 domains), SDC1⌬Cyto (without the cytoplasmic tail), and eSDC1 (the SDC1 ectodomain fused to the CD4 transmembrane domain), which were transfected in HEK293 cells. After 24 h, confocal microscopy analysis revealed that all constructs were localized at the cell membrane, indicating that these deletions did not affect plasma membrane targeting. 3 Cells expressing each construct were then transfected with polyplexes containing the pTT-DsRed vector, and the percentage of cells expressing both fluorescent proteins was determined by flow cytometry 24 hpt. The results are shown in Fig. 3a and are expressed relative to fulllength SDC1-transfected cells, taken as 100%. In SDC1⌬C2-expressing cells, the percentage of DsRed-expressing cells dramatically dropped to a level similar to that obtained with SDC2 (ϳ20%), indicating that the C2 domain is crucial for PEI-mediated transfection. A similar drop in DsRed-expressing cells was obtained for SDC1⌬VC2, SDC1⌬Cyto, and eSDC1 (Fig. 3a).
Confocal microscopy analysis showed that, as observed previously, 10 min after adding polyplexes to wild-type SDC1-expressing cells, PEI rapidly induced formation of clusters with all truncated forms of SDC1 except for eSDC1. 3 Interestingly, at 6 hpt (Fig. 3b), all constructs but eSDC1 were endocytosed and colocalized with polyplexes. In contrast to SDC2, these observations indicate that the inhibitory effect of these constructs on cell transfection could not be due to delayed polyplex endocytosis, but suggest that the intracellular targeting of polyplexes was affected. These results also demonstrate that the C2 domain of SDC1 is critical for its gene delivery capacity. The SDC2 Ectodomain, but Not Its Cytoplasmic Domain, Inhibits Gene Transfer-We next wanted to determine the role of the SDC1 and SDC2 ectodomains during PEI-mediated gene delivery. To address this issue, we constructed two chimeras termed eSDC1tcSDC2 and eSDC2tcSDC1 by exchanging the ectodomains of SDC1 and SDC2. Cells expressing the chimeras were transfected with polyplexes containing the pTT-DsRed expression vector to evaluate their effect on transfection efficiencies as monitored by flow cytometry (Fig. 4a). The results show that replacement of the SDC2 ectodomain with the SDC1 ectodomain (eSDC1tcSDC2) restored transfection. In contrast, replacement of the SDC1 ectodomain with the SDC2 ectodomain (eSDC2tcSDC1) inhibited cell transfection to the same level as did wild-type SDC2, suggesting that the SDC2 ectodomain is responsible for the observed negative effect on cell transfection.
Confocal microscopy analysis was performed with chimera-expressing cells following polyplex addition. As observed previously for SDC1 and SDC2, clusters of both chimeras colocalizing with RITC-labeled polyplexes could be observed 10 min following polyplex addition. 3 Interestingly and in stark contrast to observations obtained wild-type SDC2, endocytosis of both chimera⅐polyplex complexes could be observed at 6 hpt (Fig.  4b). However, although endocytosed polyplexes were almost completely dissociated from eSDC1tcSDC2, as already observed for wild-type SDC1, they remained fully associated with eSDC2tcSDC1 (as for wild-type SDC2) (see Fig. 2b, 6 h panel). This suggests that the ectodomain of SDC2 is responsible for the negative effect on polyplex endocytosis and gene expression.
SDC2 Exerts a Dominant-negative Effect on Transfection-As both SDC1 and SDC2 have the capacity to bind polyplexes and form clusters but dramatically differ in terms of polyplex-induced endocytosis and transfection efficiency, we next examined the impact of their coexpression on these parameters. Cells coexpressing SDC1 and SDC2 were transfected with polyplexes containing the pTT-DsRed plasmid. At 24 hpt, the percentage of double-labeled cells (green for SDC expression and red for DsRed expression) was determined by flow cytometry. The results are expressed relative to the SDC1 control, set as 100%, where 50% of the coding plasmid was replaced with empty pTT5 vector to maintain the DNA/PEI ratio of 1/2. Coexpression of SDC2 with SDC1 resulted in an ϳ60% reduction in transfection efficiency compared with expression of SDC1 alone, indicating that SDC2 has a negative effect on SDC1-mediated PEI transfection (Fig. 5a).
As PEI-mediated transfection in SDC2-expressing cells was probably prevented as a direct or indirect result of a slowed polyplex endocytosis, confocal microscopy analysis was performed after addition of RITC-labeled polyplexes (Fig. 5b). To be able to distinguish between SDC1 and SDC2, GFP fused at the C-terminal end of the full-length SDC2 protein was replaced with BFP. At 10 min post-transfection, GFP-tagged SDC1, BFP-tagged SDC2, and RITC-labeled polyplexes all colocalized in clusters at the plasma membrane. 3 Interestingly, SDC1⅐SDC2⅐polyplex complexes were hardly detected within cells at 6 hpt, as they remained mainly at the cell surface, as observed previously when SDC2 was expressed alone. This result dramatically contrasts with the profile obtained when SDC1 was expressed alone (Fig. 2a), suggesting that SDC2 has a dominant-negative effect on SDC1-mediated polyplex endocytosis and gene transfer to the nucleus.

DISCUSSION
To achieve high levels of recombinant protein expression, the ideal non-viral gene delivery vehicle should possess the following properties: proficient binding to and penetration of target cells, capacity to bypass or escape the endocytotic pathways, efficient routing to the nucleus, and capacity to dissociate from plasmid DNA once in the nucleus. Compared with viruses, which have evolved to enable efficient infection, PEIs display a relatively weak gene transfer capability. The underlying mechanisms responsible for membrane binding and subsequent endocytosis of polyplexes remain largely unknown. In particular, the identity of membrane receptor(s) for polyplexes and the intracellular fate of receptor⅐polyplex complexes remain to be clearly elucidated.
It has been shown that HSPGs are involved in the cellular uptake of polylysine⅐nucleic acid and cationic lipid⅐nucleic acid complexes (42). In mammals, the dominating cell-surface HSPGs are syndecans and glypicans. These molecules are involved in many cellular processes such as differentiation, adhesion, and migration (19). Most of these processes are induced following HSPG binding to various extracellular proteins such as matrix proteins, proteases and their inhibitors, lipases, lipoproteins, and growth factors and their receptors. In the HSPG-negative (and difficult to transfect) cell line Raji, SDC1 overexpression has been found to promote cationic lipidmediated transfection (43). However, no study to date has examined the role of SDCs in PEI-mediated gene delivery. In addition, whether all SDCs share the same capacity to bind and endocytose polyplexes in a way that ultimately leads to efficient gene expression has not been investigated. In this work, we have demonstrated that this is not the case, as SDC1 and SDC2 have opposite effects on PEI-mediated gene transfer in HEK293 cells. As HEK293 cells do not seem to express detectable levels of SDC1 (as observed following Western blot detection and reverse transcription-PCR), the natural receptor(s) to which polyplexes bind and enter cells is not yet known. Nonetheless, the levels of naturally occurring receptor(s) seem not to be limiting in HEK293 cells, as only a slight but significant improvement in gene transfer was observed following SDC1 overexpression.
Clustering of SDCs appears to be an important initial step that mediates their internalization. For example, Fuki et al. (44) reported that internalization of lipase-enriched lipoproteins by SDC1 requires the formation of such clusters. Using GFPtagged SDC1 and SDC2, we have clearly shown that addition of polyplexes rapidly and efficiently induces formation of clusters that colocalize together. These results are consistent with the expected function of SDCs in PEI-mediated transfection and provide the first direct evidence demonstrating that PEI⅐DNA complexes bind to and efficiently induce clustering of SDCs. On the other hand, no cluster could be observed following 293fectin or Lipofectin addition, suggesting that cationic liposomemediated transfection does not involve SDC clustering. These observations appear to contradict a previous study in which HSPGs and particularly SDC1 were shown to favor cationic liposome⅐DNA complex-based gene delivery (43). However, as no evidence of lipoplex-induced SDC1 clustering was provided, it is possible that HSPGs simply promote lipofection by enhancing recruitment of lipoplexes at the cell surface. In addition, it has been reported that the role of HSPGs in cationic lipid-mediated transfection is mostly to protect cells against lipid-mediated cytotoxicity (45).
The integrity of lipid rafts has been shown to be essential for the clustering and biological activity of a wide variety of membrane proteins (46), including SDC1 and SDC4 (40,41). Using M␤CD, we have shown that lipid raft integrity is also required for polyplex-induced clustering of SDC1 and SDC2. Preliminary results indicate that PEI-mediated gene expression is inhibited in a dose-dependent manner by M␤CD treatment. 3 Surprisingly, in our system, the impact of SDC1 and SDC2 expression on PEI-mediated gene expression differs dramatically. Although by their nature it was expected that all SDCs would potentially be membrane receptors for polyplexes, we found that whereas SDC1 potentiated PEI-mediated gene expression, SDC2 significantly compromised it. Confocal microscopy analysis revealed that endocytosis of polyplexes bound to SDC1 occurred rapidly, whereas that of polyplexes bound to SDC2 took hours. Interestingly, many large SDC-free polyplexes could be observed in the cytoplasm at 24 hpt in both cases, suggesting that the presence of polyplexes in the cyto- plasm is not sufficient to enable transfection of cells. This is further supported by previous observation showing that plasmid DNA encoding BFP remains in the cytosol for up to 96 hpt in both BFP-expressing and BFP-nonexpressing cells (47). As SDC1 was tagged with GFP at its C-terminal end, we cannot rule out the possibility that polyplexes were still associated with cleaved SDC1 ectodomains at 24 hpt, although the inhibitory effect of SDC2 was not rescued by SDC1 coexpression. This shows that SDC2 expression has a dominant-negative effect on both overexpressed SDC1 and the naturally expressed receptor(s) present in HEK293 cells. Confocal microscopy analysis also revealed that SDC1 and SDC2 colocalized during PEI-mediated transfection, suggesting that their hetero-oligomerization may occur. The kinetics of polyplex endocytosis in SDC2/ SDC1-expressing cells is, as for SDC2, greatly delayed compared with SDC1-expressing cells. This suggests that, in addition to endocytosis, the kinetics by which polyplexes enter cells may be an important parameter for efficient gene transfer. Taken together, these results support the possibility that the delay in polyplex endocytosis could be responsible for the weak transgene expression observed in SDC2-expressing cells, but further investigation is required to confirm this hypothesis.
The fact that SDC1 but not SDC2 is able to trigger transgene expression raises the question regarding SDC domains that are important for polyplex-mediated clustering, endocytosis, and gene expression. Using a series of SDC1 deletion mutants, we have shown that the C2 domain is crucial for PEI-mediated gene expression, but not for clustering and endocytosis. As the C2 domain can bind proteins containing a PDZ domain such as syntenin (27), CASK/Lin-2 (28), synbindin (29), and synectin (26), syndecan-binding PDZ proteins may play a critical role in PEI-mediated transfection. Although binding to C-terminal peptides of partner proteins appears to be the typical mode of PDZ domain interaction, these domains can also interact with internal peptide sequences (48). It is thus possible that the presence of GFP at the C termini of SDC1 and SDC2 does not interfere with their interaction with PDZ proteins such as syntenin. In fact, the presence of the SDC1 C2 domain is critical because its deletion significantly impairs PEI-mediated gene expression, but not clustering or endocytosis. This is in agreement with a previous study showing that abrogation of the SDC PDZ-binding motif does not impair SDC endocytosis (49) In addition, the fact that polyplexes induced clustering of SDC1⌬Cyto is in good agreement with a previous study showing that the SDC1 cytoplasmic domain is not required for its association with lipid rafts (50). Interestingly, endocytosis of SDC1⌬Cyto was not impaired, which contrasts with previous reports showing that the SDC1 cytoplasmic domain is required for its association with the cytoskeleton (50, 51) and subsequent endocytosis (40). It is possible that polyplexes could link HEK293 endogenous receptor(s) with SDC1⌬Cyto, thus promoting its endocytosis. This is unlikely, however, as no cluster formation could be observed in eSDC1-expressing cells, a result in good agreement with previous studies showing that the SDC transmembrane domain is needed for SDC dimerization and oligomerization (26).
The use of SDC1/SDC2 chimeras revealed that the ectodomain of SDC2 is responsible for the observed inhibition of PEI-mediated gene expression. These results indicate that when fused to the SDC1 ectodomain, the cytoplasmic and transmembrane domains of SDC2 also have the capacity to trigger endocytosis and subsequently gene expression. This indicates that SDC ectodomains may represent an important barrier to polycation-mediated gene expression, depending on the subtype that is prevalently expressed in a particular cell.
Interestingly, regardless of the SDC1 truncation or chimera used, except for eSDC1tcSDC2, clustering and polyplex endocytosis were not prevented, whereas PEI-mediated gene expression was greatly affected. This clearly shows that endocytosis of polyplexes is not sufficient to trigger gene expression, suggesting that proper intracellular routing or trafficking of polyplexes following endocytosis is a critical step for PEI-mediated gene transfer. This study thus provides new insights into the involvement of syndecan domains in polycation-mediated gene expression.
In conclusion, we have shown here for the first time that SDC1 and SDC2 are directly involved in polyplex binding and that SDC2 strongly delays polyplex endocytosis and inhibits PEI-mediated gene expression. Taken together, this work sheds some light on the mechanisms involved during polyplex-mediated transfection and also provides a new function for SDC1 and SDC2. Our results also suggest that the role of each member of the HSPG family in PEI-mediated gene delivery should be investigated individually rather than collectively. In addition, this work implies that the nature of the SDCs expressed in a particular tissue or cell line would have a significant impact in gene therapy applications.