Contribution of syndecans to cellular internalization and fibrillation of amyloid-β(1–42)

Intraneuronal accumulation of amyloid-β(1–42) (Aβ1–42) is one of the earliest signs of Alzheimer’s disease (AD). Cell surface heparan sulfate proteoglycans (HSPGs) have profound influence on the cellular uptake of Aβ1–42 by mediating its attachment and subsequent internalization into the cells. Colocalization of amyloid plaques with members of the syndecan family of HSPGs, along with the increased expression of syndecan-3 and -4 have already been reported in postmortem AD brains. Considering the growing evidence on the involvement of syndecans in the pathogenesis of AD, we analyzed the contribution of syndecans to cellular uptake and fibrillation of Aβ1–42. Among syndecans, the neuron specific syndecan-3 isoform increased cellular uptake of Aβ1–42 the most. Kinetics of Aβ1–42 uptake also proved to be fairly different among SDC family members: syndecan-3 increased Aβ1–42 uptake from the earliest time points, while other syndecans facilitated Aβ1–42 internalization at a slower pace. Internalized Aβ1–42 colocalized with syndecans and flotillins, highlighting the role of lipid-rafts in syndecan-mediated uptake. Syndecan-3 and 4 also triggered fibrillation of Aβ1–42, further emphasizing the pathophysiological relevance of syndecans in plaque formation. Overall our data highlight syndecans, especially the neuron-specific syndecan-3 isoform, as important players in amyloid pathology and show that syndecans, regardless of cell type, facilitate key molecular events in neurodegeneration.

brains and through modulating Aβ aggregation and clearance they promote amyloid pathology 11 . In other studies, SDC1-3 were also found to be associated with the majority of senile plaques in AD brains 36 .
Due to their highly sulfated polyanionic glycosaminoglycan (GAG) chains, SDCs interact with myriad of extracellular cationic ligands and transmit signals from the extracellular space towards the cellular interior, hence influencing cellular metabolism, transport and information transfer 37-39 . As key regulators of cell signaling and biological functions, SDCs also play important roles in the pathogenesis of several human diseases [40][41][42] . Expression of SDCs follow a tissue specific pattern: SDC1 is widely expressed in epithelial and plasma cells, SDC2 in mesenchymal cells, whereas SDC3 is mainly localized to neural tissues. In contrast to other SDCs, SDC4 is abundant in many cell types 38,43 . SDCs share similar structure: a conserved short, one span transmembrane domain (TM) and the approximately 30 amino acid length cytoplasmic domain (CD). Through their CDs, SDCs effect a large number of signaling cascades 44 . The N-terminal extracellular domains (ectodomains) of SDCs contain three GAG attachment sites for HS near the N terminus and may bear chondroitin sulfate (CS) at the juxtamembrane region 45 . The SDC4 ectodomain also compromises a cell-binding domain (CBD) mediating cell to cell attachment 46 . SDCs also mediate cellular uptake of a wide variety of macromolecules, along with viruses and bacteria [47][48][49][50] . During endocytosis, ligand-or antibody-mediated clustering induces the redistribution of SDCs to lipid rafts and stimulation of a lipid raft-dependent, but clathrin-and caveolae-independent endocytosis of the SDC-ligand complex 51 . Contrary to SDCs, the cell surface HSPG glypicans mediate internalization of their ligands primarily through caveolin-dependent endocytosis 52 .
Due to their involvement in endocytosis, SDCs are attractive targets to deliver macromolecules into the cells [53][54][55] . Throughout the years, our research group has been exploring the SDC-mediated uptake of cell-penetrating peptides (CPPs), small cationic peptides enabling the efficient intracellular delivery of membrane-impermeable biomolecules [56][57][58][59] . Cationic CPPs such as Tat, penetratin or polyarginine all utilize SDC-dependent macropinocytic pathways to enter the cells. Since CPPs internalized via SDC-mediated routes can bypass lysosomal degradation, aiding CPP-attached cargoes to exert their bioactivity intracellularly, SDC-mediated uptake and intracellular accumulation of fibrillar Aβ1-42 aggregates could have indeed damaging effects on neurons. The different expression pattern of SDCs in the CNS also makes SDCs exciting molecular targets to be explored in the course of neurodegeneration 60 .
Considering the evidence on the potential involvement of SDCs in membrane attachment and subsequent cellular delivery of Aβ1-42, we decided to analyze the internalization of Aβ1-42 in cellular models expressing distinct SDC isoforms. In order to analyze SDCs' effect of without the interference of other HSPGs or caveolae-mediated endocytosis, we chose the K562 cell line -a human myelogenous leukemia cell line with reportedly no expression of any members of the syndecan or glypican HSPG families, along with no caveolae expression -as model cells to express the various SDC isoforms [61][62][63][64] . Quantitative uptake assays with SDC transfectants enabled quantitative assessment of SDCs' contribution to Aβ1-42 uptake, while studies with SDC structural mutants revealed the role of SDC domains in interaction with the peptide. Utilization of Thioflavin T (ThT) fluorescence assays and scanning electron microscopy also helped to assess the effect of SDCs on Aβ1-42 fibrillation, the molecular mechanism responsible for the formation of toxic amyloid oligomers. Observations obtained on stable SDC transfectants (established in K562 cells) were also confirmed in SH-SY5Y neuronal cells.
Overall our data highlight SDCs, especially the neuron-specific SDC3 isoform, as important players in the amyloid pathology of AD and show that SDCs, regardless of cell type, facilitate key molecular events in neurodegeneration.

Results
Contribution of sDCs to Aβ1-42 uptake and fibrillation. While cell surface HSPGs have been already identified as key targets for the cellular attachment and internalization of Aβ1-42, the exact contribution of transmembrane SDCs to Aβ1-42 endocytosis has not been assessed yet. To enable the exact assessment of SDCs' contribution to cellular uptake of Aβ1-42, while minimizing the interfering effects of other HSPGs or caveolae-mediated endocytosis, stable transfectants of SDCs were created in the K562 cells, a cell line with reportedly low HSPG and no syndecan or glypican expression, along with no detectable levels of caveolin-1, the main component of caveolae [61][62][63][64][65] . The low expression of membrane HSPGs (see Supplementary Fig. S1. for comparison of HS expression of K562 cells to that of SH-SY5Y, a widely used neuronal cell line) and the inability to form caveolae, the source of caveolar endocytosis, makes K562 an ideal human model cell line to study the effect of SDC overexpression on Aβ1-42 uptake without the interference of other HSPGs or caveolae-mediated endocytosis. As the role of HS in the cellular attachment of Aβ1-42 has been already demonstrated, SDC transfectants were standardized according to their HS expression ( Supplementary Fig. S2). (It is worth noting that contrary to HS, we could not detect any CS expression on wild-type K562 cells and SDC transfectants [ Supplementary Fig. S3]). Thus SDC transfectants with similar level of HS expression were selected and, along with wild-type (WT) K562 cells, treated with FITC-labeled Aβ1-42 or transferrin (Trf), the marker of clathrin-mediated endocytosis, for 1, 3, 6 and 18 hours. Contribution of SDCs to cellular uptake and binding of FITC-Aβ1-42 or FITC-Trf was then quantified with quantitative flow cytofluorometric assays. Protein internalization was assessed by adding trypan blue (dissolved at a concentration of 0.25% in ice-cold 0.1 M citrate buffer) 1 min before the analyses, thus extracellular fluorescence of surface bound fluorescent proteins (Aβ1-42 or Trf) was quenched 66 . Extracellularly attached proteins were quantified by the difference of the fluorescence measures of protein-treated samples unquenched and quenched with trypan blue. As Fig. 1. shows, internalization of Aβ1-42 was highest in SDC3 transfectants at every time points. SDC4 transfectants surpassed WT K562 cells in uptake of Aβ1-42 after 6 h of incubation, while SDC2 bested K562 cells in uptake and attachment of Aβ1-42 only at 18 h. Compared to K562 cells, cellular uptake of Aβ1-42 into SDC1 transfectants were lower until 18 h of incubation. On the other hand, uptake of Trf, the marker of clathrin-mediated endocytosis, was higher in K562 cells than in SDC1, 2 and 3 transfectants, while SDC4 exhibited slightly higher Trf uptake levels than WT K562 cells. Attachment of Aβ1-42 was highest in SDC3 transfectants and peaked at 6 h of incubation. Extracellular attachment of Aβ1-42 in SDC4 transfectants was also higher compared to WT K562 cells, while both SDC1 and SDC2 exhibited similar extracellularly attached Aβ1-42 levels as WT K562 cells. In case of Trf, attachment of all cells were quite similar, except for SDC1 that seemed to bind less Trf.
After assessing uptake kinetics of Aβ1-42, we moved on to study the fibrillation of the peptide. It is widely established, that polyanions, like heparin and HS facilitates the fibrillation of Aβ1-42 34,67,68 . After demonstrating SDCs' contribution to Aβ1-42 uptake, we also analyzed SDCs' effect on Aβ1-42 fibrillation. ThT fluorescence studies quantitatively showed that SDCs, especially SDC3 and SDC4 significantly (p < 0.05) facilitate the fibrillation of Aβ1-42 (Fig. 2a). Compared to WT K562 cells with relatively low surface HS expression, SDC3 and SDC4 induce the fibrillation of Aβ1-42 even after 1 h of incubation, as shown by increased ThT fluorescence of Aβ1-42-treated SDC3 and 4 transfectants. Simultaneous scanning electron microscopy studies also confirmed the fibrillation triggering effect of SDCs -especially SDC3 and SDC4 -on surface attached Aβ1-42 (Fig. 2b). Thus scanning electron microscopy also showed increased number of fibrillar Aβ1-42 assemblies on SDC3 and SDC4 overexpressing cells, even after 1 h incubation. Confocal microscopy studies then revealed ThT labeled fibrillar structures inside of Aβ1-42 treated cells, especially in SDC3 and SDC4 transfectants, suggesting while triggering Aβ1-42 fibrillation, SDCs also mediate the intracellular translocation of Aβ1-42 oligomers (Fig. 2c).
Considering the results of the initial uptake kinetics and fibrillation studies, we moved on to analyze the cellular internalization of Aβ1-42 at 18 h of incubation, when differences in Aβ1-42 uptake and fibrillation was the greatest between SDC transfectants and WT K562 cells. Compared to WT K562 cells, all SDC transfectants exhibited higher uptake of Aβ1-42 at 18 h (p < 0.01). Among SDC transfectants, the neuron specific SDC3 internalized Aβ1-42 the most and Trf the least (Fig. 3a-d). Thus the previously observed difference in Aβ1-42 and Trf uptake (along with cellular attachment) became indeed obvious: while SDC1, 2 and 3 all increased internalization of Aβ1-42, cellular uptake of Trf, the marker of clathrin-mediated endocytosis were reduced in all these SDC  attachment to SDC3 transfectants was significantly (p < 0.01) reduced (Fig. 3c,d). Microscopic visualization demonstrated similar results as Aβ1-42-treated SDC transfectants exhibited higher intracellular fluorescent signals (Fig. 3e). On SDC transfectants, significant amount of fluorescence also arose from externally bound FITC-Aβ1-42. Trf treated cells on the other hand, exhibited fluorescent vesicle-like intracytoplasmic structures, a characteristic feature of clathrin-mediated endocytosis, but lacked surface attached aggregate-like structures observed on Aβ1-42-treated SDC transfectants (Fig. 3e). CLSM colocalization studies showed strong intracellular colocalization of SDCs with Aβ1-42 (the Mander's overlap coefficients [MOC] between Aβ1-42 and SDCs were all greater than 0.7, referring to strong colocalization), suggesting the common intracellular pathway SDCs and Aβ1-42 follow during cellular entry (Fig. 4a,b). It is worth noting that like neurons, SH-SY5Y neuroblastoma cells endogenously express SDC3 ( Supplementary Fig. S4) and colocalization of SDC3 with Aβ1-42 could be also demonstrated in the SH-SY5Y cell-line (MOC = 0.81 ± 0.39; Fig. 4e,f). Co-immunoprecipitation on Aβ1-42 treated SDC transfectants and WT SH-SY5Y cells also confirmed that Aβ1-42 binds SDC3 ( Supplementary Fig. S5). Unlike Aβ1-42, Trf did not exhibit colocalization with any SDCs after 18 h of incubation (MOC ≈ 0.25; Fig. 4c,d and f).

SDC-mediated uptake of Aβ1-42 occurs through membrane microdomains.
It has been reported, that HSPGs are endocytosed through a clathrin-and caveolin-independent, but flotillin-dependent route 69 . The flotillin family of membrane proteins, namely flotillin 1 (FLOT1) and flotillin 2 (FLOT2) define specific microdomains in the plasma membrane. Flotillins have been long considered markers of lipid rafts because they are detergent insoluble and float in sucrose density gradients. CLSM studies showed the intracellular colocalization of both flotillins with Aβ1-42 and SDCs in SDC transfectants, while WT SH-SY5Y cells also exhibited intraneuronal colocalization of flotillins with Aβ1-42 and SDC3, suggesting that SDC-dependent uptake of Aβ1-42 occur through lipid rafts ( Fig. 5a-f). With affinity-based proteomics FLOT1 and -2 were detected in a pull-down experiment of SDC4 with 43/210/79.9% and 41/203/72.7% frequency (number of unique peptide/peptide count/coverage, respectively), thus confirming the findings of the CLSM colocalization studies (Supplementary Table S1).

Effect of SDC domains on Aβ1-42 uptake and fibrillation.
To understand the involvement of SDC domains in the interaction with Aβ1-42, we set up studies where various parts of the SDC ectodomain were affected. First we induced undersulfation of SDCs' HS chains with sodium chlorate (NaClO3), a well-known inhibitor of protein sulfation 70,71 . Promoting undersulfation of SDCs with NaClO3 exerted an evident and significant (p < 0.05) inhibition of Aβ1-42 attachment in all cells lines, however the inhibitory effect on uptake was only significant (p < 0.05) in case of SDC3 and 4 transfectants (especially SDC3 and 4) than in WT K562 cells (Fig. 6c). Scanning electron microscopy studies also confirmed the inhibitory effect of undersulfation as fibrillary assemblies of Aβ1-42 were reduced due to NaClO3 treatment (Fig. 6d). Next we created structural mutants of SDC4, an isoform that exhibited high uptake efficacy of Aβ1-42 in the previous uptake assays. As SDC4 has a specific cell-binding-domain (CBD) that mediates cell-to-cell attachment, we also addressed the potential contribution of CBD to interaction with Aβ1-42. Thus a number structural mutants of SDC4 were created, including the CBD. pSi4 made of the CBD and the secretion signal sequence (Si), but lacking any HS chains. HSA.pSi4 mutants on the other hand lack CBD, but contain the HS attachment site (HSA) with the HS chains, while pSi4 mutants possess a truncated extracellular domain made of only the Si of SDC4 (Fig. 7a). All of the above mentioned SDC4 mutants, along with one coding WT SDC4, were tagged with GFP at the juxtamembrane region and expressed in K562 cells. Clones with equal extent of GFP, hence SDC expression were then selected with flow cytometry and -along with WT K562 cells as controls -treated with Aβ1-42 for 18 h at 37 °C. Cellular uptake and attachment was then analyzed with flow cytometry, as well as confocal and scanning electron microscopy. Compared to K562 cells, WT SDC4 transfectants exhibited the most significant (p < 0.01) increase in Aβ1-42 uptake (Fig. 7b,c). Mutants lacking HS chains (i.e. pSi4 or CBD.pSi4) did not induce any increase in Aβ1-42 uptake, suggesting that neither the signal sequence nor the CBD domain interacts significantly with Aβ1-42. Mutants with truncated ectodomain made of only the HSA with the HS chains on the other hand increased Aβ1-42 internalization at an extent similar to SDC4 transfectants, showing the paramount importance of polyanionic HS chains in the interactions with Aβ1-42. Similarly to SDC4 transfectants, microscopic visualization also confirmed the intracellular colocalization of HSA.pSi4 with Aβ1-42 (MOC = 0.74), suggesting that Aβ1-42 enters the cells attached to HS chains of SDC4 (Fig. 7d,e). Scanning electron microscopy then demonstrated a less pronounced fibrillary assemblies of Aβ1-42 on SDC4 mutants lacking HS chains (Fig. 7e).
After assessing SDCs' contribution to Aβ1-42 fibrillation and cellular entry in SDC overexpressing K562 cells, we also analyzed the contribution of SDC3 to Aβ1-42 fibrillation and uptake in the more complex SH-SY5Y neuronal model cell line. Thus SDC3 overexpressing SH-SY5Y cells were created and the effect of SDC3 overexpression on Aβ1-42 fibrillation and uptake cellular was studied. In SH-SY5Y cells, SDC3 overexpression resulted in increased HS expression, along with increased cellular uptake and fibrillation of Aβ1-42 (Fig. 8a-e). Thus a very definite correlation was found between SDC3 overexpression and Aβ1-42 uptake (r = 0.63) or fibrillation (r = 0.80) in SH-SY5Y neurons (Fig. 8c,d). HS expression also exhibited high correlation with Aβ1-42 uptake (r = 0.74) and fibrillation (r = 0.79), further confirming the roles of HS chains in interactions with Aβ1-42 (Fig. 8c,d).

Discussion
There is compelling evidence that Aβ1-42, rendered in β-sheet dominated senile plaques, is a hallmark of Alzheimer's disease (AD) in the brain of late-life dementia patients. Despite of growing criticism on the Aβ hypothesis, new lines of evidence support the concept that an imbalance between production and clearance of Aβ1-42 peptides is a very early, often initiating factor in Alzheimer's disease (AD) 72 . Intraneuronal accumulation of Aβ aggregates due to Aβ re-uptake has been observed to precede extracellular plaque formation in AD 18   exploiting the SDC-mediated macropinocytic delivery of peptides that enables the efficient intracellular transport of bioactive macromolecules 56,58 . Considering the similarities on the cellular uptake of CPPs and Aβ1-42, we investigated the cellular uptake of Aβ1-42 in cellular models specific to various SDC isoforms. Our studies explored the exact contribution of each SDC isoform to Aβ1-42 internalization: overexpression of all SDCs increased the uptake of Aβ1-42, however the increase was fairly different among the various SDC isoforms. Thus the neuron specific SDC3 increased Aβ1-42 uptake from the earliest time points and SDC4, the universally expressed isoform of the family, increased Aβ1-42 uptake at a pace similar to SDC3. Contrary to SDC3 and 4, SDC1-and 2-mediated internalization of Aβ1-42 was markedly slower. SDC-mediated cellular entry of Aβ1-42 via SDCs was shown to occur through lipid rafts and attachment to polyanionic HS chains. Among the various parts of the SDC ectodomain, the HS chains proved to have a major role in interacting with Aβ1-42. As SDC transfectants applied expressed equal amount of HS, the observed difference between the various SDCs in Aβ1-42 fibrillation could be attributed to the structural differences in HS chains of the distinct SDC isoforms. Also the different SDC isoforms interact with different set of intracellular signaling molecules and that could also contribute to the observed difference in Aβ1-42 uptake. The high affinity of Aβ1-42 towards the neuron specific SDC3 isoform, along with the fast and marked uptake into SDC3 transfectants, could account for the tendency of Aβ1-42 affecting neurons: SDC3 on neurons would bind and internalize Aβ1-42 even before it could be cleared off the extracellular space by phagocytic cells. The observation SDC3 overexpression triggers Aβ1-42 fibrillation suggest that significant part of Aβ1-42 enter SDC3 expressing cells in fibrillar form, thus implying the negative effect of SDC3 overexpression on neurons, as observed by Liu et al. in their present study on postmortem human AD brains 11 . The decreased uptake of Trf, the marker of clathrin-dependent endocytosis, into SDC3 transfectants also implies that SDC3 overexpression affects the classical endocytosis, thus clearance of Aβ1-42, leading to intraneuronal accumulation of oligomeric Aβ1-42 via lipid-raft dependent, alternative endocytic pathways.
Overall, our data represents SDCs, especially SDC3 and SDC4 as a preferred cellular targets for Aβ1-42 to enter the cells. Considering significant overexpression of SDC3 and SDC4 in human AD brains, our SDC dependent cellular assays provided results relevant to human conditions and revealed unknown details of SDC-mediated endocytosis and fibrillation of Aβ1-42. The observation that SDC3 overexpression, regardless of cell type, can contribute to fundamental biological processes (i.e. uptake and fibrillation) related to Aβ1-42 pathology, highlights SDCs, especially SDC3, the neuron specific member of the SDC family with increased expression in AD brains as important players in the pathophysiologic events of neurodegeneration.

Peptides.
Fluorescently labeled (FITC, 5-TAMRA and HiLyte ™ Fluor 647) Aβ1-42 was purchased from BACHEM and AnaSpec, respectively. FITC-labeled human Trf, the marker of clathrin-mediated endocytosis was purchased from Thermo Fisher Scientific. SDC constructs, cell culture and transfection. Full-length SDC1-4 and SDC4 deletion mutants were amplified and subcloned in the mammalian expression plasmid obtained from Clontech (pcDNA3, pEGFP). Structural SDC4 mutants were constructed by inserting green fluorescent protein (GFP) into the juxtamembrane region of the extracellular segment. The signals were kept in all cases to orient the proteins into the membrane. Human SDC DNA constructs were prepared as described above and transfected into K562 cells with DMRIE-C (Invitrogen) to establish stable cell lines permanently expressing human SDCs or SDC4 structural mutants 56 . Transfections were carried out according to the recommendations of the manufacturer. After 24 h, cells were incubated with selection medium, containing 0.4 mg/ml G418 (Sigma). Selection medium was changed every second day. After 2 weeks, G418-resistant colonies were analyzed for the SDC expression with flow cytometry using APC-labeled antibodies specific for the given SDC isoforms (R&D Systems). In case of GFP-tagged SDC4 mutants, expression was analyzed with flow cytometry measuring fluorescence intensities of the GFP tags. Thus colonies exhibiting marked SDC expression were chosen for further studies. The erythroleukemia cell line K562 and its SDC clones were grown as a suspension culture in DMEM/F12 medium (Thermo Fischer Scientific) supplemented with 10% fetal calf serum (FCS; Gibco) at 37 °C in a humified 5% CO 2 containing air environment. In case of SDC3 overexpressing SH-SY5Y cells, transfection and selection of SDC3 overexpressing clones were carried out as described above. SDC3 overexpressing SH-SY5Y clones, along with WT SH-SY5Y cells were then grown in Advanced MEM medium (Thermo Fischer Scientific) supplemented with 10% FCS (Gibco) at 37 °C in a humified 5% CO2 containing air environment.

Flow cytometry analysis of HS and CS expression.
As HS was shown to attach Aβ1-42, HS expression of wild-type (WT) K562 cells, SDC transfectants and SH-SY5Y cells were measured with flow cytometry by using anti-human HS antibody (10E4 epitope [Amsbio]) and FITC-or Alexa 647-labeled goat anti-mouse IgG (Sigma) according to the manufacturers' protocols. SDC transfectants with almost equal amount of HS expression were selected for further uptake studies. SDC transfectants were also assessed for CS expression, using monoclonal anti-CS primary antibody (Clone CS-56, Sigma) and FITC-labeled goat anti-mouse IgG (Sigma) according to the manufacturers' protocols. cells, after incubation and washing, the cells (WT K562, SH-SY5Y and SDC transfectants) were resuspended in 0.5 ml of physiological saline. Equal volumes of this suspension and a stock solution of trypan blue (Merck KGaA; 500 μg/ml dissolved in ice-cold 0.1 M citrate buffer at pH 4.0) were allowed to mix for 1 min before the flow cytometric analyses. In this way, sample pH was lowered to pH 4.0, thereby optimizing the quenching effect of trypan blue to quench the extracellular fluorescence of surface bound fluorescent proteins 66 . In the case of the SDC4 mutants treated with HiLyte ™ Fluor 647-labeled Aβ1-42, extracellular fluorescence of surface attached peptides was removed by trypsinization according to the method described by Nakase et al. 78 . Cellular uptake and attachment was then measured by flow cytometry using a FACScan (Becton Dickinson). Cellular attachment was calculated by subtracting intracellular fluorescence (i.e. those quenched with trypan blue or trypsin) from measures of overall fluorescence. A minimum of 10,000 events per sample was analyzed. Viability of cells was determined by appropriate gating in a forward-scatter-against-side-scatter plot to exclude dead cells and debris. After fixation in 4% paraformaldehyde (Sigma), nuclei were stained with DAPI (1:5000) for 5 min, the after three washing with PBS, the samples were embedded in Fluoromount G and distribution of fluorescence was analyzed on an Olympus FV1000 confocal laser scanning as described above. For colocalisation analyses (SDCs with Aβ1-42 or Trf; SDCs with flotillins, SDC3 with Aβ1-42 or flotillins) the images were analyzed in the ImageJ software (NIH, Bethesda, Maryland, US) with the plug-in JACoP28 as described by Wesén et al. 13 21 images (7 images per sample, experiment performed in triplicate) were analysed, and the data is presented as mean ± SEM.

Co-immunoprecipitation experiments. Stable SDC3 transfectants or WT SH-SY5Y cells were incubated
with or without FITC-labeled Aβ1-42 at a concentration for 18 h at 37 °C. After incubation the cells were washed twice with ice cold PBS and treated with cold Pierce IP lysis buffer. Then the cells were scrapped off to clean Eppendorf tubes, put on a low-speed rotating shaker for 15 min and centrifuged at 14,000 g for 15 min at 4 °C. The supernatant were then transferred to new tubes and combined with 5 µg of the human SDC3 antibody (R&D Systems). The antigen sample/SDC3 antibody mixture was then incubated for overnight at 4 °C with mixing. The antigen sample/SDC3 antibody mixture then were added to a 1,5 ml microcentrifuge tube containing pre-washed Pierce Protein A/G Magnetic Beads (Thermo Fisher Scientific) and after incubation at room temperature for 1 hour with mixing, the beads were then collected with a MagJET Separation Rack magnetic stand (Thermo Fisher Scientific) and supernatants were discarded. To eluate the antigen, 100 µl of SDS-PAGE reducing sample buffer were then added to the tubes and samples were heated at 96 °C for 10 minutes and the samples were proceeded to SDS-PAGE.

Affinity-based proteomics of SDC4 interactions.
To reveal the SDC4 protein interactome, whole cell extracts were prepared from stable transfectant of GFP-tagged SDC4, and then a GFP co-immunoprecipitation (GFP co-IP) assay was performed. Protein concentration of the whole cell extract (WCE) was determined by using Bradford Protein Assay (Bio-Rad, Hercules, California, USA), according to the manufacturer's instructions. Total protein extracts (2 mg) were immunopurified (IP) using 40 µl anti-GFP antibody-coupled 50 nm superparamagnetic beads (µMACS GFP Isolation Kit, Miltenyi Biotec, Germany). The unbound material was removed by washing the beads with 2 ml (equal to 50x beads volume) detergent-free buffers as follows: three times with 1x TBS and once with 25 mM ABC(NH 4 HCO 3 ) buffer. The immunopurified proteins were desalted after on-bead-digestion with trypsin (Promega, Germany) 79  acceptance criteria were set as previously published. Close homologues were only reported if at least three unique peptides matched to the protein. Spectral counting was used to estimate relative abundance of individual proteins in the samples: peptide counts of the individual proteins were normalized to the total number of peptide identifications in each sample 81 . Proteins (i) with reproducible detection (|log 2 fold-change| <0.67 between biological replicates), (ii) with at least two identified peptides, (iii) with at least 5% coverage and (iv) with a median-normalized protein binding affinity score above a previously defined cut-off value were considered as proteins that specifically associate with SDC4 82 .
Thioflavin T binding assays. WT  Promoting undersulfation. To study the effect of proteoglycan sulfation, cells were incubated with 60 mM sodium chlorate (NaClO 3 ; Sigma) for 48 h, the processed for the flow cytometry and microscopy studies as described above.
Statistical analysis. Results are expressed as means ± standard error of the mean (SEM). Differences between experimental groups were evaluated by using one-way analysis of variance (ANOVA). Values of p < 0.05 were accepted as significant. Pearson's correlation coefficient was calculated for correlation between the SDC3 or HS expression and Aβ1-42 uptake and fibrillation values of SH-SY5Y cells. For colocalisation analyses, the Mander's overlap coefficient (MOC) was calculated as described by Wesen et al. 13 .