Laminin-guided highly efficient endothelial commitment from human pluripotent stem cells

Obtaining highly purified differentiated cells via directed differentiation from human pluripotent stem cells (hPSCs) is an essential step for their clinical application. Among the various conditions that should be optimized, the precise role and contribution of the extracellular matrix (ECM) during differentiation are relatively unclear. Here, using a short fragment of laminin 411 (LM411-E8), an ECM predominantly expressed in the vascular endothelial basement membrane, we demonstrate that the directed switching of defined ECMs robustly yields highly-purified (>95%) endothelial progenitor cells (PSC-EPCs) without cell sorting from hPSCs in an integrin-laminin axis-dependent manner. Single-cell RNA-seq analysis revealed that LM411-E8 resolved intercellular transcriptional heterogeneity and escorted the progenitor cells to the appropriate differentiation pathway. The PSC-EPCs gave rise to functional endothelial cells both in vivo and in vitro. We therefore propose that sequential switching of defined matrices is an important concept for guiding cells towards desired fate.

Discovery of coating condition appropriate for endothelial differentiation from mesodermal progenitors. Since the vast majority of day 3 cells were positive for KDR, we next explored more appropriate conditions for their differentiation to endothelial lineage. We investigated various matrices onto which day 3 cells were plated and cultured for an additional 4 days in the presence of VEGF. As a result, we found that the non-coated and laminin 411 (LM411)-coated conditions reproducibly induced endothelial commitment with higher purity than other conditions (Fig. 1a-c, Supplementary Fig. 2). Of particular note, LM411 reproducibly presented a higher yield than the non-coated condition while maintaining comparable purity ( Supplementary  Fig. 3). The ECs derived from PSC-EPCs on LM411 possessed the capacities for Ac-LDL uptake and endothelial tube formation (Fig. 1d,e). Interestingly, matrices suitable for undifferentiated human PSCs such as Matrigel and laminin 511(LM511) 7 showed relatively low purity (Fig. 1b), while LM411 could not support PSCs (data not shown). Taken together, these results demonstrated that LM411 acts as a suitable matrix for producing highly purified PSC-EPCs from mesodermal progenitors in day3 cells.
The LM411-E8 fragment improved the endothelial cell yield and angiogenesis capacity.
Laminins are a common ECM component and responsible for various forms of cell-to-basement membrane adhesion 7 . There are 15 laminin isoforms in mammals including humans, among which laminin 411 (LM411) is the major isoform that lines the basal membrane of endothelial cells in capillary vessels and binds mainly to the cell surface transmembrane receptors integrin α 6β 1 and α 7X1β 1 8 . Based on the observation that laminins bind to integrins at their C-terminal region, we generated E8 fragments, which is the truncated form of the laminins that represent the C-terminal region 9 . E8 fragments retain full binding activity toward integrins but lack binding activities to other components, such as heparin/heparan sulfate. E8 fragments of LM511 and LM332 (LM511-E8 and LM332-E8, respectively) possess greater activity of PSC adhesion than their intact forms 10 . Accordingly, we next  Fig. 4a). Undifferentiated PSCs hardly adhered onto LM411-E8 coated dish, as with full-length LM411 (data not shown), but LM411-E8 showed significantly stronger adhesive properties with respect to day3 cells than did LM411 (Fig. 2a). In particular, cell adhesion increased in a dose-dependent manner, even with a higher cellular density (Fig. 2b), and the purity of PSC-EPCs on LM411-E8 continued to exceed 95%, which exceeded that on LM411, resulting in a significant increase in the cell number of PSC-EPCs (Fig. 2c,d, hereafter, mesodermal progenitors fitted to LM411-E8-mediated endothelial differentiation were referred to PSC-MPCs). Endothelial surface markers such as CD146 11 , CD143 12 , Tie-2 13 and EphB4 14 were expressed on day 9 cells, while a pluripotent marker Tra-1-60 was scarcely expressed (Supplementary Fig. 4b). Notably, PSC-EPCs derived on LM411-E8 yielded the endothelial cells that form longer and more branched tubes than those on LM411 ( Fig. 2e-g,  Supplementary Fig. 4c). In addition, PSC-EPCs on LM411-E8 also internalized Ac-LDL ( Supplementary Fig. 4d).
In order to further evaluate the characteristics of PSC-EPCs on LM411-E8, we performed RNA-seq analysis of PSC-EPCs under several coating conditions. The transcriptional profiles of PSC-EPCs were closer to those of the primary endothelial cell-derived cell lines, such as HUVECs and HAECs, than parental ESCs (Supplementary Data are presented as the mean ± SEM (n = 3) and were statistically analyzed using Student's t-test. Representative results of at least three independent experiments are shown. (e-g) Evaluation of angiogenesis properties of PSC-ECs differentiated on non-coated dishes, LM411 or LM411-E8. (e) Representative images of tube-like structures formed by PSC-ECs derived from KhES-1. Scale bar: 500 μ m. (f) Total tube length per well. Data are presented as the mean ± SEM (n = 3) and were statistically analyzed using Student's t-test. Representative results of at least two independent experiments are shown. (g) Branching points per well. Data are presented as the mean ± SEM (n = 3) and were statistically analyzed using Student's t-test. Representative results of at least two independent experiments are shown. (h) PSC-MPC adhesion was abrogated by LM411-E8 (EQ). Data are presented as the mean ± SEM (n = 3) and were statistically analyzed using Student's t-test. Representative results of at least two independent experiments are shown. Fig. 5a,b). As for specific gene expressions, some endothelial development-associated genes, such as LMO2, NOTCH1, HEY2, DLL4 and NRP1 11,15,16 , are upregulated, while function-associated genes, such as IL8, IL6, IL18, NFKB1, EDN1 and CAV1 [17][18][19] , are downregulated in PSC-EPCs. However, representative common endothelial characteristic markers, such as CDH5, PECAM1 and MCAM, are expressed both on PSC-EPCs and primary ECs at comparable levels ( Supplementary Fig. 5d). Moreover, the PSC-EPCs on LM411-E8 belonged to a cluster distinct from those on the full-length LM411/non-coating dish ( Supplementary Fig. 5b), and a gene ontology analysis revealed that PSC-EPCs on LM411-E8 express lines of genes associated with vasculature development and angiogenesis at higher levels compared to those derived under other conditions ( Supplementary Fig. 5c). Taken together, those data support the contribution of LM411-E8 to the endothelial induction in our system both in function, yield and purity.
Initial adhesion of PSC-MPCs to LM411-E8 is dependent on integrin α6β1. We next explored the mechanisms underlying endothelial differentiation on LM411-E8. We first investigated whether the initial adhesion of PSC-MPCs onto LM411-E8 is dependent on the integrin-laminin axis, as adhesion to ligands other than integrins is usually abrogated in laminin-E8 fragments. For this purpose, we pre-treated PSC-MPCs with specific blocking antibodies against the α 6 or β 1 integrin subunits, both of which are expressed on PSC-MPCs ( Supplementary Fig. 6a). Consequently, the adhered cells were significantly decreased to a level comparable to that observed under the non-coating condition ( Supplementary Fig. 6b). Furthermore, we evaluated the degree of cell adhesion onto a mutant fragment of LM411-E8 (LM411-E8(EQ)), in which the Glu residue (E) in the C-terminal tail region of the γ 1 chain of LM411-E8 was replaced with Gln (Q) to nullify integrin binding activity ( Supplementary Fig. 6c). Notably, the adhesion of PSC-MPCs on LM411-E8 (EQ) was significantly decreased and comparable to that seen in the uncoated dishes (Fig. 2h). Overall, these data indicate that surface integrin family proteins, especially integrin α 6β 1, mediate the initial adhesion of PSC-MPCs onto LM411-E8.

LM411-E8 guides PSC-MPCs into endothelial lineage in VEGF-dependent manner.
We next investigated whether LM411-E8 possess the ability to facilitate the endothelial differentiation of adhered PSC-MPCs. For this purpose, we carried out a double-switching assay (Fig. 3a) in which PSC-MPCs once adhered onto LM411-E8 were harvested again and subsequently seeded onto Matrigel, LM511-E8 or LM411-E8. LM411-E8 increased the purity of PSC-EPCs compared to Matrigel or LM511-E8 (Fig. 3b,c), indicating that LM411-E8 not only selects cells via integrin-dependent adhesion but also facilitates the endothelial differentiation of adhered progenitor cells. In the absence of VEGF, however, few PSC-EPCs were induced, even under the LM411-E8-coated condition ( Supplementary Fig. 7).

Single-cell RNA-seq analysis reveals involvement of Rho pathway in the endothelial differentiation of cells on LM411-E8.
In order to investigate the effects of the matrices on endothelial differentiation at the single cell level, we next performed single-cell RNA-seq analysis. We plated PSC-MPCs onto fresh Matrigel-or LM411-E8-coated dishes and sampled single cells at different time points (Fig. 4a, Supplementary  Fig. 8). Consequently, we detected a comparable number of genes in each cell ( Supplementary Fig. 9a,b). Principal component analysis (PCA) depicted a seamless transition of the transcriptional profiles during differentiation at single-cell resolution (Fig. 4b). In this panel, we could putatively classify each sample into six differentiation stages: A (pluripotent stem cells), B (early mesodermal progenitors), C (mesoderm progenitors), D (early endothelial progenitors), E (intermediate endothelial progenitors), F (late endothelial progenitors) and others (Fig. 4b). According to this classification, expression of literature-based representative genes related to mesoderm and endothelial differentiation, such as KDR (NM_002253), CDH5 (NM_001795) and PECAM1 (NM_000442), were confirmed to be orderly up-regulated while a pluripotent stem cell marker ZFP42 (REX1, NM_174900) was down-regulated ( Fig. 4c). Notably, the profiles of PSC-MPCs plated onto LM411-E8 were relatively homogenous and followed an aligned transition from day 5 to day 7. On the other hand, those on Matrigel were relatively heterogeneous, though a small subset of cells kept pace with those on LM411-E8 ( Fig. 4b and Supplementary Fig. 10a). Overall, these results suggest that LM411-E8 facilitated uniform endothelial differentiation.
We next looked for the pathways contributing to endothelial differentiation in our culture. Gene set enrichment analysis (GSEA) on each stage compared to undifferentiated stage A revealed the highly significantly up-represented pathways (FDR < 0.25, p < 0.0075) along the orderly differentiation ( Fig. 4d and Supplementary Table 1). As shown in Fig. 4d, GSEA indicated that our culture successfully transform initial differentiation cells (Stage B; oxidative phosphorylation was elevated from pluripotent stage, compatible to previous reports 20,21 ) into endothelial stages (Stage C to F). Interestingly, expanded enrichment maps (Fig. 4e, Supplementary Fig. 10b,c) figured out the close relationships between cytoskeleton-related Rho pathway and proliferation-related pathways during differentiation, especially in earlier endothelial (D and E) stages. Moreover, additional GSEA analysis revealed that Rho pathway is significantly enriched in cells on LM411-E8 than those on Matrigel in stage D and E, suggesting the function of LM411-E8 in stimulating proliferation of early stage endothelial progenitors via Rho pathway (Fig. 4f,g, Supplementary Table 2). Indeed, though expression of Rho itself was not upregulated, its' activated form in PSC-EPCs show the clear increase on LM411-E8 (Fig. 5a), and chemical inhibition of the Rho-GTPase pathway drastically decreased the yield of PSC-EPCs (Fig. 5b) while maintaining the high purity (Fig. 5c). Considering that Rho signaling pathway is activated downstream to VEGF as well as various adhesion molecules including integrins 22,23 , these data highlight the important role of the Rho signaling pathway in achieving a higher endothelial yield under LM411-E8 and VEGF co-existing conditions. Modification of initial mesodermal induction improved the efficiency of endothelial differentiation. Although our differentiation system produces PSC-EPCs with high purity, the yield of PSC-EPCs is not satisfactory (see Fig. 2c). We considered that this might be because the number of PSC-MPCs adapted to Scientific RepoRts | 6:35680 | DOI: 10.1038/srep35680 LM411-E8-guided differentiation system is still low. Therefore, to improve the yield of PSC-EPCs, we tried to modify the condition of initial differentiation into PSC-MPCs. Since endogenous activation of the Wnt-β -catenin pathway potentiates PSC differentiation towards the primitive streak and mesoderm 24-26 , we used CHIR99021, a potent inhibitor of GSK3β 27 , to activate the Wnt-β -catenin pathway during initial differentiation. Additionally, we adopted a perfectly chemically-defined medium and recombinant LM511-E8 coating during initial differentiation that is applicable to future regenerative medicine (Fig. 6a). This modified defined system maintained the same level of purity without cell sorting (Fig. 6b) but drastically increased the yield of PSC-EPCs, producing 10 to 15 PSC-EPCs from one PSC (Fig. 6c). The modified protocol increased the growth rate in initial differentiation (Fig. 6d). These findings were compatible with the results of a quantitative expression assay, in which the Wnt pathway genes leading to cell cycle activation were up-regulated in PSC-MPCs under defined condition compared to both undifferentiated PSCs and PSC-MPCs under original undefined condition ( Supplementary Fig. 11a,b, Supplementary Tables 3 and 4). The expansion rate of PSC-EPCs from PSC-MPCs was also increased in the defined system (Fig. 6e). The tube formation assay and LDL-uptake assay verified that the PSC-ECs in this system were functional (Fig. 6f,g). Indeed, the representative genes deeply associated with the endothelial function, such as vWF, eNOS and PECAM1, were up-regulated in PSC-EPCs (Fig. 6h). Furthermore, the expression of ICAM-1, a leukocyte adhesion molecule, was up-regulated by TNF-α stimulation (Fig. 6i). Importantly, LM411-E8 still showed significantly higher purity of PSC-EPCs than LM511-E8 (Fig. 6j). This finding supported the importance of guiding function by LM411-E8 for mesodermal progenitor cells, even in the Wnt-activated condition. Finally, we confirmed that the PSC-ECs could give rise to functional blood vessels in vivo. In the Matrigel plug assay, the transplanted PSC-EPCs successfully presented with vascularization perfused with mouse blood, verifying the functional activity of PSC-EPCs plated on LM411-E8 in vivo (Fig. 6k).

Discussion
Here we found that matrix switching onto LM411 and its C-terminal short fragments efficiently guides PSC-MPCs towards endothelial lineage. Our results highlight the importance of signal transduction through cell-matrix interaction for the fate decision of endothelial progenitor cells.  LM411-coating yielded PSC-EPCs at higher purity compared to Matrigel and the other laminin matrix LM511. The part of these results may be associated with the specific binding spectrum of LM411 to integrin family proteins. One possibility is that the narrower binding spectrum of LM411 may increase the apparent specification efficacy by excluding the deviant population at initial adhesion. As shown in the binding assay using specific blocking antibodies to integrin ( Supplementary Fig. 6b), adhesion of PSC-MPCs onto LM411-E8 is exclusively dependent on integrin α 6β 1. Undesirable binding of LM511 to α 3β 1 and α 7β 1 in addition to α 6β 1 8 may hold non-committed progenitors on the matrix. The other possibility is that integrin signaling other than α6β1 during differentiation hamper the commitment of PSC-MPCs into endothelial lineage. While integrin α 6β 1 facilitates endothelial commitment through activation of Etv2, integrin α 3β 1 inhibits this pathway 28 . Since both LM111, a major component of Matrigel, and LM511 bind integrin α 3β 1 in addition to α 6β 1, these matrices might transduce conflicting signals to the progenitor cells, thereby reducing their specific commitment. The results of double-switching assay (Fig. 3a,b) support this possibility, as PSC-MPCs selected on LM411 did not differentiate into endothelial lineage efficiently on subsequent culture on LM511 or Matrigel. RNA expression analysis provided us interesting suggestions about intracellular functions responsible for present results. Previous studies have pointed out Rho family proteins as an important mediator of endothelial cell functions such as angiogenesis and stabilization. Especially, recent studies showed that RhoA mediates VE-cadherin expression and localization, which in turn negatively regulate angiogenesis by blocking VEGF signal as homeostatic mechanism. Present study may suggest the specific role of LM411-integrin α 6β 1 axis for supporting continual signal transduction in specific situation, e.g. embryonic vasculogenesis. However, further investigation will be needed since the possibility that unknown off-target binding to LM411 required for efficient endothelial differentiation still remains.
The sequential switch from LM511-E8 to LM411-E8 in chemically defined medium successfully provided a novel defined system for inducing endothelial cells from PSCs, with more than 95% purity, without cell sorting: this rate is more efficient than that described in previous reports with a purity of 40% 12 , 55% 29 and 55.5% 30 . The yield of PSC-EPCs per single PSC was approximately 10 to 15, which is also larger than most previous reports 31 . Although the yield in our system remains unsatisfactory, the possibility of combining our protocol with the scalable expansion of endothelial progenitor cells 11 should be taken into account to facilitate future clinical applications.
In conclusion, the present study demonstrated the concept of matrix switching for the field of in vitro directed differentiation culture. In addition to the endothelial lineage, recent studies have reported LM511-E8 32 and laminin 111 33,34 to be useful matrix components for delivering midbrain dopaminergic neurons and hepatocytes, respectively. We therefore propose that sequential switching of defined ECMs depending on the lineage and stage of cells is a reasonable strategy for establishing efficient differentiation systems.   anti-mouse TER-119 antibodies (BioLegend), rat anti-human CD49f antibodies (GoH3, BD), anti-human CD29 antibodies (AIIB2; developed by Caroline H. Damsky and obtained from Developmental Studies Hybridoma Bank) were used in this study. Rabbit polyclonal antibody against ACID/BASE coiled-coil peptides was produced by immunizing with ACID/BASE coiled-coil peptides as an immunogen. HRP-conjugated streptavidin was purchased from Thermo Scientific (Rockford, IL).

Maintenance of human pluripotent stem cells (PSCs). PSCs were maintained on SNL feeder cells in
ES medium supplemented with 5 ng/mL of bFGF. At passage, the cells were dispersed with CTK solution (0.25% trypsin, 0.1% collagenase IV, 20% KSR and 1 mM CaCl 2 ) for 30 seconds at room temperature, and SNL feeder cells were removed as previously reported 36 . Dish and plate coating. BD falcon non-coating tissue culture plates and dishes were used in this study.
Differentiation of PSC-EPCs. PSC clumps were plated onto GFR-Matrigel-coated plates at a density of 2 clumps/cm 2 and cultured in mTeSR1. When the colonies grew to a diameter of 750 μ m, the medium was replaced with fresh mTeSR1 containing BMP4 (80 ng/mL). After three days, the cells were dissociated to the single-cell level using TrypLE Express for 20 minutes at 37 °C and subsequently plated onto a matrix-coated plate and cultured for an additional four days in Stempro34SFM (Life technologies) containing VEGF (80 ng/mL). In the modified system, PSC clumps were plated onto LM511-E8-coated plates at a density of 5 clumps/cm 2 and cultured in mTeSR1. When individual colonies grew up to the diameter of 750 μ m, the medium was replaced with Essential 8 (Life technologies) containing CHIR99021 (4 μ M), BMP4 (80 ng/mL) and VEGF (80 ng/mL). After two days, the whole culture was dissociated, plated onto LM411-E8-coated plates and cultured for four days in Stempro34SFM containing VEGF (80 ng/mL).

Flow cytometric analysis.
Flow cytometric analysis data were collected using LSRFortessa (BD) and subsequently analyzed with the FlowJo software package (Treestar).

LDL-uptake assay and immunocytochemistry. Cells were reacted with Ac-LDL in Endothelial
Serum-Free Medium (ESFM) for 5 hours at 37 °C in 5% CO 2 and then washed with PBS (− ) twice, fixed with Cytofix for 5 minutes and permeabilized with Perm/Wash at room temperature for 30 minutes. The primary antibody reaction with sheep anti-human CD31 antibodies (1:10) was performed at 4 °C overnight, and the cells were then washed with Perm/Wash twice. The secondary antibody reaction with FITC-conjugated anti-sheep IgG antibodies (1:100) was performed at room temperature for one hour. After washing with Perm/Wash, fluorescence imaging was performed using a FluoView FV10i confocal microscope (OLYMPUS).
Tube formation assay. The tube formation assay was performed as previously reported 28 . For gel preparation, 50 μ L of Matrigel was dispensed into each well of a 96-well plate and incubated at 37 °C for 30 minutes. The cells were suspended in ESFM supplemented with 80 ng/mL of VEGF, seeded at a density of 4 × 10 4 /well and cultured overnight. For measurement of the tube length and number of blanching points, the Image J software package (NIH) was used.
Matrigel plug assay. The Matrigel plug assay was performed according to a previous report with some modifications 37 . Briefly, 1 × 10 6 cells suspended in 100 μ L of GFR-Matrigel supplemented with bFGF (300 ng/mL) were subcutaneously injected into the back skin of immunodeficient NOG mice (approximately 6 weeks old) 38 . The Matrigel plugs were harvested 21 days after transplantation and subjected to immunofluorescent staining.

Immunohistochemistry.
Matrigel block was fixed with 4% paraformaldehyde overnight at 4 °C, equilibrated in 20% sucrose/PBS (− ) overnight at 4 °C and frozen embedded into O.C.T. compound. The embedded block was sectioned at a thickness of 6 μ m and adsorbed onto glass slides. The slides were dried in air, fixed with Cytofix for 5 minutes at room temperature and permeabilized with Perm/Wash for 30 minutes at room temperature.

RNA extraction and real-time quantitative RT-PCR analysis.
RNA samples were first prepared by silica gel membrane-based spin-columns (RNeasy -Kit; Qiagen, Valencia, CA, USA) in accordance with the manufacturer's instructions. RNAs were then subjected to an RT reaction followed by cDNA amplification in accordance with the previously reported method 39 with minimal modification. For targeted detection of four genes (vWF, eNOS, PECAM1 and GAPDH) and pathway-focused profiling (PrimerArray ® Wnt signaling pathway [#PH010] and PrimerArray ® cell-cycle [#PH002], TaKaRa), quantitative RT-PCR experiments were performed with a StepOnePlus ™ Real-Time PCR System (Applied Biosystems) using a SYBR Premix Ex Taq II kit (#RR820S, TaKaRa), in accordance with the manufacturer's instructions. Quantitative assessment of the expression was performed using the standard Δ CT method applying GAPDH as a single internal control (targeted detection), or a combination of GUSB, HPRT1, PGK1, ACTB, GAPDH, TBP, B2M and PPIA for multiple references (PrimerArray).

RNA sequencing.
Total RNA was extracted using the RNeasy kit as described in previous subsection, and the quality was tested with Bioanalyzer (Agilent). Ribosomal RNA was depleted by RiboZero Gold (Epicentre). RNA libraries were prepared with the Illumina TruSeq Stranded Total RNA Sample Prep kit and sequenced in the 100 cycle Single-Read mode of HiSeq2500 (Illumina). All sequence reads were extracted in FASTQ format using BCL2FASTQ Conversion Software 1.8.4 in the CASAVA 1.8.2 pipeline and mapped to hg19 reference genes, downloaded on 10 th December, 2012, using TopHat v2.0.8b 40 and quantified by RPKM for Genes 41 , downloaded on 19th October, 2012. Gene Ontology analysis was done using David v6.7 42 . Total RNA samples were applied to RNA sequencing, and the data were analyzed using the Gene Spring software package.
Single-cell RNA sequencing. Cells were dispersed with TrypLE Express for 20 minutes at 37 °C. Single cells were sorted using FACS Aria II and dropped into 96-well plates filled with 10 μ L of Reaction buffer of SMARTer Ultra Low Input RNA -HV kit (Clontech), followed by cDNA synthesis and amplification according to the manufacturer's instruction. Sequencing libraries were constructed using Nextera XT DNA Sample Prep kit (Illumina). The libraries were sequenced and all sequence reads were extracted, mapped and quantified as above.
Gene Set Enrichment Analysis (GSEA). GSEA (http://software.broadinstitute.org/gsea/index.jsp) was used to estimate biological signatures. Data sets filtered by the criterion requiring RPKM > 0 in at least one sample were applied for each analysis with Molecular Signatures Database v5.1 (Hallmark and BioCarta gene sets, size-filters set at 12-500) using gene-set permutation. Networks were visualized by Cytoscape (http://www.cytoscape.org) (P-value Cutoff 0.05, FDR Q-value Cutoff 0.25 and Overlap Coefficient Cutoff 0.01 were used). GSE accession number. RNA-seq datasets deposited in the GEO database can be accessed with the GEO accession number GSE85784.
Cell adhesion assay. The cell adhesion assay was performed as previously reported 10 . Ninety-six-well plates coated with matrices were blocked with 1% BSA in DMEM/F12 for 1 hour at room temperature. The cells were suspended in DMEM/F12 containing 0.1% BSA and plated onto coated wells. After 30 minutes of incubation, non-adherent cells were removed via pipetting with DMEM/F12 twice. The remaining adherent cells were fixed with 10% formaldehyde for 15 minutes at room temperature and post-fixed with 100% ethanol for 5 minutes at room temperature. The fixed cells were then stained with 0.4% crystal violet in methanol (100%) for 5 minutes at room temperature. After washing extensively with deionized water, the remaining crystal violet in the cells was eluted with 1% SDS and quantified by measuring the optical density at 570 nm using a multi-well plate reader. For the functional inhibition assay, cells were pre-incubated with neutralizing antibodies in DMEM/F12 containing 0.1% BSA for 30 minutes on ice before plating.