Probing expression of E-selectin using CRISPR-Cas9-mediated tagging with HiBiT in human endothelial cells

Summary E-selectin is expressed on endothelial cells in response to inflammatory cytokines and mediates leukocyte rolling and extravasation. However, studies have been hampered by lack of experimental approaches to monitor expression in real time in living cells. Here, NanoLuc Binary Technology (NanoBiT) in conjunction with CRISPR-Cas9 genome editing was used to tag endogenous E-selectin in human umbilical vein endothelial cells (HUVECs) with the 11 amino acid nanoluciferase fragment HiBiT. Addition of the membrane-impermeable complementary fragment LgBiT allowed detection of cell surface expression. This allowed the effect of inflammatory mediators on E-selectin expression to be monitored in real time in living endothelial cells. NanoBiT combined with CRISPR-Cas9 gene editing allows sensitive monitoring of real-time changes in cell surface expression of E-selectin and offers a powerful tool for future drug discovery efforts aimed at this important inflammatory protein.


INTRODUCTION
Selectins are a family of Ca 2+ -dependent C-type lectins present on the surface of numerous cell types in the cardiovascular system including endothelial cells (E-and P-selectin), platelets (P-selectin), and leukocytes (L-selectin). 1 They interact with cell surface glycans to promote adhesion of hematopoietic cells to vascular surfaces and promote rolling of circulating leukocytes and their delivery to sites of inflammation. [2][3][4] E-selectin (CD62E) is a 115 kDa adhesion molecule that is expressed exclusively by vascular endothelial cells. 2 It is a single-chain transmembrane glycoprotein consisting of an N-terminal calcium-dependent (C-type) lectin domain, an epidermal growth factor domain, a chain of six consensus repeats, a transmembrane domain, and an intracellular cytoplasmic tail. 2,5 Although E-selectin is constitutively expressed on the surfaces of endothelial cells of bone marrow and skin microvessels 2,6,7 in most tissues, E-selectin expression is induced in response to inflammatory cytokines such as tumor necrosis factor alpha (TNFa), interleukin-1b (IL-1b), or lipopolysaccharide (LPS). 8,9 E-selectin expression has also been associated with tumor angiogenesis and metastasis in a variety of cancers. [10][11][12] E-selectin binds to the tetrasaccharide sialyl Lewis X (SLe x ) structure. 12 Ligands for E-selectin which possess sLe x -motifs include P-selectin glycoprotein ligand-1, CD44, and E-selectin ligand-1. 4,13 In addition, a peptide that contains the sequence DITWDQLWDLMK that can bind E-selectin with high affinity has been discovered (E-selectin-binding peptide) using phage display. 12,14 This peptide sequence has been used to target N-(2-hydroxypropyl)-methacrylamide-based co-polymer-doxorubicin conjugates to the tumor vasculature. 12,15 Despite the well-established changes in E-selectin expression deduced by immunohistochemistry in vascular endothelial cells in response to inflammatory mediators, the detailed pharmacological characteristics of E-selectin expression in living endothelial cells have been more difficult to study. This is largely because of the lack of reagents to interrogate the properties of this C-type lectin in live cells in real time, and the limitations that are inherent in its normal low expression in endothelial cells in the absence of inflammation.
To investigate this further, we have used nanoluciferase 16 and NanoLuc Binary Technology (NanoBiT) [17][18][19] to investigate ligand-induced changes in the cell surface expression of E-selectin. To ensure that ligandinduced changes in cell surface expression were monitored in physiologically relevant cells under the control of their native promoters, we have used CRISPR-Cas9 genome editing to attach the 11 amino acid high-affinity NanoBiT tag HiBiT 17,19 to the N terminus of endogenous E-selectin expressed in human umbilical endothelial cells (HUVECs). Following addition of exogenous purified 18-kDa membrane-impermeable LgBiT to HUVECs, it self-complements with HiBiT to reinstate the full-length nanoluciferase enzyme with subsequent luminescence identifying tagged proteins at the surface of living cells. [17][18][19] Using this approach, we demonstrated ligand-induced expression of HiBiT-tagged E-selectin in primary HUVECs in response to TNFa, IL-1a, IL1-b, LPS, vascular endothelial growth factor (VEGF), and histamine.

E-selectin expression in response to TNFa in fixed HUVECs
Initial experiments were undertaken in wild-type HUVECs to confirm that E-selectin was induced by TNFa over 6 h. Immunofluorescence detection of E-selectin in fixed cells using a monoclonal anti-E-selectin antibody and an Alexa Fluor 488-conjugated secondary antibody confirmed that robust E-selectin was observed in most cells following 6 h stimulation with 1 nM TNFa ( Figure 1A; representative image of n = 5). Similar data were obtained in telomerase reverse transcriptase (TERT2)-immortalized HUVECs ( Figure 1A). The response to TNFa in HUVECs was further characterized in fixed cells using an alkaline phosphatase-conjugated secondary antibody and its substrate p-nitrophenol phosphate to allow quantification of E-selectin expression as a color change that could be read on an absorbance plate reader ( Figure 1B). These data showed that 1 nM TNFa produced a significant increase (p < 0.001) in optical density following 6 h incubation ( Figure 1B). The response was concentration dependent and yielded an EC 50 of 0.016 G 0.003 nM (n = 5) ( Figure 1C). Time course studies showed that the response to iScience Article 1 nM TNFa peaked between 6 and 8 h and required more than 1 h stimulation to induce a significant change in expression ( Figure 1D).

E-selectin expression in CRISPR-Cas9 genome-edited HUVECs
In order to evaluate E-selectin expression in living HUVECs, we undertook genome editing to introduce an 11 amino acid HiBiT tag onto the N terminus of endogenous human E-selectin in HUVECs. Initially, a 370 bp fragment of the N-terminal region of E-selectin was generated by PCR amplification and sequenced to confirm the lack of single-nucleotide polymorphisms in the target PAM site. A PAM site immediately following the signal sequence of E-selectin was then identified and used for CRISPR-Cas9 genome editing to introduce a HiBiT tag at the start of the coding sequence for E-selectin ( Figure 2). Initial experiments with 1 nM TNFa stimulation of wild-type HiBiT-tagged HUVECs confirmed that this cytokine produced a robust stimulation of cell surface E-selectin expression (following re-complementation of HiBiT with exogenous membrane-impermeable purified LgBiT; Figure 3A) that was concentration dependent and yielded an EC 50 value of 0.057 G 0.011 nM (n = 5; Figure 3B).
However, one limitation of wild-type HUVECs is the limited number of passages (usually <9 passages) that they can be used over which they maintain their endothelial cell phenotype. As a consequence, the same genome-editing strategy was also undertaken in immortalized TERT2-HUVECs that do not have such strict passage limitations. These cells have been transfected with human TERT to prevent the normal reduction in telomerase expression and subsequent shortening of telomeres that leads to cell senescence. 20 Mixed populations of TERT2-HUVECs expressing HiBiT-tagged E-selectin (TERT2 (mx) HUVECs) demonstrated iScience Article a robust response to 1 nM TNFa that yielded an EC 50 of 0.10 G 0.03 nM (Figures 3C and 3D; n = 5) that was very similar to that observed in mixed populations of edited wild-type HUVECs. Both wild-type and TERT2-HUVECs retained their endothelial phenotype (as judged by expression of the endothelial marker CD31) in the presence and absence of genome editing ( Figure S1). Mixed populations of TERT2-HiBiT-HUVECs were then subjected to dilution cloning to identify cells that were homozygous for HiBiT-E-selectin insertion.
Several clonal lines were identified with clones B4, H3, and C8 producing strong luminescence responses following TNFa stimulation ( Figure S2). B4 and H3 were heterozygous while clone C8 was homozygous for HiBiT-E-selectin. Figure S3 shows the sequencing for clone C8 in the region of the HiBiT insertion. Short tandem repeat genetic profiles of wild-type and HiBiT-E-selectin gene-edited TERT2-HUVECs (clone C8) confirmed identical genome profiles for the TERT2-HUVEC and TERT2-HUVEC-HiBiT-E-selectin clone C8 cell lines (Table S1). Both cell lines had 86.7% identity with the genome profile of the original TERT2-HUVEC cell line deposited with ATCC (CRC-4053). This is above the 80% threshold generally accepted for declaring a match when accounting for genetic drift. 21 The homozygous clone C8 was selected for further study and produced a large concentration-dependent increase in HiBiT-E-selectin expression following 6 h stimulation with TNFa ( Figures 3E and 3F). The EC 50 value obtained for TNFa in this clonal line was 0.06 G 0.02 nM ( Figure 3F; n = 5). Analysis of the HiBiT insert in the genome sequence of this clone confirmed that it was homozygous for HiBiT insertion and yielded a single 45 bp increase in the size of the N-terminal region ( Figure 2C). In contrast, other clones showed two bands equivalent to untagged E-selectin and HiBiT-tagged E-selectin consistent with heterozygous editing of the HiBiT tag on a single allele. In clone C8, the luminescence obtained following stimulation with 1 nM TNFa was sufficiently bright to image using bioluminescence imaging ( Figure 2D). Significant increases in HiBiT E-selectin cell surface expression were also obtained in clone C8 in response to LPS, IL-1a, and IL-1b ( Figure 4 and Table 1).  Figure S4A). Furthermore, similar responses were seen in Western blots of HiBiT-E-selectin TERT2-HUVECs probed with a HiBiT monoclonal antibody over four independent experiments (Figures S5). To ensure that the phenotype and physiological responses of wild-type HUVECs and genome-edited TERT2-HUVECs were similar, we evaluated their ability to support angiogenesis and form new blood vessels. [22][23][24] This was demonstrated by their ability to form endothelial microtubes (microvessels) when grown in a three-dimensional (3D) matrix. [23][24][25] Wild-type HUVECs, TERT2-HUVECS, and HiBiT-E-selectin TERT2-HUVECs were able to elicit tube formation when grown in Geltrex ( Figure 5).

Effect of VEGF 165 a and histamine on basal and TNFa-stimulated E-selectin expression
VEGF and histamine have previously been reported to either induce E-selectin expression (VEGF 165 a) or enhance TNFa-induced expression (Histamine) in vascular endothelial cells. 26,27 In the present study, both VEGF 165 a (100 nM) and histamine (100 nM) were able to produce a significant increase in basal Hi-BiT-E-selectin expression ( Figures 6A and 6C). Furthermore, a significant increase in the response to 1 nM TNFa was obtained in the presence of 100 nM VEGF but not histamine (100 nM) ( Figures 6B and  6D). Similar effects were also seen in non-edited HUVECs using the alkaline phosphatase E-selectin antibody assay in fixed cells ( Figure S6).

Real-time kinetic analysis of TNFa-induced E-selectin expression in living cells
To monitor the real-time kinetics of TNFa-induced expression of E-selectin in living cells, experiments were undertaken with the homozygous TERT2-HUVEC HiBiT E-selectin clone C8. As these kinetic experiments were performed over 15 h, the long acting caged nanoluciferase substrate endurazine was used. This is hydrolyzed at a slow rate by cellular esterases and releases furimazine throughout the experiment. 28,29 Incubation with 1 nM TNFa resulted in a large stimulation of HiBiT-E-selectin cell surface expression that began after 2 h and peaked at 8 h ( Figure 7A). The difference in the peak cell surface expression of HiBiT E-selectin between TNFa (1 nM) and vehicle-treated cells was substantial (p < 0.001; Figure 7B). The stimulation by  Figures 7D-7F). These data confirmed the exquisite sensitivity of this assay to monitor real-time changes in E-selectin expression over 15 h. Furthermore, the peak responses obtained from these kinetic experiments gave an EC 50 value (0.38 nM) for TNFa which was similar to that obtained from endpoint assays ( Figure 7E). Log peak luminescence values obtained at all concentrations of TNFa were significantly different from vehicle controls (p < 0.0001; one-way ANOVA with Dunnett's multiple comparison test; Figure 7F).

DISCUSSION
E-selectin is a Ca 2+ -dependent C-type lectin present on the surface of endothelial cells that can interact with cell surface glycans to promote adhesion and rolling of hematopoietic cells such as circulating leukocytes and neutrophils. [2][3][4] In the present study, we have used NanoBiT 18,19 to investigate ligand-induced changes in the cell surface expression of this C-type lectin. To ensure that ligand-induced changes were monitored in physiologically relevant cells under the control of their native promoters, we used CRISPR-Cas9 genome editing to attach the 11 amino acid HiBiT tag 18,19 to the N terminus of endogenous E-selectin expressed in both HUVECs and immortalized TERT2-HUVECs. 20 Measurement of cell surface expression of HiBiT-tagged E-selectin was then made following addition of exogenous membrane-impermeable LgBiT to achieve self-complementation of the full-length nanoluciferase enzyme. [17][18][19] For the studies in TERT2-HUVECs, we were able to isolate a homozygous clonal cell line that expressed HiBiT on both alleles.
Using this approach, we were able to demonstrate ligand-induced expression of HiBiT-tagged E-selectin in both gene-edited primary HUVECs and gene-edited immortalized TERT2-HUVECs in response to TNFa, IL-1a, IL1-b, LPS, VEGF 165 a, and histamine. The EC 50 values for TNFa, IL-1a, IL1-b, and LPS were very similar between gene-edited primary HUVECs and the clonal gene-edited TERT2-HUVEC cell line (Table 1) Figure 4) while the maximal responses to IL-1a and IL-1b were much lower and accounted for circa 15% of the responses seen with TNFa and LPS (Figure 4). This probably reflects their very different mechanisms of action and receptors involved in stimulating intracellular signaling. 30-32 VEGF 165 a and histamine have previously been reported to either induce E-selectin expression (VEGF 165 a) or enhance TNFa-induced expression (histamine) in vascular endothelial cells. 26,27 In the present study, both VEGF 165 a (100 nM) and histamine (100 nM) were able to produce a small but significant increase in basal HiBiT-E-selectin expression in immortalized TERT2-HUVECs. A small increase in the response to 1 nM TNFa was also observed with VEGF 165 a but not histamine. Similar results were obtained with VEGF 165 a and histamine in fixed un-edited HUVECs using immunohistochemistry ( Figure S6), although only the effect of histamine on the TNFa response was significant in the latter case. iScience Article Time course analysis of the response to TNFa in fixed HUVECs confirmed that the response required more than 1 h stimulation and that the peak was achieved following 6-8 h of stimulation with 1 nM TNFa. However, the gene-edited cells provided an opportunity to dynamically monitor expression of HiBiT-tagged E-selectin in real time in living cells. To facilitate this, we used a stabilized derivative of the nanoluciferase substrate furimazine (endurazine) that slowly releases furimazine following the action of extracellular esterases. 28 This allowed us to monitor the expression of HiBiT-tagged E-selectin on the surface of living geneedited TERT2-HUVECs in real time. LgBiT and endurazine were present for the full 15 h incubation and showed that E-selectin expression in response to 1 nM TNFa was increased following a 2 h lag period and reached a well-maintained plateau between 6 and 8 h. There was no evidence that LgBiT was depleted during the 15 h total incubation since the luminescence detected at the end of 15 h in the continued presence of 50 nM LgBiT was very similar to that achieved if LgBiT was added for 20 min at the end of the 15 h incubation period. Furthermore, the evaluation of low concentrations of TNFa (0.1 nM) in the kinetic assay format yielded exquisitely sensitive responses to very low concentrations of TNFa that showed very similar kinetic profiles to those obtained with higher concentrations.
In summary, the present study has shown that the use of genome editing to apply an 11 amino acid tag

Limitations of the study
The use here of NanoBiT in conjunction with CRISPR-Cas9 gene editing allowed the real-time quantification of cell surface E-selectin expression on HUVECs. However, a limitation of using this technique in wild-type HUVECs (which have a limited window of passages before phenotypic changes are observed) was the inability to clone these cells to produce a homogeneous population of cells positive for the HiBiT insertion. This meant that each experimental replicate required its own electroporation to deliver the CRISPR-Cas9 reagents to the HUVEC. This factor, combined with the inherently lower editing efficiency typically seen for ''knock-in'' edits, resulted in decreased luminescence outputs more influenced by any variability in cell seeding. To address this issue, we generated a CRISPR-Cas9 TERT2-HUVEC clone HiBiT E-selectin (clone C8). This required single-cell seeding of a mixed population of immortalized TERT2-HUVECs and screening of clones for expression and homozygosity iScience Article of HiBiT insertion, which can be a lengthy process. The clone we generated was homozygous for the HiBiT edit, resulting in considerably greater experimental windows with more consistent luminescence output per experiment. This allowed detailed live cell kinetics of E-selectin to be monitored in real time and provides a powerful tool for future drug discovery efforts at this important inflammatory protein. Future drug-screening hits can then be confirmed in wild-type HUVECs and related endothelials cells using CRISPR-Cas9 editing. Phenotypic validation (e.g. CD31 immunolabeling or endothelial microtubes (microvessels) formation when grown in a 3D matrix) should, however, always be performed for all edited cell populations (both mixed and clonal) and/or following dilution cloning. The approach used here was performed under a materials transfer agreement for non-commercial use from ATCC. Any future drug-screening application will, however, require an additional commercial use license from ATCC.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Materials availability
Materials developed from this study are available from the lead author on reasonable request.

Data and code availability
d All data reported in this paper will be shared by the lead contact upon request.
d This study did not generate datasets or code.
d Any additional information required to reanalyze the data reported is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS
Male human umbilical vein endothelial cells (HUVECs) and female human TERT2 immortalised HUVECs were used in this study.

Cell culture
All cell types were grown at 37 C/5% CO 2 and passaged when they reached 70% confluency using phosphate buffered saline (PBS; D8537; Lonza, Germany) trypsin (0.25% w/v or 1% w/v in versine; T4174; Sigma Aldrich CRISPR RNA (crRNA) was designed to be homologous to the N-terminal region of interest on the N-terminus of E-selectin (CGTGGAGGTGTTGTAAGACC) immediately after its signal sequence (TGCTTCTCATTAAAGAGAGTGGAGCC). The sequences were designed using the CRISPOR program design tool (http://crispor.tefor.net), by inputting this N-terminal section of E-selectin genomic sequence (found on University of California Santa Cruz (UCSC) using the latest upload; Dec.2013 (GRCh38/hg38 human assembly) into the software. 35 Potential guide sequences for protospacer adjacent motif (PAM) regions were listed and the highest predicted efficiency sequence and location were selected. Cas9 protein requires a PAM site (specifically NGG for Cas9, where 'N' is any nucleotide base) to serve as a binding signal for Cas9. The selected guide sequence had an MIT specificity score of 88, indicating a low likelihood of offtarget effects. 36 The complete RNA guide sequence used was 5'-CGUGGAGGUGUUGUAAGACC GUUUUAGAGCUAUGCU-3' (with the target sequence in bold).
The repair template was designed to contain the desired HiBiT sequence (GTGAGCGGCTGGCGGCTGT TCAAGAAGATTAGC), with a Glycine-Serine-Serine-Glycine (GSSG) linker (GGGAGTTCTGGC), containing left and right homology arms (71 base pairs in length) on either side to aid the insertion of the HiBiT-GSSG sequence.
The sequence used was 5'CAACAGTACCAAACTCTACCATTTCTTTTCTTTTTCTCCCACTAGTGCTTCTCATTAAAGAGAGTGGAGCC GTGAGCGGCTGGCGGCTGTTCAAGAAGATTAGCgggagttctggcTGGTCTTACAACACCTCCACGGAAGCT ATGACTTATGATGAGGCCAGTGCTTATTGTCAGCAAAGGTACAC-3'. Within the repair template, the PAM site (CCT in bold) was disrupted by the insertion of the HiBiT (underlined) and the GSSG linker (lower case) sequences to prevent re-cutting by residual Cas9. The repair template was synthesized as a single stranded oligo DNA nucleotide (ssODN). All CRISPR/Cas9 reagents (including the tracrRNA) were purchased from Integrated DNA Technologies, Inc. (IDT; Iowa, USA).
Generation of HiBiT E-selectin mixed population of wildtype and TERT2 HUVECs 48h before, HUVECs or TERT2 HUVECs were passaged to attain 80% density at the time of electroporation. On the day of electroporation, HUVECs or TERT2-HUVECs were treated with trypsin (1% w/v in versine) and pelleted. Cells (500,000) were then re-pelleted before resuspending in 100ml warm Ingenio Electroporation Solution. The ribonucleoprotein (RNP) complex was formed by annealing equal parts E-selectin guide RNA and tracrRNA for 5 min at 95 C, before addition of purified Cas9 protein and sterile PBS (20 min at room temperature) to achieve a 2:1 ratio of guide RNA:Cas9 protein (1500nM guideRNA:750nM Cas9 final concentration in electroporation cuvette). The RNP mix was then incubated with HiBiT E-selectin repair template and Electroporation Enhancer, before mixing with resuspended cells gently but thoroughly. Using the Ingenio plastic cell dropper, the mixture was gently transferred into the electroporation cuvette avoiding bubbles. Cells were then electroporated using Nucleofector 2b Device electroporator (HUVEC -human A-034 setting). CRISPR/Cas9 edited HUVEC cells were then resuspended in an appropriate volume of warm Medium 200 before they were seeded at 20,000 cells per well in 0.1% gelatin coated white sided, flat bottomed Greiner 96 well plate (Greiner Bio-one; 655098), containing 50ml pre-warmed Medium 200. Medium 200 was replaced 24h after electroporation. Gene-edited TERT2 HUVECs were resuspended in warmed Medium 200 supplemented with 2.2% LVES and grown in a 25cm 2 tissue culture flask (T25) which had been pre-coated with 0.1% gelatin (20min, 37 C). Media was replaced every 2 days to aid growth of ll OPEN ACCESS