Optimization of a Western blot protocol for the detection of low levels of tissue factor in human cells

Abstract

Background

The tissue factor (TF)/activated factor VII complex is the major activator of the coagulation system. TF is expressed by a variety of cells, including activated monocytes and tumor cells. Increased TF expression can cause thrombosis in different diseases, including sepsis, viral infections, and cancer. We have previously described a method for analyzing human TF in high-expressing cells by Western blotting.

Objectives

The goal of this study was to establish a method for detecting human TF in low-expressing cells.

Methods

We examined the ability of 3 different antibodies to detect TF in low-expressing cell lines: rabbit polyclonal anti-human TF antibody NBP2-15139 (Novus Biologicals), goat polyclonal anti-human TF antibody AF2339 (R&D Systems), and rabbit monoclonal anti-human TF antibody ab252918 (clone EPR22548-240; Abcam). We also used the Abcam antibody to measure TF expression in lipopolysaccharide-stimulated peripheral blood mononuclear cells.

Results

We found that sensitivity was affected by various factors, including the blocking conditions, the detection method, and the primary and secondary antibodies. Both the R&D and Abcam antibodies were more specific in assessing TF expression than the Novus antibody; however, the Abcam antibody was the best of the 3 in evaluating TF in low-expressing cell lines. We detected TF in lipopolysaccharide-stimulated human peripheral blood mononuclear cells using the new method with the Abcam antibody.

Conclusion

Researchers should consider each step in Western blotting when establishing a method for detecting low-abundance antigens, such as TF.

Essentials

• •TF is the main initiator of blood coagulation and is present at low levels.
• •We tested the ability of 3 antibodies to detect low TF levels in human cells by Western blotting.
• •The antibody ab252918 (clone EPR22548-240) from Abcam was the best of the 3 antibodies tested.
• •Antibodies should be validated contextually and include appropriate positive/negative controls.

1. Introduction

Tissue factor (TF) is a receptor and cofactor for factor (F)VII/activated FVII [1]. The TF-activated FVII complex is the major physiological activator of blood coagulation [1]. Increased levels of TF in the blood on activated monocytes and extracellular vesicles are associated with disseminated intravascular coagulation and thrombosis in endotoxemia, sepsis, viral infection, and cancer [[2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12]]. Bacterial lipopolysaccharide (LPS) stimulation of human monocytes or the human acute monocytic leukemia cell line THP-1 induces TF expression and the release of TF-positive extracellular vesicles [[13], [14], [15], [16], [17]]. However, the TF levels observed in monocytes and monocytic cells in response to LPS are much lower than those in cancer cell lines, like the human pancreatic adenocarcinoma cell line BxPC-3. It has been estimated that an LPS-stimulated monocyte expresses ∼17,000 TF molecules per cell compared with >350,000 TF molecules per cell for BxPC-3 [18,19]. Therefore, optimizing protocols for detecting low levels of TF protein is useful to the field.

Western blotting is a core technique used to measure the level of a specific protein in a given sample using antibodies directed against the protein of interest. Antibodies obtained from commercial sources should be independently validated. The International Working Group for Antibody Validation provided guidelines for antibody validation [20]. These guidelines included assessing antibody specificity through genetic strategies (ie, knockout [KO] or knockdown cell lines) and reporting antibody information (eg, catalog [cat] number, lot number, and concentration of the antibody). The committee also suggested additional validation may be required, as changes in experimental contexts can impact the ability of a given antibody to detect its antigen [20]. For example, antibodies can have differing affinities for the native and denatured forms of an antigen [21]. In addition, antibodies for Western bloting can be further validated using approaches such as deglycosylation, which results in a band shift. Importantly, TF is a glycosylated protein that has an apparent molecular weight of ∼50 kilodaltons (kDa) due to glycosylation rather than the predicted molecular weight of ∼36 kDa based on the amino acid sequence [1].

We have previously described a method for detecting TF by Western blotting in BxPC-3 cells [22]. The goal of this study was to establish a Western blotting protocol for the detection of human TF in low-expressing cell lines and LPS-stimulated peripheral blood mononuclear cells (PBMCs).

2. Methods

2.1. Antibodies

The primary and secondary antibodies used in this study are listed in Table 1. Antibodies were prepared and stored according to the manufacturer’s instructions. Antibodies suspended in phosphate-buffered saline (PBS) were stored in single-use aliquots to avoid repeated freeze/thaw cycles.

Table 1.

Antibodies used in this study.

Antibody	Supplier	Catalog no.	Lot no.	Stock concentration (mg/mL)	Dilution
Rabbit anti-human TF polyclonal	Novus Biologicals	NBP2-15139	43005	0.64	1:1000
Goat anti-human TF polyclonal	R&D Systems	AF2339	VUC0324011	0.2	1:1000
Rabbit anti-human TF monoclonal (EPR22548-240)	Abcam	ab252918	GR3270793-2, 1096425-7	0.493, 0.514	1:1000
Rabbit anti-human TF monoclonal (EPR22548-232)	Abcam	ab228968	1060895-25	0.521	1:1000
Mouse anti–β-actin monoclonal	Santa Cruz Biotechnology	sc-47778	B2724	0.2	1:5000
Mouse anti–β-actin monoclonal	Cell Signaling Technologies	3700	21	0.312	1:5000
Rabbit anti–α-actinin polyclonal	Cell Signaling Technologies	3134	2	0.005	1:1000
IRDye 680RD goat anti-rabbit IgG	LI-COR Biosciences	925-68071	D40618-21	1	1:10,000
Goat anti-mouse IgG Alexa Fluor 790	Thermo Fisher Scientific	A11357	2983151	2	1:10,000
Donkey anti-rabbit IgG DyLight 800	ImmunoReagents	DkxRb-003-F800NHSX	29-55-080312	1	1:10,000
Goat anti-rabbit IgG-HRP	Thermo Fisher Scientific	31460	ZI399931	0.4	1:5000
Rabbit antigoat IgG-HRP	Thermo Fisher Scientific	31402	094179I	0.4	1:5000
Donkey antigoat IgG-HRP	Thermo Fisher Scientific	A15999	94-105-092923	0.5	1:5000
Mouse antigoat IgG-HRP	Santa Cruz Biotechnology	sc-2354	D1823	0.4	1:5000
Horse anti-mouse IgG-HRP	Cell Signaling Technologies	7076P2	39	0.184	1:5000

HRP, horseradish peroxidase; Ig, immunoglobulin; TF, tissue factor.

2.2. Cell lines

BxPC-3 (human pancreatic adenocarcinoma, cat number CRL-1687) and THP-1 (human acute monocytic leukemia, cat number TIB-202) cell lines were obtained from the American Type Culture Collection. HAP-1 wild-type (WT; cat number C631) and HAP-1 TF KO (cat number HZGHC004781c001) cells (human myelogenous leukemia) were obtained from Horizon Discovery. HAP-1 TF KO cells were generated by creating an 8-base pair deletion in exon 2 of F3 (TF) using CRISPR-Cas9, which introduced a frameshift mutation.

BxPC-3 and THP-1 cells were cultured in RPMI 1640 Medium (Gibco, cat number 11875-093) supplemented with 10% fetal bovine serum (FBS; Omega Scientific, cat number FB-02), 1% penicillin-streptomycin (Gibco, cat number 15140-122), and 1% L-glutamine (Gibco, cat number 25030-081). HAP-1 WT and HAP-1 TF KO cells were cultured in Iscove’s Modified Dulbecco’s Medium (Gibco, cat number 12440-053) supplemented with 10% FBS and 1% penicillin-streptomycin. All cells were maintained in a 5% CO₂ incubator at 37 °C. THP-1 cells were incubated with Dulbecco’s PBS (DPBS; Gibco, cat number 14190-144) at 37 °C with 5% CO₂ (control) or stimulated with LPS (serotype O111:B4, Millipore Sigma, cat number L2630-100MG, 10 μg/mL) for 5 hours [14]. LPS was reconstituted in PBS at a concentration of 1 mg/mL and stored in single-use aliquots at −20 °C until use. All cell lines were routinely tested for mycoplasma using a Mycoplasma PCR Detection Kit (Applied Biological Materials, cat number G238), and were negative for mycoplasma.

2.3. Collection of blood from human volunteers

Four healthy human volunteers provided informed written consent prior to blood collection. The donors were aged 25 to 35 years, and the majority were male (75%). This study was approved by the Institutional Review Board of the University of North Carolina at Chapel Hill. All blood was collected and processed within 1 hour of collection. Whole blood was collected into a vacutainer containing 3.2% sodium citrate (BD, cat number 366560) using a 21-gauge needle. The first 3 mL of blood was discarded, and a total of 10 mL (2.7 mL per vacutainer) was collected from each volunteer for further analysis.

2.4. Isolation of PBMCs from human whole blood and stimulation with LPS

Ficoll-Paque PLUS density gradient medium (Cytiva, cat number 17144002) was warmed to room temperature prior to use. Three milliliters of Ficoll-Paque were added to a 15 mL centrifuge tube. The whole blood (2 mL) was diluted 1:1 with room temperature DPBS. The blood-DPBS mixture was layered on top of the Ficoll-Paque medium. The samples were centrifuged at 400 × g for 30 minutes at 20 °C with no brake. The upper layer of the gradient, containing plasma, was removed. The next layer, containing PBMCs, was then removed and transferred to a new tube. PBMCs were washed with 3 volumes of DPBS and centrifuged at 400 × g for 10 minutes at 20 °C. Cell pellets were resuspended in 8 mL of DPBS and centrifuged again at 400 × g for 10 minutes at 20 °C. Pellets were then resuspended in 200 μL of RPMI 1640 Medium supplemented with 10% FBS per tube. Resuspended cells from the same donor were then combined, and the cells were counted manually. Cells were diluted to 4 × 10⁶ cells/mL.

For each donor, a sample of the PBMCs was immediately processed (control), and another was stimulated with 0.1 μg/mL of LPS serotype O128:B12 (Millipore Sigma, cat number L2887) for 5 hours with gentle rocking at 37 °C. A total of 2 × 10⁶ cells in 0.5 mL were stimulated in each tube. Following stimulation, samples from the same donor were pooled (total of 6 × 10⁶ cells per donor) and centrifuged at 2500 × g for 15 minutes at room temperature. The cell pellets were washed with DPBS and centrifuged again at 2500 × g for 15 minutes at room temperature. The cell pellets were then stored at −80 °C until they were used for Western blotting.

2.5. Quantification of F3 (TF) mRNA

Total RNA was extracted from cell pellets using an RNeasy Mini Kit (Qiagen, cat number 74104). The optional on-column DNase treatment was also performed according to the kit instructions. RNA concentration was determined using a NanoDrop (Thermo Fisher Scientific). Complementary DNA (cDNA) was synthesized using 500 ng of total RNA (25 ng cDNA/μL) and the iScript Reverse Transcription Supermix cDNA synthesis kit (Bio-Rad, cat number 1708841). F3 (TF) mRNA was quantified using real-time quantitative polymerase chain reaction with SYBR Green (Thermo Fisher Scientific, cat number 4309155). Primers were synthesized by Integrated DNA Technologies. The primers for F3 were forward, 5′-gtagagtgtatgggccagga-3′ and reverse, 5′-ttcagtggggagttctcctt-3′. 18S ribosomal RNA was used as a control. The primers for 18S ribosomal RNA were forward, 5′-ctcaacacgggaaacctcac-3′ and reverse, 5′-cgctccaccaactaagaacg-3′. The reactions were prepared according to the kit instructions. Reactions were performed at half volume (25 μL instead of 50 μL). A total of 50 ng of cDNA was input into each reaction, and the cycling conditions were performed according to the manufacturer’s instructions. Amplification was measured using the QuantStudio 3 Real-Time PCR System (Thermo Fisher Scientific). Results were analyzed by determining the ΔCycle threshold (CT) of each sample (ΔCT = F3 (TF) CT−18S ribosomal RNA CT), where a lower number signifies a high level of F3 mRNA.

2.6. Cellular TF activity assay

Frozen cell pellets were washed 3 times in 1 mL of 25 mM HEPES buffer (Corning, cat number 25-060-CI). For each wash, cells were centrifuged at 500 × g at 4 °C for 5 minutes, and the supernatants were discarded. After the final wash, pellets were resuspended in 150 μL of 25 mM HEPES buffer. Samples were then sonicated 3 times at 4 °C under the following conditions: amplitude = 50, process time = 1 minute, pulse on = 10 seconds, pulse off = 10 seconds (∼8000 J each; 24,000 J total) using a Q700 Sonicator (Qsonica). A STart4 coagulation analyzer (Diagnostica Stago) was used to measure the coagulation time. For each reaction, 50 μL of human plasma (Thermo Fisher Scientific, cat number 100595) and 50 μL of standard or sample were incubated for 1 minute at 37 °C in a cuvette (Diagnostica Stago, cat number 38876) with a metal bead (Diagnostica Stago, cat number 26441). Then, 50 μL of 25 mM CaCl₂ (Thermo Fisher Scientific, cat number C69-500) was added to the reaction, and the reaction timer started. Clotting times were recorded for each standard or sample. A standard curve was created using recombinant human TF (Innovin, Thermo Fisher Scientific, cat number 10873566). TF activity was normalized using the total protein concentration of the sample as determined by the Pierce Bicinchoninic Acid Protein Assay Kit (Thermo Fisher Scientific, cat number 23227).

2.7. Preparation of cell lysates for sodium dodecyl sulfate–polyacrylamide gel electrophoresis

Adherent cells were washed once with DPBS, followed by trypsinization. Trypsinized cells were suspended in cell growth media. Suspension and trypsinized adherent cells were pelleted by centrifugation at 3000 × g at 4 °C for 5 minutes. Supernatants were removed, and the cell pellets were washed once with DPBS. Cells were centrifuged again at 1000 × g at 4 °C for 5 minutes, and the cell pellets were stored at −80 °C until use. To prepare cell lysates, frozen pellets were thawed on ice and resuspended in 100 μL of RIPA Lysis Buffer (Thermo Fisher Scientific, cat number 89900) containing a protease inhibitor cocktail (cOmplete Protease Inhibitor Cocktail, Millipore Sigma, cat number 11697498001). Resuspended cell pellets were vortexed for 10 seconds and placed on ice for 10 minutes. Vortexing and resting on ice were repeated for a total of 3 times. Lysates were then centrifuged at 10,000 × g at 4 °C for 10 minutes. The supernatant, containing the soluble protein fraction, was then transferred to a new tube. The protein concentration of the lysates was quantified using the bicinchoninic acid assay.

2.8. Acetone precipitation

Due to the low protein concentration of the samples, acetone precipitation was performed for all samples (except for 1 μg of total protein) to permit loading of 40 μg of total protein per sample onto the gel. To each volume of cell lysate, 4 volumes of 100% ice-cold acetone were added. Samples were incubated at −20 °C overnight. Proteins were pelleted by centrifugation at 15,000 × g at room temperature for 10 minutes. Supernatants were removed, and the pellets were air-dried for 30 minutes at room temperature. Dried pellets or 1 μg of total protein were resuspended in 12 μL of 1× reducing loading buffer containing 2.5% β-mercaptoethanol (GenScript, cat number M00676). Samples were incubated for 10 minutes at 95 °C prior to loading.

2.9. Deglycosylation

Deglycosylation was performed using the PNGase F kit (New England Biolabs, cat number P0704S) according to the manufacturer’s instructions for denaturing conditions. Up to 20 μg of total protein per 1× reaction was added to 1 μL of Glycoprotein Denaturing Buffer (10×) and brought up to 10 μL with water. The reactions were incubated at 100 °C for 10 minutes before being placed on ice and centrifuged briefly. Then, 2 μL GlycoBuffer 2 (10×), 2 μL 10% NP-40, and 6 μL H₂O were added to each reaction. Next, 1 μL of PNGase F was added. For control reactions, 1 μL of water was added instead of PNGase F. Reactions were incubated at 37 °C for 1 hour. Due to the final sample volume, samples were acetone precipitated following deglycosylation.

2.10. Western blotting

Samples were run under reducing conditions and loaded onto 4% to 20% Mini-PROTEAN TGX Gels (Bio-Rad, cat number 4561096). A protein molecular weight marker (PageRuler Plus Prestained Protein Ladder, Thermo Fisher Scientific, cat number 26619) was used. Proteins were separated by electrophoresis at a constant ampere (0.01-0.02 A) using tris-glycine-sodium dodecyl sulfate buffer (Bio Basic, cat number A0030). Proteins were transferred to 0.45 μm polyvinylidene difluoride membranes (MilliporeSigma, cat number IPFL00010) via wet transfer. Transfer buffer (2.4 g tris base [Thermo Fisher Scientific, cat number BP152500], 14.4 g glycine [Thermo Fisher Scientific, cat number G46500], and 20% methanol [Thermo Fisher Scientific, cat number BP1105-4] in a total volume of 1 L of ultrapure water) was made in-house. Proteins were transferred for 2 hours at 100 V at 4 °C. All the membranes were blocked for 1 hour at room temperature using either Pierce Protein-Free Blocking Buffer (Thermo Fisher Scientific, cat number 37572) or 5% nonfat milk in Tris-buffered saline (TBS)-Tween (Walmart, nonfat dry milk powder), as indicated in the figure legends. The blots were incubated with a primary antibody overnight at 4 °C with rocking. The membranes were then washed 3 times with TBS (Boston Bioproducts, cat number BM-300) with 0.1% Tween 20 (Thermo Fisher Scientific, cat number BP-337-100) for 10 minutes each. The membranes were then incubated with a secondary antibody diluted in blocking buffer for 1 hour at room temperature with rocking. Following this incubation, the membranes were washed 2× for 10 minutes each and 1× for 5 minutes with TBS-Tween. We used 2 different detection methods (fluorescence or chemiluminescence) with the exposure times indicated in the figure legends. The blots were visualized using an iBright FL1000 imaging system (Thermo Fisher Scientific). Following visualization of anti-TF antibody binding to the membrane, membranes were stripped for 10 minutes at room temperature using the Restore Western Blot Stripping Buffer (Thermo Fisher Scientific, cat number 21059) according to the manufacturer’s instructions and reblocked using blocking buffer. The membranes were then reprobed for the loading control proteins β-actin or α-actinin.

3. Results

3.1. Measurement of TF expression in human cell lines expressing high and low levels of TF

We selected a human pancreatic cancer cell line that expresses a high level of TF (BxPC-3), which we used in our previous publication on TF Western blotting [22]. We also chose 2 other human cancer cell lines expressing much lower levels of TF (HAP-1 WT and THP-1), and 1 that does not express TF protein (HAP-1 TF KO). The frameshift mutation present in the HAP-1 TF KO cells does not eliminate mRNA synthesis but can cause transcription to decrease and/or the resulting mRNA to be unstable, leading to degradation. Therefore, HAP-1 TF KO cells express a low level of F3 (TF) mRNA but do not express TF protein. Additionally, monocytes and monocytic cells, like THP-1 cells, express more F3 mRNA and have increased TF activity following LPS stimulation compared to unstimulated cells [[13], [14], [15]].

We measured the levels of F3 mRNA and TF activity in the different cell lines. As expected, we found that BxPC-3 cells have considerably more F3 mRNA (lower ΔCT) and TF activity compared with the other cell lines (Table 2). THP-1 cells stimulated with LPS had more F3 mRNA and TF activity compared with unstimulated THP-1 cells. HAP-1 TF KO cells had the lowest levels of F3 mRNA and no TF activity (Table 2).

Table 2.

Tissue factor expression in the cell lines used in this study.

Cell line	ΔCT (F3 − 18S)	TF activity (pg/μg total protein)	TF activity (fold change relative to HAP-1 WT)
BxPC-3	6.89 ± 0.21	219.23 ± 15.18	79.47 ± 5.50
THP-1 control	17.00 ± 0.45	0.96 ± 0.24	0.35 ± 0.09
THP-1 + LPS	13.93 ± 0.50	2.79 ± 0.47	1.01 ± 0.17
HAP-1 WT	15.53 ± 0.62	2.76 ± 0.35	1 ± 0.13
HAP-1 TF KO	18.64 ± 1.19	0.24 ± 0.07	0.09 ± 0.03

F3 (TF) mRNA levels were quantified using real-time quantitative polymerase chain reaction and normalized to 18S messenger RNA levels; data are shown as an input-normalized level (ΔCT). TF activity was measured using a one-stage coagulation assay and normalized to total protein; data are shown as TF activity fold change relative to HAP-1 WT. All data are presented as the mean ± SD.

KO, knockout; LPS, lipopolysaccharide; TF, tissue factor; WT, wild-type.

3.2. Effect of different blocking buffers and detection methods on the measurement of human TF by Western blotting

We have previously reported on a method to detect TF by Western blotting in high TF-expressing cells [22]. Interestingly, we found that this method detected a high level of TF (40 μg of BxPC-3 lysate) but failed to detect a low level of TF (1 μg of BxPC-3 lysate; Figure 1A). We used 2 loading controls (β-actin and α-actinin). The advantage of using α-actinin is that it has a higher molecular weight (100 kDa) compared with TF, whereas β-actin has a similar molecular weight (42 kDa). Because the detection method can affect sensitivity [23], we changed the secondary antibody to allow for chemiluminescent-based detection. We found that this change led to the detection of a low level of TF (1 μg of BxPC-3 lysate; Figure 1B). We then also modified the blocking protocol to utilize 5% nonfat dry milk suspended in TBS-Tween rather than the Pierce Protein-Free Blocking Buffer. This change further increased the sensitivity of TF detection (Figure 1C).

Figure 1.

Effects of different blocking buffers and detection methods on the measurement of tissue factor (TF) antigen by Western blotting. BxPC-3 cell lysates (40 μg and 1 μg) were run on separate gels, and proteins were transferred to membranes. TF was detected using the Novus antibody (catalog [cat] number NBP2-15139) diluted 1:1000 in blocking buffer (A and B) or 5% bovine serum albumin in Tris-buffered saline (TBS)-Tween (C). (A) The blot was blocked in Pierce Protein-Free Blocking Buffer (Thermo Fisher Scientific, cat number 37572), and anti-TF antibody binding was detected using a fluorescent goat anti-rabbit secondary antibody (LI-COR, cat number 925-68071) diluted 1:10,000 in blocking buffer. β-Actin and α-actinin were used as loading controls. β-Actin was detected using a mouse anti–β-actin antibody (Cell Signaling Technologies, cat number 3700) diluted 1:1000 in blocking buffer. Anti–β-actin antibody binding was detected using a fluorescent goat anti-mouse Alexa Fluor 790 antibody (Thermo Fisher Scientific, cat number A11357) diluted 1:10,000 in blocking buffer. α-Actinin was detected using a rabbit anti–α-actinin antibody (Cell Signaling Technologies, cat number 3134) diluted 1:1000 in blocking buffer. Anti–α-actinin antibody binding was detected using a fluorescent donkey anti-rabbit DyLight 800 antibody (ImmunoReagents, cat number DkxRb-003-F800NHSX) diluted 1:10,000 in blocking buffer. For imaging, the blot was exposed for 1 minute (TF), 24 seconds (β-actin), and 1 minute (α-actinin). (B) The blot was blocked using the Pierce Protein-Free Blocking Buffer, and anti-TF antibody binding was detected using an HRP-linked goat anti-rabbit secondary antibody (Thermo Fisher Scientific, cat number 31460) diluted 1:5000 in blocking buffer. We added Clarity Max Western Enhanced Chemiluminescence (ECL) Substrate (Bio-Rad, cat number 1705062) for 10 seconds prior to imaging. β-Actin diluted 1:5000 and α-actinin diluted 1:1000 were probed as described above. An HRP-linked horse anti-mouse HRP secondary antibody (Cell Signaling Technologies, cat number 7076) diluted 1:5000 in blocking buffer was used to detect anti–β-actin antibody binding. The binding of the rabbit anti–α-actinin antibody was detected using an HRP-linked goat anti-rabbit secondary antibody (Thermo Fisher Scientific, cat number 31460) diluted 1:5000 in blocking buffer. Clarity Western ECL Substrate was added for 10 seconds prior to imaging. For imaging, the blot was exposed for 5 seconds (TF), 10 seconds (β-actin), and 1 second (α-actinin). (C) The blot was blocked in 5% nonfat milk (Walmart, nonfat dry milk powder) in TBS-Tween buffer, and anti-TF antibody binding was detected using the same HRP-detection used in B. β-Actin and α-actinin were detected as described above in B, except the primary antibodies were diluted in 5% bovine serum albumin in TBS-Tween buffer. Clarity Western ECL Substrate was added for 10 seconds prior to imaging. For imaging, the blot was exposed for 5 seconds (TF), 10 seconds (β-actin), and 10 seconds (α-actinin). The blots were visualized using the iBright FL1000 imaging system (Thermo Fisher Scientific). kDa, kilodalton.

3.3. Effect of different secondary antibodies on the measurement of human TF by Western blotting

We evaluated the effect of different secondary antibodies on the detection of anti-TF antibody binding by Western blotting. We found that the intensity of the TF bands in the HAP-1 WT and recombinant TF samples was altered when different secondary antibodies were used with the same primary antibody (R&D Systems, cat number AF2339) and blotting conditions (Figure 2). We believe that the higher molecular weight species at ∼60 kDa observed with the donkey secondary antibody in the lane with recombinant TF was bovine serum albumin (BSA), which is used to stabilize the recombinant TF.

Figure 2.

Effect of different secondary antibodies on the detection of tissue factor (TF) by Western blotting using the same primary antibody. Cell lysates from HAP-1 wild-type (WT) and HAP-1 TF knockout (KO; both 40 μg) and recombinant TF (rTF; Innovin, 500 pg, Thermo Fisher Scientific, catalog [cat] number 10873566) were run on a gel, and proteins were transferred to a membrane. The blots were blocked with 5% nonfat milk (Walmart, nonfat dry milk powder) in Tris-buffered saline (TBS)-Tween. TF was detected using a goat anti-TF polyclonal antibody (R&D Systems, cat number AF2339) diluted 1:1000 in 5% bovine serum albumin in TBS-Tween. Anti-TF antibody binding was detected using either a mouse antigoat horseradish peroxidase (HRP)-linked secondary antibody (Santa Cruz, cat number sc-2354), a rabbit antigoat HRP-linked secondary antibody (Thermo Fisher Scientific, cat number 31402), or a donkey antigoat HRP-linked secondary antibody (Thermo Fisher Scientific, cat number A15999), all diluted 1:5000 in blocking buffer. The membranes were incubated with Clarity Max Western Enhanced Chemiluminescence Substrate (Bio-Rad, cat number 1705062) for 10 seconds. β-Actin was used as a loading control and was detected using a mouse anti–β-actin antibody (Santa Cruz, cat number sc-47778) diluted 1:5000 in 5% bovine serum albumin in TBS-Tween. An HRP-linked horse anti-mouse HRP antibody (Cell Signaling Technologies, cat number 7076) diluted 1:5000 in blocking buffer was used to detect anti–β-actin binding. Clarity Western Enhanced Chemiluminescence Substrate was added for 10 seconds prior to imaging. For imaging, the blot was exposed for 5 seconds (TF) and 10 seconds (β-actin). The blots were visualized using the iBright FL1000 imaging system (Thermo Fisher Scientific). kDa, kilodalton.

3.4. Detection of TF in low TF-expressing cell lines

We compared the ability of 4 commercially available anti-human TF antibodies to detect human TF in low-expressing cell lines: rabbit anti-human TF antibody NBP2-15139 (Novus Biologicals), goat anti-human TF antibody AF2339 (R&D Systems), rabbit anti-human TF antibody ab252918 (clone EPR22548-240, designated Abcam 1), and rabbit anti-human TF antibody ab228968 (clone EPR22548-232, designated Abcam 2; Figure 3). The Novus and R&D antibodies are both polyclonal antibodies that were affinity-purified using TF, whereas the Abcam antibodies are monoclonal antibodies that were protein A affinity-purified. All of the antibodies are suitable for Western blotting according to their product listings.

Figure 3.

Detection of tissue factor (TF) in different cell lines using 4 different anti-TF antibodies. Cell lysates (BxPC-3 [1 μg] and THP-1 control; THP-1 + lipopolysaccharide [LPS]; HAP-1 wild-type [WT]; and HAP-1 TF knockout [KO; all 40 μg]) were run on separate gels and transferred to membranes. TF was detected using 1 of 4 antibodies: (A) Novus NBP2-15139 (rabbit polyclonal), (B) R&D AF2339 (goat polyclonal), and (C) Abcam ab252918 (lot number 1096425-7, designated Abcam 1, rabbit monoclonal) and (D) Abcam ab228968 (designated Abcam 2) diluted 1:1000 in 5% bovine serum albumin in Tris-buffered saline (TBS)-Tween. The blots were blocked with 5% nonfat milk (Walmart, nonfat dry milk powder) in TBS-Tween. The binding of the rabbit anti-TF antibodies was detected using an HRP-linked goat anti-rabbit secondary antibody (Thermo Fisher Scientific, cat number 31460), and the binding of the goat anti-TF antibody was detected using an HRP-linked rabbit antigoat secondary antibody (Thermo Fisher Scientific, cat number 31402). Both secondary antibodies were diluted 1:5000 in 5% milk in TBS-Tween. The membranes were incubated with Clarity Max Western Enhanced Chemiluminescence Substrate (Bio-Rad, cat number 1705062) for 10 seconds. α-Actinin was used as a loading control and detected using a rabbit anti–α-actinin antibody (Cell Signaling Technologies, cat number 3134) diluted 1:1000 in 5% bovine serum albumin in TBS-Tween. The binding of the rabbit anti–α-actinin antibody was detected using an HRP-linked goat anti-rabbit secondary (Thermo Fisher Scientific, cat number 31460) diluted 1:5000 in blocking buffer. Clarity Western Enhanced Chemiluminescence Substrate was added for 10 seconds prior to imaging. For imaging, the blots were exposed for 10 seconds (A and B) or 2 seconds for TF detection (C and D) . All blots were exposed for 10 seconds for α-actinin detection. The blots were visualized using the iBright FL1000 imaging system (Thermo Fisher Scientific). For α-actinin, the brightness of the entire blot was adjusted. kDa, kilodalton.

The Novus antibody detected TF in 1 μg of total protein from the high TF-expressing cell line BxPC-3 using the new blocking and detection methods (Figure 3A). This antibody detected a band consistent with the molecular weight of glycosylated TF in the HAP-1 WT sample but not the HAP-1 KO sample (Figure 3A). In the THP-1 samples, the Novus antibody detected 2 bands at ∼30 and ∼45 kDa. The ∼45 kDa band and TF have similar molecular weights, preventing visualization of a distinct TF band (∼50 kDa), even in the LPS-stimulated sample (Figure 3A). Following deglycosylation, however, the TF band became apparent (Figure 4A); the ∼45 kDa band decreased in intensity, and a new band at ∼37 kDa appeared following deglycosylation of protein lysates from THP-1 cells stimulated with LPS. The molecular weight of this new ∼37 kDa band is consistent with the molecular weight of deglycosylated TF.

Figure 4.

Effect of deglycosylation on lysates from BxPC-3, HAP-1 wild-type (WT), HAP-1 tissue factor (TF) knockout (KO) cells, and lipopolysaccharide (LPS)-stimulated THP-1 cells. Cell lysates (BxPC-3 [1 μg]; HAP-1 WT and HAP-1 TF KO; and LPS-stimulated THP-1 cells [all 40 μg]) were treated with no enzyme or PNGase F. The lysates from the LPS-stimulated THP-1 cells were run on a separate gel from the other lysates. After running the gels, the proteins were transferred to membranes. The blots were blocked with 5% nonfat milk (Walmart, nonfat dry milk powder) in Tris-buffered saline (TBS)-Tween. TF (glycosylated TF [TF] and deglycosylated TF [dg-TF]) was detected using 1 of 3 antibodies: (A) Novus NBP2-15139, (B) R&D AF2339, and (C) Abcam ab252918 (lot number GR3270793-2, designated Abcam 1) diluted 1:1000 in 5% bovine serum albumin in TBS-Tween. The binding of the rabbit anti-TF antibodies was detected using an HRP-linked goat anti-rabbit secondary antibody (Thermo Fisher Scientific, cat number 31460), and the binding of the goat anti-TF antibody was detected using an HRP-linked rabbit antigoat secondary antibody (Thermo Fisher Scientific, cat number 31402). Both secondary antibodies were diluted 1:5000 in 5% milk in TBS-Tween. The membranes were incubated with Clarity Max Western Enhanced Chemiluminescence Substrate (Bio-Rad, cat number 1705062) for 10 seconds. α-Actinin was used as a loading control and detected using a rabbit anti–α-actinin antibody (Cell Signaling Technologies, cat number 3134) diluted 1:1000 in 5% bovine serum albumin in TBS-Tween. The binding of the rabbit anti–α-actinin antibody was detected using an HRP-linked goat antir-abbit secondary antibody (Thermo Fisher Scientific, cat number 31460) diluted 1:5000 in blocking buffer. The membranes were incubated with Clarity Western Enhanced Chemiluminescence Substrate for 10 seconds. For imaging, the blots were exposed for 2 seconds for TF detection and 10 seconds for α-actinin detection. The blots were visualized using the iBright FL1000 imaging system (Thermo Fisher Scientific). For α-actinin, the brightness of the entire blot was adjusted. kDa, kilodalton; NS, non-specifc.

The R&D and both Abcam 1 and 2 antibodies detected TF in both the 1 μg of total protein from BxPC-3 and the lower TF-expressing cells, THP-1 and HAP-1 WT, using the new blocking and detection methods (Figure 3B–D). There was an increase in TF expression in the sample from THP-1 cells stimulated with LPS compared with unstimulated cells with all 3 antibodies, and there was no band detected in the HAP-1 TF KO sample (Figure 3B–D). These results are consistent with the mRNA and activity data of these cells showing increased TF protein expression in LPS-stimulated THP-1 cells compared with unstimulated cells, as expected (Figure 1). The Abcam 1 antibody produced a slightly more intense band compared with the R&D and Abcam 2 antibodies (Figure 3B–D). Additionally, we observed a ∼127 kDa non-specific band that was consistent with the total protein input in all samples using the Abcam 1 antibody (data not shown). To simplify our study, we proceeded with further validation of the Novus, R&D, and Abcam 1 antibodies. The loading control was also changed from β-actin to α-actinin because we could not completely remove the Abcam 1 anti-TF antibody from the blot by stripping, resulting in carryover signal on the β-actin blot. Notably, the molecular weight of β-actin (42 kDa) is similar to that of TF (∼50 kDa), whereas α-actinin is larger (100 kDa).

3.5. Validating different anti-TF antibodies using deglycosylation of cell lysates

In addition to using genetic strategies (ie, KO), deglycosylation can be used to further validate detected bands as TF. Therefore, we performed deglycosylation on the cell lysates of our various cell lines and examined how this affected the TF band detected using the Novus, R&D, and Abcam 1 antibodies.

As shown in Figure 4, all 3 antibodies detected both the glycosylated and deglycosylated forms of TF in 1 μg of total protein from the high-expressing line BxPC-3. However, the 3 antibodies differed greatly in their ability to detect TF in the low-expressing line HAP-1 WT. The faint band at ∼50 kDa detected by the Novus antibody in the HAP-1 WT sample disappeared after deglycosylation. However, deglycosylated TF was difficult to observe due to the presence of a nonspecific band at ∼30 kDa, which had a similar molecular weight to deglycosylated TF (∼37 kDa; Figure 4A). Using this antibody, we also observed a reduction in the intensity of the ∼45 kDa band from LPS-stimulated THP-1 cells following deglycosylation. We believe the ∼45 kDa band is a combination of TF (∼50 kDa) and a nonspecific band (∼45 kDa; Figure 4A). The R&D antibody showed good specificity, having a band in both the control and deglycosylated HAP-1 WT samples and no band in the HAP-1 TF KO samples (Figure 4B). The blot probed with the R&D antibody also exhibited a downward shift in the band observed in the deglycosylated sample from LPS-stimulated THP-1 cells (Figure 4B). The Abcam 1 antibody also showed good specificity with no band in the HAP-1 TF KO samples (Figure 4C). There was also a downward shift in the bands observed in the samples from the HAP-1 WT and LPS-stimulated THP-1 cells following deglycosylation using the Abcam 1 antibody (Figure 4C). In conclusion, the Abcam 1 antibody demonstrated the best sensitivity among the 3 antibodies tested (Figure 4C).

3.6. Detection of TF in LPS-stimulated human PBMCs by Western blot

Because the Abcam 1 antibody had a higher sensitivity compared with the Novus and R&D antibodies, we used it for the detection of TF in LPS-stimulated PBMCs (Figure 5). There was no detectable TF in the unstimulated PBMCs. Following 5 hours of LPS stimulation, however, there was a prominent TF band in the PBMC samples (Figure 5). We observed a range of responses to LPS using the PBMCs from the different donors (Figure 5), which has been reported previously [24].

Figure 5.

Detection of tissue factor (TF) in unstimulated and lipopolysaccharide (LPS)-stimulated human peripheral blood mononuclear cells (PBMCs). PBMCs from 4 donors were unstimulated or stimulated with LPS for 5 hours. Lysates from PBMCs (40 μg) were run on a gel, and proteins were transferred to a membrane. The blot was blocked in 5% nonfat milk in Tris-buffered saline (TBS)-Tween. TF was detected using the Abcam anti-human TF antibody ab252918 (lot number 1096425-7, designated Abcam 1) diluted 1:1000 in 5% bovine serum albumin in TBS-Tween. Antibody binding was detected using an HRP-linked goat anti-rabbit secondary antibody (Thermo Fisher Scientific, cat number 31460) diluted 1:5000 in blocking buffer. The membrane was incubated with Clarity Max Western Enhanced Chemiluminescence Substrate (Bio-Rad, cat number 1705062) for 10 seconds. α-Actinin was used as a loading control and detected using a rabbit anti–α-actinin antibody (Cell Signaling Technologies, cat number 3134) diluted 1:1000 in 5% bovine serum albumin in TBS-Tween. The binding of the rabbit anti–α-actinin antibody was detected using an HRP-linked goat anti-rabbit secondary (Thermo Fisher Scientific, cat number 31460) diluted 1:5000 in blocking buffer. The membrane was incubated with Clarity Western Enhanced Chemiluminescence Substrate for 10 seconds. For imaging, the blot was exposed for 2 seconds for TF detection and 10 seconds for α-actinin detection. The blot was visualized using the iBright FL1000 imaging system (Thermo Fisher Scientific). For α-actinin, the brightness of the entire blot was adjusted. kDa, kilodalton.

4. Discussion

In this study, we examined the effects of varying several different parameters on the detection of TF in low-expressing cell lines, including the blocking buffer, the detection method, and the primary and secondary antibodies. First, we modified the blocking and detection methods of our previously published method, which detected TF in high TF-expressing cell lines [22]. We found that a high level of TF (40 μg of BxPC-3 cell lysate), but not a low level of TF (1 μg of BxPC-3 cell lysate), was detected using the previously published protocol with the Novus antibody. We increased the sensitivity of TF detection using the Novus antibody by changing the blocking buffer (from Pierce Protein-Free Blocking Buffer to 5% nonfat milk in TBS-Tween) and the detection method (from fluorescent to chemiluminescent). It is well established that fluorescence-based detection is less sensitive than chemiluminescent-based detection, with chemiluminescent-based methods offering up to femtogram-level sensitivity [21,23]. Indeed, using these new conditions, we were able to detect TF in 1 μg of total protein from BxPC-3 cells with the Novus antibody. We also found that using different secondary antibodies with the same primary antibody (R&D Systems) affected the sensitivity of TF detection.

As expected, we found that the primary antibody was a major contributor to the detection of TF. Our data demonstrated that the Abcam 1 antibody had increased sensitivity for the detection of TF compared with the Novus and R&D polyclonal antibodies. Both the R&D and Abcam 1 antibodies also showed better specificity than the Novus antibody.

Investigators commonly use Western blotting to measure the level of TF in cell lysates and tissue samples. Many different commercial and noncommercial monoclonal and polyclonal anti-TF antibodies have been used in the literature for the detection of TF by Western blotting. Some of the different anti-TF antibodies used in these studies are highlighted in Table 3 [[25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43]]. By searching PubMed using the keywords “tissue factor” and “Western blot,” as well as searching company websites, we identified 126 publications that performed Western blotting using samples derived from tissues or cell lysates. Of those, only 19 papers provided sufficient methodological details (eg, primary antibody and dilution; secondary antibody and dilution) to replicate the method. Additionally, when reporting the results, many articles show tightly cropped bands, which limits the ability to assess blots for the presence of nonspecific bands. Journals, such as Research and Practice in Thrombosis and Haemostasis, require at least several bandwidths above and below the band of interest and the identification of molecular weight markers within the cropped blot. Other journals, such as the Journal of Clinical Investigation, require the submission of uncropped blots with boxes indicating where bands in the final figure originated.

Table 3.

Commercially available anti-human tissue factor antibodies used in the literature.

Primary antibody	Primary clonality	Sample type
Goat anti-human TF AF2339 (R&D Systems)	Polyclonal	Bronchial epithelial cells [25,26]
		Kidney epithelial cell [27]
		Acute promyelocytic leukemia cell line NB4 [28]
Goat anti-TF 4501 (American Diagnostica)	Polyclonal	Vascular smooth muscle cells [29]
Mouse anti-human TF H00002152-B01P (Abnova)	Polyclonal	Pericytes [30]
Rabbit anti-human TF 47769 (Cell Signaling)	Polyclonal	Pancreatic cancer cell lines [22]
Rabbit anti-human TF NBP2-15139 (Novus)	Polyclonal	Pancreatic cancer cell lines [22]
Mouse anti-human TF clone HTF-1a,b	Monoclonal	Endothelial cells [31]
		Pancreatic cancer cell lines [22,32]
		Squamous cell carcinoma cell line [33]
Mouse anti-human TF 4509, clone IIID8c (American Diagnostica)	Monoclonal	Adherent monocytes [34]
Mouse anti-human TF 010-5059, clone TF9-10H10 (Bio-Rad)d	Monoclonal	Pancreatic cancer cell lines [22]
Mouse anti-human TF sc-374441, clone H-9 (Santa Cruz Biotechnology)	Monoclonal	Breast and colorectal cancer cell lines [35]
		Endothelial cells [36]
Rabbit anti-human TF ab252918, clone EPR22548-240 (Abcam)e	Monoclonal	Myeloid leukemia cell line HAP-1 [37]
Rabbit anti-human TF ab228968, clone EPR22548-232 (Abcam)f	Monoclonal	Platelets and endothelial cells [38,39]
		Liver cancer cell line HepG2 [40]

TF, tissue factor.

^aOriginally produced by Carson et al. [41].

^bMultiple commercial sources, including BD Biosciences (catalog number 550252) and Thermo Fisher Scientific (catalog number 16-1429-82).

^cOriginally produced by Magdolen et al. [42].

^dOriginally produced by Morrissey et al. [43].

^eDesignated as Abcam 1 in the current study.

^fDesignated as Abcam 2 in the current study.

Many researchers often assume that companies validate the antibodies they sell, but this is not always the case. For example, the Abcam rabbit monoclonal anti-human TF antibody ab151748 (clone EPR8986) was sold for many years and was used in 29 papers. Of the 29 papers, 15 used the antibody to detect what they thought was TF by Western blotting [22]. In a validation experiment performed by Abcam, however, the EPR8986 antibody detected a band in samples from HAP-1 TF KO cells. This led to the antibody being withdrawn from the market and raises significant concerns about the TF data presented in the 29 papers.

Currently, Abcam lists several anti-human TF antibodies on theirwebsite. We determined that the 2 rabbit anti-human TF monoclonal antibody clones (“EPR” number) used in this study are listed under different cat numbers (“ab” number) based on their formulation, including or excluding BSA/sodium azide. The antibodies were validated by the company using a variety of cell lines, including HAP-1 WT and HAP-1 TF KO cells. Interestingly, there are marked similarities between the blots used to validate the 2 different antibody clones, which is a concern.

The antibody designated Abcam 2 in this study has been reported to detect a 33 kDa band in platelets [38,39]. However, both we and another group found the Abcam 2 antibody detected a ∼50 kDa band [40]. At present, it is unclear if the 33 kDa band observed in platelets is TF. Likewise, the Novus antibody used in this study detects TF in BxPC-3 cells. However, nonspecific bands at molecular weights similar to glycosylated and nonglycosylated TF interfered with validating whether the Novus antibody detects TF in THP-1 cells. Thus, while an antibody may appropriately detect its target in some cell types, this is not necessarily true for all cell types and requires further validation for each cell type.

In summary, we validated the R&D (AF2339) and Abcam (ab252918; Abcam 1) antibodies for the detection of low levels of human TF in cells by Western blotting using a combination of TF WT and KO cells and deglycosylation. We do not recommend using the Novus antibody for detecting low levels of human TF because it detects nonspecific bands in HAP-1 TF KO cells and THP-1 cells. We also show that the Abcam 1 antibody has increased sensitivity for TF detection by Western blotting compared with the R&D antibody. All the steps of a Western blotting protocol should be considered during protocol design and development. Additionally, appropriate positive and negative controls should be utilized during the validation of an antibody.