An Investigation of the Specificity of Research Antibodies against NF-κB-subunit p65

The transcription factor nuclear factor kappa B (NF-κB) plays a pivotal role in a variety of biological processes in different mammalian tissues and organs, including the immune and central nervous systems (Hayden and Ghosh 2008). Many of the studies on the NF-κB signaling in different cellular systems were, at least partly, based on immunological analyses of the distribution of the NF-κB subunit p65. Recently, different p50 and p65 antibodies were analyzed and demonstrated that many of the tested antibodies were not fully specific for their antigen (Herkenham et al. 2011). Such improper epitope recognitions may lead to misinterpretations of the distribution, localization and activity of NF-κB. Further elucidating the specificity of commercially available antibodies against p65, we tested six different antibodies using immunocytochemistry (ICC) and western blots. Murine embryonic stem cells (mESCs) protein samples were used as a stringent negative control, as these cells are known to be p65-negative at the protein level (Lüningschrör et al. 2012), whereas protein from GFP-p65 knock-in mice (De Lorenzi et al. 2009) served as positive controls. We demonstrate inappropriate cross-reactivity of several commercially available p65 antibodies.

We treated murine embryonic fibroblast (MEFs) with the cytokine tumor necrosis factor alpha (TNF-α) to trigger NF-κB signaling for the detection of nuclear immunoreactivity of the p65 antibodies using ICC. In all approaches, such TNF-α–dependent nuclear signals were observable (Fig. 1A). Non-stimulated MEFs showed no expected cytosolic signals for NF-κB p65, except when using antibodies sc-8008, sc-372 and E498 (Fig. 1A). All antibodies are summarized in Table 1 and all data in Table 2. Using western blots from cell lysates, the observations from ICC could partly be confirmed (Fig. 1B). Interestingly, sc-7151 marked a single band at the size of p65 after having shown inappropriate cross-reactivity in ICC. On the contrary, E498 did not mark a band at all. Apparently, this antibody is highly specific for p65 but only in its native form and not after a denaturing SDS-PAGE; although, according to the manufacturer, this antibody is suitable for western blotting. It should be noted, however, that this antibody is no longer available for purchase. As a stringent negative control, we tested the antibodies on mESCs in western blots as well as in ICC. In western blots, all antibodies, except sc-372, did not mark a band (Fig. 1B) and also demonstrated no immunoreactivity in ICC. One representative staining using MAB3026 is shown in Figure 1C. The sc-372 antibody demonstrated strong cytosolic immunoreactivity in repeated approaches (Fig. 1C), confirming the result from the western blot, where it marked one single band at a size comparable with p65 (Fig. 1B).

Figure 1.

Antibodies against NF-κB p65 show cross-reactivity in mouse embryonic fibroblasts (MEFs) and mouse embryonic stem cells (mESCs). (A) Immunocytochemical staining of TNF-α–treated MEFs using six different commercially available antibodies against p65. Cells were cultivated on sterilized and etched coverslips in DMEM (Life Technologies) supplemented with 10% FCS (Life Technologies; Carlsbad, CA), penicillin/streptomycin (5 ml/50 mg; PAA Laboratories GmbH, Pasching, Austria), amphotericin B (5 ml/ 1.25 mg; PAA) and L-glutamine (200 mM; PAA) to a density of 1×10⁴ cells per coverslip. Before treatment with TNF-α, the cells were synchronized for 4 hr at 4C in DMEM containing 10% FCS and 15 mM HEPES buffer (PAA). Afterwards, the cells regenerated at 37C for 3 hr. Stimulation with TNF-α (10 ng/ml, Calbiochem; San Diego, CA) was performed for 30 min at 37C, followed by a washing step with phosphate-buffered saline (PBS) and subsequent fixation using phosphate-buffered 4% paraformaldehyde. After 3 further washing steps with PBS, the cells were permeabilized with PBS containing 0.02% Triton X-100. Incubation with the primary antibody lasted 1 hr at room temperature or was done overnight at 4C. Secondary fluorochrome-conjugated antibodies (Life Science, Darmstadt, Germany) were applied for 1 hr at room temperature. Sytox green (0.5 µM, Molecular Probes, Göttingen, Germany) was used for nuclear counterstaining. In TNF-α–treated MEFs (upper panel) all antibodies demonstrated primarily nuclear immunoreactivity. In control cells (unstimulated MEFs), only application of the sc-372, sc-8008 and E498 led to clear cytosolic signals. Immunoreactivity of the sc-109 was distributed throughout the whole cell, whereas sc-7151 and MAB3026 led to nuclear signals even in the untreated cells. (B) Western blot of whole cell lysates from MEFs and mESCs. mESCs were cultivated according to Lüningschrör et al. (2012). The protein concentration was determined via Nanodrop UV spectrophotometry. Protein (20 µg) was separated using 12% denaturing SDS-PAGE. Separated proteins were then transferred onto PVDF membranes with a semi-dry blotter. After a blocking step using phosphate-buffered 5% milk powder, membranes were incubated with p65 antibodies overnight at 4C. Secondary HRP-conjugated antibodies were applied for 1 hr at room temperature. Visualization of the results was achieved with enhanced chemiluminescence. All antibodies were tested on western blots from whole cell lysates of MEFs and (except the E498) mESCs. The sc-8008, sc-7151, sc-372, and MAB3026 antibodies all marked one single band on lysates from MEFs. The sc-109 antibody marked multiple bands, whereas the E498 antibody did not mark a band. Only the sc-372 antibody exhibits a band for mESCs lysates at a size comparable with p65. (C) Immunocytochemistry of mESCs as a negative control. ICC was performed as described above. Upper panel: In mESCs, ICC with the sc-372 antibody led to a strong cytosolic signal. The remaining p65 antibodies did not show antigen binding in mESCs. Lower panel: MAB3026 antibody was chosen as a representative example. Scale bar: 20 μm, arrows: ESC colonies, arrowheads: feeder layer of Mitomycin C-inactivated MEFs.

Table 1.

List of NF-κB p65 Antibodies Used in this Study.

Antibody (clone)	Species of origin	Epitope	Dilution (ICC)	Dilution (WB)	Manufacturer
sc-372	rabbit	polyclonal to C-terminus	1/50 – 1/100	1/500	Santa Cruz Biotechnology, Inc (Dallas, TX0
E498	rabbit	polyclonal to residues around Glu498	1/50 – 1/100	1/500	Cell Signaling Technology, (Beverly, MA)
sc-7151	rabbit	polyclonal to N-terminus (aa 1-286)	1/50 – 1/100	1/500	Santa Cruz Biotechnology, Inc (Dallas, TX0
MAB3026 (12H11)	mouse	monoclonal to region of NLS	1/50 – 1/100	1/500	Chemicon, (Temecula, CA)
sc-109	rabbit	polyclonal to N-terminus	1/50 – 1/100	1/500	Santa Cruz Biotechnology, Inc (Dallas, TX0
sc-8008 (F-6)	mouse	monoclonal to N-terminus (aa 1-286)	1/50 – 1/100	1/500	Santa Cruz Biotechnology, Inc (Dallas, TX0

Table 2.

Summary of All Observations Regarding p65 Antibody Specificity.

Antibody (clone)	sc-372	E498	sc-7151	MAB3026 (12H11)	sc-109	sc-8008 (F-6)
Western blot	+	–	+	o	–	+
MEFs				(Very large band at approx. p65)
Western blot	–	Not tested	+	+	+	+
mESCs
ICC	+	+	–	–	o	+
MEFs
ICC	–	Not tested	+	+	+	+
mESCs
ICC	+	Not tested	–	–	–	+
GFP-p65 MEFs

+, specific result; o, ambiguous result; –, non-specific result.

Additionally, we studied the potential co-localization of GFP-expression in a GFP-p65 knock-in mouse line. These mice express a GFP-p65 fusion protein from the endogenous p65 locus that functionally substitutes p65 (De Lorenzi et al. 2009). Therefore, MEFs derived from this mouse line—hereafter referred to as GFP-p65 MEFs—are highly suitable for performing co-localization studies to gain further insight into the specificity of the p65 antibodies.

For sc-8008 and MAB3026 antibodies, co-immunostaining was performed with ab290 anti-GFP antibody (Abcam; Cambridge, MA). All antibodies demonstrated nuclear co-localization with the GFP signals in TNF-α–treated cells. A representative immunostaining of TNF-α–treated cells using the sc-8008 antibody is shown in Figure 2A. In non-stimulated GFP-p65 MEFs, only signals from sc-372 and sc-8008 antibodies co-localized with the GFP signal (Fig. 2A, B). To summarize the presented data, only the sc-8008 antibody showed the expected immunoreactivity in all approaches of our test models. Interestingly, this is in contrast to the results by Herkenham et al. (2011), wherein presumed non-specificity of sc-8008 was indicated by the presentation of bands of variable sizes in western blots of different tissues from TNF receptor 1/p65 double-knockout mice (Herkenham et al. 2011). Additionally, they presumed that the sc-372 antibody was very specific for p65, whereas we demonstrate its inappropriate cross-reactivity in mESCs. Remarkably, the above presented contrary results regarding sc-372 and sc-8008 might be due to batch fluctuations. It is unlikely that the same batches were used in our analyses and in the 2011 study by Herkenham and colleagues.

Figure 2.

Immunocytochemistry of GFP-p65 murine embryonic fibroblasts (MEFs) for potential co-localization of GFP with anti-p65 antibodies. (A) Immunostaining using the ab290 anti-GFP antibody and the sc-8008. ICC was performed as described for wildtype MEFs in Figure 1. Nuclear counterstaining was performed using DAPI (0.5 µg/ml; AppliChem, Darmstadt, Germany). Both in unstimulated and TNF-α–treated cells, the signals showed complete co-localization. (B) Immunostaining of unstimulated GFP-p65 MEFs. Only immunostaining using the sc-372 led to a complete co-localization of the p65-driven GFP signal and p65 antibody. Immunoreactivity of MAB3026, sc-7151 and sc-109 antibodies only showed partial co-localization with the GFP signal.

In the case of MAB3026, we demonstrated non-specificity of binding, which was analogous to the findings of Herkenham et al (2011). With the western blot results being not completely unambiguous regarding MAB3026 (see Table 2), the ICC staining revealed non-specific nuclear immunoreactivity independent of TNF-α treatment. MAB3026, originally created in our laboratory, is intended to mark “active p65”, as its epitope includes the nuclear localization signal, which is only exposed when p65 is activated. Then named alpha-p65M, it passed stringent tests for validity, marking exclusively p65 in p65-overexpressing HEK 293 cells in western blot and immunostaining (Kaltschmidt et al. 1995). It was transferred to Boehringer-Mannheim as Clone 12H11, resold to Roche and finally bought by Chemicon, and it is now sold as MAB3026. Being generated in the early ‘90s, transferred from one company to the next, we presume that there may be a mutation in Clone 12H11. With the NLS being rather conserved, a mutation that changes the recognition of the epitope slightly might lead to the recognition of other NLS-containing proteins. This hypothesis would be in agreement with our observations of TNF-α–independent nuclear immunoreactivity.

The antibodies sc-109 and sc-7151 both demonstrated inappropriate cross-reactivity and usage of these antibodies, especially in ICC, is therefore not recommended.

Our results indicate that not all of the commonly used antibodies against p65 exclusively bind to p65, at least in the batches that we have tested. Even antibodies that mark specifically p65 in western blotting do not necessarily show specific immunoreactivity in ICC. Low amounts of p65 in cells require higher concentrations of the antibody, which increases the risk of non-specific binding. Therefore, the usage of these antibodies should be conducted with awareness of the limitations of each antibody, and great care should be taken to exclude false-positive results, which can lead to misinterpretations of the localization and activity of p65. To avoid this, rigorous testing of every new batch of antibody prior to its application is highly recommended.