The dark side of EGFP: defective polyubiquitination

Abstract

Enhanced Green Fluorescent Protein (EGFP) is the most commonly used live cell reporter despite a number of conflicting reports that it can affect cell physiology. Thus far, the precise mechanism of GFP-associated defects remained unclear. Here we demonstrate that EGFP and EGFP fusion proteins inhibit polyubiquitination, a posttranslational modification that controls a wide variety of cellular processes, like activation of kinase signalling or protein degradation by the proteasome. As a consequence, the NF-kappaB and JNK signalling pathways are less responsive to activation, and the stability of the p53 tumour suppressor is enhanced in cell lines and in vivo. In view of the emerging role of polyubiquitination in the regulation of numerous cellular processes, the use of EGFP as a live cell reporter should be carefully considered.

Figure 1. EGFP inhibits NF-κB and JNK activation in 293T cells

(A) Activation of an NF-κB luciferase reporter by a Flag-tagged API2-MALT1 fusion protein is inhibited by EGFP in a dose dependent manner. (B) EGFP does not block NF-κB dependent luciferase activity induced by expression of the p50/p65 subunits of NF-κB (C) Activation of an NF-κB luciferase reporter by Flag-API2-MALT1 is inhibited by EGFP mutated for the TRAF6 binding motif (D) EGFP but not pmaxGFP prevents TNF-α induced activation of the NF-κB luciferase reporter and phosphorylation of IκB-α and JUN in 293T cells. EGFP does not increase HSP70 levels or induce HSP70B′ in 293T cells. Fluorescence intensities (Ex485/Em520nm) for EGFP and pmaxGFP were both ∼100 fold above background. (E) N- and/or C-terminal EGFP fusion proteins (for API2, CHIC2, NXF5a, NXF1, MALT1, Actin, Rab5, Syndecan binding protein 2 (SDCBP2) or β-Tubulin) and EGFP with a nuclear localization signal (NLS) or a farnysilation site (pEGFP-F) reduce TNF-α-induced NF-κB luciferase reporter activity in 293T cells. Bottom: In humans, HSP70 is constitutively expressed under normal conditions, but Hsp70B′ is only induced in response to stress. There is no basal expression of Hsp70B′. As a positive controle for HSP70B′ expression, 293T cells (lane 2) were heat shocked at 44°C for two hours (lane 3) and allowed to recover at 37°C for 5 (lane 4) or 18 (lane 5) hours before harvest. NF-κB-dependent luciferase activity is represented for each experiment as fold induction of vector transfected cells and is represented graphically as the mean and standard deviation of at least three independent experiments. All molecular weight standards are in kDa.

Figure 2. EGFP blocks Lys63- and Lys48-linked polyubiquitination.

(A) 293T cells transfected with the indicated constructs and treated for 4 hours with 20 ng/ml TNF-α (lane 5 and 6) were immunoblotted with anti-Flag (API2-MALT1), anti-HA (HA-Ub-K63) and anti-EGFP antibodies (left panel) or anti-IKKγ immunoprecipitates were immunoblotted with anti-Flag (IKKγ) or anti-HA (ubiquitin) (right panel). (B) EGFP affects K48-linked polyubiquitination. 293T cells were transfected with a Ubiquitin construct with only Lys48 available for polymerization (HA-Ub-K48), treated for 4 hours with 20 ng/ml TNF-α or left untreated and cell lysates were immunoblotted with anti-HA (Ub-K48) and anti-EGFP antibodies. Fluorescence intensities (Excitation 485/Emission 520 nm) for EGFP and pmaxGFP were comparable (∼100 fold higher then background values), expression of pDs-Red was confirmed by Fluorescence microscopy. (C) EGFP stabilizes exogenous API2-Myc in 293T cells via reduction of its Lys48-linked auto-ubiquitination and proteasomal degradation. (D) Stable expression of EGFP in the merkel cell carcinoma cell line MCC14.2 reduces polyubiquitination and enhances endogenous p53 expression levels. The average ratio and standard deviation of p53 to actin signals are given (three independent experiments), (Ub)ⁿ : polyubiquitinated proteins.

Figure 3. EGFP affects polyubiquitination-dependent processes in vivo.

(A) Western blots of spleen and liver extracts from FVB wild-type (wt) and EGFP transgenic mice detected with antibodies against Ubiquitin, EGFP and actin (loading control). (B) EGFP reduces phosphorylation of IκB-α in anti-IgM/anti-CD40 stimulated B-lymphocytes purified from EGFP mice. Experiments performed in triplicate, a representative image of one experiment is shown. The average and standard deviations of ratios of IκB-α-P to actin relative to un-stimulated cells are given. (C) The deficit of B220⁺/CD40⁺ B-cells in the bone marrow of API2-MALT1 mice is restored in EGFP/API2-MALT1 double transgenic mice. Experiments performed in triplicate, the average and standard deviations are depicted. eMalt1: endogenous Malt1, *a-specific band (D) EGFP mice show increased p53 levels in the liver and have enhanced p53/p21 responses upon γ-irradiation induced DNA damage. The average and standard deviations of the ratios of p53/p21 to actin signals relative to that of sample 1 are given. (E) Recombinant EGFP (rEGFP) does not prevent polyubiquitination of a Biotin-Lysozyme substrate in vitro. (Ub)ⁿ: polyubiquitinated proteins.