Quantitative Analysis of NF-κB Transactivation Specificity Using a Yeast-Based Functional Assay

Abstract

The NF-κB transcription factor family plays a central role in innate immunity and inflammation processes and is frequently dysregulated in cancer. We developed an NF-κB functional assay in yeast to investigate the following issues: transactivation specificity of NF-κB proteins acting as homodimers or heterodimers; correlation between transactivation capacity and in vitro DNA binding measurements; impact of co-expressed interacting proteins or of small molecule inhibitors on NF-κB-dependent transactivation. Full-length p65 and p50 cDNAs were cloned into centromeric expression vectors under inducible GAL1 promoter in order to vary their expression levels. Since p50 lacks a transactivation domain (TAD), a chimeric construct containing the TAD derived from p65 was also generated (p50TAD) to address its binding and transactivation potential. The p50TAD and p65 had distinct transactivation specificities towards seventeen different κB response elements (κB-REs) where single nucleotide changes could greatly impact transactivation. For four κB-REs, results in yeast were predictive of transactivation potential measured in the human MCF7 cell lines treated with the NF-κB activator TNFα. Transactivation results in yeast correlated only partially with in vitro measured DNA binding affinities, suggesting that features other than strength of interaction with naked DNA affect transactivation, although factors such as chromatin context are kept constant in our isogenic yeast assay. The small molecules BAY11-7082 and ethyl-pyruvate as well as expressed IkBα protein acted as NF-κB inhibitors in yeast, more strongly towards p65. Thus, the yeast-based system can recapitulate NF-κB features found in human cells, thereby providing opportunities to address various NF-κB functions, interactions and chemical modulators.

Fig 1. Relative transactivation potential of p50 and p65 homodimers towards a panel of κB Response Elements.

A) κB target site preferences of p50TAD and p65 homodimers. Luciferase assays were performed to quantify relative transactivation capacity of p50TAD and p65 homodimers towards 17 different κB-REs. Reporter strains were grown in selective media containing 0.032% galactose for 16 hours reaching near stationary phase. For each isogenic reporter strain, the luciferase activity is calculated as fold-induction with respect to the values obtained with empty vector transformants. The average normalized activity and the standard error of four biological repeats are presented. κB-REs are ranked based on increasing transactivation potential with p50TAD. The same rank is used to plot the results obtained with p65 (lower panel). To the right are presented κB-RE sequences that are selectively responsive to either p50TAD or p65 (see text for sequence match to optimized consensus for p50 or p65). B, C) Web logo representations of the groups of κB-REs that were active or inactive with p50TAD and p65, respectively. D) Western blots presenting the relative expression of p50TAD and p65 proteins at different amounts of galactose. Yeast cells transformed with the GAL1-based expression vectors for NF-κB proteins were cultured for 16 hours at the indicated concentrations of galactose. An antibody directed against the p65 transactivation domain, which is also present in the p50TAD construct, was used for immunodetection. PGK1 endogenous protein provides a loading control.

Fig 2. Additive or weak cooperative transactivation between adjacent κB REs.

A) Yeast-based luciferase assays were performed at moderate to high expression levels of p50TAD or p65. Results were normalized and plotted as in Fig 1. The impact of tandem duplication (2d) of the decameric (1d) κB-RE or the indicated combination of two different κB-REs was evaluated. The REs were chosen based on the results from Fig 1 to include sequences exhibiting different levels of transactivation potentials. B) RE1 and RE6 were inactive as decameric κB-REs and there was no transactivation for two decamers in tandem or high expression of NF-κB proteins (up to 1% galactose). C) Functional interactions between two different κB-REs derived from the MCP-1 promoter. The M1 and M2 decameric κB-REs exhibited different responsiveness to p50TAD and p65, when examined separately, but are both derived from the MCP-1 promoter and they are located in close distance (19 nucleotide spacer) in the natural context. The combined responsiveness of the M1 and M2 κB-REs was examined, taking into account the impact of the distance between the two decamers and orientation relative to the transcriptional start site. Tandem repeats of M1 or M2 were included as controls. Yeast reporter strains, transformed with the indicated expression vector were cultured for 16 hrs with the indicated low amount of galactose. Relative activity refers to the average light units normalized for cell number (measured by optical density at 600nm). Average and standard error of four biological repeats are presented. The NF-κB-independent reporter activity (empty vector) is also presented as reference. Interestingly, the M2+M1 strains exhibited higher basal level of reporter expression. The M1+M2 sp strain contains the M1 and M2 κB decamers separated by 18 nt as in the human gene (see Table A in S1 File).

Fig 3. Functional interactions between p50 and p65 towards a panel of κB-REs.

A) and B) Eight different κB-REs, each comprising two adjacent copies of the decameric κB sequences were examined for the transactivation potentials of p50TAD, p65 as well as upon co-expression of both proteins. Cells were grown in lower (0.008%, panel A) and high (0.064%, panel B) levels of galactose. C) A non-chimeric p50 construct lacking the TAD domain was also studied at the higher galactose level for all REs, except RelBCons. In all panels, luciferase assays were conducted and results plotted as described in Fig 1.

Fig 4. Functional evaluation of four selected κB-REs in MCF7 cells.

A-B) MCF7 cells were transiently transfected with four pGL4.26 derived vectors containing different κB-REs along with a control pGL4.26 empty vector. Twenty-four hours after transfection cells were treated for 4 or 8 hours with TNFα (10ng/ml for panel A or 50ng/ml for panel B) alone or in combination with BAY11-7082 (20μM for 8 hours, only for panel B). Presented are the average fold-induction relative to the empty pGL4.26 vector and the standard deviation of at least three independent biological replicates. C) Western blot analysis showing p50 and p65 protein levels from nuclear-cytoplasmic fractions after the indicated treatments at the following doses: TNFα (50ng/ml) and BAY11-7082 (20μM). GAPDH and Histone 3 were used as reference controls for the cytoplasmic and nuclear fraction respectively.

Fig 5. IκBα inhibits p65-dependent transactivation in yeast.

The highly responsive M2 strain was used to test the impact of co-expressing IκBα with the NF-κB proteins p50TAD or p65. A) Luciferase assays results were obtained and plotted as described in Fig 1. Control transformants lacking the IκBα expression construct were obtained using the pRS315 empty vector. Cells were cultured in 0.032% galactose for 16 hours to achieve moderate expression of p50TAD or p65. IκBα is expressed under the constitutive ADH1 promoter. For all conditions the light units were normalize for the optical density of the cultures. The relative luciferase activity, obtained with cells transformed with empty vectors was set to 1 and used to obtained the fold of reporter induction due to the expression of NF-κB proteins. Bars plot the average and standard errors of four biological replicates. B) A western blot image revealing the impact of IκBα on p50TAD or p65 protein levels. Transformants with two p50TAD expression vectors that differ for the selection marker gene (LEU2 for p50TAD #1 and TRP1 for p50TAD #2) were included. PGK1 was used as loading control.

Fig 6. Effect of the small molecules BAY11-7082 and ethyl pyruvate on NF-κB activity in yeast.

The strains containing the REs M2, RE4 and RelBCons were grown overnight (16 hours) in selective media containing low levels of galactose (0.008%) with or without the addition of different concentrations of BAY11-7082 (10μM, 20μM) and EP (1.5mM, 2.5mM, 5mM, 10mM). A-C) Average luciferase assays and standard errors of four biological repeats are presented. Results obtained with cells transformed with an empty expression vector are included to take into account the impact of the small molecules on the NF-κB-independent, basal expression of the reporter. D, E) Western blot images showing p50TAD, p50TAD + p65 and p65 protein levels in BAY and EP treated and untreated cells. PGK1 was used as a loading control.

Fig 7. Comparison between predicted DNA binding affinity and yeast-based transactivation.

A), B) The relative binding affinities of p50 or p65 towards 17 κB-REs were obtained from (http://thebrain.bwh.harvard.edu/nfkb) and compared with the relative transactivation potential measured in yeast at moderate levels of galactose induction. REs are ordered from left to right based on increasing Z-score for DNA binding affinity. The highest affinity RE (RE5 for p50 and I1 for p65) is set to 100. REs with Z-score lower than 4, considered equivalent to background, are labeled by *. C) Similarly, Z-scores of the p50-p65 heterodimers and transactivation potentials are compared for the 6 κB-REs that were tested in the co-expression experiments in yeast.