Identification and functional characterization of novel binding site on TNF-a promoter
Xiaoren Tang, Matthew J. Fenton, and Salomon Amar
Department of Periodontology and Oral Biology, School of Dental Medicine, and The Pulmonar Center, School of Medicine, Boston University, Boston, MA 02118
Communicated by Susan E. Leeman, Boston University School of Medicine, Boston, MA, January 29, 2003 (received for review November 11, 2002)
Tumor necrosis factor (TNF)-a is a pleiotropic cytokine that is produced by monocytes and macrophages. The transcription of TNF-a is highly controlled by lipopolysaccharide (LPS), although the exact mechanisms are unknown. In human monocytes, transcription factor lipopolysaccharide-induced TNF-a factor (hLITAF) regulates the expression of TNF-a. Tang et. al (2002) uses DNase footprinting to identify a CTCCC region (-515 to –511) in the promoter of hTNF-a that is necessary for binding of hLITAF. The paper also identifies amino acids 165-180 of hLITAF as the region that specifically mediates binding with TNF-a. GST-hLITAF fusion proteins were used to determine the binding domain of hLITAF for TNF-a. The experiments revealed that the amino acids 165-180 are sufficient to induce expression of TNF-a when the CTCCC region of the promoter is present.
TNF-a, a proimmflamotory cytokine of the immune system, has effects on lipid metabolism, coagulation, insulin resistence, and endothelial function (OMIM, 2003). Because TNF-a can has both beneficial and pathological effects, its expression is highly regulated both during and after transcription. Previous studies have identified multiple transcription factors that play a role in TNF-a expression. NF-kB, Ets, NF-AT, activating protein 1, cAMP response element-binding protein, signal transducers and activators of transcription, and lipopolysaccharide-induced TNF-a factor are all transcriptions factors that bind to the promoter region of TNF-a, but the relative impact of each on the expression of TNF-a is unknown. LPS stimulates monocytes and macrophages by activating transcription factors that in turn activate TNF-a and other inflammatory proteins. LPS up-regulates certain transcription factors, but is also responsible for down-regulating other transcription factors. Previous experiments showed that LPS up-regulates transcription factor NF-kB which increases the expression of hTNF-a. When NF-kB is inhibited however, hTNF-a continues to be expressed, but at lower levels. This finding led the authors to identify transcription factor hLITAF as mediating the expression of hTNF-a. Previous experiments showed that when hLITAF was removed from cells, levels of hTNF-a were reduced.
In this paper, the authors aim to elucidate the role of hLITAF in hTNF-a expression. They use DNase footprinting and electrophoretic mobility-shift assay (EMSA) to determine the sites of interaction between hLITAF and TNF-a. The CTCCC region is identified in TNF-a as being necessary for binding of hLITAF, and amino acids 165-180 of hLITAF are sufficient to express hTNF-a.
In order to determine the sites of interaction between hLITAF and TNF-a, hLITAF and hTNF-a fragments with different deletions were generated by PCR. The hLITAF fragments were fused with GST so that the proteins could be purified with a monoclonal antibody. Figure 1 is a schematic representation of the hLITAF and hTNF-a constructs. Three different regions of the hLITAF protein were identified as A, B, and C by sequence analysis. The hLITAF constructs include a wildtype fusion protein (amino acids 1-228, containing peptides A, B, and C), amino acids (aa) 1-75, aa 1-151 (contains peptide A), aa 76-151, aa 76-228 (peptide B and C), aa 152-228 (peptides B and C), aa 152-228 with a deletion of 165-180 (contains peptide C), aa 152-228 with a deletion of 180-195 (contains peptide B).
The hTNF-a promoter constructs include the wildtype promoter (-991 to 1, contains CTCCC region), region –991 to 1 with a deletion of –515 to –511 (CTCCC region is deleted), region –550 to –487 with the TATA box (contains CTCCC region), and region –550 to –487 with the TATA box and a deletion of –515 to –511 (CTCCC region is deleted). In the schematic representation, boxes represent potential binding sites for the various transcription factors of TNF-a.
DNase footprinting was used to determine the binding site within the hTNF-a promoter of hLITAF. DNase footprinting can be used to detect a region of DNA that binds with a protein of interest. It is based on the principle that DNase will degrade all DNA that is not bound to a protein, thus the region of DNA that is bound to the protein is left intact. For this experiment, the hTNF-a promoter was designed with a HindIII site at the 3’ end so that only the 5’ end would be radioactively labeled with [32P]ATP; digestion of the hTNF-a DNA with HindIII removed the radioactive label from the 3’ end. The different hLITAF constructs were added individually to hTNF-a DNA and then the DNA was degraded by DNase I from the 3’ end for 5 minutes. The fragments were then gel electrophoresed and detected by autoradiography. The controls in this figure show that the TNF-a promoter alone was fully degraded by DNase. Three hLITAF constructs (wildtype (A, B, C), aa 76-228 (B, C), and aa 152-228(B, C)) revealed footprints; in all three cases, region –515 to –511 of hTNF-a showed protection from DNase. The control lanes and lanes that contained hLITAF sequences with no peptides or peptide A only were not protected from DNase. Part b shows the sequence of the TNF-a promoter from –550 to –487. The region of the promoter that was protected from DNase digestion (-515 to –511) was determined to be CTCCC.
Electrophoretic mobility-shift assay (EMSA) was used to further analyze the interaction between hTNF-a DNA and hLITAF. Radioactively labeled hTNF-a promoter was used as the probe for the hLITAF. Three different fusions proteins of hLITAF were used in the assay: amino acids 1-151 (first half of protein, contains peptide A), aa 152-228 (second half of fusion protein, contains peptide B and C) and aa 1-228 (wildtype LITAF). One control lane contained only the [32P]ATP-labeled hTNF-a promoter and another control lane contained the promoter with only the GST fusion protein. Each of the three hLITAF fusion proteins was added to the promoter and tested both with and without an unlabeled competitor. A band shift was observed in lanes 5 and 7, indicating that the fusion proteins with aa 152-228 and aa 1-228 were able to bind with the probe (only when the competitor was not present), thus causing the molecular weight of the probe to become heavier and shift upwards on the gel. The EMSA test in part A revealed that the second half of the fusion protein (aa 152-228) containing peptides B and C is sufficient for binding to the hTNF-a promoter.
To determine which portion of the hLITAF protein is responsible for binding to the hTNF-a promoter, they developed one construct that contained a deletion for peptide B and another construct that contained a deletion for peptide C (depicted in Figure 1). For the EMSA experiment in part B, two radioactively labeled probes were used: the -550 to –487 region of the hTNF-a promoter and the –550 to –487 region with a deletion from –515 to –511 (CTCCC binding site). Appropriate controls were applied and the fusion protein constructs were tested individually. A band shift was detected in the lanes containing fusion proteins aa 152-228 (peptides B and C), aa 1-228 (peptides A, B, and C), and aa 152-228 with a deletion from 180-195 (peptide B only). The most important part of this figure is that a band shift was observed for the fusion protein construct aa 152-228 with the deletion of aa 180-195 (construct contained peptide B only) but no band shift was observed for construct aa 152-228 with deletion of aa 165-180 (construct contained peptide C only). This data shows that hLITAF fails to bind to hTNF-a when peptide B is absent but is able to bind to hTNF-a when peptide B is present. The data also further supports the finding that the CTCCC region of hTNF-a is necessary for the binding of LITAF. All GST fusion protein constructs failed to bind to the hTNF-a promoter with the deleted CTCCC region.
The authors indicate that while data from Fig. 3 suggests that peptide B of hLITAF is sufficient for hTNF-a binding, it is not conclusive. Three peptides were synthesized (A, B, and C) that correspond to three separate regions in the amino acid sequence of LITAF. Peptides were individually added to THP-1 cells at four different concentrations by Chariot (a commercial kit for protein transduction). Hemagglutinin antigenic (HA) peptide was also individually introduced into cells and served as a control peptide for the experiment. After 24 hours, the cells were centrifuged and the TNF-a found in the supernatant was quantified by ELISA. ELISA uses specific antibodies to purify TNF-a and measure the amount of the protein present. At the higher concentrations of peptides (100 ng/ml and 1 ug/ml), the amount of hTNF-a protein was significantly higher in those cells that had the added peptide B. The levels of hTNF-a in cells with added peptide A and C remained relatively constant with the control HA cells. The data from this experiment show that peptide B is sufficient to increase the levels of hTNF-a protein, and therefore must be sufficient for binding to the promoter region and stimulating its transcription. Peptides A and C did not show data indicating they are sufficient for increasing expression of hTNF-a.
At this point, the authors wanted to determine if the CTCCC sequence in the hTNF-a promoter is necessary for the peptide B region of LITAF to increase expression levels. The different hTNF-a constructs described in Figure 1b were individually transfected into THP-1 cells. LPS, peptide A, B, C, or HA were then added to the cells and the activity of the hTNF-a promoter was measured by the luciferase assay. In Figure 5a, the activity of wildtype hTNF-a construct (-991 to 1) and the –991 to 1 construct with a deletion from –515 to –511 (CTCCC region) were analyzed. LPS (a positive control), stimulated both wildtype and mutant promoter constructs. The HA, A, and C peptides did not increase activity in either of the promoter constructs. When peptide B was added, a significant increase in activity (about twice that of HA) was noted for the wildtype promoter, but not the mutant promoter containing the deletion. This data indicates that peptide B is sufficient for stimulating hTNF-a when the CTCCC region is present but fails to do so when the CTCCC region is absent.
Figure 5b repeats the experiment in Figure 5a, but hTNF-a promoter constructs of –550 to –487 plus the TATA box and –550 to –487 with a deletion of –515 to –511 (deleted the CTCCC region) plus the TATA box. This time, LPS was able to stimulate the hTNF-a construct containing the CTCCC region, but failed to stimulate the construct that did not contain the CTCCC region. Similarly, peptide was sufficient to stimulate the hTNF-a construct that contained the CTCCC region, but failed to stimulate the construct without the CTCCC region. Peptide B resulted in 2.3 times more activity of the hTNF-a promoter with the CTCCC region present than the HA control. All other peptides (HA, A, and C) failed to stimulate either of the hTNF-a constructs. The data from Figure 5a and b show that peptide B (aa 165-180) of hLITAF is the binding region for hTNF-a and is sufficient for activating the hTNF-a promoter.
Overall, I think the claims made in this paper are supported by their data, the data is adequately displayed in the figures, and it is well written and logical. The results accurately support the authors’ claims; they have convinced me that the CTCC region of the hTNF-a promoter is necessary for binding and the peptide B region of hLITAF is sufficient for activating the promoter of hTNF-a. The authors clearly state their hypotheses and designed appropriate experiments to carry them out. DNase footprinting is a very useful method to determine the region of DNA to which a protein binds; EMSA is also an appropriate method for determining which region of the LITAF protein the DNA probe will bind to. All experiments were well controlled and the EMSA experiments produced data that clearly. The abstract of the paper was clearly and concisely written; it gave the reader an accurate picture of the information discussed in the paper. The introduction also was nicely written. It contained sufficient background information so that the reader can understand the implications of the current experiment and why the expression of TNF-a must be highly regulated. The results section is organized nicely; the authors present their reasoning behind each experiment and describe why the results from the previous experiment led them to the next experiment. In the discussion, the authors highlight the important aspects of their finding and again described the logic behind their experiments. At the end of the discussion, they state that their findings will help design treatments for TNF-related diseases. I was unaware that there were diseases related to TNF; it would have been interesting if they had expanded on this topic so that we could more fully understand the impact of TNF and its expression.
The authors designed an extensive number of probes. I think they developed a sufficient amount of mutant constructs in order to test for the site of interaction between hTNF-a and LITAF. The different constructs were clearly displayed and labeled in Figure 1; it was easy for the reader to visualize which constructs were used for each test. The only confusing aspect of this figure was what exactly peptides A, B, and C were. It did not become clear to me until later that these were sequences in LITAF that they determined themselves based on the amino acid sequence.
In Figure 2, the DNase footprint experiment contained good controls and amino acid markers. Lanes 1 and 2 (promoter/DNase and promoter/DNase/GST, respectively) served as a good standard to which results could be compared and lane 6 (wildtype LITAF with DNase and promoter) served as a good positive control for the footprinting results of the promoter/LITAF interactions. No lane contained a construct with only peptide B or only peptide C; it may have been helpful to include that, but they were still able to support their results without those lanes. The DNase footprinting was also easy to detect; in lanes 6, 7, and 8, the fusion protein was clearly protecting the DNA by binding.
Figure 3 is also well controlled (both positive and negative). The data turned out very clearly, as it is easy to detect the bands that are shifted upwards due to binding of the protein fragment with the probe. This figure is successfully able to demonstrate that the amino acids 165-180 (peptide B) must be present in order for hLITAF to bind with the CTCCC region of hTNF-a. The authors are also critical of themselves; they realize that they must conduct further experiments to determine if amino acids 165-180 are sufficient for inducing the expression of hTNF-a (as opposed to binding only).
Figures 4 and 5 are good experiments for determining the expression of hTNF-a mediated by LITAF rather than just the binding. The authors have successfully shown that peptide B is sufficient for increasing the levels of transcription of hTNF-a. These experiments are also well controlled, but perhaps they could have included a control in which no peptide was added to the cells. It is helpful to use HA as a control peptide for binding to hTNF-a. In Figure 5b, I think it is somewhat confusing as to why LPS does not increase the levels of transcription for the mutant TNFP3 (-550 to –487 with a deletion from –515 to –511) when LPS increase expression levels for mtNFP1 (also containing a deletion from –515 to –511), although they do address this issue in the discussion. It would have helped to make it more clear in the results section that LPS did not increase expression for mtTNF3 because it lacked sufficient binding sites for the other transcription factors, as well as for LITAF.
TNF-a is a fascinating gene; expression of the gene can have both beneficial and pathological effects on an organism. Because of the complexity of its functions, it is a necessity for TNF-a to be highly regulated. The role of LPS, the numerous transcription factors for TNF-a, and the specific binding sites illustrate just how tightly controlled this gene is. This paper was able to elucidate the role of hLITAF in hTNF-a binding and identify the binding sites of each. The authors talk quite a bit about the other transcription factors that are also responsible for regulating the expression of hTNF-a, however their roles have not been clearly defined. In future studies, I would like to clearly determine the binding sites of other transcription factors for hTNF-a. The transcription factors Ets, NF-AT, and AP-1 could be characterized in a manner similar to the methods in this experiment for hLITAF. I would create multiple GST fusion proteins for each transcription factor that contained different deleted portions of the transcription factor. I would also create constructs of hTNF-a with different deleted portions of the promoter. DNase footprinting and EMSA could once again be used to determine the sites of interaction between hTNF-a and the transcriptions factors. I would also want to do a functional test to determine the expression levels of hTNF-a when in the presence of the different transcription factor constructs. These experiments would help to characterize all of the transcription factors that are involved in the expression of hTNF-a.
Another interesting experiment would be to determine which transcription factors are responsible for activating TNF-a in the presence of a particular pathogen. I would use mouse models that each contain a different knockout mutation for one of the transcription factors. Then I would administer different pathological agents to the mice. I would detect the transcription levels of TNF-a with a northern blot (I would use TNF-a DNA as a probe for the mRNA). I would expect to see decreased levels of TNF-a transcription in a mouse that contains a knockout mutation for the transcription factor that is involved in the response to the specific antigen administered. If a particular transcription factor were not involved in the response pathway for a specific antigen, then I would expect the mouse models with that knockout mutation to have wildtype levels of TNF-a transcription.
The authors mention that understanding the factors that mediate expression of hTNF-a will be useful for diseases related to hTNF-a mutants. Several mutants containing a mutation in TNF-a have been identified in murine models. A mutation in TNF-a causes it to lose affinity for binding with its receptors, thus resulting in a loss of function; mutant TNF-a are no longer able to inhibit malignant cells lines (OMIM 2003). To study possible treatments for TNF-a diseases, I would use a mouse strain containing a mutation in TNF-a. I would then inject mice with high levels of TNF-a in an attempt to correct the phenotype. In order to determine if TNF-a is able to bind with its receptors, I would use a western blot with an antibody to the receptor as a probe. I would use a control lane of the receptor alone determine its molecular weight. Then I would probe the proteins from the mouse to determine if the molecular weight had changed. If TNF-a had bound to the receptor, I would expect the molecular weight of the receptor to increase (and the band to shift upwards on the gel). The same experiment could be conducted with an antibody to TNF-a to measure the molecular weight of the protein alone and the protein-receptor complex. To determine if TNF-a has corrected the phenotype of the mouse (rather than just gained affinity to bind with the receptor), I could either assess the activity level of the receptor or give the mouse a pathological agent to observe if the phenotype is corrected. I think applying the findings from the current study to solutions for genetic diseases would be very beneficial and rewarding.
[OMIM] Online Medelian Inheritance in Man. 2003. Tumor Necrosis Factor. <http://www.ncbi.nlm.nih.gov/entrez/dispomim.cgi?id=191160>. Accessed 2003 May 2.
Davidson College Biology Department
Davidson College Molecular Biology
Please send questions and comments to Sarah Baxter at firstname.lastname@example.org. Thanks!