Identification and functional characterization of a novel binding site on TNF-alpha promoter
Xiaoren Tang*, Matthew J. Fenton**, and Salomon Amar* **
Proceedings of the National Academy of Sciences, USA, April 1, 2003. Vol. 100, no. 7, 4096-4101.
*Department of Periodontology and Oral Biology, School of Dental Medicine, and **The Pulmonary Center, School of Medicine, Boston University, Boston, MA 02118
Tumor Necrosis Factor (TNF) is a pleiotropic cytokine that is most often derived from monocytes/macrophages and involved in the inflammatory response processes. The fact that TNF is pleiotropic gives it dual expression capabilities: its expression can be beneficial or pathological. Thus, the expression of TNF alpha is strictly controlled at both transcriptional and post-transcriptional levels. It is thought that TNF is transiently induced by lipopolysaccharide (LPS), and multiple binding sites, coined NF-kB sites, have been determined to be responsible in the mouse. However, the regulation of human TNF alpha (hTNF-alpha) is not known beyond a simple description of transcription factors such as NF-kB, Ets, NF-AT, AP-1, LPS and others, and no specifics about the relative contributions of these factors are known.
LPS has been identified as an endotoxic component located in the outer membrane of Gram-negative bacteria and is known to play a major role in the production and secretion of major inflammatory mediators. Furthermore, LPS is known to up-regulate inducible transcription factors such as NF-kB and down-regulate constitutive transcription factors. It was previously known that a mutation or deletion within the NF-kB binding motif of hTNF-alpha promoter caused a failure of response to LPS, yet the inhibition of NF-kB in humans did not cause full deactivation of TNF-alpha expression. (Yao et. al., Kuprash et. al., Goldfeld et. al.) Therefore, it was evident that NF-kB was not the only determinant of hTNF-alpha promoter induction. It was hypothesized that TNF-alpha promoter may contain several binding sites associated with more than one transcription factor. In previous studies performed by members of this group, a factor in addition to NF-kB was identified and coined hLITAF (lipopolysaccharide induced TNF-alpha factor). (Myokai et. al., Takashiba et. al.) Furthermore, hLITAF mRNA was found in four expected places within the body: placenta, peripheral blood leukocytes, lymph nodes, and spleen. Conclusively, NF-kB and hLITAF must both contribute to the induction of hTNF-alpha promoter.
Utilizing the new-found information that hLITAF is a binding factor in the expression of TNF-alpha, the group used DNase I footprinting to find a 6 nucleotide sequence on TNF that binds hLITAF. Subsequently, the group used electrophoretic mobility-shift assays (EMSA) and ELISA with peptide constructs to determine the peptide region on hLITAF that is necessary and sufficient for mediation of binding to TNF-alpha promoter, and thus the strict regulation of its expression, in an effort to apply these data to pharmacology study of TNF-related diseases.
Cloning constructs and mutagenesis were performed in DH5alpha strain of E.Coli and clones (in GST fusion protein) underwent purification in strain BL21. The cell culture used was of THP-1 monocytes. hLITAF DNA fragments were created by PCR and subcloned into pGEX4T-1 vector. (Primer sequences used for each construct can be found within the Materials and Methods section of the text.)
DNase I footprinting was performed using 2 oligonucleotides: a (-)487 - (-)550 LITAF template with incorporated HindIII site at 5' end; and a (-)550 - (-)535 primer fragment. The template and fragment were then incubated together in Klenow solution.
EMSA was used to explore protein-DNA interaction by running GST-hLITAF fragments with a radiolabeled oligo dsDNA on non-denatured 6% polyacrylamide in TBE.
ELISA was used to explore the sufficiency and necessity of three peptides on hLITAF. Peptides A, B, C and negative control, hemagglutinin (HA), were solubilized in DMSO and delivered to THP-1 cells. Cell supernatants were harvested and TNF-alpha secretion was measured.
Fig.1A depicts 8 hLITAF constructs subcloned into pGEX4T-1 and fused with GST fusion protein. The first construct is a full construct including a depiction of peptides A, B and C. Constructs 2-8 contain partial segments of hLITAF including various combinations of peptides. Constructs 7 and 8 display the deletions created in peptide B and peptide C, respectively.
Fig. 2A depicts hTNF-alpha promoter/pGL3 constructs. The first construct is a full hTNF-alpha promoter, showing 4 regions for the transcription factor binding sites, the TATA box and a luciferase reporter (Luc). Constructs 2, 3 and 4 contain deletions for LITAF, the other three binding sites, and the other three binding sites plus a deletion in LITAF, respectively.Results
DNase I Footprinting - Exploratory Confirmation of CTCCC hLITAF Binding Site
To find the specific hLITAF binding site on hTNF-alpha promoter, DNase I footprinting was used with a 3' HindIII site designed into the DNA and probed with a 32P ATP labeled DNA fragment. This design allowed the group to make probe hTNF-alpha promoter constructs and digest the fragments in HindIII restriction enzyme to ensure the probe was only radiolabeled at its 5' end. When the probe binds, the dsDNA fragments are degraded base by base with DNase I for five minutes (long enough to produce fragments of every length). Figure 2 shows the results of DNase I footprinting.
Fig. 2A is a radiographed gel exploring the interaction of hLITAF and hTNF-alpha DNA. The image includes a full negative control lane (making sure 5 minutes of digestion in DNase I produced the desired spectrum of fragment lengths), a partial negative control lane with only GST fusion protein probe (making sure GST is digested as expected), and the first 6 GST-hLITAF constructs depicted in Fig. 1A. Protected, undigested regions are shown in constructs 1, 5 and 6, those which contain peptides B and C. These regions are noted on the gel by lightness matching background levels closer than areas where DNA is present. The researchers claim that hTNF-alpha promoter DNA probed with hLITAF constructs that contain only peptides B and C (not peptide A) show slightly more protection. The researchers further provide their conclusion that this region spans nucleotides (-)511 - (-) 515 on hTNF-alpha promoter by including a molecular weight marker that was run to the left of the gel.
Fig. 2B shows the nucleotide sequence for which this protected region corresponds. This motif is indicated with the sequence CTCCC.
EMSA - Narrowed Exploratory Confirmation of the CTCCC/LITAF Peptide B Complex
Using the data from the DNase footprint, the researchers narrow the search for the specific binding site to Fig.1A constructs 5 and 6 (containing only peptides B and C). Firstly, Fig. 1A constructs 1, 3, 6 and GST fusion protein alone were tested for their complexing with hTNF-alpha promoter by 32P labeled hTNF-alpha promoter oligo through EMSA. Secondly, the specific amino acids in hLITAF that are responsible for binding to hTNF-alpha promoter were detected using EMSA with mutated constructs 7 and 8 depicted in Fig. 1A.
Fig. 3A is a radiograph of a 32P labeled hTNF-alpha promoter/GST-hLITAF EMSA. GST/hLITAF constructs 1, 3 and 6 from Fig. 1A contained full, peptide A inclusive and peptide B inclusive wild-type fragments, respectively. Negative controls were run to show background levels and shifts were detected in constructs 1 and 6, indicating that either hLITAF peptide B and/or peptide C were necessary for complex formation with hTNF-alpha promoter. It is notable that the band shift for the full construct 1 is significantly higher on the gel than the band for construct 6. This may indicate the increased molecular weight and size of the protein/DNA complex when all three peptides are potentially involved in complexing.
Fig. 3B is a similar EMSA radiograph showing a more narrowed approach. In addition to constructs 1, 3 and 6, panel B includes constructs 7 and 8, containing deletions in peptides B and C, respectively. Controls similar to panel A were used, including a negative 32P-hTNF-alpha mutant promoter probe which provides evidence for the specific CTCCC motif necessary for hLITAF binding. Shift was detected in construct 8, indicating that peptide B alone is sufficient to complex with hTNF-alpha promoter. Similar shifts for constructs 1 and 6 were present in panel B as in panel A.
ELISA - Function Confirmation of hLITAF Specific Binding
hLITAF peptide regions were solubilized and transfected into THP-1 cells (along with an HA negative control) to determine the sufficiency of each in hTNF-alpha secretion. The levels of secretion were measured by ELISA
Fig. 4 is a bar graph depicting hTNF-alpha levels of secretion for each peptide (A, B, C, HA) for increasing concentrations of peptide. As concentration levels of peptide increase linearly, it appears that levels of secreted hTNF-alpha in A and C treated cells similarly increases linearly. However, levels of secreted hTNF-alpha treated with peptide B appear to increase exponentially. Quantitatively, the researchers estimate that peptide B increases levels of hTNF-alpha secretion by 2.4 fold above unstimulated, background levels (namely, levels in HA treated cells).
Luciferase Assays - Functional Confirmation of the CTCCC Motif
To functionally confirm the CTCCC motif as a DNA binding region, hTNF-alpha promoter/ Luc reporter constructs (from Fig.1B) were transiently transfected into THP-1 cells and subsequently stimulated by either LPS, peptides A, B, C or HA.
Fig. 5A is a bar graph showing the triplicate luciferase assay results for transfection and stimulation of constructs 1 and 2 from Fig. 1B. Construct 1 is wild-type hTNF-alpha promoter/Luc reporter and construct 2 contains a deletion in the LITAF binding region. LPS stimulated both constructs and provides a positive control background. Similarly, the transfection with HA provides a negative background control. Compared to the HA background control, tranfection of peptide B into wild-type hTNF-alpha promoter/reporter cells causes a 2 fold increase in luciferase expression. Contrastingly, luciferase production in the mutant constrcut was not observed through any of the variable peptides.
Fig. 5B is a bar graph evaluation of luciferase expression similar to panel A. Mutant hTNF-alpha promoter/Luc reporter constructs 3 (containing only LITAF binding region, TATA box and Luc reporter exon) and 4 (containing only TATA box and Luc reporter) from Fig. 1B were transfected into THP-1 cells. LPS stimulation causes expression in construct 3 but not in construct 4. Similarly, the presence of peptide B mirrors the expression seen under stimulation by LPS; the researchers claim that peptide B causes a 2.4 fold increase in luciferase expression (compared to HA background) when transfected into construct 3 and no expression in construct 4.
It can be concluded from figure 5 that the presence of both peptide B of LITAF and the appropriate hTNF-alpha promoter LITAF binding site is necessary for hTNF-alpha expression.
Although it had already been determined that TNF-alpha promoter is regulated by NF-kB and possibly other factors in the mouse, it can be concluded that a similar system is present in the human (hTNF-alpha promoter). hTNF-alpha can be expressed when stimulated with LPS if some of the helper DNA binding regions are present despite the absence of hLITAF binding region. However, a specific interaction between peptide B of hLITAF (aa 165-180) and nucleotides (-)515 - (-)511 of hTNF-alpha promoter sequence is sufficient for normal hTNF-alpha promoter activation.
The first half of the discussion focuses on a detailed summary of the conclusions made via DNase I footprinting, EMSA, ELISA and luciferase assays. A more concise conclusive statement can be found above. The second half begins with an explanation of LPS stimulation of the mtTNFP1-pGL3basic construct (dark bar) in Figure 5A. Why does LPS successfully stimulate hTNF-alpha promoter expression if the hLITAF binding sequence has been deleted? An explanation offered by the authors is that, while endogenous hLITAF will not be able to function if the CTCCC site is deleted, this construct contains other inducible sites (NF-kB, AP-1, etc.) that allow full LPS to induce hTNF-alpha promoter. The authors continue by offering another affirmation that, since there are clear differences in expression in Fig.5B, the 165-180 amino acid region (peptide B) of hLITAF is an independent domain that is sufficient for expresssion of TNF-alpha gene.
Abstract: The abstract is a very thorough, concise account of the methods and conclusions made by this team of researchers. It begins with a brief broadening of the previous experimentation in the mouse, which is necessary for deeper involvement and acknowledgement of the purpose in specifying a similar system in humans. Additionally, the end of the abstract provides insight into the greater application of this specific knowledge to pharmacology and prevention of TNF-related diseases, enticing the reader to a deeper exploration of the paper.
Introduction: Focusing on background information about TNF and LPS, respectively, the introduction provides a concise account of the complexity of this protein/DNA interaction. By offering a healthy amount of detail of previous experiments, the researchers provide a strong background with which to build in their own experimentation. A summary of the conclusion is well-stated at the end of the introduction.
Materials and Methods: A detailed description of techniques allows the reader to follow along and reproduce the experiment if he so desires. The inclusion of full primers sequences for all of the constructs provides muffled reading for the novice, but critical information for the expert. Figure 1 is critical for continuity and unity throughout the paper and essential for the novice reader, yet a more specific description including nucleotides and amino acid labelings on the cartoons would help any reader through the paper. Furthermore, in the synthesis of hLITAF/pGEX4T-1 constructs and hTNF-alpha promoter/pGL3 constructs, all pertinent combinations are seen, giving the reader foreshadowed insight into the types of analyses seen further along in the paper.
Figure 2: A well-contrasted gel is produced, with expected controls, confirming proper digestion and labeling and negation of any artifacts caused by GST fusion protein. The MW marker is a much needed component as well, yet it would be desired to have a more detailed marker (i.e. every 5 bases or so). Overall, the authors' conclusion based on the data Fig. 2 provides is accurate and the technique for ensuring one-sided DNase I digestion by a 3' HindIII site is ingenious. However, their interpretation that there exist stark differences in the footprints of constructs 1, 5 and 6 is questionable. It does not seem necessary that Figure 2B is included as part of the selected data shown.
Figure 3: EMSA is a logical next step in the process of narrowing down the protein/DNA interaction. It is a highly sensitive technique and it is evident in the resolution of Figure 3 that the authors knew that these data would afford them a major stepping block. The inclusion of part A shows the authors honesty and care in creating a foundation for panel B by positively identifying the presence of protein/DNA complexing. The different detection between construct 7 and 8 in panel B is the most compelling evidence in the paper. Both panels A and B were well-controlled by the authors and the (+/-) array above each radiograph gives the reader a simple and clear understanding of the ingredients of each lane. However, the inclusion of the 32P hTNF-alpha promoter (-550 ~ -487, oligo DNA) is the key control in this experiment. It shows that, in addition to the presence of specific amino acids of hLITAF, the CTCCC motif must be present to allow the protein/DNA complex to be formed. The depiction in Figure 3 is professional and clear, and it appears that the authors are confident in their conclusions.
Figure 4: Less important than the following data, this figure merely shows that the authors were able to understand that peptide B, corresponding to amino acids 165-180, caused an increase in TNF-alpha secretion. Fig. 4 provides a first step in the functional/positive confirmation of the specificity of binding, however, it lacks a positive control which would aid in its interpretation. Here the authors get by with declaring a 2.4 fold increase without the presence of a positive control.
Figure 5: Knowing that LPS might allow induction of hTNF despite a lack of some binding sites, the authors smartly used LPS stimulation of cells as a positive control. Inclusion of the deletions, to refresh the reader, are important and included for good reason. Overall, Fig. 5 provides compelling evidence in a form easily digestable by the reader. Again, however, the authors get by with a fuzzy number account of the increase in hTNF-alpha expression. A more quantified method would be appreciated.
Overall, the progression in data is logical and easy to follow, from the first apperance of the constructs, to the narrowing of identification of binding and finally to the functional/positive confirmation of this protein/DNA induction, both its necessity and its sufficiency.
Discussion: Other than consolidating and restating the detailed conclusions provided in the progression of data, the discussion does little to provide foresight into explanations and possible future experiments that the reader was directed to at the end of the abstract. An explanation of the more detailed aspects of Figure 5 touches on some of the hypotheses from previous experiments, stated in the introduction, yet the final statement of the paper leaves the reader empty-handed: "The elucidation of these mechanisms should help in the design of new pharmacological approaches aimed at addressing TNF-related diseases." The authors' keen insight so well-shown through the data seems to falter when faced with the possibility of providing a detailed explanation of future ideas for experimentation. It seems Tang, Fenton and Amar were ready to hit the beach.
The large picture that this study begins to address involved the cooperation of various transcription factors in the induction process of hTNF expression. Full knowledge of the mechanism of regulation of hTNF will be a major key to unlocking the mysteries of tumor growth and regulation. Therefore, the next step is to determine the capabilities of other transcription factors that may bind to hTNF-alpha promoter.
It is noted that NF-kB is thought to play a major role in determination in mice and possibly a similar role in humans. How important is the role of NF-kB and/or other aforementioned inducible factors in relation to hLITAF? Is there analogous necessity and sufficiency for hTNF-alpha promoter? A logical next step would be to determine whether presence of NF-kB binding site alone is sufficient for expression under LPS stimulation. A construct, similar to those in Fig. 1B, containing only the potential NF-kB binding site could be used in experiments analogous to those run in this paper (DNase footprinting, EMSA, ELISA and luciferase assays). It could be hypothesized that NF-kB might interact mostly with its correlating binding site on hTNF-alpha promoter and that the specific amino acids that induce this function could be identified by these methods. Other transcription factors (Ets, NF-AT, AP-1, AP-2, STAT1, etc.) could be analyzed in similar fashions, using EMSA complexing data as a main driving force to conclusions - it is evident from the authors' use of EMSA, that it can be a truly powerful tool.
Another approach is to investigate the importance of peptides A and C of hLITAF. It would be hypothesized that these peptides interact with hTNF-alpha promoter in some way, but how specifically? What conclusions can be made about these peptides after it has been determined that they are not necessary for hTNF expression under LPS stimulation. Using a more detailed, elongated construct of hTNF-alpha promoter region containing all possible transcription factor binding sites and hLITAF protein constructs containing only peptide A and peptide C, protein/DNA interactions could be investigated by DNase I footprinting coupled with EMSA. These findings might help to construct a detailed scheme of all induction and transcription regulation involved in hTNF.
If it is found that other transcription factors exhibit sufficiency analogous to hLITAF, it could be concluded that the different combinations of these factors lead to different expressions of hTNF, either beneficial or pathological. If it is concluded that hLITAF/hTNF-alpha promoter interaction is the primary mode of hTNF regulation, it could be concluded that the secondary effects of other transcription factors might lead to the observed, strict hTNF regulation. Either way, the effects of inducible transcription factor binding sites other than NF-kB and hLITAF must be investigated.
Once a confirmed mechanistic scheme is determined, different hTNF-alpha promoter constructs could be subcloned into normal human genomic monocyte/macrophage cell lines. This subcloned construct could then be inserted into human subjects at: the localized regions noted in the introduction (placenta, peripheral blood leukocytes, lymph nodes and spleen); the blood stream of the patient; and the localized tumor site of patients. Variable presence of particular binding sites could be tested. Similarly, inducible high copy vectors of the factors NF-kB, hLITAF and others could be transfected into these sites. Over time, the size and virility of tumors could be measured. With carefully controlled experiments, it is hoped that an optimal combination of factors and binding sites could be identified for desired pharmacological effects. The batch of ingredients that induced this beneficial expression of hTNF could be therapeutically transfected into effected humans on a large scale with hopeful outcomes.
Goldfeld, A. E., Doyle, C. & Maniatis, T. (1990) Proc. Natl. Acad. Sci. USA 87, 9769-9773.
Myokai, F., Takashiba, S., Lebo, R., & Amar, S. (1999) Proc. Natl. Acad. Sci. USA 96, 4518-4523.
Purves, et. al. Life: The Science of Biology, 6th ed. Sinauer Associates, Inc: Sunderland, MA. 2001, 368.
Takashiba, S., Van Dyke, T.E., Shapira, L. & Amar, S. (1995) Infect. Immuno. 63, 1529-1534.