B Signaling

NEGATIVE FEEDBACK:© AAASNF-κB is held inactive in the cytoplasm by three IκB isoforms. Cell stimulation activates the kinase IKK which leads to phosphorylation and degradation of IκB. This frees NF-κB to enter the nucleus and activate genes including IκBα. IκBβ and -ε are synthesized at a steady rate, allowing for complex temporal control including negative feedback. (From A. Hoffmann et al., Science, 298:1241–5, 2002.)The transcription fac

Aileen Constans(aconstans@the-scientist.com)
Mar 27, 2005


NF-κB is held inactive in the cytoplasm by three IκB isoforms. Cell stimulation activates the kinase IKK which leads to phosphorylation and degradation of IκB. This frees NF-κB to enter the nucleus and activate genes including IκBα. IκBβ and -ε are synthesized at a steady rate, allowing for complex temporal control including negative feedback. (From A. Hoffmann et al., Science, 298:1241–5, 2002.)

The transcription factor NF-κB exists in unstimulated cells as a cytoplasmic homo- or heterodimer bound to inhibitory IκB protein. NF-κB has received a great deal of attention since its discovery more than two decades ago, and for good reason. It regulates genes implicated in innate immunity, inflammation, cancer, and apoptosis. And the molecules associated with the NF-κB signaling pathway are prime drug targets.

This issue's Hot Papers focus on distinct parts of the NF-κB pathway. An interdisciplinary approach combining computer modeling of IκB...


NF-κB is regulated by three isoforms of the inhibitor IκB: IκBα, IκBβ, and IκBε. IκB binding keeps NF-κB localized to the cytoplasm. In the so-called canonical pathway, activation of NF-κB by a stimulus such as TNF-α leads to phosphorylation, ubiquitination, and degradation of the IκB isoforms. This in turn permits NF-κB translocation to the nucleus, where it binds DNA and activates target genes, including the gene for IκBα. Newly synthesized IκBα then binds to NF-κB, inhibiting it.

A Hot Paper from an interdisciplinary group of researchers at the California Institute of Technology and Johns Hopkins University sought to find the precise roles of each IκB isoform and to determine how NF-κB regulates different genes at different times.1 To approach the IκB roles, the researchers developed a computational model based on the assumption that IκBβ and IκBε, which are not regulated by NF-κB, serve to dampen the effect of increased IκBα synthesis. They then tested and confirmed their model in mouse embryonic fibroblasts (MEFs) in which each isoform was knocked out, using experimental data to iteratively refine their model. The resulting model provided evidence that NF-κB signaling is bimodal: Oscillations in NF-κB activation and inhibition time the regulation of downstream target genes.

Data derived from the Science Watch/Hot Papers database and the Web of Science (Thomson Scientific, Philadelphia) show that Hot Papers are cited 50 to 100 times more often than the average paper of the same type and age.

"The IκB-NF-κB signaling module: temporal control and selective gene activation," Hoffmann A, Science , 2002 Vol 298, 1241-5 (Cited in 112 papers, Hist Cite Analysis)"The phosphorylation status of nuclear NF-κB determines its association with CBP/p300 or HDAC-1," Zhong H, Mol Cell , 2002 Vol 9, 625-36 (Cited in 121 papers, Hist Cite Analysis)"Distinct roles of the IκB kinase a and b subunits in liberating nuclear factor κB (NF-κB) from IκB and in phosphorylating the p65 subunit of NF-κB," Sizemore N, J Biol Chem , 2002 Vol 277, 3863-9 (Cited in 104 papers, Hist Cite Analysis)

Yale immunobiologist Sankar Ghosh points out that the work was not only technically challenging, as the authors had to prepare multiple knockout cell lines, but also significant and unique in that it was the first time anyone had analyzed the damping down of the NF-κB pathway via a mathematical model. " [The authors] took a fresh look at something we have known about this whole system, which is that it's an inducible transcription factor that gets activated and then gets deactivated," says Ghosh.

But coauthor Alexander Hoffmann, then a postdoc in David Baltimore's laboratory at CalTech, says that initially so many doubts were expressed about the usefulness of the computational approach that he was reluctant to discuss the research until it was almost ready for publication. "People thought the approach was completely useless," he says. "I think in biology in general, people assume that [if] it's a very complex system, any attempt to formalize it using mathematical equations is making simplifying assumptions that are so radical that you lose the point." Though it took some time, Hoffmann and colleagues eventually demonstrated that the model was predictive of cell behavior.


Indeed, mathematics may have been the only way to understand fully the NF-κB pathway in a time-resolved manner. Hoffmann, now at the University of California, San Diego, notes that previous signal-transduction experiments using transient transfection of cell lines allowed researchers to identify signaling molecules, but they failed to see the temporal dimension of gene expression, or how timing can control which genes are being expressed at a given time in the pathway. "The same pathway can be used for a multitude of different effects, if you are able to control the timing," he says. "Different temporal profiles of the signaling events in the pathway can control different sets of genes."

Andre Levchenko of Johns Hopkins points out that it would have been difficult to use conventional experimental techniques alone to understand the oscillations in a signal transduction pathway, because the complexity of the system is counter-intuitive. "Especially if there are any feedback interactions involved, because our human mind is not trained to think in terms of circular logic, and this is what you have here, a vicious circle of feedback," he says.

Hoffmann notes that current global gene-expression and protein-interaction maps are not yet predictive of how a cell will respond to a particular signal – an ultimate goal in systems biology. Such a map, he says, would need to incorporate time-resolved data.

University of Manchester chemist Douglas Kell adds that a truly predictive model would take single-cell behavior into account rather than rely on bulk cell measurements. "If you average a whole lot of [cells] and they're all out of phase, you don't see any oscillations at all. Whereas if you look at the individual cells, you see very strong oscillations," Kell says.

Mike White and his colleagues from the University of Liverpool, along with Kell and colleagues, used single-cell time-lapsed fluorescent imaging of the movement of NF-κB-GFP fusion proteins to demonstrate that nuclear-to-cytoplasmic oscillations predicted by the Hoffmann/Levchenko model occur in single cells.4 They also showed that persistent oscillations were required to maintain gene expression from an NF-κB-regulated firefly luciferase reporter gene. Kell and colleagues published a sensitivity analysis to show the parts of the model that are most important to the system's behavior.5

"It's not clear from first principles how raising the concentration of NF-κB could affect one or the other of the various cellular processes downstream from the NF-κB signaling [e.g., apoptosis and proliferation], given that all you're doing is changing the same thing. However, if what determines cell fate is not the amplitude of the NF-κB but the frequency, then you can have the [processes] encoded separately," Kell says.



© 2004 AAAS

SK-N-AS cells were designed to express the NF-κB component RelA-DsRed (red) and IκBα-EGFP (green). Time lapse photography (minutes) indicates asynchronous nucleus-to-cytoplasm oscillations after stimulation with TNF-α. The arrow marks one oscilating cell. Scale bar = 50 μm. (from D.E. Nelson et al., Science, 306:704–8, 2004.)

In a second highly cited work, a group from Yale elucidated another way that NF-κB's activity can be modulated. Previously, Sankar Ghosh and colleagues demonstrated that phosphorylation of the p65(RelA) subunit is responsible for recruiting a transcriptional activator, CBP/p300, onto NF-κB. The group's Hot Paper demonstrated that the phosphorylation state of p65(RelA) "seems to act as a switch that determines whether NF-κB protein either binds transcriptional coactivators or binds transcriptional repressors, like histone deacetylases [HDACs]," says Ghosh, who adds that this type of switch had not been reported in any other transcription factor.2

Virologist Warner Greene at the Gladstone Institute of Virology and Immunology at UC-San Francisco has used these findings as a "blueprint" for his studies of HIV latency. Using a model cell line containing a latent HIV provirus, Greene and colleagues found that p50 homodimers bound to an HDAC actively repress the HIV long terminal repeat contributing to the latent state.6 "Then, when we activate the cells with TNF or other mitogens, the repressor is replaced by a transcriptional activator, and the latent virus is expressed," Greene explains.

University of Dundee researcher Neil Perkins has also looked at how the phosphorylation state of NF-κB affects gene regulation. He argues that Ghosh's switch is not the whole story and that other phosphorylation events regulating NF-κB association with HDACs can be found in the p65(RelA) transactivation/transrepression domain. His group has found, for instance, that a residue in the p65(RelA) transactivation domain, threonine 505, is phosphorylated in response to induction of the ARF tumor suppressor.7"This seems to be one of the other phosphorylation events that flips NF-κB into being a repressor of transcription," Perkins says.


The third Hot Paper, from George Stark's lab at the Cleveland Clinic Foundation, probed the roles of the IκB kinase (IKK) subunits, IKKβ and IKKα, in the canonical pathway of NF-κB activation.3 Coauthor Nywana Sizemore explains how earlier studies in knockout mice showed that IKKβ, but not IKKα, was responsible for the cytokine-induced degradation of IκB, but that cells lacking IKKα were deficient in the induction of several NF-κB dependent mRNAs in response to the inflammatory cytokines IL-1 and TNF-α. This suggested to the researchers that IKKα could be required for a second NF-κB activation pathway. Size-more and colleagues used MEF knockout cell lines to study the roles of the β and α IKK subunits in NF-κB activation and phosphorylation of the p65 subunit of NF-κB in response to the IL-1 and TNF-α. They found that IKKα is required for activation of NF-κB through the PI3K/AKT-mediated pathway.

"I think the most important thing that came out of this study was that we identified a novel substrate for the IκB kinases, as well as showed that there was a second pathway necessary for full activation of NF-κB, which was phosphorylation and activation of the p65 subunit of NF-κB, which leads to full transactivation," says Sizemore.

Several concurrent papers confirmed these observations. Work from Albert Baldwin's lab at the University of North Carolina, Chapel Hill, revealed that IKKα accumulates in the nucleus after cytokine exposure and that it regulates NF-κB expression via phosphorylation of histone H3.8 In another study appearing soon after the Sizemore work, Kennneth Marcu's group at the State University of New York at Stony Brook performed an extensive gene-chip study of IKKα, IKKβ, and IKKγ knockout cell lines and showed that cells lacking the IKKα subunit also lacked the inflammatory gene response.9

"This was significant because previous to this work, a lot of people studied the degradation pathway of IκBα, which is mediated primarily by IKKβ, and had kind of written off IKKα as sort of a secondary activator for degradation of IκB and hadn't really looked at other functions that it may have," says Sizemore.

Marcu notes that these results were somewhat controversial at that time, as prior work from Michael Karin's lab at UC, San Diego suggested that IKKα was not involved in the canonical NF-κB activation pathway (in which IκBα is phosphorylated, allowing NF-κB to migrate to the nucleus and activate target gene transcription). Indeed, it remains possible that the disparate conclusions are in part the result of differences between the experiments themselves. Karin's research was performed in mouse knock-out and knock-in models, whereas the Stark paper used MEFs, and it's possible that IKKα's nuclear role might also be related to the cell's becoming immortalized or transformed, says Marcu. Further exploration of the in vivo role of IKKα would require the development of cell type-specific IKKα conditional mouse knockouts, which are currently underway.

Though influential, the Hot Papers represent only a small part of what's proving to be a burgeoning research field in which many questions remain. Dundee's Perkins, for example, is looking at how the functionality of NF-κB is determined by its mechanism of activation. And, Greene at UCSF points out that scientists are currently probing how specific genes are regulated by NF-κB and recruited by transcriptional coactivators, as well as how posttranslational modifications at key residues can change the biology of the pathway. Says Greene: "This is an amazingly rich area, the posttranslational modification of NF-κB itself, as well as the histone and the chromatin that surround NF-κB-responsive genes. It's an unbelievably active area of research right now."

Interested in reading more?

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?