Regulating Inducible Transcription Through Controlled Localization

See allHide authors and affiliations

Science's STKE  17 May 2005:
Vol. 2005, Issue 284, pp. re6
DOI: 10.1126/stke.2842005re6


Many signaling pathways regulate the activity of effector transcription factors by controlling their subcellular localization. Until recently, the cytoplasmic retention of inactive transcription factors was mainly attributed to binding partners that mask the nuclear localization signals (NLSs) of target proteins. Inactive transcription factors were thought to be exclusively cytoplasmic until their activation, after which the NLSs were unmasked to allow nuclear translocation. There is now a growing body of evidence, however, that challenges this simple model. This Review discusses recent reports that suggest that inducible transcription factors can constantly shuttle between the cytoplasm and the nucleus, and that their apparent cytoplasmic retention can be achieved by binding partners that mask the NLSs, tether the transcription factor to cytoplasmic structures, or mark the transcription factor for proteasomal degradation. We also discuss the possibility that this more complex model of cytoplasmic retention might be applicable to a broader range of transcription factors and their associated signaling pathways.


The regulation of gene expression by extracellular signals often requires that the effector transcription factors remain inactive until a signal promotes their activation. This is achieved in many signaling pathways through sequestration of inducible transcription factors in the cytoplasm. Upon activation, these transcription factors translocate to the nucleus, where they either activate or repress expression of specific target genes. One well-studied example of a transcription factor that is regulated through cytoplasmic-to-nuclear translocation is nuclear factor κB (NF-κB). In unstimulated cells, NF-κB is retained in the cytoplasm through masking of the nuclear localization signals (NLSs) on NF-κB dimers by inhibitory proteins known as IκBs [see (1, 2) for review]. Exposure of cells to stimuli that activate NF-κB leads to the activation of a protein kinase, the IκB kinase (IKK) complex, which phosphorylates IκBs on two conserved N-terminal serine residues. Phosphorylated IκB is then ubiquitinated and degraded, thus releasing NF-κB, which is recognized by the nuclear import machinery and quickly shuttled into the nucleus to regulate NF-κB–dependent gene expression. Such a unidirectional model of inducible transcription (Fig. 1) is also seen in other signaling pathways, including steroid receptor (SR) and Wnt signaling pathways (3, 4). However, this simple paradigm appears insufficient to explain how the localization of several transcription factors is controlled. Recent work suggests that many inducible transcription factors are not statically localized in the cytoplasm or the nucleus, but instead constantly shuttle between these compartments (1). In addition, cytoplasmic retention of inducible transcription factors is a complex process that is not achieved simply by masking of the NLSs (5, 6). Creating a more detailed and accurate model of how inducible transcription is regulated is necessary to fully understand how gene expression can be tailored in response to changes in the extra- and intracellular environment. In this review, we summarize recent work that has revealed additional mechanisms through which the localization and activity of certain inducible transcription factors are regulated. We do not intend to provide an exhaustive discussion of all mechanisms regulating inducible transcription factors, but rather to highlight a few pathways that we believe illustrate certain general principles that are likely to apply to many other signaling pathways.

Fig. 1.

Original model of the subcellular localization of inducible transcription factors. In unstimulated cells (left), a retention factor (RF) sequesters the transcription factor (purple) in the cytoplasm by masking the nuclear localization signals (NLSs). In stimulated cells (right), the RF dissociates from the transcription factor and unmasks the NLSs, enabling nuclear translocation.

Rethinking an Old Paradigm: Nucleocytoplasmic Shuttling of NF-κB

One of the more surprising discoveries in recent years has been the finding that NF-κB is not passively retained in the cytoplasm in unstimulated cells. Rather, NF-κB–IκBα complexes appear to continuously shuttle between the cytoplasm and the nucleus. The most compelling evidence for such nucleocytoplasmic shuttling of NF-κB–IκBα complexes was achieved by inhibiting CRM1, a key mediator of nuclear export (7). In unstimulated cells treated with leptomycin B, a specific inhibitor of CRM1, practically all of the NF-κB–IκBα complexes were nuclear, which suggests that NF-κB–IκBα is constantly being imported into the nucleus and then actively exported. The mechanism of this shuttling has been further investigated with respect to the most abundant NF-κB heterodimer, p65-p50, and is now attributed to incomplete masking of the NF-κB NLS by IκBα (8). Although the IκBβ isoform masks the NLSs on both p65 and p50 and therefore remains exclusively cytoplasmic, IκBα does not fully mask the p50 NLS. The exposed NLS on p50 enables the constant nuclear import of NF-κB–IκBα complexes, which may be aided by the nonclassical NLS on IκBα. However, after nuclear import, NF-κB–IκBα complexes are rapidly exported to the cytoplasm because exposed nuclear export signals (NESs) are present on IκBα and p65 (1, 2). Therefore, the steady-state cytoplasmic localization of NF-κB–IκBα complexes in unstimulated cells reflects an imbalance between the contribution of the export and import signals on NF-κB and IκBα proteins (7). The potent NESs on IκBα and p65 effectively overpower the two exposed NLSs on IκBα and p50. Although the biological function of NF-κB–IκBα shuttling is not clear, it is possible that this dynamic state facilitates the rapid detection and response to stimulation and also the rapid inactivation of NF-κB by nuclear export in the absence of signal.

Another mechanism of regulating NF-κB activity that has particular relevance to this review is the sequestration of NF-κB–associated coactivators. One such coactivator is SIMPL (signaling molecule that interacts with pelle-like kinase), which was initially identified as an adapter molecule between IRAK-1 [interleukin-1 (IL-1) receptor-associated kinase–1] and IKK and now is also known to specifically coactivate NF-κB p65–dependent transcription (9, 10). In resting cells, SIMPL is found in both the cytoplasm and the nucleus, with a preference for cytoplasmic localization. In cells stimulated by tumor necrosis factor–α (TNF-α), but not IL-1, SIMPL translocates to the nucleus through action of its C-terminal NLS. The mechanism through which SIMPL is retained in the cytoplasm in unstimulated cells is currently unknown, but it may be mediated by IRAK-1 or IKK. Upon cell stimulation, SIMPL may be modified by a phosphorylation event that causes it to detach from one or more retention factors and increases the affinity of SIMPL for activated NF-κB p65. This potential method of selectively releasing SIMPL under activated conditions is similar to that proposed for the coactivator p/CIP (see below). Although the mechanisms of controlling the localization of SIMPL are not well understood, evidence of its cytoplasmic retention has implications for how the response of an inducible transcription factor can be fine-tuned to a specific stimulus. Not only is the appropriate relocalization of NF-κB required for TNF-α–dependent gene regulation, the relocalization of the coactivator is also necessary. Therefore, it can be imagined that IL-1–induced NF-κB will not transactivate TNF-α–responsive genes because nuclear SIMPL is absent.

Although a complete discussion of NF-κB regulation is beyond the scope of this review, the mechanisms discussed thus far illustrate key principles that are used for modulating inducible transcription through controlled localization (Fig. 2). Thus, although inducible transcription factors are generally enriched in the cytoplasm under unstimulated conditions, the steady-state subcellular localization of the transcription factor in unstimulated cells may mask a dynamic nucleocytoplasmic shuttling of the transcription factor, where the default subcellular localization is determined by the relative strength of the nuclear import versus export signals; moreover, coactivators necessary for the activity of inducible transcription factors may themselves be retained in the cytoplasm under unstimulated conditions. We next illustrate the applicability of these mechanisms to a few inducible transcription factors and their associated signaling pathways.

Fig. 2.

A current model of the subcellular localization of inducible transcription factors. This figure depicts a more complex, comprehensive model of how inducible transcription factors are regulated by controlled localization. In unstimulated cells (left), transcription factors may be found in one of three locations: (i) distal from the nucleus and close to the site of induction (purple), (ii) perinuclear (yellow), or (iii) constitutively shuttling between the cytoplasm and the nucleus (red). In stimulated cells (right), retention factors (RFs) dissociate from cytoplasmic transcription factors to enable their nuclear translocation (purple and yellow) while a fraction of the transcription factors may continue to shuttle (red).

Nucleocytoplasmic Shuttling with Periods of Retention: Smads

Smad proteins are a family of inducible transcription factors that mediate signals from the transforming growth factor–β (TGF-β) superfamily of ligands, including TGF-β and bone morphogenetic protein (BMP). Smads can be divided into three classes: (i) receptor-regulated Smads (R-Smads 1, 2, 3, 5, and 8), within which Smad1 is specific to BMPs and Smads 2 and 3 are specific to TGF-β; (ii) mediator Smad (Smad4); and (iii) inhibitory Smads (I-Smads 6 and 7) (6). It was originally proposed that Smads 2, 3, and 4 remain in the cytoplasm in unstimulated cells; after exposure of cells to TGF-β, Smad2 or Smad3 and Smad4 would oligomerize and then translocate to the nucleus (11). As seen in the case of NF-κB, this model of TGF-β–dependent Smad relocalization has proven to be too simplistic. In fact, Smads 2, 3, and 4 undergo constitutive nucleocytoplasmic shuttling marked by discrete periods of cytoplasmic or nuclear retention (12, 13). This dynamic pattern of Smad localization probably allows for the rapid activation and inactivation of Smads, thereby assuring that Smad-mediated transcription directly reflects the temporal pattern of TGF-β stimulation. Specifically, the preferential cytoplasmic localization of inactive Smads places these factors near their intracellular activators. In cells exposed to TGFβ, Smads 2 and 3 and Smad4 are phosphorylated and oligomerize, and the preferential nuclear localization of this activated complex allows Smads to recruit the necessary coactivators and drive transcription. Soon after Smad nuclear translocation, nuclear phosphatases rapidly dephosphorylate the active Smad complex and promote its dissociation/inactivation (14).

Although these potential functions of Smad shuttling and retention are compelling, the study of Smad localization has traditionally relied on in vitro analyses that assess the steady-state localization of cytoplasmic and nuclear Smads. These studies have clearly shown a TGF-β–dependent change in Smad localization, whereas shuttling of Smads has only been inferred from indirect experiments. Recently, these inferences have been confirmed by exciting work that provides a real-time demonstration of the dynamic nature of Smad localization (15). In addition, the mechanism of Smad nucleocytoplasmic shuttling has also been determined. With respect to Smad2 and Smad3, continuous shuttling is mediated by an exposed NLS (14) and active, CRM1-independent export (12). In contrast, Smad4 is shuttled into the nucleus via its NLS but is actively exported from the nucleus by a CRM1-dependent process (13). Modulation of this shuttling favors either the cytoplasmic or nuclear localization of Smads, depending on whether a stimulus is absent or present. Therefore, individual Smads display a distinct pattern of localization under basal and stimulated conditions (15). In resting cells, the majority of Smad2 and Smad3 is cytoplasmic. Similarly, Smad4 is predominantly cytoplasmic in unstimulated cells but undergoes rapid, continuous nucleocytoplasmic shuttling. After stimulation of cells with TGF-β, Smads 2 and 3 and Smad4 become predominantly nuclear and the degree of nucleocytoplasmic shuttling decreases, which suggests that Smads may be actively retained in the nucleus of stimulated cells. The cytoplasmic localization of Smads in unstimulated cells could be brought about by binding of monomeric Smads to cytoplasmic retention factors in unstimulated cells. (15). This mechanism is supported by the finding that Smads 2, 3, and 4 bind to microtubules (MTs) and that this binding inhibits Smad2 activation and translocation to the nucleus (6). Stimulation of cells with TGF-β triggers the dissociation of Smads from MTs. This idea has, however, been challenged by reports that MT disruption does not consistently release Smad2 from cytoplasmic retention (15). Another possible cytoplasmic retention factor is the Smad anchor for receptor activation (SARA), which binds and sequesters Smads 2 and 3 (14). Although SARA does bind a subset of Smad2 and Smad3, the cytoplasmic localization of SARA to the early endosome is inconsistent with the pancytoplasmic distribution of Smads 2 and 3 (15). Consequently, SARA is unlikely to be the predominant cytoplasmic retention factor for Smads. With respect to nuclear Smad retention factors, associated cofactors such as the forkhead class transcription factor FAST-1 and ECSIT (evolutionarily conserved signaling intermediate in Toll or TGF pathways), or weak scanning interactions with DNA, might be responsible for retaining active Smads in the nucleus (1416).

A different mechanism for Smad3 cytoplasmic localization and inducible activation suggests that the localization of Smads may be controlled by various unexpected binding partners and can be cell type–specific. In this case, the insulin-stimulated protein kinase Akt (protein kinase B) sequesters Smad3 at the cell membrane and thereby prevents Smad3 activation by TGF-β (17). Because TGF-β–induced Smad3 promotes apoptosis and insulin-induced Akt prevents apoptosis, the controlled localization of Smad3 by Akt provides a mechanism through which these two signaling pathways can communicate and modulate cell survival. Indeed, costimulation with insulin and TGF-β induces apoptosis in cells with a low Akt/Smad3 ratio (such as Hep3B cells), whereas the same stimulation encourages survival in cells with a high Akt/Smad3 ratio (such as RIE-1 cells). Although this example is specific to Smad3-dependent apoptosis, it demonstrates that cytoplasmic sequestration of inactive Smads (Smads 2, 3, and 4) can fulfill the unanticipated function of integrating multiple signaling pathways in addition to ensuring that TGF-β stimulation is rapidly detected and propagated.

Inactivation by Cytoplasmic Sequestration

Both Smads and NF-κB proteins are regulated by cytoplasmic localization brought about by their interaction with specific retention factors. In the case of NF-κB, the release of NF-κB from IκBα is directly mediated by the proteasomal degradation of IκBα, a result of a well-characterized pathway of phosphorylation and ubiquitination. NF-κB–dependent de novo synthesis of IκBα results in the reformation of NF-κB–IκBα complexes and the resumption of cytoplasmic localization. However, although numerous interactions between retention factors and inducible transcription factors have been reported, the mechanisms whereby these interactions are regulated are less well defined. In particular, it is often unclear how a retention factor selectively releases its binding partner in response to appropriate stimulation. For example, the events governing the affinity of Smads for MTs remain to be defined. A series of recent studies has shed new light on diverse mechanisms by which other transcription factors and their coactivators are retained in the cytoplasm and selectively released.


Nrf2 (NF-E2–related factor 2) is a member of the antioxidant response element (ARE) family of transcription factors and is a major regulator of cytoprotective genes expressed in phase 2 of a detoxification response (18). In resting cells, Nrf2 is localized to the cytoplasm because of its association with Keap1 (Kelch-like ECH associating protein 1), which regulates Nrf2 activity through two mechanisms (19). First, Keap1 functions as part of a novel E3 ubiquitin ligase complex that targets Nrf2 for proteasomal degradation (20). Second, Keap1 anchors any intact Nrf2 to the actin cytoskeleton, which localizes cytoplasmic Nrf2 in a perinuclear region near the 26S proteasome and the endoplasmic reticulum (ER), where phase I enzyme products are released (21). Upon stimulation with these enzymes or cytotoxic stressors, Keap1 undergoes posttranslational modifications, many of which are not well understood. However, certain cytotoxic agents induce modifications of the thiol groups on Keap1 residues Cys151, Cys273, and Cys288, thereby inhibiting its E3 ligase activity and allowing Nrf2 to accumulate and translocate into the nucleus (20). In this case, it is clear that the retention factor is not only sensitive to the relevant stimulation, but also localizes the inactive transcription factor such that it can readily respond to activating stimuli.


As a negative regulator of lipid biosynthesis in yeast, Opi1p remains inactive until stimulated by the presence of high intracellular levels of the lipid inositol and subsequent biosynthesis of phosphatidylinositol (PI) (22). In the absence of free inositol, Opi1p is anchored to the ER by binding to Scs2p. Scs2p is an integral ER membrane protein, and its association with the FFAT [two phenylalanines (FF) in an acidic tract] domain of Opi1p is necessary for the cytoplasmic retention of this transcription factor (23). Once localized to the ER, the Opi1p NLS binds phosphatidic acid (PA) in the ER membrane and is effectively masked (24). The masking of the Opi1p NLS by PA is important for its function because PA is consumed during the biosynthesis of PI. Consequently, in the presence of high concentrations of intracellular inositol, PI biosynthesis reduces the PA content of the ER, thereby exposing the Opi1p NLS. The accompanying step needed to release Opi1p from Scs2p has not been determined, but the change in ER membrane composition could itself interfere with the interaction of Opi1p and Scs2p. Alternatively, inositol or a downstream product may induce a posttranslational modification of Opi1p or Scs2p, resulting in a reduction in their binding affinity. It is interesting to speculate about why Opi1p uses both PA and Scs2p as retention factors. PA has the advantage of being highly sensitive to the inductive signal (inositol), but this phospholipid may not bind Opi1p with enough avidity to prevent nuclear translocation. In addition, PA is not a major membrane component, and even a small reduction in its levels could release Opi1p. By using Scs2p as an additional retention factor, inactive Opi1p will be robustly bound to the ER membrane and will only be released upon reduction of PA. Alternatively, Scs2p may be necessary to quickly sequester and inactivate Opi1p, whereas the renewal of ER PA levels after inositol-induced consumption might take a substantial amount of time. Although it is purely speculative, it is possible that Scs2p plays a predominant role in the retention of Opi1p while PA plays a predominant role in initiating its stimulus-induced release.


TFII-I is a mitogen-activated transcription factor that, in the absence of growth factors, is retained in the cytoplasm by the p190A Rho guanosine triphosphatase activating protein (GAP). The binding of TFII-I to p190A RhoGAP appears to depend on the FF domain of p190A, and stimulation by various growth factors such as platelet-derived growth factor (PDGF) and epidermal growth factor (EGF) induces a tyrosine phosphorylation event within the first p190A FF repeat, which may disrupt its interaction with TFII-I (25). Although little is known about p190A-dependent retention of TFII-I, the presence of an FF domain in p190A is reminiscent of the key phenylalanines (FFAT) that mediate the retention of Opi1p in yeast (23). The FF domain is a highly conserved protein interaction motif that is specifically involved in controlling the nuclear localization of proteins (26). FF domains are found in many eukaryotic proteins and might be a shared component of many cytoplasmic retention factors.


p/CIP (p300/CBP-interacting protein, also known as SRC-3 in humans) is a steroid receptor coactivator (SRC) (27). p/CIP binds to and coactivates various transcription factors, including NF-κB, AP-1, STATs, ETS, p53, and E2F (28). Although the subcellular distribution of p/CIP varies among cell types, in the majority of cells, p/CIP is exclusively cytoplasmic under unstimulated conditions (29). Many of the transcription factors with which p/CIP associates are themselves retained in the cytoplasm until stimulation, thereby creating a situation analogous to that already discussed for NF-κB and SIMPL. The factors that cause the cytoplasmic retention of p/CIP are not well defined, but separate fractions of unstimulated p/CIP associate with MTs and the IKK complex (29, 30). Considering MTs as a retention factor for p/CIP presents a conundrum because an intact MT cytoskeleton is necessary for the nuclear translocation of activated p/CIP. A possible explanation is that p/CIP is transported along MTs during nuclear translocation but that this transport is inhibited in unstimulated cells. According to this speculative model, anchoring p/CIP to the MT transport machinery would prime p/CIP for nuclear translocation upon cell stimulation. In addition, the gross reorganization of the MT cytoskeleton that accompanies cell division may mediate the cell cycle–specific relocalization of p/CIP that has been observed in certain cell types. Retention by IKK is also particularly relevant because TNF-α–activated IKK phosphorylates p/CIP, enabling its translocation and selective association with nuclear NF-κB. Interestingly, the pattern of p/CIP residues phosphorylated by TNF-α–activated IKK is different from that induced by other stimuli that activate p/CIP (28). These different patterns of phosphorylation determine to which specific transcription factors activated p/CIP binds in the nucleus. It can therefore be imagined that a particular stimulation activates a specific kinase, which in turn phosphorylates p/CIP on residues that not only mediate its nuclear translocation, but also determine the transcription factor to which it will bind.


Controlling the subcellular localization of inducible transcription factors and their coactivators is a complex process that is specific to each signaling pathway. In this review, we have tried to highlight common themes that unite these otherwise diverse processes. For example, many inducible transcription factors are sequestered in the cytoplasm by retention factors. Retention is achieved by either masking the transcription factor’s NLS, anchoring the factor to cytoplasmic structures, or by promoting its degradation (Table 1). In addition to sequestering transcription factors from target genes, retention factors also appear to selectively localize inducible transcription factors such that proximity to the relevant stimulus is maintained. Furthermore, as illustrated by Nrf2 and p/CIP, retention factors often localize their targets just outside of the nucleus or may tether them to cytoplasmic proteins that may facilitate their nuclear translocation. Another common characteristic of cytoplasmic retention factors is their ability to selectively respond to relevant inductive signals. Therefore, upon activation, not only is the transcription factor modified to lose its affinity for its retention factor, but the retention factor can also undergo changes that lead to loss of its affinity for its target.

Table 1. Mechanisms of cytoplasmic sequestration, as illustrated by the localization of the discussed transcription factors in unstimulated cells. The respective mechanisms of retention are listed, in addition to the proposed retention factors responsible for the cytoplasmic sequestration. Coactivators are indicated by an asterisk.

In the case of NF-κB and Smads, cytoplasmic retention is accompanied by nucleocytoplasmic shuttling. As more is understood about other inducible transcription factors, it may be revealed that constitutive shuttling is a widespread phenomenon. However, it can be inferred from the pathways discussed in this review that shuttling occurs only when an inductive signal is diffusely localized (i.e., there are multiple stimulatory inputs) or is localized near the cell membrane, as occurs with many receptor-mediated signals. Under these conditions, shuttling may allow the transcription factor to constantly survey the cytoplasm for stimuli and then rapidly relay that signal to the nucleus. Conversely, when the inductive signal is generated in a focused location, as is the case for Nrf2 and Opi1p, it would be advantageous to anchor the transcription factor close to the origin of this signal. Therefore, it is likely that the static or dynamic localization of transcription factors might depend on the location of the stimulus to which they respond.

It is now clear that regulating gene expression through controlled localization involves much more than simply sequestering effector transcription factors away from target genes until the time of induction (Fig. 1). Inactive transcription factors may be fully retained in the cytoplasm or may shuttle constantly between the nucleus and the cytoplasm (Fig. 2). In addition, coactivators may likewise be sequestered in the cytoplasm until the time of induction. Finally, sequestered factors may be localized near the source of induction and/or near the nucleus. Each, or a combination, of these features ensures that inducible transcription is inactive in the absence of signal and is readily activated in the presence of signal. With these features in mind, other well-known pathways in which the subcellular localization of inducible transcription factors is modulated should be reexamined. For example, the IL-6– and γ interferon (IFN-γ)–dependent activation of STATs (signal transducers and activators of transcription), which has recently been implicated in inflammatory bowel disease, is dependent on the nuclear translocation of STAT proteins (31). Future studies of these signaling pathways that consider the principles discussed above may ultimately enable us to modulate these pathways for therapeutic purposes.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
View Abstract

Stay Connected to Science Signaling

Navigate This Article