Tunable signal processing through modular control of transcription factor translocation (original) (raw)

Science. Author manuscript; available in PMC 2013 Aug 19.

Published in final edited form as:

PMCID: PMC3746486

NIHMSID: NIHMS480773

Nan Hao

1Harvard University Faculty of Arts and Sciences Center for Systems Biology, Cambridge, MA 02138, USA

2Howard Hughes Medical Institute, Department of Molecular and Cellular Biology, and Department of Chemistry and Chemical Biology, Harvard University, Cambridge, MA 02138, USA

Bogdan A. Budnik

1Harvard University Faculty of Arts and Sciences Center for Systems Biology, Cambridge, MA 02138, USA

Jeremy Gunawardena

3Department of Systems Biology, Harvard Medical School, Boston, MA 02115, USA

Erin K. O’Shea

1Harvard University Faculty of Arts and Sciences Center for Systems Biology, Cambridge, MA 02138, USA

2Howard Hughes Medical Institute, Department of Molecular and Cellular Biology, and Department of Chemistry and Chemical Biology, Harvard University, Cambridge, MA 02138, USA

1Harvard University Faculty of Arts and Sciences Center for Systems Biology, Cambridge, MA 02138, USA

2Howard Hughes Medical Institute, Department of Molecular and Cellular Biology, and Department of Chemistry and Chemical Biology, Harvard University, Cambridge, MA 02138, USA

3Department of Systems Biology, Harvard Medical School, Boston, MA 02115, USA

Abstract

Signaling pathways can induce different dynamics of transcription factor (TF) activation. We explored how TFs process signaling inputs to generate diverse dynamic responses. The budding yeast general stress responsive TF Msn2 acted as a tunable signal processor that could track, filter, or integrate signals in an input dependent manner. This tunable signal processing appears to originate from dual regulation of both nuclear import and export by phosphorylation, as mutants with one form of regulation sustained only one signal processing function. Versatile signal processing by Msn2 is crucial for generating distinct dynamic responses to different natural stresses. Our findings reveal how complex signal processing functions are integrated into a single molecule and provide a guide for the design of TFs with “programmable” signal processing functions.

Many transcription factors (TFs) display diverse activation dynamics in response to various external stimuli (1–4). To investigate how TFs process upstream signals, we studied the S. cerevisiae general stress responsive TF Msn2 (5). In the absence of stress, Msn2 is phosphorylated by Protein Kinase A (PKA) and localized to the cytoplasm; in response to stress, Msn2 is dephosphorylated and translocates to the nucleus where it induces gene expression (5).

Natural stresses elicit highly variable dynamics of Msn2 nuclear translocation (Fig. 1A) (6, 7), which are thought to result from oscillatory signaling inputs (presumably PKA activity) (8). To study how Msn2 processes oscillatory PKA inputs, we used an engineered yeast strain (6) carrying mutations in all three PKA isoforms that enable selective inhibition of PKA activity by a cell-permeable inhibitor, 1-NM-PP1 (9). We used this synthetic system and a microfluidics platform (10) mounted on a microscope to produce oscillatory inputs of PKA inhibition and monitored translocation of Msn2 to the nucleus. The input amplitude was chosen on the basis of the steady state amount of Msn2 nuclear localization in response to sustained inputs: high amplitude input (3 μM 1-NM-PP1) led to maximal nuclear localization of Msn2; whereas low amplitude input (0.2 μM 1-NM-PP1) induced an intermediate amount of nuclear localization (Fig. 1B, black circles). The pulse duration of oscillatory input was selected on the basis of duration of pulsatile Msn2 nuclear bursts in the physiological response to glucose limitation (6). With high amplitude oscillatory input, each input pulse induced a high amount of nuclear localization (Fig. 1C, left). In contrast, oscillatory input with low amplitude barely elicited any localization responses, although sustained input with the same amplitude led to a half-maximal amount of nuclear localization (Fig. 1C, right). Therefore, Msn2 filters temporal fluctuations of the input in an amplitude-dependent manner such that it tracks high amplitude inputs, but responds in a limited manner to low amplitude signals.

An external file that holds a picture, illustration, etc. Object name is nihms480773f1.jpg

Tunable signal processing behaviors of Msn2

(A) Illustration of the distinct single cell dynamic responses of Msn2 to various stresses. (B) Steady state abundance of Msn2 in the nucleus in response to various concentrations of 1-NM-PP1. In response to each concentration of 1-NM-PP1, Msn2 exhibited uniform and stable nuclear localization in single cells and did not exhibit stochastic fluctuations as observed in natural stress responses. Open circles: responses to different concentrations of 1-NM-PP1; closed circles: responses to 3 μM and 0.2 μM 1-NM-PP1, which are used as high and low amplitude inputs, respectively, for the following analysis. AU – arbitrary unit. (C) Averaged single-cell time traces of Msn2 nuclear translocation (Lower panels: n: ~50 cells; error bar: single-cell variances) in response to oscillatory inputs with high and low amplitudes (Upper panels). Left: high-amplitude input produced by 3 μM 1-NM-PP1; right: low amplitude input produced by 0.2 μM 1-NM-PP1. Pulse duration - 3 min; pulse interval - 2 min. To emphasize the fact that 3 μM 1-NM-PP1 elicits a steady state response that is approximately twice the response elicited by 0.2 μM 1-NM-PP1, the y-axes of the upper panels are not presented on a linear scale.

To understand how Msn2 translates signaling inputs into different translocation responses, we characterized Msn2 phosphorylation, which controls nuclear translocation (5, 11). We detected phosphorylation of eight PKA consensus sites, primarily located within the nuclear export signal (NES) and nuclear localization signal (NLS) domains (11) (Fig. S1). Two sites in the NES (S288, S304) and four sites in the NLS (S582, S620, S625, S633) were functionally important for regulation of nuclear transport (Fig. S2).

To intuitively understand the behaviors of the translocation system, we conducted a steady-state analysis that incorporated the separation of timescales for nuclear transport and phosphorylation. For simplicity, we represented regulation of nuclear export and import by phosphorylation of one site in the NES and a second site in the NLS, acting independently of each other (Fig. 2A). We assumed that the slowest timescales occur for nuclear import when the NLS is phosphorylated (kin) and for nuclear export when the NES is unphosphorylated (kout). In contrast, when the NLS is unphosphorylated or when the NES is phosphorylated, nuclear import and export, respectively, both were assumed to occur on faster timescales (kin′, kout′) (12). This scheme gives four phosphoforms with distinct combinations of transport rates (Fig. 2B). Finally, we assumed that phosphorylation and dephosphorylation are fast relative to the translocation timescales (13, 14), so that we could treat transitions between the four phosphoforms, which are triggered by input of PKA inhibition, as effectively instantaneous. PKA and Msn2 phosphatases localize to both the cytoplasm and the nucleus in yeast (15–17), so Msn2 can be phosphorylated and dephosphorylated in both compartments.

An external file that holds a picture, illustration, etc. Object name is nihms480773f2.jpg

A theoretical analysis of transcription factor translocation

(A) Nomenclature used to define the status of phosphorylation and localization of the TF: the 1st “P/U” - the NES is phosphorylated (P) or unphosphorylated (U); the 2nd “P/U” - the NLS is phosphorylated (P) or unphosphorylated (U); “c/n” - in the cytoplasm (c) or nucleus (n). (B) Phosphorylation states determine the rate constants of nucleocytoplasmic transport. Unphosphorylated (U) or phosphorylated (P) NES has slow (dashed line) or fast (solid line) nuclear export rates (kout, kout′), respectively; unphosphorylated (U) or phosphorylated (P) NLS has fast (solid line) or slow (dashed line) nuclear import rates (kin′, kin), respectively. Thus, each phosphoform has a specific combination of nuclear import and export rates. (C) Schematic of the translocation model. Left: schematic of WT and phosphosite mutants; right: model structures and reaction flows (grey arrows) in response to strong or weak inputs and input removal. 1st row: WT; 2nd row: NLS 4A - S582A, S620A, S625A, S633A; 3rd row: NLS 4E – S582E, S620E, S625E, S633E; 4th row: NES 2A – S288A, S304A. We did not specifically study the case in which the NES sites are constitutively phosphorylated because Ser-to-Glu mutants of the NES sites behaved similarly to Ser-to-Ala mutants, suggesting that Glu cannot mimic phosphorylation on NES sites (Fig. S2A). (D) Predicted responses to various dynamic inputs: 1st column: oscillatory high-amplitude input; 2nd column: oscillatory input with varied amplitudes; 3rd column: input fluctuating between high and low amplitudes. Black: responses of WT; blue: NLS 4A; green: NLS 4E; red: NES 2A. The ranges of input timescales necessary to generate the predicted responses are determined by the fast and slow timescales of transport rates and are listed above each column. Model output was generated by a steady-state analysis of the translocation system (Supplementary Materials).

For purposes of illustration, we used a weak and a strong input to represent the amplitude of PKA inhibition (Fig. 2C). Because sites in the NLS are more preferred for PKA phosphorylation than those in the NES (11, 18) (Fig. S3), we assumed that weak input (partial inhibition of PKA) would lead to dephosphorylation of only the NES, and the NLS phosphorylated form (U_Pc) would then go to the nucleus with a slow import rate (kin). In contrast, strong input would lead to dephosphorylation of both the NES and NLS and the fully unphosphorylated form (U_Un) would be transported into the nucleus with a fast rate (kin′). Upon input removal, the NES and NLS are re-phosphorylated and the doubly-phosphorylated form (P_Pn) is expected to be exported with a fast export rate (kout′) (Fig. 2C, 1st row). In accordance with this analysis, strong input (3 μM 1-NM-PP1) led to rapid Msn2 translocation whereas weak input (0.2 μM 1-NM-PP1) resulted in slower translocation, and export is rapid when PKA inhibition is removed (Fig. 3AB, WT).

An external file that holds a picture, illustration, etc. Object name is nihms480773f3.jpg

Distinct signal processing by WT, NLS and NES phosphosite mutants of Msn2

(A) Averaged single-cell time traces of Msn2 nuclear translocation in response to sustained inputs with low (0.2 μM 1-NM-PP1, solid triangles) or high (3 μM 1-NM-PP1, solid circles) amplitudes. Inputs were applied at time point zero. (B) Averaged single-cell time traces of Msn2 nuclear translocation in response to removal of high-amplitude input (3 μM 1-NM-PP1). (C) Averaged single-cell time traces of Msn2 nuclear translocation in response to oscillatory high-amplitude (3 μM 1-NM-PP1) inputs. Pulse duration – 3 min; pulse interval – 4 min. (D) Time traces of Msn2 nuclear translocation in response to oscillatory inputs with a mixture of low (0.2 μM 1-NM-PP1) and high (3 μM 1-NM-PP1) amplitude pulses. (E) Time traces of Msn2 nuclear translocation in response to input fluctuating between high (3 μM 1-NM-PP1) and low (0.2 μM 1-NM-PP1) amplitudes. For (A)–(E), data points are average single-cell time traces (n: ~50 cells; error bar: single-cell variances). The simple model in Fig. 2 has been fit to the time trace data in this figure and the solid lines in (A)–(E) are model fitting results (see Supplemental Material - “Model parameters are constrained by experimental data” for details). The dependence of the responses on the timescales of input and transport rates is presented in Supplemental Material - “The relationship between timescales of input and timescales of transport rates”.

We then analyzed how Msn2 might respond to oscillatory inputs. In response to a strong oscillatory input, Msn2 would go in and out of the nucleus with import and export rates (kin′, kout′) that are fast relative to the input pulse duration and inter-pulse interval. Hence, Msn2 responded fully to each pulse and tracked the input dynamics (model: Fig. 2D, 1st row, left; data: Fig. 3C, WT). In response to a weak oscillatory input, Msn2 would enter the nucleus with a slow import rate (kin) relative to the timescale of the input pulse, and therefore only a small amount of Msn2 entered the nucleus, effectively filtering out low amplitude signals (model: Fig. 2D, 1st row, middle; data: Fig. 3D, WT). In response to an input fluctuating between high and low amplitudes, because Msn2 would go out of the nucleus with a slow export rate (kout) relative to the timescale of the inter-pulse intervals, Msn2 was not fully exported during the interval and integrated the input fluctuations (model: Fig. 2D, 1st row, right; data: Fig. 3E, WT).

To further test the model and explore the influence of regulation of both import and export on signal processing, we studied cases in which only nuclear import or export, but not both, was regulated by phosphorylation because the functional phosphosites within the NES or NLS were mutated (Fig. 2C and D, Fig. 3).

In the case in which the NLS sites could not be phosphorylated (NLS 4A), Msn2 would enter the nucleus with a constitutively fast import rate (kin′) and go out of the nucleus with a fast export rate (kout′) upon input removal (Fig. 2C, 2nd row; data: Fig. 3A–B, NLS 4A). Hence, in response to oscillatory inputs with high amplitude, low amplitude, or fluctuating between high and low amplitude, Msn2 would have fast import and export rates and fully entered the nucleus during a pulse and exited the nucleus during inter-pulse intervals (model: Fig. 2D, 2nd row; data: Fig. 3C–E, NLS 4A).

If NLS sites were mutated to mimic constitutive phosphorylation (NLS 4E), Msn2 would enter the nucleus with a constitutively slow import rate (kin) (Fig. 2D, 3rd row; data: Fig. 3A, NLS 4E). In response to oscillatory inputs, when the input duration is short relative to the timescale of the slow import rate, Msn2 went into the nucleus slowly and reached low concentrations (model: Fig. 2D, 3rd row; data: Fig. 3C–E, NLS 4E).

If NES sites cannot be phosphorylated (NES 2A), Msn2 would exit the nucleus with a constitutively slow export rate (kout) (Fig. 2D, 4th row; data: Fig. 3B, NES 2A). In response to oscillatory inputs with high amplitude, low amplitude or fluctuating between high and low amplitude, when the interval is short relative to the time scale of the slow export rate, Msn2 would have a slow export rate, could not fully exit the nucleus during intervals, and therefore integrated responses to rapidly changing inputs (model: Fig. 2E, 4th row; data: Fig. 3C–E, NES 2A). In summary, NLS 4A, NLS 4E, or NES 2A “tracks”, “filters” or “integrates” the oscillatory inputs, respectively, whereas WT Msn2 exhibits a combination of all these processing behaviors.

To study the processing of natural stress signals, we monitored WT and mutant Msn2 translocation in response to different stresses (Fig. 4A–C and S5). We also monitored the dynamics of WT Msn2-mCherry and mutant Msn2-YFP expressed together in the same cells – this allowed us to directly compare the responses of WT and mutant Msn2 to the same stochastic input signals triggered by natural stress (Fig. 4D and S6). Glucose limitation induced sporadic pulses of rapid nuclear localization of WT Msn2 with frequency regulated by stress intensity; osmotic stress elicited a single pulse of nuclear accumulation; and oxidative stress led to sustained nuclear localization (Fig. 4A–C, and S5, WT). NLS 4A, which tracks the inputs, was more responsive to inputs and exhibited a high frequency of small rapid bursts of nuclear translocation (Fig. 4D, 1st row), and thus produced similar response frequency to low and high levels of glucose limitation (Fig. 4A–C, 2nd row, panels marked with blue asterisks; Fig. S5D, left, blue circles). By contrast, NLS 4E filtered out the sporadic translocation bursts in response to glucose limitation (Fig. 4D, 2nd row) and therefore exhibits similar dynamics to both glucose limitation and osmotic stress (Fig. 4A–C, 3rd row, panels marked with green asterisks). NES 2A integrated the sporadic bursts in response to strong glucose limitation (Fig. 4D, 3rd row) and exhibited prolonged nuclear accumulation, similar to that of oxidative stress responses (Fig. 4A–C, 4th row, panels marked with red asterisks). Consistent with the analysis of artificial inputs, WT Msn2 and the mutants differed in how they processed signaling inputs triggered by natural stresses and therefore generate different responses. WT Msn2 with dual regulation of nuclear import and export generates distinct translocation responses to different stress conditions, whereas the mutants that have only one mode of nuclear transport regulation fail to fully differentiate the different stresses into distinct translocation outputs. Cells may use these diverse TF translocation patterns to induce distinct gene expression programs (6), or to elicit different levels of noise in single cell responses, both of which might be beneficial for survival under stress conditions.

An external file that holds a picture, illustration, etc. Object name is nihms480773f4.jpg

Distinct responses of WT, NLS, and NES phosphosite mutants of Msn2 to natural stresses

Single-cell responses of WT, NLS 4A, NLS 4E, and NES 2A to glucose limitation (A), osmotic stress (B), and oxidative stress (C) (n: ~50 cells, each stress condition). Representative single-cell time traces of Msn2 nuclear translocation are shown. Asterisks emphasize the conditions under which the mutants fail to distinguish two different stresses. Quantification of the time traces is presented in Fig. S4. (D) Time traces of WT Msn2-mCherry and mutant Msn2-YFP, monitored in the same cells, in response to glucose limitation (black – WT, blue – NLS 4A, green – NLS 4E, red – NES 2A). More single-cell traces are shown in Fig. S5.

Nucleocytoplasmic translocation of many mammalian TFs, such as NFATs, STATs, and Smads (19–21), is controlled by regulation of both their nuclear localization and nuclear export signals. Hence, the proposed dual regulation mechanism may represent a general mechanism for shaping the dynamic behaviors of these TFs. Complex signal processing behaviors can be achieved by signaling circuits comprised of multiple molecules (22–29). We reveal a single TF molecule can also mediate sophisticated signal processing functions by assembling independent functional modules. These functions are “tunable” by phosphorylation at multiple sites in each module and “programmable” by mutating or reassembling functional modules.

Supplementary Material

Supps and Figs

Acknowledgments

We thank A. Murray, P. Cluzel, S. Ramanathan, B. Stern, and M. Rust for comments on the manuscript, and S. Mukherji, X. Zhou, A. S. Hansen and other members of the O’Shea lab for helpful discussions. J.G. is supported by NIH R01 GM081578. E.K.O. is an Investigator of the Howard Hughes Medical Institute.

References

1. Werner SL, Barken D, Hoffmann A. Stimulus specificity of gene expression programs determined by temporal control of IKK activity. Science. 2005 Sep 16;309:1857. [PubMed] [Google Scholar]

2. Batchelor E, Loewer A, Mock C, Lahav G. Stimulus-dependent dynamics of p53 in single cells. Mol Syst Biol. 2011 May 10;7:488. [PMC free article] [PubMed] [Google Scholar]

3. Lee TK, et al. A noisy paracrine signal determines the cellular NF-kappaB response to lipopolysaccharide. Sci Signal. 2009;2:ra65. [PMC free article] [PubMed] [Google Scholar]

5. Gorner W, et al. Nuclear localization of the C2H2 zinc finger protein Msn2p is regulated by stress and protein kinase A activity. Genes Dev. 1998 Feb 15;12:586. [PMC free article] [PubMed] [Google Scholar]

6. Hao N, O’Shea EK. Signal-dependent dynamics of transcription factor translocation controls gene expression. Nat Struct Mol Biol. 2012 Jan;19:31. [PMC free article] [PubMed] [Google Scholar]

7. Cai L, Dalal CK, Elowitz MB. Frequency-modulated nuclear localization bursts coordinate gene regulation. Nature. 2008 Sep 25;455:485. [PMC free article] [PubMed] [Google Scholar]

8. Garmendia-Torres C, Goldbeter A, Jacquet M. Nucleocytoplasmic oscillations of the yeast transcription factor Msn2: evidence for periodic PKA activation. Curr Biol. 2007 Jun 19;17:1044. [PubMed] [Google Scholar]

9. Bishop AC, et al. A chemical switch for inhibitor-sensitive alleles of any protein kinase. Nature. 2000 Sep 21;407:395. [PubMed] [Google Scholar]

10. Hersen P, McClean MN, Mahadevan L, Ramanathan S. Signal processing by the HOG MAP kinase pathway. Proc Natl Acad Sci U S A. 2008 May 20;105:7165. [PMC free article] [PubMed] [Google Scholar]

11. Gorner W, et al. Acute glucose starvation activates the nuclear localization signal of a stress-specific yeast transcription factor. EMBO J. 2002 Jan 15;21:135. [PMC free article] [PubMed] [Google Scholar]

12. Komeili A, O’Shea EK. Roles of phosphorylation sites in regulating activity of the transcription factor Pho4. Science. 1999 May 7;284:977. [PubMed] [Google Scholar]

13. Shulga N, et al. In vivo nuclear transport kinetics in Saccharomyces cerevisiae: a role for heat shock protein 70 during targeting and translocation. J Cell Biol. 1996 Oct;135:329. [PMC free article] [PubMed] [Google Scholar]

14. Adams JA, Taylor SS. Phosphorylation of peptide substrates for the catalytic subunit of cAMP-dependent protein kinase. J Biol Chem. 1993 Apr 15;268:7747. [PubMed] [Google Scholar]

15. Huh WK, et al. Global analysis of protein localization in budding yeast. Nature. 2003 Oct 16;425:686. [PubMed] [Google Scholar]

16. Tudisca V, et al. Differential localization to cytoplasm, nucleus or P-bodies of yeast PKA subunits under different growth conditions. Eur J Cell Biol. 2010 Apr;89:339. [PubMed] [Google Scholar]

17. Tkach JM, et al. Dissecting DNA damage response pathways by analysing protein localization and abundance changes during DNA replication stress. Nat Cell Biol. 2012 Sep;14:966. [PMC free article] [PubMed] [Google Scholar]

18. Neuberger G, Schneider G, Eisenhaber F. pkaPS: prediction of protein kinase A phosphorylation sites with the simplified kinase-substrate binding model. Biol Direct. 2007;2:1. [PMC free article] [PubMed] [Google Scholar]

19. Okamura H, et al. Concerted dephosphorylation of the transcription factor NFAT1 induces a conformational switch that regulates transcriptional activity. Mol Cell. 2000 Sep;6:539. [PubMed] [Google Scholar]

20. Reich NC, Liu L. Tracking STAT nuclear traffic. Nat Rev Immunol. 2006 Aug;6:602. [PubMed] [Google Scholar]

21. Chen X, Xu L. Mechanism and regulation of nucleocytoplasmic trafficking of smad. Cell Biosci. 2011;1:40. [PMC free article] [PubMed] [Google Scholar]

22. Levy S, Kafri M, Carmi M, Barkai N. The competitive advantage of a dual-transporter system. Science. 2011 Dec 9;334:1408. [PMC free article] [PubMed] [Google Scholar]

23. Longo DM, Hoffmann A, Tsimring LS, Hasty J. Coherent activation of a synthetic mammalian gene network. Syst Synth Biol. 2010 Mar;4:15. [PMC free article] [PubMed] [Google Scholar]

25. Locke JC, Young JW, Fontes M, Hernandez Jimenez MJ, Elowitz MB. Stochastic pulse regulation in bacterial stress response. Science. 2011 Oct 21;334:366. [PMC free article] [PubMed] [Google Scholar]

26. O’Shaughnessy EC, Palani S, Collins JJ, Sarkar CA. Tunable signal processing in synthetic MAP kinase cascades. Cell. 2011 Jan 7;144:119. [PMC free article] [PubMed] [Google Scholar]

27. Behar M, Hoffmann A. Understanding the temporal codes of intra-cellular signals. Curr Opin Genet Dev. 2010 Dec;20:684. [PMC free article] [PubMed] [Google Scholar]

28. Bennett MR, et al. Metabolic gene regulation in a dynamically changing environment. Nature. 2008 Aug 28;454:1119. [PMC free article] [PubMed] [Google Scholar]

29. Bashor CJ, Helman NC, Yan S, Lim WA. Using engineered scaffold interactions to reshape MAP kinase pathway signaling dynamics. Science. 2008 Mar 14;319:1539. [PubMed] [Google Scholar]

30. Zaman S, Lippman SI, Schneper L, Slonim N, Broach JR. Glucose regulates transcription in yeast through a network of signaling pathways. Mol Syst Biol. 2009;5:245. [PMC free article] [PubMed] [Google Scholar]

31. Filonov GS, et al. Bright and stable near-infrared fluorescent protein for in vivo imaging. Nat Biotechnol. 2011 Aug;29:757. [PMC free article] [PubMed] [Google Scholar]

32. Cox J, Mann M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat Biotechnol. 2008 Dec;26:1367. [PubMed] [Google Scholar]

33. Ong SE, Mann M. A practical recipe for stable isotope labeling by amino acids in cell culture (SILAC) Nat Protoc. 2006;1:2650. [PubMed] [Google Scholar]