
Opinion A post-translational modification code for transcription factors: sorting through a sea of signals Be´ re´ nice A. Benayoun1,2,3 and Reiner A. Veitia1,2,3 1 Institut Jacques Monod, Baˆ timent Buffon, 15 Rue He´le` ne Brion, Paris Cedex 13, France 2 CNRS-UMR 7592, Institut Jacques Monod, Baˆ timent Buffon, 15 Rue He´le` ne Brion, Paris Cedex 13, France 3 Universite´ Paris Diderot, Paris 7, Paris, France Cellular responses to environmental or physiological The typical textbook signaling pathway is activated by cues rely on transduction pathways that must ensure the binding of a ligand to a transmembrane receptor, which discrimination between different signals. These cas- in turn modulates the activity of cytoplasmic transducers. cades ‘crosstalk’ and lead to a combinatorial regulation. After one or more steps of signal amplification through This often results in different combinations of post- these transducers, an endpoint is often the activation or translational modifications (PTMs) on target proteins, inhibition of specific transcription factors (TFs), which which might act as a molecular barcode. Although ultimately modulate expression of a specific set of genes appealing, the idea of the existence of such a code for [3]. Many transducers in signaling pathways are post- transcription factors is debated. Using general argu- translational modification (PTM) enzymes, the substrates ments and recent evidence, we propose that a PTM code of which can be specific amino acid residues of TFs is not only possible but necessary in the context of embedded within appropriate consensus sequences. The transcription factors regulating multiple processes. position of TFs in signaling networks makes them, at least Thus, the coding potential of PTM combinations should in theory, good candidates to act as integrators of various both provide a further layer of information integration stimuli, especially because the activity of specific TFs can from several transduction pathways and warrant highly be simultaneously modulated by several signaling path- specific cellular outputs. ways. The existence of a histone PTM code in which expres- Introduction sional information would be encoded as combinatorial Biological responses to environmental or physiological nucleosomal PTMs has been under debate for years. How- cues rely on signal transduction pathways that must ever, as shown later, there is a growing body of evidence ensure discrimination among a wide panoply of signals. supporting the existence of some sort of code [4–9]. The These pathways must also enable discrimination between ongoing debate about the existence of a PTM code regulat- noise, owing to random fluctuations of signals, and a true ing non-histone proteins is more recent and far from being input. Signal transduction will ultimately induce modu- settled [5,10–13]. In particular, in a recent opinion paper, lation of the cellular proteome in response to a molecular Sims and Reinberg [10] have challenged the relevance of effector (e.g. a hormone) or another type of signal (e.g. the concept of PTM code outside the realm of histones. stress). A particular cell can be subjected to multiple Here, we present a different perspective. We argue that the environmental and physiological signals, and their integ- existence of such a code is not only possible but necessary ration is mandatory to elicit a coherent response. Until in the case of at least some (if not all) TFs, and we provide recently, signaling cascades were perceived as rather lin- evidence supporting our claims. ear pathways. Now, these cascades have been shown to communicate, to ‘crosstalk’, ensuring a combinatorial regu- The ‘histone code’ paradigm lation. Such interconnections between pathways form net- Histones are the recipients of a large panel of PTMs, works, which elicit integrated responses in a way that can including serine/threonine phosphorylation, lysine acety- be assimilated to a coherent code [1,2]. By definition, a code lation, lysine or arginine methylation, glutamine ADP- is a system of elements that are linked by rules so as to ribosylation, and conjugation to ubiquitin and ubiquitin- convert pieces of information (for instance, stimuli) into like proteins such as small ubiquitin-related modifiers other forms of representation or response (for instance, a (SUMO) [4]. Distinct combinations of post-translational cellular outcome or a biological response; http://en.wikipe- modifications of histone tails have been proposed to dia.org/wiki/code). Using a code, a limited set of elements function as a molecular ‘code’, which is able to modulate can be assorted into combinations and specify different the chromatin transcriptional status. This adds a layer of meanings or outputs (Box 1). complexity to gene-expression regulation [14].These modifications are set, maintained or removed by an ever-expanding number of enzymes [4]. Interestingly, Corresponding author: Veitia, R.A. ([email protected]). high-throughput analyses have shown that combinations 0962-8924/$ – see front matter ß 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.tcb.2009.02.003 Available online 26 March 2009 189 Opinion Trends in Cell Biology Vol.19 No.5 of histone PTMs cluster in regions of similar transcrip- Positive PTM interactions occur, for instance, in the case tional activity. For instance, acetylation of lysines of the Forkhead TF FOXO1 (Box 2). Indeed, phosphoryl- H2AK7, H3K9, H3K14, H3K18, H4K5 and H4K12 cor- ation of one particular serine of FOXO1 by protein kinase B relates positively with transcription at different loci [15], (PKB, also known as AKT) creates a casein kinase I (CKI) whereas H3K9 lysine methylation, which is often accom- phosphorylation consensus, which, after modification, cre- panied by histone hypo-acetylation, correlates with tran- ates another CKI consensus. The presence of these PTMs scriptional repression [4,6]. These correlations indicate eventually leads to a robust and quick nuclear exclusion that specific PTM marks somehow ‘encode’ the transcrip- [18]. tional status of chromatin, at least in a binary way. The The existence of many proteins, besides histones, dis- existence of a ‘histone code’ is now supported by a large playing multiple PTM isoforms therefore calls for an exten- body of experimental evidence. Differentially modified sion of the histone code paradigm. Naturally, the use of the residues of histone tails would provide interaction sur- term ‘PTM code’ for non-histone proteins does not imply faces recognized by proteins (adaptors) which, upon that ‘one particular site modification equals one particular binding, would affect chromatin structure and gene downstream event’. A PTM code is more likely to be expression accordingly [4]. combinatorial and context-dependent in nature. For The fact that the same histone PTMs can be interpreted instance, phosphorylation of Thr28, Ser193 and Ser258 differently according to the cellular context, which is one of of FOXO4, stimulated by Insulin and/or IGF1 (Insulin/ the arguments of Sims and Reinberg against a PTM code, IGF1) signaling, promotes nuclear exclusion but stress- does not invalidate the existence of a histone code, at least induced phosphorylation of Thr447 and Thr451 by Jun N- in its wider sense. In our biological framework, we must terminal kinase (JNK) is sufficient to promote nuclear accept that the cell is somehow able to grasp the right import, whatever the PTM status of Thr28, Ser193 and ‘contextual meaning’ of the PTM-encoded message, how- Ser258 [11]. However, in any given environmental or ever complicated and context-dependent the decoding pro- physiological context, cellular outputs should be similar, cess is. so particular PTM protein isoforms emerging in response to this context might help encode these ‘stereotyped’ Extending the ‘PTM code’ hypothesis to other proteins responses. A recent phospho-proteomic study in HeLa cells has The existence of a combinatorial code would stem from revealed that phosphorylation is a major modification of the very nature of signal transduction, which is also com- a vast repertoire of cellular proteins (>6000 phosphoryl- binatorial. Thus, to ensure signal discrimination, the ation sites identified in >2000 proteins), and that a good response to a particular signal must be encoded by the proportion of these modifications are modulated in triggering of a combination of signaling cascades. This response to extra-cellular signals (i.e. 14% of those phos- combinatorial use of signaling pathways can be regarded phorylations changed at least twofold in abundance after as an evolutionary strategy (and necessity), thereby EGF treatment). Interestingly, at least 46 known tran- increasing the number of possible outputs using a scriptional regulators were found among those phospho- restricted number of cascades. This should lead to the proteins [16]. It is tempting to propose that extensive PTM emergence of some sort of ‘pathway code’ (Figure 1a,b). of proteins, leading to several co-existing isoforms, com- Just try to figure out how a cell can distinguish between pensates for the relative paucity of genes in vertebrate extracellular signals S1, S2...Sn, when they all activate genomes by enabling a single gene product to accomplish the same modifying enzyme (e.g. PKA or AKT): clearly, different roles according to their PTMs. Interestingly, other concurrent pathways must be specifically activated PTMs have been found to cooperate with one another, or inhibited conjointly according to the signals to ensure either positively
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages9 Page
-
File Size-