Shining a light on post-translational modification

Abstract

Post-translational modification, such as phosphorylation or glycosylation, provides a mechanism for increasing the diversity of protein structures in the cell and regulating biological activity. In addition, such modifications may result in the localization of proteins to specific cellular organelles, with incorrect targeting being associated with a number of diseases. The simplest strategy to identify the functional importance of post-translational modifications is to use mutagenesis methods to replace the residue that is post-translationally modified by one that cannot undergo the relevant chemical transformation. Merely causing “loss of function” does not, however, address questions concerning how cellular function depends on the timing of post-translational changes and∕or the movement of modified proteins between organelles. The recent demonstration that genetically encoded “photocaged” proteins can be employed to resolve such issues therefore represents an exciting advance in this research area, and is an elegant illustration of the power of combining the power of chemical synthesis and methods for manipulating the biological machinery of protein synthesis.

The concept of using “caged” species to initiate cellular events was first realized in the creation of small molecules that could be used to release ATP with temporal and spatial control (McCray et al., 1980). In extending this strategy to proteins, protecting groups must be attached to specific side chains so as to cause a loss of function. Their removal by irradiation with light “activates” the protein, thereby allowing it to perform its role in cellular metabolism [Fig. 1A]. The use of photochemical reactions to release the “caged” protein offers several advantages in studies of biological systems. First, irradiation is a non-invasive procedure that initiates biochemical function at a specific time because the reaction that releases the active protein typically takes place in less than a few seconds. Second, photoactivation can be performed at a well-defined spatial location using multiple-photon confocal microscopy (Mayer and Heckel, 2006). On the other hand, the need to introduce (or synthesize) the photoactivatible protein in the cell is a drawback of this strategy, presenting a number of technical challenges. These include (i) the development of simple and efficient synthetic routes for preparing “non-natural” amino acids containing unreactive, albeit photoactivatible, functional groups; (ii) the identification of photochemical reactions used to “release” the protein from its non-functional “caged” state, which ideally should have a high quantum yield and employ light that does not damage the cell; and (iii) the development of cell-based strategies for the site-specific incorporation of non-natural amino acids into the protein of interest using the translational machinery of the cell (Xie and Schultz, 2004).

Figure 1. (A) Photoactivation of a “caged” protein.

A functionally important residue, in this case serine, is replaced by a genetically encoded analog in which the side chain hydroxyl is protected by the DMNB group. Subsequent irradiation then releases the serine side chain to give the active protein, which can then be phosphorylated. (B) Schematic representation of the phosphorylation sites in PHO4. Color-coding indicates the different functional domains of the transcription factor. Phosphorylation sites are shown together with the number of the serine residue that is post-translationally modified. (C) Photochemical reaction of DMNB-serine to yield the active protein and a nitrosoaldehyde by-product.

In order to demonstrate the utility of genetically encoded “photocaged” proteins in probing cell biology, Lemke et al. (2007) studied the correlation of cellular function and post-translational modification in Saccharomyces cerevisiae PHO4, a transcription factor that plays a critical role in the signaling cascade used by yeast to grow at different concentrations of exogenous phosphate (Oshima, 1997). The removal of PHO4 from the nucleus prevents transcription of genes involved in mediating the yeast response to phosphate starvation. When phosphate is present at high concentration, PHO4 becomes phosphorylated by the PHO80∕PHO85 cyclin∕cyclin-dependent kinase complex at a number of serine residues (Kaffman et al., 1994). Moreover, experiments using PHO4 variants in which specific serine residues were replaced by alanine showed that each site plays a separable role in regulating PHO4 activity, with the presence of multiple sites resulted in overlapping levels of control (Komeili and O’Shea, 1996) [Fig. 1B]. For example, phosphorylation at both Ser-114 and Ser-128 is required for nuclear export of the post-translationally modified PHO4 into the cytoplasm, while phosphorylation at Ser-152 prevents uptake of the transcription factor into the nucleus. Importantly, the serine residues in PHO4 that undergo phosphorylation are ideally suited for “caging” because they are located on the solvent accessible surface of the protein. This is an important point because it is easy to imagine that introducing relatively large photoactivatible moieties onto the side chains of buried amino acids will disrupt protein folding pathways. Moreover, the serine hydroxyl group can be easily converted into its ortho-nitrobenzyl derivative using straightforward and efficient chemistry (Lawrence, 2005).

UV irradiation can often lead to cell damage with the concomitant activation of repair mechanisms. In order to avoid this problem in their studies of “photocaged” PHO4, Lemke et al. (2007) prepared a serine derivative in which the side chain hydroxyl group was protected with the dimethoxynitrobenzyl (DMNB) moiety. This protecting group undergoes cleavage on irradiation with visible light (405 nm wavelength) to release the hydroxyl group [Fig. 1C]. Although few details were provided, the DMNB-protected serine derivative was taken up by Saccharomyces cerevisiae when introduced into the culture medium at millimolar concentration, and incorporated into PHO4-GFP variants using an interesting strategy that can, in principle, be utilized for any non-natural amino acid (Xie and Schultz, 2006). The approach relies on the presence of (i) a unique tRNA-codon pair, for which the tRNA is not a substrate for endogenous aminoacyl tRNA synthetases; and (ii) a corresponding “orthogonal” aminoacyl-tRNA synthetase capable of attaching the non-natural amino acid onto the unique tRNA with high specificity. In the studies of the photocaged PHO4-GFP variants, specific serine codons in the gene encoding the protein were replaced by the amber nonsense sequence (TAG), and an Escherichia coli leucyl suppressor tRNA_CUA (Leu5_CUA) (which has U35 and A37 in the anticodon loop) was introduced into the yeast strain. The endogenous yeast aminoacyl-tRNA synthetases cannot employ this bacterial tRNA as a substrate (Wu et al., 2004). A sophisticated screening procedure, involving both positive and negative selection (Xie and Schultz, 2006) was then employed to “evolve” the Escherichia coli leucyl-tRNA synthetase capable of using Leu5_CUA as a substrate into an enzyme for aminoacylating the tRNA with DMNB. The identification of an “orthogonal” tRNA synthetase with the desired specificity was, of course, critical to obtaining the desired PHO4-GFP variants. On this point, the active site of Escherichia coli leucyl-tRNA synthetase is particularly amenable to “directed evolution” as it is a large cavity containing only side chains and no backbone elements. As a result, large numbers (∼10⁷) of variants could be generated in the libraries that were screened for the desired aminoacylation specificity. Even after several screening rounds, it was found that the “evolved” enzyme still retained a low level of activity with leucine as a substrate. This problem was overcome, however, by the introduction of a “rational” mutation (Lincecum et al., 2003) that resulted in an increased hydrolysis rate for the undesired “mis-acylated” leucyl derivative thereby leading to enhanced fidelity for aminoacylation of Leu5_CUA with the photocaged serine derivative.

Having solved these “engineering” challenges, Lemke and co-workers then examined whether the resulting photocaged PHO4-GFP constructs could be used to “unravel” the effects of site-specific phosphorylation in a cellular milieu. The presence of GFP at the C-terminus of PHO4, which provides a convenient fluorescent marker for visualizing PHO4 localization and trafficking, does not affect biological function within Saccharomyces cerevisiae (O’Neill et al., 1998). An important issue in employing photocaged proteins in vivo is the possibility that the presence of the DMNB moiety affects translation and folding of the photocaged protein, or localization and transport to (and between) specific cell organelles. That this was not the case was demonstrated by control experiments in which the PHO4-GFP construct containing a photocaged serine at the S3 site was shown to localize to the nucleus due to phosphorylation at S4, where it was transcriptionally active. Photolysis with blue light in the presence of high exogenous phosphate then resulted in transport of the PHO4-GFP construct from the nucleus showing that removal of the DMNB group had taken place permitting phosphorylation at S3. By monitoring changes in nuclear fluorescence using an image series, Lemke et al. (2007) were also able to elucidate the kinetics of the phosphorylated PHO4-GPF construct from the nucleus. Such an experiment would not have been possible by merely replacing the S3 serine by alanine. In a further intriguing set of experiments, photocaged PHO4-GFP variants were used to deconvolute the effects of differential phosphorylation at the S2 and S3 sites, which prior experiments had shown to be phosphorylated at about the same rate. Thus, two PHO4-GFP variants were prepared in which the S2 site was removed and the S3 site protected as its DMNB derivative, and vice versa. Subsequent photolysis then revealed that when the S3 serine was replaced by alanine the resulting PHO4-GFP double mutant remained in the nucleus. On the other hand, photochemical removal of the DMNB group from the S3 site when alanine was substituted in the S2 site resulted in a slow rate of export from the nucleus. Hence, phosphorylation at the S2 and S3 sites gives rise to differential effects in the rate of nuclear export, raising the possibility that transcriptional control of the cellular response systems might be an additional, as yet uncharacterized, mechanism for controlling gene expression.

This work demonstrates the utility of photocaged proteins in evaluating the cellular effects of post-translational modifications, and the power of combining chemical synthesis with molecular biological methods in investigating the mechanisms of cellular regulation (Benner and Sismour, 2005). Although these recent studies on PHO4 employed photoactivatible serine derivatives, genetically encoded proteins containing similar protecting groups for tyrosine and cysteine side chains have been also been described (Dieters et al., 2006; Wu et al., 2004). Photocaged variants of asparagine have also been prepared (Ramesh et al., 1993), meaning that the extension of the genetically encoded strategy to studies of glycosylation is entirely feasible, assuming that an appropriate aminoacyl tRNA synthetase can be identified. Of course, in extending this strategy to other systems, a number of other technical challenges must be addressed. For example, it is clearly preferable not to employ microinjection methods to introduce the photocaged amino acid into the cell in order to preclude unexpected damage, and so cells must be able to take up the non-natural amino acid in a form that is a substrate for the engineered “orthogonal” aminoacyl tRNA synthetase. Second, the by-product of the photochemical cleavage reaction is an o-nitrosoaldehyde, which is a reactive species that can potentially interfere with cellular metabolism. The final issue is that of employing an optimized reporter system for evaluating the effects of uncaging the functionalized protein. In these experiments on PHO4, the GFP tag added to the C-terminal tail of the photocaged variants was known not to interfere with the ability of PHO4 to control transcription of its target genes. This may not, however, be true for other proteins of biochemical interest.