Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Am Chem Soc. Author manuscript; available in PMC 2010 November 11.
Published in final edited form as:
PMCID: PMC2783612

Synthetic Control of Green Fluorescent Protein (GFP)


Semi-synthetic green fluorescent proteins (GFPs) can be prepared by producing truncated GFPs recombinantly and assembling them with synthetic β-strands of GFP. The yield from expressing the truncated GFPs is low, and the chromophore is either partially formed or not formed. An alternative method is presented in which full-length proteins are produced recombinantly with a protease site inserted between the structural element to be removed and the rest of the protein. The native peptide can then be replaced by cutting the protease site with trypsin, denaturing in guanidine hydrochloride to disrupt the complex, separating the native peptide from the rest of the protein by size exclusion, and refolding the protein in the presence of a synthetic peptide. We show that this method allows for removal and replacement of the interior chromophore containing helix, and that the GFP barrel is capable of inducing chromophore formation in a synthetic interior helix.

We recently demonstrated that semi-synthetic green fluorescent protein (GFP) can be assembled by adding a synthetic peptide to a truncated protein that is produced recombinantly1. This in vitro assembly mimics the in vivo assembly of “split GFP” used for protein solubility assays2 and protein colocalization3,4. Unfortunately, recombinant expression of truncated GFPs often results in low sample yield5 with the chromophore either not formed1,6 or partially formed7. Here we present a method that overcomes both of these limitations and enables efficient synthetic control of all residues, including those in the chromophore-containing interior α-helix. This method can be used to produce samples of split GFP for in vitro biophysical characterization and should inform efforts for improvement or interpretation of in vivo experiments.

Figure 1 displays the 11 β-sheets and the interior α-helix (ih) along with the modifications made to the GFP primary sequence to introduce loops that can be selectively digested with proteases and circular permutation of the C- and N-termini8. Figure 2 shows our synthetic strategy beginning with the full-length GFP which is cleaved at the loop insertion, stripped of the small terminal peptide, and reassembled with a synthetic peptide. This method is reminiscent of the preparation of split ribonuclease S, which is generated by proteolysis of ribonuclease with subtilisin9, except that in this case a trypsin cleavage site is specifically engineered into the normally trypsin-resistant GFP10 with loop insertions. After digesting the loop the GFP remains intact and spectrally indistinguishable from the uncut protein. The original strand is removed by denaturation and then replaced by a fully synthetic strand with any desired sequence containing natural or unnatural amino acids.

Figure 1
The 11 β-strands and α-helices of GFP represented as arrows and cylinders, respectively. The dashed line shows the loop insertion used to make GFP:loop:s11 (Fig 2A), and the dash-dotted line shows both the circular permutation and the ...
Figure 2
Methods for replacing (A) β-strand 11 (or any of the 11 β-strands by circular permutation, c.f. Fig. 1) or (B) the interior helix (ih) that contains the 3 amino acids that become the GFP chromophore. In each case, a GFP with a loop insertion ...

Due to the manifold distinct protein constructs, we developed the systematic notation illustrated in Fig. 2. Anything to the left of the term GFP is on the N-terminal side of the protein, and anything to the right of the term GFP is on the C-terminal side. ‘Loop’ refers to the sacrificial loop insertions (Fig. 1), ‘s11’ refers the 11th stave of the β-barrel11, and ‘ih’ refers to the interior α-helix. A strike through ‘loop’ implies the loop was removed with trypsin, and a strike though ‘ih’ or ‘s11’ implies the original peptide was removed by denaturation and size exclusion12. Any synthetic peptide is underlined, and the dot (•) implies that a noncovalent complex has been formed between the protein and the synthetic peptide preceding and following the dot.

Following removal of s11 and the denaturant (Fig. 2A), the absorption of GFP: loop:s11 is quite different from that of the native protein (Fig. 3, dashed). Interestingly, this protein is much more fluorescent than expected (Fig. 3, dashed red), since the chromophore itself, when outside the protein environment, is non-fluorescent in aqueous solution13. This suggests that there is residual structure in GFP: loop:s1114 that prevents non-radiative decay pathways available to the free chromophore. Consistent with this, the UV CD of GFP: loop:s11 is very similar to GFP itself (supporting information). Upon addition of s11 with the identical sequence, a protein whose absorbance and fluorescence spectra are indistinguishable from the original protein is formed (Fig. 3, solid)15. The method described above can be used to gain synthetic access to any strand in GFP11.

Figure 3
Absorbance and fluorescence spectra of elements in method described in Fig. 2A for the removal and replacement of strand 11. GFP: loop:s11, GFP: loop:s11, and GFP: loop:s11s11 and spectra are shown by dotted, dashed, and solid lines, respectively. ...

Remarkably this strategy also works for the interior helix (Fig. 4). ih:loop:GFP has the entire chromophore-containing helix removed. In order to determine if this “empty barrel”14 can catalyze chromophore formation in a synthetic ih upon reconstitution, ih with the S65T mutation (ih S65T) was introduced to ih:loop:GFP using the scheme in Figure 2B. The fluorescence excitation spectrum of ih:loop:GFP•ih S65T overlays with that of S65T ih:loop:GFP but not with that of ih:loop:GFP, the starting material from which ih:loop:GFP was derived (Fig. 4). The conversion of the characteristic absorption and fluorescence from that of the native chromophore with Ser at position 65 to that characteristic of the S65T mutation, unambiguously demonstrates that ih:loop:GFP induces chromophore formation in ih S65T. Following the methods described in the supplementary materials, the reconstituted fluorescence develops with a half-life of roughly 1 day, and the maximum fraction reconstituted has thus far been ~20%15,16.

Figure 4
Fluorescence excitation spectra of the elements shown in Fig. 2B. The native ih containing the chromophore incorporating serine 65 is removed, yielding a colorless protein ( ih:loop:GFP), followed by addition of a synthetic ih or a synthetic ih with the ...

This strategy has the advantage of producing high-yield samples with a mature chromophore, unlike previous methods of producing semi-synthetic GFPs1,6,7. Moreover, we have shown that perhaps the largest perturbation, removing and replacing the interior α-helix, is possible. Replacement of the interior α- helix with a synthetic peptide allows facile introduction of modifications to the chromophore structure using unnatural amino acids or other synthetic elements.

Supplementary Material



This work was supported in part by a grant from the NIH (Grant GM27738). LMO is supported by an ARCS Foundation Stanford Graduate Fellowship.


Supporting Information Available: Protein preparation, mass spectra, and amino acid sequences. This material is available free of charge via the Internet at


1. Kent KP, Childs W, Boxer SG. J Am Chem Soc. 2008;130:9664–65. [PubMed]
2. Cabantous S, Terwilliger TC, Waldo GS. Nat Biotechnol. 2005;23:102–7. [PubMed]
3. Shyu YJ, Hu CD. Trends Biotechnol. 2008;26:622–30. [PubMed]
4. Ozawa T. Anal Chim Acta. 2006;556:58–8. [PubMed]
5. Degradation or precipitation of truncated GFPs likely causes low yields.
6. Sakamoto S, Kudo K. J Am Chem Soc. 2008;130:9574–82. [PubMed]
7. Huang YM, Bystroff C. Biochem. 2009;48:929–40. [PubMed]
8. Despite these modifications, the E. coli expression yields of the full-length protein containing the chromophore remain high (>100mg/L). The parent GFP (see Supplementary Materials for full sequences) is derived from a split GFP developed for solubility assays2, except that position 65 is reverted back to serine, which provides an optical probe for the integrity and photochemistry of GFP17,18 (see Fig. 4)
9. Richards FM, Vithayathil PJ. J Biol Chem. 1959;234:1459. [PubMed]
10. Sniegowski JA, Phail ME, Wachter RM. Biochem Biophys Res Comm. 2005;332:657–63. [PubMed]
11. Any of the 11 β-strands can be removed by circular permutation, but we use the natural C-terminal 11th strand for illustration. We used this method to remove s7; this will be reported in subsequent papers.
12. GFP 1-10mat referred to previously1 is GFP:loop:s11 in this scheme.
13. Niwa H, Inouye S, Hirano T, Matsuno T, Kojima S, Kubota M, Ohashi M, Tsuji FI. Proc Natl Acad Sci USA. 1996;93:13617–22. [PubMed]
14. The structures of GFP:loop:s11 and ih:loop:GFP are not yet known.
15. The kinetics of reassembly of ih and s11 with ih:loop:GFP and GFP:loop:s11, the incorporation of unnatural amino acids, and the rate of chromophore formation in ih will be described in subsequent work.
16. We note that native chemical ligation could be used to fuse ih to the N-terminus of ih:loop:GFP.
17. Tsien RY. Annu Rev Biochem. 1998;67:509–44. [PubMed]
18. Chattoraj M, King BA, Bublitz GU, Boxer SG. Proc Natl Acad Sci USA. 1996;93:8362–7. [PubMed]