New aldehyde tag sequences identified by screening formylglycine generating enzymes in vitro and in vivo.

Formylglycine generating enzyme (FGE) performs a critical posttranslational modification of type I sulfatases, converting cysteine within the motif CxPxR to the aldehyde-bearing residue formylglycine (FGly). This concise motif can be installed within heterologous proteins as a genetically encoded "aldehyde tag" for site-specific labeling with aminooxy- or hydrazide-functionalized probes. In this report, we screened FGEs from M. tuberculosis and S. coelicolor against synthetic peptide libraries and identified new substrate sequences that diverge from the canonical motif. We found that E. coli's FGE-like activity is similarly promiscuous, enabling the use of novel aldehyde tag sequences for in vivo modification of recombinant proteins.

Formylglycine generating enzyme (FGE) was identified in 2003 as the posttranslational machinery that activates type I sulfatases in eukaryotes.(1) The enzyme oxidizes a cysteine residue within a ∼13 amino acid consensus sequence, also termed the “sulfatase motif”, forming an aldehyde-bearing formylglycine (FGly) residue (Figure 1) that is critical for the sulfatases’ catalytic function.(2) In eukaryotes, FGE requires a minimal submotif, CxPxR,[3,4] that is highly conserved among all type I sulfatases. However, in prokaryotes either CxP/AxR(5) motifs or serine-based SxPxR(6) motifs are found within sulfatases. Prokaryotic FGEs, first characterized from Mycobacterium tuberculosis and Streptomyces coelicolor,(7) recognize CxPxR, while anaerobic sulfatase-maturating enzymes (anSMEs) act on both CxAxR and SxPxR.(5) FGEs and anSMEs have distinct sulfatase substrates and catalytic mechanisms.[7,8]

Figure 1

Reaction catalyzed by FGE.

In addition to its intriguing biological function, FGE has also attracted attention as a tool for protein engineering. Conversion of cysteine to FGly introduces a uniquely reactive aldehyde group at a specific site dictated by the sulfatase motif. Recently, we reported that a six-residue sulfatase submotif (LCTPSR) can be introduced into heterologous proteins while maintaining in vivo conversion to FGly during expression in E. coli.(9) Once the aldehyde group was posttranslationally installed, chemoselective ligation with aminooxy- or hydrazide-functionalized molecules enabled site-specific protein modification. We employed this genetically encoded “aldehyde tag” for site-specific labeling of proteins with probes and polyethylene glycol (PEG) groups.

Although the six-residue aldehyde tag is a relatively small motif, its foreign sequence may perturb local structure or confer immunogenicity on therapeutic proteins. These potential liabilities prompted us to focus on expanding the repertoire of aldehyde tag sequences, with the ultimate objective of designing motifs that minimally perturb the host protein. Perusal of bacterial genomes that encode putative FGEs revealed sulfatase submotifs that differ from the canonical sequence CxPxR.(10) Therefore, naturally occurring FGEs might recognize a spectrum of motifs that could serve as diverse aldehyde tags for protein engineering.

In this work, we probed the specificities of FGEs from M. tuberculosis and S. coelicolor using an alanine-scanning peptide substrate library. We developed an in vitro assay (Figure 2) that monitors conversion of cysteine to FGly within synthetic N-terminally biotinylated peptide substrates. The peptides were first incubated with FGE, after which the newly formed aldehydes were reacted with aminooxy-functionalized 2,4-dinitrophenyl (2,4-DNP) conjugate 1.(11) The resulting oxime-linked products were captured on NeutrAvidin-coated microtiter plates. Colorimetric detection was accomplished by incubation with a commercial anti-2,4-DNP antibody conjugated to alkaline phosphatase (α-2,4-DNP-AlkPhos) followed by treatment with p-nitrophenyl phosphate (pNPP).(12)

Figure 2

A high-throughput assay for FGE activity.

We generated two peptide libraries based on 13-residue motifs found in putative sulfatases from the two prokaryotes (ICTPARASLLTGQ and LCTPSRGSLFTGR, from S. coelicolor and M. tuberculosis, respectively). Each residue within the sequences was probed by alanine substitution to generate a total of 28 peptides including the two wild-type sequences (native alanine residues within the S. coelicolor sequence were substituted with glycine). The percent conversion of cysteine to FGly was quantified for each alanine- (or glycine)-substituted peptide relative to that of the corresponding wild-type sequence.

As shown in Figure 3, the two FGEs displayed different tolerances for alanine mutations within the sulfatase motifs. Substitution at any position in the native sequence recognized by S. coelicolor FGE resulted in significant reduction in FGly formation (Figure 3a, blue bars). Replacement of Thr3, Pro4, Arg6, or Leu9 with alanine was particularly detrimental. A similar specificity profile was observed with the library derived from the M. tuberculosis sulfatase motif (Figure 3b). Human FGE, which has a 51% amino acid sequence identity to S. coelicolor FGE, also has a strict requirement for Pro and Arg within the CxPxR sequence.[3,4] However, the human enzyme is known to tolerate substitutions corresponding to Thr3 or Leu9,(4) indicating species-specific variation in substrate preference. Surprisingly, M. tuberculosis FGE displayed a much greater tolerance for alanine substitution in both sulfatase motif libraries (Figure 3a and b, red bars). Notably, replacement of Pro4 or Arg6 with alanine was well tolerated, as were substitutions in the C-terminal region.

Figure 3

FGE activity on peptide substrates. (a) Relative activity of S. coelicolor (blue) and M. tuberculosis (red) FGEs on peptides derived from the S. coelicolor sulfatase motif. (b) Relative activity of S. coelicolor (blue) and M. tuberculosis (red) FGEs on peptides derived from the M. tuberculosis sulfatase motif. Error bars represent the standard deviation of three replicates.

Despite the 46% amino acid sequence identity shared by M. tuberculosis and S. coelicolor FGEs, their response to alanine substitutions in peptide substrates is very different. To gain insight into the molecular basis of substrate discrimination, we generated structural models of FGE-peptide complexes using the S. coelicolor enzyme’s crystal structure(7) and a homology model of M. tuberculosis FGE (Figure 4).(14) These models indicated that the substrate’s conserved Pro residue binds within a pocket that varies considerably in size between the species homologues. The pocket in the M. tuberculosis FGE model (Figure 4b) appears more open, potentially accommodating a greater spectrum of amino acid alterations in the peptide substrate. The S. coelicolor FGE pocket, by contrast, appears to be more confined around the bound Pro residue.

Figure 4

Models of prokaryotic FGE active sites with peptide substrate bound. (a) Crystal structure of S. coelicolor FGE with modeled peptide substrate. (b) Homology model of M. tuberculosis FGE with modeled peptide substrate. The substrate peptide shown in cyan is CTPSR. Colors indicate electrostatic potential (blue, positive; red, negative).

The data in Figure 3 suggest that FGEs from certain prokaryotes are capable of modifying alternative aldehyde tag sequences that diverge from the canonical motif. In previous work, we showed that E. coli possesses an FGE-like activity that converts Cys to FGly in heterologous proteins possessing the canonical sequence LCTPSR.(9) Although its molecular identity is not known, the FGE-like activity’s presence in this popular protein expression host enables the production of aldehyde-tagged proteins without need for exogenous FGE. To determine whether E. coli’s FGE-like activity exhibits substrate promiscuity, we expressed the maltose-binding protein (MBP) possessing various aldehyde tag sequences at the C-terminus downstream of a His6 tag (Figure 5). Control proteins bearing the corresponding C-to-A mutation or the wild-type sulfatase motif (LCTPSR) were expressed similarly. The isolated proteins were reacted with Alexa Fluor C5-aminooxyacetamide and analyzed by SDS-PAGE and fluorescence imaging (Figure 5).

Figure 5

SDS-PAGE of MBP constructs bearing the C-terminal aldehyde tag sequences shown above each lane. The proteins were expressed in E. coli, purified on Ni-NTA spin columns, and reacted with Alexa Fluor 647 C5-aminooxyacetamide (Aminooxy Alexa Fluor 647). Fluorescence images of the gel are shown. Top, Alexa Fluor 647. Bottom, protein loading as determined by Sypro Orange.

The E. coli machinery converted all three sequences tested—LCTPSR (wild-type), LCTASR, and LCTASA—at comparable levels, while no signal was observed for any of the C-to-A mutants. Alanine substitution of the conserved Pro and Arg residues within the canonical sequence did not significantly reduce conversion efficiency. This striking observation suggests that a wide range of aldehyde tag sequences are recognized in E. coli, offering a practical system for expression of modified proteins.

In summary, peptide library screening revealed noncanonical sequences that are recognized by M. tuberculosis FGE in vitro and the E. coli FGE-like activity in vivo. This finding expands the range of aldehyde tag sequences for protein engineering. An important future goal is to identify the molecular nature of E. coli’s machinery.