An AGT-based protein-tag system for the labelling and surface immobilization of enzymes on E. coli outer membrane

Abstract The use of natural systems, such as outer membrane protein A (OmpA), phosphoporin E (PhoE), ice nucleation protein (INP), etc., has been proved very useful for the surface exposure of proteins on the outer membrane of Gram-negative bacteria. These strategies have the clear advantage of unifying in a one-step the production, the purification and the in vivo immobilisation of proteins/biocatalysts onto a specific biological support. Here, we introduce the novel Anchoring-and-Self-Labelling-protein-tag (ASLtag), which allows the in vivo immobilisation of enzymes on E. coli surface and the labelling of the neosynthesised proteins with the engineered alkylguanine-DNA-alkyl-transferase (H5) from Sulfolobus solfataricus. Our results demonstrated that this tag enhanced the overexpression of thermostable enzymes, such as the carbonic anhydrase (SspCA) from Sulfurihydrogenibium yellowstonense and the β-glycoside hydrolase (SsβGly) from S. solfataricus, without affecting their folding and catalytic activity, proposing a new tool for the improvement in the utilisation of biocatalysts of biotechnological interest.


Introduction
The term immobilised enzymes refers to 'enzymes physically confined or localised in a certain defined region of space with retention of their catalytic activities, and which can be used repeatedly and continuously' 1 . The immobilisation of enzymes on solid supports is historically very important for overcoming their general instability in harsh operational conditions and their low shelf-life, as well as the need for their recycling more times 2 . Furthermore, the physical separation of the biocatalyst from the reaction mixture avoids the protein contamination of the products. Although a reduction in reaction rates sometimes occurs, because the enzyme cannot mix freely with the substrate or a particular conformational change is needed for the biocatalyst efficiency, there are many examples of increased activity and stability of immobilised enzymes 3 . Many chemical or physical methods for the enzyme immobilisation are currently available, from the physical adsorption to the covalent coupling on supports ( Figure  1(a-c,e-f)) [3][4][5][6] . Recently, the use of protein-tags based on an engineered version of the human O 6 -alkyl-guanine-DNA-alkyl-transferases (hAGT) is an effective alternative for the covalent immobilisation of proteins and enzymes (Figure 1(d)) [7][8][9] . AGTs (or OGTs, MGMTs; E.C.: 2.1.1.63) are DNA repair enzymes, which irreversibly transfer the alkyl group from the damaged DNA containing O 6 -alkyl-guanines to their cysteine residue in the active site [10][11][12][13] . In 2003, Johnsson and his group demonstrated that most enzymes of this class display relatively low substrate specificity, making them reactive also with free O 6 -benzyl-guanines (O 6 -BG) nucleobases 14 . This led to the development of the so-called SNAPtag TM technology, which uses derivatives of O 6 -BG potentially conjugated with an unlimited number of chemical groups [15][16][17][18] . This system allows the immobilisation on O 6 -BG-derivatised surface of the protein expressed in fusion with the SNAP-tag 18 (Figure 1(d)). However, all these approaches mainly depend on the high costs due to the isolation and purification of the biocatalysts. This limitation can be easily overcome by the heterologous expression of enzymes and their in vivo direct immobilisation on the surface of bacterial hosts, by the utilisation of transmembrane protein domains, as the ice nucleation protein (INP) of the Gram-negative bacterium Pseudomonas syringae (Figure 1(g)) 19,20 . This protein is composed of an N-terminal domain (N, 175 residues) structurally separated from a C-terminal domain (C, 49 residues) by a repetitive central domain 21 . Both domains play a role in the anchoring of proteins to the outer membrane 21 . The use of INP as anchoring carrier is considered of great interest in biotechnological applications, ranging from the development of bacterial cell surface-display systems for vaccine delivery to the fabrication of whole-cell biocatalysts and biosensors [22][23][24] . The N-terminal domain of INP (INPN) was recently and successfully used for the one-step procedure immobilisation (Figure 1(g)) [22][23][24][25][26][27][28][29] . Moreover, Capasso et al. demonstrated that the amount of a thermostable carbonic anhydrase [30][31][32][33] fused to the INPN domain and expressed on the bacterial cell surface had a hydratase activity similar to that of the enzyme covalently immobilised onto magnetic nanoparticles 30,34 . Here, we introduce a novel protein-tag system, (hereinafter Anchoring-and-Self-Labelling-protein-tag or ASL tag ), which simultaneously allows the in vivo immobilisation of the enzyme of interest on the E. coli surface and its quantitative determination ( Figure 2). The ASL tag system is formed by the INPN domain fused to an engineered and thermostable variant of the alkylguanine-DNAalkyl-transferase (H 5 ) from the hyperthermophilic archaeon Sulfolobus solfataricus 35,36 . This enzyme was extensively characterised, suggesting its biotechnological role as thermostable alternative to the commercial SNAP-tag TM and its utilisation as protein-tag for heterologous expression of proteins of interest in E. coli and, for the first time, in thermophilic organisms as Thermus thermophilus and Sulfolobus islandicus [35][36][37] . Thus, using the substrate of H 5 , a fluorescein derivative of the O 6 -BG (Figure 2), we successfully estimated the expression of the ASL tag in E. coli cells, by in vitro gel-imaging techniques, as well as by in vivo fluorescent microscopy. Furthermore, we demonstrated that the activity and the stability of the enzymes of interest (SspCA, the a-carbonic anhydrase from Sulfurihydrogenibium yellowstonense; and SsbGly, the b-glycoside hydrolase from S. solfataricus) fused to the ASL tag and exposed on the surface of E. coli cells were not affected by the presence of this novel protein-tag.

Reagents
All DNA restriction and modification enzymes and the fluorescent substrate for the OGT activity (SNAP-Vista Green TM , hereinafter BG-FL) were purchased from New England Biolabs (Ipswich, MA); molecular biology kits for the plasmid preparations and DNA gel extractions (NucleoSpin V R Gel and PCR Clean-up V R ) were from Macherey-Nagel GmbH (Germany); Lyophilised Thrombin Protease from GE Healthcare (Illinois, US). Eurofins Genomics (Germany) performed the oligonucleotides synthesis and the DNA sequencing service.

DNA constructs
To obtain the pET-ASL tag construct, we replaced the a-carbonic anhydrase (SspCA) gene with the ogtH 5 gene in the previously described vector pET-22b/INPN-SspCA 30 . By the latter and the pQE-ogtH 5 35 plasmid as template, the DNA fragments relative to the INPN domain and H 5 were respectively amplified with the INPN-and H 5 -Fwd/Rev oligonucleotides pairs (listed in Table 1), under the following conditions: an initial denaturation at 95.0 C for 5 min, 30 cycles of 30 s at 95.0 C, 30 s at 50.0 C and 30 s at 72.0 C, followed by a final extension of 5 min at 72.0 C. DNA products were fused to each other in a further PCR amplification, taking advantage of the total complementarity of the INPN-Rev to the H 5 -Fwd oligonucleotide, obtaining the final ASL tag DNA fragment. Subsequently, this fragment and the pET-22b/INPN-SspCA vector were digested with Hind III and Xba I restriction endonucleases, gel-purified, and ligated. The ligation mixture was used to transform the E. coli DH5a strain, and positive colonies were confirmed by colony PCR and DNA sequencing.
The same cloning strategy was used to achieve the pET-ASL tag -SspCA construct: the ASL tag DNA fragment obtained this time using the H 5 -Rev2 oligonucleotide was used as template to further fuse to the SspCA gene (obtained by amplification with SspCA Fwd/Rev oligonucleotide pairs, Table 1). Again, positive colonies after ligation and transformation were confirmed as above.
Finally, the pET-ASL tag -lacS plasmid preparation started with the achievement of the DNA fragment relative to the ogtH 5 gene fused to the b-glycoside hydrolase from the thermophilic archaea Sulfolobus solfataricus (lacS), obtained by the Hind III/BamH I digestion from the pQE-ogtH 5 -lacS plasmid 36 . The pET-ASL tag plasmid was similarly digested to obtain the pET22 recipient for the first ligation/transformation round. Positive blue colonies were selected by the hydrolase activity of the lacS gene product (SsbGly) on the Ampicillin selective Luria-Bertani (LB) Agar plates supplemented with 5-bromo-4-chloro-3-indolyl-b-D-glucopyranoside (X-Glc), and the insertion confirmed using PCR analysis. This intermediate vector was further digested with BamH I, treated with Alkaline Phosphatase Calf Intestinal (CIP) and ligated to the BamH I/BamH I INPN DNA fragment, derived from the digestion of the pET-ASL tag plasmid. In this case, positive blue colonies on This tag can be further fused to a gene of interest (goi, in blue), for a one step procedure of the expression and immobilisation of an enzyme (E). The presence of the H 5 moiety allows the quantitative estimation of the yield of E by the irreversible alkyl-transferase assay using a fluorescent O 6 -benzyl-guanine derivative (BG-FL). Between inpn and ogtH 5 , a spacer (in pink), a thrombin cleavage site (shown as an orange triangle) and a 6 Â His-tag (in black) were inserted, for the easy separation and purification of any H 5 -E fusion protein.
LB-Amp-X-Glc agar plates were confirmed by Pst I restriction enzyme digestion analysis.

Determination of the protein expression by a fluorescent assay based on the H 5 activity
All constructs were used to transform E. coli BL21(DE3) cells. Cultures were grown at 37.0 C in LB selective medium supplemented with 100.0 mg/L ampicillin and 30.0 mg/L chloramphenicol; expression was induced with 1.0 mM isopropyl-thio-b-Dgalactoside (IPTG) when an absorbance value of 0.5-0.6 A 600 nm was reached, or in the ZY auto-induction medium (AI) 38 , supplemented with the same selective antibiotics. After an overnight incubation, whole cells were collected and assayed by using the BG-FL fluorescent H 5 substrate previously described [35][36][37]39 for a qualitative measurement of the protein expression, an aliquot of 1.0 mL of cells was centrifuged at 4000 Â g and the pellet was resuspended in 50.0 lL of 5.0 lM BG-FL in phosphate-buffered saline (PBS 1Â, 20.0 mM phosphate buffer, NaCl 150.0 mM, pH 7.3). After an incubation for 2.0 h at 37.0 C, reactions were stopped by denaturing the samples in Leammli Buffer 1Â and directly loaded on SDS-PAGE, followed by gel-imaging on a VersaDoc 4000 TM system (Bio-Rad), by applying a blue LED/530 bandpass filter as excitation/emission parameters, respectively. Finally, the fluorescence intensity of each band was normalised to the intensity of the signal obtained from the Coomassie Blue staining analysis.
For a quantitative determination of the expression, whole cells were opportunely diluted to achieve an OD 600nm of 1.0. By following the same above-mentioned assay, three different volumes of whole cells were loaded in the same gel with defined amounts (in the range of 0.8-50.0 pmols) of purified free H 5 protein, after the reaction on BG-FL in the same conditions. The obtained values of the relative fluorescence as a function of the purified loaded H 5 were fitted by a linear equation, whose parameters were then used for the estimation of the amount of expressed H 5 -derivated fusion proteins, assuming that the activity of the H 5 moiety in the fusions is not affected by the presence of the other protein partner(s). Given that the concentration of E. coli cell cultures of 1.0 OD 600nm is ca. 8.0 Â 10 8 cells/mL, and the amount of wet cells is 1.7 g/L, it is possible to calculate the yield of expressed proteins in terms of pmol/mg of cells 40 .

Membranes fractionation
The E. coli outer (OM) and inner (IM) membranes were purified by following a procedure previously described 27 . Briefly, harvested bacterial cells were resuspended 1:20 (g/mL) in 25.0 mM Tris/HCl buffer, pH 8.0 and disrupted by sonication on ice (10 cycles of 10 s: 50 s on:off treatment). Cell extract was centrifuged at 120,000 Â g for 1.0 h, and the supernatant containing the cytoplasmic fraction was discarded. Both IM and OM fractions were recovered in the pellet and resuspended in 20.0 mL of PBS 1Â, containing 0.01 mM MgCl 2 and 2.0% Triton X-100. After incubation at room temperature for 30.0 min, the solution was centrifuged as described above. Then, a defined amount of the OM fraction obtained was assayed for the H 5 activity, leading to the quantitative determination of the total amount of fusion protein, as previously described.

Thrombin assay
The ASL tag on the OM was cleaved with the Thrombin Protease, in order to separate the H 5  Images were collected using a DM6 fluorescence microscope and Hamamatsu camera under the control of Leica LAS AF 6000 software; excitation and emission wavelengths used suitably for AlexaFluor488 dye were k ex ¼ 490 nm; k em ¼ 525 nm, respectively.

b-glycoside hydrolase assay
The b-glycoside hydrolase assay was performed as previously described 41 at different temperatures in 50 mM sodium phosphate buffer at pH 6.5, in the presence of 2 Np-and 4 Np-Glc substrates at 5.0 mM final concentration. OM fractions containing ASL tag and relative fusion proteins amounts ranging from 1.0 to 5.0 lg were used in each assay. For the correction of the spontaneous hydrolysis of the substrates, all the reactants except the enzyme (blank mixture), was taken into account. The enzymatic activity was calculated on the basis of the molar extinction coefficient (e M ) values of 2-and 4-nitrophenol in 50 mM sodium phosphate buffer pH 6.5 at different temperatures, as previously reported 41 . We defined as one unit of enzyme activity the amount of enzyme hydrolyzing 1.0 lmol of substrate in 1.0 min, under the abovedescribed conditions.

Carbonic anhydrase assay
CA activity assay was a modification of the procedure described by Capasso et al. 33 . Briefly, the hydratase assay was performed at 0 C using CO 2 as substrate following the pH variation due to the catalyzed conversion of CO 2 to bicarbonate. Bromothymol blue was used as pH indicator. The production of hydrogen ions during the CO 2 hydration reaction lowers the pH of the solution leading to a colour transition of the dye. The time required for the colour change is inversely proportional to the amount of CA present in the sample. The Wilbur-Anderson units (WAU) were calculated according to the following definition: one WAU of CA activity is defined as the ratio (T 0 À T)/T, where T 0 (the time needed for the pH indicator color change for the uncatalyzed reaction) and T (the time needed for the pH indicator color change for the catalyzed reaction) are recorded as the time (in seconds) required for the pH to drop from 8.3 to the transition point of the dye (pH 6.8) in a control buffer and the presence of enzyme, respectively.

Expression analysis and localisation of ASL tag in E. coli
The in vivo alkyl-transferase fluorescent assay of H 5 on whole bacterial cells is a useful method to determine the heterologous expression of this protein and/or relative fusion proteins in the mesophilic E. coli and in the thermophilic species T. thermophilus  Table 2.  and S. islandicus, without any purification procedure [35][36][37] . The in vivo assay is possible since the bacterial cells are permeable to OGT fluorescent substrates (as the commercially available SNAP-Vista Green TM or SNAP-Cell TMR TM ); and the catalytic activity at mesophilic temperatures (25-37 C) of H 5 mutant is one order of magnitude higher than the SsOGT wild-type counterpart and comparable to the commercial SNAP-tag TM36 . Despite GFPs utilisation, the covalent conjugation of H 5 with the benzyl-fluorophore moiety of the substrate (Figure 2) allows the denaturation of the whole cells and the direct loading of the samples on SDS-PAGE for the gel-imaging analysis, as described in the Materials and Methods [35][36][37]39 .
To evaluate the expression of the ASL tag , E. coli BL21(DE3)/pET-ASL tag cells were grown in LB medium and induced with IPTG or in the AI. In the latter case, although the expression level of the fusion protein was satisfactory, the presence of only the H 5 signal (ca. 20% of the total fluorescence intensity) in the gel-imaging analysis is difficult to rationalise (Figure 3(a)). Probably, during the expression of the protein (e.g., in the advanced stage of the growth) could occur interruptions or failures in the translocation process of the ASL tag on the outer membrane of E. coli with following breaking/cleavage events, especially in the spacer region between INPN and H 5 (Figure 2). On the other hand, after the IPTG induction, a strong and clear signal at the expected molecular weight is visible, without any fragmentation ( Figure  3(a)). This is an important result since, to date, the heterologous expression of SsOGT wild-type and relative variants in E. coli has been generally performed in the ABLE C strain, which keeps low the number of copies of the ogt-containing plasmids 35,36 . As proof, transformed E. coli BL21(DE3) strain with the pQE-ogtH 5 plasmid 36 showed a very low expression of free H 5 (Supplementary Figure S1), whereas the fusion with the INPN domain and the consequent translocation on the outer cell membrane made the H 5 expression possible in this strain. The comparative fluorescent analysis with a defined amount of free H 5 enzyme, allowed us to quantitatively measure the heterologous expression of ASL tag as pmol/mg of the whole wet cells (Figure 3(b) and Table 2). The assay on H 5 confirmed the anchoring function of the INPN trans-membrane protein: only the whole cells and the fraction containing the outer membrane displayed a fluorescent band corresponding to the ASL tag fusion protein, whereas the signal was missed in the lanes relative to the cytoplasmic and the inner membrane fractions (Figure 4(a)). Besides, the evidence of the INPN on the bacterial outer membrane was confirmed by treating the whole bacterial cells with the thrombin, too ( Figure 2). A cleavage site for this protease was localised between the INPN domain and the H 5 moiety (Figure 2). The fluorescent signal corresponding to the MW of the H 5 protein was present only in the supernatant fraction when the protease was added at the same time with the BG-FL substrate (Figure 4(b)). Finally, we checked the OGT activity of ASL tag in living cells by microscope analysis, upon the labelling with the fluorescent substrate. The obtained images of living E. coli BL21(DE3) cells showed a strong and specific fluorescent signal only in those transformed with the ASL tag -containing plasmid (Figure 4(c)), indicating that the fusion protein is stable and proficient to labelling. This data suggest that ASL tag is suitable for localisation and analysis of membrane proteins, and provide an opportunity for further in vivo analyses of ASL tag -tagged proteins of interest under physiological conditions.

In vivo immobilisation of thermostable enzymes by ASL tag
As described in the literature, it has been demonstrated that the monomeric a-carbonic anhydrase (SspCA) from the thermophilic bacterium S. yellowstonense can be actively expressed onto the outer membrane of E. coli 30 . Following this strategy, we realised a plasmid expressing the ASL tag -SspCA construct by inserting the ogtH 5 gene between the INPN and SspCA. The expression profile analyzed by following the H 5 activity on BG-FL confirmed that in the AI multiple fluorescent signals are present ( Figure 5(a)), mainly represented by the full-length ASL tag -SspCA (70.2 kDa; ca. 65% of the total fluorescence intensity) and lower band (ca. 35%) closer to 37 kDa than 50 kDa ( Figure 5(a)). This signal is compatible to the ASL tag (42.0 kDa) as well as the H 5 -SspCA moiety (46.0 kDa), suggesting that both translation interruption and translocation failure events in this growth conditions can be not excluded. Again, we detected an optimal achievement of the full-length fusion protein under IPTG-based expression in LB medium (ca. 95%; Figure 5(a)). In this condition and considering the INPN:H 5 :SspCA ratio as 1:1:1, the amount of the whole fusion protein expressed was estimated as ca. 60.0 pmol/g cells, which corresponds to ca. 1.5 lg of the sole immobilised carbonic anhydrase per mg of cells (Table 2). Preliminary assays on ASL tag -SspCA indicated that the presence of H 5 does not hamper the hydratase activity of the SspCA, if compared with that of the previously expressed INPNanchored enzyme 30 (data not shown).
ASL tag system was also tested with another thermostable enzyme, the b-glycoside hydrolase (SsbGly) from the thermophilic archaea S. solfataricus 42 . We previously demonstrated that the cytoplasmic H 5 -SsbGly fusion protein is stable and active for both the OGT and the b-glycosidase assays, suggesting that the presence of one enzyme does not interfere with the folding and activity of the other 36 . Interesting to note that anchoring this protein fusion to the bacterial outer membrane by the INPN domain is particularly challenging because SsbGly is active only in its tetrameric form 43,44 . The presence of blue colonies on LB agar plate in the presence of a glucoside chromogenic derivative (X-Glc), which is a preferred substrate of SsbGly 41 but not of the E. coli LacZ (a b-galactosidase enzyme) was a first and convincing indication of the oligomerisation of this thermostable enzyme. Although in this case the amount of the expressed fusion protein was lower than the above examples ( Figure 5(b) and Table 2), IPTG-inducted E. coli BL21(DE3)/pET-ASL tag -SsbGly cells displayed a fluorescent signal of expected molecular weight (98.8 kDa), corresponding to one monomer of SsbGly fused to one ASL tag unit. However, a higher band is clearly visible in the fluorescence analysis (marked with an asterisk in Figure 5(b)), out of the molecular weight marker range used: as for the cytoplasmic H 5 -SsbGly fusion, it could suggest that it corresponds to a partially denatured part of the tetrameric form of ASL tag -SsbGly (ca. 400.0 kDa), which is particularly resistant to thermal denaturation 36,42,43,45 . Finally, an amount of 0.24 lg/mg of immobilised SsbGly on the OM fraction (on the basis of the calculated H 5 pmol) was assayed on 2 Np-and 4 Np-Glc at three different temperatures. The results show an activity of 12.6 ± 0.7 and 8.8 ± 0.4 (50 C), 30.3 ± 0.4 and 20.3 ± 0.9 (60 C), 51.5 ± 0.9 and 30.8 ± 1.5 U/mg (70 C), respectively, whereas OM fraction containing the sole ASL tag did not result in any b-glucosidase activity, as expected (data not shown). These values are correctly related to the activity of SsbGly 45 , clearly indicating that the formation of the quaternary structure on the E. coli OM occurs. However, since the 3D structure of this thermostable glycoside hydrolase showed that it is not laying on a surface 43 , we hypothesised an invagination of the E. coli outer membrane to allow the assembly of four units of the ASL tag -SsbGly (Supplementary Figure S2).

Conclusions and perspectives
In the present work, we introduced and demonstrated the utility of a novel protein-tag, composed by the N-terminal domain of the INP protein fused to a DNA repair enzyme. From our results, it is readily apparent that ASL tag offers: (i) an easy expression and in vivo one-step procedure of enzyme immobilisation on biological supports (e.g., E. coli outer membrane); (ii) significantly reduces the costs of the enzyme purification and those of the immobilisation support, allowing a direct exposition of the enzyme to the solvent 20,30 ; (iii) an indirect labeling, by the reaction of a thermostable variant of the SNAP-tag TM (H 5 ) 46,47 , which covalently links desired chemical groups conjugated to its benzyl-guanine substrate 14,35 . ASL tag favoured the expression of a monomeric protein (e.g., the thermostable SspCA) and an enzyme having a complex quaternary structure (e.g., the thermophilic SsbGly), without compromising their overall folding and enzymatic activity. Moreover, we showed that the utilisation of a fluorescein-derivative of the BG led to the localisation and the quantitative determination of the yield of the expressed ASL tag and the relative fusion proteins ( Table 2). On the other hand, despite the GFPs utilisation limited only to all fluorescence-based applications, the possibility to conjugate different groups to the BG-for the H 5 reaction 36 dramatically expands the biotechnological potential of this novel protein-tag. For example, it will be possible to modulate the activity of biocatalysts (by introducing inhibitors/activators), or connecting them with other proteins for the improvement of cascade reactions in the presence of bi-functional chemical groups ( Figure 6).