The development of fluorescent protein tracing vectors for multicolor imaging of clinically isolated Staphylococcus aureus

Recent advances in fluorescent protein technology provide a wide variety of biological imaging applications; however current tools for bio-imaging in the Gram-positive bacterium Staphylococcus aureus has necessitated further developments for fluorescence intensity and for a multicolor palette of fluorescent proteins. To enhance the expression of multicolor fluorescent proteins in clinical S. aureus strains, we developed new fluorescent protein expression vectors, containing the blaZ/sodp promoter consisting of the β-lactamase gene (blaZ) promoter and the ribosome binding site (RBS) of superoxide dismutase gene (sod). We found S. aureus-adapted GFP (GFPsa) driven by the blaZ/sodp promoter was highly expressed in the S. aureus laboratory strain RN4220, but not in the clinical strains, MW2 and N315, harboring the endogenous blaI gene, a repressor of the blaZ gene promoter. We therefore constructed a constitutively induced blaZ/sodp promoter (blaZ/sodp(Con)) by introducing substitution mutations into the BlaI binding motif, and this modification allowed enhanced expression of the multicolor GFP variants (GFPsa, EGFP, mEmerald, Citrine, Cerulean, and BFP) as well as codon-optimized reef coral fluorescent proteins (mCherry and AmCyan) in the S. aureus clinical strains. These new fluorescent probes provide new tools to enhance expression of multicolor fluorescent proteins and facilitate clear visualization of clinical S. aureus strains.

well as to establish a S. aureus infection model using fluorescent proteins, in which optimization of codon usage and replacement of the region surrounding the ribosome binding sequence (RBS) have been reported to enhance the expression of the fluorescent proteins in S. aureus [12][13][14][15][16] . However, the current methods often entail several limitations as to color palette and its brightness and therefore necessitated the development of new fluorescent vectors that can efficiently enhance fluorescence intensity and the multicolor palette in S. aureus strains. Many proteins are generally difficult to highly express in heterologous host organisms. In E. coli, extensive research has established many different types of promoters and a great number of engineered laboratory strains for heterologous protein production 17 . However, a well-established method for the heterologous protein expression is still lacking in S. aureus, much less clinically isolated strains. Previous studies have focused on the β-lactamase gene (blaZ) promoter P blaZ for the gfp expression in S. aureus. Notably, the GFPmut2 was constitutively expressed in the laboratory strain RN4220 [18][19][20] . This study therefore developed fluorescent protein expression vectors with the P blaZ promoter to highly express GFP and RCFP variants in clinically isolated S. aureus.
Here, we describe novel fluorescent protein expression vectors, which were shown to exhibit greater fluorescence intensity in clinical S. aureus strains. To improve the expression of an exogenous fluorescent protein in S. aureus clinical strains, we used the blaZ/sodp(Con) promoter and codon-optimized fluorescent protein genes, and showed that the fluorescence intensity in the series of GFP and RCFP variants were significantly enhanced in the clinical strains. These new tools efficiently expressing fluorescent proteins in clinical S. aureus strains are valuable for understanding the pathogenic mechanism of S. aureus.

Results
Adaptation of the green fluorescent protein to S. aureus. We aimed to develop fluorescent protein expression vectors that can produce sufficient fluorescence intensity to visually identify fluorescing colonies of S. aureus on agar plates with the naked eye. S. aureus strain RN4220 colonies containing pS1GFP in which GFPmut3b was expressed under the control of the β-lactamase gene (blaZ) promoter exhibited faint fluorescence on TSB agar plates (Fig. 1b). When we replaced the 13-bp sequence containing RBS sequence of the blaZ gene with the corresponding region of the superoxide dismutase gene (sod) (Fig. 1a) that has been reported to enhance the expression of fluorescent protein in S. aureus 13 , the fluorescent activity of the colonies expressing GFPmut3b was significantly improved on the agar plates (Fig. 1b). Further, the fluorescence intensity increased by an approximately 10-fold using the microplate assay reader (Fig. 1c). The high expression of GFPmut3b was confirmed as a major protein band corresponding to its molecular weight, 26.8 kDa, in SDS-PAGE (Fig. 1d). Although the GFPmut3b is one of the faster folding GFP variants that has been optimized for bacteria and has minimal toxicity 21 , overexpression of GFPmut3b in S. aureus showed adverse effects such as cell growth inhibition, producing smaller colonies, and frequent co-occurrence of larger colonies exhibiting no fluorescence signal (Fig. S1a). To overcome these adverse effects, GFPmut3b was modified by introducing Cycle3 mutations (F99S/M153T/V163A) that are known to improve the solubility and reduce the toxicity 9, 22 , yielding S. aureus-adapted GFPmut3b, GFP sa (S65G/S72A/F99S/M153T/V163A). As a result, the mutations antagonized the appearance of non-fluorescent colonies and maintained high levels of the fluorescent intensity (Figs 1b,c and S1b) and GFP sa expression (Fig. 1d) without affecting its excitation and emission wavelengths (Fig. 1e,f). Further, confocal laser scanning microscopic analysis revealed that almost all of S. aureus cells expressed GFP sa protein and were clearly visualized (Fig. 1g).
Introducing suppressor mutations into the blaZ gene promoter. S. aureus-adapted GFP sa demonstrated higher expression under the control of the hybrid blaZ/sodp promoter in laboratory strain RN4220. However, in clinical strains, MW2 and N315, with the same vector, the expression of GFP sa was very low, and the fluorescence intensity was poor (Fig. 2b,c,d). BlaZ plasmid is prevalent in clinical strains 23,24 , and the blaIRZ gene operon is natively present on a plasmid in both clinical strains, MW2 and N315, but not in RN4220 [25][26][27] . The BlaI and MecI specifically bind to the same dyad symmetry (TACA/TGTA) sequence and repress the blaZ promoter activity (Fig. 2a) 28 . To minimize the possible negative effect by the endogenous BlaI/MecI on the blaZ/sodp promoter activity, we introduced suppressor mutations into the BlaI/MecI binding motif (TACA/TGTA), yielding a constitutively induced blaZ/sodp(Con) promoter (Fig. 2a). As a consequence, GFP sa expression was significantly elevated, and the resulting fluorescence intensity was highly improved in both N315 and MW2, whereas the fluorescence intensity was slightly reduced in strain RN4220, compared with the parental sequence (Fig. 2b,c,d).
Taken together, the data demonstrated the modified blaZ/sodp promoter, (blaZ/sodp(Con)) can effectively enhance the expression of GFP sa in S. aureus clinical strains.

Expression of multicolor GFP variants in clinical strains.
A. victoria GFP and its variants contribute to multicolor imaging with spectral profiles ranging in color from blue to yellow, and have been engineered by site-directed mutagenesis [29][30][31][32][33][34] . Using GFP sa as the standard, we next constructed these GFP color variants (EGFP, pFK51, and pFK52 in S. aureus RN4220. SDS-PAGE and Western blot analysis show the relative quantities of the GFP in the whole cell lysates. The Western blot gel was cropped and the full-length image is included in Supplemental Fig. 5(a). (e) Excitation and (f) Emission spectra for GFPmut3b and GFP sa in S. aureus RN4220. Each spectrum was normalized to a maximum value of 1. Excitation spectra were recorded with emission at 540 nm. Emission spectra were recorded with excitation at 460 nm. GFPmut3b and GFP sa were depicted by a solid line and dotted line, respectively. (g) S. aureus RN4220 containing pMK4blaZ/sodpGFP sa (pFK52) was grown on TSB containing chloramphenicol and visualized using an FV1000 confocal scanning laser microscope (Olympus). Panels show (1) DAPI, (2) GFP, (3) overlay of DAPI and GFP images, (4) differential interference contrast (DIC), (5) overlay of DAPI, GFP, and DIC images. mEmerald, Citrine, Cerulean, and BFP) using amino acid substitutions considering the codon usage patterns in S. aureus. Since several BFP variants were engineered using substitution of the T66H mutation; we constructed three BFP variants, EBFP (F64L/S65T/Y66H/Y145F), 1EMF (F64L/Y66H/V163A), and P4-3E (F64L/Y66H/Y145F/ V163A) and evaluated which BFP variants would be more suitable for S. aureus MW2 and N315. Fluorescence assay showed that homogenates of P4-3E and EBFP exhibited somewhat greater fluorescence intensity than 1EMF when S. aureus MW2 were grown in TSB medium (Fig. S2a), while only S. aureus expressing P4-3E was visually identified as fluorescing colonies on agar plates by the naked eye (Fig. S2b). Each strain expressing the multicolor GFP variant exhibited sufficient fluorescence intensity to visually identify fluorescing colonies by the naked eye ( Fig. 3a), and SDS-PAGE and Western blot analyses showed that like GFP sa all of these multicolor variants were expressed with high yield in S. aureus MW2 harboring the corresponding vectors with the blaZ/sodp(Con) promoter (Fig. 3b,c). The multicolor GFP variant genes driven by the blaZ/sodp(Con) promoter also exhibited high fluorescence intensity in another clinical strain, N315 (Fig. S3a). Further, fluorescence excitation and emission spectrum analysis showed these strains expressing the multicolor GFP variants reflected the previously published spectrum profiles (Fig. S4). This data indicated that multicolor GFP variants were highly expressed in clinical strains MW2 and N315 with the blaZ/sodp(Con) promoter.

Codon optimization of AmCyan and mCherry for S. aureus.
Like Aequorea GFP, reef coral fluorescent proteins (RCFP) and their variants also contribute to multicolor imaging with spectral profiles ranging in color from cyan to far red 1, 5 . We constructed the RCFP expression vectors for S. aureus clinical strains, in which the amCyan or mCherry gene was expressed under the control of the hybrid blaZ/sodp(Con) promoter. Unexpectedly, S. aureus expressing the commercially available amCyan or mCherry gene formed faintly fluorescent colonies on agar plates (Fig. 4a), we therefore optimized the codon usage of amCyan and mCherry genes by replacing rare GC-rich codons with AT-rich ones. As a consequence, the codon-optimized AmCyan (AmCyan(S.a)) and mCherry (mCherry(S.a)) exhibited sufficient fluorescence intensity to visually identify fluorescing colonies on agar plates by the naked eye (Fig. 4a), and the corresponding proteins were confirmed to their molecular weights, 25.2 kDa and 26.7 kDa, respectively (Fig. 4b) and detected by Western blot (Fig. 4c). The fluorescence assay shows the codon adapted genes resulted in increased fluorescence intensity of AmCyan and mCherry (Fig. 4d,e). The amCyan and mCherry genes driven by the blaZ/sodp(Con) promoter also exhibited a significant fluorescence intensity in strain N315 (Fig. S3b). Further, fluorescence excitation and emission spectrum analysis showed these strains expressing AmCyan or mCherry reflected previously published spectrum profiles (Fig. S4). Taken together, the expression of AmCyan and mCherry were significantly improved in S. aureus clinical strains. Identification of S. aureus cells expressing a specific fluorescent protein in co-culture systems. We next aimed to discriminate S. aureus cells expressing a specific fluorescent protein from the co-existence of bacteria expressing different fluorescent proteins in a co-culture experiment. Confocal laser scanning microscopic analysis revealed that each S. aureus clinical strain expressing different fluorescent proteins was selectively detected in the co-culture system in which Cerulean, Citrine, and mCherry(S.a) were differently expressed in the strain N315, TY34 35 , and MW2, respectively (Fig. 5a). Further, S. aureus N315 cells expressing mCheery(S.a) were clearly distinguished from E. coli DH5α cells expressing EGFP (Fig. 5b). These results demonstrated that our fluorescent protein expression vectors have the potential to sensitively detect particular S. aureus cells in co-culture with different strains or other bacteria.

Discussion
In this study, we developed new fluorescent protein vectors to express a bright fluorescence intensity enough to visually identify fluorescent colonies of S. aureus macroscopically on agar plates. Practically available high-level expression of fluorescent protein has not been achieved in previous studies in which fluorescent proteins were detected only with fluorescence microscopy [10][11][12][13][14] . Our findings provide new applications to enhance the expression  of multicolor GFP variants and would allow us to sensitively detect particular S. aureus cells in bacterial populations and in animal infection models.
To enhance fluorescent brightness, we replaced the sequence from the Shine-Dalgarno (SD) sequence to the start codon with sod that was previously reported with the sod RBS to lead to the highest fluorescence intensity with the sarA promoter 13 . Consequentially, we not only modified the SD sequence of blaZ gene to an optimal SD sequence (AGGAGG), but also altered the distance from the SD sequence to the start codon. The distance between the RBS and the start codon is also known to affect the efficiency of translation initiation 36 . Our results support the concept that the improvement of the sequence from the SD sequence to the start codon sequence significantly enhances the expression of the GFP variant, although the exact mechanism responsible for this effect remains unclear (Fig. 1b). However, recent studies suggest the interaction among the RBS, the initiation codon, the 5′-coding region in translation initiation, and RNA secondary structure at the 5′ terminus affects the protein expression in bacteria 37,38 . Therefore, the enhancement of GFP production may be accounted for by the decrease of free fold energy of the 5′ end of mRNA transcripts.
At the beginning of this study, we evaluated GFP sa expression vectors with the blaZ/sodp promoter in a laboratory strain, RN4220. This strain is easily genetically manipulated; however, it carries a number of genetic mutations that may affect the virulence of the strain 20,27 . Like RN4220, laboratory strains may not be suitable for the evaluation of pathogenesis, because laboratory strains often lack important pathophysiological characters 39 . Hence, clinical strains should be used to properly evaluate the pathogenesis and virulence of S. aureus. S. aureus harboring the bla gene locus appeared in the 1940s after the introduction of penicillin; and at present, most clinical isolates carry the blaIRZ gene on plasmids or on the chromosome 23,40,41 . Because the GFP sa expression driven by the blaZ/sodp promoter was not detectable in S. aureus clinical strains, MW2 and N315 (Fig. 2), we hypothesized the endogenous BlaI may inhibit the transcription of the blaZ/sodp promoter in trans; and we therefore generated the constitutively induced blaZ/sodp(Con) promoter to overcome the limitation of the clinical strains. The blaZ/sodp(Con) promoter can be adapted to enable expression of not only fluorescent proteins but also various exogenous proteins (toxins) from clinical S. aureus strains.
Codon optimization frequently plays a key role in exogenous protein expression 37,42 . The GC contents of human codon-optimized amCyan and mCherry are 46.8% and 62.5%, respectively, showing strong preferences for G + C at the third codon position is distinct from the codon usage patterns in S. aureus with 33.5% GC content. Exchanging the 92 nucleotides of amCyan gene and 76 nucleotides of mCherry gene, respectively, both fluorescence intensities were significantly improved overcoming the codon usage bias in S. aureus. However, SDS-PAGE analysis showed the level of RCFPs production did not reach the level of GFP variants (Figs 3c and 4b). These results suggest the possibility that further investigation with these vectors could improve the expression level of RCFPs by further adapting of codon usage or decreasing the free folding energy of the initial 5′-coding region.
In summary, we have developed new multicolor fluorescent protein vectors that efficiently enhance fluorescence intensity in S. aureus clinical strains where the greater fluorescence intensity of multicolor fluorescent proteins may facilitate clear visualization of S. aureus clinical strains. Ultimately, these findings may help in better understanding the pathogenic mechanisms of S. aureus.

Methods
Bacterial strains, plasmids and growth conditions. The bacterial strains and plasmids used in this study are in Table 1. S. aureus and E. coli were cultured at 37 °C with shaking at 140 rpm in test tube (25 mm × 150 mm) containing 3 ml of trypticase soy broth (TSB) (Becton, Dickinson and Company) or 3 ml of lysogeny broth (5 g yeast extract, 10 g polypeptone, 10 g NaCl per liter; pH 7.2), respectively. Ampicillin (Amp, 100 µg/ml) and chloramphenicol (Cp, 10 µg/ml) were added to the medium if necessary. The plasmids encoding AmCyan or mCherry were purchased from Clontech, TaKaRa Bio Inc., Japan.

Improvement of GFP expression vector using site-directed mutagenesis. To increase expression
of fluorescent proteins in S. aureus clinical strains, we improved a GFP expression vector based on pS1GFP carrying GFPmut3b gene (a kind gift from Prof. M. Krönke) 19 . The hybrid blaZ/sodp promoter was constructed by replacing the sequence containing the RBS in the blaZ gene in the sod gene with blaZPR and GFP-F primers using the inverse PCR method with KOD Plus Neo DNA polymerase (TOYOBO, Japan). The PCR product was phosphorylated with T4 Polynucleotide Kinase (TaKaRa Bio Inc, Japan), and circularized by self-ligation with Ligation high Ver.2 (TOYOBO, Japan); and then the circular DNA was transformed into E. coli DH5α. Plasmid DNA was extracted from the transformed E. coli DH5α using FastGene TM Plasmid Mini Kit (Nippon Genetics Co., Ltd. Japan) and the resultant plasmid was verified using ABI 3130 DNA sequencer (Applied Biosystems). To express the fluorescent protein in S. aureus clinical strains, the substitution of the repressor BlaI/MecI binding motif (TACA/TGTA) was performed with the blaZ-mutF and blaZ-mutR primers using inverse PCR as above, yielding the constitutively induced blaZ/sodp(Con) promoter.
Construction of the multicolor GFP variants using amino acid substitution. Amino acid substitution was performed using inverse PCR with pMK4blaZ/sodp(Con)GFPmut3b (pFK53) as the initial template with the KOD Plus Neo DNA polymerase (TOYOBO, Japan). The primers described in Table 2 were used to construct the GFP variants ( Table 3). The resulting plasmids were confirmed using DNA sequencing.

Construction of codon-optimized RCFP expression vectors.
The amCyan and mCherry genes were combined with the hybrid blaZ/sodp(Con) promoter using overlap extension PCR and cloned into the pKAT vector 43 . Plasmid pKAT derived from pND50 44 is an E. coli-S. aureus shuttle vector containing the replication origins of pUB110 (S. aureus) and the pUC19 (E. coli) lacZ(α) gene from pUC19. This enables a simple blue-white screening for clones in E. coli, and enables the cat gene conferring resistance to chloramphenicol in both E. coli and S. aureus. All restriction enzyme sites in the multiple cloning site (MCS) located in the lacZ(α) gene can be used for cloning into pKAT. In the first PCR, the hybrid blaZ/sodp(Con) promoter region, amCyan gene, and mCherry gene were amplified from pFK53, pAmCyan (Clontech, TaKaRa Bio Inc., Japan), and pmCherry (Clontech, TaKaRa Bio Inc., Japan) with the following primer sets (AmCyan: blaZp-F and blaZPR-Cyan and blaZP-CyanF and pUC-RH; mCherry: blaZp-F and blaZPR-Cherry and blaZP-mCherryF and pUC-RH), respectively. The second PCR was performed with the mixture of two PCR fragments as the template using the primer set (blaZp-F and pUC-RH). The resulting PCR products were digested with HindIII and cloned into the same site in pKAT. Codon optimization was then repeatedly performed using inverse PCR as mentioned above with the primers described in Table 2 that were designed to optimize the codon usage of amCyan and mCherry gene using the Kazusa Codon Usage Database (http://www.kazusa.or.jp/codon/). The DNA sequences of codon-optimized amCyan and mCherry genes have been submitted to the GenBank and are available under accession numbers LC088723 (amCyan) and LC88724 (mCherry).

Transformation of S. aureus and quantification of fluorescence intensity.
In brief, S. aureus was individually transformed using electroporation as described previously 45 . Each plasmid was first transformed into S. aureus RN4220 20 and selected as chloramphenicol-resistant colonies, then the resulting modified plasmids were isolated and electroporated into S. aureus MW2 and N315. S. aureus was grown as described in the growth conditions. Bacterial cells from overnight cultures were washed with phosphate-buffered saline (PBS) and re-suspended to an optical density at 660 nm of 0.2. The cell suspension was then dispensed into triplicate wells (100 µl/well) of a U-bottom 96 well cell culture plate (Greiner Bio-One). The fluorescence intensity was measured using a Varioskan Flash Multimode Reader (ThermoFisher Scientific Inc.) with two independent samples.

Quantification of fluorescent proteins using SDS-PAGE and Western blot analysis. Fluorescent
proteins were detected from whole cell lysates. S. aureus was cultured overnight as described in the growth conditions. The pre-cultured cells were adjusted to an optical density at 660 nm of 0.02 with TSB with chloramphenicol, and 3 ml of the culture were transferred into test tube (25 mm × 150 mm) and then incubated at 37 °C with shaking at 140 rpm for 16 h. The bacterial cells from 1 ml cultures were harvested by centrifugation, and then the whole cell lysates were prepared as follows: cells were re-suspended in 200 μl CS buffer (100 mM Tris-HCI,   Confocal laser scanning microscopic analysis. S. aureus strains and E. coli were cultured in test tube (25 mm × 150 mm) containing 3 ml of Brain Heart Infusion (BHI) broth with chloramphenicol or ampicillin at 37 °C with shaking at 140 rpm overnight. The pre-cultured cells were mixed and diluted 1:000 into 3 ml fresh BHI, and then incubated at 37 °C with shaking at 140 rpm for 6 h. Bacterial cells were harvested by centrifugation, washed twice with PBS buffer, fixed with 3% parafomaldehyde in PBS, and washed twice with PBS. Bacterial cells were counterstained with 0.1 μg/ml of 4′,6′-diamidino-2-phenylindole (DAPI) if necessary. The cell suspensions were spotted onto glass slides and air-dried at room temperature. The coverslips were mounted using VECTASHIELD H-1000 (Vector Laboratories, Inc. Burlingame, CA). All confocal images were recorded using an confocal laser scanning microscopy (Olympus, Fluoview FV1000, Japan). GFPmut3b  S65G, S72A  21   GFPsa  S65G, S72A, F99S, M153T, V163A  This study   EGFP  F64L, S65T  4   mEmerald  F64L, S65T, S72A, N149K, M153T, I167T, A206K  4   Citrine  S65G, S72A, V68L, Q69M, T203Y  30   Cerulean  F64L, S65T, Y66W, S72A, Y145A, N146I, H148D, M153T, V163A Table 3. GFP variants used in this study.