Identification of an N-terminal tag (580N) that improves the biosynthesis of fluorescent proteins in Francisella tularensis and other Gram-negative bacteria

Abstract

Utilization of fluorescent proteins is widespread for the study of microbial pathogenesis and host-pathogen interactions. Here, we discovered that linkage of the 36 N-terminal amino acids of FTL_0580 (a hypothetical protein of Francisella tularensis) to fluorescent proteins increases the fluorescence emission of bacteria that express these recombinant fusions. This N-terminal peptide will be referred to as 580N. Western blotting revealed that the linkage of 580N to Emerald Green Fluorescent Protein (EmGFP) in F. tularensis markedly improved detection of this protein. We therefore hypothesized that transcripts containing 580N may be translated more efficiently than those lacking the coding sequence for this leader peptide. In support, expression of emGFP_Ft that had been codon-optimized for F. tularensis, yielded significantly enhanced fluorescence than its non-optimized counterpart. Furthermore, fusing emGFP with coding sequence for a small N-terminal peptide (Serine-Lysine-Isoleucine-Lysine), which had previously been shown to inhibit ribosomal stalling, produced robust fluorescence when expressed in F. tularensis. These findings support the interpretation that 580N enhances the translation efficiency of fluorescent proteins in F. tularensis. Interestingly, expression of non-optimized 580N-emGFP produced greater fluorescence intensity than any other construct. Structural predictions suggested that RNA secondary structure also may be influencing translation efficiency. When expressed in Escherichia coli and Klebsiella pneumoniae bacteria, 580N-emGFP produced increased green fluorescence compared to untagged emGFP (neither allele was codon optimized for these bacteria). In conclusion, fusing the coding sequence for the 580N leader peptide to recombinant genes might serve as an economical alternative to codon optimization for enhancing protein expression in bacteria.

1. Introduction

Fluorescent proteins are extremely useful molecular tools that allow investigators to explore protein trafficking, reporter gene expression, protein-protein interaction, and many other dynamic properties of proteins [1–4]. Moreover, expression of fluorescent proteins by bacteria facilitates the detection of these microorganisms during pathogenesis [5,6]. Additionally, many important host-pathogen interactions have been revealed through the labeling of both mammalian and microbial proteins with contrasting fluorescent molecules [7,8].

Francisella tularensis is a category A bioterrorism agent due to its low infectious dose, <10 colony forming units (CFU), and high mortality rate among infected individuals [9–12]. Furthermore, F. tularensis is an intracellular pathogen capable of replicating within multiple host cell types including macrophages, neutrophils, epithelial cells, and hepatocytes; this bacterium can also invade erythrocytes [13–17]. Therefore, engineering this bacterium to stably express fluorescent proteins could provide insight into its pathogenesis and host-pathogen interactions.

In 2008, the Francisella glucose-repressible promoter (FGRp) located upstream of FTL_0580 was characterized and used to drive gene expression of tdtomato (encodes a red fluorescent protein [3,18]) from a shuttle plasmid [14]. This construct produced robust red fluorescence in F. tularensis during intracellular infection of host macrophage cells [14]. Prior to this aforementioned study, the only other promoters used for ectopic gene expression in Francisella were those of acpA (the respiratory-burst-inhibiting acid phosphate gene), the promotor of FTN_1451 (an uncharacterized gene from F. novicida), and the strong groE promoter [6].

In a previous study by Zaide et al., 2011, data suggested that while groEp was classified as a strong promoter, it failed to produce measurable or useful expression levels of green fluorescent protein (GFP) within F. tularensis [6]. Intriguingly, only the strongest promoter identified (Pbfr) was capable of producing visible green fluorescence while driving expression of gfp by Francisella bacteria cultured on an agar plate [6]. These results suggest that factors additional to gene expression contribute to the production and/or maintenance of stable fluorescent proteins in Francisella.

Even with the development of molecular tools and controllable promotor systems, the expression of recombinant proteins by bacteria can still be sub-optimal. One strategy to overcome meager recombinant protein expression is to utilize a peptide tag fusion [19–21]. For instance, the linkage of the secretory signal peptide from a Bacillus thuringiensis insecticidal protein (Cry1Ia; Iasp) to eGFP (enhanced GFP) led to increased expression of this fluorescent protein [22]. Similarly, another study showed that the insertion of sequence encoding for Serine-Lysine-Isoleucine-Lysine (SKIK) adjacent to the start codon enhances protein expression in Escherichia coli [23]. The SKIK peptide leads to increased protein expression by inhibiting ribosomal stalling by interfering with arrest peptides thereby enhancing translation [24]. Therefore, the identification and characterization of additional unique peptide fusions that enhance protein expression in bacteria will not only augment our ability to express recombinant proteins, but will also increase our knowledge of fundamental molecular interactions.

Here we show that fusing the coding sequence for a small N-terminal segment of FTL_0580 (580N) and emGFP (Emerald Green Fluorescent Protein [25] or tdTomato substantially increases the fluorescence emission of the bacteria producing the encoded fluorescent proteins. Moreover, this fusion is required for detection of EmGFP in F. tularensis by western blotting. Evidence presented here suggests that 580N enhances translation in F. tularensis when using coding sequence for fluorescent proteins that had not been codon optimized. The enhanced signal afforded by linking 580N to fluorescent proteins was also observed in E. coli and Klebsiella pneumoniae. In summary, the addition of 580N represents a simple and straightforward strategy to enhance the expression fluorescent proteins and perhaps other recombinant proteins in bacteria.

2. Material and methods

2.1. Bacterial strains and growth conditions

From frozen stock cultures, Francisella tularensis live-vaccine strain (LVS; Table 1) was initially cultured on chocolate II agar plates and incubated at 37 °C with 5% CO₂ for 3–4 days [13,26,27]. For cultivation in broth, Tryptic Soy Broth (BD Difco) supplemented with 0.1% cysteine (TSBc) was used [28,29]. F. tularensis LVS broth cultures were incubated at 37 °C for 12–18 h with shaking. Escherichia coli NEB5α (Table 1) was used for genetic manipulations and for fluorescent protein expression studies. E. coli and K. pneumoniae (ATCC 33495) bacterial strains were cultivated on Luria-Bertani (LB) agar, or LB broth (BD Difco). Inoculated LB agar plates were incubated at 37 °C for 24 h, and LB broth cultures were incubated at 37 °C for 12–18 h with shaking. For strains containing plasmid vectors, antibiotics were added to the media; for F. tularensis LVS kanamycin was used at 10 μg/ml; for E. coli, 100 μg/ml ampicillin and 35 μg/ml kanamycin was used; for K. pneumoniae, 50 μg/ml of neomycin was used for selection.

Table 1.

Bacterial strains, plasmids, and primers used in this study. 5Phos, 5′ phosphate.

	Description
Strains
Francisella tularensis LVS	Francisella tularensis subsp. holarctica live vaccine strain, a gift from Karen Elkins
Escherichia coli NEBSα	Competent E. coli used for the uptake of foreign DNA, New England Biolabs
Klebsiella pneumoniae	ATCC 33495; strain C122 that was originally isolated from a human urinary tract
Plasmids
pTC3D	pF8tdTomato with DNA (n.t. 557161 to 558861; 580N) from the LVS chromosome cloned upstream of tdtomato. [14]
pSC3	pFNLTP8 containing emGFP under control of groE promoter
pSC11	FGRp (56 bp minimum sequence) created using pSig1 and 2 oligos annealed together and cloned into the KpnI EcoRI of pFNLTP8 [14].
pSC13	pFNLTP8 containing emGFP (promoterless) [26]
pSC18	Plasmid containing emGFP under control of FGRp and lacking 580N. emgfp amplified with SC18/19 cloned into the NdeI BamHI site of pSC11 [26].
pSC34	tdtomato cloned into the NdeI BamHI site of pSC11
pKHEG	Plasmid containing emGFP under control of FGRp with 580N
pMKC1	pFNLTP8 containing emGFP under the control of FGRp (the equivalent portion of this promoter present) in pKHEG.
pKG24C	Plasmid containing emGFP under control of groEp with 580N (correct orientation)
pGRP	E. coli, Francisella shuttle vector containing the FGRp promotor adjacent to the multicloning site [14].
pABST	E. coli, Francisella shuttle vector containing the groEp promotor adjacent to the multicloning site [29].
pMKC2	emGFP that was codon-optimized for F. tularensis (emGFPFt) under the control of FGRp
pMKC3	580N-emGFP that was codon-optimized for F. tularensis (580N-emGFPFt) under the control of FGRp
Primers and oligonucleotides
SC18	5′ GGTGGTCATATGGTGAGCAAGGGCGAG 3′
SC19	5′ GGTGGTGGATCCTTACTTGTACAGCTCGTCCATG 3′
FGRp + 580N F	5′ CATGGGTACCCTATAGTTCTTCATCATACTTATGCTGCTCTTG 3′
FGRp + 580N R	5′ CATGCATATGGGTGGCGGGATCTTCAT 3′
emGFP F	5′ GCACAAGCTGGAGTACAACTA 3′
emGFP R	5′ GATGTTGTGGCGGGTCTT 3′
lpnA F	5′ CAGGTTTAGCGAGCTGTTCTA 3′
lpnA R	5′ CTTGCTCAGTAGTAGCTGTCTG 3′
580N F	5′ CATGGAATTCATGTTAGAACGTCATCCATTAGTCAAA G 3′
580N R	5′ CATGGAATCCGGTGGCGGGATCTTCAT 3′
LeaderF	5′ CATGGGTACCCTATAGTTCTTGATGATAGTTATGCTCTTGGCGAA 3′
Full-prom-R-NdeI	5′ ACGTCATATGTTAAATGCTCCTAGCAATTTTTGATTTATAACTTAAC 3′
CO-Eco-F	5′ CCAGAGCTCGTCAGGAATTC 3′
CO-Bam-R	5′ CTGTCGGATCCTTATTTATATAATTCATC 3′
Full-prom-R-EcoRI	5′ ATCGGAATTCTTAAATGCTCCTAGCAATTTTTGATTTATAACTTAAC 3′
SKIK-top	5’/5Phos/TAATGAGTAAAATTAAACA 3′
SKIK-bottom	5’/5Phos/TATGTTTAATTTTACTCAT 3′

2.2. Plasmid construction

Plasmids and primers used in this study are listed in Table 1. All enzymes were purchased from New England Biolabs. To generate a vector in which native tdtomato (FPbase ID: PGG5S) was expressed from FGRp [14], the tdtomato gene was amplified from pTC3D [14] using primer pair SC18/19 and cloned into the NdeI BamHI site of pSC11 [26]. This yielded pSC34 (Table 1).

To generate a plasmid encoding 580N-EmGFP under the control of FGRp, FGRp and the N-terminal domain of FTL_0580 (580N) sequences were amplified from pTC3D using the 580N + FGRP F and 580N + FGRP R primers (Table 1). This amplicon was cloned into the KpnI and NdeI sites of pSC13 to produce pKHEG. To generate a plasmid encoding emGFP (FPbase ID: GLAD6) under the control of FGRp, FGRp was amplified from pTC3D using primers LeaderF and Full-prom-R-NdeI (Table 1). This amplicon was cloned into the KpnI and NdeI sites of pSC13 to produce pMKC1 (Table 1).

To create a vector which has emGFP under control of the groEp promoter from F. tularensis, the emGFP gene from pRSET/EmGFP (Invitrogen) was excised by restriction with EcoRI and SalI and cloned into pABST [29] digested with the same enzymes. The resulting plasmid was designated pSC3.

To create a plasmid encoding 580N-emGFP under the control of the groEp, FGRp and the N-terminal domain of FTL_0580 (580N) sequences were amplified from pTC3D using the 580N F and 580N R primers and this amplicon was cloned into the EcoRI site of pSC3 to produce pKG24C. Proper orientation of this insert was confirmed (data not shown).

To create a vector in which coding sequence for SKIK was cloned adjacent to the start codon of emGFP, pMKC1 was digested with KpnI-HF, treated with rSAP, and ligated with the oligonucleotides SKIK-top and SKIK-bottom. This generated the plasmid pHR1. Confirmation of the correct orientation of the sequence encoding SKIK was conducted through the generation of an amplicon by PCR using SKIK-top as the forward primer and emGFP R as the reverse primer.

Once validated by restriction digestion, electrophoresis, and DNA sequencing, plasmids were mobilized into F. tularensis LVS and K. pneumoniae via electroporation [29].

To generate constructs encoding emGFP or 580N-emGFP that had been codon-optimized for F. tularensis, synthetic gBlocks fragments were designed and synthesized by IDT-DNA (sequences available upon request). This material was amplified by PCR using the primers, CO-Eco-F and CO-BamR and the amplified material was cloned into the EcoRI and BamHI sites of pSC11. Subsequently, the FGRp promotor was amplified from pTC3D with LeaderF and Full-Prom-R-EcoRI. This fragment was cloned into the EcoRI and NdeI sites to produce pMC2 (emGFP_Ft) and pMC3 (580N-emGFP_Ft).

2.3. Fluorescence stereoscopy

F. tularensis bacteria cultured on chocolate II agar or E. coli bacteria cultured on LB agar were visualized using an Olympus SZX7 stereoscope. F. tularensis strains not expressing EmGFP were used as a control for background fluorescence and was used to set the darkp oint for all images; the bright point was set for the most intense fluorescence producer in experimental set without saturating the image as we have done previously [26]. The same settings were applied for all images across an experimental set. Moreover, equivalent exposure times were used for all images in a single experiment. Composite images were generated by merging the brightfield and pseudocolored fluorescence channels using cellSens software (Olympus).

2.4. Fluorescence microscopy

Bacteria cultivated in broth to stationary phase were spotted onto 1% agarose pads (in PBS) formed onto the surface of glass slides as we have done previously [26]. Microscopy images were captured using an Olympus IX73 microscope equipped with a 100 × , 1.45-numerical-aperture phase objective and an ORCA-Flash4.0 LT + digital CC11440–42U CMOS camera (Hamamatsu). F. tularensis strains not expressing EmGFP were used as a control for background fluorescence and was used to set the dark point for all images; the bright point was set for the most intense fluorescence producer in experimental set without saturating the image as we have done previously [26]. The same settings were applied for all images across an experimental set. Composite images were produced by merging the phase contrast and pseudocolored fluorescence channels using cellSens software.

2.5. Fluorescence quantification

Broth cultures grown to stationary phase and were normalized to account for minor differences in bacterial numbers (all cultures used here for fluorescence quantification produced comparable culture densities [within 0.2 OD₆₀₀ units]). F. tularensis LVS TSBc cultures that were incubated with aeration at 37 °C for 16–24 h typically reached stationary phase at an OD₆₀₀ of 2.0–3.0 in TSBc when measured with a Spectronic 200 spectrophotometer. Samples were added to 96-well plate and fluorescence was measured using an Eppendorf PlateReader AF2200 (green fluorescence, 485 nm excitation [α20 nm], 535 nm emission [α25 nm] filter set; red fluorescence, 535 nm excitation [α10 nm], 595 nm emission [α10 nm]) or a Bio-Tek Synergy H1 plate reader (green fluorescence, 485 nm excitation, 528 nm emission; red fluorescence, 554 nm excitation, 581 nm emission).

2.6. Gel electrophoresis and western blot

All cultures were grown to stationary phase (OD₆₀₀ = 2.0–3.0) and were normalized to account for minor differences in bacterial numbers (all cultures used here produced comparable culture densities [within 0.2 OD₆₀₀ units]). Normalized cultures were centrifuged at max speed (~17,000×g) for 3 min to collect cell pellets. For SDS-PAGE, the pellets were resuspended in 1X Laemmli +2.5% 2-Mercaptoethanol. Samples were boiled at 105 °C for 10 min and were subsequently loaded onto a 14% polyacrylamide gel and run at 150 V for 45 min utilizing Tris-Hepes-SDS running buffer. Proteins were transferred onto nitrocellulose paper at 250 mA (100V for 20 min). The nitrocellulose paper was blocked using phosphate buffered saline containing 0.5% casein, 0.5% bovine serum albumin, 100 mg/L Phenol Red, and 0.02% Sodium Azide pH 7.4 (killer filler) for 25–30 min [30], then treated with primary antibody (rabbit anti-GFP [Life Technologies], 1:1000 dilution) overnight with shaking at room temperature. The nitrocellulose paper was washed using phosphate buffered saline (PBS, pH = 7.2–7.4) three times at room temperature. The nitrocellulose paper was treated with secondary antibody (goat anti-rabbit IgG [MilliporeSigma], 1:1000 dilution) for at least 1 h at room temperature. The nitrocellulose paper was washed twice with PBS and once with Tris buffer (pH = 8.0). The bands on the nitrocellulose paper were developed until red bands appeared using napthol as-ms phosphate (Research Products International) and fast red tr salt zinc chloride (Aldrich) dissolved in Tris buffer [31,32]. Quantification of band intensity was determined using ImageJ. To confirm loading consistency, separate gels were treated with Imperial^™ Protein Stain (Thermo Fisher) according to the instructions of the manufacturer.

2.7. RNA extraction

Broth cultures that had been grown to stationary phase were normalized to approximate OD₆₀₀ values and pelleted in a microcentrifuge at maximum speed (~17,000×g) for 3 min. The pellets were resuspended in Trizol (Ambion by Life Technologies) and incubated for 5 min at room temperature. Following incubation, chloroform was mixed in until the sample turned a milky pink; samples were transferred to organic matrix tube. The tubes were centrifuged for 15 min at room temperature. Subsequently, the liquid phase (no white pellet) was transferred into a new tube. A 1:1 ratio of 70% EtOH was mixed into the sample; the sample was transferred into spin columns (PureLink, ThermoFisher). Spin columns were centrifuged for 15 s (~17,000×g) at room temperature and the flow through was discarded. To ensure elimination of DNA, PureLink DNase (ThermoFisher) was added to the spin column and incubated for 15 min at room temperature. Samples were washed using Wash Buffer 1 and spun for 15 s. Flow though was discarded and Wash Buffer 2 was added. The samples were spun for 15 s (~17,000×g), then washed again using Wash Buffer 2. Spin columns were transferred to recovery tubes and RNase free H₂O was added to elute the RNA. The columns were centrifuged and RNA concentration was determined using a NanoDrop spectrophotometer. Samples were stored at −80 °C.

2.8. Reverse transcriptase-polymerase chain reaction with SYBR green

RNA samples were diluted to 50 ng/μl and heated at 65 °C for 5 min to disrupt any secondary structure. Diluted RNA was added to 96-well plate containing SYBR green master mix, SuperScript III Platinum Taq Reverse Transcriptase (Invitrogen), and primers specific for either lpnA or emGFP. Initial holding step was 50 °C for 3 min, followed by a 95 °C for 5 min holding step. Following the two holding steps, the thermocycler cycled through 95 °C for 15 s, 51 °C for 15 s, and 72 °C for 1 min, for 50 cycles. Fluorescent multicomponent data was used to estimate Ct values. The ΔΔCt value was calculated according to Livak & Schmittgen, 2001 [33]. Here, we compared the target molecule C_T (emGFP) to the reference gene (lpnA). Following calculation of ΔCt, the values for LVS/pKHEG were compared to LVS/pSC18, allowing the calculation for the ΔΔCt. The ΔΔCt was then normalized to the calibrator value [33].

2.9. Bacterial growth curve and EmGFP stability assay

TSBc cultures that had been inoculated with equivalent levels of bacteria were transferred to a clear-bottom 96-well plate. This plate was incubated at 37 °C with double orbital rotation in a Bio-Tek Synergy H1 plate reader. OD₆₀₀ measurements were recorded at the time points indicated. To determine whether 580N affected EmGFP stability, we used an antibiotic-chase assay [34,35]. Here, F. tularensis LVS bacteria expressing EmGFP_Ft (pMKC2) or 580N-EmGFP_Ft (pMKC3) were incubated in TSBc at 37 °C with double orbital shaking in a Synergy H1 plate reader reader (Bio-Tek). Wells were either treated with gentamicin (50 μg/ml) or were left untreated. Green fluorescence readings (485 nm excitation, 528 nm emission) were measured at the time points indicated. Fluorescence retention was calculated as the level of relative fluorescence produced from the gentamicin-treated bacteria divided by the relative fluorescence produced by the untreated bacteria.

2.10. RNA secondary structure predictions

Coding sequence for the 5’ untranslated region and coding sequence for the N-terminal 65–70 codons was analyzed by the RNAfold web server [36].

2.11. Statistical analyses

All statistical calculations were conducted using GraphPad Prism software.

3. Results

3.1. Fusion with the N-terminal segment of FTL_0580 (580N) enhances fluorescence of TdTomato in F. tularensis

Our laboratory group was interested in generating genetic constructs to produce fluorescence in F. tularensis. We previously published work detailing the plasmid, pTC3D (encodes tdtomato under the control of the Francisella glucose-repressible promoter, FGRp) [14]. pTC3D was originally generated by cloning random F. tularensis LVS genomic fragments upstream of a promoterless-tdtomato [14]. F. tularensis LVS bacteria expressing pTC3D produced robust red fluorescence that was readily detectable in fluorescence microscopy, stereoscopy, or by using a plate reader [14]. Upon designing and producing additional fluorescence constructs, we and others observed suboptimal fluorescence produced by F. tularensis LVS regardless of the type of fluorescent protein being expressed or the strength of the promoter [6,37]. However, since previous studies showed that F. tularensis LVS/pTC3D produced strong fluorescence we decided to reanalyze the sequence of the randomly cloned fragment of F. tularensis DNA in this vector. This analysis revealed that the coding sequence of the 36 amino acid N-terminus of FTL_0580 plus a three amino-acid linker (consisting of proline, alanine, and threonine) was fused to tdtomato in pTC3D (Fig. 1). This fused N-terminal domain will be referred to as 580N throughout this manuscript.

Fig. 1.

Nucleotide and putative amino acid sequence of the N-terminal domain of Francisella tularensis LVS FTL_0580 (580N) fused to tdTomato from pTC3D. The plasmid, pTC3D which encodes the FGRp controlling expression of tdtomato was previously generated by ligating a random Sau3AI F. tularensis genomic fragment into the BamHI site upstream of the coding sequence for this fluorescent reporter [14]. Further analysis of this cloned F. tularensis genomic fragment revealed an in-frame fusion between the coding sequence for the N-terminal region of FTL_0580 followed by 3 additional amino acids (PAT) and tdtomato. This N-terminal domain is referred to as 580N. The start codon of FTL_0580 is indicated in bold, italicized, and underlined once, whereas the start codon for tdtomato is shown in bold, italicized, and double underlined.

To determine whether the sequence encoding 580N affected the fluorescence of TdTomato, we created a plasmid devoid of this sequence that encoded merely the FGRp promotor upstream of tdtomato (pSC34). Interestingly, deletion of the sequence encoding 580N substantially diminished observable red fluorescence in F. tularensis LVS compared to bacteria expressing 580N-tdtomato (Fig. 2A&B). Quantification of the fluorescence produced by F. tularensis LVS expressing these constructs confirmed the observations made using microscopy (Fig. 3). Specifically, in F. tularensis LVS, a construct encoding 580N-tdtomato produced significantly more fluorescence (~50-fold increase) than one encoding tdtomato alone or the empty vector (Fig. 3). Moreover, there was no difference in fluorescence between F. tularensis LVS expressing a construct encoding tdtomato alone and the empty vector control (Fig. 3). Altogether, these data suggest that the linkage of 580N to TdTomato enhances the fluorescence of F. tularensis bacteria expressing this fluorescent protein.

Fig. 2.

Fusion with the coding sequence for 580N to tdtomato is required for stereoscopic and microscopic detection of red fluorescence in F. tularensis LVS. F. tularensis bacteria harboring the constructs indicated were streaked onto chocolate II agar (containing kanamycin), incubated at 37 °C for three days, and were subjected to fluorescence stereoscopy (A). All images taken under the red fluorescent channel were captured using identical settings. The merged images were produced by combining the brightfield and pseudocolored red fluorescence image. To visualize individual bacteria, F. tularensis LVS broth cultures that had been grown to stationary phase at 37 °C with agitation were spotted onto agarose pads formed onto the surface of glass slides (B). All images taken under the red fluorescent channel were captured using identical settings. Composite images were produced by merging the phase contrast and pseudocolored red fluorescence channels. Empty vector, pGRP; FGRp tdtomato, pSC34; FGRp 580N-tdtomato, pTC3D. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3.

580N-fusion with TdTomato is required for detection of red fluorescence in F. tularensis. Bacteria were cultured to stationary phase and fluorescence was measured with a plate reader (554 nm excitation, 581 nm emission), tdtomato expression was driven by FGRp in F. tularensis. Empty vector, pGRP; FGRp tdtomato, pSC34; FGRp 580N-tdtomato, pTC3D. A one-way ANOVA with Tukey’s multiple comparison test was used to analyze the data presented in each panel. P < 0.0001****; F. tularensis LVS empty vector vs. FGRp tdtomato, not significant; F. tularensis LVS empty vector vs. FGRp 580N-tdtomato****; F. tularensis FGRp tdtomato vs. FGRp 580N-tdtomato*****.

An Alphafold structure prediction [38] revealed that FTL_0580 encodes a YdcH Family protein, a protein of unknown function (DUF 465). Because the function of YdcH proteins has not been elucidated, structural predictions will not likely provide any insight into the potential function of 580N. We therefore utilized molecular biology to characterize the impact of 580N in the expression of fused fluorescent proteins in F. tularensis.

3.2. 580N-fusion enhances fluorescence of EmGFP in F. tularensis

To determine whether the enhanced fluorescence observed in the 580N-TdTomato fusion was specific to this particular red fluorescent protein, we generated pKHEG and pSC18 which encode 580N-emGFP and emGFP respectively (each under the control of FGRp). These constructs produced a similar pattern of fluorescence as observed for their red counterparts. Specifically, we observed robust green fluorescence from F. tularensis LVS bacteria expressing 580N-emGFP, but little to no green fluorescence from F. tularensis expressing emGFP alone using microscopy (Fig. 4A&B). Quantification of the fluorescence produced by these bacteria confirmed and extended the observations made using microscopy (Fig. 5A). Specifically, in F. tularensis LVS, a construct encoding 580N-emGFP produced ~9-fold more green fluorescence than one encoding emGFP alone (Fig. 5A). To rule out any influence of 580N expression on the overall health of the bacteria, we analyzed the growth curves of F. tularensis bacteria expressing EmGFP, 580N-EmGFP, or an empty vector. All three strains exhibited comparable growth curves (Fig. S1 in the supplemental material) suggesting that the increased fluorescence mediated by 580N is unlikely to be due to disparity in bacterial numbers or metabolism differences. These data confirm that the 580N fusion enhances the fluorescence intensity of the F. tularensis bacteria expressing these constructs.

Fig. 4.

Fusion with the coding sequence for 580N to emGFP is required for stereoscopic and microscopic detection of green fluorescence in F. tularensis LVS. F. tularensis bacteria harboring the constructs indicated were streaked onto chocolate II agar (containing kanamycin), incubated at 37 °C for three days, and were subjected to fluorescence stereoscopy (A). All images taken under the green fluorescence channel were captured using identical settings. The merged images were produced by combining the brightfield and pseudocolored green fluorescence image. To visualize individual bacteria, F. tularensis LVS broth cultures that had been grown to stationary phase at 37 °C with agitation were used to seed agarose pads formed onto the surface of glass slides (B). All images taken under the green fluorescent channel were captured using identical settings. Composite images were produced by merging the phase contrast and pseudocolored green fluorescence channels. Empty vector, pABST; FGRp emGFP, pSC18; FGRp 580N-emGFP, pKHEG; groEp emGFP, pSC3; groEp 580N-emGFP, pKG24C. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 5.

580N-fusion with EmGFP is required for detection of green fluorescence in F. tularensis. Bacteria were cultured to stationary phase and fluorescence was measured with a plate reader (485 nm excitation, 528 nm emission). emGFP expression was either driven by FGRp in F. tularensis (A) or groEp in F. tularensis (B) Empty vector, pGRP (A) or pABST (B); FGRp emGFP, pSC18; FGRp 580N-emGFP, pKHEG; groEp emGFP, pSC3; groEp 580N-emGFP, pKG24C. A one-way ANOVA with Tukey’s multiple comparison test was used to analyze the data presented in each panel. P < 0.0001****; F. tularensis LVS empty vector vs. FGRp emGFP, not significant; F. tularensis LVS empty vector vs. FGRp 580N-emGFP****; F. tularensis FGRp emGFP vs. FGRp 580N-emGFP***** (A). P < 0.0001****; F. tularensis LVS empty vector vs. groEp emGFP****; F. tularensis LVS empty vector vs. groEp 580N-emGFP****; F. tularensis groEp emGFP vs. FGRp 580N-emGFP***** (B).

To determine whether the inclusion of the 580N sequence led to increased production of the fused fluorescent protein, we conducted western blotting on F. tularensis LVS strains harboring plasmids encoding emGFP or 580N-emGFP, each under the control of FGRp (Fig. 6). The predicted molecular weight of EmGFP is ~26.9 kDa and 580N-EmGFP is predicted to be ~31.6 kDa. While EmGFP alone was difficult to detect from F. tularensis LVS/pMKC1, LVS/pKHEG bacteria expressing 580N-EmGFP produced an obvious doublet band near the predicted molecular weight (Fig. 6A). The presence of the doublet could suggest protein cleavage, modification, or alterations in folding. Imperial staining of an identical SDS-PAGE gel was used to ensure equivalent protein loading (Fig. 6B). Quantification of band intensity clearly showed increased protein levels for 580N-EmGFP vs. EmGFP alone (approximately 22-fold increase; Fig. 6C). Together with the fluorescence quantification (Fig. 5), the western blot data (Fig. 6) suggest that this N-terminal peptide fusion is required to produce robust levels of fluorescent proteins in F. tularensis LVS. This could suggest that 580N either increases protein stability or enhances the synthesis of the linked protein.

Fig. 6.

Western blotting of EmGFP and 580N-EmGFP. Cell lysates were suspended in Laemmli buffer containing β-mercaptoethanol, boiled, and subjected to SDS-PAGE. The proteins were electroblotted onto a membrane. After blocking, this membrane was treated with an anti-GFP antibody, washed, and a secondary antibody conjugated to alkaline phosphatase was used as a probe for detection (A). To ensure equivalent protein levels between each sample, a separate SDS-PAGE gel that was loaded with these same cell lysates was stained with Imperial^™ stain (B). Band intensities from four separate western blots was quantified using ImageJ (C). Data shown represent average signal intensity ±SD. A Student’s t-test was used to analyze western blot quantification data, P = 0.0017.

3.3. 580N does not increase transcription of fused emGFP

We next sought to determine the nature of the observed differences in fluorescent protein expression between F. tularensis LVS/pKHEG (580N-EmGFP) and LVS/pMKC1 (580N). Expression of the fluorescence genes were driven by FGRp in both constructs. FGRp is the native promotor of the source of 580N (FTL_0580) in F. tularensis. We therefore hypothesized that 580N may be enhancing transcription from FGRp. We tested this hypothesis in two ways. Firstly, we generated constructs in which emGFP or 580N-emGFP was driven by a different promotor (groEp). Regardless of the promotor, EmGFP fluorescence was no different than the background levels in F. tularensis unless fused to 580N (Figs. 4 and 5). Specifically, a microscopic analysis revealed that F. tularensis LVS bacteria expressing 580N-emGFP under the control of groEp produced obvious green fluorescence, while LVS expressing untagged emGFP with the same promotor did not (Fig. 4A&B). Because this similar trend was observed using two distinct promotors, it is unlikely that 580N exerts any transcription effect on the associated fluorescence gene. Quantification of the fluorescence produced by these bacteria confirmed the observations made using microscopy (Fig. 5B). Specifically, F. tularensis LVS bacteria expressing 580N-emGFP by using groEp produced four-fold greater fluorescence emission than F. tularensis LVS expressing emGFP alone or the empty vector (Fig. 5B). Moreover, there was no measured difference in fluorescence between F. tularensis LVS bacteria expressing emGFP alone under the control of groEp and the empty vector control (Fig. 5B).

Secondly, to confirm that 580N does not enhance transcription of emGFP, we also conducted real-time reverse transcription PCR (rtPCR) on RNA extracted from F. tularensis LVS bacteria harboring a plasmid that encodes 580N-emGFP or emGFP alone (both under the control of FGRp). Here, we conducted real-time rtPCR using primers specific for emGFP and lpnA (used as a reference gene) (Fig. S2 in the supplemental material). This analysis showed no difference in the level of emGFP expression between these strains further indicating that the impact of 580N on green fluorescence production in F. tularensis is not due to increased transcription.

3.4. 580N enhances translation efficiency of emGFP in F. tularensis

Collectively, these data presented here could suggest that 580N enhances the translation efficiency in F. tularensis. To test this, we synthesized constructs encoding EmGFP or 580N-EmGFP that had been codon-optimized for F. tularensis (pMC2 [emGFP_Ft] and pMC3 [580N-emGFP_Ft). Codon optimization of emGFP_Ft led to a ~4-fold increase in fluorescence compared to F. tularensis LVS bacteria expressing the non-optimized version of this fluorescent protein (Fig. 7A). This suggested that codon selection contributed to the lack of expression of untagged EmGFP in F. tularensis in previous experiments. However, the non-optimized version of 580N-emGFP produced more robust fluorescence than both codon-optimized emGFP_Ft and 580N-emGFP_Ft (Fig. 7A). This indicates that translation efficiency alone is not solely responsible for increased fluorescence produced by fluorescent proteins tagged with 580N in bacteria. Further, F. tularensis LVS bacteria expressing codon-optimized 580N-emGFP_Ft produced reduced fluorescence compared to those expressing the untagged emGFP_Ft indicating that codon optimization abolishes the fluorescence enhancement in bacteria we observed in previous experiments.

Fig. 7.

580N fusion enhances translation efficiency of emGFP. F. tularensis LVS bacteria expressing the EmGFP variant indicated were cultured to stationary phase and fluorescence was measured with a plate reader (485 nm excitation, 528 nm emission). Empty vector, pGRP; EmGFP, pMKC1; 580N-EmGFP, pKHEG (A&B), EmGFP_Ft, pMKC2; 580N-EmGFP_Ft, pMKC3 (A); SKIK-EmGFP, pHR1 (B). A one-way ANOVA with Tukey’s multiple comparison test was used to analyze the data presented in each panel. P < 0.0001 with all groups significantly different from one another (P values not shown) (A&B).

A previous study showed that the N-terminal SKIK peptide enhances the translation of fusion proteins in bacteria by interfering with arrest peptides [24]. Therefore, if ribosomal stalling is the reason why expression of emGFP alone does not produce robust fluorescence in F. tularensis, then fusion of coding sequence for SKIK adjacent to the N-terminal methionine should allow for increased synthesis of the fused fluorescent protein (SKIK-EmGFP). F. tularensis bacteria producing SKIK-EmGFP produced ~3-fold greater levels of green fluorescence than those expressing EmGFP alone (Fig. 7B). This finding indicates that emGFP expression in F. tularensis is hampered by inefficient translation, and further supports that like SKIK, fusion with 580N likely antagonizes ribosomal stalling leading to robust protein synthesis. Notably, F. tularensis bacteria producing 580N-EmGFP produced more green fluorescence than those expressing SKIK-EmGFP (Fig. 7B).

Given the codon optimization data and the finding that the N-ternminal SKIK fusion increased fluorescence of F. tularensis bacteria producing EmGFP, it is unlikely that the increased fluorescence afforded by 580N is due to increased protein stability. To test this, however, an antibiotic-chase study was conducted. Here, F. tularensis LVS bacteria expressing either emGFP_Ft or 580N-emGFP_Ft were treated with gentamicin. Green fluorescence from these cultures was compared to those left untreated to determine whether 580N influences protein stability. This analysis revealed no difference in fluorescence loss in F. tularensis bacteria expressing emGFP_Ft vs. 580N-emGFP_Ft (Fig. S3 of the supplemental material) suggesting that 580N does not affect protein stability of EmGFP.

Because F. tularensis bacteria expressing non-codon optimized 580N-emGFP produced greater green fluorescence intensity than those expressing the codon-optimized emGFP_Ft or 580N-emGFP_Ft alleles, we hypothesized that codon optimization could affect RNA secondary structure. In turn, the RNA secondary structure could influence translation efficiency [39,40]. To predict the secondary structure of transcripts encoded here, we used the RNAfold web server [36]. This analysis showed that the Shine-Dalgarno (SD) region was predicted to be masked on the mRNA from the non-optimized emGFP (Fig. 8). Addition of 580N unveiled this ribosomal binding site (Fig. 8). However, the codon-optimized 580N-emGFP_Ft transcript contains a predicted hairpin juxtaposed to the SD (Fig. 8) which may sterically hinder ribosomal binding [41]. These structural predictions support the hypothesis that codon optimization may affect RNA secondary structure and could negatively impact translation in some instances.

Fig. 8.

RNA secondary structure predictions. Coding sequence for the 5′ untranslated region and coding sequence for the N-terminal 65–70 codons was analyzed by the RNAfold web server. The color of the nucleotide represents the probability of the predicted base-pairing. Arrows indicate the location of the Shine-Dalgamo sequence. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.5. 580N-fusion to EmGFP enhances fluorescence of gram-negative bacteria

To test whether the enhanced fluorescence produced by 580N-EmGFP was applicable to other organisms, we mobilized plasmids encoding this and the unfused fluorescent counterpart into E. coli as well as the opportunistic pathogen, Klebsiella pneumoniae (pKHEG and pSC18 respectively). In both E. coli and K. pneumoniae, the green fluorescence produced from bacteria expressing 580N-emGFP was much more intense than that produced from bacteria expressing emGFP alone (Fig. 9 A&B).

Fig. 9.

580N-fusion with EmGFP enhances green fluorescence in E. coli and K. pneumoniae. Bacteria were cultured overnight to stationary phase and fluorescence was measured with a plate reader (485 nm excitation, 528 nm emission). Empty vector, pGRP; emGFP, pSC18; 580N-emGFP, pKHEG. A one-way ANOVA with Tukey’s multiple comparison test was used to analyze the data. (A) P < 0.0001****; E. coli empty vector vs. FGRp emGFP****; E. coli empty vector vs. FGRp 580N-emGFP****; E. coli FGRp emGFP vs. FGRp 580N-emGFP****. (B) P < 0.0001****; K. pneumoniae empty vector vs. FGRp emGFP, P = 0.05*; K. pneumoniae empty vector vs. FGRp 580N-tdtomato****; K. pneumoniae FGRp tdtomato vs. FGRp 580N-tdtomato*****.

4. Discussion

We identified that the coding sequence for the 36 N-terminal amino acids of FTL_0580 plus a three amino acid linker (PAT) were fused to tdtomato in pTC3D [14]. Removal of the 580N peptide diminished fluorescence to undetectable levels in F. tularensis. Namely, 580N-EmGFP was detectable in F. tularensis while EmGFP was not, regardless of the promotor used to drive the expression of genes encoding these fluorescent proteins. Western blotting revealed clear detection of 580N-EmGFP in F. tularensis, with minimal detectable EmGFP (Fig. 6), consistent with the microscopy and fluorescence quantification data (Figs. 2–5, 7). This increased fluorescence intensity was also observed in E. coli and K. pneumoniae bacteria (Fig. 9). Regarding the protein expression disparity in F. tularensis, we determined that this was not due to transcriptional differences (first by using the alternative groE promotor to drive gene expression and secondly by conducting real-time rtPCR on mRNA extracted from F. tularensis harboring a plasmid either encoding emGFP or 580N-emGFP) (Fig. 5 & Fig. S2). Data presented here indicate that while F. tularensis bacteria produced robust 580N-EmGFP, this microorganism did not synthesize detectable EmGFP protein without the peptide under similar conditions, unless the untagged protein was codon-optimized (Fig. 7). While the disparity in fluorescent protein production by F. tularensis was not due to transcriptional differences, it is most likely that 580N-emGFP transcripts were more efficiently translated than emGFP mRNA [42]. In support of this interpretation, codon optimization of emGFP_Ft led to increased expression over the non-optimized counterpart in F. tularensis (Fig. 7A). This result indicated that translation efficiency was likely responsible for the disparity in fluorescence levels between bacteria expressing the 580N-linked and untagged fluorescent proteins. Likewise, the N-terminal SKIK peptide linked to EmGFP produced robust green fluorescence in F. tularensis (Fig. 7B) likely by overcoming ribosomal stalling, as this has been observed in other bacterial systems [24]. This result further suggests that linkage to 580N enhances the translation efficiency of EmGFP in F. tularensis in a similar mechanism to SKIK. Twenty-five percent of the amino acids in 580N contain basic R groups. Based on the presence of the two lysine residues in SKIK and the high abundance of basic R groups in 580N, it is possible that these positively charged moieties play a role in antagonizing arrest peptides to prevent ribosomal stalling [24]. Finally, 580N did not apparently affect EmGFP protein stability suggesting that this leader peptide did not extend the half-life of the fused fluorescent protein (Fig. S3) [34].

Interestingly, the non-optimized 580N-emGFP produced increased fluorescence compared to the codon-optimized 580N-emGFP_Ft indicating that codon optimization does not always result in pinnacle protein expression (Fig. 7A). This could be due to the alteration of RNA secondary structure [40] and could suggest that gene sequences have evolved to balance codon selection and RNA secondary structure for optimal gene expression (Fig. 8) [39]. Given the cost of codon optimization and the enhanced fluorescence produced by the 580N fusions, incorporating this leader peptide onto recombinant proteins might serve as an economical alternative to enhance recombinant protein expression in F. tularensis and other bacteria.

In our hands, bacteria expressing 580N-linked to fluorescent proteins had no discernible growth defects compared to wild-type strains or those expressing the native fluorescent proteins (Fig. S1 in the supplemental material). Also, we saw no evidence of protein aggregation under the microscope in bacteria producing fluorescent proteins tagged with 580-N, instead uniform diffuse fluorescence was observed.

Whether linkage of 580N to heterologous proteins increases translation efficiency in Gram-positive bacteria or eukaryotes remains to be determined. These possibilities should be the focus of future investigations. However, in the F. tularensis LVS system, 580N may also be useful for the synthesis of heterologous proteins other than the fluorescent variety which could have implications in vaccine design. The live-vaccine strain has been shown to give people long-term immunological memory, ranging up to 30 years [43]. Due to this capability, the potential exists that recombinant F. tularensis LVS expressing heterologous proteins may provide protection not only against F. tularensis, but other targeted pathogenic microorganisms. Robinson et al., 2015, previously used LVS to express Pseudomonas aeruginosa proteins, thus allowing it to potentially immunize against two pathogenic bacteria [29]. However, the production of the heterologous P. aeruginosa proteins by F. tularensis LVS was weak [29]. The addition of 580N linked to proteins from other pathogenic bacteria may allow more robust expression within F. tularensis LVS making potentially more effective vaccine strains. Moreover, the Horwitz group has shown that ectopic expression of bacterial antigens in the recombinant F. tularensis LVS ΔcapB strain has produced protective vaccines against the select agents, Bacillus anthracis and Yersinia pestis [44–46]. We speculate that linking 580N to the coding sequence of vaccine targets may enhance protection afforded by the LVS ΔcapB platform.

Data availability

Data will be made available on request.