Fluorescent proteins (FPs) have been instrumental to our understanding of bacterial pathogenesis [1]. FPs have been used widely in plasmid- or chromosome-based strategies in in vitro and in vivo studies [1,2]. Plasmid-based GFPs generally have higher copy number and expression, benefitting brightness at the cost of cell-to-cell variation (due to different copy numbers in different cells), plasmid instability, and fitness defects due to plasmid carriage or high GFP expression. Fitness defects in particular then complicate studies of pathogenesis, which more often manifest in vivo. In contrast, chromosomal GFPs tend to have lower expression and lower brightness, limiting utility for visualizing small (or individual) bacterial collections while mitigating these other problems with variability, stability, and fitness.

Urinary tract infections (UTIs) are one of the most common bacterial infections of humans, accounting for over $2.3 billion in medical expenditures annually in the US [3]. Most UTIs are caused by strains of E. coli, thus the term uropathogenic E. coli (UPEC). As with other infectious diseases [1,4], fluorescent proteins have been instrumental for many discoveries of the pathogenic mechanisms utilized by UPEC, including the development of intracellular bacterial communities (IBCs) [5,6], quiescent intracellular reservoirs (QIRs) [7], and avoidance of neutrophil killing [6] by the cystitis strain UTI89 [8]. For these studies, two strains are commonly used, both of which express the GFPmut3 variant of GFP [1]: UTI89 carrying plasmid pANT4 [9] and UTI89 att_HK022::COM-GFP [10]. Both of these strains have been used to monitor formation of intracellular structures during UTI by microscopy [7,10,11], but to date UTI89/pANT4 has not been further characterized for other infection phenotypes.

Since the identification of GFPmut3, new variants of GFP demonstrate various improved properties [12]. One of these in particular, superfolder GFP (sfGFP) [13], has higher brightness and faster folding kinetics than the currently used GFPmut3. We have further improved the brightness of sfGFP by fusing it to a GFP-specific single domain antibody [14] using the vGFP strategy to create a monomeric fluorophore with 30%–50% increased brightness and pH resistance [15]. We refer to this improved sfGFP as vsfGFP-9.

We here report the creation of new derivatives of UTI89 carrying vsfGFP-9 on the chromosome or on a derivative of the pANT4 plasmid that provide nearly 10× increased brightness to the commonly used UTI89 att_HK022::COM-GFP and UTI89/pANT4, respectively. We demonstrate that these derivatives, despite the markedly higher brightness, have no fitness defects relative to the strains they are intended to replace. Furthermore, we find that the plasmid-based strains (UTI89/pANT4 and SLC-638) have an equivalent fitness defect relative to UTI89 as measured by infection load. In contrast, chromosomal expression of vsfGFP-9 produces brightness approaching that of UTI89/pANT4 without a defect in IBC formation or infection load. These new, brighter strains should be useful in future studies of the pathogenic mechanisms of UTI89, and the strategies employed here can be similarly applied to improve fluorescent derivatives of other UPEC strains.

Strains and growth conditions. UTI89 [8], UTI89 att_HK022::COM-GFP [10], and UTI89/pANT4 [9,16] have been previously described. Lysogeny broth (LB) was used for all growth. Ampicillin was supplemented as required at 100 μg/mL, and kanamycin at 50μg/mL. For mouse infections, bacteria were grown under type 1 pili inducing conditions as previously described [17]. Bacterial growth curves were measured using a Bioscreen C MBR (Bioscreen, Finland) in LB medium at 37 °C for 24 h.

Construction of sfGFP and vsfGFP-9 expressing strains. All plasmids used in this study are listed in Table S1. All primers used for cloning and homologous recombination were purchased from Sigma (Singapore) and are listed in Table S2. The genes encoding sfGFP and vsfGFP-9 were amplified by PCR using primer pairs P1-P2 from plasmids pSLC-253 and pSLC-255 respectively. The PCR products were then digested and cloned into pANT4 (replacing GFPmut3) using XbaI and HindIII restriction sites to produce pSLC-282 (containing sfGFP) and pSLC-284 (containing vsfGFP-9). Nonfluorescent colonies from the pSLC-282 cloning contained plasmids without GFP, giving pSLC-306 (empty vector control for pANT4). pSLC-282 was transformed into UTI89 to give SLC-634; pSLC-284 was transformed into UTI89 to give SLC-638.

For chromosomal GFP strains, annealed complementary pairs of oligonucleotides (primer P3 and its reverse complement) containing a rrnB_T2 transcriptional terminator [18] upstream of a modified σ70 promoter [19] were digested with ClaI and XbaI and cloned into plasmids pSLC-253 and pSLC-255 digested with the same enzymes to generate pSLC-293 (sfGFP) and pSLC-294 (vsfGFP-9). We integrated sfGFP or vsfGFP-9 between chromosomal coordinates 1044461-62 in UTI89, which is the att_HK022 region, using a two-step positive-negative selection system [20]. Briefly, the kan-P_rhaB-relE cassette was amplified from pSLC-217 [20] using primers P4 and P5 and integrated into the att_HK022 region using Red recombinase-mediated recombineering [21] to generate SLC-653. Primer pairs P5 and P6 were used to amplify fragments containing sfGFP or vsfGFP-9 from plasmids pSLC-293 or pSLC-294; these PCR products were used to replace the kan-P_rhaB-relE cassette by Red recombinase-mediated recombineering with negative selection on M9 media supplemented with 2% rhamnose, giving SLC-717 (sfGFP) and SLC-719 (vsfGFP-9).

Western blots. Primary antibodies (and dilutions) used were anti-GFP mouse monoclonal antibody (1:6000; Santa Cruz Biotechnology, Shanghai, China), anti-RNA Polymerase β (1:6000; Santa Cruz Biotechnology, Shanghai, China). The secondary antibodies were ECL^™ Anti-rabbit IgG and ECL^™ Anti-mouse IgG, both conjugated with Horseradish peroxidase (HRP) and used at a 1:10,000 dilution.

Mouse infections. Infections were performed as previously described [17]. Data shown is the result of two separate experiments performed on separate days with four to six mice per strain and per time point. Six to seven-week old C3H/HeN female mice were obtained from Harlan (Israel). In co-infections with UTI89 and SLC-719 (chromosomal vsfGFP-9), SLC-719 colonies were differentiated from UTI89 cells by visualization of green fluorescence under 10× magnification because neither strain carries an antibiotic resistance cassette. In coinfections with UTI89 and UTI89 att_HK022::COM-GFP, UTI89 att_HK022::COM-GFP colonies were quantified by plating on LB supplemented with kanamycin, and UTI89 colonies were calculated by subtraction of UTI89 att_HK022::COM-GFP titers from total titers quantified on LB plates. In each experiment, two to three of these LB plates were also used to quantify UTI89 att_HK022::COM-GFP titers using green fluorescence under 10× magnification; in all cases, the fluorescence-based quantification was within 10% of the kanamycin-based quantification.

IBCs were quantified in single infections for UTI89, UTI89 att_HK022::COM-GFP, UTI89/pANT4, SLC-717, SLC-719, and SLC-638. At 6 hpi, infected mice were sacrificed, and harvested bladders were hemisected, stretched onto silicone pads, and fixed with 3% paraformaldehyde (Sigma) as previously described [17]. IBCs from GFP-expressing strains were quantified by fluorescence at 10× magnification. All samples were then subjected to X-Gal staining for independent quantification as previously reported [11]. Data shown is the result of two separate experiments performed on separate days with four mice per strain per experiment.

Flow cytometry. Flow cytometry analysis for GFP expression was performed on a S3^™ Cell Sorter (Bio-Rad, Hercules, CA, USA).

We have created new plasmid- and chromosome-based GFP expressing derivatives of UTI89. Through using a new vsfGFP-9 gene as well as optimization of expression, these are nearly 10× brighter than previously published GFP UTI89 derivatives, and have no in vitro or in vivo defects compared with previous GFP-expressing strains. These strains should be useful for future studies of UTI89 pathogenesis and for the creation of vsfGFP-9 derivatives of other UPEC and E. coli.