Optimizing a high‐sensitivity NanoLuc‐based bioluminescence system for in vivo evaluation of antimicrobial treatment

Abstract

Focal and systemic infections are serious threats to human health. Preclinical models enable the development of new drugs and therapeutic regimens. In vivo, animal bioluminescence (BL) imaging has been used with bacterial reporter strains to evaluate antimicrobial treatment effects. However, high‐sensitivity bioluminescent systems are required because of the limited tissue penetration and low brightness of the BL signals of existing approaches. Here, we report that NanoLuc (Nluc) showed better performance than LuxCDABE in bacteria. However, the retention rate of plasmid constructs in bacteria was low. To construct stable Staphylococcus aureus reporter strains, a partner protein enolase (Eno) was identified by screening of S. aureus strain USA300 for fusion expression of Nluc‐based luciferases, including Nluc, Teluc, and Antares2. Different substrates, such as hydrofurimazine (HFZ), furimazine (FUR), and diphenylterazine (DTZ), were used to optimize a stable reporter strain/substrate pair for BL imaging. S. aureus USA300/Eno‐Antares2/HFZ produced the highest number of photons of orange‐red light in vitro and enabled sensitive BL tracking of S. aureus in vivo, with sensitivities of approximately 10 CFU from mouse skin and 750 CFU from mouse kidneys. USA300/Eno‐Antares2/HFZ was a powerful combination based on the longitudinal evaluation of the therapeutic efficacy of antibiotics. The optimized S. aureus Eno‐Antares2/HFZ pair provides a technological advancement for the in vivo evaluation of antimicrobial treatment.

Impact statement

Bioluminescence (BL) imaging with bacterial reporters has remained a powerful strategy to evaluate antibiotic treatment effects in vivo. In this study, we showed that NanoLuc (Nluc) worked effectively in bacteria, while the retention rate of plasmid in bacteria was low. A partner molecule enolase (Eno) was screened out from Staphylococcus aureus strain USA300 for stable fusion expression of Nluc‐based luciferases. We demonstrated that Nluc‐based luciferases showed superior sensitivity for BL imaging of S. aureus in vitro and in vivo. Our study may provide a model for generating other bacterial reporters for BL imaging applications.

INTRODUCTION

Focal infections, such as skin abscesses, pyelonephritis, and pneumonia, as well as systemic infections, including bacteremia and sepsis, are commonly encountered in hospitals ¹ , ² , ³ , ⁴ . The increase of bacterial antimicrobial resistance complicates the control of infectious diseases and results in high disability and mortality rates ⁵ . A study based on 471 million individual records and 7585 study‐location‐years revealed that six leading drug‐resistant bacteria, namely, Escherichia coli, Staphylococcus aureus, Klebsiella pneumoniae, Streptococcus pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa, were responsible for 3.57 million deaths in 2019 ⁵ . If no action is implemented, then 10 million people will have died every year due to antimicrobial resistance by 2050 ⁶ . The World Health Organization has suggested a priority list of antibiotic‐resistant bacteria for the discovery and development of new antibiotics ⁷ . Preclinical animal models are crucial in assessing the antimicrobial efficacy of new drugs and therapeutic regimens ⁸ , and a noninvasive and affordable strategy allowing for the accurate tracking of pathogens in vivo is urgently needed for drug screening.

Bioluminescence (BL) imaging relies on the catalysis of luciferase on its substrate, which is termed luciferin ⁹ , and has been commonly used for imaging target cells or reporting biological events in living animal models ¹⁰ , ¹¹ . More than 40 bioluminescent systems, including eukaryotic and prokaryotic luciferase‐based BL systems, have been established ¹² . The classical and most applicable bacterial BL system is genetically encoded by the single luxCDABE operon derived from the insect pathogen Photorhabdus luminescens ¹³ , ¹⁴ , and such a system built into metabolically active bacteria can naturally emit blue‐green light with a maximum wavelength of 490 nm ¹³ , ¹⁵ , ¹⁶ . Previous studies showed that the lux system constructed in S. aureus can be used for drug discovery or the evaluation of antistaphylococcal efficacy in vitro, although the brightness of the luxCDABE reporter system is low ¹⁷ , ¹⁸ . An improved lux operon, designated ilux, was generated in E. coli via coexpression of an additional flavin mononucleotide reductase to provide sufficient substrate and a mutant lux operon optimized by error‐prone mutagenesis of the wild‐type luxCDABE operon ¹⁹ . The brightness of the ilux system increases by approximately sevenfold compared with that of lux when expressed in E. coli. Moreover, the ilux system can achieve a single‐cell image in vitro. However, both lux and ilux emit a photon peak at 490 nm, and light wavelengths shorter than 600 nm decrease rapidly in mammalian tissue, thereby limiting the use of the lux system in vivo ⁹ , ²⁰ .

Many luciferases derived from diverse organisms, such as firefly luciferase (Fluc), which presents low catalytic activity with a peak emission of 578 nm ²¹ , Renilla luciferase (Rluc), which shows high catalytic activity but emits natively at wavelengths below 500 nm with a low bioluminescent quantum yield ²² , and luciferase (Luc) from the click beetle Pyrophorus plagiophtalamus, which produces red light with a long emission wavelength of >600 nm, have been selected to develop BL systems ²³ . NanoLuc (Nluc) is a new small luciferase subunit (19 kDa) from the deep‐sea shrimp Oplophorus gracilirostris that demonstrates an ∼150‐fold increase in catalytic activity compared with other common luciferases, including Fluc and Rluc ²⁴ . Nluc utilizes coelenterazine in an ATP‐independent reaction to produce blue light that peaks at 454 nm ²⁴ . Nluc is rarely used in BL imaging because of its short‐wavelength light emission. Intramolecular BL resonance energy transfer (BRET) was applied, and a novel molecule, designated Antares, was designed by fusing Nluc with two domains of the cyan light‐excitable orange‐red fluorescent protein (CyOFP1) to produce photons with long wavelengths ²² , ²⁵ , ²⁶ . Antares is the most red‐shifted among the variants because it emits >430‐fold more red photons at >600 nm per molecule compared with Fluc ²⁰ , ²¹ , ²² . A BRET‐based Antares2 reporter was developed by replacing the wild‐type Nluc with the high‐catalytic activity mutant Teluc (Nluc carrying D19S, D85N, and C164H mutations) in Antares to further improve BL signals in deep tissues ²⁰ . The Antares2/diphenylterazine (DTZ) reporter can emit 3.8‐fold more photons above 600 nm than Antares/DTZ and 65‐fold stronger signals than Fluc/d‐luciferin while easily tracking tumor cells in deep tissues in vivo ²⁰ . However, tracking bacterial cells with BL reporters in vivo is still a challenge.

Bioluminescent Nluc and LuxCDABE were expressed with plasmids in S. aureus, E. coli, and P. aeruginosa in the present study. Nluc presented higher BL signals than LuxCDABE in the tested bacteria. However, the retention rate of recombinant plasmids in bacteria was low. A carrier molecule enolase (Eno) was identified by screening of S. aureus strain USA300 for fusion expression of Nluc‐based luciferases, and three reporter strains, including USA300/Eno‐Nluc, USA300/Eno‐Teluc, and USA300/Eno‐Antares2, were generated. Further testing revealed that the USA300/Eno‐Antares2/hydrofurimazine (HFZ) combination enabled highly sensitive BL imaging for tracking S. aureus in vivo and was powerful in the longitudinal evaluation of the antimicrobial effect.

RESULTS

Nluc functioned differently in bacterial species and strains

The majority of luciferase reporters are constructed with plasmids ¹⁶ , ²³ , ²⁷ . The nluc gene was amplified from the plasmid pNL1.1 (Promega) and cloned into the E. coli–S. aureus shuttle plasmid pLI50 ²⁸ under the control of the promoter of the S. aureus fhuD2 gene that encodes a protein involved in iron metabolism ²⁹ to detect the performance of an engineered luciferase Nluc (19 kDa) from the deep‐sea shrimp O. gracilirostris in bacteria ²⁴ . The bioluminescent luxCDABE operon of the bacterial insect pathogen P. luminescens from the plasmid pAKlux2.1 (Addgene) was also constructed into pLI50 ¹⁶ , ³⁰ . The resulting plasmid, pFH‐nluc or pFH‐lux (Figure S1A,B and Table S1), was transformed in sequence into E. coli DH5α and S. aureus RN4220 and then transformed into S. aureus Newman and USA300 to achieve reporter bacteria. In addition, the nluc gene was directly fused with the gentamycin acetyltransferase gene in the pUCU24 plasmid ³¹ , and a P. aeruginosa reporter was also constructed by transforming pUCP24‐nluc into strain PAO1 (Figure S1C).

Bacterial reporters with the luxCDABE operon construct can naturally produce blue‐green light, while a substrate is required for those with the nluc gene ¹⁴ , ¹⁶ , ³² . BL signals of the reporter bacteria with pFH‐nluc were determined by mixing 1 × 10⁷ CFU/ml of bacterial cells harvested at the mid‐logarithmic phase with an equal volume of substrate HFZ (100 µM). The staphylococcal fhuD2 promoter‐controlled nluc gene functioned correctly in S. aureus strains RN4220, Newman, and USA300 (Figure 1A). Notably, we revealed that the pFH‐nluc construct also functioned correctly in E. coli (Figure 1A), thereby indicating the activity of the fhuD2 gene promoter across bacterial species. The P. aeruginosa reporter carrying pUCP24‐nluc could also catalyze HFZ to produce brightness (Figure 1A). However, the brightness of reporter strains varied even under the same culture conditions. As shown in Figure 1B, S. aureus Newman/pFH‐nluc and USA300/pFH‐nluc generated similar brightness, which was approximately twofold higher than that of RN4220/pFH‐nluc and DH5α/pFH‐nluc and around fivefold higher than that of PAO1/pUCP24‐nluc after 6 h of culture. The bacterium‐carried nluc construct generated considerably higher brightness than those with lux constructs (Figure 1A,B). BL signals of reporters may be affected by the growth phase of bacteria. We revealed that pFH‐nluc or pUCP24‐nluc reporter strains produce relatively stable brightness compared with the strain carrying the pFH‐lux plasmid, which decreased substantially after 9 h of culture (Figure 1C,D). Moreover, Nluc‐catalyzed HFZ produced ~2–10⁶‐fold higher BL signals than its lux operon counterpart at different growth phases (Figure 1D).

Figure 1.

Comparison of Nluc and LuxCDABE engineered bacteria for in vitro bioluminescence (BL) imaging. (A) Representative BL images of 50 µl of pFH‐nluc‐ or pUCP24‐nluc‐ transformed reporter bacterial culture (1 × 10⁷ CFU/ml) mixed with 50 µl of HFZ (100 µM) and 100 µl of pFH‐lux‐transformed reporter bacterial culture (5 × 10⁶ CFU/ml). (B) Quantification of total flux produced from reporter bacteria after 6 h of culture. Data are presented as mean ± standard error of the mean (SEM). Statistical significance was analyzed using the unpaired two‐tailed t‐test. *p < 0.05 and ***p < 0.001. (C) Change of BL signals of the reporter bacteria cultured within 24 h. The cultures of diverse bacteria were harvested at the time indicated. BL signals were measured by mixing 50 µl of 1:100 diluted pFH‐nluc‐ or pUCP24‐nluc‐transformed bacterial culture with 50 µl of HFZ (100 µM) in a black 96‐well plate. The signals from the wells carrying 100 µl of 1:200 diluted pFH‐lux‐transformed bacterial cultures were also determined using the IVIS® Lumina LT system. Representative BL images are shown. (D) BL intensity for each reporter strain over time. Data are presented as mean ± SEM. (E) Stability determination of plasmids in bacteria cultured in vitro. After 48 h of culture, the bacterial CFU was determined using the plate dilution method with a BHIA plate and BHIA supplemented with 10 µg/ml of Cm, or using an LB agar plate and an LB agar plate supplemented with 100 µg/ml of Amp or 20 µg/ml of Gm. Each experiment was repeated at least three times in triplicate. The mean number of viable bacteria grown on BHIA/LB plates was adjusted to 100%, while the relative numbers of viable bacteria grown on BHIA/LB plates supplemented with Cm, Amp, or Gm were calculated and are indicated. Data are presented as mean ± SEM. Amp, ampicillin; BL, bioluminescence; CFU, colony‐forming unit; Cm, chloramphenicol; Gm, gentamicin; p/s, photons/seconds.

A recombinant plasmid will be progressively lost in the absence of antibiotic‐selective pressure ²⁷ . We then detected the stability of plasmids in the reporter strains grown in vitro using the plate dilution method ²⁷ . Compared with the stability of pFH‐nluc in E. coli DH5α (86%) and pUCP24‐nluc in P. aeruginosa PAO1 (72%), only 0.15%–47.4% of S. aureus strains retained the pFH‐nluc plasmid after 48 h of culture (Figure 1E). In vivo studies also showed that only 0.0018% of USA300/pFH‐nluc exists in the kidneys of mice 5 days postinfection through the tail vein (Figure S1D). Overall, these data demonstrated that Nluc achieved better BL performance than LuxCDABE in both Gram‐positive and Gram‐negative bacteria. However, the brightness of reporters varied in different bacterial species and strains. The stability of the nluc construct also varied in bacterial strains and species, with the minimum stability of pFH‐nluc observed in S. aureus USA300. These results highlighted the high requirement for BL reporter system optimization.

Screening of S. aureus partner molecules for fusion of Nluc‐based luciferases

The plasmid‐mediated nluc (pFH‐nluc) in the hypervirulent methicillin‐resistant S. aureus strain USA300, which was isolated in 2000 and has become a major source of community‐acquired infections worldwide ³³ , was unstable in vitro and in vivo (Figures 1E and S1D). We next attempted to construct a stable reporter strain by inserting the gene encoding Nluc‐based luciferase into the chromosome of USA300. A fusion carrier molecule is necessary for the stable expression of exogenous genes ³⁴ . We hypothesized that a partner for fusion expression of Nluc‐based luciferases must meet several criteria: (i) expression is stable during the entire growth phase; (ii) expression is unaffected by stress conditions, such as antibiotic treatment; and (iii) fusion with the exogenous gene of interest fails to arrest bacterial growth. Among the proteins that met these criteria, four evolutionarily conserved proteins in S. aureus, namely, penicillin‐binding protein 2a (PBP2a) ³⁵ , enolase (Eno) ³⁶ , NAD‐dependent deacetylase (CobB) ³⁷ , and the global regulator SarA ³⁸ , were randomly selected to determine their availability as a fusion partner. Western blot analyses revealed that the expression of PBP2a and Eno was stable throughout the growth phase of S. aureus USA300, while that of CobB and SarA varied (Figure 2A,B). PBP2a, a target of β‐lactams, increased in expression in response to oxacillin treatment but demonstrated the opposite effect with Eno (Figure S2A). A study showed that Eno acts as a predominant component of S. aureus membrane vesicles and can carry foreign proteins to membrane vesicles ³⁹ . This finding implied that Eno could be an appropriate candidate for the delivery of exogenous Nluc‐based luciferases.

Figure 2.

Construction of stable and functional Nluc‐based luciferase‐producing Staphylococcus aureus reporter strains. (A) Western blot analysis of the expression of PBP2a, Eno, CobB, and SarA in S. aureus USA300 over time. The protein gel served as a loading control (Lc), and the molecular weights of the protein markers (M) are indicated on the left. (B) Semi‐quantitative analysis of the gray values of the indicated bands in each lane of (A) using ImageJ software. Data are presented as mean ± SEM. Statistical significance was calculated by two‐way analysis of variance (ANOVA); ns indicates no significance, **p < 0.01, and ***p < 0.001. (C) Western blot analysis of Eno fusion proteins in the total cell lysates using mouse anti‐Eno polyclonal antibodies. (D) Bioluminescence (BL) photographs of different S. aureus reporter strain and substrate combinations. (E) Representative BL images of 50 µl of S. aureus reporter strains (1 × 10⁷ CFU/ml) mixed with 50 µl of different doses of HFZ, FUR, or DTZ (3.125–100 µM). (F) Quantification of total flux produced from S. aureus reporter strains with different concentrations of substrates. Data are presented as mean ± SEM. Statistical significance was analyzed by two‐way ANOVA; ns indicates no significance, *p < 0.05, **p < 0.01, and ***p < 0.001. (G) Images of BL signals from USA300/Eno‐Nluc, USA300/Eno‐Teluc, and USA300/Eno‐Antares2 over a range of bacterial loads with 50 µl of HFZ (100 µM) in the black 96‐well plates. (H) Good correlation between the BL intensity and bacterial number of S. aureus USA300/Eno‐Nluc, USA300/Eno‐Teluc, or USA300/Eno‐Antares2. The experiment was repeated at least three times in triplicate. Data are presented as mean ± SEM. x, bacterial number; y, BL intensity.

Genes that encode Nluc‐based luciferases, namely, Nluc, Teluc, and Antares2, were chemically synthesized and genetically fused in‐frame with the 3ʹ‐terminal of eno in the chromosome of S. aureus USA300 to achieve USA300/Eno‐Nluc, USA300/Eno‐Teluc, and USA300/Eno‐Antares2 reporter strains, respectively (Figure S2B). The expected fusion proteins of Eno‐Nluc, Eno‐Teluc, and Eno‐Antares2 were confirmed via Western blot analysis (Figure 2C). The soluble substrate HFZ with acceptable solubility ¹¹ , the imidazopyrazinone substrate furimazine (FUR) ²⁴ , and the coelenterazine analog DTZ ⁹ were used to detect the BL intensity of reporter strains in vitro. The results showed that S. aureus reporter strains USA300/Eno‐Nluc, USA300/Eno‐Teluc, and USA300/Eno‐Antares2 could catalyze HFZ, FUR, and DTZ to produce macroscopic light (Figure 2D). Furthermore, the hemolytic activities and growth curves of all reporter strains of interest were similar to those of the wild‐type strain USA300 (Figure S2C,D). We demonstrated that all three luciferase reporter strains could catalyze HFZ to generate stable BL signals within 24 h of testing (Figure S2E,F). The brightness of luciferase reporter/HFZ pairs was also assessed under different temperatures, pH, and other physiochemical conditions such as primary solubilizing agent PEG‐300 ¹¹ , cell culture medium RPMI‐1640 ⁴⁰ , RPMI‐1640 plus 10% fetal bovine serum, phosphate‐buffered saline, 2‐[4‐(2‐hydroxyethyl)‐1‐piperazinyl] ethanesulfonicacid (HEPES) buffer, Tris‐EDTA buffer, and normal saline (NS, 0.9% NaCl). The results demonstrated that all reporter combinations were stable at 30°C–40°C, pH 6–8 (Figure S2G–J), and produced higher BL signals under other physiochemical conditions (Figure S2K,L).

The bioluminescent intensities of strains USA300/Eno‐Nluc, USA300/Eno‐Teluc, and USA300/Eno‐Antares2 were evaluated with different concentrations of substrates, given their high stability. As shown in Figure 2E,F, reporter strains catalyzed HFZ to produce the highest brightness signals, followed by FUR and DTZ. Although the USA300/Eno–Teluc/HFZ pair generated the highest total emission, the USA300/Eno–Antares2/HFZ pair emitted >2.5‐fold more photons of orange‐red light than other pairs (Figures 2D and S3A–C), and HFZ did not inhibit the growth of USA300/Eno‐Antares2 (Figure S3D,E). We also found that the BL signals of USA300/Eno‐Nluc/HFZ, USA300/Eno‐Teluc/HFZ, and USA300/Eno‐Antares2/HFZ were highly correlated with the total viable bacterial CFU counts in vitro (R ² > 0.99) (Figure 2G,H). However, even low doses of toxic compounds, including nitrogen mustard ⁴¹ , dimethyl sulfoxide, and the antibiotics erythromycin (ERM) and vancomycin (VAN), could inhibit the growth of S. aureus USA300/Eno‐Antares2 and result in BL signal reduction (Figure S3F–H). These results indicated that the S. aureus USA300/Eno‐Antares2/HFZ pair had potential application in vivo, and any factor that affects bacterial growth had an effect on the BL signal.

USA300/Eno‐Antares2/HFZ enabled sensitive BL imaging for tracking S. aureus in a mouse bacteremia model

USA300/Eno‐Teluc/HFZ and USA300/Eno‐Antares2/HFZ pairs showed excellent performance in vitro, especially regarding total BL emission and orange‐red light photon production (Figures 2D–F and S3A–C). The engineered USA300/Eno‐Teluc and USA300/Eno‐Antares2 strains were chosen to infect BALB/c mice intravenously through a tail‐vein injection, and different substrates were individually administered to evaluate the performance of reporter strain/substrate pairs for S. aureus deep‐tissue BL imaging. BL signals could be detected in the kidneys on the dorsal sides of mice challenged with S. aureus USA300/Eno‐Antares2/HFZ, USA300/Eno‐Antares2/DTZ, USA300/Eno‐Antares2/FUR, and USA300/Eno‐Teluc/DTZ, but not in the kidneys of mice injected with USA300/Eno‐Teluc/HFZ and USA300/Eno‐Teluc/FUR (Figures 3A and S4A,B). However, bacterial loads in the kidneys of infected mice were comparable in each group (Figure S4C). The BL intensity was normalized with the total BL signal relative to the total viable bacterial count in each kidney. The results showed that the normalized BL intensity of the USA300/Eno‐Antares2/HFZ pair was the highest among all reporter/substrate pairs (Figures 3B and S4D).

Figure 3.

The USA300/Eno‐Antares2/HFZ pair is superior to other pair combinations for deep‐tissue bioluminescence (BL) imaging. (A) BL imaging of BALB/c mice that were infected intravenously with 1 × 10⁷ CFU of USA300/Eno‐Antares2 or USA300/Eno‐Teluc. After 24 h of infection, the mice (n = 3) were imaged after an intraperitoneal injection of 1 µmol of HFZ. 1 × 10⁷ CFU of USA300‐infected mice served as negative controls. Numbers 1, 3, and 5 represent the left kidneys, and 2, 4, and 6 indicate the right kidneys of the challenged mice. (B) Quantitative comparison of the normalized BL intensity of USA300/Eno‐Antares2 with diverse substrates indicated. Data are presented as mean ± SEM. Statistical significance was analyzed using the two‐tailed Mann–Whitney U test, *p < 0.05 and **p < 0.01. (C) Determination of BL sensitivity. BALB/c mice (n = 9) were each intravenously inoculated with 1 × 10⁶ CFU of USA300/Eno‐Antares2, and 1 µmol of HFZ was administered after 6 h of infection. In vivo BL signals were measured. S. aureus USA300‐infected mice served as negative controls. The left and right kidneys of mice are shown with specific numbers as indicated. (D) BL intensities of three representative mice to monitor signal decay over time (left and right kidneys, respectively). (E) Correlation between in vivo BL signals and bacterial loads in the right and left kidneys; the red arrow denotes the lowest CFU (750) detected by BL imaging. x, bacterial CFU; y, BL intensity. (F) BL images of the dorsal and ventral sides of mice (n = 3/group) at the time indicated. S. aureus USA300‐infected mice served as negative controls. (G) Correlation between total BL signals in vivo of the dorsal and ventral sides of mice and total bacterial CFU counts in kidneys and livers. The linear regression lines and correlation coefficients for the time span indicated are shown. x, bacterial CFU; y, BL intensity.

The sensitivity of deep‐tissue BL imaging is usually determined using bacterial colonization from animal kidneys. However, the bacterial count from kidneys varies among individual animals despite being challenged with the same number of bacteria ²³ . Mice were infected with 1 × 10⁶ CFU of USA300/Eno‐Antares2 via a tail‐veil injection, and in vivo BL signals were captured 6 h postinfection after administering 1 µmol of HFZ to detect the sensitivity of USA300/Eno‐Antares2/HFZ for deep‐tissue BL imaging. Compared with the BL signals of mice challenged with USA300 (negative control), the BL signals from the dorsal sides of mice were variable, and signals were absent in two of the infected mice (Figure 3C). The BL signals of three infected mice were collected every 3 min for 24 min after administering HFZ to monitor signal decay over time. The results demonstrated that BL signals were stable in vivo (Figure 3D). Bacterial loads in the right and left kidneys of mice 6 h postinfection were determined, and a satisfactory correlation between BL signals and viable bacterial numbers within 6 h of infection was observed (R ² = 0.9466) (Figure 3E). A sensitivity of approximately 750 CFU from the kidneys was detected via in vivo BL imaging, and this value was 40‐fold higher than that of the S. aureus luc reporter system ²³ .

Emitted in vivo BL signals are inaccurate during invasive and deep‐seated bacterial infections, and the actual in vivo bacterial load is underestimated ⁴² , ⁴³ . BL signals for deep‐tissue BL imaging may vary on the ventral and dorsal sides of experimental mice ²³ . Therefore, BL signals were collected from kidneys on the dorsal side and livers on the ventral side of mice at 6, 12, and 24 h and 2, 4, and 6 days after being challenged with 1 × 10⁷ CFU of USA300/Eno‐Antares2 (Figures 3F and S4E). Bacterial burdens in the kidneys and livers were also determined, and the normalized BL intensity was calculated with the total BL signals (dorsal side plus ventral side) divided by the total number of viable bacteria (kidney plus liver). Notably, we revealed that in vivo BL signals of USA300/Eno‐Antares2 were highly correlated with the numbers of bacterial CFU within 24 h of infection (R ² = 0.9703 for 1 day, Figure 3G), but not after more than 1 day (R ² = 0.3403 for 2–6 days, Figure 3G). Ex vivo bacterial CFU counts from the kidneys and livers of infected mice dynamically changed over time (Figure S4F), and abscesses formed in the kidneys of mice 6 days after infection (Figure S4G). We speculated that the formation of an abscess may impair BL signals by impeding the entrance of Nluc substrate. Therefore, the reporter system of USA300/Eno‐Antares2/HFZ could provide a useful tool for monitoring deep‐seated bacterial dissemination during the early stages of bacteremia and mainly within 24 h of postinfection.

BL imaging of USA300/Eno‐Antares2/HFZ in the mouse skin infection model

S. aureus frequently causes skin and soft tissue disorders, such as impetigo, folliculitis, cellulitis, and wound infections ¹⁸ . We next assessed the BL intensity of the USA300/Eno‐Antares2/HFZ pair as a quantitative approach for evaluating the skin colonization of S. aureus based on the high‐brightness signals of the USA300/Eno‐Antares2/HFZ pair in vivo (Figures 3F and S4E). A mouse skin infection model was constructed by injecting the USA300/Eno‐Antares2 reporter strain into the skin on the dorsal side of the back ³⁸ . Mice were injected subcutaneously with a mixture containing different CFU of bacteria (1 × 10¹–1 × 10⁵ CFU) and an equal volume of HFZ (100 μM) to examine the in vivo sensitivity of BL imaging in the mouse skin. As shown in Figure 4A, 1 × 10¹–1 × 10⁵ CFU of reporter strain‐challenged mice showed dose‐dependent focal BL signals. The total BL intensity (log₁₀) was linearly correlated with the number of injected USA300/Eno‐Antares2 bacteria (R ² = 0.9632) (Figure 4B). Therefore, sensitivity of approximately 10 CFU from the mouse skin was achieved through the optimized USA300/Eno‐Antares2/HFZ BL system. Three mice were subcutaneously challenged with USA300/Eno‐Antares2/HFZ and BL signals were collected every 5 min for 50 min to determine the optimal time for BL detection in skin. The results showed that in vivo BL signals from mouse skin were stable within 15 min and then decreased significantly (p < 0.01, Figure S5A).

Figure 4.

BL imaging of the USA300/Eno‐Antares2/HFZ pair in mouse skin infection. (A) Representative BL signals in mouse skins administered with 1 × 10¹–1 × 10⁵ CFU of USA300/Eno‐Antares2, respectively. USA300 injected mice were served as negative controls. Each animal was anesthetized and then subcutaneously administered with 100 µl of mixture containing 50 µl of the indicated amount of the USA300/Eno‐Antares2 reporter strain and 50 µl of substrate HFZ (100 µM) (n = 3/group). (B) Quantitative analysis of BL signals produced 5 min after injection of reporter strain/HFZ. The red arrow indicates the lowest CFU capable of being detected in mouse skin by in vivo BL imaging (10 CFU). Data are presented as mean ± SEM. x, bacterial number; y, BL intensity. (C) Mice (n =  6/group) were subcutaneously infected with 1 × 10⁷ CFU of the reporter strain USA300/Eno‐Antares2, then 100 µl of HFZ (100 μM) was injected around the abscess, and BL signals at the indicated time after infection were captured. Mice administered 1 × 10⁷ CFU of the wild‐type USA300 and 100 µl of HFZ (100 μM) served as negative controls. (D) Bacterial CFU counts from the skin of mice at 6, 12, and 24 h, and 2, 4, and 6 days after infection with 1 × 10⁷ CFU of USA300/Eno‐Antares2. Data are presented as mean ± SEM. Statistical significance was analyzed by one‐way analysis of variance, ns indicates no significance, **p < 0.01, and ***p < 0.001. (E) Correlation between in vivo total BL signals and the viable bacterial loads in the skin of mice at 6 h, 12–24 h, and 2–6 days postinfection of USA300/Eno‐Antares2. The linear regression lines and correlation coefficient values are shown. x, Bacterial CFU; y, BL intensity.

Thirty‐six mice were injected subcutaneously with 1 × 10⁷ CFU of USA300/Eno‐Antares2 to explore the potential application of the USA300/Eno‐Antares2/HFZ system for detecting S. aureus in vivo. Body weights and skin abscess areas of infected mice were measured daily. The animals were imaged at 6, 12, and 24 h and 2, 4, and 6 days postinfection (n = 6 for each time point) after a subcutaneous injection of 100 µl of HFZ (100 µM) around the abscess. Infected mice were euthanized, and bacterial loads in the skin abscesses were determined after BL imaging. As shown in Figure S5B, skin abscess areas changed dynamically after being challenged with bacteria, while body weights were comparable among the groups (Figure S5C). BL signals could be detected in all infected mice at the time points tested. However, the BL intensity was variable with prolonged infection time (Figure 4C) and the ex vivo bacterial CFU counts changed (Figure 4D). Linear correlations between BL signals and USA300/Eno‐Antares2 bacterial loads were achieved at 6 h (R ² = 0.99) and 12–24 h (R ² = 0.9094), as well as 2–6 days (R ² = 0.9181) postinfection (Figure 4E). The toxicity of HFZ to BALB/c mice was also investigated, and comparable body weights and serum interleukin 6 (IL‐6) and tumor necrosis factor‐α (TNF‐α) levels were observed (Figure S5D–F). This finding suggested that a strong association existed between in vivo BL signals and skin bacterial colonization. Hence, USA300/Eno‐Antares2/HFZ was a powerful reporter/substrate pair for tracking and quantifying bacteria in S. aureus skin infections.

BL system of USA300/Eno‐Antares2/HFZ for the therapeutic efficacy evaluation of antibiotics against S. aureus skin infections

Longitudinal and noninvasive monitoring of bacterial loads can provide key information about the disease pathogenesis and infectious course during antimicrobial therapy ⁸ , ²³ . We attempted to evaluate the efficacy of antibiotics against S. aureus skin infections with in vivo BL imaging on the basis of a high‐level correlation between in vivo BL signals and ex vivo CFU of the engineered USA300/Eno‐Antares2 reporter strain in the mouse skin infection model (Figure 4C–E). Wild‐type S. aureus USA300 is resistant to erythromycin (ERM, MIC > 128 µg/ml) but sensitive to vancomycin (VAN, MIC = 0.25 µg/ml) ³³ , ⁴⁴ . Therefore, a mouse skin infection therapy model was constructed to investigate the capacity of the USA300/Eno‐Antares2/HFZ pair for the longitudinal evaluation of antibiotic efficacy against S. aureus in vivo (Figure 5A). As shown in Figure 5B, the administration of HFZ could track the location of the S. aureus USA300/Eno‐Antares2 reporter in mouse skin for 11 days. Consistent with the BL signals on the dorsal back of mice, the skin abscess areas in the groups treated with normal saline (NS) and different concentrations of ERM or VAN increased immediately after infection, and then gradually decreased over time (Figures 5B,C and S6A). BL intensities decreased faster in the 32 mg/kg of VAN‐treated group than in the NS‐treated group (Figure 5C). Further regression analysis demonstrated that BL intensities in the skin abscesses were linearly correlated with the concentration of VAN at 11 days (Figure 5D, R ² = 0.9858), but not with ERM (Figure S6B). Compared with mice treated with NS, ERM, 8 mg/kg or 16 mg/kg of VAN, bacterial loads in the skin abscesses and IL‐6 levels in the sera were significantly lower in mice treated with 32 mg/kg of VAN at 11 days (Figures 5E and S6C). In addition, mouse body weights significantly increased in the 32 mg/kg VAN treatment cohort at 9 and 11 days compared with those in the NS treatment cohorts (Figure S6D). However, the comparable skin abscess sizes observed among the seven groups of infected mice (Figure S6E,F) indicated that the skin abscess size may be a poor index for bacterial skin colonization. Overall, these results demonstrated that VAN was an effective treatment, whereas ERM was not, and that S. aureus USA300/Eno‐Antares2 could be successfully used in the longitudinal evaluation of the therapeutic efficacy of antimicrobial agents in vivo. The optimized BL imaging system based on the USA300/Eno‐Antares2/HFZ pair provided a technological advancement for investigating the preclinical efficacy of anti‐S. aureus agents.

Figure 5.

Evaluation of the treatment efficacy of antibiotics against Staphylococcus aureus skin infection using USA300/Eno‐Antares2 BL imaging. (A) Schematic diagram of the timeline in mouse skin model construction and antibiotic administration. The back hairs of BALB/c mice (n = 5/group) were depilated completely with sodium sulfide, and then 1 × 10⁷ CFU of USA300/Eno‐Antares2 were injected subcutaneously into each mouse. After 24 h of inoculation, the infected mice received vancomycin (VAN), erythromycin (ERM), or normal saline (NS) treatment for 11 days twice a day (every 12 h) through an intraperitoneal injection. During the therapeutic process, mice were imaged every other day, and the abscess area and body weight were also measured. (B) Longitudinal BL images of mice treated with NS and different concentrations of ERM or VAN as indicated. (C) BL intensities from each mouse in groups of NS‐treated and different concentrations of VAN‐treated were measured at the indicated time and presented. Data are presented as mean ± SEM. Statistical significance was analyzed using a two‐tailed Mann–Whitney U test; ns indicates no significance and **p < 0.01 relative to the NS‐treated group at the same time. (D) Correlation of BL intensities in the skin abscesses with different concentrations of VAN for treatment of 11 days postinfection of 1 × 10⁷ CFU of USA300/Eno‐Antares2. Data are presented as mean ± SEM. x, concentration of VAN; y, BL intensity. (E) Bacterial loads in the skins of mice after 11 days of VAN therapy. Data are presented as mean ± SEM. Statistical significance was analyzed by one‐way analysis of variance; ns indicates no significance and ***p < 0.001 relative to the NS‐treated group at 11 days postinfection.

DISCUSSION

The preclinical evaluation of antimicrobial efficacy in animal models is crucial for the development of new agents and therapeutic regimens ⁸ . In vivo BL imaging with bacterial reporters has remained a powerful strategy for the noninvasive and nondestructive monitoring of bacterial loads in the infection process ¹⁵ . In vivo imaging of bacterial pathogens using luciferases is challenging ²³ . The high catalytic activity of Nluc enables the production of bright and sustained BL signals that peak at 454 nm ²⁰ , ²⁴ . Several studies have attempted to engineer novel Nluc‐based luciferases with enhanced brightness and red‐light emission or develop substrates with satisfactory solubility for in vivo mammal cell imaging ¹¹ , ²⁰ , ³² . In the present study, the deep‐sea shrimp‐derived nluc gene and insect bacterial pathogen‐carried luxCDABE operon were constructed into plasmids, and their BL capacities were compared using diverse bacteria, including S. aureus, E. coli, and P. aeruginosa. We demonstrated that the BL brightness of Nluc and Lux varied according to the bacterial species and strains. The BL intensity of Nluc‐expressing bacteria was generally ~2–10⁶‐fold higher than that of Lux‐expressing bacteria (Figure 1D). However, the stability of recombinant plasmids was dependent on the bacterial species. Approximately 86% of E. coli DH5α and 72% of P. aeruginosa PAO1 could maintain pFH‐nluc and pUCP24‐nluc plasmids, respectively, while only 0.15%–47.4% of S. aureus strains could retain the pFH‐nluc plasmid after 48 h of culture in vitro (Figure 1E). Moreover, the retention rate of pFH‐nluc in S. aureus USA300 in vivo was low, and only 0.0018% of bacteria existed with plasmids 5 days postinfection of USA300/pFH‐nluc into mice through the tail vein (Figure S1D). Similar findings were also reported in the study of Bacconi et al., in which the stability of plasmids in S. aureus was improved by cloning toxin/antitoxin and partition systems into a plasmid of interest ²⁷ . However, the complicated strategy involving stability modification of a certain plasmid has rarely been used.

Insertion of the target gene into the bacterial chromosome is an effective way to improve gene stability ⁴⁵ . A suitable carrier molecule encoded by genomic DNA may be important for the successful expression of the exogenous gene in this case ³⁴ . Therefore, an evolutionarily conserved Eno that plays an important role in S. aureus glycolysis and energy production ³⁶ was screened for the construction of Nluc‐based luciferases, including Nluc, Teluc, and Antares2, in S. aureus. The growth rate and hemolytic activity of S. aureus USA300 remained unaffected by the fusion of nluc, teluc, and Antares2 with eno (Figure S2B–D). Furthermore, the fusion‐expressed Eno‐Nluc, Eno‐Teluc, and Eno‐Antares2 in S. aureus USA300 could catalyze various substrates to produce macroscopic light (Figure 2D). This finding indicated a successful approach for fusion expression of Nluc‐based luciferases with Eno in S. aureus. The carrier molecule‐based optimizing strategy may provide an acceptable example in the construction of other bacterial reporters, such as E. coli and S. pneumonia ¹⁵ , ¹⁹ , ⁴⁵ .

Photons emitted by luciferases shorter than 600 nm may strongly interact with mammalian tissues and show poor BL signals, particularly in deep tissues, during in vivo bacterial BL imaging ²⁰ . Moreover, the use of Nluc‐based luciferases in vivo is limited by the poor water solubility and bioavailability of the substrate FUR or DTZ ¹¹ , ³² . The discovery of HFZ with excellent aqueous solubility allows for high‐dose substrate usage and improves light production in tracking the location and growth of tumor cells in mice ¹¹ . We demonstrated that the USA300/Eno‐Teluc/HFZ pair generated the highest total BL emission, while the USA300/Eno‐Antares2/HFZ pair produced the most photons of orange‐red light in vitro when screening combinations of reporter strains and substrates (Figure S3C). This finding implied that the optimized and engineered bacterial reporters had high application potential. We then showed that the combination of S. aureus USA300/Eno‐Antares2 and HFZ was superior to other reporter/substrate pairs in BL imaging of deep tissues, such as the kidney and liver, using the mouse bacteremia model (Figure 3A,B). Approximately 750 CFU of USA300/Eno‐Antares2 in mouse kidneys could produce detectable BL signals (Figure 3E), thereby indicating a sensitivity 40‐fold greater than that of the S. aureus luc reporter system (about 3 × 10⁴ CFU from kidneys) ²³ . Miller et al. showed a high‐level correlation between the in vivo BL data and ex vivo CFU counts of S. aureus luc reporter strain AH4775 from the kidneys of infected mice 16 h postinoculation (R ² = 0.9924) ²³ . Deep‐seated BL imaging revealed that signals from the ventral and dorsal sides of mice are variable ²³ . Therefore, the normalized BL intensity was measured and calculated with the total BL flux (dorsal side plus ventral side) divided by the total number of bacterial CFU (kidney plus liver). Notably, we revealed that the total in vivo BL signals of S. aureus USA300/Eno‐Antares2 highly correlated with the ex vivo bacterial CFU within 24 h of infection (R ² = 0.9703, curve fitting with the data for 6, 12, and 24 h after inoculation) but not at the time of more than 1 day (R ² = 0.3403 for 2–6 days postinfection, Figure 3G). This phenomenon was likely attributed to the dynamic variation and distribution of bacteria in the internal organs of model animals (Figure S4F), individual differences in the host immune response, and substrate permeability inhibition by bacterial abscess formation (Figure S4G) ⁴⁶ . Therefore, the optimal USA300/Eno‐Antares2/HFZ system may be useful to explore deep‐seated bacterial dissemination in the early stages, particularly within 24 h postinfection. New in vivo BL reporters with photon emission longer than 600 nm or novel substrates with high tissue or bacterial abscess permeability require further investigation to monitor deep‐seated bacterial dissemination.

Compared with the bacteremia model, the bacterial count from the mouse skin infection model confirmed that an acceptable correlation existed between BL data and bacterial CFU in the skin of mice at 6 h, 12–24 h, and 2–6 days postinoculation (Figure 4E). This result was consistent with those of previous studies ¹⁸ , ²³ . Additionally, the combination of USA300/Eno‐Antares2/HFZ showed high sensitivity of 10 CFU in monitoring S. aureus in mouse skin infections (Figure 4B). Therefore, the optimal USA300/Eno‐Antares2/HFZ pair was used for BL imaging to evaluate the antimicrobial efficacy of certain antibiotics against bacterial skin infections. The therapeutic model was designed on the basis of the resistant nature of S. aureus USA300, which is resistant to ERM and sensitive to VAN ³³ , ⁴⁴ . The measured BL signals from infected skin samples remarkably decreased in mice treated with VAN compared with those treated with ERM or NS (Figures 5B,C, and S6A). BL intensities in the skin abscesses were linearly related to the VAN concentrations at 11 days (Figure 5D, R ² = 0.9858). Correspondingly, the bacterial loads in the skin abscesses of 32 mg/kg VAN‐treated mice were substantially reduced compared with those of 8 mg/kg or 16 mg/kg of VAN‐, ERM‐, and NS‐treated mice at 11 days (Figure 5E). These findings suggested that VAN can be an effective treatment, while ERM is not. Different dosages of antibiotic treatment were used to confirm that the USA300/Eno‐Antares2/HFZ system could be successfully applied to evaluate the treatment performance of antimicrobial agents in vivo based on its high sensitivity (Figure 5A,B).

Overall, data from the present study demonstrated that Nluc‐based luciferases could catalyze their substrates to produce higher brightness signals than the classical luxCDABE system in BL imaging. The optimized USA300/Eno‐Antares2/HFZ pair improved the stability and brightness of BL signals in vitro and in vivo. Moreover, this combination provided reliable BL signals for monitoring the bacterial burden and lesion healing during antimicrobial treatment in a preclinical S. aureus skin infection model. The USA300/Eno‐Antares2/HFZ combination may be used as an in vivo BL evaluation tool for S. aureus infection at the early stages of systemic bacterial dissemination. The carrier molecule‐based optimization strategy for BL systems may provide a model for generating other bacterial reporters to establish noninvasive, stable, cost‐effective, and accurate animal models for focal and systemic antimicrobial treatment evaluation.

MATERIALS AND METHODS

Bacterial strains

Bacterial strains used in the present study are listed in Table S1. E. coli DH5α strain was purchased from TIANGEN Biotech (China). S. aureus USA300 strain FPR3757 (GenBank accession no. CP000255.1) was kindly provided by Min Li (Renji Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China), Newman (NCTC 8178) was provided by Prof. Yu Lu (Jilin University, China), and RN4220 (NCTC 8325–4) was provided by Baolin Sun (University of Science and Technology of China). P. aeruginosa strain PAO1 was maintained in our lab. E. coli and P. aeruginosa strains were cultivated in Luria broth (LB) medium (Oxoid). S. aureus strains were cultured with brain–heart infusion (BHI) medium (Oxoid) or BHI agar (BHIA). When appropriate, the broth media or agar plates were supplemented with ampicillin (Amp, 100 μg/ml), gentamicin (Gm, 20 μg/ml), or chloramphenicol (Cm, 10 μg/ml).

Construction of Nluc‐ and LuxCDABE‐producing bacterial strains

Plasmids used in the present study are listed in Table S1. A fragment of fhuD2 gene promoter DNA (213 bp) was amplified from the genomic DNA template of S. aureus USA300 with the primer pair fhud2P‐F/R ²⁷ , and inserted into the E. coli–S. aureus shuttle vector pLI50 to construct the pFH plasmid. Then, the DNA of nluc from the plasmid of pNL1.1 (Promega Corp.) or the luxABCDE operon from the pAKlux2.1 plasmid (Addgene) was amplified with the primer pair nluc‐F/R or lux‐F/R, and subsequently inserted into the pFH plasmid under the control of the fhuD2 gene promoter to obtain pFH‐nluc and pFH‐lux plasmids, respectively. Similarly, the nluc gene was also amplified with primers nluc‐F1/R1 and fused with the gentamycin acetyltransferase gene of the pUCP24 vector ³¹ to achieve the pUCP24‐nluc plasmid. The pFH‐nluc and pFH‐lux plasmids in E. coli DH5α were prepared, respectively, and transformed into S. aureus strains RN4220, Newman, and USA300. The pUCP24‐nluc plasmid was transformed into P. aeruginosa PAO1 to obtain Nluc‐producing strain PAO1/pUCP24‐nluc. All primers used in this study are listed in Table S2.

In vitro BL imaging of Nluc‐ and LuxCDABE‐expressing bacterial strains

The imidazopyrazinone substrate FUR ²⁴ , coelenterazine analog DTZ ⁹ , and soluble substrate HFZ ¹¹ were separately dissolved in a polyethylene glycol‐300 (PEG‐300) formulation made up of 10% glycerol, 10% ethanol, 10% hydroxypropylcyclodextrin, and 35% PEG‐300 in distilled water ¹¹ . BL imaging was performed using the IVIS® Lumina LT Series III system (Perkin Elmer). Images were analyzed using Living Image 4.4 software. For BL imaging of Nluc‐producing bacteria in vitro, 50 µl of pFH‐nluc or pUCP24‐nluc transformed bacterial culture (1 × 10⁷ CFU/ml) was, respectively, mixed with 50 µl of diverse substrates (FUR, DTZ, and HFZ, 100 µM for each) in a black 96‐well plate and then measured. For BL imaging of LuxCDABE‐expressing bacteria, 100 µl of pFH‐lux transformed bacterial culture (0.5 × 10⁷ CFU/ml) was added to the wells of a black 96‐well plate and then detected by the IVIS® Lumina LT system. For the determination of BL signal stability, bacterial strains of interest were cultured in BHI or LB media at 37°C with shaking. The next day, 10 µl of each overnight bacterial culture was added to 2 ml of fresh BHI or LB broth and continued to culture at 37°C with shaking. Then, 10 µl of pFH‐nluc or pUCP24‐nluc plasmid‐carried bacterial culture was taken every hour up to 24 h, and separately added to 1 ml of Phosphate‐buffered saline (PBS) (1:100 dilution). After that, 50 µl of each sample was mixed with 50 µl of HFZ (100 µM) in a black 96‐well plate, and BL signals were measured with the IVIS® Lumina LT system. As for pFH‐lux transformed bacteria, a 10 µl sample taken from fresh culture at each hour was 1:200 diluted in PBS, and 100 µl of bacterial sample was added to the black 96‐well plate and determined. The wild‐type bacteria, including E. coli DH5α, S. aureus USA300, and P. aeruginosa PAO1, were used as negative controls. BL data were collected using the IVIS® Lumina LT system 1 min after the addition of a substrate. The parameters set for image acquisition were as follows: emission filter open for total BL; exposure time, 1–10 s; binning factor, 4; field of view, 12.5 cm; and f‐stop, 1.

Plasmid stability in bacteria

The stability of plasmids in diverse reporter strains was detected using the plate dilution method as described ²⁷ . Briefly, 10 µl of the overnight culture of pFH‐nluc transformed S. aureus USA300 was inoculated into 2 ml of fresh BHI broth in the absence of antibiotics and cultured at 37°C for 24 h, and this process was repeated once the next day. After 48 h of culture in BHI without antibiotics, 100 µl of the sample was 10‐fold serially diluted with fresh BHI, seeded onto a BHIA plate with or without antibiotics (Cm, 10 µg/ml), and incubated at 37°C for 18 h. Each experiment was repeated at least three times in triplicate. The mean number of viable bacteria grown on BHIA plates without antibiotics was adjusted to 100%, while the relative number of viable bacteria grown on BHIA containing Cm was calculated. The stability of plasmids in other reporter strains grown in vitro was also detected as described above.

To evaluate plasmid stability in vivo, BALB/c mice were intravenously injected with 1 × 10⁷ CFU of pFH‐nluc transformed S. aureus USA300 (approximately 30% lethal dose). After 5 days of infection, the infected mice were killed and the kidneys were homogenized in 1 ml of PBST (PBS with 0.1% TritonX‐100) using the instrument of Minibeadbeater 16 (Biospec) ⁴⁷ . The homogenized sample was then serially diluted in PBS and seeded onto BHIA plates with or without Cm (10 µg/ml). The retention rate of the pFH‐nluc plasmid in S. aureus USA300 in vivo was calculated as described above.

Screen of carrier molecules for expression of Nluc‐based luciferases in S. aureus

To construct stable Nluc‐based luciferase expressing S. aureus reporters, several evolutionarily conserved proteins in S. aureus USA300, such as PBP2a ³⁵ , Eno ³⁶ , CobB ³⁷ , and the global regulator SarA ³⁸ , were randomly selected to detect their potential as carrier molecules for fusion expression of exogenous luciferases. The expression levels of the candidate proteins in diverse growth phases of S. aureus USA300 were examined by Western blot. Briefly, the overnight culture of S. aureus USA300 was 1:100 diluted with 200 ml of BHI broth and cultivated at 37°C with shaking. Bacterial cells in 3 ml of culture harvested at 4, 6, 8, 10, 12, 14, 16, 18, and 20 h postinoculation were collected by centrifugation at 10,000g for 10 min at 4°C. The cell pellet was washed twice with PBS, resuspended in 1 ml of cold PBS supplemented with 1% (m/v) β‐mercaptoethanol (Sigma) and 1 mM PMSF (Beyotime). Cells were broken by addition of 0.1 mm diameter zirconia/silica beads and shaking on the Minibeadbeater 16 instrument (Biospec). Cell debris was removed after centrifugation at 10,000g for 10 min at 4°C. The protein concentration was determined using the Bradford Protein Assay Kit (Beyotime). The proteins of interest in S. aureus USA300 were separated by 12% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS‐PAGE) and transferred onto a polyvinylidene difluoride (PVDF) membrane (GE Healthcare). The first mouse anti‐PBP2a, anti‐Eno, anti‐CobB, and anti‐SarA polyclonal antibodies and a secondary sheep anti‐mouse antibody coupled with horseradish peroxidase (Abmart) were used in the Western blot. The target protein signals were visualized by the SuperSignal West Atto Substrate (Thermo Fisher Scientific) and photographed.

Construction of Eno‐fused Nluc‐based luciferase expressing S. aureus reporters

The nucleotide sequences of nluc, teluc, and antares2 according to the codon usage bias of S. aureus USA300 were chemically synthesized by BGI‐Shenzhen (China) and cloned into the pUC18 plasmid to achieve pUC‐nluc, pUC‐teluc, and pUC‐antares2, respectively (Table S1). Based on the stable expression of Eno in S. aureus USA300, the nluc, teluc, and antares2 genes in the plasmids of interest were individually amplified with polymerase chain reaction (PCR) and in‐frame fused with the eno gene within the genome of S. aureus USA300 using homologous recombinant strategy ³⁸ . Briefly, the nluc gene was amplified using primers eno‐nluc‐F1/R1 from the pUC–nluc plasmid template. For site‐specific homologous recombination, the left and right homologous arms across the stop codon of the eno gene were designed and amplified from S. aureus USA300 genomic DNA with primers up‐eno‐F1/R1 and down‐eno‐F1/R1 (Table S2). The fusion fragment (left arm–nluc–right arm) was obtained by over‐lap PCR with the primer pair of up‐eno‐F1/down‐eno‐R1 and cloned into the temperature‐sensitive shuttle vector pBT2 with one step cloning strategy to generate pBT2‐nluc (Table S1). After transformation of the pBT2‐nluc plasmid into S. aureus USA300, the integration of plasmid into bacterial chromosome was induced by cultivating plasmid‐carried USA300 at 42°C for 20 h, followed by cultivating at 25°C for 20 h to achieve plasmid missing strains. The USA300/Eno‐Nluc strain was confirmed by PCR and DNA sequencing. Similar strategies were used to construct USA300/Eno‐Teluc and USA300/Eno‐Antares2, respectively. The expressions of the fusion proteins, such as Eno‐Nluc, Eno‐Teluc, and Eno‐Antares2, in the engineered S. aureus reporters were verified by Western blot.

Hemolytic activity assay

To detect whether the fusion expression of Nluc‐based luciferase with Eno in S. aureus affects the bacterial pathogenicity, the hemolytic activity assay was performed. S. aureus reporter strains were each cultivated at 37°C to an optical density at 600 nm (OD₆₀₀) of 2.0, the culture was 1:100 diluted in PBS, and 5 μl of bacterial suspension was spotted onto 5% (vol/vol) sheep blood agar plates and incubated at 37°C for 24 h. After culture, the plates were photographed.

Bacterial growth curve

The growth curves of S. aureus engineering strains USA300/Eno‐Nluc, USA300/Eno‐Teluc, and USA300/Eno‐Antares2 were determined as previously described ⁴⁸ . Briefly, S. aureus strains of interest were cultured in BHI broth at 37°C overnight with shaking, and 0.2 ml of each culture was added to 20 ml of fresh BHI broth in a sterile 50 ml flask. The OD₆₀₀ values were measured every hour for 24 h after inoculation. The OD₆₀₀ values over the cultivation time were used to draw the growth curves.

Detection of BL signals of S. aureus reporter strains in tubes

S. aureus reporter strains USA300/Eno‐Nluc, USA300/Eno‐Teluc, and USA300/Eno‐Antares2 were separately cultured in BHI broth at 37°C to an OD₆₀₀ of 2.0. Next, the bacterial cells were harvested by centrifugation at 10,000g for 10 min at 4°C and adjusted to 1 × 10⁹ CFU/ml with PBS. BL images of the test tubes containing 100 µl of S. aureus reporter suspension and 10 µl of individual substrate (HFZ, DTZ, or FUR, 3 mM for each) were taken in a dark room using a V40 digital camera (HONOR) with parameters of exposure time of 1 s, ISO 16,200, and f‐number of 1.9.

Evaluation of the stability of S. aureus reporter strains in vitro

The strains of interest were adjusted to 1 × 10⁶ CFU/ml with PBS (pH 7.3), 50 µl of bacterial solution was mixed with 50 µl of HFZ (100 µM) in a thermostatic metal bath at different temperatures (0°C–50°C) or bacteria were prepared with PBS under different pH (4–10). BL signals were detected using the IVIS® Lumina LT system 1 min after the addition of a substrate.

The stability of S. aureus reporter strains was also tested in other physiochemical conditions, including PEG‐300 formulation, HEPES (1 M, pH 7.3), TE (1×, pH 7.3), NS, RPMI‐1640, or RPMI‐1640 plus 10% FBS.

BL imaging of the engineered S. aureus reporter strains in vitro

To compare the BL intensity of the engineered S. aureus reporter strains in vitro, the strains of interest were cultured in BHI and adjusted to 1 × 10⁷ CFU/ml with PBS. A total of 50 µl of S. aureus suspension was individually mixed with an equal volume of the substrate at diverse concentrations (3.125, 6.25, 12.5, 25, 50, and 100 µM) in a black 96‐well plate and then BL signals were measured using the IVIS® Lumina LT system. To analyze the linear correlation between BL signals and bacterial numbers, S. aureus reporter strains USA300/Eno‐Nluc, USA300/Eno‐Teluc, and USA300/Eno‐Antares2 were cultured, and bacterial solutions of 1 × 10² to 1 × 10⁶ CFU/ml in PBS were prepared, 50 µl of bacterial solution was mixed with 50 µl of HFZ (100 µM) in a black 96‐well plate; BL images and signals were collected using the IVIS® Lumina LT system 1 min after the addition of a substrate. Wild‐type S. aureus USA300 was used as a negative control. The parameters for image acquisition were as follows: emission filter open for total BL flux or a 591 ± 10 nm emission filter used for red BL signals; exposure time, 1 to 10 s; binning factor, 4; field of view, 12.5 cm; and f‐stop, 1.

To determine whether HFZ has any effects on the reporter strain, 0.2 ml of USA300/Eno‐Antares2 culture was added to 20 ml of fresh BHI broth containing diverse concentrations of HFZ (12.5, 25, 50, and 100 µM) in a sterile 50 ml flask. The OD₆₀₀ values were measured every hour for 24 h after inoculation. The growth curve was drawn. Then, the bacterial number was counted using the plate dilution assay.

The toxicities of different concentrations of NM (1.25, 2.5, 5.0, and 10 µM), dimethyl sulfoxide (1.25% (v/v), 2.5%, 5%, and 10%), ERM (4, 8, 16, and 32 µg/ml), or VAN (0.125, 0.25, 0.5, and 1 µg/ml) to the USA300/Eno‐Antares2 reporter strain were also evaluated with black 96‐well plates in vitro.

Emission spectra analysis of BL signals from the engineered S. aureus reporters

To determine the emission spectra of Eno‐fused Nluc‐based luciferases in S. aureus, the reporter strains of interest were separately cultured at 37°C to an OD₆₀₀ of 2.0, and then bacterial cells were harvested by centrifugation at 10,000g for 10 min at 4°C and adjusted to 1 × 10⁹ CFU/ml with PBS. 100 µl of S. aureus suspension of USA300/Eno‐Nluc, USA300/Eno‐Teluc, or USA300/Eno‐Antares2 was mixed with 10 µl of individual substrate (FUR, DTZ, or HFZ, 3 mM for each) in a black 96‐well plate. After 1 min of reaction, the BL spectra were collected using a SpectraMax M2 (Molecular Devices) with 1‐nm increment from 400 to 700 nm. The spectra of Eno‐Teluc and Eno‐Antares2 were normalized to the Nluc emission peaked at 460 nm. The experiment was repeated three times in triplicate.

Selection of the reporter strain/substrate pair for deep‐tissue BL imaging

Based on the emission spectra analysis in vitro, a mouse S. aureus bacteremia model was generated and used to screen the reporter strain/substrate combination for deep‐tissue BL imaging. Female BALB/c mice (n = 3 for each combination) were administered intravenously with 1 × 10⁷ CFU of the USA300/Eno‐Teluc or USA300/Eno‐Antares2 reporter strain. After 24 h, the infected mice were injected with 1 µmol of substrate (HFZ, DTZ, or FUR) into the peritoneal cavity. Wild‐type USA300‐challenged mice served as negative controls. BL signals in the kidneys from the dorsal sides of mice were detected using the IVIS® Lumina LT system 3 min after injection of a substrate ²² . Images were acquired every 3 min and the maximal signal data were chosen for subsequent analysis. The bacterial loads in the kidneys of infected mice were also determined using the plate dilution assay as described previously ⁴⁵ . The normalized BL intensity was calculated by dividing the BL signals by the total number of viable bacteria in each kidney.

BL imaging of S. aureus USA300/Eno‐Antares2 in a mouse bacteremia model

To determine the sensitivity of BL imaging of S. aureus USA300/Eno‐Antares2 systemic dissemination, BALB/c mice (n = 9) were infected intravenously with 1 × 10⁶ CFU of the reporter strain. After 6 h of infection, 1 µmol of HFZ was intraperitoneally injected, and in vivo BL signals were obtained from the dorsal sides of mice 3 min after injection of the substrate. To monitor signal decay over time, the BL signals on the dorsal sides of mice were measured every 3 min up to 24 min. The bacterial loads in the kidneys of infected mice were determined using the plate dilution assay. In vivo deep‐tissue BL sensitivity was determined as the lowest bacterial CFU in the kidneys of mice that can still emit a bioluminescent signal.

To evaluate the correlation between deep‐tissue BL signals and bacterial CFU numbers, BALB/c mice (n = 3/group) were infected intravenously with 1 × 10⁷ CFU of S. aureus USA300/Eno‐Antares2. Then, mice were intraperitoneally administered with 1 µmol of HFZ at 6, 12, and 24 h, and 2, 4, and 6 days postinfection. The in vivo BL signals were collected from the dorsal and ventral sides of mice. The bacterial loads in the kidneys and livers of infected mice were also determined using the plate dilution assay ⁴⁷ . The total flux from in vivo deep tissues (dorsal + ventral) and total bacterial CFU (kidney + liver) were used to assess the correlation between deep‐tissue BL signals and bacterial loads in mice.

BL imaging of S. aureus USA300/Eno‐Antares2 in a mouse skin infection model

To investigate the lower limit of sensitivity for BL imaging of S. aureus USA300/Eno‐Antares2 in mouse skin infections, BALB/c mice (n = 3 for each dose group) were fully anesthetized with 1% (m/v) pentobarbital sodium (50 mg/kg) and the back hairs were depilated completely with 6% (m/v) sodium sulfide as described ³⁸ . A mixture containing 50 µl of 1 × 10¹ to 1 × 10⁵ CFU of S. aureus USA300/Eno‐Antares2 and 50 µl of HFZ (100 µM) was subcutaneously injected into the dorsal back skin of a mouse. 1 × 10⁵ CFU of wild‐type USA300 injected mice served as negative controls. BL signals were measured using the IVIS® Lumina LT system 1 min after injection of the mixture. The lower limit of sensitivity for skin BL imaging was described as the lowest bacterial CFU in the skin of infected mice that can still emit a bioluminescent signal.

For skin‐tissue BL imaging, BALB/c mice (n = 6 per group) were infected subcutaneously with 1 × 10⁷ CFU of S. aureus USA300/Eno‐Antares2 in the dorsal skin as described previously ³⁸ . After 6, 12, and 24 h, and 2, 4, and 6 days of infection, each mouse was injected subcutaneously with 100 µl of HFZ (100 µM) around the abscess and imaged. The mice injected with USA300 served as negative controls. After BL imaging, mice were killed, and bacterial burdens in the skin were determined. Total BL flux and skin bacterial CFU were used to assess the correlation between BL signals and skin bacterial CFU loads.

Determination of HFZ toxicity in mice

Female BALB/c mice (n = 3) were subcutaneously infected with 100 µl of HFZ (100 µM) in PBS every day for 11 days. The body weights of the mice were measured every other day. Blood samples were collected 12 h after the last injection of HFZ, and the levels of IL‐6 and TNF‐α in mouse sera were determined using an enzyme‐linked immunosorbent assay (ELISA).

Evaluation of the antimicrobial efficacy of antibiotics against S. aureus infection using BL imaging

According to the linear correlation between in vivo BL signals and bacterial CFU counts, a mouse S. aureus skin infection model was used to longitudinally evaluate the antibiotic therapeutic efficacy using USA300/Eno‐Antares2/HFZ BL imaging. Briefly, BALB/c mice (n = 5/group) were subcutaneously infected with 1 × 10⁷ CFU of S. aureus USA300/Eno‐Antares2 in the dorsal skin. The dose of vancomycin in patients was recommended to be 30 mg/kg/day or greater ⁴⁹ . Therefore, the mice in the treatment group were each injected intraperitoneally with 8 mg/kg, 16 mg/kg, or 32 mg/kg of ERM or VAN twice a day for 11 days. The NS‐injected mice served as controls. The abscess area was assessed every other day by the maximal length × width of the developing ulcer, as previously described ⁵⁰ . The body weights of mice were also measured every other day for up to 11 days. On 1, 3, 5, 7, 9, and 11 days postinfection, mice were anesthetized and each mouse was injected subcutaneously with 100 µl of HFZ (100 µM) around the skin abscess. Total BL signals from the dorsal sides of mice were measured using the IVIS® Lumina LT system 3 min after injection of a substrate. Bacterial loads in the skin of mice after 11 days of antibiotic therapy were determined and compared, and IL‐6 levels in mouse sera were determined by enzyme‐linked immunosorbent assay.

Statistical analysis and reproducibility

Statistical analysis was carried out using GraphPad Prism (GraphPad Software, version 9.2.0.) and SPSS (IBM SPSS Statistics 24). Replicate times and statistical tests for certain experiments are presented in the figure legends. Error bars indicate the standard error of the mean.

AUTHOR CONTRIBUTIONS

Weilong Shang: Data curation (equal); formal analysis (equal); funding acquisition (equal); investigation (equal); methodology (equal); visualization (equal); writing—original draft (lead). Zhen Hu: Data curation (equal); methodology (equal); project administration (equal); writing—original draft (supporting). Mengyang Li: Data curation (equal); investigation (equal); methodology (equal); visualization (equal); writing—original draft (supporting). Yuting Wang: Data curation (equal); methodology (equal); software (equal). Yifan Rao: Methodology (equal); software (equal); validation (equal). Li Tan: Methodology (supporting); software (supporting); supervision (supporting). Juan Chen: Investigation (supporting); methodology (supporting); software (supporting). Xiaonan Huang: Investigation (supporting); methodology (supporting); validation (equal). Lu Liu: Methodology (supporting); software (supporting). He Liu: Data curation (supporting); investigation (supporting); methodology (supporting). Zuwen Guo: Investigation (supporting); methodology (supporting); visualization (supporting). Huagang Peng: Data curation (supporting); investigation (supporting). Yi Yang: Methodology (equal); validation (equal); visualization (equal). Qiwen Hu: Methodology (supporting); supervision (supporting). Shu Li: Investigation (supporting); supervision (supporting). Xiaomei Hu: Conceptualization (equal); supervision (equal); writing—review and editing (supporting). Jiao Zou: Conceptualization (supporting); investigation (equal); methodology (equal); supervision (equal); validation (equal); writing—review and editing (supporting). Xiancai Rao: Conceptualization (lead); funding acquisition (equal); resources (equal); supervision (equal); writing—review and editing (lead).

ETHICS STATEMENT

Female BALB/c mice (6–8 weeks old, 16–20 g) were purchased from the Laboratory Animal Center of Third Military Medical University (Army Medical University). All animal experiments were approved by the Institutional Animal Care and Use Committee of the Third Military Medical University (protocol no. #SYXK‐YU‐20170002).

CONFLICT OF INTERESTS

The authors declare no conflict of interests.

Supporting information

Supporting information.

Shang W, Hu Z, Li M, Wang Y, Rao Y, Tan L, et al. Optimizing a high‐sensitivity NanoLuc‐based bioluminescence system for in vivo evaluation of antimicrobial treatment. mLife. 2023;2:462–478. 10.1002/mlf2.12091

DATA AVAILABILITY

All the data are available in the main text or in the supplementary information materials. All vector information is provided in Table S1.