Longitudinal proliferation mapping in vivo reveals NADPH oxidase-mediated dampening of Staphylococcus aureus growth rates within neutrophils

Upon the onset of inflammatory responses, bacterial pathogens are confronted with altered tissue microenvironments which can critically impact on their metabolic activity and growth. Changes in these parameters have however remained difficult to analyze over time, which would be critical to dissect the interplay between the host immune response and pathogen physiology. Here, we established an in vivo biosensor for measuring the growth rates of Staphylococcus aureus (S. aureus) on a single cell-level over days in an ongoing cutaneous infection. Using intravital 2-photon imaging and quantitative fluorescence microscopy, we show that upon neutrophil recruitment to the infection site and bacterial uptake, non-lethal dampening of S. aureus proliferation occurred. This inhibition was supported by NADPH oxidase activity. Therefore, reactive oxygen production contributes to pathogen containment within neutrophils not only by killing S. aureus, but also by restricting the growth rate of the bacterium.

www.nature.com/scientificreports www.nature.com/scientificreports/ dye-based proliferation measurement approaches have been established for bacteria, the necessity of fluorophore loading or chemical pulsing of fluorescence expression limited the use of most of these systems to only few hours after infection [14][15][16] .
Here, we established an in vivo biosensor for measuring S. aureus proliferation on a single cell-level at any given time point post infection. This has enabled us to map the bacterial growth rate side by side with the induction of the innate immune response. We show that upon neutrophil recruitment to the site of infection, the bacteria were efficiently taken up into the phagocytes. Concomitantly, S. aureus growth was significantly reduced, which was aided by functional NADPH oxidase. Taken together, our data suggest that ROS production can contribute to the pathogen containment by non-lethal restriction of S. aureus proliferation within neutrophils.

Results
A biosensor for real-time measurement of S. aureus growth rate. It has been difficult so far to determine S. aureus growth rates in vivo, especially beyond the first few rounds of cell division after inoculation 15,16 . Therefore, we generated a proliferation biosensor based on dilution and de novo production of a photoconvertible fluorescence protein 17,18 . In this approach, the green mKikumeGR 19 is expressed constitutively from a plasmid (termed pKikume hereafter) with optimized regulatory sequences (see Supplementary methods and Supplementary Fig. S1a-d). The protein could be photoconverted by violet light (Fig. 1a,b and Supplementary  Fig. S1e) at any given time point. Thus, we reasoned that bacterial proliferation should be measurable by the recovery from photoconversion through dilution of red (photoconverted) and de novo production of green (non-photoconverted) protein taking place in growing, but not in proliferation-inactive bacteria (Fig. 1a). To evaluate the approach, S. aureus carrying the pKikume reporter plasmid (S. aureus-pKikume) was examined by time lapse microscopy after photoconversion in vitro. Recovery of the original, green fluorescence was detectable after photoconversion in dividing bacteria in a time-dependent manner, whereas the non-dividing bacteria retained the red, photoconverted fluorescence (Fig. 1b,c). Quantitative analysis at a single bacteria level showed a very robust change in the mKikume red/green fluorescence ratio over time for dividing, but not for non-dividing bacteria (Fig. 1c,d). Furthermore, flow cytometry analysis of exponentially growing S. aureus showed that complete recovery from photoconversion occurred within 90 min (Fig. 1e,g).
Recovery from photoconversion could, in principle, also occur through fluorescence protein turnover in non-dividing bacteria. In order to address the contribution of such a growth-independent impact on biosensor readout, we employed division-incompetent, but metabolically active S. aureus generated using psoralen and UV DNA crosslinking 20,21 . These bacteria were completely non-proliferative, but unaffected in their membrane potential (Fig. 2a,c). However, for the division-incompetent S. aureus-pKikume, we observed a significant reduction in recovery from photoconversion as compared to controls (Fig. 2d). Therefore, indeed bacterial proliferation, and not protein turnover, is mainly responsible for the fluorescence recovery readout of the biosensor.
Longitudinal S. aureus growth measurement in vivo shows maximal proliferation dampening upon the arrival of neutrophils at the site of infection. Neutrophils represent the first line of cellular defenses against S. aureus infection. Upon recruitment from the bloodstream, these professional phagocytes are crucial for the containment and degradation of microorganisms at the site of infection 5 . Consequently, while S. aureus burden in the tissue increased significantly between 3 and 16 h p.I., bacterial numbers had dropped below initial numbers by 48 h p.I. (Fig. 3a). This was in line with a significant recruitment of CD45 + leukocytes at the site of infection at 16 h p.I., which mainly consisted of Ly6G + CD11b + neutrophils (Fig. 3b). The decline of pathogen numbers could be in principle achieved by killing of a constantly growing bacterial population, or, alternatively, by also inhibiting the individual bacterial growth rates. In order to analyze whether S. aureus growth rate was changed upon the massive neutrophil recruitment, we compared bacterial recovery from photoconversion at 3 versus 16 h p.I. by intravital 2-photon microscopy. To this end, B6 albino mice were infected intradermally in the ear with S. aureus-pKikume. Control immunofluorescence staining of S. aureus in fixed ear tissues showed that all bacteria at the site of infection expressed the mKikume protein and would thus be detectable during intravital 2-photon microscopy ( Supplementary Fig. S2). Bacteria at the site of infection were photoconverted by 405 nm illumination at 3 h post infection (p.I.) and immediately imaged by intravital 2-photon microscopy for subsequent 60 min (Fig. 3c). Similar to the observations in the in vitro system, substantial increase in green and decrease in red mKikume fluorescence was detected after photoconversion ( Supplementary Fig. S3a, Supplementary movie 1). Importantly, the mKikume red/green fluorescence ratio and consequently a proliferation index calculated from this ratio, reached a plateau within 60 min ( Supplementary Fig. S3b,c), which was comparable with the kinetics observed in vitro (see Fig. 1f,g). This suggests that 3 h p.I., S. aureus proliferation in the tissue can reach rates that are comparable to exponential growth in vitro. Strikingly, we observed a significant decline in the rate of recovery from photoconversion at 16 h p.I. compared to 3 h p.I. (Fig. 3d,e, Supplementary movie 2). Of note, recovery from photoconversion was not completely abrogated at 16 h p.I. Thus, we concluded that at 16 h p.I., most of the observed S. aureus remained alive initially, but exhibited a reduced growth as compared to 3 h p.I. As the bacterial numbers at 48 h p.I. were nearly 100 times lower than at 16 h p.I. and thus complicated a robust determination of proliferation rates from intravital 2-photon microscopy, we devised a quantitative confocal microscopy approach based on cryosections of S. aureus-pKikume infected ears fixed 60 min after photoconversion. Automated determination of mKikume red and green fluorescence of these samples allowed for measuring proliferation rates in a large number of tissue sections in a standardized fashion. Confirming our intravital 2-photon microscopy observations, proliferation rates determined from automated analysis of tissue sections were dramatically lower at 16 h versus 3 h p.I., but did not significantly deline further at 48 h p.I. (Fig. 3f,g). Thus, concomitantly with the onset of the innate immune reaction, the growth rate of the S. aureus at the site of infection rapidly decreases.
www.nature.com/scientificreports www.nature.com/scientificreports/ NADPH oxidase activity contributes to the dampening of S. aureus growth. The production of reactive oxygen by NADPH oxidase constitutes a major antimicrobial effector mechanism of neutrophils. In addition to its importance for oxidative burst acting directly against phagocytosed microorganisms, superoxides produced by this enzyme can trigger the release of NETs 7,8 . Whether NADPH oxidase activity can impair pathogen proliferation rates or mediate direct killing of S. aureus without overt impact on its proliferation has remained completely unknown. We therefore set out to analyze S. aureus proliferation rates in NADPH oxidase-deficient cybb −/− mice. We observed an increased recruitment of both neutrophils and monocytes to the infected ears of cybb −/− animals, suggesting that NADPH oxidase-deficiency resulted in a loss of pathogen containment and increased inflammation (Fig. 4a, Supplementary Fig. S4). Also, at 16 h. p.I., bacterial burden was found to be significantly increased in cybb −/− animals (Fig. 4b). Strikingly, analysis using the proliferation biosensor revealed a significantly higher S. aureus growth rates in cybb −/− mice compared to wildtype both at 16 h and 48 h, but not at 3 h p.I. (Fig. 4c,d). Of note, the bacterial growth rates were not completely rescued to the levels found at 3 h p.I. in the cybb −/− mice, suggesting that besides reactive oxygen production, other mechanisms limit bacterial growth, which could also explain that ultimately, the infection is controlled in both experimental groups in our cutaneous www.nature.com/scientificreports www.nature.com/scientificreports/ infection model (see Fig. 4b). Nevertheless, our data show that NADPH oxidase contributes to the reduction of bacterial numbers by dampening S. aureus growth.

S. aureus is located within neutrophils at time points of efficient confinement. Neutrophil
NADPH oxidase can be involved in control of both extracellular bacteria via NETs 8 , as well as intracellular bacteria via oxidative damage 7,22 . In order to visualize the interaction of neutrophils with S. aureus upon recruitment to the site of infection and to analyze bacterial localization, we employed Catchup IVM mice, in which neutrophils exhibit red fluorescence dependently of neutrophil-specific Ly6G expression 23 . When we infected these mice with GFP-expressing S. aureus (S. aureus-pGFP), we observed using intravital 2-photon microscopy 3 h p.I. the uptake of bacteria by neutrophils ( Fig. 5a,b, Supplementary movie 3). Interestingly, quantitative analysis of S. aureus cellular localization revealed that while the small number of neutrophils present at 3 h p.I. had taken up a fraction of the bacteria, the majority of S. aureus was found within these cells at 16 h p.I. (Fig. 5c-e). Thus, we conclude that S. aureus is internalized by recruited neutrophils right after the onset of the innate immune response and bacterial growth is mainly controlled intracellularly. The intracellular environment in cybb −/− versus wildtype host cells could differentially impact on the proliferation biosensor readout instead of proliferation itself. To exclude this possibility, we compared the recovery from photoconversion of untreated versus division-incompetent S. aureus after uptake into bone marrow-isolated neutrophils from cybb −/− and wildtype mice. S. aureus which were division-incompetent (see Fig. 2) showed substantially less recovery from photoconversion both in cybb −/− as well as wildtype neutrophils, indicating that proliferation measurement per se is not affected by the presence or absence of NADPH oxidase ( Supplementary Fig. 5a). In line with our observations in vivo (see Fig. 4), we could observe that dilution of mKikume red fluorescence in cybb −/− neutrophils is slightly enhanced in comparison to both wildtype neutrophil infection as well as division-incompetent S. aureus ( Supplementary Fig. 5b). Thus, also in vitro, cybb −/− neutrophils might be permissive for S. aureus division.
Therefore in summary, our data suggest that NADPH oxidase-mediated dampening of S. aureus growth rates upon uptake into neutrophils contributes to control of the infection. www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
The emergence and spread of antibiotic-resistant S. aureus demands a better understanding of the persistence mechanisms, lifestyle and physiological state of S. aureus at sites of infection. The bacterial growth rate is a fundamentally important parameter for elucidating these questions, but has remained difficult to measure so far. Determination of the bacterial burden via colony forming units is insufficient for this purpose, as it cannot distinguish whether pathogen numbers are controlled via killing of the bacteria, or reduction of the proliferation rate, which in principle can be non-lethal for the microbes. To circumvent this, dilution-based approaches have been applied for measuring the proliferation of the bacteria shortly after infection, e.g. by pulsing S. aureus with www.nature.com/scientificreports www.nature.com/scientificreports/ fluorescent dyes 16,24,25 , or with chemical inducers that trigger a short peak of fluorescence protein production 14 . However, this limits the window of measurement to only few division cycles after infection, or to the homogenous accessibility of the tissue of interest for the inducer, respectively. Protein maturation-dependent fluorescent timers have been introduced as a pulse-free alternative for measuring the de novo production of bacterial protein, but are strongly influenced by the oxidative state of the microenvironment 26 . As this approach is not based on the fluorescence protein levels, but maturation kinetics, its use for the investigation of innate immune defense mechanisms is complicated 27 .
We show that a photoconversion-based biosensor can be used for probing bacterial growth in a non-invasive fashion. Very much in contrast to alternative fluorescence dilution-based approaches, our system allowed bacterial proliferation measurement at any time point during the ongoing infection, even days after inoculation. This has enabled us to dissect for the first time the impact of NADPH oxidase on S. aureus growth rates in vivo not only before and after neutrophil recruitment, but also at later phases when the pathogen is about to be cleared. Of note, the arrival of neutrophils and phagocytosis of the S. aureus coincided with a reduction of bacterial growth. Importantly however, we show a slight recovery from photoconversion of S. aureus also at 16 h p.I., when the vast majority of the bacteria is located within recruited neutrophils. It is possible that such a residual proliferative activity is limited to the small population of bacteria not locates within neutrophils at this time point. However, the observation is interesting with regard to in vitro data suggesting that S. aureus growth is not completely abolished within professional phagocytes. For example, bacterial proliferation has been shown in the lysosomal compartment of macrophages 15 and dendritic cells 28 . Similarly, it was shown that the bacteria can grow within Kupffer cells in the liver after a systemic infection 16 . For neutrophils, intracellular survival has been shown, however direct evidence for intracellular replication is still lacking 10 . Of note, we show that compared to the mainly extracellular bacteria at 3 h p.I., pathogen growth is substantially reduced at phases at which the S. aureus have been phagocytosed by neutrophils. The situation could be different in NADPH oxidase deficient neutrophils: Although we found that S. aureus burden was significantly elevated in NADPH oxidase deficient mice after, but not before neutrophil recruitment to the site of infection, the strong reduction in the proliferation of S. aureus after neutrophil recruitment was partially rescued under NADPH oxidase deficiency. Therefore, we conclude that the recruited phagocytic cells rely on NADPH oxidase not only to directly kill S. aureus, but also to restrict bacterial growth www.nature.com/scientificreports www.nature.com/scientificreports/ non-lethally. At time points at which NADPH oxidase deficiency results in elevated pathogen burden, most S. aureus are located within neutrophils. This suggests that the observed contribution of ROS production to the inhibition of bacterial growth is taking effect within the recruited phagocytes after uptake of the pathogen.
Taken together, our proliferation reporter system enables us to dissect non-invasively, and throughout the whole course of the infection in the tissue, the impact of the neutrophil response on S. aureus in vivo. This approach has the potential of providing a better understanding of the interplay between host immune defenses and the physiological adaptations of the pathogen during infection.
All Bacterial strains, media and cultivation. Bacteria were cultivated in Brain Heart Infusion Broth (Carl Roth) at 37 °C with shaking or on 1.5%-agar plates (Carl Roth), supplemented with 12.5 µg/mL chloramphenicol (Roth). For OD 600 measurement BHI medium was used as reference and the culture was diluted with BHI medium to obtain values between 0.01 and 1.0. For calculations OD 600 = 1 was assumed to correspond to 10 8 bacteria/ml.

Generation of division-incompetent, but metabolically active bacteria.
To generate division-incompetent, but metabolically active bacteria, pyrimidine bases of the DNA were crosslinked by psoralen with long-wavelength UVA light 20,21 . For this, 10 µM of 4′-(aminomethyl)-4,5′,8-trimethylpsoralen hydrochloride (AMT, Sigma Aldrich) were added to the S. aureus culture. 0.5 ml of this sample were illuminated in a petri dish on ice with violet light at 357 nm wavelength using an assembly of 3 × 3 LED diodes (Strato, half-viewing angle: 10°; Radiant Power: 10 mW) for 10 min in a distance of 0.8 cm.

Measurement of membrane integrity. To measure the viability of the bacteria, the BacLight TM Bacterial
Membrane Potential Kit (Molecular Probes, Invitrogen) was used according to manufacturer's instructions and measured by flow cytometry. As control for intact membrane potential, a S. aureus day culture was used in which the membrane potential was abolished using the proton ionophore carbonyl cyanide 3-chlorophenylhydrazone (CCCP). As indicator, 3,3′-Diethyloxa-carbocyanine iodide (DiOC 2 ) was used, in which green fluorescence (measured at 488 nm excitation and 530/30 nm emission) shifts to red fluorescence (measured at 488 nm excitation and 670/30 nm emission) by self-association in high cytosolic concentrations.

S. aureus infection.
A day culture was inoculated by an overnight culture (16-18 h) to an OD 600 0.04-0.06.
The bacteria are harvested for infection at OD 600 0.4-0.6, which represented the early exponential growing phase. After washing with cold PBS, the OD 600 was adjusted to 2.5 in PBS by centrifugation (9000 g, 5 min, 4 °C) and resuspension. The ear skin of mice was intradermally infected at three sites per ear, each with 5 × 10 4 bacteria using a 35 gauge syringe.

Determination of bacterial tissue burden.
To analyze colony-forming units in the ear, bacteria were isolated from infected ears by homogenization with 15 mL disposable tissue grinders (Fisherbrand) or 15 mL tapered tissue grinders (Wheaton). The cells were centrifuged (16500 g, 3 min, 4 °C) and lysed with 1% triton-X 100 (Roth) in H 2 O. The bacteria were serially diluted in triplicates and plated on agar plates. photoconversion. For in vitro photoconversion, S. aureus-pKikume were illuminated in a 96-well plate with violet light at 405 nm wavelength using an assembly of 2 × 2 LED diodes (Strato, half-viewing angle: 15°; Radiant Power: 10 mW) for 1 minute in a distance of 1.7 cm. In Fig. 1a, the in vitro photoconversion of S. aureus-pKikume was done by illumination in a petri dish with violet light at 357 nm wavelength by assembling 3 × 3 LED diodes (Strato, half-viewing angle: 10°; Radiant Power: 10 mW) for 5 minutes in a distance of 0.8 cm.
Ears of anaesthetized mice were photoconverted with violet light at 405 nm wavelength using an assembly of 3 × 3 LED diodes (Strato, half-viewing angle: 15°; Radiant Power: 10 mW), for intravital 2-photon microscopy illumination tuime was set to 1-2 minutes in a distance of 2 cm, for confocal microscopy, the ears were illuminated from each side for 1 minute in a distance of 1.5 cm Flow cytometry. Flow cytometry of bacteria was performed in culture medium or diluted to suitable concentrations with cold PBS. The red mKikume fluorescence signal was measured at 488 nm excitation and 610/20 nm emission, the green mKikume fluorescence signal was measured at 488 nm excitation and 530/30 nm emission.
Intravital two-photon microscopy. The mice were anesthetized and prepared for intravital two-photon imaging. The mouse was placed on a heating stage adjusted to 37 °C and the ear was fixed to a metal platform. A coverslip sealed to a surrounding parafilm blanket was placed onto the ear and glued to the platform. Two-photon imaging was performed Zeiss LSM 700 equipped with a Mai Tai DeepSee Ti:Sa laser (Spectra-Physics) tuned at 920 nm (S. aureus-pKikume in B6 albino mice) or 960 nm (S. aureus-GFP in Catchup IVM mice) and a W Plan-Apochromat 20x/1.0 DIC VIS-IR dipping objective (Zeiss). For S. aureus-pKikume, the emitted signal was split by 490 nm, 520 nm and 555 nm long pass and dichroic mirrors and filtered with 509/22 nm, 500/50 nm (green mKikume) and 589/54 nm (red mKikume) filters before collection with non-descanned detectors. For S. aureus-GFP in Catchup IVM mice the emitted signal was splitted by 555 nm long pass and 490 nm long pass and dichroic mirrors and filtered with 565/10 nm (red signal, neutophils), 485 nm short pass (second harmonics) and 500/50 nm (GFP signal) filters before collection with non-descanned detectors. Measurement was performed with the TCS SP8 confocal laser scanning microscope (Leica Microsystems) using a 63x/1.40 Oil CS2 objective. For proliferation analysis the red mKikume fluorescence signal was measured at 561 nm excitation and 598-638 nm emission, while the green mKikume fluorescence signal was measured at 488 nm excitation and 526-570 nm emission.
To extract data of S. aureus uptake by neutrophils, the red tdTomato fluorescence signal (neutrophils in the Catchup IVM mice) was measured at 561 nm excitation and 600-669 nm emission, while the green GFP fluorescence signal (S. aureus-pGFP) was measured at 488 nm excitation and 499-565 nm emission. Image analysis. Bacteria in intravital 2-photon microscopy movies were segmented using the surface detection function of the Imaris software (Bitplane) from combined red and green mKikume fluorescence channels. Fluorescence values and time points of the segmented bacteria for the individual channels were converted into flow cytometry datasets using the DiscIT software 29 and analyzed using the FlowJo X software (FlowJo, LLC).
For analysis of confocal microscopy data, 144 × 144 μm frames spaced at least 2.5 μm apart were used to detect bacteria in a combined green and red mKikume channel using the threshold and analyze particle functions of the Fiji software (NIH, http://rsb.info.nih.gov/ij/). Fluorescence data from the resulting regions of interest were extracted, converted into flow cytometry datasets using the DiscIT software 29 and analyzed using the FlowJo X software (FlowJo, LLC). The combined proliferation indices of all detected bacteria within one frame were used for data representation.

Data Availability
All relevant data shown included in the manuscript and supplemental material, raw data are available without restriction from the corresponding author upon request.