Intravital two-photon microscopy of host–pathogen interactions in a mouse model of Staphylococcus aureus skin abscess formation
Corresponding authorCurrent affiliation:
Interfakultäres Institut für Mikrobiologie und Infektionsmedizin Tübingen, Abteilung Medizinische Mikrobiologie und Hygiene, Universität Tübingen, Tübingen, Germany
Program of Molecular Pathogenesis, Helen L and Martin S. Kimmel Center for Biology and Medicine, Skirball Institute of Biomolecular Medicine, New York University School of Medicine, New York City, NY, USA
Program of Molecular Pathogenesis, Helen L and Martin S. Kimmel Center for Biology and Medicine, Skirball Institute of Biomolecular Medicine, New York University School of Medicine, New York City, NY, USA
Program of Molecular Pathogenesis, Helen L and Martin S. Kimmel Center for Biology and Medicine, Skirball Institute of Biomolecular Medicine, New York University School of Medicine, New York City, NY, USA
Staphylococcus (S.) aureus is a frequent cause of severe skin infections. The ability to control the infection is largely dependent on the rapid recruitment of neutrophils (PMN). To gain more insight into the dynamics of PMN migration and host–pathogen interactions in vivo, we used intravital two-photon (2-P) microscopy to visualize S. aureus skin infections in the mouse. Reporter S. aureus strains expressing fluorescent proteins were developed, which allowed for detection of the bacteria in vivo. By employing LysM-EGFP mice to visualize PMN, we observed the rapid appearance of PMN in the extravascular space of the dermis and their directed movement towards the focus of infection, which led to the delineation of an abscess within 1 day. Moreover, tracking of transferred labelled bone-marrow neutrophils showed that PMN localization to the site of infection is dependent on the presence of G-protein-coupled receptors on the PMN, whereas Interleukin-1 receptor was required on host cells other than PMN. Furthermore, the S. aureus complement inhibitor Ecb could block PMN accumulation at thesite of infection. Our results establish that 2-P microscopy is a powerful tool to investigate the orchestration of the immune cells, S. aureus location and gene expression in vivo on a single cell level.
Intravital two-photon (2-P) microscopy is a powerful tool to study the dynamics of immune responses and host–pathogen interactions in living animals (Germain et al., 2006; Velazquez et al., 2007; Richter-Dahlfors et al., 2012). Applied to various infection models this technique has been used to investigate the complex environment of the host in response to bacteria, parasites, fungi and viruses (Mansson et al., 2007a; Chtanova et al., 2008; Peters et al., 2008). The possibility to acquire information about cell behaviour in space and time makes intravital microscopy especially suited to study leucocyte recruitment in different organs, which is a hallmark of many types of infection and inflammatory processes (Kreisel et al., 2010; Herz et al., 2012).
Staphylococcus (S.) aureus is a major human pathogen that causes localized infections (predominantly in the skin and soft tissue) as well as systemic diseases (e.g. sepsis, endocarditis and organ abscesses). Greatly complicating the treatment of S. aureus infections has been the emergence of multi-resistant and hyper-virulent strains over the last decades (Lowy, 1998; Otto, 2010). An important component of S. aureus virulence is the organism's ability to withstand and undermine the host immune response, mediated by the production of virulence factors such as toxins, proteases, phenol-soluble modulins, inhibitors of neutrophil responses and antagonists of the complement cascade among others (Chan et al., 1998; de Haas et al., 2004; Foster, 2005; Rooijakkers et al., 2005; Jongerius et al., 2007). Global gene regulators control the expression of many of the virulence genes in S. aureus with the accessory gene regulator (Agr) being one of the best studied. The agr locus in S. aureus encodes a typical autoactivation circuit consisting of four components (AgrA–D), which translates bacterial cell density into different gene expression profiles (Ji et al., 1997; Dunman et al., 2001; Novick and Geisinger, 2008). AgrB and AgrD combine to generate the auto-inducing peptide (AIP). AgrA and AgrC represent a classical two-component signalling system. Extracellular AIP is sensed by AgrC, which activates AgrA, which in turn upregulates the agr P2 and P3 promoters. The downstream effects of Agr activation are mediated by a regulatory RNA (RNAIII), which is transcribed from the agr P3 promoter (Novick and Geisinger, 2008). Overall, low density (low AIP concentration) favours the expression of cell-surface molecules, which enable the bacteria to bind to host cells and intercellular matrix, whereas high cell density activates the quorum-sensing (QS) system, leading to the production of secreted virulence factors. Although SarA also belongs to the family of global gene regulators in S. aureus, the promoter SarA P1 has been used for constitutive labelling of bacterial cells in in vitro and in in vivo studies, because P1 becomes active very early during bacterial growth and exhibits strong activity (Cheung et al., 1998; Malone et al., 2009)
The cellular immune response to S. aureus is mainly mediated by neutrophils (PMN), which are rapidly recruited to the site of infection after intracutaneous inoculation (Molne et al., 2000). The importance of these cells in clearing the infection is underlined by the increased susceptibility to S. aureus of humans or mice with low PMN counts or compromised effector functions, such as defects in the enzyme NADPH oxidase, which mediates the killing of the pathogen via reactive oxygen species. Various animal studies have addressed the question of how S. aureus is recognized by host cells and what mediators are involved in PMN recruitment in different infection models. A role has been established for TLR2 (Kielian et al., 2005), NOD2 (Deshmukh et al., 2009), IL-1β (Miller et al., 2006; 2007), C5a (von Köckritz-Blickwede et al., 2010) and CXCR2 (Kielian et al., 2001), but there is little information available regarding the spatiotemporal orchestration of the early PMN migration during abscess formation, which is a hallmark of S. aureus infections (Cheng et al., 2011).
Here, we describe an experimental system, in which 2-P microscopy is successfully used to visualize the process of S. aureus skin abscess formation in a mouse model. We employed LysM-EGFP mice (Faust et al., 2000) to analyse PMN migration in response to infection with fluorescent S. aureus reporter strains. Signalling pathways in these migration events were investigated using adoptively transferred PMN. We also used the experimental set-up to visualize Agr activation in bacterial cells at the site of infection. Furthermore, we used the S. aureus complement inhibitor Ecb to analyse the role of the complement cascade for PMN recruitment in the skin.
Beyond the specific results, these data establish intravital 2-P microscopy as a valuable tool to investigate the dynamics of the cellular immune response in the skin after S. aureus infection, and to study host–pathogen interactions and bacterial gene expression at the single cell level in vivo.
Generation of fluorescent S. aureus reporter strains
In order to visualize S. aureus bacteria using 2-P microscopy, we first generated a set of bacterial reporter strains (Table 1). These express different fluorescent proteins (FP) from plasmids constructed with the pCN shuttle vector system (Table 2) (Charpentier et al., 2004). pCN54 was chosen as a starting point, because it encodes GFPmut2, which is suitable for expression in Gram-positive organisms (Cormack et al., 1996). Many transgenic or knock-in mouse strains, which are used for 2-P microscopy, also employ GFP as a reporter. For this reason we created S. aureus strains, which express the CFP-variant Cerulean (CER) (Rizzo et al., 2004) or the YFP-variant Venus (VEN) (Nagai et al., 2002), thereby enabling to discriminate between host cells and bacteria in imaging experiments. In order to adapt CER and VEN for the expression in Gram-positive (gp) organisms, we synthesized the genes encoding gpCER and gpVEN by using gfpmut2 as a template and introducing mutations, which encode the amino acid substitutions that lead to the previously described shift in the excitation/emission spectra (Nagai et al., 2002; Rizzo et al., 2004) (Supplementary Fig. S1).
In a first step, the Pcad promoter of pCN54 was replaced by the constitutively active sarA P1 promoter (Cheung et al., 1998; Malone et al., 2009), which yielded pJL-sar-GFP (Fig. 1A and Table 2). Then, we constructed pJL-sar-CER and pJL-sar-VEN by replacing the gfpmut2 gene in pJL-sar-GFP by the genes encoding gpCER (gpcer) and gpVEN (gpven) respectively (Fig. 1A and Table 2). After electroporation of these plasmids into the DNA acceptor strain RN4220 (Kreiswirth et al., 1983), the vectors were transduced to S. aureus RN6734 (Novick et al., 2000), which has been used for bioluminescence imaging in a skin infection model before (Wright et al., 2005). We observed both homogeneously labelled and brightly fluorescent cells in confocal images of the resulting strains RN6734-sar-GFP, RN6734-sar-CER and RN6734-sar-VEN respectively (Fig. 1B and Table 1, and data not shown). By replacing the sarA P1 promoter we then generated a set of reporter plasmids, in which the FP expression is controlled by the agr P3 promoter (pJL-agr-GFP, pJL-agr-CER, pJL-agr-VEN; Fig. 1A and Table 2). When we introduced pJL-agr-GFP into RN6734 (resulting in RN6734-agr-GFP), we observed that repeated dilution of cultures in stationary phase, and thereby reducing AIP concentration, led to a 42.1-fold decrease of the mean fluorescence intensity when analysed by FACS (Fig. 1C). This indicates that the transcriptional fusion construct faithfully reports Agr activity. As expected, the RN6734-sar-GFP reporter strain exhibited stable fluorescence, which decreased only 1.7-fold after dilution of the culture (Fig. 1C). When we investigated fluorescence in bacterial micro-colonies on solid media by time-lapse microscopy over a time span of 13 h, we observed the development of density-dependent fluorescence in colonies of RN6734-agr-CER, which started to emerge in the centre of the colony after 7 h (Supplementary Movie S1). In contrast, fluorescence was readily detectable and homogenously distributed in micro-colonies of RN6734-sar-GFP (Supplementary Movie S1). These data confirm the density-dependent expression of the Agr-reporter constructs.
Then we introduced the reporter plasmids into S. aureus strain Newman, which has been used extensively in various S. aureus infection models (Baba et al., 2008) (Table 1). In this strain, the sarA P1-reporter plasmid also led to constitutive labelling of the bacterial cells, whereas expression of FP from agr P3-reporter plasmids depended on cell density (Fig. 1D, and data not shown). As expected, a deletion mutant of the Newman strain, which lacks the agr locus (NewmanΔagr) and is therefore not able to participate in QS, was fluorescent only when GFPmut2 expression was controlled by the sarA P1 promoter, but not by the agr P3 promoter (Fig. 1D). This shows that the reporter system can be used in different relevant S. aureus model strains.
To test the virulence of the FP expressing strains and the stability of the plasmids in vivo, we injected 107 colony-forming units (cfu) Newman or Newman-sar-CER intradermally into the shaved flank skin of C57BL/6 mice in the absence of antibiotic selection for the plasmids. We found that mice in both groups developed dermonecrotic lesions, and the number of colonies recovered from the skin after 5 days did not differ between the two groups (Fig. 1E). To determine whether expression of FP was maintained in the absence of antibiotic selection in vivo, we plated serial dilutions of homogenized abscesses from five mice infected with Newman-sar-CER. After overnight growth on TSB agar plates (without antibiotics), we randomly picked 24 colonies from each abscess and checked for fluorescence using confocal microscopy. We found that of 120 tested colonies 118 (98.3%) still exhibited fluorescence. These findings show that the sarA reporter maintains the fluorescent protein gene during in vivo infection. Overexpression of FP does not affect S. aureus virulence, demonstrating its suitability for 2-P microscopy experiments.
Intravital 2-P microscopy of dermal S. aureus infection
To study the in vivo immune response during dermal S. aureus infection, we adopted 2-P microscopy to visualize the infection focus in the skin. Mice were infected in the shaved lateral flank skin by intradermal injection of the bacteria in a small volume (5–10 μl) of PBS. Immediately prior to imaging, a crescent incision to preserve blood flow was made in the skin of the deeply anaesthetized mouse and the dermis, after careful separation from the subcutaneous tissue around the infection site, was fixated on a glass coverslip (Fig. 2A). With the coverslip facing down, the mouse was mounted on a heated stage to maintain body temperature, and the stage was then placed in the light path of an inverted 2-P microscope. The mouse received oxygen via a nose cone during the experiment. Intravenous injection of quantum dots or labelled Dextran immediately before the surgery or during image acquisition demonstrated intact blood flow in the skin flap (Fig. 2B, and data not shown), which is a prerequisite for sufficient oxygenation of the tissue under imaging conditions and for access of blood-delivered host cells to the infection site.
One feature of 2-P microscopy is the detection of signals from second-harmonic generation (SHG), which indicates the presence of centrosymmetric structures, and which is often used to visualize extracellular matrix (e.g. collagen fibres) (Zipfel et al., 2003; Williams et al., 2005). When 106 cfu Newman-sar-VEN were injected into the skin of a C57BL/6 mouse 20 min before imaging, S. aureus embedded in the dermal tissue could be readily detected at the infection site (Fig. 2C). Z-stack analysis of the infection site revealed a band-like appearance of the bacterial inoculum and the presence of a layer of intact dermal matrix, indicating the restriction of the bacteria to the dermis without penetration into the subcutis (Fig. 2C). Single bacteria or clusters of S. aureus could be visualized at high magnification even when an inoculum as low as 1000 cfu was injected (Supplementary Fig. S2). This represents an inoculum, which is orders of magnitude smaller than the inocula conventionally used in S. aureus infection models (Bunce et al., 1992; Tarkowski et al., 2001; Wright et al., 2005).
Next, we set out to test whether our experimental system is also useful for investigating the spatial regulation of bacterial gene expression in vivo on a single cell level. Therefore, we employed our Agr-reporter strains and used 2-P microscopy to analyse the distribution of Agr activity within the infection focus. Previous studies using bioluminescence imaging have demonstrated that Agr activity reaches a maximum within hours after infection (Wright et al., 2005). Since the resolution of this technique is limited, it provides no information about the spatial distribution of Agr activity within the focus of infection. In order to be able to locate the infection site in the skin and investigate Agr activity, we injected a 1:1 mixture of Newman-sar-CER and Newman-agr-VEN. Repeated dilution of the culture of Newman-agr-VEN before infection ensured that Agr was inactive at the time point of infection. Surprisingly, when we visualized the infection site 8 h after injection of S. aureus, only a few discrete Agr-positive regions could be observed, and these were mainly located in the periphery of the bacterial focus (Fig. 2D). This was unexpected, because the concept of quorum sensing predicts that Agr upregulation should start in a location with high bacterial density (and high AIP concentration). These experiments demonstrate that 2-P microscopy is a useful approach to investigate bacterial gene expression at the single cell level in vivo.
PMN recruitment in response to S. aureus infection
In order to visualize the cellular immune response in our skin infection model, we took advantage of LysM-EGFP mice (Faust et al., 2000), in which myelomonocytic cells are labelled green. It has been demonstrated in the skin and other organs that PMN and monocytes/macrophages can be distinguished by their brightness: monocyte/macrophages express low levels of GFP (GFPlo), whereas neutrophils are GFPhi cells (Peters et al., 2008; Kreisel et al., 2010). However, it must be noted that during Listeria infection the levels of GFP in monocytes increased to levels similar to those in PMN over time (Waite et al., 2011). Our early time points in this study reflect the steady-state condition in which PMN are brightest. When we examined the flank skin of uninfected LysM-EGFP mice in our 2-P microscopy system, we observed the presence of GFPlo cells throughout the dermis, which most likely represent tissue macrophages (Fig. 3A, Supplementary Movie S2). When we analysed the migratory parameters of these cells, we found that they move with low velocity through the dermal matrix (Fig. 3B). Accordingly, they also exhibited low spatial displacement (distance between the location in the first and last image) over time (Fig. 3C). These findings are consistent with other reports of relatively sessile nature of macrophages in other sites (Lindquist et al., 2004; Schwickert et al., 2007). Only a few GFPhi cells could be observed, and these were mainly located in blood vessels (Fig. 3A), which indicates that these cells are circulating PMN in accordance with previous reports (Zinselmeyer et al., 2008; Kreisel et al., 2010). Prolonged imaging (> 2 h) lead to influx of GFPhi cells into the skin of uninfected LysM-EGFP mice, which we interpret to be a result of surgical trauma (data not shown). Therefore, in subsequent experiments imaging time was limited to < 2 h during which changes in PMN recruitment can be attributed in large part to manipulations prior to the surgery such as infection. Sham injections with sterile saline did not increase PMN recruitment. Mouse surgery was performed c. 15 min before the start of image acquisition.
Next, we infected LysM-EGFP mice intradermally with fluorescent S. aureus and visualized the focus of infection and the surrounding dermal tissue using 2-P microscopy. We noticed that within 2 h after infection with 104 cfu Newman-sar-CER, PMN have left the blood vessels and become detectable in the interstitial tissue in the vicinity of the bacterial focus (Fig. 3D). This was also the case when we used a 100-higher lower inoculum (106 cfu) (data not shown). Due to the nature of the tissue preparation, in which the dermis is held parallel to the 2-P laser-scanning plane by the coverslip, the movement of the cells could be tracked by acquiring x-y tile scans with a depth of excitation of 5 μm. The ability to focus on one plane and acquire x-y images was fortuitous, as the rapid movement of the PMN required that images were acquired every 15 s over a large area with sufficient resolution. The PMN exhibited clearly visible movement towards the bacterial focus (Fig. 3D, Supplementary Movie S3), which was confirmed by statistical analysis of the Psi (Ψ) angles (the angle between the trajectory of the cell during the time-lapse movie and the vector pointing towards the centre of the infection focus) (Chtanova et al., 2008; Waite et al., 2011) (Fig. 3E). If cells migrate randomly and unrestrictedly in all directions, the track angles are equally distributed for 2D tracking (Beltman et al., 2009). After S. aureus infection, the Ψ angles of the PMN tracks exhibited a skewing towards angles below 90°, which indicates preferential migration towards the focus of infection (Fig. 3E). Closer examination revealed that some of the PMN were already associated with individual S. aureus cocci or clusters in the vicinity of the infection site. It has been reported that PMN show a decreased cell speed after uptake of Leishmania parasites (Peters et al., 2008). Therefore, we analysed the migratory parameters of PMN that are in the immediate vicinity of bacteria and compared them to PMN in areas with no bacteria, but in infected skin, which is required to recruit PMN (Fig. 3F). We found that interaction with bacteria leads to a markedly decreased cell speed (Fig. 3, left panel) and lower track displacement (Fig. 3F, middle panel). The meandering index (cell displacement divided by track length), which is an indicator of track straightness, was also lower in this PMN subset (Fig. 3F, right panel). These data suggest that once PMN enter the extravascular tissue, they first show a directed migration towards the S. aureus focus with high velocity. Once they are in the immediate vicinity of S. aureus, PMN display a different migration pattern that is characterized by much lower cell speed and loss of directionality, resulting in the rapid accumulation of PMN within and around the S. aureus inoculum.
Surprisingly, high-resolution time-lapse movies demonstrated that PMN often passed single bacteria while migrating towards the centre of the infection. Occasionally, the PMN seemed to ‘sample’ bacterial clusters without uptake of S. aureus before moving on (Fig. 3G, Supplementary Movie S4). These data show that PMN are rapidly recruited to a S. aureus focus in the dermis, but that single cocci or clusters do not necessarily act as a stop-signal at 2 h after infection.
Clusters of PMN have accumulated at the site of S. aureus deposition 6 h after infection, while PMN in the surrounding tissue were still migrating towards the focus (Supplementary Fig. S3, Supplementary Movie S5). To confirm that the migrating GFPhi cells are indeed PMN, we treated LysM-EGFP mice 1 day prior to infection with a PMN depleting dose of RB6-8C5 antibody and visualized the site of infection by 2-P microscopy 6 h after injection of S. aureus. We observed that the cellular infiltrate of GFPhi cells was completely absent in PMN depleted mice, whereas mice injected with an isotype control antibody exhibited unaltered recruitment of PMN (Fig. 4A and B). Closer examination of the infection site revealed that GFPlo cells were still present in the dermal tissue of RB6-8C5-treated animals (Fig. 4A, right panel) and that these cells did not inhibit S. aureus to grow in large clusters. These findings show that the extravascular GFPhi cells in the dermis after S. aureus infection are indeed PMN and relate the established role of PMN in limiting S. aureus growth in vivo to specific patterns of PMN migration.
Soft tissue infections with S. aureus may lead to the development of an abscess, which is defined as the accumulation of pus inside a capsule. When we analysed the focus of S. aureus infection established by 106 cfu S. aureus after 24 h by 2-P microscopy, two well-delineated zones reflected the abscess development: the centre of the infection was characterized by a large accumulation of PMN, which was distinctly separated from a peripheral zone harbouring smaller PMN numbers (Fig. 4C, upper panel). Bacteria could be visualized in the centre of the developing abscess as well as in the periphery (Fig. 4C, lower panels). Time-lapse movies and subsequent track analysis revealed that the PMN inside the developing abscess exhibited lower speed (Fig. 4D, Supplementary Movie S6) in contrast to the peripheral cells, which migrated with higher speed (Fig. 4D, Supplementary Movie S6), intermediate between PMN moving to and among S. aureus 2 h after infection (Fig. 3G). Interestingly, compared with the 2 h time point, peripheral PMN around the abscess exhibited much lower directional movement towards the focus of infection, which was reflected by the almost equal distribution of the Ψ track angles (Fig. 4E).
Signalling pathways that lead to PMN recruitment
In order to investigate the signalling requirements that lead to PMN recruitment to the site of S. aureus infection in the dermis, we first treated mice with Pertussis Toxin (PTx) prior to infection. This abolishes Gαi-protein-coupled receptor (GPCR) signalling, which involves formyl peptide receptors, complement C3a/C5a receptors and chemokine receptors (Kaslow and Burns, 1992; Luther and Cyster, 2001). Mice treated with a single dose of PTx 1 day before S. aureus flank skin infection exhibited a significantly higher S. aureus burden after 5 days (Supplementary Fig. S4), which is consistent with insufficient PMN recruitment in the initial phase of the infection. Since we established that a large number of PMN were recruited within the first hours after infection, we visualized PMN migration to the site of infection after 6 h by 2-P microscopy. In contrast to PBS injected mice, PTx treated animals exhibited greatly reduced PMN infiltrations in the dermis (Fig. 5A and B), which was accompanied by the presence of large S. aureus clusters (Fig. 5A, right panel). We observed PMN undergoing rolling adhesion in blood vessels in the vicinity of the infection site. Surprisingly, some PMN were arrested on the vessel wall, which was not expected based on the classical multistep paradigm in which Gi-coupled GPCR play a critical role in transition from rolling adhesion to firm adhesion (Fig. 5C, Supplementary Movie S7). These data demonstrate that 2-P microscopy can also be used to analyse intermediate steps in PMN recruitment in response to S. aureus infection.
To be able to address the signalling requirements more specifically, we established a model in which PMN were isolated from mouse bone marrow and labelled in vitro before being transferred to a recipient mouse by intravenous injection. When we transferred labelled PMN from C57BL/6 wild-type mice into syngeneic recipients and then tracked their migration in the skin of an infected LysM-EGFP mouse, we found that the transferred cells were co-recruited with the endogenous PMN and thus appeared to be functional despite the ex vivo manipulation, because they exhibited similar cell speed, meandering index and directionality towards the focus of infection when compared with endogenous PMN (Supplementary Movie S8, and data not shown). Notably, the transferred PMN represented only a minor fraction of the total number of recruited PMN.
To elucidate whether GPCR are required on the PMN themselves for recruitment to the site of infection, we transferred labelled PMN that were treated with PTx in vitro and mixed them at a 1:1 ratio with untreated, but differently labelled PMN. Six hours after infection we used 2-P microscopy to investigate the recruitment of these cells to the infection site. Consistent with our previous result, we found that untreated PMN migrated towards the bacteria in the skin, but PTx treated PMN were almost completely absent from the focus of infection (Fig. 5D and E). This demonstrates that GPCR on the PMN are required for their extravasation and recruitment towards the site of infection.
It was previously shown in bone-marrow transfer experiments that IL-1R signalling on radio-resistant host cells is important for the recruitment of PMN 1 day after skin infection with S. aureus (Miller et al., 2006). To further investigate the influence of IL-1R signalling on early PMN migration on a single cell level, we isolated PMN from the bone marrow of C57BL/6 WT and IL-1R−/− mice, mixed them at a 1:1 ratio after labelling with different dyes, and then transferred them into WT or IL-1R−/− animals prior to an infection with S. aureus. After 6 h, the sites of infection were visualized by 2-P microscopy (Fig. 5F, Supplementary Movie S9). When we analysed the images for recruitment of PMN, we found that the percentage of WT and IL-1R−/− PMN at the site of infection was not significantly different in either WT or IL-1R−/− mice (Fig. 5G). But when we assessed total numbers of recruited cells, we found that far fewer PMN were located at the S. aureus inoculation site in IL-1R−/− mice compared with WT mice. Although the absolute number of recruited PMN showed a variation from experiment to experiment probably due to competition with endogenous unlabelled PMN, the relative fraction of recruited PMN in IL-1R−/− mice compared with WT controls was consistent at 14.7 ± 0.6% (n = 3) (Fig. 5H). This confirms that IL-1R signalling plays a dominant role in PMN recruitment in the early phase of dermal S. aureus infection, whereas the absence of IL-1R on PMN does not affect their capacity to migrate towards the infection focus.
Collectively, these data also demonstrate the suitability of our 2-P microscopy set-up to dissect multiple steps in the PMN recruitment after infection on a single cell level.
Using virulence factors to uncover PMN recruitment mechanisms
Finally, we set out to test whether our experimental system would also allow for visualization of the effect of S. aureus immune evasion factor Ecb on PMN migration. Ecb is a protein secreted by S. aureus that inhibits complement activation. Parallel systemic and local administration of the purified protein resulted in reduced PMN recruitment in an immune complex-mediated peritonitis model (Jongerius et al., 2007). We therefore probed the role of complement activation in PMN recruitment to sites of S. aureus inoculation with purified recombinant Ecb protein. We treated LysM-EGFP mice by injecting Ecb intravenously just prior to inoculation and intradermally together with S. aureus and compared PMN counts in treated and untreated mice by visualizing the site of infection using 2-P microscopy. We observed that Ecb reduces the number of PMN in the skin 6 h after infection (Fig. 6A and B). These results demonstrate that recombinant S. aureus virulence factor Ecb is able to inhibit PMN recruitment during skin infection. Thus, the complement cascade contributes to the PMN recruiting mechanisms in response to S. aureus in the skin.
Conventional light microscopy as well as confocal microscopy and other imaging techniques have been widely applied in various in vivo models to gain insight into host cell migration after infection as well as immune evasion and virulence mechanisms. However, histological examination of the infected tissue provides only a ‘snapshot’ of the dynamics of the immune response, whereas non-invasive luminescence imaging approaches lack the necessary resolution to describe host–pathogen interactions on a single cell level. Intravital 2-P microscopy is a technique to overcome these disadvantages and has been used successfully to study infections with diverse pathogens such as Toxoplasma gondii, Leishmania major, Listeria monocytogenes, lymphocytic choriomeningitis virus and uropathogenic Escherichia coli (Velazquez et al., 2007; Mansson et al., 2007a; Chtanova et al., 2008; Peters et al., 2008; Kim et al., 2009; Waite et al., 2011). The possibility to generate time-lapse movies has allowed for extensive analysis of host cell recruitment and migration during infection (Kreisel et al., 2010; Herz et al., 2012). Furthermore, this technique is capable of providing detailed information about the complex pathophysiology and local changes during infection including important parameters such as blood flow and PO2, which was shown in a kidney infection model using uropathogenic E. coli (Melican et al., 2008; Richter-Dahlfors et al., 2012).
Although a detailed protocol is available for imaging immune responses in the ear skin of mice (Li et al., 2012), we chose the flank skin to apply 2-P microscopy to investigate PMN migration during S. aureus infection, because this location is very established in infection models (Bunce et al., 1992) and allows for injection of higher inocula, which lead to abscess formation.
Various FP have previously been used to study different properties of S. aureus (Cheung et al., 1998; Malone et al., 2009). We achieved high expression of constitutive SarA-dependent as well as Agr-dependent fluorescent reporters by using different FP based on GFPmut2 (Cormack et al., 1996). Adaptation of the reporter genes to the codon usage of Gram-positive organisms proved to be crucial for obtaining sufficient fluorescence in our hands. gpmCherry, a codon-optimized red-shifted FP (Gauger et al., 2012), exhibited low fluorescence when expressed in S. aureus during infection in vivo and excited with the 2-P laser (data not shown). Therefore, to achieve spectral separation we used bacteria labelled with CER, which were well distinguishable from GFP-labelled cells in 2-P microscopy experiments (see below). The replicon cassette in our vector system (pT181 cop-wt repC) leads to copy numbers of 20–25 per cell, which enables promoter activity studies (Charpentier et al., 2004). One concern with these vector-based reporters is their loss in vivo in the absence of antibiotics, which provide the pressure to maintain the plasmid in vitro. Surprisingly, nearly all bacterial colonies recovered after 5 days from the skin lesions of mice without antibiotic treatment exhibited fluorescence in our experimental set-up. Also, the presence of the plasmid did not influence the virulence of the bacteria in the skin infection model. The versatility of this cassette-based shuttle vector system is underlined by the fact that we were able to introduce the plasmids into the S. aureus model strains RN6734 and Newman by phage transduction, thereby circumventing the use of adapted and less virulent laboratory strains [e.g. RN4220, which exhibits delayed activation of the Agr mediated QS (Traber and Novick, 2006)] for animal infections.
Even low numbers of S. aureus embedded in the extracellular matrix of the dermis were readily detectable by 2-P microscopy in a mouse skin flap with intact blood flow. By employing an Agr reporter strain, we found that Agr induction was detectable 8 h after infection, which is in line with previous reports that found an early peak of Agr activity using a bioluminescent reporter strain (Wright et al., 2005). The concept of QS as a mechanism to perceive bacterial density would predict preferential upregulation in the middle of the infection focus, where the concentration of AIP should be the highest. This was indeed observed when we analysed bacterial micro-colonies in vitro. In contrast and to our surprise, we found Agr positive bacterial cells in the border region of the S. aureus infection focus in vivo. One explanation for this could be provided by the concept of ‘diffusion sensing’, in which spatial confinement (represented by reduced AIP diffusibility) is sensed by the bacteria rather than cell density (Redfield, 2002). Indeed, Agr upregulation can be observed in single cells or small clusters after physical isolation by entrapment in nanostructured droplets (Carnes et al., 2010). Since bacterial cells in the border region of the infection are the first to encounter recruited PMN from the bloodstream, we speculate that Agr induction might be an effect that follows uptake of small bacterial clusters by these arriving leucocytes, comparable to what was reported after internalization by epithelial cells in vitro (Qazi et al., 2001). Nevertheless, we cannot rule out that uptake by resident macrophages leads to Agr upregulation in vivo, or that complex interstitial flow patterns near the site of infection result in a higher level of AIP at the periphery of the focus.
To visualize the immune response against S. aureus in the skin, we took advantage of LysM-EGFP mice, in which PMN are GFPhi, whereas macrophages and monocytes are GFPlo (Faust et al., 2000; Peters et al., 2008; Kreisel et al., 2010). We observed GFPlo cells distributed throughout the dermis of uninfected mice by intravital 2-P microscopy, which displayed low migration velocity and displacement, and which most likely represent sessile macrophages. This is consistent with behaviour of macrophages observed in other sites during 2-P microscopy studies (Lindquist et al., 2004; Schwickert et al., 2007). PMN were located in blood vessels in uninfected mice, but rapidly emigrated to the interstitial tissue after infection. Treatment with a monoclonal antibody (RB6-8C5) abolished PMN recruitment. Although RB6-8C5 has been reported to also deplete inflammatory monocytes (Daley et al., 2008), we still observed GFPlo cells in the dermis, which indicates no effect of the used dosage on sessile macrophages.
In a murine footpad infection model with L. monocytogenes the PMN migration speed after extravasation was reported as 7.5 μm min−1 (Zinselmeyer et al., 2008). This is similar to the velocity of PMN moving towards Listeria foci in the spleen (Waite et al., 2011), but slower than the observed values in our model (Fig. 3F). This can be explained by fact that different pathogens have different capabilities to elicit host signals, which lead to PMN recruitment. Listeria reside inside phagocytic cells during an infection and therefore may be much less likely to activate complement, which is important for extracellular S. aureus mediated PMN recruitment. Additionally, S. aureus was shown to produce phenol-soluble modulins that can enhance PMN recruitment at low concentrations (Kretschmer et al., 2010).
In a study using a kidney infection model with uropathogenic E. coli Mansson et al. reported that PMN constituted only approximately 20–40% of nucleated cells recruited to the infection site at the early hours of infection (Mansson et al., 2007a,b). Since non-transgenically labelled cells are not visible in our infection system, it remains unclear, if and what numbers of other cells get recruited to the site of S. aureus infection in the skin. Nevertheless, Peters et al. showed that the vast majority of dermal GFPhi cells after infection with L. major are Gr-1+ and CD11b+, which is consistent with the PMN phenotype, and that these cells are associated with the pathogen (Peters et al., 2008).
Navigation of leucocytes to sites of infection has been proposed to use a series of guidance cues including bacterial products and mediators produced by the host in response to infection, which result in correct localization of the cells to the site of the lesion (Foxman et al., 1997; Heit et al., 2002; McDonald et al., 2010; Ng et al., 2011). This dominance of chemotactic gradients may provide an explanation for the observed ‘ignorance’ of PMN towards bacterial clusters. On the other hand, PMN that were closely associated with S. aureus or had taken up bacterial cells changed their migratory characteristics by decreasing cell speed and displacement. Similar findings have been reported for PMN encountering L. major (Peters et al., 2008), suggesting that this might be a mechanism to localize and confine PMN to the area of infection. To test the involvement of G-protein-coupled receptors in our system, which includes chemokine receptors (such as CXCR2) as well as formylated peptide receptors, we treated mice with PTx prior to infection, which led to an increase in S. aureus cfu counts 5 days after infection. This finding is congruent with reports that show increased S. aureus loads in a brain abscess model using CXCR2 knockout mice (Kielian et al., 2001) and skin lesions in PMN depleted mice (Molne et al., 2000). PMN did not leave the bloodstream and did not migrate towards the focus of infection after PTx treatment.
PTx has broad effects on cell migration, since it affects migrating cells and activated endothelium, which can result in considerable toxicity in vivo, if the dose is not chosen carefully. To avoid these effects, we developed an intravenous adoptive transfer model of in vitro labelled bone-marrow PMN. The transferred PMN exhibited migration characteristics that were comparable to endogenous cells and which could be blocked by ex vivo treatment with PTx prior to transfer. This strategy circumvents potential artefacts that might occur if the toxin-treated cells are injected directly into the skin (Ng et al., 2011). Additionally, this strategy allows for simultaneous transfer of differently labelled PMN populations. We exploited this approach to investigate the influence of IL-1R on the localization of PMN after S. aureus infection. When we co-transferred PMN from WT and IL-1R−/− mice, we observed that their recruitment depended on the mouse background, but not on the presence of IL-1R on the cells themselves. This is in line with previous reports, which show that skin lesions of IL-1R−/− mice showed decreased numbers of PMN 1 day after infection and that IL-1R on resident skin cells rather than on myeloid cells is required for sufficient cell recruitment (Miller et al., 2006).
Overall, the in vivo experiments show that PMN GPCRs, host IL-1R signalling and complement activation all were essential for significant PMN recruitment at 2–6 h. The early role of GPCR blockade was not surprising as these receptors play a key role in the multistep extravasation paradigm by activating integrins (Lawrence and Springer, 1991; von Andrian et al., 1991). Therefore, we were somewhat surprised to observe some arrest of pertussis toxin treated PMN in blood vessels, but most PMN were limited to rolling adhesion. It was surprising that IL-1R deficiency so completely suppressed leucocyte infiltration. S. aureus possesses mechanisms to suppress inflammasome activation by phagocytes (Shimada et al., 2010) suggesting that host reliance on IL-1R signalling may allow the bacterium to control the tempo of the early neutrophil response. The role of IL-1 is likely to upregulate adhesion molecules on endothelial cells to facilitate extravasation (Pober et al., 1986), to upregulate tissue adhesion molecules to support interstitial migration (Dustin et al., 1986), and to initiate fibroblast activation to eventually form the abscess wall (Kielian et al., 2004). The reliance on complement may relate to guidance cues such as C5a, which are attenuated by multiple mechanisms in S. aureus infection (Jongerius et al., 2010; Laarman et al., 2011). Recruitment of PMN by S. aureus may be sufficiently attenuated that impairing any one mechanism abrogates the response. Alternatively, robustly activated steps mediated by IL-1 on endothelial cells to upregulate adhesion and GPCRs on leucocytes responding to anaphylatoxins may be sufficiently distinct that eliminating either of these steps eliminates extravasation.
Finally, we also investigated the influence of a S. aureus immune evasion protein on PMN recruitment in our model. Ecb is an inhibitor of the C3 convertase and strongly inhibits the generation of C5a, which has chemotactic and pro-inflammatory properties. In contrast to other immune evasion molecules like CHIPS and SCIN, which are human specific, purified Ecb is able to block complement activation in the mouse (Jongerius et al., 2007). In 2-P microscopy experiments, we could demonstrate that Ecb is able to block PMN recruitment to the site of infection. This became apparent when the protein was injected intradermally and intravenously. Since the concentrations of Ecb secreted by S. aureus in vivo have not been described, we adopted the dosage and mode of administration (local and systemic) from a previous report (Jongerius et al., 2007). Our results are in line with a recent study, which compared the virulence of S. aureus Newman and an isogenic mutant that lacks Ecb and the closely related protein Efb (Jongerius et al., 2012). It was shown that the ΔEcbΔEfb mutant exhibited decreased virulence in different mouse infection models (pneumonia, sepsis and kidney abscess formation) and that Ecb and Efb block the influx of PMN in the lung during S. aureus pneumonia. Thus, the complement system plays an important role in PMN recruitment after S. aureus infection in the skin and in other organs.
Taken together, we describe here an experimental system that allows for visualization of the spatiotemporal orchestration of the cellular immune response to S. aureus inside an infected animal at single cell resolution. Studying this pathogen in its physiological environment during an infection is a new approach to gain insight into host–pathogen interactions and bacterial virulence mechanisms, which potentially lead to the development of new therapeutic strategies against this clinically important bacterium.
This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the Public Health Service (National Institutes of Health). The New York University School of Medicine Institutional Animal Care and Use Committee (IACUC) approved this protocol. All surgery was performed under Ketamine, Xyalazine and Acepromazine anaesthesia, and all efforts were made to minimize suffering.
Generation of fluorescent reporters
pJL-sar-GFP was derived from pCN54 by replacing the Pcad promoter with a SarA P1 promoter fragment, which was generated by PCR amplification from chromosomal DNA of RN6734 using the primers 5′-GTTGTTGCATGCCTGATATTTTTGACTAAACCAAATGC-3′ and 5′-TCGATAGAATTCGATGCATCTTGCTCGATACATTTG-3′, and cloned using the SphI/EcoRI restriction sites (underlined). The Pcad promoter in pCN54 was also replaced with a DNA fragment from pRN7105 containing the agr P3 promoter to generate pJL-agr-GFP.
The genes for gpCerulean and gpVenus were commercially synthesized (DNA2.0) using the gfpmut2 gene as a template and by introducing the described mutations (Supplementary Fig. S1). The genes were adjusted in a way that they could be amplified using the primers 5′-GGATCCGAATTCTTAGGAGGATGATTATTTATGAGTAAAGGAGAAGAACTTTTCACTG-3′ and 5′-GGATCCGGCGCGCCTTACTTGTACAGCTCGTCCATGCCGC-3′. The restriction sites of EcoRI and AscI (underlined) were used to replace gfpmut2 in pJL-sar-GFP and pJL-agr-GFP with gpcer and gpven, respectively, to yield the reporter plasmids listed in Table 2.
The reporter plasmids were purified from E. coli DH5α and subsequently introduced into the restriction-defective S. aureus strain RN4220, by electroporation. From here, the plasmids were transferred to the indicated S. aureus strains by phage transduction (Table 1).
Confocal microscopy of S. aureus cells
Staphylococcus aureus reporter strains were imaged after fixation of 104 bacterial cells per ml to coverslips or to the bottom of chamber slides using 2% paraformaldehyde. Image acquisition was performed in the confocal mode of an inverted Zeiss LSM 710 NLO microscope equipped with a spectral detector and employing a Zeiss Plan-Apochromat 63×/1.40 oil objective. The following excitation wavelengths, laser sources and detection spectra were used: gpCerulean: 405 nm/diode laser/454–581 nm; GFPmut2: 488 nm/argon laser/493–598 nm; gpVenus: 514 nm/argon laser/519–612 nm.
FACS was performed on a FACSCalibur (BD Biosciences) equipped with a 488 nm laser for excitation of GFPmut2. FlowJo software (Tree Star) was used for data analysis.
C57BL/6NTac (B6) wild-type mice were purchased from Taconic Farms. LysM-EGFP mice (Faust et al., 2000) on a C57BL/6 background were housed in our animal facility. For experiments these mice were bred to heterozygosity (LysMEGFP/+) and used at 6–12 weeks of age. B6.129S7-Il1r1tm1Imx/J (IL-1R−/−) mice and wild-type controls were purchased from The Jackson Laboratory.
Mouse skin infection model
Five millilitres of Tryptic Soy Broth (TSB) was inoculated from frozen bacterial stocks and cultured for 16 h (37°C, shaking at 225 r.p.m.). Erythromycin (10 μg ml−1) was added, if necessary to maintain the plasmid. A fresh culture was then inoculated 1:1000 (without Erythromycin) and bacteria were grown to early/mid-logarithmic phase (c. 3 h) before being washed two to three times with sterile PBS. After resuspension in PBS, S. aureus density was calculated from OD600 measurements, and the inoculum was adjusted to the desired infection dose with sterile PBS. Mice were anaesthetized with premixed Ketamin (50 mg kg−1)/Xylazine (10 mg kg−1)/Acepromazine (1.6 mg kg−1) and infected intradermally on their right shaved flank skin using a 28G 3/10cc Insulin syringe (BD).
Mouse surgery and intravital microcopy
The surgical preparation of the mouse adopted the technique described by von Andrian for imaging the inguinal lymph node (von Andrian, 1996). The deeply anaesthetized mouse was placed on a heated surgical stage and a midline incision was performed. While keeping the surgical site moisturized with sterile PBS, this incision was then extended laterally, yielding a crescent-shaped skin flap (Fig. 2A), from which the subcutaneous tissue was carefully removed with a moistened synthetic-tipped applicator. The focus of infection or the region of interest was then covered with a 45 × 50 mm microscope cover glass (0.16–0.19 mm thickness, Fisher Scientific), and the mouse was placed on a microscope stage, which is surrounded by an incubation chamber, and maintained at 37°C. Oxygen (1–2 l min−1) was delivered via a face mask. Qtracker 655 non-targeted quantum dots (Invitrogen) were diluted 1:20 in sterile PBS and injected intravenously to visualize blood vessels in some experiments. Laser excitation was performed using a tunable Mai Tai laser (Spectra-Physics) at a wavelength of 910 nm on an inverted microscope system (Zeiss LSM 710 NLO). Serial non-descanned detectors (NDD) and appropriate filters/mirrors were used to detect signals in the following ranges: 420–465 nm (second harmonic generation), 465–500 nm (gpCerulean), 500–555 nm (GFPmut2, gpVenus), 575–645 nm (SNARF-1) and 660–760 nm (Qtracker 655). ZEN 2009 software (Zeiss) was used for operating the microscope, image acquisition and data handling.
For cell tracking purposes the site of the S. aureus inoculum was located in the dermis under a layer of unperturbed dermal connective tissue. A Zeiss LD LCI Plan-Apochromat 25×/0.8 multi-immersion objective was used to acquire xy-scans, which were tiled (usually 9–16 images) to cover large areas of the site of S. aureus deposition. Images were acquired every 15 s for a total duration of 15–20 min. One to three movies were recorded per animal with special caution to keep total imaging periods as short as possible in order to avoid artefacts that resulted from surgical manipulation.
Image processing and statistics
Image analysis was performed with Imaris (Bitplane Scientific Software). QuickTime PRO (Apple) was used to add labels to movie files. Prism 5 (GraphPad Software) and Excel (Microsoft) were employed for statistical data analysis. The two-tailed Student's t-test was used to compare sample groups with Gaussian distribution, whereas the Mann–Whitney U-test was used for samples with non-Gaussian distribution.
Transfer of labelled PMN
Bone marrow was harvested by flushing the femurs and tibias of the mice with cold HBSS (without calcium and magnesium)/0.1% FCS. After passing the cells through a 40 μm cell strainer (BD) to remove larger particles and lysis of red blood cells with ACK Lysing Buffer (Lonza), the cells were washed with HBSS (without FCS) and resuspended in 1 ml of HBSS. A discontinuous 81% (v/v)/62% gradient was prepared from sterile, low endotoxin Percoll PLUS (GE Healthcare) [100% = 90% Percoll PLUS + 10% 10× HBSS (without calcium and magnesium)] and HBSS, and the bone-marrow cells were carefully applied on top of the gradient. Centrifugation was then carried out at room temperature for 20 min at 1400 g in a swinging bucket rotor. The cells at the interface of the 81% and 62% Percoll gradient, which represent the PMN, were carefully removed and washed with HBSS at room temperature. Labelling of the cells was performed with CFSE or SNARF-1 (Invitrogen) at a final concentration of 5 μM. After washing, PMN were transferred to the recipient mouse by intravenous injection.
Ecb was produced recombinantly in E. coli as described previously (Jongerius et al., 2007).
Mouse Genome (MGD) accession ID numbers
LysM (Lyz2) MGI:96897, LysM-eGFP (Lyz2tm1.1Graf) MGI:2654931, Gr-1 (Ly6g) MGI:109440. Mouse Genomics Informatics (MGI), The Jackson Laboratory, Bar Harbour, ME. URL: http://www.informatics.jax.org/.
We thank Victor J. Torres for providing the S. aureus Newman strains. We are very grateful to the Novick and Dustin labs for advice on S. aureus genetics and two-photon microscopy respectively.