Fluorescence Lifetime Imaging Unravels C. trachomatis Metabolism and Its Crosstalk with the Host Cell

Chlamydia trachomatis is an obligate intracellular bacterium that alternates between two metabolically different developmental forms. We performed fluorescence lifetime imaging (FLIM) of the metabolic coenzymes, reduced nicotinamide adenine dinucleotides [NAD(P)H], by two-photon microscopy for separate analysis of host and pathogen metabolism during intracellular chlamydial infections. NAD(P)H autofluorescence was detected inside the chlamydial inclusion and showed enhanced signal intensity on the inclusion membrane as demonstrated by the co-localization with the 14-3-3β host cell protein. An increase of the fluorescence lifetime of protein-bound NAD(P)H [τ2-NAD(P)H] inside the chlamydial inclusion strongly correlated with enhanced metabolic activity of chlamydial reticulate bodies during the mid-phase of infection. Inhibition of host cell metabolism that resulted in aberrant intracellular chlamydial inclusion morphology completely abrogated the τ2-NAD(P)H increase inside the chlamydial inclusion. τ2-NAD(P)H also decreased inside chlamydial inclusions when the cells were treated with IFNγ reflecting the reduced metabolism of persistent chlamydiae. Furthermore, a significant increase in τ2-NAD(P)H and a decrease in the relative amount of free NAD(P)H inside the host cell nucleus indicated cellular starvation during intracellular chlamydial infection. Using FLIM analysis by two-photon microscopy we could visualize for the first time metabolic pathogen-host interactions during intracellular Chlamydia trachomatis infections with high spatial and temporal resolution in living cells. Our findings suggest that intracellular chlamydial metabolism is directly linked to cellular NAD(P)H signaling pathways that are involved in host cell survival and longevity.


Introduction
The obligate intracellular bacterium Chlamydia trachomatis (C. trachomatis) has two metabolically different developmental forms, which ensure its infectivity and replication. The infectious form, the elementary body, enters the host cell and differentiates into a metabolically active form, the reticulate body. Chlamydial reticulate bodies grow within the host cell in an intracellular membrane-bound compartment called the chlamydial inclusion. Within 24 hours post infection (hpi), numbers and size of reticulate bodies are maximized and chlamydiae start to re-differentiate to infectious elementary bodies. Cell lysis and release of the elementary bodies occur at around 48 hpi [1].
C. trachomatis is a sensitive marker organism for host cell metabolic changes because it strongly depends on ATP and metabolites generated by the host. Although an ADP/ATP transporter has been found [2,3], the genome sequence also unraveled the existence of several glucose metabolizing enzymes [4]. Since then, it has been speculated that C. trachomatis not only uses host cell ATP, but also is capable to produce its own energy [5]. However, the metabolic pathways of C. trachomatis are often truncated. Thus, C. trachomatis might directly import the substrates required to compensate for the incomplete metabolic pathways [4]. Using microarray technology, it was shown that the ADP/ ATP translocase and the ATP requiring oligopeptide transporters are expressed as immediate early genes. Furthermore, some metabolic enzymes such as the malate dehydrogenase (which requires nicotinamide-adenine dinucleotide (NAD) as cofactor) are also expressed in the early phase of infection [6]. Interestingly, no pathways for the biosynthesis of NAD and no NAD kinase for the synthesis of phosphorylated NAD have been found in the chlamydial genome. Although it seems obvious that a system to import NAD(P) from the host cell must exist similarly to that of environmental chlamydiae, no NAD(P) transporter has yet been identified in C. trachomatis according to sequence homology searches [7]. It is therefore reasonable to assume that intracellular chlamydial development strongly depends on host cell NAD availability.
The lack of suitable methods to investigate chlamydial metabolism separately from host cell metabolism has hindered scientific progress in studying host and pathogen metabolic interactions. Current knowledge on chlamydial metabolism is restricted to micro-array and RT-PCR analyses about the expression of metabolic genes during different intracellular developmental stages, the characterization of recombinant chlamydial metabolic enzymes, and the biochemical analysis of infected cells [1,5,6,[8][9][10]. Recently, it was shown by Raman microspectroscopy that amino acid uptake and protein synthesis is preserved in C. trachomatis after extracellular incubation [11]. Increases in glucose consumption and ATP levels were observed in cells infected with chlamydiae, however these studies could not distinguish between host cell and bacterial metabolism [12,13].
Fluorescence lifetime imaging (FLIM) of NAD(P)H using twophoton laser scanning microscopy has been used to study cellular metabolism in living cells by utilizing the autofluorescence properties of the reduced form of the metabolic cofactor, NAD [NAD(H)] and its phosphorylated form, NADP [NAD(P)H] [14][15][16][17]. At 730-750 nm excitation, the cellular autofluorescence as measured by two-photon microscopy is dominated by NAD(P)H [14,16,18,19]. The fluorescence lifetime is the time that a molecule spends to return to its ground state from its excited state. This exponential decay rate can be measured by time correlated single photon counting (TCSPC). The fluorescence lifetime of a molecule is thereby independent of its total concentration but strongly depends on its respective protein binding within a given microenvironment [20]. Thus, NAD(P)H functions are determined by binding to different cellular proteins. NAD(H) is a cofactor for catabolic reactions in the cytosol and the mitochondria while the phosphorylated NADP(H) is a cofactor in anabolic reactions and plays an important role in the cellular anti-oxidative defense system [21,22]. NADH is generated within the glycolytic pathway by glyceraldehyde-phosphate-dehydrogenase (GAPDH) in the host cell cytosol while NADPH is produced by the enzymes of the pentose-phosphate pathway. In the mitochondria, NAD(P)H cannot pass through the inner mitochondrial membrane and only electrons from NAD(P)H are carried across by the malate-aspartate shuttle for oxidative phosphorylation. By contrast, free NAD(P)H can cross the nuclear envelope by diffusion through the nuclear pores; therefore, changes of cytosolic levels of free NAD(P)H are represented in the nucleus. An important binding protein of NADH in the nucleus is the transcriptional corepressor, C-terminal binding protein (CTBP) [23]. Consequently, nuclear NADH functions to control gene transcription by regulating CTBP. [24]. Importantly, fluorescence lifetimes of NAD(P)H significantly differ between cytosol, mitochondria and nucleus in one single cell [19]. Differences of the NAD(P)H fluorescence lifetimes that are based on the different proteinbindings of NAD(P)H reflect the diverse physiological functions in these cellular compartments. Thus, monitoring of NAD(P)H fluorescence lifetimes allow for a compartmentalized analysis of cellular metabolic changes over time. In addition, calculating the bi-exponential fluorescence decay enables not only the monitoring of different components of the fluorescence lifetimes but also the ratio of free to protein-bound NAD(P)H [14,15]. The fluorescence lifetime of protein-bound NAD(P)H [t 2 -NAD(P)H] is longer than that of free NAD(P)H [t 1 -NAD(P)H] [20]. As free NAD(P)H is readily diffusible, regulation of its binding proteins is correlated to the relative amount of free NAD(P)H [24]. The value of t 2 -NAD(P)H and the ratio of free to protein-bound NAD(P)H are established markers to sensor metabolic activity of eukaryotic cells and have been used to monitor changes in host cell glycolysis or oxidative phosphorylation [15,19,25]. Inhibition of glycolysis results in an increase of t 2 -NAD(P)H while the inhibition of oxidative phosphorylation decreases t 2 -NAD(P)H. The shift from oxidative phosphorylation to glycolysis has been made responsible for the decrease of t 2 -NAD(P)H during cancer development [15]. Furthermore, FLIM is the only currently available method to determine compartmentalized redox state of NAD(P)H in living cells.
We applied two-photon FLIM of NAD(P)H to discriminate and thoroughly characterize chlamydial and host cell metabolism. By measuring t 2 -NAD(P)H values and determining the ratio of free to protein-bound NAD(P)H in C.trachomatis-infected cells, we observed significant changes both in host cell and chlamydial metabolism during the intracellular developmental cycle.

Compartmentalization of NAD(P)H in C. trachomatisinfected cells
In non-infected HEp-2 cells, the majority of cellular autofluorescence at 730 nm excitation originated from the mitochondria. The cytosol and nuclei showed reduced fluorescence, representing the compartmentalized distribution of NAD(P)H in the cells ( Figure 1A). To confirm that NAD(P)H is the major contributor to cellular autofluorescence at 730 nm excitation, the autofluorescence spectra were measured in four different emission channels. The highest emission intensity of autofluorescence was between 380 nm and 500 nm in all cellular compartments, corresponding to the peak of NAD(P)H fluorescence emission (Figures S1A and S1B). In C. trachomatis-infected cells a new compartment is formed, the chlamydial inclusion which also showed a strong autofluorescence signal ( Figure 1A). The spectra of the chlamydial fluorescence did not show any shift compared to the autofluorescence of the other cellular compartments, indicating that the fluorescence inside the chlamydial inclusion also originated primarily from NAD(P)H (Figures S1A and S1B). The chlamydial inclusions could be clearly distinguished from the host cells due to the different fluorescence intensity of NAD(P)H inside the chlamydial inclusion ( Figures S2A and S2B). The selected region of interest (ROI) was then used to determine fluorescence lifetimes of NAD(P)H inside the chlamydial inclusion and other cellular compartments. We observed a broad distribution of t 2 -NAD(P)H

Author Summary
Separate analysis of host and pathogen metabolic changes in intracellular C. trachomatis infections is arduous and has not been comprehensively realized so far. A more detailed understanding about the metabolic activity and needs of C. trachomatis and its specific interactions with the host cell would be the basis for the development of novel treatment strategies. We therefore applied fluorescence lifetime imaging (FLIM) of the metabolic coenzymes NAD(P)H using two-photon microscopy to directly visualize metabolic changes of host cells and pathogens in living cells. NAD(P)H fluorescence was detected both on the chlamydial inclusion membrane and inside the inclusion. Interestingly, changes in chlamydial growth and progeny induced by glucose starvation and IFNc treatment were directly linked to significant changes of the NAD(P)H fluorescence lifetimes inside the inclusions. Furthermore, measurement of the NAD(P)H fluorescence lifetime in the host cell nucleus revealed that infected cells were programmed for starvation during the metabolically active phase of intracellular chlamydial growth. Our findings highlight for the first time a direct interaction between host and pathogen metabolism in intracellular bacterial infections that exceeds sole competition for nutrients. In conclusion, fluorescence lifetime imaging of NAD(P)H by two-photon microscopy enables real-time analysis of metabolic host-pathogen interactions in intracellular infections with high spatial and temporal resolution.
that indicates heterogeneity of NAD(P)H protein binding in each cellular compartment ( Figure S2C). We could confirm that the mean values of t 2 -NAD(P)H in the cytosol, nucleus, mitochondria and chlamydial inclusion were significantly different (overall significance p,0.0001). The mean lifetime values are listed in Table S1. In addition, t 2 -NAD(P)H was significantly increased inside the chlamydial inclusion at 24 hpi compared to all the other cellular compartments ( Figure 1B).
Non-mitochondrial NAD(P)H fluorescence surrounding the chlamydial inclusion NAD(P)H fluorescence showed an increased intensity surrounding the chlamydial inclusion. To test whether the high intensity signal originated from the host cell mitochondria, we co-incubated the cells with a mitochondria marker, tetramethylrhodamineethyl-ester (TMRE). TMRE showed complete co-localization with the high intensity NAD(P)H fluorescence in non-infected HEp-2 cells (Figures 2A and S3A). By contrast, in C. trachomatis-infected cells, high intensity signals only partially co-localized with TMRE. Most of the high intensity fluorescence which was found around the chlamydial inclusion did not co-localize with host cell mitochondria that were labeled with TMRE (Figures 2A and  S3B). This high intensity signal around the inclusion also did not show co-localization with the GFP-tagged lysosomal and golgi markers that were overexpressed in C. trachomatis-infected cells ( Figure 2B). Overexpression of the GFP-tagged 14-3-3b protein that was previously shown to be localized on the inclusion membrane [26] co-localized with NAD(P)H indicating that the high intensity NAD(P)H fluorescence derives from the inclusion membrane in C. trachomatis-infected cells ( Figure 2C).

Fluorescence lifetime changes of NAD(P)H in the chlamydial inclusions during the developmental cycle
To verify our hypothesis that changes in the values of t 2 -NAD(P)H inside the chlamydial inclusion are an indicator of chlamydial metabolism, we followed the alterations of t 2 -NAD(P)H during the developmental cycle of C. trachomatis ( Figure 3A). Increased metabolism of reticulate bodies during the mid-phase of intracellular growth was directly correlated with an increase of t 2 -NAD(P)H between 12 and 4 hpi. By contrast, t 2 -NAD(P)H did not change further during the late phase of the infection (48 hpi) when infectious elementary bodies were formed ( Figure 3B). However, the distribution of t 2 -NAD(P)H was more heterogeneous at 48 hpi ( Figure S4A) whereas the ratio of free to protein-bound NAD(P)H (a 1 /a 2 ratio) inside the chlamydial inclusion significantly decreased between 12 and 24 hpi ( Figure  S4B).The distribution and intensity of fluorescence in the chlamydial inclusions also changed during the developmental cycle. While highest fluorescence in the mid-phase of the chlamydial development was observed at the inner border of the inclusion, the average fluorescence intensity increased and was homogeneously distributed within the inclusion within 24 hpi. During late infection, large areas inside the inclusion showed no NAD(P)H fluorescence, indicating replacement of metabolically active reticulate bodies by infectious but metabolically inert elementary bodies ( Figures S5A, S5B and S5C). An increase of t 2 -NAD(P)H was also observed in inclusions from early C. pneumoniae infected cells, followed by a decline in t 2 -NAD(P)H during later time points of the infection ( Figures S6A and S6B).

Inhibition of host cell metabolism directly interferes with the development and metabolism of C. trachomatis
To further prove that t 2 -NAD(P)H characterizes the metabolic activity of C. trachomatis, we monitored t 2 -NAD(P)H changes inside the chlamydial inclusion during inhibition of host cell metabolism. We used glucose starvation to inhibit host cell glycolysis and antimycin A to inhibit the mitochondrial complex III and subsequent oxidative phosphorylation. Both inhibitors decreased host cell ATP levels ( Figure S7A). The intracellular development of C. trachomatis was monitored by immunofluorescence staining and electron microscopic (EM) imaging of chlamydial inclusions ( Figure 4A). Sizes of chlamydial inclusions were quantitatively analyzed at 24 hpi and the numbers of infectious C. trachomatis progeny were measured by recovery assays (Figures 4B and 4C). Both glucose starvation and treatment with antimycin A resulted in smaller and aberrant chlamydial inclusions when compared to untreated cells. Quantification of the fluorescence lifetimes showed a significant decrease of t 2 -NAD(P)H inside the C. trachomatis inclusions when the host cells were treated with the metabolic inhibitors, as an indicator for decreased metabolism of C. trachomatis ( Figures 4D and S7B). The observed decrease in the t 2 -NAD(P)H values was directly correlated to the decreased recovery rate of infectious C. trachomatis.

Decreased t 2 -NAD(P)H and relative amount of proteinbound NAD(P)H in persistent chlamydial inclusions
Interferon-c (IFNc) is known to induce C. trachomatis persistence that is defined as a viable but non-cultivable developmental stage. Persistent C. trachomatis are characterized by enlarged reticulate bodies in morphological aberrant inclusions [27]. To determine the metabolism of reticulate bodies in persistence, we measured the    Table S2. Images show enlargement of representative ROIs used for analysis. ROIs were selected by signal intensities. Low intensity areas were selected as cytoplasma and nucleus, high intensity areas as mitochondria. For the analysis of the inclusions, ROIs were selected inside the inclusion (see Figure S2A and S2B). doi:10.1371/journal.ppat.1002108.g001 stages of the infection, we measured the ratio of free to proteinbound NAD(P)H (a 1 /a 2 ) and the values of t 2 -NAD(P)H in the nucleus. The ratio of a 1 /a 2 was decreased and t 2 -NAD(P)H was increased in the nucleus of C. trachomatis-infected cells at 24 hpi ( Figure 6A-C). Similar effects were observed when host cell glycolysis was inhibited by 2-fluoro-deoxy-glucose (2FDG) or by glucose deprivation (data not shown) that mimic cellular starvation. The results indicate decreased levels and altered protein binding of NAD(P)H in the nucleus as a consequence of metabolic changes in the host cell cytosol.

Discussion
C. trachomatis was speculated to be an energy parasite that totally depends on the supply of energy and metabolic co-enzymes from the nutrient-rich cytoplasm of an infected host cell. In recent years, when genome data became available and with the use of functional enzymatic competition assays, it became obvious that chlamydiae contain the capacity to produce its own energy and reducing power [5]. However, separate analysis of host and pathogen metabolism is still challenging and requires novel experimental and technical set-ups that facilitate dynamic monitoring of metabolic changes inside the intracellular chlamydial inclusion separately from the host cell in living cells. Imaging of NAD(P)H by its autofluorescence is an ideal tool to study metabolism in C. trachomatis-infected cells as it permits the spatial separation of metabolic changes in different cellular compartments [19]. By use of the two-photon microscopy technique lateral resolutions of below 1 mm are achieved with a maximum reduction in photo-damage over time. Changes in the NAD(P)H fluorescence lifetimes thereby reflect different protein binding characteristics of NAD(P)H and allow to calculate the relative amount of free NAD(P)H in ratio to protein-bound NAD(P)H [14].
First we studied the compartmentalized distribution of NAD(P)H fluorescence in C. trachomatis-infected cells. This uncovered a high intensity NAD(P)H fluorescence around the chlamydial inclusion which was not of mitochondrial origin.
Previous findings that the C. trachomatis-inclusion does not colocalize with the mitochondria of the host cells were confirmed [28]. Unlike the nuclear pore, the chlamydial inclusion is passively impermeable to small molecules rendering sole diffusion of NAD(P)H across the inclusion membrane unlikely [29]. The colocalization of the non-mitochondrial NAD(P)H signal with the inclusion marker 14-3-3b host cell protein indicates that it might represent a potential NAD(P)H transporter or the active site where NAD(P)H is synthesized for chlamydial use. The existence of NAD(P) transporter in environmental chlamydiae [7] suggests that human pathogenic chlamydiae also harbour a transporter rather than a biosynthetic pathway because of its smaller genome size.
Our data on the distribution of NAD(P)H fluorescence within the chlamydial inclusion further supports the model from Wilson et al. that reticulate bodies first occupy the juxta-membrane space during the formation of the chlamydial inclusion and later move to the centre for differentiation [30]. In the centre of the inclusion, limited supply of essential metabolites may ultimately trigger their re-differentiation to infectious elementary bodies [31].
The significantly increased t 2 -NAD(P)H values inside the chlamydial inclusion define a novel, metabolically distinct cellular compartment within infected cells. Comparison of the classical NAD(P) binding domains in humans and C. trachomatis using the SUPERFAMILY database search (http://supfam.cs.bris.ac.uk/ SUPERFAMILY_1.73/) [32] indicated that the Rossmann-fold [33] is present in 318 proteins in humans but only in 14 proteins of C. trachomatis (L2b/UCH-1/proctitis). Our measurements of t 2 -NAD(P)H values during the developmental cycle indicate that NAD(P)H within C. trachomatis inclusion binds to different proteins as its metabolism changes. In addition, our results indicate that t 2 -NAD(P)H strongly correlates with the metabolic activity of chlamydial reticulate bodies but not with the inclusion size. The increase in t 2 -NAD(P)H in the mid-phase of infection was accompanied by a decrease in the relative amount of free NAD(P)H, which might indicate the presence of an oxidative type of energy metabolism within the chlamydial inclusion [4]. Similar changes of t 2 -NAD(P)H were measured in C. pneumoniae inclusions. The longer developmental cycle of the bacteria allowed also monitoring the decrease in metabolism towards the end of the infection cycle. However the absolute changes that we could measure were also less pronounced.
IFNc induces persistence of C. trachomatis [34]. The molecular mechanism involves tryptophan depletion through the induction of the indoleamine 2,3-dioxygenase (IDO) by IFNc [35]. By measuring t 2 -NADP(H) and the ratio of a 1 /a 2 in the persistent chlamydial inclusions we could prove that the metabolism of the chlamydial reticulate bodies are largely reduced. In addition, we could visualize the enlarged reticulate bodies inside the inclusions.
Reduction of host cell ATP synthesis by inhibition of glycolysis or oxidative phosphorylation induced aberrant chlamydial inclusions and decreased t 2 -NAD(P)H values in the chlamydial inclusions. Therefore, t 2 -NAD(P)H values predicted metabolic activity of growing and differentiating reticulate bodies that resulted in infectious chlamydial progeny.
Looking more in detail on the ratio of free to protein-bound NAD(P)H in the host cell nucleus, we could also show that intracellular chlamydial infection programmed the cell for starvation. Free NAD(P)H, which is approximately 80% of the total amount of NAD(P)H, represents the freely diffusing NAD(P)H molecules as described by measuring fluorescence lifetime of NADH in solution [36]. As there is no barrier for the diffusion of free NAD(P)H between the cytosolic and nuclear compartments, nuclear NAD(P)H concentrations directly reflect the free cytosolic NAD(P)H concentrations [24]. The t 2 -NAD(P)H  Table S3A. doi:10.1371/journal.ppat.1002108.g003 changes in the nuclei of C. trachomatis-infected cells were consistent with those found in 2FDG treated cells in which cellular glycolytic flux is inhibited and cellular starvation is induced within 24 hours [19,24]. It is well accepted that calorie restriction and cellular starvation go along with the induction of gene expression patterns that allow energy conservation and directly regulate the lifespan of cells [37]. The deacetylase sirtuin1 (SIRT1) is a master regulator of longevity in the host cell nucleus and is directly regulated by its substrate NAD [38] while its transcription is regulated by the NADH-binding transcriptional corepressor, carboxy-terminal binding protein 1 (CTBP) [23,24,39].
A limitation of NADP(H) autofluorescence imaging is that conversion of NAD(P)H fluorescence intensity values to absolute concentration values is not straightforward because the different quantum yields of free and protein-bound NAD(P)H have to be calculated [40,41]. In addition, the fluorescence decay parameters of the phosphorylated and non-phosphorylated forms of reduced NAD are the same and are indistinguishable. Although estimations of cellular concentrations suggest that a substantial part of the cellular fluorescence originates from NADPH rather than from NADH [21], we and others have demonstrated significant NAD(P)H fluorescence lifetime changes by inhibiting glucose metabolism corresponding to projected changes of cellular NADH concentration [15,40,42]. We would also expect that NAD(P) transport or synthesis is initiated very early in the chlamydial developmental cycle as gene-expression analysis indicates that metabolic enzymes requiring NAD as cofactor are expressed as immediate early genes [6]. However, because of the small size of the intracellular C. trachomatis inclusions in the first 12 hours of infection, a reliable analysis of the NAD(P)H fluorescence in the early phase of infection could not be performed.
A more detailed understanding about the metabolic activity and needs of C. trachomatis during the intracellular growth phase is needed to conceive novel therapeutic strategies to target the pathogen in its intracellular growth phase without affecting the host. Fluorescence lifetime imaging using two-photon microscopy bares new insights into the crosstalk between host and pathogen metabolism and suggests C. trachomatis-induced changes in subcellular NAD(P)H contents to directly interfere with nuclear NAD(P)H signaling pathways that are involved in cellular survival and longevity. In the process of understanding on how intracellular pathogens interfere with host cell metabolism, metabolic profiling of infected cells by FLIM of NAD(P)H will be an invaluable tool that complements established large scale genomic and proteomic approaches.  Table S4. doi:10.1371/journal.ppat.1002108.g004

Materials and Methods
Cell culture and propagation of C. trachomatis and C. pneumoniae HEp-2 epithelial cells (ATCC CCL-23) were cultured in Dulbecco's modified Eagle's Medium (DMEM) with glucose (4.5 g/l) (PAA,), or for glucose starvation studies, in glucose free DMEM (PAA, Germany) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 4 mM L-glutamine, 110 mg/l sodium-pyruvate, 10 mg/ml gentamicin and 30 mM Hepes. For two-photon microscopy, 5610 5 HEp-2 cells per dish were cultured in 50 mm culture dishes. For recovery assays, 3610 5 HEp-2 cells per well were cultured in 6-well culture plates. Cells were allowed to adhere for 24 hours prior to infection with C. trachomatis or C. pneumoniae. Cells were grown at 37uC, in 5% CO 2 humidified air. C. trachomatis strain L2 (ATCC VR-902B) and C. pneumoniae CWL029 strain (ATCC VR-1310) were purified on discontinuous density gradients. HEp-2 cells were infected with 1 inclusion forming unit (IFU) C. trachomatis per cell. HEp-2 cells were infected with 4 IFU C. pneumoniae per cell in the presence of 1 mg/ml cycloheximide by using centrifugation.

FLIM of NAD(P)H by two-photon microscopy
For two-photon microscopic studies, HEp-2 cells were grown on cover glass in 50 mm culture dishes and infected with C. trachomatis or C. pneumoniae as described above. Cover glasses were examined in a MiniCeM chamber for microscopy (JenLab, Jena, Germany) fitted to a heated stage, which enabled live cell imaging. The twophoton microscope (DermaInspect; Jenlab) was equipped with a Chroma 640DCSPXR dichroic mirror (AHF analysentechnik AG, Tübingen, Germany) and a 406/1.3 Plan-Apochromat oilimmersion objective (Zeiss, Göttingen, Germany). A tunable infrared titanium-sapphire femtosecond-laser (710-920 nm tuning range; MaiTai; Spectra Physics, Darmstadt, Germany) was used as an excitation source at 730 nm excitation for FLIM of NAD(P)H. Residual excitation light was blocked from the FLIM detector by a blue emission filter (BG39, Schott AG, Mainz, Germany).

Detection of spectrally distributed autofluorescence and fluorescent labeling of mitochondria
Fluorescence light was collected in four emission channels simultaneously using a set of dichroic mirrors and four photomultiplier tubes (Hamamatsu R1294A and R1295A). Fluorescence emission was spectrally distributed as follows: channel 1: 380-450 nm, channel 2: 450-500 nm, channel 3: 500-580 nm, channel 4: 580-680 nm. Residual excitation light was blocked by a two-photon emission filter (E680SP, Chroma Technology Corp., Bellows Falls, VT). Fluorescence intensities in each emission channels were analyzed in medium and in ROIs in host cell mitochondria and chlamydial inclusion by ImageJ. Background fluorescence intensity was recorded in a dark measurement without excitation light and was subtracted from the autofluorescence intensity values. The peak of emission in channel 2 (emission between 450-500 nm) at 730 nm excitation by two-photon microscopy corresponds to the peak of NAD(P)H fluorescence spectrum [18,43]. For the fluorescent detection of mitochondria, cells were pre-incubated with the mitochondrial membrane potential sensitive dye, tetramethylrhodamine-ethylester (TMRE) (10 nM) for 15 minutes. Because of the overlap of the two-photon excitation cross-section between TMRE and NAD(P)H, simultaneous excitation of both fluorescence signals was possible at 730 nm. Fluorescence of channel 2 (NAD(P)H) and fluorescence of channel 4 (TMRE) were used for creating merged images by ImageJ.

GFP Imaging and co-localization analysis
GFP was excited at 840 nm by the two-photon microscope described above. Cellular autofluorescence was minimal at 840 nm when low excitation power was used. Two-photon excitation of GFP fluorescence at 730 nm provided minimal absorption when GFP overexpression was at a low level. To analyze co-localization, intensity images were exported to the Keyence BZ9000 Analysis software (Osaka, Japan) and line intensity profiles of the fluorescence were calculated.

Chlamydial inclusion size measurements
Chlamydial inclusion sizes were analyzed by measuring the area of inclusions after exporting the grey-scale FLIM intensity images from SPCImage software to ImageJ (NIH, Bethesda, MD). 54 inclusions were analyzed from three independent experiments.

IFNc induced persistence
To induce persistence, HEp-2 cells were treated with 10 units/ ml IFNc for 24 hours prior to and during C. trachomatis infection.

Chlamydial recovery
To determine the burden of infectious C. trachomatis elementary bodies after intracellular development, titration experiments were performed. Infected HEp-2 cells (24 hpi) were mechanically detached with a cell scraper and re-suspended in fresh growth medium. Serial dilutions of suspension were inoculated in confluent cycloheximide-treated HEp-2 cell monolayers with centrifugationassisted inoculation. Development of chlamydial inclusions after 30 hours was analyzed on methanol-fixed slides using FITC-labeled monoclonal chlamydial-LPS antibodies (Dako, Glostrup, Denmark). Chlamydial recovery was calculated as IFU/ml by observation of 10 microscopy fields (406magnification).

Metabolic inhibition of HEp-2 cells
To inhibit host cell glycolysis, glucose starvation or 2FDG (Sigma) treatment were used. To inhibit host cell oxidative phosphorylation, antimycin A (Sigma) treatment was used. HEp-2 cells were allowed to adhere for 24 hours. For glucose starvation, media was changed to media containing no glucose. For 2FDG and antimycin A treatment, media was changed to media containing 5 mM 2FDG or 3 nM antimycin A respectively. 15 minutes after the media was changed, cells were infected with C. trachomatis or were left un-infected. After 24 hours of infections, FLIM measurements, recovery assays, ATP measurements and immunofluorescence stainings were performed.

Immunofluorescence staining
Growth media was changed to media containing no glucose or containing antimycin A (3 nM) prior to infection with C. trachomatis. After 24 hours of infection, cells were fixed with  Table S5A. (C) Treatment of cells with 2FDG or infection with C. trachomatis decreases the a 1 /a 2 ratio inside the nucleus. Quantitative analysis of a 1 /a 2 ratio (n = 54 ROIs from three independent experiments; mean 6SEM). Detailed results of statistical analysis are shown in Table S5B. doi:10.1371/journal.ppat.1002108.g006 methanol and chlamydial inclusions were stained with FITClabelled monoclonal chlamydial-LPS antibodies (Dako). For better visualization cells were counterstained with 0.1% Evans-blue.

ATP measurements
ATP was measured using a luminometric ATP assay kit (ABD Bioquest, Sunnyvale, CA) and a microplate reader (Tecan Infinite 200 PRO, Maenedorf, Switzerland). HEp-2 cells were grown in 96-well plates (2000 cells/ well) and treated with the metabolic inhibitors as described above. ATP assays were performed according to the manufacturer's instructions.

Electron microscopy
Cells were fixed with 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M cacodylate buffer for 1 hour. Postfixation was performed with 1% OsO4 in O.1 M cacodylate buffer for 2 hours. Samples were dehydrated with graded ethanol series and embedded in araldite (Fluka, Buchs, Switzerland). Ultrathin sections were stained with uranyl acetate and lead citrate and were examined with a Philips EM 400 transmission electron microscope.

Statistical hypothesis testing
For all comparisons, we used nonparametric models for longitudinal data as described previously [44]. Due to small sample sizes, ANOVA-type statistics were applied with a Boxapproximation. Adjustment for multiple testing: Overall, six sets of statistical tests were carried out ( Figures 1B, 3B, 4D, 6B, 6C, S4B). The overall significance level was set to 0.05. Within each set, adjustment for multiple testing was performed using the hierarchical procedure by Bonferroni-Holm (Bonf-Holm). To account for the testing of six sets of hypotheses, a Bonferroni correction was performed overall. Table S1 Quantitative analysis of NAD(P)H FLIM. Fluorescence lifetimes, relative amounts and fluorescence quantum yields of free and protein-bound NAD(P)H in the cytosol, mitochondria and nucleus of non-infected HEp-2 cells and in the C. trachomatis inclusion. t 1 : fluorescence lifetime of free NAD(P)H, t 2 : fluorescence lifetime of protein-bound NAD(P)H, a 1 : relative amount of free NAD(P)H, a 2 : relative amount of protein-bound NAD(P)H, q 1 : fluorescence quantum yield of free NAD(P)H, q 2 : fluorescence quantum yield of protein-bound NAD(P)H (n = 54; mean 6 SD).

(DOC)
Table S2 Statistical analysis of t 2 -NAD(P)H in cytosol, mitochondria, nucleus of non-infected HEp-2 and in chlamydial inclusion at 24 hpi. The model included experimental days (three per group, hence six in total) and compartment (cytosol, mitochondria, nucleus, inclusion) as independent factors and images per day (six) as well as cells per image (three) as repeated measures with all main effected and interactions. The dependent variable was t 2 -NAD(P)H in nucleus. Compartment was tested overall as well as comparing inclusion with every other compartment.