Intracellular bacterial pathogens (IBPs) represent a group of disease-causing bacteria that are able to invade a variety of differentiated mammalian cells of various differentiation states, including immune cells (in particular those of the IIS), and to survive and proliferate in specialized membrane-surrounded vacuoles or in the cytosol (Figure 1) for extended periods of time before leaving the host cells by cell disruption or cell-to-cell spread (Cossart and Sansonetti, 2004; Kumar and Valdivia, 2009; Ray et al., 2009; Fredlund and Enninga, 2014). Most of the IBPs are facultative intracellular, i.e., they are able to replicate within host cells, but also extracellularly in tissues or mucosal excretions, in natural environments and in more or less complex cell-free (“axenic”) media (Omsland et al., 2009). The efficient replication of a minority of IBPs (so-called “obligate IBPs”) appears to be strictly dependent on host cells (Wood et al., 2014).

Most IBPs have a typical Gram-negative cell envelope structure and inject (often many) effector proteins by specific protein secretion/injection systems (mainly type 3, and 6 secretion systems, T3SS, T4SS or T6SS; see Figure 1) into the host cells cytosol (Costa et al., 2015). Most of these effectors, whose functions are currently intensively investigated, are apparently involved in the uptake and/or the survival/proliferation processes of these IBPs by interfering mainly with different signaling pathways of host cells affecting cytoskeletal dynamics, survival, innate immune responses and possibly also metabolism of the contacted host cells (Alto and Orth, 2012; Eisenreich et al., 2013). The best-studied representative among the less frequently occurring Gram-positive IBPs is Listeria monocytogenes that replicates in the cytosol of infected host cells. The pathogenic Mycobacterium species have a unique cell envelope different from the typical cell envelopes of Gram-positive and Gram-negative bacteria. These pathogens possess type 7 protein secretion systems (ESX/T7SS) that secrete proteins some of which are clearly involved in pathogenicity and host cell interaction (Simeone et al., 2009; Houben et al., 2014).

Numerous virulence factors essential for invasion, intracellular survival and proliferation of the IBPs have been characterized (Cossart and Sansonetti, 2004). Although L. monocytogenes as Gram-positive pathogen lacks a protein injection apparatus, it produces a variety of secreted and cell-bound internalins that interact as ligands with different host cell receptors thereby performing in part similar trigger functions as the Gram-negative effector proteins (Bierne et al., 2007; Mcgann et al., 2007).

The expression of the major virulence genes of IBPs is often controlled by master transcription regulators, with links to the metabolism of the IBP and even of the host cell (Stoll et al., 2008; Poncet et al., 2009; De Las Heras et al., 2011; Gillmaier et al., 2012; Reniere et al., 2015). Not surprisingly, all IBPs are heterotrophic, aerobic or facultative anaerobic bacteria, able to survive and replicate under the normoxic and hypoxic conditions which they may encounter during their infection cycles. The metabolic potentials of most IBPs are reduced compared to the metabolic capacity of typical heterotrophic generalists (e.g., E. coli; see Figure 2). The extent of the metabolic reduction differs, however, strongly among these IBPs in view of the catabolic as well as the anabolic capabilities (Eisenreich et al., 2010; Fuchs et al., 2012), suggesting that the dependency on nutrient supply to be delivered by the host cell and the adaptation of the bacterial metabolism to that of the host cell must be specific for each IBP. Interestingly, there are, however some metabolic functions that are maintained by all IBPs and hence might be indispensable for intracellular bacterial survival (Figure 3).

The metabolic burden put on the host cell by these IBP infections has a strong impact on the host cells metabolism, causing activation of specific signaling pathways which may result in increased catabolic, energy and anabolic metabolism but also in induction of apoptosis/pyroptosis and autophagy (Eisenreich et al., 2013). These metabolic changes may in turn influence the metabolism of the intracellular bacteria. Thus, a complex network of metabolic interactions between the IBPs and host cells exists which determines the fate of IBPs within host cells, especially immune cells of the IIS (Eisenreich et al., 2015).

In contrast to the highly diverse metabolic potentials of the IBPs, the metabolic equipment of mammalian cells is quite similar (Figure 4). But in spite of this, the metabolic activities of host cells can change dramatically depending on the differentiation state, external and internal signals and (in case of cancer cells) genetic alterations.

Without external stimuli (e.g., growth factors) most terminally differentiated mammalian cells, including the immune cells of the IIS are non-proliferating or only slowly proliferating, i.e., these cells are in a (more or less) metabolically quiescent state (Figure 4A). In this state they take up, at a low rate, glucose—dependent on the cell type—by one of several glucose transporters (GLUTs) (Chen et al., 2015). Glucose is then phosphorylated by hexokinase (HK) to glucose-6-phosphate (G6P) and further metabolized in the cytosol through the glycolytic pathway (GL) and/or the pentose phosphate pathway (PPP). G6P is oxidized in the GL to pyruvate which is shuttled into mitochondria where it is further oxidized to carbon dioxide by pyruvate dehydrogenase (PDH) and the tricarboxylic acid cycle (TCA), generating NADH/H⁺, FADH₂, GTP, and intermediates, essential for anabolic pathways (especially fatty acids/lipids, amino acids). G6P can be also oxidized by G6P dehydrogenase (G6PDH) and 6-phosphogluconate dehydrogenase (6PGDH) to ribulose-5P (Ru5P) in the oxidative branch of the PPP. Both reactions generate NADPH, an important co-factor for reductive biosynthetic pathways (e.g., fatty acids/lipids, deoxyribonucleotides) and scavenging of reactive oxygen species (ROS). In the non-oxidative branch of the PPP, Ru5P is reversibly converted by ribose-5-phosphate isomerase (RPI) to ribose-5P (R5P) which is an essential precursor to many biomolecules, especially nucleotides, and to xylulose-5P (X5P) by ribulose-5-phosphate 3-epimerase (RPE). This non-oxidative PPP branch consists of additional reversible reactions catalyzed by transketolase (TKT) and transaldolase (TAL) and recruits the glycolytic intermediates fructose-6P (F6P) and glyceraldehyde-3P (G3P), which can thereby also be converted into pentose phosphates (and vice versa) thus linking GL and PPP (see Figure 4). Indeed, GL and PPP are coordinately regulated to support cell growth and survival as discussed below in more detail.

Under non-activated cell conditions, ATP production occurs mainly by OXPHOS in the electron transport chain (ETC) whereby the electrons of NADH/H+ (generated in GL and TCA cycle) are transferred to oxygen as final acceptor. Under these normal physiological conditions growth factors are available only in limited and strictly controlled amounts (Rathmell et al., 2000). Hence, most terminally differentiated cells perform a catabolic metabolism which allows optimal efficiency of ATP production from limited nutrient supply (DeBerardinis et al., 2006; Vander Heiden et al., 2009; Ward and Thompson, 2012).

When growth factor concentration is increased, nutrient uptake (most notably by glucose and glutamine transporters) is induced and the intracellular nutrient pools rise. These cells now adopt a more anabolic metabolism (i.e., enhanced biosynthesis of nucleotides, “non-essential” amino acids, fatty acids/lipids and the macromolecules), double the cellular biomass and initiate cell division (DeBerardinis et al., 2008; Vander Heiden et al., 2009). Compared to the non-activated differentiated cells, the growth factor-stimulated, proliferating cells do not optimize efficiency of ATP production but rather the production of larger amounts of intermediary metabolites and NADPH for glutathione production (essential for antioxidant defense) and other biosynthetic reactions (Figure 4B). ATP is now mainly produced through glycolysis by substrate phosphorylation. Excess pyruvate generated as final product in the GL is reduced to lactate which is secreted. The reaction also restores NAD necessary for the maintenance of the glycolytic flux. This kind of metabolism (“aerobic glycolysis”), which is energetically inefficient but leads to fast ATP production and favors anabolic processes, is characteristic for many fast proliferating cancer cells (DeBerardinis, 2008; Ngo et al., 2015). Although the cells of the mononuclear phagocyte system (MPS), comprising monocytes (MOs), MPs and DCs, do not proliferate in tissue after extravasation, they can switch their metabolism upon transition from the quiescent to the activated state. The metabolism in the activated state of MP cells shows similarities to that of cancer cells (O'Neill and Pearce, 2016).

Expression of the pathways involved in carbon and energy metabolism of mammalian cells is controlled by a complex net of nutrient sensors, growth hormone receptors, several downstream signaling pathways, canonical transcription factors including NF-κB, AP-1, NFAT, ChREBP, PPARα, SREBP as well as non-coding RNAs (Eberle et al., 2004; Postic et al., 2007; Fritz and Fajas, 2010; Cairns et al., 2011; Thompson, 2011; Harris et al., 2012; Iizuka et al., 2013; Ganeshan and Chawla, 2014; Jiang et al., 2014; Pollizzi and Powell, 2014; Yu et al., 2015). In the last decade, studies in particular on the dysregulated metabolism of cancer cells also highlighted the significance of (proto)oncogenes and tumor suppressors as key regulators of cellular metabolism. Most important players are here PI3K/AKT1, mTORC1, AMPK, RAS, HIF-1, cMYC, and p53 (Figure 5). Whereas, the impact of these factors on the regulation of growth cycle, differentiation, proliferation, immune responses, survival, and apoptosis of mammalian cells were already extensively investigated in the past, their involvement in the regulation of central metabolic pathways was recognized only recently. For recent reviews, see (Yin et al., 2012; Iurlaro et al., 2014). In the following, we focus mainly on the impact of the latter factors for the regulation of the central carbon metabolism, since they may be primary targets for the interaction with specific effector proteins of IBPs resulting in the reprogramming of the host cell metabolism during infection (Table 1).

Resting NPs which contain few mitochondria and consume little oxygen (Van Raam et al., 2008) are mainly activated by TLR agonists, formylated peptides or by the local growth factor G-CSF. Activated NPs show increased oxygen consumption, glucose uptake, and now produce ATP exclusively by aerobic glycolysis (Gabig et al., 1979). PPP activity required especially for the generation of NADPH is also induced. NADPH is essential for the induction of NADPH oxidase responsible for the production of H₂O₂, the characteristic microbicidal product of NPs. Netosis, the special ability of NPs to release extracellular nets containing DNA and microbicidal proteins (Urban et al., 2009) is also dependent on NADPH oxidase (Kirchner et al., 2012).

DCs, a heterogeneous cell population, reside in tissues exposed to the external environment and represent a surveillance system sensing infections and the first line of defense against invading pathogens (Dominguez and Ardavin, 2010; Arnold-Schrauf et al., 2014; Mildner and Jung, 2014; Durai and Murphy, 2016). DCs also act at the transition from IIS and AIS by presenting microbial antigens to T cells and providing inflammatory signals that modulate T cell differentiation (Steinman, 2012).

Resident DCs undergo, upon interaction of microbial MAMPs and cellular DAMPs with the corresponding PRRs, a rapid switch from the resting metabolic state (characterized by fatty acid oxidation and OXPHOS) to an activated state. This is accompanied by the extensively studied induction of cytokines, chemokines and other factors functioning in IIS (and AIS). Most of the encoding genes are controlled by NF-κB, the expression of which is also highly induced under these conditions. The activated metabolic state of (MO-derived) DCs (triggered e.g., by E. coli or only by its LPS) is characterized by the induced expression of genes encoding proteins involved in glucose catabolism via glycolysis and PPP and transcription factors activating these pathways, in particular HIF-1 (Huang et al., 2001; Jantsch et al., 2008). An in-depth analysis (Krawczyk et al., 2010) showed that the interactions of TLR2, TLR4, and TLR9 with their corresponding ligands, i.e., bacterial PGN, LPS, and CpG, respectively, cause a switch from the OXPHOS-engaged resting state to the activated state characterized by aerobic glycolysis and lactate production similar to the “Warburg effect” in cancer cells (Warburg et al., 1927). These so-called Tip-DCs² induce expression of iNOS resulting in increased NO production from arginine which seems to be responsible for the breakdown of mitochondrial respiration (Everts et al., 2012). This TLR-induced metabolic change is driven by the PI3K/Akt pathway (probably in a MyD 88-dependent manner (Gelman et al., 2006) and relies on HIF-1α (Jantsch et al., 2008). AMPK and the anti-inflammatory IL-10 antagonize the TLR-induced activation process (Krawczyk et al., 2010; Carroll et al., 2013). In contrast to cancer cells and activated T and B cells in which the metabolic switch to aerobic glycolysis causes enhanced cell division, the Warburg effect in DCs is mainly associated with increased biosynthetic activities (in particular production of immune mediators, like IL-12, IL-6, and TNF) rather than with cell proliferation. A more detailed overview on the present knowledge of the metabolic changes in the various activated DC subtypes is given in a recent review (Pearce and Everts, 2015).

MPs can be functionally grouped, in a simplified view, into pro-inflammatory bactericidal M1 and anti-inflammatory M2 subsets (Mantovani et al., 2005; Mosser and Edwards, 2008). In vivo studies show, however, that a broad spectrum of MP subtypes seems to exist with considerable plasticity rather than solely the M1- or M2-polarized subtypes (Murray and Wynn, 2011; Stewart et al., 2012; Vergadi et al., 2017).

Classical activation of resident, metabolically quiescent MPs, e.g., by LPS and IFN-γ, induces the M1 phenotype which is characterized (similar to the activated DC state) by HIF-1 activation, leading to increased glucose uptake and enhanced aerobic glycolysis, increased flux through the PPP, inhibition of mitochondrial OXPHOS and enhanced lipid biosynthesis (Rathmell, 2012). The LPS/TLR4-mediated activation also results in increased RNI and ROS production which mainly contributes to the antimicrobial activity of M1 macrophages. ROS is not only produced by the induced phagosomal NADPH oxidase, but also in the mitochondria by translocation of the tumor necrosis factor receptor-associated factor 6 (TRAF6) into these organelles (West et al., 2011). These MPs use significant amounts of the glycolytically generated ATP to maintain the mitochondrial membrane potential, thereby preventing apoptosis (Garedew et al., 2010). The M1 state is characterized by the production of pro-inflammatory cytokines TNF-α, IL-1ß, and IL-12.

Alternatively activated M2 macrophages (activated by IL-4 and IL-13) exhibit AMPK-controlled enhanced mitochondrial OXPHOS, fatty acid oxidation and a low rate of glycolysis (Vats et al., 2006). These MPs produce the anti-inflammatory cytokines IL-4 and IL-5. IL-4 induces the expression of arginase 1, a characteristic marker of M2 macrophages. This enzyme converts L-arginine to L-ornithine (and urea) which may lead via glutamic semialdehyde to L-proline. In contrast, in M1 macrophages imported L-arginine is used for the generation of NO catalyzed by iNOS (Gordon and Martinez, 2010; Odegaard and Chawla, 2011; Martinez and Gordon, 2014).

An interesting finding concerning the M1/M2 polarization has been reported by Haschemi et al. (2012). These authors showed that the sedoheptulose kinase SHPK (formerly carbohydrate kinase-like CARKL), which catalyzes an orphan reaction in the PPP, is an important modulator for M1/M2-polarization during LPS activation. SHPK downregulation (or loss) induced by LPS leads to induction of M1-like polarization while SHPK upregulation results in M2-like activation of MPs.

In summary, the metabolic hallmarks of activated NPs, DCs, and M1 macrophages are aerobic glycolysis, absence of OXPHOS and enhanced PPP leading to increased biosynthetic activity but not to cell proliferation. The high rate of glycolysis seems to be necessary to meet the increased demand of energy and precursors for the biosynthesis of antimicrobial metabolites and pro-inflammatory proteins. In contrast, M2 macrophages perform OXPHOS, fuelled mainly by fatty acid oxidation, while aerobic glycolysis does not occur.

Surface epithelia, such as the mucosal epithelia and the skin, are only one or a few cell layers thick but represent effective protective barriers against most microorganisms. Both, MECs, and SKs are non-professional phagocytes, but produce a diversity of antimicrobial proteins that directly kill or inhibit the growth of most of the frequently encountered microorganisms. Due to their antimicrobial potential (Gallo and Hooper, 2012), MECs and SKs are also considered as part of the IIS. Many IBPs are able to actively invade the epithelial cells and to replicate in the intracellular environment during infection (Pizarro-Cerdá and Cossart, 2006). The metabolic responses of MECs and SKs caused by IBP infections are poorly studied.

In response to disturbance of the tissue integrity, due to different events including infections by IBPs, MOs and NPs are recruited to the sites of infection where the activated phagocytes contribute to local inflammation. These inflammatory sites are characterized by low concentrations of oxygen and glucose, but high levels of lactate and ROS (Saadi et al., 2002; Campbell et al., 2014). The surrounding epithelial cells are affected by the oxygen depletion and ROS generation leading (among others) to HIF-1 stabilization (Campbell and Colgan, 2015) and enhanced HIF-regulated expression of antimicrobial factors (Peyssonnaux et al., 2008) and probably also to HIF-triggered shifts of metabolic pathways (Kominsky et al., 2010).

Different IBPs prefer different carbon compounds for energy generation and production of necessary catabolic intermediates (Eisenreich et al., 2010; Abu Kwaik and Bumann, 2013; Barel et al., 2015). The few metabolic generalists among the IBPs, mainly the facultative intracellular Enterobacteriaceae (Salmonella, Shigella and the related enteroinvasive E. coli) and possibly also Brucella spp. seem to use glucose as preferred carbon source for their intracellular catabolic and anabolic activities (Bowden et al., 2009; Götz et al., 2010; Xavier et al., 2013; Waligora et al., 2014). However, glucose is also the primary carbon source for metabolically activated immune cells that serve as host cells for IBPs. Extensive glucose deprivation by the IBPs may therefore lead to cell death by apoptosis or pyroptosis (Fink and Cookson, 2007; Suzuki et al., 2007; Lembo-Fazio et al., 2011).

On the other hand, IBPs that have adapted to a mainly intracellular life, like Rickettsia, Chlamydia, Coxiella, Francisella, and Legionella are more or less auxotrophic and often lack pathways and reactions for efficient glucose uptake and/or catabolism. These IBPs, but even the above mentioned metabolic generalists may pursue an intracellular metabolic strategy to achieve optimal intracellular replication which we recently termed “bipartite metabolism” (Grubmüller et al., 2014; Abu Kwaik and Bumann, 2015; Eisenreich et al., 2015). In this scenario, a combination of several host cell-derived carbon compounds drives the intracellular microbial metabolism and thereby allows a fine-tuned adaptation to the host cell (Figure 6). One (or more) host cell-derived energy-rich carbon substrate(s), less critical for the host cell metabolism than glucose, may function as major energy source for ATP production by substrate phosphorylation or oxidative phosphorylation via the ETC. These substrates include glycolysis-derived compounds, like glycerol(-3P), pyruvate and lactate, but also TCA intermediates, like succinate and malate, or amino acids that can be converted to such substrates, such as serine, alanine, aspartate, glutamate, and fatty acids.

More specifically, for this purpose L. monocytogenes seems to use glycerol (Eylert et al., 2008), Shigella pyruvate (Kentner et al., 2014), Francisella tularensis various amino acids, glycerol, glycerol-3P, or pyruvate (Brissac et al., 2015), Salmonella enterica glycerol, fatty acids and lactate (Steeb et al., 2013), Legionella pneumophila amino acids especially serine (Price et al., 2011; Schunder et al., 2014), Brucella abortus possibly glutamate (Zuniga-Ripa et al., 2014), Chlamydia trachomatis possibly malate (Iliffe-Lee and Mcclarty, 2000; Mehlitz et al., 2017), and Mycobacterium tuberculosis fatty acids, cholesterol (Ehrt et al., 2015; Lovewell et al., 2016) or glycolytic C₃ compounds (Beste et al., 2013), respectively. These carbon substrates are not necessarily utilized for the generation of hexoses (and subsequently other carbohydrates needed for cell envelope synthesis) via gluconeogenesis. Indeed, several IBPs (Rickettsia, Chlamydia, Coxiella, and Legionella) lack the typical bacterial type I-III fructose-1,6-diphosphatases (FBPase, an essential enzyme for gluconeogenesis). It is, however, possible that these IBPs, some of which (e.g., L. pneumophila) are definitely able to perform gluconeogenesis, might compensate the missing conventional FBPase by alternative FBPase activities, e.g., by the reversible pyrophosphate-dependent phosphofructokinase or fructose-6-phosphate aldolase (Schürmann and Sprenger, 2001; Arimoto et al., 2002; Ganapathy et al., 2015). On the other hand, in Brucella abortus which possesses type I and type II FBPases, both enzymes are dispensible for intracellular replication (Zuniga-Ripa et al., 2014), whereas in Francisella tularensis inactivation of the glpX gene (encoding a type II FBPase) leads to reduced intracellular multiplication (Brissac et al., 2015), indicating the need of gluconeogenesis for intracellular replication of this IBP particularly in those stages of infection in which glucose is limited. However, when glucose is available most IBPs take up glucose in addition to the above mentioned carbon source(s) from the host cell, but use it for anabolic rather than for energy-producing catabolic pathways (Figure 6). In this context, it is interesting to note that most IBPs (exceptions are the facultative intracellular Enterobacteriaceae) lack the highly efficient glucose-specific phosphoenolpyruvate-dependent transport system (PTS), but may possess less efficient ABC or major facilitator superfamily GLUTs (Aguilar et al., 2012). Interestingly, L. monocytogenes possess a large number of PTS, but lack the most efficient glucose-specific PTS-G (Stoll and Goebel, 2010). Several IBPs are able to import phosphorylated hexoses (especially glucose-1P and/or glucose-6P, deriving eventually from host cell glycogen) by means of specific transporters: UhpT in case of L. monocytogenes (Chico-Calero et al., 2002) and UhpC in case of Chlamydia pneumoniae (Schwöppe et al., 2002) and Legionella pneumophila (Schwöppe et al., 2002; Chien et al., 2004). The expression of these transporters may be induced under the intracellular conditions (Joseph et al., 2006).

The imported glucose or glucose-P seems to be mainly channeled into the PPP. NADPH, produced in the oxidative branch of the PPP (present in IBPs with the exception of Francisella, Coxiella, and Rickettsia spp.), serves as cofactor for the reductive biosynthesis of nucleotides and fatty acids and for the generation of glutathione which provides protection against oxidative damage. Through the non-oxidative arm of the PPP (present in all IBPs except Rickettsia spp.) specific phosphorylated carbohydrates are generated which are required for the biosynthesis of aromatic amino acids, nucleotides and bacterial cell envelope components.

Most other basic anabolic compounds, especially amino acids (even those which in principle could be synthesized by the IBPs), are imported from the host cell (Grubmüller et al., 2014). An exception is meso-diaminopimelate (mDAP), an essential component of peptidoglycan synthesis which cannot be delivered by the host cell and has to be apparently produced by the IBPs (Figure 1). Among all IBPs for which genome sequences are known, only Francisella tularensis and other Francisella spp. lack most genes encoding the enzymes necessary for mDAP biosynthesis.

The withdrawal of amino acids from the host cells by the IBPs will cause amino acid starvation in the host cells (Tattoli et al., 2012; Tsalikis et al., 2013) which may lead, as discussed below, to induction of “prosurvival autophagy” thus prohibiting early host cell death by apoptosis/pyroptosis. Some IBPs possess anti-autophagic properties and in this case the induced autophagy of the host cell may even support intracellular replication of the IBPs (see below).

The concept of bipartite metabolism may also explain the frequently observed persistent state of IBPs, i.e., their intracellular survival without active replication. Interruption of the major glucose (or glucose-6P) supply to the IBP will stop bacterial cell wall synthesis and cell division, but as long as the catabolism of an alternative carbon substrate provides sufficient ATP (and eventually necessary amounts of phosphorylated carbohydrates generated by gluconeogenesis and PPP) to maintain the integrity of indispensable macromolecules (especially of DNA and cell envelope structures), the IBP will survive but not multiply.

As outlined above, the well-characterized defense mechanisms executed by the immune cells of the IIS include ROS and RNI production, synthesis of antibacterial peptides, induction of inflammasomes with the concomitant synthesis of inflammatory cytokines (IL1-ß and IL18) and subsequent pyroptosis as well as induction of autophagy (Sauer et al., 2010). These defense mechanisms are predominantly triggered by the interactions of different bacterial MAMPs with their corresponding TLRs and NLRs. Certain T3SS- and T4SS-effector proteins, e.g., SopE and SopB of S. Typhimurium secreted by T3SS-1, can also promote inflammatory host defense responses (Bruno et al., 2009).

The induced expression of these antimicrobial host responses requires activation of the metabolism of the immune cells involved (Lewis et al., 2011; Amiel et al., 2012; Everts et al., 2012). Interactions of LPS with TLR4 and of LTAs and lipoproteins with TLR2 and TLR6—presumably early events in the encounter of Gram-negative and Gram-positive IBPs, respectively with the IIS immune cells in vivo—trigger these defense reactions and activate the metabolism of the initially metabolically quiescent phagocytes (Krawczyk et al., 2010). This activation step leads to PI3K/Akt-mediated induction of aerobic glycolysis and NF-kB activation causing (among others) increased production of ROS and NO with subsequent loss of OXPHOS and inhibition of the TCA cycle (Everts et al., 2012). Induced expression of PKM2 seems to play an important role in this metabolic switch and is characteristic to inflammatory immune cells, such as M1 macrophages, Tip-DCs (M1-related cells that release TNF and NO, see above) (Schmid et al., 2012; Guilliams et al., 2014; Martinez and Gordon, 2014; Murray et al., 2014) and Th17 cells (Korn et al., 2009). This type of inflammatory cell metabolism represents, however, a rather hostile setting for the intracellular replication of invading IBPs (Price and Vance, 2014; Palsson-Mcdermott et al., 2015). Indeed, mice with impaired NO generation show dramatically enhanced bacterial proliferation in Kupffer cells upon infection with S. Typhimurium (Vazquez-Torres et al., 2004), suggesting that, without the antimicrobial NO production, intracellular replication of the S. Typhimurium in these metabolically activated MPs proceeds much more efficient.

To counteract the plethora of antimicrobial host cell reactions, IBPs have developed a variety of defense mechanisms: (a) inhibition of MAMP/PRR recognition through modification and shutting down the production of MAMPs, like LPS, LTA, and flagellin, (b) inactivation of ROS by induced production of superoxide dismutase and catalase, (c) reduction of the NO level through induction of arginase, (d) inhibition of inflammasome formation and pyroptosis, and (e) suppression of autophagy (Hedrick, 2004; Shen and Higgins, 2006; Bhavsar et al., 2007; Coats et al., 2007; Phalipon and Sansonetti, 2007; Ishii et al., 2008; Pieters, 2008; Taxman et al., 2010; Kullas et al., 2012; Fabrik et al., 2013; Smith and May, 2013; Al-Khodor et al., 2014; Okumura and Nizet, 2014; Wynosky-Dolfi et al., 2014). The realization of most of these bacterial countermeasures against the host cell defense reactions requires, however, a host cell metabolism that allows active intracellular replication of the IBPs and it remains to identify the appropriate metabolic host cell conditions that activate the IBP-specific countermeasures.

The paradox situation that the same immune cells may carry out antimicrobial (i.e., kill IBPs) but also promicrobial functions (i.e., serve as convenient host cells for the IBPs) was discussed by the example of MPs in a recent review (Price and Vance, 2014). An important message was that the anti-inflammatory M2 but not the inflammatory M1 phenotype of activated MPs represents a comfortable replication niche for IBPs. Indeed, recent studies (Eisele et al., 2013; Xavier et al., 2013) showed that alternatively activated M2 macrophages represent the predominant MP population inhabited by replicating Salmonella and Brucella strains, respectively, during chronic infection in the mouse model. This ability correlates with higher levels of unconsumed glucose in the M2 compared to M1 macrophages. Signal transduction through the nuclear peroxisome proliferator-activated receptors (PPAR-γ and PPAR-δ), respectively, was shown to be crucial for this process. Induction of these metabolic regulators enhance OXPHOS and fatty acid ß-oxidation in M2 macrophages thereby leading to increased levels of unconsumed glucose. M1 macrophages on the other hand catabolize glucose through the highly activated glycolytic pathway and thereby withdraw the bacteria this essential carbon source.

These findings led to the suggestion that the PPAR-induced increase in intracellular glucose availability may be a rather general mechanism promoting growth of persistent pathogens within M2 macrophages (Xavier et al., 2013). Indeed, M2 polarization of human macrophages was recently shown to favor also replication of Chlamydia pneumoniae (Buchacher et al., 2015).

However, the typical M2 metabolism does not seem to be the only metabolic condition which supports replication and proliferation of IBPs. Established MO- and MP-like cell lines, such as J774A.1, P388.D1, RAW264.7, THP-1, and U-973, which are routinely used as host cells for studying intracellular replication of IBPs, perform a permanently activated metabolism which is mainly triggered by the constitutive expression of oncogenes (e.g., Myc), or by the inactivation of tumor suppressors (e.g., p53). In these cell lines, most IBPs are able to replicate efficiently, indicating that this type of activated, glycolysis-driven host cell metabolism also supports successful intracellular replication of many IBPs (Fuchs et al., 2012; Eisenreich et al., 2013; Kentner et al., 2014). The dysregulated metabolism of these transformed cell lines is clearly different from the metabolism of M2 macrophages (characterized by ß-oxidation of fatty-acids driven OXPHOS), but apparently provides all nutrients necessary for the bipartite metabolism of most IBPs.

It should also be noted that in contrast to Salmonella and Brucella, the intracellular replication capacity of Listeria, Franciscella and Mycobacterium strains is not influenced by the modulation of the PPAR levels and hence the intracellular glucose concentration of the host cells (Eisele et al., 2013).

These data show that not only the strategy exemplified by M1/M2-polarized MPs (strategy A, outlined in Figure 7A), but also other (still rather hypothetical) strategies (exemplified e.g., by Figures 7B–D) may result in a host cell metabolism that supports IBP replication: (A) in an already activated immune cell population different subsets may exist (e.g., M1/M2 macrophages and their subgroups Martinez and Gordon, 2014) that are either metabolically favorable or adverse for intracellular replication of IBPs, (B) anti- or pro-replicative metabolic conditions may be induced in individual cells of a quiescent immune cell population depending on the interaction between different host cell receptors and different IBP ligands, (C) before or shortly after internalization the IBPs may translocate effector protein(s) that specifically counteract the antimicrobial activities (e.g., ROS or RNI production) without changing the metabolic conditions of the host cell that are otherwise favorable for IBP replication, or (D) phagocytosis of the IBPs and activation of pro-replicative metabolic conditions in the immune cells may be separately triggered by interaction of different host cell receptors with different IBP-specific ligands.

The data discussed in Interactions between host cell targets and bacterial effectors leading to metabolic changes within host cells that may support intracellular replication of IBPs, suggest that the exclusive use of glucose is not necessarily the best strategy for fueling the intracellular metabolism of all IBPs. Rather, IBPs according to their different metabolic capacities may depend—for performing an optimally balanced intracellular metabolism and hence efficient proliferation—on IBP-specific combinations of catabolic and anabolic metabolites that are provided by the host cell. This in turn requires IBP-specific metabolic conditions within (non-cancer) host cells which (in vivo) may already exist in a subset of a potential host cell population or may be induced by invading pathogens in metabolically quiescent (primary) cells through the interaction of specific bacterial ligands (e.g., effector proteins or other virulence factors) and appropriate host cell receptors (Figure 7).

As discussed in Metabolic activation of quiescent host cells that may prohibit or favor intracellular growth of IBPs, the alternatively activated M2 macrophages seem to represent such a subpopulation performing a metabolism that supports intracellular replication of certain IBPs (Price and Vance, 2014). Alternatively, specific T3SS- or T4SS-effector proteins or other virulence factors of IBPs which directly or indirectly interact with the above described central regulators of host cell metabolism, including PI3K, Akt, mTOR, cMyc, HIF-1α, PKM2, P53, TIGAR, PTEN, and AMPK, might also be candidates for modulating the metabolism of (quiescent) primary host cells in such a way that it becomes supportive for intracellular IBP replication. Indeed, interactions of IBP effectors with several of these oncogenes and tumor suppressors have already been demonstrated (Table 1). However, the outcome of these interactions has been so far mainly studied with respect to the pathogen internalization and/or immune responses, while the likely impact on the host cell metabolism has hardly been addressed. To the best of our knowledge, only few reports (see below) show how IBPs may manipulate the host cell metabolism, in particular that of immune cells by such interactions.

Gillmaier et al. (2012) showed that primary murine bone marrow-derived MPs (BMDMs) which perform in the uninfected state the typical quiescent OXPHOS-based metabolism (see above), are metabolically highly activated upon internalization of wild-type L. monocytogenes. However, in contrast to LPS- and IFN-γ-mediated metabolic activation, the L. monocytogenes-induced BMDM metabolism does not only lead to induction of the glycolytic pathway but also to induction of the initial steps of the TCA cycle thus enabling synthesis of citrate, its subsequent export into the cytosol and the generation of acetyl-CoA and oxaloacetate by ACL. This may allow enhanced anabolic activities in the BMDMs (especially fatty acid/lipid and amino acid biosynthesis). This type of induced host cell metabolism is similar to that occurring in the murine macrophage cell line J774 (its metabolism is permanently activated due to constitutively expressed Myc) where L. monocytogenes replicates without significantly altering the metabolism of these host cells. The efficiency and the mode of intracellular metabolism of L. monocytogenes are similar in the activated BMDM and in J774 cells (Gillmaier et al., 2012).

Interestingly, only a small population of the BMDMs (<5%) contains a large number (>20 bacteria/host cell) of replicating listeriae, suggesting that this type of metabolic activation, obviously favorable for intracellular replication of L. monocytogenes, may be induced only in a subset of the BMDMs. Notably, this L. monocytogenes-triggered metabolism of the BMDMs mimics more the growth factor-induced metabolism governed by the PI3K/Akt/mTORC1 pathway which leads to the activation of glycolysis, PPP and TCA (Kaplan et al., 2000; Jones and Thompson, 2009; Salminen and Kaarniranta, 2010). It has been shown that internalin B (InlB) of L. monocytogenes has structural and functional similarities to the hepatocyte growth factor (HGF), recognizes the same receptor Met as HGF and InlB/Met interaction activates PI3K (Ireton et al., 1996; Shen et al., 2000; Bierne and Cossart, 2002; Li et al., 2005; Niemann, 2013; Gessain et al., 2015). It is therefore intriguing to speculate that InlB/Met interaction in a subset of the BMDM population (Galimi et al., 2001; Moransard et al., 2010) might be responsible for the induction of the above described metabolism that supports intracellular L. monocytogenes replication. Direct experimental evidence for this hypothesis is still missing, but it has been shown (Gillmaier et al., 2012) that a non-pathogenic prfA mutant (also unable to express inlB) and heat-inactivated L. monocytogenes wild-type bacteria are unable to induce this type of metabolism in BMDM.

As discussed above, the tumor suppressor p53 blocks glucose uptake, glycolysis and PPP (Vousden and Ryan, 2009) whereas inactivation or down-regulation of p53 leads to the activation of these metabolic pathways which occurs in many cancer cells. Caco-2 cells contain a deleted and mutated p53 gene and no detectable p53 protein (Djelloul et al., 1997). These metabolically activated cells are excellent hosts for many IBPs including L. monocytogenes. Indeed, down-regulation of p53 in RAW264.7 and HeLa cells and p53 knockout mice also results in significantly increased intracellular replication of L. monocytogenes, whereas intracellular replication is inhibited when p53 was overexpressed in the two cell lines (Wang et al., 2016).

Shigella flexneri (which similar to L. monocytogenes replicates in host cells cytosol) may invade and multiply within MPs but rapidly induce pyroptotic cell death of these host cells (Suzuki et al., 2007). Epithelial cells of the intestine are the preferred host cells which allow efficient replication of Shigella upon oral infection. In the infected cells, degradation of p53 has been reported (Bergounioux et al., 2012) which is mainly caused by calpain protease activation, promoted by the T3SS effector protein VirA. The p53 degradation could activate the host cell metabolism, but the calpain activation causes also cell death and thus elimination of this replication niche (Bergounioux et al., 2012). However, IpgD, another T3SS effector of S. flexneri, dephosphorylates PIP2 and the resulting phosphatidylinositol-5-phosphate leads to the activation of the PI3K pathway which promotes phosphorylation of Akt (Pendaries et al., 2006). This Akt activation could also activate the host cell metabolism.

Down-regulation of p53 has also been demonstrated in human cells upon infection by different Chlamydia species (González et al., 2014; Siegl et al., 2014) and this down-regulation appears to be crucial for the efficient intracellular replication of C. trachomatis which in this case mainly occurs in a membrane-bound compartment, the inclusion. However, it is still unknown which chlamydial factor(s) is responsible for the p53 down-regulation and what are the metabolic consequences.

Mycobacterium tuberculosis (Mbt) infection is initiated by inhalation into the alveolus. Although studies in lungs of aerosol-infected mice show extensive replication of Mbt in alveolar epithelial cells and other non-macrophage cells and dissemination to other organs early in infection (Ryndak et al., 2015), alveolar MPs are the typical host cells of Mbt where the bacteria primarily reside in phagosomes. Fatty acids and cholesterol are considered to be the most important carbon sources for optimal growth and persistence of Mbt during infection (Lovewell et al., 2016). These two nutrients are involved in the production of energy but also of intermediates for mycolic acids, the major and specific lipid components of the cell envelope of Mbt (Marrakchi et al., 2014). Mbt is also able to utilize several other substrates depending on the growth conditions (Marrero et al., 2013; Beste et al., 2013; Ehrt et al., 2015; Hampel et al., 2015; Rücker et al., 2015) Glucose has been reported to be used for growth in vivo and both, glycolysis and gluconeogenesis, are important pathways for the intracellular carbon metabolism (Ehrt et al., 2015).

Not surprisingly, Mbt significantly reprograms the metabolism of the infected host MPs. Metabolomic profiling in mice infected with Mbt showed increased levels of several metabolites in the lung (presumably within MPs) of infected mice which suggest elevated ß-oxidation of lipids and glyoxylate shunting, increased nucleotide and amino acids metabolism and increased anti-oxidative stress response in this tissue upon Mbtinfection (Shin et al., 2011). A hallmark of mycobacterial infection is the formation of foamy MPs characterized by the accumulation of large lipid bodies. The foamy phenotype appears to be associated with the temporary increase of the peroxisome proliferator-activated receptor gamma (PPAR-y) and dependent on TLR-2 (Almeida et al., 2012). It was further shown (Singh et al., 2012) that Mbt induces the foamy phenotype by diverting the glycolytic pathway of the host MPs toward ketone body synthesis. This metabolic dysregulation leads to the activation of the anti-lipolytic G protein-coupled receptor GPR109A which causes perturbations in lipid homeostasis and consequent accumulation of lipid bodies in the MP. ESAT-6, a secreted virulence factor of Mbt interacting with TLR2, seems to be responsible for this feedback loop. The increased MP glycolysis and the accompanied lipid droplet accumulation is in line with the observation that liver X transcriptional receptors (LXRA and LRXB) which control fatty acid, cholesterol and glucose homeostasis are critical regulators of lipid metabolism during intracellular Mbt replication (Han et al., 2014).

Autophagy is a catabolic process that sequesters and degrades dispensible cellular contents, like waste proteins and damaged organelles, but also microorganisms reaching the cytosol (reviewed by Gomes and Dikic, 2014; Randow and Youle, 2014). Induction of autophagy occurs in response to depletion of ATP, amino acids and glucose and is controlled mainly by AMPK and mTORC1. As outlined above, AMPK is activated by high AMP/ATP ratios while mTORC1 activity is modulated by amino acid levels (Bar-Peled et al., 2012; Inoki et al., 2012).

In the autophagic process, the targeted cytosolic components are engulfed in double membrane surrounded autophagosomes which then fuse with lysosomes leading to the degradation of the autosomal contents by digestive enzymes. This process results in the release of metabolites (in particular amino acids and carbohydrates) that can be reused for anabolic and catabolic processes, especially for protein biosynthesis and energy generation.

“Prosurvival autophagy” may thus promote cell survival under nutrient (especially amino acid) starvation conditions (Dibble and Manning, 2013; Jewell and Guan, 2013), but it may also support microbial replication. Manipulation of both, the energy-sensing AMPK and the nutrient-sensing mTORC1 kinase activities, has been shown to be crucial for the proliferation of several viral pathogens (Brunton et al., 2013); the resulting metabolic changes in the host cells which support viral replication have, however, hardly been analyzed.

The significance of these two key regulators of cell metabolism for the intracellular replication of bacterial pathogens has also been discussed and the importance of autophagy for intracellular growth (through the supply of additional nutrients) has been clearly demonstrated for Franciscella tularensis (Steele et al., 2013).

Other IBPs adapted to intracellular life in the host cells cytosol, like L. monocytogenes and Shigella flexneri, also induce autophagy in MPs (Mostowy et al., 2011). These IBPs have developed specific mechanisms which protect themselves from the autophagic attack. L. monocytogenes protects itself by ActA and InlK (Yoshikawa et al., 2009; Dortet et al., 2012) while S. flexneri escapes autophagy mainly by IcsB (Ogawa et al., 2005). Autophagy induced in the host cells infected by these IBPs may therefore provide additional nutrients, especially amino acids (Grubmüller et al., 2014) to these intracellular bacteria thus supporting their intracellular replication.

As discussed above, activated NPs infiltrating the intestinal epithelium infected by certain IBPs, e.g., S. Typhimurium, will affect the metabolism of the surrounding epithelial cells (e.g., by HIF-1 activation) and, hence, possibly also the replication of IBPs capable of invading these epithelial cells.

Activated NPs may, however, also positively affect the metabolism and proliferation of IBPs in the extracellular state (Winter et al., 2010a; Lopez et al., 2012). The best studied example is S. Typhimurium which induces acute inflammation in the intestine by T3SS-1 effectors translocated into epithelial cells and MPs. ROS produced by the attracted activated NPs convert luminal thiosulfate to tetrathionate which serves as Salmonella-specific electron acceptor in anaerobic respiration (Winter et al., 2010b). Specific S. Typhimurium strains lysogenized by a bacteriophage carrying the sopE gene are also able to enhance the production of host-derived nitrate which is an even more efficient electron acceptor than tetrathionate allowing anaerobic nitrate respiration by these bacteria (Lopez et al., 2012). Gut-derived ethanolamine has been suggested as electron donor for this anaerobic respiration chain (Thiennimitr et al., 2011). This metabolic capacity of S. Typhimurium (and possibly of other IBPs), supported by activated immune cells, enhances proliferation of these pathogens in the gastrointestinal tract at the expense of the other fermenting resident gut microbes.

The genes essential for this anaerobic metabolism of enteric S. Typhimurium appear to be part of a metabolic network which is decaying in the extraintestinal typhoid S. Typhi (Nuccio and Bäumler, 2014). S. Typhi tightly regulates the expression of T3SS-1 by the regulatory protein TviA, thereby suppressing the pro-inflammatory host responses which reduces intestinal inflammation (Winter et al., 2014) and even impairs activation and proliferation of naïve flagellin–specific CD4⁺ T cells in Peyer's patches and mesenteric lymph nodes (Atif et al., 2014). This allows increased bacterial dissemination finally leading to systemic infection.

Other metabolites produced by activated immune cells in excess, e.g., TCA intermediates like succinate, malate (due to enhanced glutaminolysis) and especially lactate (due to enhanced glycolysis) could be used by IBPs as additional catabolic carbon sources in bipartite metabolism during intracellular replication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.