Abstract
Macrophages are key innate immune cells for determining the outcome of Mycobacterium tuberculosis infection. Polarization with IFNγ and LPS into the “classically activated” M1 macrophage enhances pro-inflammatory and microbicidal responses, important for eradicating the bacterium. By contrast, “alternatively activated” M2 macrophages, polarized with IL-4, oppose bactericidal mechanisms and allow mycobacterial growth. These activation states are accompanied by distinct metabolic profiles, where M1 macrophages favor near exclusive use of glycolysis, whereas M2 macrophages up-regulate oxidative phosphorylation (OXPHOS). Here we demonstrate that activation with IL-4 counterintuitively induces protective innate memory against mycobacterial challenge. This was associated with enhanced pro-inflammatory cytokine responses and killing capacity. Moreover, despite this switch towards a phenotype that is more akin to classical activation, IL-4 trained macrophages do not demonstrate M1-typical metabolism, instead retaining heightened use of OXPHOS. Moreover, inhibition of OXPHOS with oligomycin, 2-deoxy glucose or BPTES all impeded heightened pro-inflammatory cytokine responses from IL-4 trained macrophages. Lastly, this work identifies that IL-10 negatively regulates protective IL-4 training, impeding pro-inflammatory and bactericidal mechanisms. In summary, this work provides new and unexpected insight into alternative macrophage activation states in the context of mycobacterial infection.
Introduction
Mycobacterium tuberculosis, the causative agent of tuberculosis (TB), has arguably caused the most deaths of any pathogen in human history. M. tuberculosis continues to be the cause of more than a million deaths annually: in 2018, 10 million people were newly diagnosed with active TB and 1.4 million died from the disease (WHO, 2019). The only currently licensed TB vaccine, Bacille Calmette-Guérin (BCG), was developed a century ago and although it protects children against disseminated disease, it fails to protect adults from pulmonary infection (Ryndak & Laal, 2019). The lack of an efficient vaccine, and growing antibiotic resistance, means that new ways of combatting TB are urgently needed (Choreño-Parra, Weinstein, Yunis, Zúñiga, & Hernández-Pando, 2020).
Macrophages are key host innate immune cells for controlling M. tuberculosis infection (Cohen et al., 2018). A significant feature of macrophages is dynamic plasticity, expressed by their ability to polarize towards distinct activation states (M. L. E. Lundahl, Scanlan, & Lavelle, 2017; Sica, Erreni, Allavena, & Porta, 2015). Activation with interferon gamma (IFNγ) together with lipopolysaccharides (LPS) yields the bactericidal and pro-inflammatory “classically activated” or M1 macrophages (Ferrante & Leibovich, 2012; Sica et al., 2015), whereas activation with the type 2 and regulatory cytokines, interleukin (IL)-4, IL-13, IL-10 and transforming growth factor (TGF)β, results in “alternatively activated” M2 macrophages, which enhance allergic responses, resolve inflammation and induce tissue remodeling (Bystrom et al., 2008; Mantovani et al., 2004). Another key difference between these activation states are their metabolic profiles: whereas classically activated macrophages switch to near exclusive use of glycolysis to drive their ATP production, alternatively activated macrophages instead upregulate their oxidative phosphorylation (OXPHOS) machinery (Van den Bossche et al., 2016; Van den Bossche, O’Neill, & Menon, 2017).
Whilst classifying macrophages in this manner is a simplification – the reality is a broad spectrum of various differentiation states that is continuously regulated by a myriad of signals (Sica et al., 2015) – it is nevertheless considered that for a host to control TB infection, classical macrophage activation is vital (Jouanguy et al., 1999; Philips & Ernst, 2012). Furthermore, a macrophage metabolic shift to glycolysis is crucial for effective killing and thus overall control of TB infection (Gleeson et al., 2016; Huang, Nazarova, Tan, Liu, & Russell, 2018). By contrast, alternatively activated macrophages directly oppose bactericidal responses, which has been demonstrated to lead to enhanced bacterial burden and TB pathology (Moreira-Teixeira et al., 2016; Orecchioni, Ghosheh, Pramod, & Ley, 2019; Shi, Jiang, Bushkin, Subbian, & Tyagi, 2019). Due in large part to the established ability of classically activated macrophages to kill M. tuberculosis, inducing Th1 immunity is a key aim for TB vaccine development (Abebe, 2012; Andersen & Kaufmann, 2014; Ottenhoff et al., 2010). Apart from targeting adaptive immune memory, another promising approach has emerged in recent years: bolstering innate immune killing capacity by the induction of innate training (Khader et al., 2019; Moorlag et al., 2020).
Innate training is regarded as a form of immunological memory. It is a phenomenon where a primary challenge, such as a vaccination or an infection, induces epigenetic changes in innate immune cells, which alters their responses following a secondary challenge (Arts et al., 2018; Saeed et al., 2014; van der Meer, Joosten, Riksen, & Netea, 2015). Unlike adaptive immune memory, innate training is non-specific, i.e. the secondary challenge does not need to be related to the primary challenge. For instance, BCG vaccination has been demonstrated to protect severe combined immunodeficiency (SCID) mice from lethal Candida albicans infection; reducing fungal burden and significantly improving survival (Kleinnijenhuis et al., 2012). This ability of the BCG vaccine to train innate immunity is now believed to be a core mechanism by which it induces its protection against M. tuberculosis. Apart from the BCG vaccine, it has been demonstrated that other organisms and compounds can induce innate training, such as β-glucan (van der Meer et al., 2015) which induced protective innate training against virulent M. tuberculosis, as shown by enhanced mouse survival following in vivo infection, and enhanced human monocyte pro-inflammatory cytokine secretion following ex vivo infection (Moorlag et al., 2020).
Conversely, there is a risk that certain immune challenges could lead to innate immune re-programming that hinders protective immune responses. For instance, recent reports have demonstrated how virulent M. tuberculosis (N. Khan et al., 2020) and mycobacterial phenolic glycans (M. Lundahl et al., 2020) can program macrophages to attenuate bactericidal responses to subsequent mycobacterial challenge. In the current study, macrophage activation caused by former or concurrent parasitic infections is considered regarding the possibility of innate immune re-programming that hinders protective immune responses against this disease. Geographically, there is extensive overlap between tuberculosis endemic areas and the presence of helminths (Salgame, Yap, & Gause, 2013). With regard to macrophage activation and combatting tuberculosis, this is an issue as parasites induce type 2 responses, leading to alternative macrophage activation instead of classical activation and associated bactericidal responses (Chatterjee et al., 2017; X. X. Li & Zhou, 2013). Indeed, it was recently demonstrated how products of the helminth Fasciola hepatica can train murine macrophages for enhanced secretion of the anti-inflammatory cytokine IL-10 (Quinn et al., 2019).
Unexpectedly, our data indicates that murine macrophage activation with IL-4 and IL-13 induces innate training that enhances pro-inflammatory and bactericidal responses against mycobacteria. Although macrophages trained with IL-4 and IL-13 resemble classically activated macrophages, we identify that they do not adopt their typical metabolic profile, instead retaining heightened OXPHOS activity and notably lacking a dependency on glucose and glycolysis. Lastly, we identify IL-10 as a negative regulator of this innate training response, which may have obscured previous identification of this macrophage phenotype.
Results
Prior Alternative Activation Enhances Mycobacterial Killing
Macrophages are key immune cells for combatting M. tuberculosis. Upon infection, alveolar macrophages serve as the initial hosts of the intracellular bacterium (Cohen et al., 2018) and bactericidal responses of recruited monocyte-derived macrophages (MDM) are crucial for control and killing of M. tuberculosis (Huang et al., 2018). Overall, it has been highlighted that an early Th1 driven immune response is key to early eradication of M. tuberculosis, where classical activation of macrophages results in effective bactericidal action. However, a complicating factor is the occurrence of concurrent parasitic disease, which instead drives Th2 immunity and alternative macrophage activation, which in turn has been demonstrated to enhance TB pathology (Moreira-Teixeira et al., 2016; Orecchioni et al., 2019). In this context, we sought to investigate how type 2 responses may induce innate memory, and how such memory could affect macrophage bactericidal properties.
To investigate the effect of IL-4 and IL-13 on macrophage acute and innate memory responses, an in vitro model of BCG infection was used. Murine bone marrow-derived macrophages (BMDMs) were stimulated with IL-4 and IL-13 (M(4/13)) on Day -1, followed by infection with BCG Denmark on either Day 0 or Day 6 (Figure 1A). On either day, the macrophages were exposed to roughly 30 bacteria per cell and after three hours extracellular bacteria were removed by washing and internalized bacteria were measured by colony forming units (CFUs), resulting in a multiplicity of infection (MOI) between one and four, depending on differences in uptake (Figure 1B). Internalized bacteria were measured by CFU (Figure S1A) at 27- and 51 hours post infection to determine killing capacity (Figure 1C).
Mycobacterial killing was compared to naïve BMDMs incubated with media on day -1 (media control, M(-)) on both Days 0 and 6, and BMDMs classically activated with IFNγ and LPS (M(IFNγ/LPS)) on Day 0. M(IFNγ/LPS) could not be investigated on Day 6 due to reduced viability. On Day 0, M(IFNγ/LPS) displayed significantly enhanced mycobacterial killing at 27- and 51-hours post infection, whereas M(4/13) displayed comparable killing capacity to naïve M(-) (Figure 1C). Secretion of TNFα and IL-10 were also measured at 3-, 27- and 51 hours post infection on Day 0 (Figure S1B) and Day 6 (Figure 1D). These cytokines were chosen as TNFα is critical in the early host response against M. tuberculosis (Bourigault et al., 2013; Keane et al., 2001), while IL-10 can prevent phagolysosome maturation in human macrophages (O’Leary, O’Sullivan, & Keane, 2011) and promote disease progression in mice (Beamer et al., 2008). Following BCG infection on Day 0 (Figure S1B), IL-10 was not detectable and bactericidal M(IFNγ/LPS) secreted TNFα. Consistent with previous work, activation with IL-4 and IL-13 on the other hand did not enhance BCG killing nor induce inflammatory cytokine secretion.
By contrast, on Day 6 (innate training responses), M(4/13) exhibited both heightened mycobacterial uptake (Figure 1B) and significantly greater killing capacity compared with untrained M(-), both at 27- and 51 hours post infection (Figures 1C and S1A). Furthermore, this was accompanied by a near complete abrogation of IL-10 secretion, along with a minor decrease of TNFα compared with M(-) (Figure 1D), displaying an overall shift towards a more pro-inflammatory response profile. To investigate the change in phenotype of M(4/13) between the two days tested, a more detailed characterization was carried out.
Innate Training with IL-4 and IL-13 Promotes Pro-Inflammatory Responses
The current dogma suggests that classical (M1) macrophage activation induces upregulation of antigen presentation, enhanced secretion of pro-inflammatory cytokines including interleukin (IL)-1β, IL-6 and TNFα, as well as the production of reactive oxygen and nitrogen species (ROS and RNS respectively) (Bystrom et al., 2008; Sica et al., 2015). By contrast, alternatively activated macrophages (M2) dampen inflammation, promote angiogenesis and scavenge debris (Gordon & Martinez, 2010; Sica et al., 2015). These macrophages are identified by their upregulation of chitinase-like 3-1 (Chil3), found in inflammatory zone-1 (Fizz1, Retnla) and arginase (Arg1) (Gabrilovich, Ostrand-Rosenberg, & Bronte, 2012; Murray & Wynn, 2011), as well as surface expression of lectins, such as the macrophage mannose receptor (MR, CD206) (Brown & Crocker, 2016). Flow cytometry and qPCR analysis of M(IFNγ/LPS) and M(4/13), compared with inactivated M(-), showed that these activation phenotypes tallied with the literature: M(IFNγ/LPS) had elevated expression of CD80, major histocompatibility complex class II (MHC II) and inducible nitric oxide synthase (iNOS, Nos2) – responsible for production of the reactive nitrogen species, nitric oxide (NO) – whereas M(4/13) exhibited heightened expression of CD206, MHC II, Arg1, Chil3 and Retnla (Figures S2A-C).
To examine their respective cytokine response profiles, M(IFNγ/LPS) and M(4/13) were stimulated with killed M. tuberculosis strain H37Rv (hereafter referred to as Mtb) or the TLR1/2 ligand tripalmitoyl-S-glyceryl-cysteine (PAM3CSK4) on either Day 0 or Day 6. Acutely activated M(IFNγ/LPS) demonstrated elevated secretion of TNFα, IL-6 and IL-10 in response to Mtb (Figure S2D) and elevated TNFα and IL-6 in response to PAM3CSK4 (Figure S3A), compared with M(-). By contrast, M(4/13) had attenuated secretion of each cytokine compared with M(-), following either Mtb or PAM3CSK4 stimulation. However, on Day 6 (innate memory responses), both M(IFNγ/LPS) and M(4/13) secreted greater pro-inflammatory TNFα and reduced anti-inflammatory IL-10 in response to Mtb (Figure 2A), and secreted elevated TNFα following PAM3CSK4 stimulation (Figure S3B). M(4/13) additionally exhibited increased IL-6 secretion in response to both stimuli. As with the BCG infection, M(IL-4/13) responses changed markedly between Day 0 and Day 6, shifting towards a pro-inflammatory response profile which was similar to classically activated macrophages.
To consider whether the shift in M(4/13) to a pro-inflammatory cytokine profile was dependent on epigenetic changes, the DNA methylation inhibitor 5ʹ-deoxy-5ʹ-(methylthio)adenosine (MTA), was employed. This inhibitor has been demonstrated to impede training by BCG (Kleinnijenhuis et al., 2012) and β-glucan (Quintin et al., 2012) – which promote pro-inflammatory responses – as well as training induced by helminth Fasciola hepatica total extract (Quinn et al., 2019), which enhances anti-inflammatory responses such as IL-10 and IL-1Ra secretion. The addition of MTA prior to activation with IL-4 and IL-13 on day -1 reduced TNFα, IL-6 and IL-10 secretion induced by either Mtb or PAM3CSK4 on Day 6 (Figure S3C), resulting in a profile reminiscent of acutely activated M(4/13) (Figure S2D). By contrast, inhibition of DNA methylation in M(-) resulted in comparable or even increased cytokine secretion. This suggested that DNA methylation following IL-4 and IL-13 activation contributed to the innate training and subsequent enhancement of pro-inflammatory responses.
Next, we addressed whether the shift towards pro-inflammatory responses was applicable to other stimuli, trained M(4/13) were incubated with various Toll-like receptor (TLR) agonists on Day 6 (Figure 2B). Incubation of cells with ligands for TLR1/2, TLR4, TLR7/8 or TLR 9 resulted in an increase of TNFα production, reduced IL-10 secretion or both. This demonstrated that prior activation with IL-4 and IL-13 caused a subsequent shift towards a pro-inflammatory response profile in response to a range of pathogen-related agonists.
Having considered cytokine responses, the expression of Arg1 and Nos2 and the secretion of NO were next addressed. In mice, the induction of iNOS is important for bactericidal NO production and subsequent killing of M. tuberculosis (Flynn et al., 1993; Pasula, Martin, Kesavalu, Abdalla, & Britigan, 2017). In turn, Arg1 directly impedes the bactericidal function of iNOS by sequestering the amino acid, arginine, which each enzyme uses to make their respective products: ornithine and NO. Subsequently, in the murine model of TB infection, Arg1 has been linked with increased bacterial burden and pathology (Moreira-Teixeira et al., 2016).
Without a secondary stimulation on Day 6, M(4/13) and M(IFNγ/LPS) displayed higher levels of Arg1 and Nos2 expression, compared with BMDMs incubated in media alone (Figure 2C), and M(IFNγ/LPS) exhibited markedly higher expression of Nos2 compared with M(4/13). Following stimulation with Mtb (Figure 3C) or PAM3CSK4 (Figure S3D) M(IFNγ/LPS) exhibited elevated expression of both Arg1 and Nos2 compared with untrained M(-), however the trained M(4/13) exhibited equivalent expression of Arg1, but with enhanced Nos2, demonstrating a further shift towards an M1 profile. An increase in NO production could not be detected following Mtb stimulation; however, following LPS stimulation, both M(4/13) and M(IFNγ/LPS) secreted greater levels of NO compared to M(-) (Figure 2D).
Other M1 and M2 markers were analyzed by flow cytometry, following incubation with media, Mtb or PAM3CSK4 on Day 6 (Figure 2E). Without secondary stimulation, M(4/13) exhibited an M1-typical profile: heightened expression of CD80 and MHC II, whilst CD206 expression was comparable to M(-). Following secondary stimulation with either Mtb or PAM3CSK4, M(4/13) CD80, CD206 and MHC II expression was significantly greater compared with M(-). Having observed a change in activation markers, cytokine induction and bactericidal capacity in M(4/13) between Days 0 and 6, the next question was whether there was a corresponding shift in glycolytic metabolism.
BMDMs Trained with IL-4 and IL-13 Retain OXPHOS Metabolism
The increased energy and biosynthetic precursor demand induced by various macrophage activation states are met in distinct ways. Consistent with previous studies, M(4/13) on Day 1 displayed both an increase of oxygen consumption rate (OCR) – indicative of mitochondrial OXPHOS activity – and extracellular acidification rate (ECAR) – an indirect measurement of lactic acid secretion and thus indicative of glycolysis (Figure 3A). This demonstrated how alternative activation is intrinsically linked with enhanced mitochondrial OXPHOS activity, via the tricarboxylic acid (TCA) cycle (Van den Bossche et al., 2016; Wang et al., 2018). The TCA cycle is in turn driven by glycolysis, glutaminolysis and fatty acid oxidation (Viola, Munari, Sánchez-Rodríguez, Scolaro, & Castegna, 2019). By contrast, acutely activated M(IFNγ/LPS) displayed an increase in ECAR and reduced OCR (Figure 3A), which was indicative of augmented glycolysis to meet the increased need for ATP, whilst ATP synthesis via OXPHOS is hindered (Liu et al., 2016). This glycolytic shift is critical for pro-inflammatory responses induced by classically activated macrophages and moreover results in mitochondrial dysfunction (Van den Bossche et al., 2016).
Due to the link between glycolysis and classical activation, it is not surprising that prior work has proposed that a macrophage glycolytic shift is crucial for effective killing and thus overall control of TB infection (Gleeson et al., 2016; Huang et al., 2018). Furthermore, it has recently been proposed that M. tuberculosis impedes this glycolytic shift as an immune evasion strategy (Hackett et al., 2020). As we had observed that innate memory responses induced by IL-4 and IL-13 caused a switch towards a pro-inflammatory and bactericidal response profile, akin to a classically activated macrophage, it was pertinent to address whether there was an accompanying glycolytic shift. Previous work by Van den Bossche et al. has highlighted that because IL-4 activated human MDMs retain their metabolic versatility, they are able to be “re-polarized” to a classical phenotype (Van den Bossche et al., 2016). This was demonstrated by activating MDMs for 24 hours with IL-4, before the cells were washed and re-stimulated with IFNγ and LPS for another 24 hours. MDMs previously activated with IL-4 secreted higher concentrations of TNFα, IL-6 and IL-12 compared with MDMs that were naïve prior to IFNγ and LPS stimulation. This adaptive quality to re-polarize is restricted to alternatively activated macrophages, as mitochondrial dysfunction in classically activated macrophages prevents them from re-polarizing to an alternative phenotype (Van den Bossche et al., 2016). Subsequently, we next addressed whether the IL-4 trained macrophages were adopting a metabolic profile similar to a classical macrophage, or retaining their more versatile metabolic profile.
As with the BCG infection studies, due to the reduced viability of the M(IFNγ/LPS) by Day 6 and 7, OCR and ECAR measurements did not reach the detection limit. To compare metabolic phenotypes of the trained macrophages at these later time points, transcription profiles were therefore analyzed. Prior work has established that the transcription factor hypoxia-inducible factor-1 alpha (HIF-1α) aids the glycolytic metabolic shift following classical activation, such as upregulating the enzyme lactate dehydrogenase (LDH) (Seth et al., 2017) to increase lactic acid fermentation. Tallying with these studies, trained M(IFNγ/LPS) on Day 7 exhibited elevated expression of LDH (Ldha) and pyruvate kinase isozyme M2 (PKM2, Pkm2) (Figure S4A). PKM2 has been shown to play a key role in stabilizing HIF-1α and is thus a crucial determinant for glycolytic metabolism re-wiring (Palsson-McDermott et al., 2015). By contrast, on Day 7, M(4/13) displayed a similar level of LDH expression and reduced transcription of PKM2 compared with M(-).
On the other hand, considering OXPHOS machinery at this time point, trained M(IFNγ/LPS) showed reduced expression of the TCA cycle enzyme succinate dehydrogenase (SDH, Sdha), whereas this downregulation was not present in M(4/13) (Figure S4A). Moreover, M(4/13) and M(IFNγ/LPS) respectively displayed increased and decreased expression of the transcription factor myelocytomatosis viral oncogene (c-Myc, Myc). IL-4 and IL-13 induce c-Myc expression in alternatively activated macrophages (L. Li et al., 2015; Luiz et al., 2020) and contrastingly LPS induced upregulation of HIF-1α occurs in tandem with downregulation of c-Myc (Liu et al., 2016). Overall, without secondary stimulation, both trained M(IFNγ/LPS) and M(4/13) retained transcriptional and metabolic profiles consistent with previous reports.
Upon secondary stimulation on Day 6 with either Mtb (Figure 3B) or PAM3CSK4 (Figure S4B), M(IFNγ/LPS) maintained the elevated expression of Pkm2, accompanied by reduced transcription of Sdha and Myc. Myc expression was induced in untrained M(-) and M(4/13) by both secondary stimuli, although Mtb did not enhance Myc in M(4/13) to the same extent as M(-). Regarding glycolytic metabolism, M(4/13) displayed relatively reduced Ldha upon Mtb incubation (Figure 3B) and further maintained a similar level of Sdha compared with M(-). These results indicated that trained M(IFNγ/LPS) and M(4/13) maintained their respective metabolic profiles following secondary stimulation with Mtb or PAM3CSK4. As such, these results indicated again that trained M(IFNγ/LPS) and M(IL-4/13) maintained their respective metabolic profiles following secondary stimulation with Mtb or PAM3CSK4.
That trained M(4/13) retain OXPHOS metabolism was furthermore supported by Mtb stimulation increasing both OCR and ECAR in M(-) and M(4/13) (Figure 3C), where M(4/13) displayed significantly greater OXPHOS driven basal respiration, ATP production and concomitant proton leakage compared with M(-) (Figure 3D). In addition, in response to either Mtb (Figure 3E) or PAM3CSK4 (Figure S4C) there was an increase in spare respiratory capacity (SRC), further signifying that trained M(4/13) retain and even elevate OXPHOS metabolism following secondary stimulation with Mtb.
To further investigate the metabolic profile of trained M(IL-4/13) vs untrained (media control) stimulated with Mtb, the relative change of intracellular metabolite abundance was furthermore assessed by liquid chromatography-mass spectrometry (LC-MS, Figures 3F-G and S4D-E). This semi-targeted analysis revealed that the trained M(4/13) appeared to have increased use of the urea cycle, as shown by reduced aspartate, arginine and argininosuccinate levels along with enhanced ornithine (Figure S4E). Furthermore, there were reduced levels of metabolites associated with glycolysis and the pentose phosphate pathway (PPP) – such as lactate and sedoheptulose-7-phosphate – whilst metabolites involved in fatty acid oxidation (FAO) and ATP synthesis regulation were enhanced (Figure 3G). In the case of FAO, reduced carnitine with enhanced carnitine-bound fatty acids (acyl carnitines) signified that fatty acids were being ferried by carnitine into the mitochondrion for FAO. FAO is upregulated during murine M2 macrophage activation and results in the production of acetyl-CoA, NADH and FADH2, which are further used to fuel the TCA cycle and downstream OXPHOS (O’Neill, Kishton, & Rathmell, 2016). Furthermore, the upregulation of creatine and phosphocreatine – as well as creatine biosynthesis precursor guanidinoacetate – intimated enhanced ATP synthesis. Creatine reacts with ATP to form ADP and phosphocreatine, transporting ATP out of the mitochondrion and preventing allosteric inhibition of ATP synthesis. In the trained M(4/13) the amount of phosphocreatine per ATP was enhanced (Figure S4F), which indicated enhanced mitochondrial ATP synthesis and supported the hypothesis that OXPHOS was enhanced. Overall, these results suggested the trained M(4/13) were upregulating their OXPHOS activity in response to mycobacterial challenge. Having observed that trained M(IL-4/IL-13) showed a distinct metabolic profile from classically activated macrophages, the roles of glycolysis and OXPHOS in driving cytokine production were next investigated with the use of inhibitors. Incubation with the glycolysis and OXPHOS inhibitor 2-deoxy glucose (2-DG) (Wang et al., 2018) prior to stimulation with Mtb (Figure S5A) significantly reduced TNFα secretion in M(IFNγ/LPS) and M(-). 2-DG also reduced PAM3CSK4 induced TNFα in M(4/13) and M(IFNγ/LPS) (Figure S5B) and reduced IL-6 secretion in M(4/13) following either Mtb or PAM3CSK4 stimulation. Secretion of anti-inflammatory IL-10 following Mtb stimulation was also reduced by 2-DG in M(IFNγ/LPS) and M(4/13). This demonstrated that glycolysis, OXPHOS or both were involved in driving cytokine responses to Mtb and PAM3CSK4 in all tested macrophages.
To further examine the role of OXPHOS, the ATP-synthase inhibitor oligomycin (OM) and the glutaminase inhibitor bis-2-(5-phenylacetamido-1,3,4-thiadiazol-2-yl)ethyl sulfide (BPTES) were used prior to secondary stimulation. The role of glutaminolysis in the trained M(IL-4/13) was of interest, as it has previously been highlighted to compensate for inhibition of glycolysis and fuel the TCA cycle during IL-4 induced macrophage activation (Wang et al., 2018). In the trained M(4/13), inhibition of either OXPHOS or glutaminolysis significantly reduced TNFα and IL-6 secretion following stimulation with either Mtb (Figure 4A) or PAM3CSK4 (Figure S5B), suggesting that both processes helped drive pro-inflammatory responses. By contrast, in trained M(IFNγ/LPS), although the markedly lower level of IL-6 was further impeded by either inhibitor, the secretion of TNFα was unaffected, supporting M(IFNγ/LPS) use of glycolysis to drive pro-inflammatory responses. OM and BPTES furthermore impeded IL-10 secretion from M(IFNγ/LPS), whereas OM reduced IL-10 secretion in M(4/13) and BPTES did not (Figure 4A). In the case of trained M(4/13) this implicated that glutamine metabolism was selectively a driver of pro-inflammatory cytokine responses.
To further cement the differences in glycolytic metabolism between trained M(IFNγ/LPS) and M(4/13), an experiment was carried out where BMDMs were incubated in either glucose depleted media or regular glucose-rich media from Day-1 (Figure 4B). Whereas glucose depletion significantly reduced the secretion of TNFα, IL-6 and IL-10 from M(IFNγ/LPS) and M(-), following Mtb stimulation on Day 6, glucose depletion did not impair pro-inflammatory cytokine secretion of M(4/13); instead the secretion of both TNFα and IL-6 was elevated under these conditions, whilst levels of regulatory IL-10 were further reduced, displaying an even greater shift towards pro-inflammatory responses. Furthermore, stimulation with PAM3CSK4 yielded similar results (Figure S5C), although IL-6 secretion remained unaltered. This highlighted that although trained M(IFNγ/LPS) and M(4/13) had similar response profiles, M(4/13) appeared to retain M2-typical metabolism (Figure 5).
IL-10 Negatively Regulates IL-4 and IL-13 Induced Training
A key consideration regarding alternative macrophage activation and TB infection is the influence of concurrent parasitic disease. Along with IL-4 and IL-13, parasites can induce production of the regulatory cytokine IL-10 (Gause, Wynn, & Allen, 2013; Roy et al., 2018), which moreover promotes alternative macrophage activation (Bystrom et al., 2008; Mantovani et al., 2004). Furthermore, it has been shown that products of the helminth F. hepatica can train macrophages to secrete more IL-10 following secondary LPS stimulation (Quinn et al., 2019). Given its association with alternative macrophage activation and parasitic infection, we next considered the potential role of IL-10 in IL-4 induced macrophage innate memory responses.
BMDMs were activated with IL-4, IL-13 and IL-10 (M(4/13/10)) to compare how this phenotype may have differed from M(4/13). Following acute activation, M(4/13/10) had similarly elevated levels of Arg1 as M(4/13) and a comparably minor induction of Nos2 (Figure 6A). The addition of IL-10 moreover caused even greater augmentation of M2-associated Chil3 and Retnla expression. Accompanying flow cytometry analysis demonstrated that both M(4/13/10) and M(4/13) exhibited reduced expression of CD80, compared with M(-), along with enhanced expression of CD206, where the M(4/13/10) had greater CD206 expression (Figure S6A). Regarding MHC II, only M(4/13) displayed enhanced expression compared with M(-). The similar upregulation of M2-characteristic markers indicated that M(4/13) and M(4/13/10) were two types of alternatively activated macrophages.
A similar comparison was made at Day 7, where both M(4/13) and M(4/13/10) retained enhanced Arg1, Retnla and Chil3 expression and displayed enhanced Nos2 expression compared with naïve M(-) (Figure S6B). Regarding accompanying CD80, CD206 and MHC II expression (Figure S6C): in all tested conditions, CD206 expression was comparable between M(4/13) and M(4/13/10), whereas CD80 expression was elevated in M(4/13/10). Furthermore, M(4/13) retained an elevated expression of MHC II, which increased two- and four-fold following stimulation with Mtb and PAM3CSK4, respectively, whereas this elevation did not occur to the same degree in M(4/13/10).
Our next question was how the addition of IL-10 during alternative activation would affect mycobacterial killing capacity. The BMDMs were incubated with roughly 30 BCG per cell on Day 0 or Day 6. There were some differences in bacterial uptake, where M(4/13) took up more bacteria per cell on Day 6, leading to an MOI of 1-4 (Figure 6B). CFU counts (Figure S1C) were used to calculate killing of internalized BCG. On Day 0, M(-), M(4/13) and M(4/13/10) displayed comparable killing capacity (Figure 6C). Moreover, BCG infection did not induce detectable levels of TNFα or IL-10 (Figure S1D). On Day 6, M(-) and M(4/13/10) maintained a similar level of BCG killing, whereas M(4/13) displayed enhanced bactericidal capacity. This indicated that IL-10 hindered the enhanced bactericidal response induced by prior IL-4 and IL-13 macrophage activation. The difference in bacterial killing capacity was further supported by confocal microscopy 27 hours after infection, where it was observed that M(-) and M(4/13/10) had markedly more bacteria per cell compared with M(4/13) (Figure 6D). Furthermore, whereas M(4/13) showed a shift towards pro-inflammatory cytokine secretion, with IL-10 secretion being reduced whilst TNFα production was largely intact, M(4/13/10) secreted significantly less TNFα than M(4/13), whilst IL-10 secretion was not enhanced (Figure 6E). This would indicate that IL-10 regulated the training induced by IL-4 and IL-13, preventing both the enhancement of killing capacity and the accompanying pro-inflammatory cytokine profile.
This difference in cytokine response profile was confirmed by carrying out similar experiments with Mtb. Whilst M(4/13), secreted lower concentrations of TNFα, IL-6 and IL-10 than M(-) on Day 1 (Figure S6D), but elevated TNFα and IL-6 by Day 7 (Figure 6F), while inflammatory cytokine secretion by M(4/13/10) was reduced on both Day 1 and Day 7. Furthermore, whilst both M(4/13) and M(4/13/10) displayed similar expression of Arg1 following Mtb stimulation on Day 7, the addition of IL-10 during activation hindered the upregulation of Nos2 observed in trained M(4/13) (Figure S6E). Overall, IL-10 appeared to impede the enhanced pro-inflammatory and bactericidal capacity induced by IL-4 and IL-13 innate training.
Discussion
Although “classical” and “alternative” activation are used to describe the two extremes of macrophage polarization, the reality is a broad spectrum of activation states, affected by a multitude of signals. Moreover, growing evidence has shown that the macrophage population during M. tuberculosis infection is highly heterogeneous, and thus elucidating which subtypes best contain the bacterium is critical for understanding disease control (A. Khan, Singh, Hunter, & Jagannath, 2019). Alternative macrophage activation is moreover relevant to TB pathology, both because alveolar macrophages – the initial hosts of M. tuberculosis upon infection – are biased towards alternative activation (Huang, Nazarova, & Russell, 2019), and also considering the influence of concurrent parasitic infection. The current study shows that alternative macrophage activation stimuli impact innate immune memory at least in part via epigenetic modification (Figure S3C); an additional dimension to the noted diversity among macrophage populations.
Whilst acute alternative macrophage activation has been demonstrated to lead to reduced control of M. tuberculosis growth (Kahnert et al., 2006), resulting in greater bacterial burden (Moreira-Teixeira et al., 2016; Orecchioni et al., 2019), there have been some conflicting data concerning the interplay between parasites and mycobacterial infections. Parasitic infection has been shown to enhance mycobacterial bacterial burden in vivo (Monin et al., 2015; Potian et al., 2011) and enhance human TB pathology (Amelio et al., 2017; Mabbott, 2018), but has in some cases also been shown to enhance protection against mycobacterial infection (Aira, Andersson, Singh, McKay, & Blomgran, 2017; O’Shea et al., 2018). Furthermore, discrepancies have been highlighted specifically regarding macrophage activation. In a model of M. tuberculosis macrophage infection in vitro, prior incubation (48hr) with antigens from Hymenolepis diminuta, Trichuris muris and Schistosoma mansoni resulted in alternative macrophage activation, but only incubation with the H. diminuta and T. muris antigens resulted in enhanced mycobacterial growth (Aira et al., 2017). Moreover, the increased growth was accompanied by enhanced secretion of IL-10. This however was not surprising, given that IL-10 has been demonstrated to prevent phagolysosome maturation in human macrophages (O’Leary et al., 2011) and promote TB disease progression in mice (Beamer et al., 2008). Herein, an additional mechanism is proposed, as IL-10 appears to be a key determinant of alternative macrophage training and subsequent control of mycobacterial challenge. BMDM activation with IL-4 and IL-13, with or without the addition of IL-10, resulted in similar activation states: comparable expression of Arg1, Retnla and Chil3 on Day 0 (Figure 6A) and 7 (Figure S6B), reduced expression of CD80 and upregulation of CD206 on Day 0 (Figure S6A), as well as reduced secretion of cytokines TNFα, IL-6 and IL-10 following stimulation with Mtb on Day 0 (Figure S6D). Whilst this hyporesponsive profile was maintained in the M(4/13/10), the trained M(4/13) on the other hand demonstrate enhanced pro-inflammatory and bactericidal mechanisms in response to mycobacterial challenge a week after initial activation (Figure 6). This intimated that IL-10 was a negative regulator of innate training induced by IL-4 and IL-13 and supports that distinct cytokine mediated modes of alternative macrophage activation differ regarding their innate memory programming.
The trained M(4/13) in response to mycobacterial challenge adopted a phenotype similar to classically activated BMDMs: a skewing towards pro-inflammatory cytokine secretion with concomitant increased expression of Nos2 and NO production (Figure 2), as well as enhanced mycobacterial killing capacity (Figure 1). Classical macrophage activation is intrinsically linked with a glycolytic shift in metabolism; it is thus not surprising that prior work has identified an enhancement of glycolysis as critical for efficient M. tuberculosis killing (Gleeson et al., 2016; Huang et al., 2018) and that impeding glycolysis attenuates macrophages ability to control TB infection (Hackett et al., 2020). Moreover, upon M. tuberculosis infection, macrophages appear to undergo a biphasic metabolic profile, where they switch from an initial increase in glycolytic metabolism to an enhancement of the TCA cycle and OXPHOS; a switch which allows mycobacterial survival and disease progression (Shi et al., 2019). As such, it was expected that the trained M(4/13) would shift towards glycolytic metabolism. However, this was not the case, as Mtb stimulation resulted in enhanced OCR and SRC, indicating the use and upregulation of OXPHOS (Figure 3). Furthermore, LC-MS analysis of metabolites indicated increased use of FAO, as well as enhanced creatine, further supporting the continued use of M2-typical metabolism. This was moreover confirmed with the use of the ATP synthase inhibitor oligomycin, where it was demonstrated that inhibition of OXPHOS reduced trained M(4/13) cytokine secretion, whereas by contrast M(IFNγ/LPS) TNFα upregulation was unaffected. Furthermore, when trained M(4/13) were incubated in a glucose depleted environment during the week preceding secondary stimulation, the glucose depletion enhanced the pro-inflammatory shift: increased TNFα and IL-6 secretion combined with reduced IL-10 production (Figure 4B). This was a stark contrast to the effect of glucose depletion on untrained M(-) and trained M(IFNγ/LPS), where the secretion of all measured cytokines was attenuated. This contrast between M(4/13) and M(IFNγ/LPS) cemented their differences regarding metabolic dependency on glucose.
It is of note that the addition of a glycolysis inhibitor reduced cytokine secretion, from M(-), M(IFNγ/LPS), as well as M(4/13) following secondary stimulation (Figures S5A-B). That inhibition of glycolysis impeded pro-inflammatory cytokine secretion, but glucose depletion did not, indicated that glycolysis in trained M(4/13) was fueling the TCA cycle and downstream OXPHOS, and upon glucose depletion other pathways, such as glutaminolysis, were compensating for this loss. Glutaminolysis has been shown to be upregulated following macrophage activation with IL-4 and has been demonstrated to compensate for glycolysis inhibition in driving the TCA cycle and OXPHOS (Wang et al., 2018). In the present study, it was moreover observed that the use of a glutaminolysis inhibitor reduced TNFα and IL-6 secretion following secondary stimulation in trained M(4/13) specifically, and not in untrained M(-) or M(IFNγ/LPS) (Figure 4A). It should be noted that prior work has identified differences in macrophage metabolic profiles in response to infection with live M. tuberculosis compared with the killed bacterium or infection with attenuated BCG (Cumming, Addicott, Adamson, & Steyn, 2018). However, the consensus is that classic macrophage activation, and an accompanying shift to glycolytic metabolism, is paramount for effective mycobacterial killing and propagation of inflammatory responses. It is surprising therefore that these results summarily intimated that trained M(4/13) maintain OXPHOS to fuel anti-mycobacterial responses.
The metabolic profile in trained M(4/13), as summarized in Figure 5, is not only distinct from classically activated macrophages, but also from that seen with other training stimuli, such as β-glucan. Innate training is being tested as a means to bolster innate immunity against M. tuberculosis (Khader et al., 2019) and it was recently reported that training with β-glucan was protective against subsequent M tuberculosis infection, as seen by enhanced human monocyte pro-inflammatory cytokine secretion and increased mouse survival in vivo (Moorlag et al., 2020). Similar to classical macrophage activation, β-glucan training results in a glycolytic shift, as demonstrated in human monocytes (Cheng et al., 2014). As such, the phenotype of the trained M(4/13) is distinct from other macrophage phenotypes previously demonstrated to protect against TB, and thus offers an additional avenue for future research regarding strategies for combatting this disease.
Van den Bossche et al. in 2016 highlighted a key adaptive distinction between classical and alternative macrophage activation: that alternatively activated macrophages can be “re-polarized” due to their metabolic versatility, whereas the mitochondrial dysfunction which occurs during classical macrophage activation prevents such reprogramming (Van den Bossche et al., 2016). In the current study, an additional adaptive capacity is identified: that activation with IL-4 and IL-13 programs BMDMs to better respond to a mycobacterial challenge, whilst crucially retaining their metabolic diversity.
In conclusion, our work presents mechanistic insight into how innate training via IL-4 and IL-13 can enhance macrophage pro-inflammatory responses and mycobacterial killing. This unexpected finding shows how macrophage plasticity belies the usual M1-M2 dichotomy and provides a new framework to explore the impact of comorbidities, such as parasitic infections, on TB disease burden.
Materials and Methods
Experimental Model and Subject Details
Animals
Mice used for primary cell isolation were eight to 16-week-old wild-type C57BL/6 mice that were bred in the Trinity Biomedical Sciences Institute Bioresources Unit. Animals were maintained according to the regulations of the Health Products Regulatory Authority (HPRA). Animal studies were approved by the TCD Animal Research Ethics Committee (Ethical Approval Number 091210) and were performed under the appropriate license (AE191364/P079).
Cell Isolation and Culture
Bone marrow-derived macrophages (BMDMs) were generated as described previously by our group.(Lebre, Hanlon, Boland, Coleman, & Lavelle, 2018) Briefly, bone marrow cells were extracted from the leg bones and were cultured in high glucose DMEM, supplemented with 8% v/v fetal bovine serum (FBS), 2 mM L-glutamine, 50 U ml-1 penicillin, 50 μg ml-1 streptomycin (hereafter referred to as complete DMEM [cDMEM]). Cells were plated on non-tissue cultured treated petri dishes (Corning) and supplemented with 25% v/v of L929 cell line conditioned medium containing macrophage colony-stimulating factor (M-CSF) on day -8. Fresh medium was added on day -5 and on day -2 adherent cells were detached by trypsinization and collected. Unless specified otherwise, for acute/polarization studies, BMDMs were seeded in 12-well plates at 0.9 × 106 BMDMs per well, and for training studies, BMDMs were seeded in 24-well plates at 0.2 × 106 BMDMs per well.
On day -1 BMDMs were cultured with medium (naïve/untrained) or activated with 25 ng ml-1 IFNγ with 10 ng ml-1 LPS (classical activation), or 40 ng ml-1 IL-4 with 20 ng ml-1 IL-13, with or without 40 ng ml-1 IL-10 (alternative activation) – all included 15% L929 conditioned media. After 24 hours, the supernatant was replaced with fresh media (with 10% L929 conditioned media). For acute activation/polarization studies, the BMDMs were left to rest for 2 hours before experiment. For training studies, the BMDMs were left to rest, fed on day 3 (media supplemented with 7% L929 conditioned media) and experiments were carried out on day 6. The amount of L929 conditioned media added on each day was consistent for all experiments.
BCG Infection
Mature BMDMs were seeded in 24-well plates at 0.5 × 106 cells (acute activation) or 0.2 × 106 cells (training) per well on day -2. For imaging of internalized BCG in untrained and trained BMDMs, 0.2 × 106 BMDMs were seeded on circular glass coverslips (placed in 24 well plates, one coverslip per well) that had been previously treated with sodium hydroxide to aid attachment.
BMDMs were incubated with media (naïve) or activated as outlined above on day -1, for 24 hours before BMDMs were washed and fresh media was added. For acute activation studies, BMDMs were infected with BCG Denmark minimum 3 hours later. For training studies, BMDMs were fed on day 3 and infected with BCG Denmark on Day 6.
Regarding infection dose (BCG per cell), for infection in acutely activated BMDMs, cell number per well was assumed to be 0.5 × 106. For infecting trained BMDMs, three extra wells were prepared for all conditions and BMDMs were removed from these wells by trypsinization on day 5, pooled and counted in triplicate. Prepared BCG single cell suspension (see below) was diluted to reach an intended infection dose of 5 BCG per cell (as measured by OD600, where 0.1 is estimated to be 10 × 106 bacteria ml-1). Each infection dose was measured via CFU counts (see below) and was thus subsequently corrected to an actual infection dose of roughly 30 BCG per cell.
3 hours post infection the supernatant was removed and the infected BMDMs were washed twice with Dulbecco’s Phosphate Buffered Saline (PBS) to remove extracellular bacteria, and fresh media was added.
To measure cytokine secretion at each time point (3-, 27- and 51 hours post infection), supernatant was collected and filtered (polyethersulfone [PES] 0.22 μm Luer lock syringe filter; Millex-GP) to remove BCG, prior to specific cytokines being measured by ELISA.
For CFU counts, at each time point the media was removed, wells were washed once with PBS before the BMDMs were lysed with 0.05% Tween20 in water. The cell lysate was diluted in pre-warmed (37 °C) 7H9 media, and 50 μl of dilutions 10-2-10-6 were spread onto enriched 7H11 agar plates by dotting. Agar plates were incubated at 37 °C for 6 weeks and colonies were counted once a week from week 2 and on. CFU counts 3 hours gave the multiplicity of infection (MOI).
For imaging internalized BCG (wells containing glass coverslips), media was removed, BMDMs were washed with PBS and then fixed with 4% paraformaldehyde (in PBS) overnight. The paraformaldehyde was then removed and coverslips were stored in PBS.
Stimulation Experiments
Naïve (media control) or activated BMDMs were incubated with media or secondary stimuli on day 0 (acute activation/polarization) or day 6 (training). Secondary stimuli: 150 μg ml-1 gamma-irradiated whole cells of Mycobacterium tuberculosis strain H37Rv (concentration measured by OD600, where 100 mg ml-1 was 0.32), 35 ng ml-1 PAM3CSK4, 25 ng ml-1 LPS, 5 μg ml-1 Poly I:C, 0.55 μg ml-1 R848, 10 μg ml-1 CpG.
For histone methylation inhibition: on day -2 BMDMs were incubated with MTA (final concentration 1 mM) one hour prior to media incubation or activation as outlined above.
For metabolic inhibition experiments: on day 6 BMDMs were pre-incubated with 1mM 2-DG for 3 hours, 2 μM oligomycin or 10 μM BPTES 1 hour prior to media incubation or secondary stimulation.
For glucose depletion training experiment: untrained or trained BMDMs were incubated with either the regular high glucose cDMEM or glucose depleted cDMEM, with DMEM devoid of glucose (Gibco), from day -2 until incubation with media or secondary stimulation on Day 6 (fed day 3 with the same media as given on day -2). Note, the glucose depleted cDMEM included 8% v/v FBS and the same amount of L929 conditioned media as outlined previously, both of which provided a source of glucose.
Seahorse
All real-time measurements of oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured by using Seahorse system: Seahorse XFe96 Analyzer (Agilent). The analysis used was a Mitochondrial Stress Test (using a standard Agilent Seahorse protocol).
On day -2, mature BMDMs were seeded in a 96-well Seahorse plate (Agilent): 100,000 cells per well for acute activation studies or 30,000 cells per well for training studies. BMDMs were activated on day -1 as outlined and analyzed on either day 0 (24 hours after activation) or day 6 (fed on day 3), with or without secondary stimulation on day 5 (training studies).
The day before analysis, a Seahorse Cartridge (Agilent) was incubated overnight at 37 °C (incubator devoid of CO2). The following day, the sterile water was replaced with Calibration Fluid at pH 7.4 incubated as before for 90 minutes before the analysis.
An hour before the analysis, in the Seahorse plate regular cDMEM was replaced with Seahorse XF DMEM, supplemented with 10 mM glucose, 1 mM pyruvate and 2 mM L-glutamine. The BMDMs were then incubated for an hour at 37 °C (incubator devoid of CO2).
15 minutes before the analysis, the cartridge was removed and reagents/inhibitors were added to their respective ports: oligomycin (port A, final concentration 10 μM), FCCP (port B, final concentration 10 μM), rotenone and antimycin-A (port C, final concentrations 5 μM each).
The analysis was carried out according to the manufacturer’s instructions. Recorded values less than zero were disregarded.
For training experiments, protein concentration was used to standardize the OCR and ECAR measurements. Supernatant was removed, cells were washed once in PBS and 10 μl of RIPA buffer was added per well. After pipetting up and down and scraping, the buffer was collected and samples from the same condition were pooled. 20 μl of pooled solution was used to measure protein concentration, using the Pierce bicinchoninic acid (BCA) assay kit (Thermo Scientific) according to the manufacturer’s instructions (microplate procedure). Absorbance was measured at 560 nm.
Metabolomics (LC-MS) sample preparation
On day -2, BMDMs were seeded in 6-well plates: 1 × 106 cells/well and activated on day -1 with IL-4/13 as outlined or incubated with media. BMDMs were fed on day 3 and incubated with irradiated M. tuberculosis on Day 6 for 24 hours. Prior to metabolite extraction, cells were counted using a separate counting plate prepared in parallel and treated exactly like the experimental plate. Supernatant was removed and cells were washed once in PBS. After aspiration, the BMDMs were kept at -80 °C or on dry ice. Metabolites were extracted by adding chilled extraction buffer (500 μl/1 × 106 cells), followed by scraping (carried out on dry ice). Buffer was transferred to chilled eppendorf tubes and shaken in a thermomixer at maximum speed (2000 rpm) for 15 minutes at 4 °C. Following centrifugation at maximum speed for 20 minutes, roughly 80% of supernatant was transferred into labelled LC-MS vials, taking care to avoid pellet and any solid debris.
Flow Cytometry
On day -2, 0.8 × 106 mature BMDMs were seeded on non-tissue cultured treated 35 mm petri dishes (Corning) and activated as previously outlined on Day -1. The naïve or activated/trained BMDMs were harvested at two separate time points: day 0 (24 hours post activation for acute activation characterisation) or day 6 (fed on day 3), with or without secondary stimulation on day 5 as specified in figure legends (training experiments).
For analysis, BMDMs were placed on ice for 30 min before harvesting with PBS-EDTA (5 mM) solution by gently pipetting up and down and transferring to flow cytometry tubes. Cells were incubated with Fixable Viability Stain 510 at RT for 15 min. After washing with PBS, cells were stained with anti-mouse Fc block, 15 minutes prior staining with CD80-FITC, CD206-PE, F4/80-PerCP-Cy5.5, CD11b-APC-eFluor 780, MHC class II-eFluor 450 for an additional 30 min at 4 °C. The cells were washed with PBS and resuspended in flow cytometry buffer (1% FBS in PBS). Samples were acquired on a BD Canto II flow cytometer and the data was analysed by using FlowJo software.
Method Details
BCG Preparation and Plating
Bacille Calmette-Guérin (BCG) Denmark (OD600 0.1) was incubated in 7H9 media at 37 °C with rotation until the OD600 0.5-0.8 (logarithmic growth phase) was reached.
Single cell suspension was prepared as follows: bacteria were pelleted by centrifugation and 7H9 media was removed. Bacteria were vortexed with glass beads before pre-warmed (37 °C) DMEM was added and bacteria were left to sediment for 5 minutes. Bacterial suspension was carefully collected (avoiding disruption of the pellet) and centrifuged again to pellet large clumps. Suspension was collected (avoiding the pellet). Suspension was passed through a 26G needle 15 times to disaggregate bacterial clumps.
Bacterial concentration was estimated by OD600 (where 0.1 was estimated as 10 × 106 bacteria ml-1) and infection dose was confirmed by colony forming unit (CFU) counts.
7H11 agar plates (for CFU counts) were made up as follows: 10.5 g middlebrook 7H11 powder and 2.5 ml glycerol was dissolved in distilled water to reach 500 ml and autoclaved. Once cooled sufficiently, 50 ml sterile-filtered (0.22 μm; SteriCup [Millipore]) albumin-dextrose-sodium chloride (ADN) enrichment was added. For 1 liter of ADN enrichment: 50 g fatty acid-free, heat-shocked bovine serum albumin, 8.5 g sodium chloride and 20 g glucose were dissolved in sterile water to reach 1 liter.
Staining and Imaging of Internalized BCG
For staining the coverslips contained in 24-well plates, PBS was removed and 17 μM Hoechst 33342 and 7.5 U Phalloidin-Alexa Fluor 647 (in 1 ml PBS) was added to each well. After half an hour (in the dark), the solution was removed and the coverslips were washed with PBS. 200 μl Modified Auramine-O stain was added for 2 minutes (in the dark), after which the stain was removed and the wells were washed with PBS. To quench any extracellular BCG, 200 μl Auramine-O Quencher de-colorizer was added for 2 minutes (in the dark) and then the quencher was removed and the coverslips were washed with PBS. The coverslips were then fixed onto glass slides: 2 μl Vectashield mounting media was used per coverslip and clear nail polish was applied around the edge and allowed to dry completely. The coverslips were stained and imaged the same day.
Images of all samples were obtained using a Leica SP8 confocal microscope, taken with x40 magnification oil objective. Z-stack images through the entire cell were obtained and representative images were taken from these stacks.
Liquid Chromatography coupled to Mass Spectrometry (LC-MS)
Extraction buffer for isolating metabolites: 50% LC-MS grade methanol, 30% LC-MS grade acetonitrile, 20% ultrapure water (internal filtration system), valine-d8 final concentration 5 μM.
Hydrophilic interaction chromatographic (HILIC) separation of metabolites was achieved using a Millipore Sequant ZIC-pHILIC analytical column (5 µm, 2.1 × 150 mm) equipped with a 2.1 × 20 mm guard column (both 5 mm particle size) with a binary solvent system. Solvent A was 20 mM ammonium carbonate, 0.05% ammonium hydroxide; Solvent B was acetonitrile. The column oven and autosampler tray were held at 40 °C and 4 °C, respectively. The chromatographic gradient was run at a flow rate of 0.200 mL/min as follows: 0–2 min: 80% B; 2-17 min: linear gradient from 80% B to 20% B; 17-17.1 min: linear gradient from 20% B to 80% B; 17.1-22.5 min: hold at 80% B. Samples were randomized and analysed with LC–MS in a blinded manner with an injection volume was 5 µl. Pooled samples were generated from an equal mixture of all individual samples and analysed interspersed at regular intervals within sample sequence as a quality control.
Metabolites were measured with a Thermo Scientific Q Exactive Hybrid Quadrupole-Orbitrap Mass spectrometer (HRMS) coupled to a Dionex Ultimate 3000 UHPLC. The mass spectrometer was operated in full-scan, polarity-switching mode, with the spray voltage set to +4.5 kV/-3.5 kV, the heated capillary held at 320 °C, and the auxiliary gas heater held at 280 °C. The sheath gas flow was set to 25 units, the auxiliary gas flow was set to 15 units, and the sweep gas flow was set to 0 unit. HRMS data acquisition was performed in a range of m/z = 70–900, with the resolution set at 70,000, the AGC target at 1 × 106, and the maximum injection time (Max IT) at 120 ms. Metabolite identities were confirmed using two parameters: (1) precursor ion m/z was matched within 5 ppm of theoretical mass predicted by the chemical formula; (2) the retention time of metabolites was within 5% of the retention time of a purified standard run with the same chromatographic method. Chromatogram review and peak area integration were performed using the Thermo Fisher software Tracefinder 5.0 and the peak area for each detected metabolite was normalized against the total ion count (TIC) of that sample to correct any variations introduced from sample handling through instrument analysis. The normalized areas were used as variables for further statistical data analysis.
ELISA
For detection of secreted TNFα, IL-6 and IL-10 from BMDMs, supernatants were collected and cytokines quantified by ELISA according to manufacturers’ instructions (R&D Systems or Biolegend), except antibody and sample volumes were halved.
Nitric oxide (NO) secretion
NO concentrations were quantified by indirect measurement of nitrite (NO2-) via the Griess Reagent System kit (Promega). The assay was carried out as per manufacturer’s instructions and absorbance of 560 nm was measured.
Reverse Transcription Quantitative PCR (rt-qPCR)
RNA was isolated via High Pure RNA Isolation Kit (Roche) according to the manufacturer’s instructions and RNA was eluted in 50 μl water. RNA (minimum 100 ng) was reverse transcribed into complementary DNA (cDNA) with an M-MLV reverse transcriptase, RNase H minus, point mutant, in reverse transcriptase buffer, mixed with dNTPs, random hexamer primers and ribonuclease inhibitor (RNAseOUT). Quantitative PCR was performed using KAPA SYBR® FAST Rox low qPCR Kit Master Mix in accordance with the instructions provided by the manufacturer, using QuantStudio 3 System technology. Primers (Table S1) were designed in Primer BLAST and/or were also checked in Primer BLAST for specificity to the gene of interest. Where possible, primers were chosen to cross an exon-exon junction.
RNA expression was normalized to the internal references β-actin and/or TATAbox-binding protein, from the corresponding sample (Ctgene–Ctreference=ΔCt). Furthermore, ΔCt from control samples were subtracted from the ΔCt of each sample (ΔCttreatment-ΔCtctrl =ΔΔCttreatment). Fold change was calculated as 2(-ΔΔCt). These calculations were carried out using Microsoft Excel.
Quantification and Statistical Analysis
Data were evaluated on either on Prism version 8 for Windows or via R for Metaboanalyst generated analysis. Differences between two independent groups were compared via unpaired Student’s t-test. For BCG infection studies where two independent groups were compared at several time points, multiple t-test analysis was employed, with Holm-Sidak correction for multiple comparisons. Differences were considered significant at the values of * P < 0.05, ** P < 0.01, *** P < 0.001 and **** P < 0.0001.
Data and Code Availability
All data is available at Mendeley Data DOI: 10.17632/ncbph43m85.1.
Author Contributions
M.L.E.L. performed and analyzed all experiments and wrote the paper. M.M., B.S. and S.V.G provided assistance with BCG infection studies. D.G.R, N.C.W. and L.A.J.O. assisted with Seahorse experiments. M.Y. identified and quantified the metabolites via LC-MS and D.G.R and C.F. assisted with metabolomics analysis. M.L.E.L., with the assistance of F.L. and A.L.G, developed the BMDM training protocol. M.L.E.L was supervised by E.M.S. and E.C.L. and the work was supervised by E.C.L.
Declaration of Interests
The authors declare no competing interests.
Supplemental Information
Acknowledgements
The authors would like to thank Dr Gavin McManus for his assistance with the imaging studies.
M. Lundahl was funded by a Trinity College Dublin postgraduate studentship and this work is supported by Science Foundation Ireland (SFI) Research Centre, Advanced Materials and BioEngineering Research (AMBER) under Grant number 12/RC/2278_P2 E, SFI under Grant number 12/IA/1421 and 19FFP/6484 (E. Lavelle) and SFI under Grant number 15/CDA/3310 (E. Scanlan). S. Gordon and M. Mitermite acknowledge funding from Wellcome Trust PhD Studentship 109166/Z/15/A and SFI award 15/IA/3154.