Mice

All studies were performed using female C57Bl/6 mice between 6–9 weeks old. Mice were purchased from Jackson Laboratories and housed in animal biosafety level 3 facilities at Rocky Mountain Laboratories. Mice were housed in sterile cages and provided irradiated food and sterile water ad libitum. Housing was in a specific-pathogen-free facility with 12 hr light/dark cycles. All animal studies were approved by the Rocky Mountain Laboratories Animal Care and Use Committee and performed in compliance with their guidelines (ASP # 2019-054-E).

In vivo studies

Mice were anesthetized using ketamine/xylazine prior to infection. Intranasal and intradermal infections were performed as previously described [14]. Briefly, FTT stocks were diluted in PBS to either 25 CFU/25μL (intranasal) or 25 CFU/10 μl (intradermal). For intranasal infections, inoculum was instilled down the left nare of the anesthetized mouse. For intradermal infection, inoculum was loaded into a tuberculin syringe and approximately 10 μL was injected into the ear pinna of the anesthetized mouse. Mice injected with PBS served as negative (mock) controls. Concentration of inoculums was confirmed by plating on Modified Mueller Hinton (MMH) agar plates followed by incubation and enumeration of colonies. LPS preconditioning of mice was accomplished by intraperitoneal injection (i.p.) of a sublethal bolus of TLRGRADE® LPS from E. coli Serotype O111:B4 (EnzoLife Sciences) at 500 mg/kg. After 24 hours, mice were infected with FTT as described above. At 24 hours post infection, mice were euthanized, and tissues processed for enumeration of bacterial burden, inflammatory cytokine levels, and metabolites.

In vivo imaging

Mice were anesthetized and thorax and belly shaved 2 days prior to infection with FTT. Mice were housed on diamond bedding (ENVIGO) to limit dermal abrasions that can cause superficial inflammation and interfere with deep tissue signals. Two hours prior to imaging, mice were injected via the retro-orbital route with 100 μL XenoLight RediJect 2-DeoxyGlucosone 750 (RJ2-DG-750, PerkinElmer) or AnnexinVivo 750 (PerkinElmer). Near-infrared imaging of fluorescent probe distribution and intensity was carried out using an IVIS Lumina XR instrument (PerkinElmer). Immediately prior to imaging, the shave regions on the mice were delipidated using Nair© (Nair, Church and Dwight, NJ) to remove any residual hair that could interfere with the deep tissue fluorescent signal. To image, mice were anesthetized using 1.5% isoflurane/oxygen air mixture and positioned on the imaging platform within the IVIS Lumina XR in the supine position. Black masking tape (Thor Labs) and black construction paper were used to mask signal from regions of non-interest including the bladder, hands, feet, nose, and tail. Images were acquired with indocyanine green filter sets with Ex 745nm/Em820 nm. Immediately following acquisition of whole-body images, mice were euthanized, and individual tissues excised and imaged ex vivo in a petri dish to resolve anatomical specificity of signal. Image analysis was done using Living Image 4.7 software (PerkinElmer).

Enumeration of bacterial burden

Lungs, liver, spleens, lymph nodes, and ears were homogenized following methods previously described [15]. Briefly, tissues were homogenized in ice cold tissue lysis buffer (150 mM Tris-HCL, 5 mM EDTA, 10 mM Trizma base) containing protease inhibitor cocktails I, II, and III (Ag Scientific) by grinding through a fine wire mesh (Belleville Wire Cloth Co.). Tissue homogenates were then serially diluted and plated onto MMH agar plates. MMH agar plates were incubated at 37°C with 5% CO₂ and colonies enumerated after 48 hours.

Quantification of host cytokines

Tissue homogenates were centrifuged at 10,000 x g for 20 minutes to clarify allowing for isolation of the soluble fraction. IFN-γ, IL-6, MCP-1, MIP-1α, TNF-α, KC, RANTES, IL-12p70, EPO, IFN-β, or MMP-9 were measured in the whole tissue homogenates using a Cytometric Bead Array (CBA; BD Biosciences) or Biomarker Group 1 Mesoplex Array (MesoScale) according to manufactures instructions. For the CBA, beads were fixed in PFA prior to flow cytometric analysis. IL-12p40 was measured by ELISA (BD Biosciences) according to manufactures instructions.

LC-MS

For all liquid chromatography–mass spectrometry (LC-MS) analyses, LC-MS–grade solvents and reagents were used. The post caval lobe of the lung, the median lobe of the liver, half the spleen, the ear, and mediastinal or cervical lymph nodes of mice were collected into 400 μL of ice-cold methanol. Tissues were subsequently bead homogenized, and 400 μL of water added. To the suspension 400 μL of chloroform was added, and the sample was agitated via shaking at 4°C for 20 minutes. Layering was induced by centrifugation at 16,000 x g, 4°C for 20 minutes. The polar (upper) layer was collected and diluted 1:10 in 50% methanol in water. Analysis was performed consistent with previous reports [16]. Samples were injected onto a Sciex ExionLC AC system and separated using an ion pairing strategy on a Waters Atlantis T3 column (100 Å, 3 μm, 3 mm × 100 mm) with a 15-minute gradient from 5 mM tributylamine, 5 mM acetic acid in 2% isopropanol, 5% methanol, 93% water (vol/vol) to 100% isopropanol. Metabolites were detected on a Sciex 5500 QTRAP mass spectrometer using a MRM strategy consistent with previous reports [16]. QC injections were performed after every 10 injections to ensure instrument stability. All data were processed and filtered using MultiQuant Software 3.0.3 (Sciex). A 50% missing value cutoff and 30% QC coefficient of variance cutoff were imposed. Remaining missing values were replaced with the minimum group value for that feature. Following filtering, the data was normalized to either the corresponding day 0 or mock-infected vehicle control groups and scaled as the log₂ of the fold change versus the mean of the reference group. A Benjamini-Hochberg correction was applied to univariate analysis with a 10% FDR cutoff. All univariate and multivariate analysis was performed in MarkerView Software 1.3.1 (Sciex).

Comparative analysis of the inflammatory response following intranasal or intradermal infection

Differences in timing and/or magnitude of the host inflammatory response may contribute to increased replication, lymph node colonization, and time to morbidity of FTT after pulmonary versus intradermal infection. To examine this relationship, we performed a spatiotemporal analysis of several cytokine/chemokines across critically involved tissues with respect to each infection route. A biphasic pattern occurred with limited cytokine production between 24–48 hours post infection and extensive cytokine production between 48–96 hours post infection, consistent with sepsis and onset of morbidity (S1 Fig). Because of this biphasic response, we focused our analysis on the 24–48-hour timeframe to identify early factors that may correlate with increased virulence after i.n. exposure (Fig 2A–2E). Cytokine levels in the ear and cLN after i.n. infection was below the detection limit during this early timeframe (uninfected, 24–48 hours post infection) (S2 Fig), consistent with the absence of bacteria (Fig 1C). Increased IFN-γ, IL-6, and KC levels occurred in the lungs by 48 hours post infection after i.n. compared to i.d. infection consistent with higher bacterial loads in this tissue following i.n. infection (Fig 2A). However, with the exception of increased KC and a transient increase in IL-12p40 levels, no evidence of change in systemic markers of inflammation, occurred in the liver following i.n. infection within the first 48 hours (Fig 2C). In contrast to the correlation of the relative bacterial load with production of cytokine observed in the lungs and ears, route of infection greatly, and disparately, influenced the ensuing inflammatory response in the respective draining lymph nodes, i.e. mediastinal lymph node (mLN) for the lung and the cervical lymph node (cLN) for the ear. Specifically, i.n. infected animals had bacterial loads in the mLN that were greater than those detected in the draining cLN of i.d. infected mice at 48 hours (Fig 1D). However, increased bacterial burden did not translate into increased production of IFN-, IL-6, or TNF-α in the mLN compared to uninfected controls at this time point. Further, KC, MCP-1, MIP-1α, and RANTES in the draining mLN of i.n. infected animals were decreased compared to uninfected mice (Fig 2B). In contrast to i.n. infected animals, mice that were inoculated i.d. had significantly greater concentrations of both RANTES and TNF-α in the draining cLN compared to uninfected controls including at a time point in which bacteria were not detected in this tissue (24 hours after infection). This suggested that reprogramming of peripheral lymph tissue following infection was differentially regulated prior to bacterial colonization and was dependent on the route of infection. We hypothesized that the increased responses observed in the lymph nodes of i.d. infected animals may corelate with heightened production of cytokines in the ears of these animals within the first 24 hours of infection. Indeed, i.d. infection resulted in significantly greater concentrations of IL-12p40 and a trend of increased production of other cytokines assessed in the ear 24 hours after infection compared to uninfected controls (Fig 2A–2E). Moreover, by 48 hours after infection every cytokine assayed was significantly increased in the ear after i.d. infection in contrast to just IFN- and IL-6 in the lung after i.n. infection (Fig 2A and 2E). Together these data suggests that i.d. infection promotes an earlier local and systemic inflammatory response than i.n. infection that was dissociated with relative bacterial load.

Open in a separate window

Fig 2

Comparative analysis of inflammatory cytokine/chemokine response following intranasal or intradermal infection.

Whole tissue homogenates were evaluated for changes in cytokine/chemokine levels in the (A) lungs, (B) mLN or cLN, (C) spleen, (D) liver, and (E) ear (i.d. infection only). Data shown are mean ± SEM and were pooled from two separate experiments (N = 10 mice per group). *p<0.05, **p<0.01, ***p<0.001 indicates significance between i.n. and i.d. infection using an unpaired t-test corrected for multiple comparisons using the Holm-Sidak method. For the ear after i.d. infection (B), *p<0.05, **p<0.01, ***p<0.001 indicates significance to time mock infected controls using a one-way ANOVA followed by Dunnett’s correction for multiple comparisons.

During the sepsis stage of disease (72–96 hours), i.n. infection continued to promote greater inflammation at the site of inoculation, i.e. the lungs, relative to i.d. infection (S1A Fig). Intradermal infection promoted comparable cytokine production in the ear (S1B Fig). In the draining lymphatics, except for modest but significant differences in IL-6 and KC, concentrations of cytokines were similar regardless of the route of infection (S1C Fig). In the liver and spleen, a divergent temporal pattern in cytokine production was evident with i.d. infection. Specifically, increased responses in IL-6, KC, and MCP-1 were evident at 72 hours post infection, which were then reversed at 96 hours post infection with i.n. exposure promoting the greater response (S1D–1E Fig). Together, these data suggested that infection route informs transient differences in the magnitude of the inflammatory responses, with i.d. infection promoting earlier inflammation in the draining lymph nodes and a modestly reduced response at later stages of infection in the lungs relative to i.n. infection.

In addition to microbial products, dead and dying host cells can also trigger inflammatory responses. Since we observed differences in cytokine production that did not correlate with bacterial loads in some tissues, we next assessed the burden of dead and dying cells following infection. The flipping of phosphatidylserine (PS) to the outer leaflets of membranes of dead or dying cells increases binding and uptake by resident macrophages [18] and detection of this lipid on the surface of cells can be used as an additional biomarker of inflammation, tissue damage, and cell death (S3A–S3F Fig) [19]. Therefore, we next examined the uptake and tissue distribution of the far-red fluorescent annexin V derivative AnnexinVivo-750 which binds to surface exposed PS. Intradermal infection resulted in increased AnnexinVivo 750 uptake in the draining cLN as early as 48 hours post infection consistent with an earlier inflammatory response (S3E Fig). In the ear, increased AnnexinVivo 750 signal was not evident until 72 hours post infection, which was 24 hours earlier than when it presented within the lungs (Figs S3F and ​and2C).2C). No significant difference in AnnexinVivo 750 uptake was evident in the liver, lungs, or spleen between i.d. or i.n. infected mice during the first 48 hours. However, by 72 hours post infection significant inflammation and/or cellular damage occurred in the spleens of infected mice, irrespective of infection route, as indicated by increased AnnexinVivo 750 uptake (S3D Fig). Finally, we observed significantly increased AnnexinVivo 750 uptake at 96 hours post infection in all tissues assessed consistent with acute onset tissue damage and morbidity (S3A–S3F Fig). These data further strengthen the conclusion that i.d. infection results in earlier inflammation and/or tissue damage in the draining lymphatics and ear relative to the pulmonary compartment after i.n. infection.

Glucose uptake and distribution following FTT infection

Shifts in host metabolism can be an essential event in the developing inflammatory response with early upregulation of glycolysis supporting rapid activation of host inflammatory programs [9]. The distinct spatiotemporal patterns in cytokine production between intradermal and intranasal infection implicate potential differences in the propensity to increase glycolysis after FTT infection. Measuring distribution and uptake of glucose derivatives with preclinical imaging techniques can identify tissues where increased glycolysis is actively supporting inflammatory functions [19,20]. Therefore, we next assessed the tissue distribution of the glucose derivative XenoLight RediJect 2-DG 750 (RJ2-DG-750) using optical fluorescence imaging in mice following i.n. or i.d. infection. A transient increase (p = 0.055) in RJ2-DG-750 uptake in i.d. versus i.n. infected mice was evident 24 hours following infection in the shaved thorax region of the mice, which included signal from the skin and most major internal organ systems (Fig 3A). However, we could not distinguish which tissues were predominantly responsible for increased signal using whole animal imaging approaches. Therefore, tissues were excised and examined individually ex vivo to delineate whether i.d. or i.n. infection caused tissue specific variations in RJ2-DG-750 uptake over time. Consistent with whole body imaging, irrespective of infection route and consistent with onset of morbidity the predominant glycolytic response occurred by 96 hours post infection in all tissues assessed, (Fig 3B–3F). Interestingly, neither the ear nor lung increased RJ-2DG uptake until later stages of infection, suggesting the metabolic potential of the primary infection site is manipulated by FTT to limit uptake of glucose or refractory to rapid upregulation and, at least at the site of infection in the ear, is dissociated from production of pro-inflammatory cytokines within the first 48 hours of infection (Fig 3B and 3C). Consistent with this notion, (although not statistically significant) both the ear and lung appeared to have reduce RJ2-DG-750 levels at 24 hours post infection relative to mock infected controls (Fig 3B and 3C). Conversely, i.d. infection promoted earlier upregulation of glycolysis in the draining cLN relative to the mLN following i.n. infection. This suggested that while the ear may be refractory to upregulate RJ2-DG-750 uptake, the host is responding rapidly at the level of the draining cLN (Fig 3D).

At later stages of disease, temporal differences were observed in systemic tissues. We observed earlier uptake of RJ-2DG in the liver and increased uptake in the spleen at 96 hours post i.n. inoculation compared to the same tissues following i.d. inoculation (Fig 3E and 3F). Together, these data indicated that i.d. infection promoted a substantive earlier glucose uptake response consistent with earlier cytokine production. However, at later stages of disease i.n. infection drove increased glucose uptake in the liver and enhanced levels in the spleen, indicative of an exacerbated metabolic response in the periphery.

Divergence in central metabolism in the ear and lungs following FTT infection

Our data above suggested that metabolic flux and associated host metabolites may be differently regulated following i.n. and i.d. infection. Therefore, we next characterized changes in a wide range of host metabolites associated with distinct infection routes to identify pathways that may delineate common or divergent changes within or between tissues over time. We used two analytical approaches to curate this data to identify both conserved or divergent responses between infection route in primary exposed tissues (i.e. lungs versus the ear) and systemic tissues (i.e. draining lymphatics, spleen, and liver). For conserved responses, a Benjamini-Hochberg correction with a false discovery rate (FDR) set at 10% comparing mock infected controls to 96 hours after infection was employed. For divergent responses, a Benjamini-Hochberg correction for FDR set at 10% was used to identify significant differences in Log₂(Fold Change) values of metabolites between tissues at 24, 48, and 72 hours post infection. Finally, we categorically binned metabolites that passed FDR cut-offs into families based on their structural relationship (i.e. amino acids) or by association with metabolic pathways (i.e. glycolysis, TCA cycle, redox reactions). This allowed for mapping isolated changes in specific metabolites within a network or involvement of the entire network.

We first compared the ear and lungs after i.d. or i.n. infection, respectively. Conserved changes included decreases in the amino acid tryptophan in the ear and lung albeit to different magnitudes, whereas glutamine and phenylalanine increased in both tissues (Fig 4A). Itaconate, a well-defined marker of TCA cycle reprogramming involved in increased inflammatory cascades, was elevated in both infection sites, but detectable increases were evident 24 hour earlier in the ear than in the lung [21] (Fig 4B). Consistent with increased glycolysis at the end stage of infection observed via RJ-2DG uptake, decreased glucose and dihydroxyacetone phosphate (DHAP) levels, were evident indicative of increased flux in both tissues (Fig 4A). Finally, NAD+ and NADP+ levels progressively decreased in both tissues, potentially indicative of increased bioenergetic demand (Fig 4B). In contrast to these conserved signatures, we observed more metabolites that significantly diverged between the ear and lung over time. Among divergent responses, i.n. infection resulted in a transient increase in arginine and phenylalanine at 24 hours post infection in the lungs followed by a general depletion in amino acids, an effect that was absent in the ears of i.d. infected mice (Fig 4B). Additionally, a depletion or reduction in glycolytic intermediates, nucleotide monophosphates (NMP), dATP, and TCA cycle metabolites occurred over time and was unique to the lungs among i.n. infected mice compared to the ear of i.d. infected mice (Fig 4B). Finally, i.n. infection resulted in an increase in most nucleobases except for hypoxanthine, which was significantly depleted at later stages of infection. The divergence in metabolic programs between ear and lung also included dysregulation in several redox or bioenergetic intermediates (Figs ​(Figs4B4B and S4–S6). Specifically, we observed statistically significant divergence in the NADPH/NADP+ levels in the lungs of i.n. infected mice relative to the ear of i.d. infected mice (S4A Fig). The increased NADPH/NADP+ ratio in the lungs was not paralleled by increased NADH/NAD+ ratio or reduced/oxidized glutathione ratio. This suggested that the response was a function of pentose phosphate pathway regulation rather than oxidative stress or nucleotide salvage (S4F and S5A Figs). We also observed a reduction in the AMP/ATP ratio in the lungs of i.n. infected mice relative to ear after i.d. infection at 48 hours, indicative of an increased bioenergetic potential (S6F Fig). The combination of increased NADPH/NADP+ ratio and bioenergetic potential are conducive of increased anabolic functions, including fatty acid synthesis which has been reported to be an energy source for FTT replication [22,23]. Conversely, the ear had higher baseline reduced/oxidized glutathione and NADH/NAD+ ratios that decreased with time, suggesting elevated redox stress (S4F and S5A Figs). Ultimately, these data support that the intradermal compartment undergoes earlier increased glycolytic and oxidative stress responses relative to the lungs, and generally has less depletion of metabolites during disease pathogenesis.

To determine if anatomically distinct infection routes drove divergent responses in the lungs, we next compared the metabolic response of only the lungs after i.n. or i.d. infection over time. Surprisingly, metabolic changes were highly conserved (Fig 4C and 4D). There were no significant differences in NADPH/NADP+ and NADH/NAD+ ratios over time (S4B and S4G Fig). Changes in reduced/oxidized glutathione ratios, ADP/ATP and AMP/ATP ratios were markedly conserved, with significant, but transient differences observed at 24 and 48 hours, respectively (S5B and S6B Figs). Further, 29 metabolites of the 111 metabolites detected shared common directional change between infection routes, including several amino acids, glucose, DHAP, itaconate, several nucleotide/nucleoside bases, nucleotide triphosphates, and cGMP, metabolites associated with redox status, and nucleotide diphosphate sugars (Fig 4C). Conversely, only 3 metabolites Phosphoribosyl pyrophosphate (PRPP), Uracil, and UTP) passed 10% FDR cutoff for differential change, and differences were marginal or transient (Fig 4D). Therefore, the metabolic response to infection in the lung was remarkably conserved regardless of where infection was first established.

Divergence in central carbon metabolism in the draining lymph nodes after FTT infection

Significant differences in glucose uptake (Fig 3D) and inflammation (Fig 2) observed in the cLN after i.d. infection compared to the mLN after i.n. infection potentially indicated broader divergence in the local metabolic response defining the response in these tissues. Therefore, we next compared the metabolic profiles of the respective draining lymph nodes associated with i.n. and i.d. infection sites. (Fig 3D). Conserved metabolic responses included changes in isolated amino acid, glycolytic, nucleobases, and mono/di/tri nucleotide phosphates. We observed depletion of tryptophan, glucose, and nucleobases whereas there were accumulations of arginine, urate mono, di and triphosphates in both the cLN and mLN over time (Fig 5A). Conversely, divergent responses identified included depletion in cystine in the cLN after i.d infection and accumulation of this metabolite mLN after i.n. infection (Fig 5B). Additionally, arginine and citrulline increased earlier in the cLN compared to the mLN. Divergent patterns in glycolysis were also evident, including decreased fructose-6-phosphate in cLN versus increased levels in the mLN over time. The mLN also contained reduced downstream glycolytic intermediates bisphosphoglyceric acid (BPG) and PEP relative to cLN (Fig 5B). Although it did not meet the <10% FDR, we observed a transient increase in itaconate in the cLN 24 hours after infection which then fluctuated over time. Itaconate accumulation occurred at 72 hours post infection in both the cLN and mLN, however responsiveness was significantly enhanced in the mLN relative to the cLN at this time point (Fig 5B). Intradermal infection resulted in depletion of TCA cycle intermediates in the cLN, apart from succinyl-CoA which increased with time (Fig 5B). Succinyl-CoA ligase is responsible for converting Succinyl-CoA to succinate, generating either ATP or GTP in the process, and this process can occur under oxygen deficient conditions [24]. Therefore, accumulation of Succinyl-CoA and cofactor GDP may be indicative of dysregulated Succinyl-CoA ligase activity in the cLN. Inversely, Succinyl-CoA was depleted in the mLN (Fig 5B). Increased AMP and GMP further suggest significant alteration in nutrient stress in the cLN relative to the mLN. cLN also had higher baseline reduce/oxidized glutathione ratio compared to the mLN, indicative of increased redox activity (S5C Fig). Together, these data support greater metabolic responsiveness in the cLN after i.d. infection that were supportive of pro-inflammatory activity. In contrast, the mLN demonstrated less dramatic metabolic dysregulation consistent with a more inflammatory null phenotype, parallel to the lung, at earlier stages of infection.

Open in a separate window

Fig 5

Infection route promotes divergent metabolic programs in the draining lymph node.

LC-MS profiling of metabolites in the draining lymph node associated with the primary infection site including the mLN for i.n. infection of the lung, and cLN for i.d. infection of the ear. (A) Metabolites with conserved changes between the mLN and cLN 96 hours after infection. (B) Metabolites with significantly divergent patterns between the cLN and mLN over time. Data are shown as the mean Log₂FC values relative to uninfected (N = 5 mice per group). Corresponding grey boxes on the accompanying greyscale heatmap indicate significance between i.n. and i.d. infected groups for (B). Statistical significance was calculated using a Benjamini Hochberg test with Q = 10% on log transformed data. For divergent patterns, metabolite hits were identified by comparing magnitude of shift in log₂FC values between i.n. and i.d. animals at 48 and 72 hours post infection. For conserved patterns, metabolites were selected through a Benjamini-Hochberg test with Q = 10% cutoff using the 96-hour timepoint post infection compared to mock infected controls for both i.n. and i.d. infected groups. Metabolites were binned into families based on structural or network relations which were abbreviated as follows: AA = Amino Acids, GlcN = Gluconeogenesis, Glyco = Glycolysis, TCA = TCA cycle, NTP = Nucleotide Triphosphates, NDP = Nucleotide Diphosphates, NMP = Nucleotide Monophosphates, Redox Metabolism = redox, N-base = Nucleobases, PPP = Pentose Phosphate Pathway, NDPS = Nucleotide Diphosphate Sugars, and Misc = Miscellaneous.

FTT manipulates metabolism of systemic tissues prior to detectable colonization

Our data demonstrated the metabolic response in the lungs was remarkably conserved irrespective of infection route (Fig 4). However, the liver and spleen also become significantly involved in disease pathogenesis, and functional failure of these tissues has been attributed to onset of morbidity [25]. If or how infection route informs the host metabolic response in the liver and spleen has not been explored. Therefore, we evaluated changes in the metabolic milieu in the spleen and liver over time following i.n. and i.d. infection. Like the lungs, metabolic changes in the liver were highly conserved between infection routes (Fig 6A). Differences between i.d. and i.n. infection route in the liver occurred only at the 48-hour timepoint post infection and were associated with magnitude of change but not change of the trends over time (Fig 6B). Generally, i.n. infection induced a more augmented metabolic response between 24–48 hours post infection compared to i.d. exposure. This observation was surprising, since we did not detect bacterial burden within the liver at this time point. This suggested either host factors or potentially bacterial effectors produced in other organs are driving metabolic responses at distal sites. The early responses observed prior to bacterial colonization included increases in several amino acids (Methionine, Glutamate, Serine, Histidine), and decreased glycolytic intermediates. This pattern indicated potential activation of glycolytic or gluconeogenic pathways. Upon bacterial colonization of the liver, disease progression was characterized by accumulation of amino acids, increased downstream glycolytic intermediates, and reprogramming of the TCA cycle including itaconate production. At the latest stages of disease, we observed a loss of xanthine/hypoxanthine, loss of reduced glutathione and NAD+ levels, and alterations in the ADP/ATP and AMP/ATP ratios. Together these changes suggested dysregulation of the redox balance and/or bioenergetic deficits that were consistent with severe tissue damage (S5D, S6D and S6I Figs).

Assessment of changes of metabolite levels in the spleen followed a similar pattern to the liver. The metabolic responses to infection were highly conserved regardless of infection route (Figs 6C, 6D, S4E, S4J, S5E, S6E and S6J). Further like the liver, many metabolites had measurable changes as early as 24 hours post infection which is before detectable colonization of this organ following i.n. infection. Conversely, 16 out of the 111 metabolites detected diverged between i.d. and i.n. infection, but like the liver this pattern only occurred at 48 hours post infection (Fig 6D). Furthermore, significant differences were primarily due to an augmented, but transitory, response in the i.n. infected group. Notably, we observed similar accumulation of glycolytic intermediates and dysregulation in the pentose phosphate pathway, hypoxanthine/urate metabolism, and redox networks. Together, these data suggested that early after infection the site of inoculation influenced metabolic processes at distal sites and i.n. infection often resulted in a more marked reversal of metabolite trends between 24 and 48 hours post infection relative to i.d. inoculation. However, once peripheral tissues were colonized these differences were resolved with the liver and spleen entering an energetic crisis at late stages of disease progression associated with organ failure and onset of morbidity.

We would also like to acknowledge Dr. Lydia Roberts for important critical input toward the manuscript.