Inoculation dose of Mycobacterium tuberculosis does not influence priming of T cell responses in lymph nodes

Abstract

The effect of Mycobacterium tuberculosis inocula size on T cell priming in the lymph node and effector T cells in the lung remains controversial. In this study, we used a naïve mouse model, without the transfer of transgenic T cells, in conjunction with mathematical model to test whether infection with higher aerosolized inocula would lead to increased priming of M. tuberculosis-specific T cells in the lung-draining lymph node. Our data do not support that inoculum size has a measurable influence on T cell priming in the lymph nodes, but is associated with more cells overall in the lung, including T cells. To account for increased T cells in the lungs, we tested several possible mechanisms, and recruitment of T cells to the lungs was most influenced by inoculum dose. We also identified IL-10 as a possible mechanism to explain the lack of influence of inoculum dose on priming of T cells in the lymph node.

Introduction

Mycobacterium tuberculosis is a leading global health concern causing 1.7 million deaths in 2009 (1). A small percentage of infected individuals do not generate an initial protective response and progress to primary disease within 2 years, while 90% of immunocompetent people are able to control infection, often for life. Although a vaccine (BCG) is widely administered, it does not provide adequate immunity against infection or disease. Various animal models and human studies have demonstrated that certain cytokines, including IFN-γ, TNF, and IL-12, as well as CD4 and CD8 T cells and activated macrophages are essential for protection against tuberculosis (4).

Initiation of the immune response against M. tuberculosis infection is a slow process. In humans, very little is known about the events that occur during transmission and initial stages of infection, since these events are “silent”. Exposure is likely to be to a very small numbers of organisms, and in some settings it is probable that it is repeated exposure that results in successful transmission events.

Animal studies demonstrated that the microbe is inhaled into the airways where it encounters alveolar macrophages and dendritic cells, which transport bacteria to draining lymph nodes for the purpose of priming T cells (2, 3). These primed T cells migrate back to the infected lung to participate in granuloma formation, but the lymph nodes also remain infected. Studies have demonstrated that after low dose aerosol infection of mice, M. tuberculosis bacilli appear in the lymph node between days 9-11, with variation among even inbred mice; bacteria in lymph nodes is necessary to initiate a priming response (3). A recent study (7) using mice without appreciable lymph nodes suggested that priming of T cells can also occur in the lung. In normal mice, bacteria arrive in the spleen 2-3 weeks post-infection, and this is also a potential site for priming T cells. Using adoptive transfer systems with large numbers of M. tuberculosis antigen-specific transgenic T cells, priming of T cells in lymph nodes (as determined by CD69 expression) occurred between days 11-12, but significant T cell proliferation in the lymph nodes began only at day 14 (19). T cell responses can be detected in the lungs by ~2 weeks post-infection (p.i.), and by 4 weeks p.i., bacterial growth in lungs is stabilized (13); the level of bacilli in lungs remains at high levels for months as the mouse experiences progressive chronic tuberculosis. This relatively long period of time between infection and induction of T cell responses may allow M. tuberculosis to gain a foothold in the lungs without facing an adaptive immune response (3). This was also observed in a computational model of the immune response in lungs to M. tuberculosis (20). Understanding factors involved in priming of T cells in response to M. tuberculosis infection may improve our ability to design vaccines that enhance rapid recall responses in the lungs and lymph node to improve protection against disease.

Our previous studies in CD40-/- mice indicated that a 2-3 fold higher aerosol inoculum resulted in an increase of IFN-γ producing cells in the lymph node by 3 weeks and in the lungs at 4 and 5 weeks, thus improving survival of these mice (12). This suggested that antigen or bacterial load could influence priming of T cells in the lymph nodes. Two other studies, using adoptive transfer of transgenic T cells, demonstrated that the numbers of bacilli within the draining lymph nodes were positively correlated with robustness of priming (as defined by activation and proliferation of the transgenic T cells) (19, 22). The studies were conflicting in their findings of effects of inoculum size on timing of priming: one study supported that a higher inoculum could cause earlier priming of T cells (19), however effects were minimal even though large inocula (1200 CFU via aerosol) were used. The other study showed an effect of dose on magnitude of responses but not on timing of induction (22).

In the current study, we addressed the influence of inoculum size on timing and magnitude of T cell priming in lymph nodes in a naïve mouse model without transfer of transgenic T cells, to determine how normal naïve frequencies of M. tuberculosis antigen-specific T cells respond to different doses of infection. We integrated mathematical modeling of the priming response in lymph nodes with our experimental data and determined that, in an intact mouse, there was minimal effect of inoculum size on priming in the lymph node and only a modest effect on the M. tuberculosis-specific number of T cells in the lung. There were, however, higher bacterial numbers and total cell numbers (including T cells) in mice with higher inocula compared to those inoculated with fewer bacteria. We addressed 3 hypotheses for these observed increases using wet-lab and modeling approaches. Our mathematical modeling approach predicts additional factors, such as a key role for IL-10 and dendritic cells in regulating T cell priming in the lymph node and in driving effector T cell numbers in the lung in a high dose scenario.

Materials and Methods

Aerosol Infection of mice

C57BL/6 mice (The Jackson Laboratory, Bar Harbor, ME) were aerosol infected with M. tuberculosis (H37Rv or Erdman strain) using a flat chamber aerosolizer (In-Tox Products, Moriarty, NM). A low dose of 8 × 10⁵ - 3 × 10⁶ or a high dose of 8 × 10⁶ - 3 × 10⁷/ml in the nebulizer was used to deliver 10-40 or 130-550 CFU respectively to the lung as determined by plating whole lung homogenates from 4-6 mice per group on 7H10 agar (Difco, Sparks, MD) 24 hours after aerosolization. Colonies were counted after incubation of plates for 21 days at 37C and 5% CO₂. The experiments were performed six times, using different doses and timing of necropsies. Representative data are provided here. Mice were kept under specific pathogen-free conditions in a biosafety level 3 facility and all studies were approved by the University of Pittsburgh Institutional Animal Care and Use Committee.

Single cell suspensions and flow cytometry

Whole lungs and lymph nodes were crushed and single cell suspensions were obtained as previously described (8). Live cells were counted by trypan blue exclusion and 5×10⁵ cells were washed in FACS buffer (1x Phosphate buffered saline (PBS) and .01% Bovine serum albumin (BSA), SerCare Life Sciences, Inc., Milford, MA). The following cell surface antibodies were added in staining buffer (FACS buffer, 20% mouse serum (Gemini Bio-Products, West Sacramento, CA)) anti CD69 (clone H1.2F3), anti CD11b (clone M1/70), anti GR-1 (clone RB6-8CS), anti CD4 (clone RM4-5) and anti CD8 (clone 53-6.7) from BD Biosciences (San Diego, CA) and anti CD11c (clone N418) from eBioscience (San Diego, CA). MHC Class I GAP tetramer (recognizing the GAPINSATAM amino acids of Rv0125) (Beckman Coulter, Fullerton, CA and NIH Tetramer Facility, Atlanta, GA) and MHC Class II multimer ESAT-6 (recognizning aa1-20 (MTEQQWNFAGIEAAASAIQG) (Marc Jenkins, University of Minnesota, Minneapolis, MN and NIH Tetramer Facility, Atlanta, GA) were also added at 1:67 or 1:100 (respectively) final concentration. Cells were incubated with antibodies for 30 min at room temperature then washed with FACS buffer and fixed in 2% paraformaldehyde (Sigma-Aldrich, St. Louis, MO). Samples were collected using an LSRII flow cytometer (BD Biosciences, San Jose, CA) and analyzed with Flowjo (Tree Star, Ashland, OR). A table summarizing these data is included as Supplementary Table I and is a representative experiment characterizing cell types in the lung.

CFU

Bacterial counts were determined by serial dilution of lung and lymph node homogenates from mice at various time points post-infection. Dilutions were plated onto 7H10 agar (Difco, Sparks, MD) and incubated at 37C, 5% CO₂ for 21 days.

IFN-γ production by ELISPOT assay

IFN-γ was measured by ELISPOT assay. Bone marrow derived dendritic cells used as antigen presenting cells in the assay were generated as previously described (13). Dendritic cells were either infected with M. tuberculosis overnight prior to stimulation of T cells in the ELISPOT assay, or ESAT-6 peptides were added 1 hour prior to addition of T cells.

Ki67 staining for proliferation

Cell surface staining was performed as above with ESAT6 multimer and GAP tetramer, anti-CD69, anti-CD4 and anti-CD8 antibodies, as previously described. After a 1 hour incubation, cells were then washed with FACS buffer and fixed in 2% paraformaldehyde and 2x APO active fixative (Cell Technology Inc., Mt. View, CA) for 20 min to overnight at room temperature. Cells were washed twice in 1x PBS, resuspended in 100ul 1% saponin (EMD Chemicals Inc, Gibbstown, NJ) and 20 ul anti Ki67 (clone B56) from BD Biosciences. Cells were incubated for 60 minutes at room temperature then washed twice in 1% saponin/1x PBS, 1x in 2% BSA/1x PBS and resuspended in 1% paraformaldehyde. Samples were collected using an LSRII flow cytometer (BD Biosciences, San Jose, CA) and analyzed with Flowjo (Tree Star, Ashland, OR).

Statistical analysis

Data were compared among high and low dose mice at each time point. Mann-Whitney analysis was used for flow cytometry data, and Student’s t test for log-transformed CFU data and ELISPOT data. Analyses were performed in Prism (GraphPad) or Excel (Microsoft), including the linear regression analyses. Mean and standard deviation is shown in each figure.

Mathematical model

We developed a series of mathematical models that qualitatively and quantitatively characterize the cellular and cytokine network during infection in lung and lymph node (15, 16). Here we used our most recent-generation, two-compartment mathematical model of lung and lymph node dynamics (16) to capture the behavior of T cell priming in lymph nodes and T cell migration to lungs during M. tuberculosis infection. The model was built on our experimental data generated in normal mice. A mathematical modeling approach allows us to address a number of questions that are challenging to address in an in vivo system, and since this model was built on data from immune competent C57BL/6 mice infected with different inocula, it is directly relevant to our studies here. The model captures the dynamics of many cell types, such as macrophage (resting-M_R, infected-M_I, classical-CAM and alternatively activated-AAM), dendritic cell (immature-IDC and antigen-bearing or mature-MDC) and lymphocyte populations in both lung and lymph node (naïve, precursor and effector CD4 and CD8 T cells). Table I lists the T cell phenotypes used in the model (macrophages not shown, for a full explanation of model see (16)). We also model 4 cytokine concentrations (TNF, IFN-γ, IL10 and IL12) in both lymph node and lung compartments. Details of the non-linear ordinary differential equation system are shown in (16). Figure 1 shows a diagram of the main immune mechanisms captured in the two compartment mathematical model.

Table I.

T cell phenotypes described in the mathematical model.

Cell type	Lung	Lymph Node
Lymphocytes		Naïve CD4 [N4], Naïve CD8 [N8]
	Precursor Th1 [T̂1], Th1 [T1]	Precursor Th1 [ ${\widehat{T}}_1^{\mathit{LN}}$], Th1 [ $T_1^{\mathit{LN}}$]
	Precursor effector CD8 [T80], IFN-γ producing CD8 [T8], CTL [TC]	Precursor effector CD8 [ $T_{80}^{\mathit{LN}}$], IFN-γ producing CD8 [ $T_8^{\mathit{LN}}$], CTL [ $T_C^{\mathit{LN}}$]

List of all the lymphocyte phenotypes represented in the ordinary differential equations (ODEs) system capturing lung-lymph node dynamics as described in (16). Measure units are number of cells in the whole organ (lung or lymph node). The variable labels are shown in square brackets to help the reader in cross-referencing all the model details illustrated in (16).

Figure 1. Schematic cartoon representing the main immune mechanisms described in our two compartment mathematical model capturing dynamics occurring within and between LNs and Lung.

All details on the model are in Marino, et al (16).

The model describes macrophage recruitment, infection, activation and death (by age, apoptosis, bursting and cytotoxic T cell killing). Dendritic cell mechanisms captured in the model are recruitment, maturation, trafficking to the lymph node and death (by age). DCs become mature after bacterial uptake and they are the only cell types that migrate to the lymph node upon maturation. We also describe mechanisms of T cell recruitment, differentiation, proliferation, migration and death (by apoptosis and age), both in lung and lymph node. In order to exactly replicate experimental conditions, we use data on CFU dynamics as input functions for the model in the lung and lymph node. Details on how the input functions are calculated and implemented are shown in (16).

Uncertainty and Sensitivity Analysis

To further investigate the impact of inocula dose on priming dynamics, we performed uncertainty and sensitivity analyses (US/A) on simulations generated after varying the dose of infection. This method allows us to determine the amount of variability in the system and also which mechanisms predominately induce it. More generally, we seek to identify immune mechanisms that affect a change in priming and/or effector cell numbers in lung and lymph node during infection with M. tuberculosis. Uncertainty analysis is performed by an extensive and efficient statistical sampling of the parameter space called latin hypercube sampling (LHS) scheme. We use a generalized correlation coefficient called partial rank correlation coefficient (PRCC) to correlate changes in immune mechanisms as well as antigen dose to some key readouts of the system, for example the magnitude of primed (in the lymph node) and effector (in the lung) CD4 and CD8 T cells. Details and applications of these techniques in systems biology can be found in (14).

We perform a detailed uncertainty analysis on the model by varying approximately 100 parameters within ranges based on baseline scenarios described in (16). We use a sample size of 3000, with uniform a priori probability distributions for each unknown parameter. The mathematical model simulates 30 days post infection to match the experimental data. We then calculate partial rank correlation coefficients (PRCCs), focusing on T cell priming in the lymph node and on number of effector T cells in the lung. For the purpose of the analysis, we label all the effector T cells in the lymph node as primed T cells, and all the effector T cells in the lung as effector T cells. The results are grouped by T cell phenotype and physiological compartment (lung or lymph node) and are all shown in Tables II and III. We also looked at what mechanisms can impact total cell numbers in the lung.

Table II.

Sensitivity Analysis Results varying antigen dose.

Output	Positive Correlation	Negative correlation
CD4 in the Lymph Node	Max infection/uptake rate by Immature DCs (IDC) in the Lung(*) [k12] Max TNF-dependent IDC recruitment rate(*) [rc4]; Rate (likelihood) of Naïve CD4+ T cell priming when encountering a MDC (LN) [k14]; Precursor Th1 proliferation rate (LN) [k15]	Resting M0alternative activation rate(*) (LN) [k7a]; Precursor Th1 cells migration rate out of the LN into the blood (*) [ξ1]; Max rate of TNF-dependent apoptosis of Th1 cells (LN) [k22a]; Max Th1 differentiation rate, IL12 and MΦ dependent (LN) [k29a];
CD8 in the Lymph Node	Max IDC infection/uptake rate in the Lung(*) [k12] Max TNF-dependent IDC recruitment rate(*) [rc4]; Max rate (likelihood) of Naïve CD8+ T cell priming when encountering a MDC (LN) [k17] Precursor T8/CTL proliferation rate (LN) [k18]	MR alternative activation rate(*) (LN) [k7a] Max T8/CTL differentiation rate, IL-12/MDC dependent (LN) [k24a]; Max rate of TNF-dependent apoptosis of CTL cells (LN) [k28a]; Max rate of TNF-dependent apoptosis of T8 cells (LN) [k26a]
CD4 in the Lung	Max IDC infection/uptake rate in the Lung(*) [k12]; Precursor Th1 proliferation rate (LN) (*) [k15]; Rate (likelihood) of Naïve CD4+ T cell priming when encountering a MDC (LN) [k14]; Scaling factor between lung and LN compartments [ϒ]; Max TNF-independent precursor Th1 recruitment rate (Lung) [rc5]; Precursor Th1 cells migration rate out of the LN into the blood [ξ1]; Max TNF-independent Th1 recruitment rate (Lung) [rc5a] Max TNF-dependent Th1 recruitment rate (Lung) [rc6a];	Maximum Th1 differentiation rate, IL-12/MΦs dependent (LN) [k29a]; Max Th1 differentiation rate, IL-12/MDC dependent (LN) [k20a]; Max rate of TNF-dependent apoptosis of Th1 cells (LN) [k22a] Naïve CD8+ T cells max MDC-dependent recruitment rate (LN) [k16];
CD8 in the Lung	Max IDC infection/uptake rate in the Lung(*) [k12] Max TNF-dependent IDC recruitment rate(*) [rc4]; Max rate (likelihood) of Naïve CD8+ T cell priming when encountering a MDC (LN) [k17]; Naïve CD8+ T cells max MDC-dependent recruitment rate (LN) [k16]; Precursor T8/CTL proliferation rate (LN) [k18]; Max TNF-independent precursor T8/CTL recruitment rate (Lung) [rc7]; Scaling factor between lung and LN compartments [ϒ]; Precursor T8/CTL cells migration rate out of the LN into the blood [ξ2]	MR alternative activation rate(*) (LN) [k7a](*); Max T8/CTL differentiation rate, IL-12/MDC dependent (LN) k24a]; Max T8/CTL differentiation rate, IL-12/MΦs dependent (LN) [k30a] Max TNF-dependent apoptosis rate of CTL cells (LN) [k28a]; Max rate of TNF-dependent apoptosis of T8 cells (LN) [k26a]

Most important mechanisms affecting priming (CD4s and CD8+ Ts) that resulted consistently significant (p<0.001) between high and low dose LHS experiments. Only either PRCCs>0.15 or PRCCs<-0.15 are shown.

^(*)significant only in the first 10 days post infection. The parameter labels in Tables 2 and 3 are shown in square brackets to help the reader in cross-referencing all the model details illustrated in (16).

Table III.

Sensitivity Analysis Results fixing antigen dose.

NAME	LOW DOSE	HIGH DOSE
CD4 in the Lymph Node		IL10 half saturation on delaying IDC maturation (Lung) [hsI10-DC] (+) IL10 half saturation on delaying precursor Th1 proliferation (LN) [hsI10-T0LN] (*) (+)
CD8 in the Lymph Node	Max IDC infection/uptake rate in the LN (*)(+)[k12a]; Rate (likelihood) of Naïve CD4+ T cell priming when encountering a MDC (LN)[k14] (*)(+); Scaling factor between lung and LN compartments [ϒ] (*)(-);	IL10 half saturation on delaying IDC maturation (Lung) [hsI10-DC] (+) IL10 half saturation on delaying precursor Th1 proliferation (LN) [hsI10-T0LN] (*) (+) MDC migration rate from the lung to the LN [φ] (*)(+)
CD4 in the Lung		IL10 half saturation on delaying IDC maturation (Lung) [hsI10-DC] (+)
CD8 in the Lung	Max IDC infection/uptake rate in the LN(*) (+) [k12a]; Rate (likelihood) of Naïve CD4+ T cell priming when encountering a MDC (LN)[k14] (*)(+); Scaling factor between lung and LN compartments [ϒ](+);	IL10 half saturation on delaying IDC maturation (Lung) [hsI10-DC] (+)

Important mechanisms affecting priming (CD4s and CD8+ Ts) that resulted uniquely significant (p<0.001) for either high or low dose LHS experiments. Only either PRCCs>0.15 or PRCCs<-0.15 are shown. The sign of the correlation is in parenthesis.

^(*)significant only in the first 10 days post infection.

RESULTS

Higher inocula results in higher bacterial numbers in lymph node and lung, but higher cell numbers only in the lungs

C57BL/6 mice were infected via aerosol with M. tuberculosis (Erdman strain). Inoculum dose was modulated by the concentration of bacteria placed in the nebulizer; exposure time in the aerosolizer was kept constant. In this experiment one group was exposed to a low dose of 2.7 × 10⁶/ml and the other group a high dose of approximately 2.4 × 10⁷/ml in the nebulizer. At day 1, 4 mice per group were harvested to determine actual dose of bacteria delivered to the lung. The low dose mice received an average of 20 ± 5 CFU and the high dose mice an average of 555 ± 150 CFU in the lungs. Mice were harvested at early time points (4 mice per group at Days 9, 14, 18, 22, 25, 28 p.i.). Mice infected with the higher inoculum had significantly higher bacterial burdens at 14 and 28 days in the lymph nodes (Figure 2A). In the lungs, bacterial numbers were 10-fold higher in the group with the higher inoculum at every time point sampled through 28 days (Figure 2A). In a separate experiment, mice were followed to 14 weeks and those infected with a higher inoculum maintained significantly higher bacterial burdens in the lungs (data not shown).

Figure 2. Higher inoculum results in higher bacterial numbers in lymph node and lung, but higher cell numbers only in the lungs.

(A) Colony forming units (CFU) in lymph node (LN) and lung following low (white) or high dose (black) infection. Each mouse is represented with a symbol and the line is through the average at each time point (low dose: dashed line; high dose: solid line). P value at day 9 in the LN is .0519 and was not considered statistically significant due to an outlier in the high dose group. Level of detection in LN at Day 9 was 35 and at Day 1 in lung was 5. Low dose group (N=24) received a nebulizer concentration during aerosol infection of 2.7 × 10⁶/ml and the high group (N=24) 2.4 × 10⁷/ml, resulting in average day 1 CFU in the lungs of 20 in the low dose group and 555 in the high dose group. (B) Total cell number in the LN and lung with of 4 mice per group per time point. Statistical significance was determined by the unpaired student t test. * p<.05, **p<.01, ***p<.001. Standard deviation reported. This experiment was independently performed 6 times, and a representative experiment is shown here.

Total cell numbers upon homogenization were counted in the lung and lymph node using trypan blue exclusion; most of the living cells were lymphocytes or of monocytic lineage. No difference in total cell numbers was seen in the lymph nodes between the mice inoculated with low or high dose M. tuberculosis, but after 3 weeks there were significantly more cells in lungs of mice inoculated with the higher dose. (Figure 2B) These findings were consistent across 6 separate experiments with different inocula, with low dose defined as <130 CFU and high dose as >130 CFU; matched low and high dose groups were inoculated with ~10 fold differences in CFU.

IFN-γ production in lungs, but not lymph node, is associated with inoculation dose

M. tuberculosis infection results in the presence of IFN-γ producing T cells in the lung by 3-4 weeks post infection in mice (13). If inoculation dose were a factor in priming immune responses to M. tuberculosis, we would expect an increase in IFN-γ producing T cells in lymph nodes in mice that received a higher inoculum, since more bacteria (and therefore antigen) should increase priming and cytokine producing T cells. This should then cause a subsequent increase in IFN-γ producing cells in the lungs. By IFN-γ ELISPOT, lymph node cells from mice infected with low or high dose had similar frequencies of M. tuberculosis-specific IFN-γ responses from T cells (Figure 3). In contrast, in high dose infected mice, there was a higher frequency of IFN-γ producing T cells in lungs between 3-14 weeks post infection (Figure 3), compared to low dose infected mice. For this assay we stimulated lung and lymph node cells with either M. tuberculosis-infected dendritic cells (Figure 3A), or dendritic cells pulsed with ESAT-6 peptides (Figure 3B). Although there were higher responses to the M. tuberculosis-infected dendritic cells, the pattern was similar when ESAT-6 peptides were used.

Figure 3. Inoculum dose influences IFN γ production as measured by ELISPOT in the lungs but not the lymph nodes.

Frequency of IFN-γ+ cells in 150,000 LN cells or 80,000 lung cells; (A) stimulation was with M. tuberculosis-infected dendritic cells or (B) dendritic cells pulsed with ESAT-6 peptide. The graphs summarize the data of 6 mice per group per time point. H37Rv strain of M. tuberculosis was used for mouse infection. The average day 1 CFU was 10 in the low dose group (white) and 130 in the high dose group (black). Statistical significance was determined by a 2 tailed T test with unequal variance. * p<.05, **p<.01, ***p<.001. Standard deviation reported. Lymph nodes could not reliably be found prior to 7 days post-infection. Not enough cells were obtained at time points before 2 weeks in the low dose (white) lymph node and 3 weeks in the lung to perform assay. This experiment was independently performed 3 times, with representative data shown here.

ELISPOT relies on stimulation of T cells for 2 days, and may provide additional signals for T cell differentiation and cytokine production ex vivo; thus one may be measuring the potential for cells to produce cytokines, rather than the status of cytokine production in the tissue. To assess the T cell populations more directly, we used flow cytometry and fluorescent-tagged M. tuberculosis antigen specific tetramers for T cells recognizing ESAT-6 and Rv0125 (i.e. GAP) (8)) to characterize M. tuberculosis-specific CD4 and CD8 T cells (Figure 4). Ki67 and CD69 were markers used to assess proliferation and activation respectively (Figures 5 and 6). Intracellular cytokine staining to assess TNF, IFN-γ, and IL-2 production by specific T cells in response to ESAT-6 and Rv0125 was also performed (data not shown). Unexpectedly, no differences in the frequency of ESAT6+ CD4 cells and GAP+ CD8 cells in the lung or lymph node were observed between mice inoculated with high or low dose (Figure 4A). In terms of actual numbers of ESAT-6+ CD4 and GAP+ CD8 T cells, there were no differences in the lymph node; in lungs, there were more of these specific T cells in the high-dose animals at Day 22 and 25 (Figure 4B), reflecting increased numbers of overall cells. No dose-dependent differences were observed in the functions (cytokines) of ESAT-6 or GAP specific T cells in lungs or lymph nodes by intracellular cytokine staining (data not shown). We also stimulated with anti-CD3/anti-CD28 antibodies to assess total cells capable of producing cytokines, and again found no differences between mice infected with high or low dose (data not shown) in terms of function of responding cells.

Figure 4. No differences in frequency of tetramer positive M. tuberculosis-specific T cells in the lymph node or lung.

(A) Frequency of M. tuberculosis-specific CD4 or CD8 T cells in LN and lung. Low dose: white symbols; high dose: black symbols. (B) Number of M. tuberculosis-specific cells. The LN shows no differences, however there are significant differences at day 22 and 25 in the lung due to the increase in total live cells at that time. Total numbers are calculated by multiplying the % positive of the live cell gate by the total number of cells in the tissue. The data are from 4 mice per group per time point where low dose (N=20) received a nebulizer concentration of 3.3 × 10⁶/ml and the high group (N=20) 2.7 × 10⁷/ml. The average day 1 CFU was 40 in the low dose group and 550 in the high dose group. Statistical significance was determined by a Mann Whitney t test * p< .05. This experiment was independently performed 4 times, with representative data shown here.

Figure 5. Cell proliferation does not account for the difference in cell numbers in the lung between low and high inocula in mice.

Cell proliferation was measured by Ki67 expression. LN or lung cells were gated on CD4 (A, solid line) or ESAT6+ CD4 (A, dashed line) or CD8 (B, solid line) or GAP+ CD8 (B, dashed line) T cells. Low dose: white symbols; high dose: black symbols. There is no difference in %Ki67+ in the LN and only a significant difference in the lung at day 14 in GAP+ CD8 T cells between high and low dose mice. Graphs summarize the data of 4 mice per group where low dose (N=20) received a nebulizer concentration of 3.3 × 10⁶/ml) and the high group (N=20) 2.7 × 10⁷/ml. The average day 1 CFU was 40 in the low dose group and 550 in the high dose group. Statistical significance was determined by a Mann Whitney t test * p< .05. This experiment was independently performed 4 times, and a representative experiment is shown here.

Figure 6. CD69 expression on T cells is not dependent on inoculum dose in lymph nodes.

LN or lung cells were gated on CD4 (A) or ESAT6+ CD4 (C) T cells and CD8 (B) or GAP+ CD8 (D) T cells and assessed for CD69 expression. Low dose: white symbols; high dose: black symbols. Graphs summarize the data of 4 mice per group where low dose (N=20) received a nebulizer concentration of 3.3 × 10⁶/ml) and the high group (N=20) 2.7 × 10⁷/ml. The average day 1 CFU was 40 in the low dose group and 550 in the high dose group. Statistics used – Mann Whitney t test * p< .05. Experiment was independently performed 4 times. Representative data are shown here.

Proliferation of T cells is similar between high and low dose infection mice in lungs and lymph nodes

Although there appeared to be no dose-related response in terms of numbers of cells or priming in the lymph nodes, there were more cells, including T cells, in the lungs by 3 weeks post-infection in the high dose group compared to the low dose group. There are at least three possibilities that could account for this increase: increased migration from lymph node to lung, increased proliferation of T cells in the lungs, or decreased cell death, all possibly related to antigen load. We addressed these possibilities by integrating our math modeling and mouse experimental approaches. We tested whether increased proliferation could account for the increases using the mouse model, while we tested increased recruitment and decreased cell death with the mathematical model.

To assess proliferation in the lymph nodes and lungs, Ki67, a nuclear antigen expressed in actively cycling cells, but not in resting G0 cells (11), was measured on T cells by flow cytometry. ESAT6+ CD4 T cells expressing Ki67 in the lymph node increased from day 14 to day 18, with another increase in proliferation at Day 25 (Figure 5A). Although the percentage of ESAT-6 specific cells that were proliferating was between 3-8%, these dynamic changes were mirrored when the entire CD4 T cell population in the lymph node was assessed, which had much higher proliferation. ESAT6+CD4 T cell proliferation peaked at day 14 in the lungs, with 60% of ESAT-6+ CD4 T cells staining positively for Ki67, and decreased to ~20% by day 21. Although the pattern of Ki67+ M. tuberculosis-specific CD4 T cells was different between the lung and the lymph node, there were no differences among mice inoculated with low or high dose in either tissue.

Proliferation of GAP+CD8 T cells was also low in the lymph nodes; overall CD8 T cell proliferation was lower than CD4 T cells, but inoculum dose did not affect the proliferation. In lungs, the percentage of GAP+CD8 T cells in the lungs peaked in the high dose mice at day 14, but not until day 18 in the low dose mice (Figure 5B), but returned to 20% by day 21 in both groups. Therefore, increased proliferation is not responsible for the increase in cell numbers in the lungs.

Activation of T cells is increased by inoculum dose in lungs, but not lymph node

CD69 expression has been used as a marker for early activation. Previously published data using mice where large numbers of transgenic T cells were adoptively transferred suggested that a higher dose of infection caused a faster and more robust priming of T cells in the lymph nodes, based primarily on CD69 expression (19). However, in our unmanipulated mouse model, there was no dependence on inoculum dose in frequency of CD69+ CD4 (Figure 6A) or CD8 T cells (Figure 6B), either in the total population or in M. tuberculosis-specific T cells (Figure 6C, D) in the lymph node, except at Day 25. However, the frequency of activated M. tuberculosis-specific T cells was higher than the activation of the total T cell population (Figure 6C & 6D). In the lung, high dose mice had significantly higher percentages of CD69+ CD4 T cells at day 25 (Figure 6A) and of CD69+ CD8 T cells at days 14 and 25 (Figure 6B) compared to low dose infection mice. In the lung, the total number of activated CD4 T cells (ESAT-6 specific and non-specific) from the high dose mice increased over time and are significantly higher than in low dose mice at Days 21 and 25 post-infection (Supplementary Figure 1). The total number of activated CD8 T cells (GAP-specific and non specific) from the high dose group also increases over time and is significantly higher at every time point than the low dose group. There was no difference in the lymph node total number of lymph node CD69+ CD4 or CD8 T cells (both specific and non specific) at any time point examined (Supplementary Figure 1). In our unmanipulated mouse model, we do not see a dose-dependent difference in priming of T cells based on CD69 expression.

Increased migration could explain higher cell numbers in lungs in the high dose infection group

In an effort to explain the increased number of total cells measured experimentally in the lung during a high dose scenario, we used our mathematical model to test whether or not increased inocula have a significant impact on cell death, T cell proliferation, or cell recruitment. We used a baseline trajectory that fits our experimental data, and varied only the dose (as a forcing function). We grouped together all the mechanisms captured in the mathematical model that affect cell death and cell killing in the lung, namely TNF-induced apoptosis of macrophages and T cells, Fas/FasL-induced apoptosis of macrophages, cytotoxic killing and bursting of macrophages. Our analysis indicates that inoculum dose does have a marginal effect on cell death/killing in the lung. We predict that increased inocula result in increased, and not decreased, level of cell death/killing in the first 10 days. Our uncertainty and sensitivity analysis (US/A) shows a significant positive correlation between antigen dose and cell death/killing within the first 7-10 days p.i. (PRCC~0.8, p<0.001). The correlation becomes negative from day 10 until day 30 (PRCC~-0.8, p<0.001). The mathematical model also confirms our experimental finding that dose does not affect T cell proliferation in lung. On the other hand, our US/A results support a dose-dependent recruitment driving higher cell numbers in the lung during infection. Both recruitment terms (i.e., TNF-dependent and TNF-independent recruitment) show a positive correlation to dose (PRCC~0.5, p<0.001 in the first 10 days). Recruitment to the lung increases levels of macrophages, precursor and effector T cells (both CD4 and CD8) T, at higher bacterial doses. Dendritic cell numbers are negatively affected by higher doses (PRCC~-0.75, p<0.001). Figure 7A shows model simulations of total cell numbers in the lung during increased dose challenges (in line with the data of Figure 2B). Figure 7B apportions the increased cell numbers to three different mechanisms, namely cell death/killing, cell recruitment and cell proliferation. Recruitment is by far the most important mechanism in driving cell numbers in the lung (as determined by comparing the total fluxes of cells to the lung due to either recruitment, killing/death and proliferation, Fig. 7B), where cell death/killing and T cell proliferation are influencing total cell numbers in week 1 (death/killing ~13%, mainly for macrophages) and weeks 2 and 3 (T cell proliferation ~10%) week post infection, respectively. Higher doses result in higher cell recruitment and death/killing in the first 10 days post infection, likely driving the higher total cell numbers in the lung later during infection.

Figure 7. Mathematical model simulations of the effect of dose on total number of cells in the lung (Panels A and B) and on T cell priming in the lymph node (Panels C and D).

The x-axes represent days p.i., and the y-axis cell counts. A. Total cell numbers in the lung resulting from a range of CFU forcing functions, spanning from the low to the high dose time courses of Figure 2A. B. Mechanisms contributing to total cell numbers in the lung. Each curve represents the % of a single mechanism over the sum of all three. C. CD4 effector T cell dynamics, after increasing CD4 T cell priming probability (k14) from the baseline value 1e-4 up to 1e-3. D. CD8 effector T-cell dynamics, after increasing CD8 T cell priming probability (k17) from the baseline value 1e-3 up to 1e-2 (see (6) for details on model equations and parameter values).

No correlation between CFU and M. tuberculosis-specific T cells

We studied whether a correlation exists between bacterial numbers and numbers of M. tuberculosis-specific CD4 and CD8 T cells in the lymph nodes and lungs from individual mice, regardless of dose of infection or time point, to determine whether bacterial numbers were affecting T cell priming. There was no correlation with bacterial numbers in the lymph nodes of individual mice with either ESAT-6+ CD4 T cells (Figure 8A) or GAP+ CD8 T cells (Figure 8B) in those same mice, which does not support an influence of inoculum on priming of specific T cells. There was a very modest correlation in the lungs between bacterial load and specific CD4 and CD8 T cells.

Figure 8. There is no correlation between CFU and M. tuberculosis-specific T cells in the lymph nodes or lungs.

Correlation was measured by plotting the linear regression of CFU vs. # of ESAT6+ CD4 (A) T cells or GAP+ CD8 (B) T cells in the LN and lung. R² values are lower in the LN than lung. There does not seem to be a difference in priming in the LN as seen by M. tb specific CD4 and CD8 T cells which does not support the hypothesis that priming in the lymph node is dose dependent. Lines on the graphs are fitted by linear regression. R² represents the goodness of fit and P value represents the significance of slope from non-zero.

Mathematical modeling can replicate adoptively transferred transgenic T cell experiments

We used our mathematical model (16) to address the differences in our experimental findings compared to published data on adoptively transferred transgenic T cells that show an effect of inoculum size on priming of T cells in the lymph node, either in terms of timing or magnitude. The goal was to recapitulate in silico those experiments where adoptively transferred transgenic T cells were injected into mice prior to challenge (with pools of up to 5 × 10⁶ transgenic T cells). In those mouse experiments, inoculum dose did increase priming in lymph nodes (19, 22). To observe an effect of inoculation dose on priming (Figure 7C-D), we had to increase in the model the rates at which CD4 and CD8 T cells are primed (a.k.a. the priming probabilities of both naïve CD4 Ts [k₁₄, Fig. 7C] and naïve CD8 Ts, [k₁₇, Fig. 8D] in the model) by approximately 10-fold from model baseline values (estimated in (16) to match experimental data from our unmanipulated mice). Thus, the mathematical model is able to recapitulate the adoptively transferred transgenic T cells experiments (19, 22), accommodating the effect of higher cognate frequencies of naive cells (e.g. ESAT-6+ CD4 T cells or GAP+ CD8 T cells) by increasing priming probabilities that eventually influence T cell priming in lymph nodes in a dose-dependent manner.

Modeling predicts a key role for IL-10 and DC in T cell priming and in driving effector T cell numbers

We then used the tested mathematical model to explore additional factors that could affect T cell priming in the lymph node and effector T cell numbers in the lungs to help explain the increase of cells in the lungs by 3 weeks post-infection in the high dose infected mice compared to the low dose infected mice, seen experimentally. Our uncertainty and sensitivity results are shown in detail in Tables II and III.

Modeling analyses of our two-compartment model predict two mechanisms that have a strong impact on T cell priming (in lymph node) and on effector T cell numbers (in lung), regardless of antigen dose and T cell type (CD4 or CD8). The first mechanism predicted by the model is the rate of bacterial uptake by, and maturation of, immature DCs in the lung (k₁₂, Table II, positive correlation). The second mechanism predicted is the rate at which resident/resting macrophages differentiate into alternatively activated macrophages in the lymph node (k_7a, Table II, negative correlation), although the latter affects primarily an increase in CD8 T cells. Both mechanisms are only relevant early during infection (i.e., within the first 10 days post-infection).

Higher rates of IDC uptake/maturation increase antigen-bearing DC (MDC) numbers in the lung and induce higher rates of migration of MDCs to the lymph node. On the other hand, a higher rate of alternatively activated macrophage (AAM) activation in the lymph node (k_7a) likely increases IL-10 concentrations (AAMs are a major producer of IL-10). More IL10 in the lymph node has a dual negative impact on priming: the first effect is direct by reducing the proliferation of CD4 and CD8 precursor effector T cells (namely, precursor Th1 and precursor CTL)(17), while the second one is indirect by decreasing/delaying DC maturation (as shown in (5, 17)).

Overall, the model predicts that IL-10 plays a key regulatory role in priming during a high antigen dose scenario. This is consistently observed for both CD4 and CD8 T cell priming in the LN, as well as for effector CD4 and CD8 T cell numbers in the lung. Two IL-10-related mechanisms are significantly affecting priming and effector T cell numbers (Table III): i) regulation of DC maturation rates in the lung¹ and ii) regulation of precursor Th1 proliferation rates in the LN². By decreasing regulation rates of DC maturation or precursor Th1 proliferation, a negative impact on priming occurs for both CD4 and CD8 T cells in lymph nodes, as well as on effector T cell numbers in lung (Table III and Figure 7C). We postulate that this occurs primarily in a high dose inoculation scenario, since more initial uptake by and subsequent infection of alternatively activated macrophages occurs. Increased AAM infection results in a further boost towards an IL-10-dominated environment, which slows down DC maturation as well as T cell proliferation. Alternatively, IL-10 related mechanisms do not affect priming and effector T cell numbers in a low dose scenario. At low antigen doses (Table III), only CD8+ T cells are affected and the mech anisms predicted as significant are in the lymph node and pertain to the efficiency of CD4 T cell priming by an MDC (k₁₄) and the rate of bacterial uptake by immature DCs and consequent DCs maturation in the lymph node (k_12a).

Overall, IL-10 concentration is positively correlated with dose. AAM development is initially positively correlated with dose, then at approximately day 35-40, the correlation becomes increasingly negative on a very fast time scale (data not shown). This suggests that day 35 could be the switching time for an AMM-dominated to a classically activated macrophage-dominated lung environment (6).

Discussion

Priming of T cells and migration of those cells to the site of M. tuberculosis infection is crucial for preventing primary tuberculosis. Previous studies in mice indicated that the numbers of bacilli infecting a host could influence the strength or timing of initiation of the T cell response against this pathogen (3, 12, 19, 22). Here, we sought to address this issue in an unmanipulated mouse model, to determine whether inoculum dose could affect priming of T cell responses in lymph nodes in the presence of the normal but low frequency of naïve precursor T cells. Our data support that within a 10-50 fold range, inoculum size had little effect on priming of M. tuberculosis-specific T cell responses in the lymph nodes. However, in the lungs, higher infection inocula resulted in an overall increase in numbers of immune cells, a modest increase in M. tuberculosis-specific T cells, and higher bacterial numbers. A mathematical model of the lymph nodes and lung during M. tuberculosis infection in a mouse (16) provided support for the experimental data, and identified migration of T cells to the lungs in response to higher infection and antigen load as the major contributor to increased numbers of these cells in the lungs seen experimentally.

It was surprising that an increased inocula, which results in more bacteria in the lungs, and more bacteria at the early time points (days 9-14) in the draining lymph nodes, did not result in increased priming of T cells in the lymph nodes. Higher bacterial numbers translates to more antigen in the lymph nodes, which should increase priming of T cells. Our modeling data provides a possible answer to this conundrum, in that IL-10, induced by higher inocula, is predicted to suppress priming. Thus, the higher antigen levels in high-dose infected mice, which would increase priming, is balanced out by higher IL-10 levels which inhibits priming of T cells; the net result is a level of T cell priming that is similar to that seen in low dose infected animals. In the low dose scenario, IL-10 was not a factor in the priming of T cells.

The role of IL-10 is difficult to confirm in an animal model, since IL-10 can only be present or absent (e.g. in a gene knockout mouse, or by antibody-mediated neutralization). IL-10-/- mice have enhanced IFN-γ responses early in the lungs (9, 10), although specific effects on priming in the LN were not addressed. Depending on the mouse strain, the lack of IL-10 signaling results in reduced bacterial numbers in the lungs (18). In an in vivo system, it is difficult to assess effects of IL-10 on T cell priming independently of other previously reported IL-10 mediated responses. Thus mathematical models can identify and help clarify the roles of more complex interactions such as with IL-10 (21).

We also used the mathematical model to address why our results differ from previously published studies (19, 22). It was reported that increased inocula led to increased T cell priming in lymph nodes. However, these studies were performed by adoptively transferring large numbers of naïve M. tuberculosis antigen-specific transgenic CD4 T cells prior to infection, and analysis of transferred T cells. In contrast, our studies were performed by examining native populations of T cells that were induced upon infection. Transfer of transgenic T cells greatly increases (10,000-1,000,000-fold) the precursor frequency of CD4 T cells. While this certainly makes it easier to detect specific responses using flow cytometry, it may also skew the responses seen. To address whether increasing the precursor frequency could lead to an effect of inoculum dose on priming, we used our mathematical model. This model validation step was designed to show the power of our in silico representation recapitulating the transgenic T cells transfer experiment described in (19, 22). We showed how the model can capture a dose-dependent impact on T cell priming by increasing the priming probabilities of both naïve CD4 Ts (k₁₄) and naïve CD8 (k₁₇) Ts: these probabilities are proxies for many mechanisms related to antigen presentation and priming, one of which is cognate frequency. Thus, higher priming probabilities captures a scenario where transgenic (“cognate”) T cells are present in the system and the mathematical model successfully reproduced the dose-dependent transgenic T cells transfer experiment results, where priming becomes sensitive to antigen dose.

In summary, our data support that although inoculum dose of M. tuberculosis does not substantially influence T cell priming in the lymph nodes, it does affect cell numbers and bacterial burden in the lungs. It is possible that in a vaccinated host, a higher dose challenge may more quickly recall T cell responses to the lungs, since in that setting priming of new T cells is less important. Thus, in vaccine challenge models, it is important to replicate a low dose challenge, since humans are likely exposed to low levels of M. tuberculosis (often repeatedly), rather than large inocula.

Abbreviations used in this paper

AAM

alternatively activated macrophage

MDC

mature dendritic cell

US/A

uncertainty and sensitivity analyses

PRCC

partial rank correlation coefficient

p.i.

post infection