Nitric oxide inhibits the accumulation of CD4+CD44hiTbet+CD69lo T cells in mycobacterial infection

Authors


Abstract

Animals lacking the inducible nitric oxide synthase gene (nos2−/−) are less susceptible to Mycobacterium avium strain 25291 and lack nitric oxide-mediated immunomodulation of CD4+ T cells. Here we show that the absence of nos2 results in increased accumulation of neutrophils and both CD4+ and CD8+ T cells within the M. avium containing granuloma. Examination of the T-cell phenotype in M. avium infected mice demonstrated that CD4+CD44hi effector T cells expressing the Th1 transcriptional regulator T-bet (T-bet+) were specifically reduced by the presence of nitric oxide. Importantly, the T-bet+ effector population could be separated into CD69hi and CD69lo populations, with the CD69lo population only able to accumulate during chronic infection within infected nos2−/− mice. Transcriptomic comparison between CD4+CD44hiCD69hi and CD4+CD44hiCD69lo populations revealed that CD4+CD44hiCD69lo cells had higher expression of the integrin itgb1/itga4 (VLA-4, CD49d/CD29). Inhibition of Nos2 activity allowed increased accumulation of the CD4+CD44hiT-bet+CD69lo population in WT mice as well as increased expression of VLA-4. These data support the hypothesis that effector T cells in mycobacterial granulomata are not a uniform effector population but exist in distinct subsets with differential susceptibility to the regulatory effects of nitric oxide.

Introduction

Infection with Mycobacterium species can result in chronic disease and sustained stimulation of the immune response that needs to be regulated in order to control bacterial growth without destroying host tissue [1]. Nitric oxide has a wide variety of regulatory activities, which can affect the chronic host response to infection [2-5]. In the case of Mycobacterium avium, the bacteria are not susceptible to the toxic effects of nitric oxide [6], allowing us to probe the role of reactive nitrogen intermediates in regulation of the T-cell response to mycobacterial infection without the confounding factor of uncontrolled bacterial growth.

Nitric oxide acts on physiological systems with effects dependent upon concentration, the relative levels of reactive oxygen radicals and pH [7]. At low concentrations, nitric oxide acts as a signaling molecule, either in a cGMP-dependent or -independent manner, to promote vascular integrity, mediate neurotransmission, and regulate cellular respiration by altering the affinity of cytochrome C for oxygen [7, 8]. At high concentrations, nitric oxide inhibits respiration and causes nitrosative damage to proteins, lipid peroxidation, and DNA [9, 10]. The balance between nitric oxide and oxygen radicals is important, as nitric oxide can reduce oxidative stress [11] but also generates peroxynitrite, which is itself damaging [12]. The damage generated by high levels of nitric oxide is detrimental to cells and results in apoptosis [9].

The impact of nitric oxide on the immune response has been extensively analyzed with identification of both positive and negative regulatory roles [13]. In humans, nitric oxide limits IL-2 release and proliferation of T cells via activation of the cGMP-dependent protein kinase, cGK I [14]. In Trypanosoma brucei mouse models, nitric oxide inhibits the accumulation of IL-2- and IFN-γ-producing T cells [15]. In both an in vitro system [16] and a Listeria monocytogenes mouse model [17], the inhibition of nitric oxide synthase (Nos) results in improved antigen-specific T-cell responses. Nitric oxide also acts as an anti-inflammatory agent by limiting the interaction of leucocytes with the endothelial monolayer [18]. Nitric oxide can drive IL-10-producing regulatory T cells, limit the expansion of Th17 cells [19, 20], and regulate the IL-12 pathway both positively [21] and negatively [22]. Indeed, at low levels, it can augment the generation of Th1 cells by increasing expression of IL-12Rβ2 [23, 24] and augment IFN-γ−mediated signaling [25].

In mycobacterial disease, nitric oxide is essential for the control of Mycobacterium tuberculosis but dispensable for the control of M. avium [4]. It limits the accumulation of activated T cells in the Mycobacterium bovis BCG model [26], the M. tuberculosis model [27], and the M. avium model [6] with an increased IFN-γ response being seen in both M. avium [6] and M. tuberculosis infected nos2−/− mice [28]. Absence of nitric oxide in M. avium infection results in lesions with increased cellularity and collagen deposition [6, 29, 30]. Thus, nitric oxide acts to limit T-cell responses and immunopathology during mycobacterial infection and in the absence of nitric oxide, there is improved IFN-γ production [6].

The impact of nitric oxide on specific subsets of activated T cells has not been extensively studied; however, recent data shows that while some antigen-specific CD4+ T-cell effectors are able to persist within the mycobacterially induced inflammatory environment, other effector cells are not [31]. Specifically, T cells that can produce IFN-γ but which maintain the capacity to proliferate are better able to persist in mycobacterially infected mice than are T cells with higher IFN-γ production but lower proliferative capacity [31]. As nitric oxide is known to be involved in both initiation and regulation of the IFN-γ-producing CD4+ T-cell population, we investigated whether different subsets of effector CD4+ T cells were differentially susceptible to nitric oxide during mycobacterial disease.

Results

The bacterial burden and the granuloma in WT and nos2−/− mice differ

We examined bacterial burden and granuloma formation following a moderate intravenous dose of M. avium 25291. Figure 1A demonstrates that growth of M. avium 25291 was reduced in nos2−/− mice compared with that in wild-type (WT) mice and that cellular inflammation was different between the two groups [30, 32]. There was a preponderance of mononuclear phagocytes with large cytoplasm in the WT mice (Fig. 1B) while the lesions in the nos2−/− mice were more circumscribed with macrophages and lymphocytes forming a mantle around a central area of neutrophil-like cells (Fig. 1C). These data confirm that the WT and nos2−/− mice differ in response to M. avium 25291 with impaired bacterial control in WT mice and more complex granuloma development in the nos2−/− mice.

Figure 1.

Bacterial burden and inflammation differ in Mycobacterium avium infected WT and nos2−/− mice. (A) WT (black bars) and nos2−/− (white bars) mice were infected intravenously with 106 M. avium 25291, livers were harvested on the indicated days and the number of colony forming units determined. Data are shown as mean + SEM of n = 5 and are representative of three independent experiments. ***p < 0.001, Student's t-test. (B, C) Livers from infected (B) WT and (C) nos2−/− mice were also processed for histological analysis on day 120 using hematoxylin and eosin (original magnification 200×). Arrows in (C) show margin of granuloma and central area of granulocytic cells not present in (B). Data shown are representative of five animals per group and two similar experiments.

Absence of Nos2 promotes lymphocytes and granulo-cytes accumulation in the M. avium induced lesion

To better define the cells within the WT and nos2−/− lesions, we probed live sections of infected liver tissue with antibody specific for macrophage (F4/80), neutrophil (Ly6G), and lymphocyte (CD4 and CD8) markers.

We found greater numbers of CD4+ or CD8+ cells throughout the F4/80+ macrophage defined lesion in the nos2−/− mouse (Fig. 2B) compared with the WT mouse (Fig. 2A). Further, there were significantly more Ly6G+ cells within the nos2−/− lesions (Fig. 2D) compared to the WT lesions (Fig. 2C) and these appeared to coalesce in central areas (Fig. 2D). These data show that both lymphocytes and neutrophils accumulate more readily within the macrophage-defined lesions of M. avium infected nos2−/− compared to WT mice.

Figure 2.

The absence of Nos2 promotes accumulation of lymphocytes and granulocytes in the Mycobacterium avium induced lesion. WT (A, C, E, G) and nos2−/− (B, D, F, H) mice were infected intravenously with 106 M. avium 25291 and the livers harvested on day 120. (A–D) Livers were placed directly in agar and rapidly sectioned without fixation using a vibrating microtome. Tissue slices were stained with the F4/80 antibody (blue in A–D) and anti-CD4/CD8 antibody (red in A and B) or anti-Ly6G antibody (red in C and D, original magnification 200×). Data shown are representative of six total animals over three experiments. (E–H) Livers were also harvested for cryosectioning and frozen sections were stained for CD4 (red) CD8 (blue) and either p22-phox (green, E, F) or Nos2 (green G, H, original magnification 200×). Nuclear stain is white. Data shown are representative of four animals per group, and of two experiments performed.

As lymphocytes were absent from the WT lesions, we wanted to compare the environment created within the F4/80 dominated lesions of the WT and nos2−/− mice. To do this, we stained cryosections from infected WT and nos2−/− livers for the enzymes required to generate toxic oxygen and nitrogen radicals. We found that p22-phox, a critical subunit of the NADPH oxidase required for oxygen radical generation [33], was readily expressed throughout the phagocyte areas of both WT (Fig. 2E) and nos2−/− mice (Fig. 2F). The Nos2 protein was less widely expressed in the WT lesions (Fig. 2G) and was absent in the nos2−/− lesions (Fig. 2H). Both CD4+ (red) and CD8+ (blue) cells were seen in the p22-phox positive WT tissues (Fig. 2E) but were much more prevalent in the p22-phox area in the nos2−/− tissues (Fig. 2F). The CD4+ T cells were significantly increased in the sections from nos2−/− livers with an average of 321±100 CD4+ cells per section versus an average of 93±29 cells per section in the WT (p = 0.0046 by Student's t-test). These data demonstrate that while Nos2 is not as widely expressed as p22-phox, it severely affects the ability of CD4+ and CD8+ lymphocytes to accumulate within the mycobacterial granuloma.

Activated effector lymphocytes within infected organs are differentially affected by nitric oxide

Mycobacterium avium infected WT mice undergo a profound IFN-γ-dependent depletion of lymphocytes; however, the impact of Nos2 in this model is not to substantially deplete T cells but to reduce the level of the IFN-γ response [6, 34]. Figure 2 suggests that T cells are specifically excluded from the phagocytic areas in M. avium infected WT mice in a nitric oxide-dependent manner. To determine whether the histological results in the WT lesions represented the depletion of all or a specific subset of lymphocytes from the affected organ, we compared the CD4+ T cells within infected organs by flow cytometry. We found only a modest effect of nos2 deficiency on the total frequency and number of either live lymphocytes or CD4+ T cells in infected organs compared to WT mice (Supporting Information Fig. 1). This trend was seen before but had not reached statistical significance in previous studies [6, 34].

To determine whether the nos2 gene was adversely affecting activated effector cells, we compared the frequency (Fig. 3A) and number (Fig. 3B) of CD4+ T cells expressing the Th1-associated transcription factor, T-bet. We found that the CD4+ T-bet+ population was significantly and substantially increased in the nos2−/− mice relative to the WT mice in all infected organs (Fig. 3). These data demonstrate that the presence of nos2 limits the accumulation of Th1-type T cells and that these activated effector cells were either more susceptible to depletion or failed to develop in the presence of Nos2.

Figure 3.

The number of CD4+ T-bet+ T cells is increased in the organs of Mycobacterium avium infected nos2−/− mice. WT (WT) and nos2−/− mice were either left uninfected (Uninf) or infected (Inf) intravenously with 106 M. avium 25291 and spleen, lungs, and livers harvested from day 60 through day 120. The organs were processed for flow cytometry and the (A) frequency and (B) number of CD4+T-bet+ cells within the organs determined. Each symbol represents an individual sample and data are shown as mean ± SEM of n = 4–22, pooled from the five (spleen), four (lung), or two (liver) experiments performed. *p < 0.05, **p < 0.01, ***p < 0.001, ANOVA.

To investigate whether all activated T-bet+ cells were equally affected by the presence of nos2, we stained CD4+ T cells from all infected organs for both T-bet and CD69, a molecule that is upregulated upon antigen exposure [35]. The pattern of staining is shown in Fig. 4A. We found that in all three organs, the frequency and number (Fig. 4B) of CD69hi T-bet+ CD4+ T cells were only modestly affected by the absence of nos2. In contrast, the CD69loT-bet+ CD4+ T-cell population failed to accumulate in the WT mice but did accumulate in the spleen, liver, and lung of the nos2−/− mice (Fig. 4C). These data demonstrate that the nos2 gene has the capacity to limit accumulation of CD69loT-bet+ CD4+ T cells.

Figure 4.

A subset of CD4+T-bet+ cells accumulates in the absence of nos2. WT (WT) and nos2−/− mice were either left uninfected (Uninf) or infected (Inf) intravenously with 106 Mycobacterium avium 25291 and spleen, lungs, and livers harvested from day 60 through day 120. (A) Live lymphocytes were gated using doublet discrimination, CD3+, and CD4+ expression (Supporting Information Fig. 2A) and then analyzed for T-bet and CD69 expression. The frequency and number of (B) CD4+T-bet+CD69hi and (C) CD4+T-bet+CD69lo were calculated. Each symbol represents an individual sample and data are shown as mean ± SEM of n = 4–22, pooled from the five (spleen), four (lung), or two (liver) experiments performed. *p < 0.05, **p < 0.01, ***p < 0.001; ANOVA.

CD4+CD44hiCD69hi and CD4+CD44hiCD69lo cells have distinct transcriptomes in vivo

To define the basis of the increased susceptibility to nos2 of the CD69loT-bet+ cells relative to the CD69hiT-bet+ cells, we compared gene transcription immediately ex vivo in CD4+ T cells sorted based on CD44 and CD69 expression using an unbiased microarray approach. Equivalent numbers of cells differing only in the expression of CD69 (CD4+CD44hiCD69hi versus CD4+CD44hiCD69lo), were purified using fluorescence-activated cell sorting from the splenocytes of WT and nos2−/− mice infected with M. avium 80 days previously (all CD44hi cells are T-bet+ in this model, data not shown). Global RNA expression was analyzed for differential gene expression and class comparison (Fig. 5). We found that there was differential expression between the four populations of cells of 911 sequences detected by unique probes and that the patterns for individual mice within each group were reproducible (Fig. 5A). Importantly, we found that gene expression patterns were associated with both genotype of the mouse (WT versus nos2−/−) and phenotype of the cells (CD69hi versus CD69lo) and that there were differences between the gene expression patterns for the CD4+CD44hiCD69lo cells isolated from WT and nos2−/− mice (black arrows in Fig. 5A). The log intensity values of the microarray data set are available in Supporting Information Table 1.

Figure 5.

CD4+CD44hiT-bet+CD69hi and CD4+CD44hiT-bet+CD69lo cells represent populations with distinct transcriptomes in vivo. WT and nos2−/− mice were infected intravenously with 106 Mycobacterium avium 25291, spleens were harvested on day 80, and the CD4+CD44hiCD69hi and CD4+CD44hiCD69lo populations separated by FACS (Supporting Information Fig. 2B). mRNA extracted from the sorted cells underwent microarray analysis to define differential gene expression between the populations as well as between WT and nos2−/− mice (i.e. four distinct classes). (A) Differential gene expression between the WT and nos2−/− genotypes and the CD69hi and CD69lo populations (i.e. four classes). Array data were centered and normalized whereupon gene expression values were filtered to exclude genes with less than 1.5 differential expression (F-test using a p-value threshold of 0.005). Genes were clustered based on patterns of expression without bias. (Arrows mark gene expression clusters that differ between WT and nos2−/− CD4+CD44hiCD69lo populations.) (B) Biological pathway analysis was performed on the four classes of array data by screening 3079 filtered unique probes against 218 predefined gene lists (BioCarta Pathways). The “Adhesion Molecule Expression on Lymphocytes” pathway was over-represented in the data set compared with that from random chance (p = 0.00031, LS Permutation test). Multiple probes targeted the same gene resulting in three replicates of Itgb1 and four replicates of Itga4.

To probe the data sets for biological relevance, we compared the differential gene expression data against 218 predefined gene lists representing previously investigated mouse biological processes. Two pathways were identified as being significantly represented in the differentially expressed data set and both contained the genes for the heterodimeric integrin known as very late antigen-4 (VLA-4, CD49d/CD29) (Fig. 5B).

By further comparing specific gene expression within the individual samples (n = 3), we were able to define statistically different gene expression for genes of interest. We found that the CD4+CD44hiCD69lo population from both the WT and nos2−/− infected mice expressed less il2, il2ra, il2rb, and ifngr2 than did the CD4+CD44hiCD69hi population (Table 1). By comparing the expression of genes between cell subsets from the WT and nos2−/− mice, we found that bcl2 expression was reduced in the absence of nitric oxide for both of the types of cells (Table 1). However, only within the CD4+CD44hiCD69lo population was there an impact of nitric oxide on the expression of il4 and to a lesser degree on il4ra (Table 1). Interestingly, there is no difference in the expression of the tbx21 (T-bet) or gata3 master regulators for IFN-γ and IL-4 within these populations (data not shown). Taken together, the data support the fact that the activated effector cells within the mycobacterial granuloma can be grouped into potentially functional subsets by surface markers. In particular, the CD4+CD44hiCD69hi population may represent an IL-2-producing and IL-2- and IFN-γ-responsive, potentially proliferating population whereas the CD4+CD44hiCD69lo may be unresponsive to IL-2 and IFN-γ. The strong need for nitric oxide in induction of IL-4 and reduction of VLA-4 in the CD4+CD44hiCD69lo population may reflect a novel regulatory mechanism for the action of nitric oxide on Th1 effector function.

Table 1. Specific gene expression in CD4+CD44hi populations differentially expressing CD69 isolated from the spleens of Mycobacterium avium infected WT and nos2−/− mice
GeneabWT CD69lo Average ± SDcp versus nos2dp versus hinos2−/− CD69lo Average ± SDp versus hiWT CD69hi Averagey ± SDp versus nos2nos2−/− CD69hi Average ± SD
  1. a

    Gene represented by specific probes on array. Some genes (marked #) are detected by several probes, in this case, a probe result representative of all probes for that gene is shown. The probe used is listed under the gene name.

  2. b

    RNA from FACS sorted cells of specific CD4+CD44hiCD69 variable phenotype from wild type (WT) and nos2−/− mice infected 80 days previously with M. avium

  3. c

    The log2 expression signal for each of the individual samples within a class (n = 3) was measured and the average ± SD for each class is shown.

  4. d

    The values from samples within each group were compared to the values from the other groups and the statistical difference between the means was determined by the Student's t-test (i.e. between genotypes WT versus nos2−/− and phenotypes CD69lo versus CD69hi).

  5. e

    For this comparison, only one or two probes showed significance and the majority showed no significance.

il2 1449990_at3.63 ± 0.76ns 4.20 ± 0.11 6.38 ± 0.59*4.94 ± 0.11
   ** **   
il2ra(#) 1420691_at4.97 ± 0.20ns 4.86 ± 0.13 6.68 ± 0.05ns7.13 ± 0.31
   *** ***   
il2rb(#) 1417546_at7.73 ± 0.51ns 8.38 ± 0.30 8.66 ± 0.13**9.50 ± 0.25
   * **   
il4 1449864_at6.58 ± 0.23*** 4.36 ± 0.34 7.26 ± 0.36*5.88 ± 0.65
   * *   
il4ra 1421034_a_at8.85 ± 0.17** 7.91 ± 0.19 8.73 ± 0.19ns9.07 ± 0.14
   ns **   
Ifng 1425947_at12.3 ± 0.91ns 13.5 ± 0.15 11.8 ± 0.66ns12.5 ± 0.35
   ns **   
ifngr2 1423557_at5.48 ± 0.55* 3.97 ± 0.47 5.52 ± 0.24***3.87 ± 0.20
   ns ns   
bcl2(#) 1440770_at9.81 ± 0.36** 7.73 ± 0.53 8.10 ± 0.06**6.98 ± 0.22
   ** nse   
itgb1(#) 1426918_at8.54 ± 0.27** 10.4 ± 0.46 8.70 ± 0.08nse8.29 ± 0.26
   ns **   
itga4(#) 1421194_at8.11 ± 0.1*** 9.33 ± 0.07 8.05 ± 0.07nse8.42 ± 0.05
   ns ***   

Nitric oxide inhibition restricts accumulation of CD69loT-bet+ CD4+ T cells and expression of VLA-4

To confirm that nitric oxide was the active agent limiting the specific subset of activated CD4+ T cells in M. avium infected mice, we delivered a specific inhibitor of nitric oxide synthase activity to infected mice and monitored the development of specific T-cell subsets. In these experiments, we treated mice early in infection, as we wanted to be able to detect the CD69loT-bet+ CD4+ T-cell population in the WT mice. Mice were infected and either left untreated or treated with aminoguanidine [36] from day 0 to day 30 or day 20 to day 30 and the phenotype of the activated CD4+ T-cell population in the infected organs determined by flow cytometry. We found that both CD69hiT-bet+ and CD69loT-bet+ CD4+ T cells could be detected in the organs of untreated infected mice (Fig. 6A) and that the frequency of CD69hiT-bet+ CD4+ T cells was either unaffected (spleen) or modestly reduced by aminoguanidine treatment (lung and liver) (Fig. 6A, top panels). In contrast, the frequency of CD69loT-bethi CD4+ T cells significantly increased in the organs of all treated mice (Fig. 6A, lower panels). These data indicate that the presence of nitric oxide influences the expression of CD69 in activated CD4+ T cells in all infected organs.

Figure 6.

Nitric oxide inhibition results in increased accumulation of CD4+T-bet+CD69lo cells and increased expression of VLA-4 on T-bet+ CD4+ T cells. WT mice were either left uninfected (UnInf) or infected (Inf) intravenously with 106 M. avium 25291 and either left untreated (UnTx) or treated (Tx) with the Nos2 inhibitor aminoguanidine throughout the infection (0–30), or for the last 10 (20–30) days of the infection. (A) Spleens, livers, and lungs were harvested on day 30, the organs processed for flow cytometry and the frequency of T-bet+CD69hi (top) and T-bet+CD69lo (bottom) in the CD4+ population determined. (B) The expression of T-bet and CD49d were also assessed and the frequency of T-bet+CD49d+ cells within the CD4+ population in the organs of infected and treated WT mice determined. Data points are for individual mice, data are shown as mean ± SEM of n = 10, and are pooled from two experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ANOVA.

Based on the array data (Fig. 5B), we also wanted to determine the impact of nitric oxide on the expression of the VLA-4 marker in CD4+ T cells by measuring the expression of the inducible subunit CD49d in the infected and aminoguanidine-treated mice. Using flow cytometry, we found that infection resulted in increased expression of VLA-4 on T-bet+ CD4+ T cells and that inhibition of nitric oxide generation resulted in further increased expression of this marker (Fig. 6B, upper panels). There was also an increased frequency of VLA-4+Tbet+ cells in the aminoguanidine-treated mice (Fig. 6B, bottom panels). These data demonstrate that nitric oxide limits the expression of VLA-4 on the activated CD4+ T cells in mycobacterially infected mice.

Discussion

In the low dose model of infection, M. avium strain 25291 generates highly necrotic lesions resembling those induced by M. tuberculosis in humans and the development of necrosis is entirely dependent upon IFN-γ, IL-12p40, and CD4+ T cells [32, 37]. In high-dose infection models, this strain of M. avium results in loss of CD4+ T cells [38] and phagocyte-dominated lesions that do not become necrotic [32]. We wanted to determine whether the impact of nitric oxide on the granulomatous response to M. avium 25291 was related to the impact of nitric oxide on CD4+ Th1 cells. The data show that in the absence of Nos2, the inflammatory site in the liver of M. avium 25291 infected mice is indeed altered. The WT lesion is characterized by the accumulation of F4/80+, p22-phox+ monocytic phagocytes. In contrast, the absence of Nos2 resulted in increased cellular diversity in the granuloma with F4/80+ p22-phox+ staining phagocytes associating with many more lymphocytes and neutrophils accumulating in foci within the lesions. Thus while the WT mice did not control the bacterial growth as effectively as the nos2−/− mice, the lesions that developed were less complex and showed no sign of incipient necrosis. In the absence of Nos2, we show that activated T cells expressing the Th1-associated T-bet transcription factor and which are low in expression of CD69 but high in expression of VLA-4, accumulated to a much higher degree within the lesions. This accumulation of activated effector T cells was associated with the formation of a complex granuloma.

The importance of the CD4+ T-cell population during granuloma development and control of mycobacterial infection make understanding the regulation of this population an important goal. Previous data has shown that there is an increase in the IFN-γ response in infected nos2−/− mice [6] and our data complement this by showing the increased accumulation of T-bet expressing cells in the absence of nos2−/−. These T-bet expressers are likely high producers of IFN-γ [39] and their accumulation will contribute to the higher circulating level of IFN-γ in infected nos2−/− mice. It has been reported that IFN-γ and nitric oxide regulate the T-cell response in mycobacterial disease [4] but the details of this control are not fully defined. IFN-γ serves to drive T-cell apoptosis during mycobacterial infection via direct and indirect effects [26, 40] and protection against IFN-γ-induced autophagy is mediated by lrgm1 [41]. We have previously shown that in vitro generated effector cells, regardless of antigen specificity, are susceptible to the IFN-γ-mediated detrimental effects of the conditions induced by M. avium strain 25291 [34]. We now show that there is a specific subset within the pool of activated T cells that is more susceptible to nitric oxide and that these T cells can be characterized by a distinct phenotypic and transcriptional profile.

The potential function of the CD4+CD69hi and CD4+CD69lo populations in the mycobacterial granuloma is addressed by the data presented here. In particular, the IL-2 data suggest that the CD4+CD44hiCD69hi cells are more likely to be able to proliferate and that the CD4+CD44hiCD69lo cells are more akin to the highly differentiated cells seen in the tuberculosis model [31]. Similarly, higher expression of the apoptosis-related bcl2 [42] in the WT populations compared with the nos2−/−-derived populations suggests that nitric oxide does promote apoptosis in these effector cells. Most intriguing, however, is the strong difference seen in the CD4+CD44hiCD69lo population with regard to VLA-4 and IL-4 in the absence of nitric oxide. IL-4 has been shown to limit VLA-4 expression on activated CD4+ T cells and this reduces migration of cells into lesional sites [43-45]. Further, upregulation of VLA also increases pathogenicity of T cells [46] and improves SLP-76 interaction with ZAP-70 within the immunological synapse [47]. Thus, it is possible that in the absence of nitric oxide, the CD4+CD44hiCD69lo cells produce less IL-4 and it is this that allows the observed greater VLA-4 expression. The published data also support the hypothesis that increased VLA-4 will allow for improved in vivo function and improved ability to accumulate within the granuloma.

One could propose therefore that the level of nitric oxide within the lesional site can dramatically impact the local protective and immunopathological response by reducing accumulation of specific subsets of activated effector cells and by altering the potency of the lymphocytes with regard to accumulation within the lesion and cytokine production. By demonstrating the differential impact of nitric oxide on distinct functional subsets of cells, we have identified a mechanism whereby protection and pathology in mycobacterial disease are modulated by nitric oxide.

Conclusion

The development of inflammation during mycobacterial infection is an important component of the disease process and is actively modulated by CD4+ T cells. Herein, we demonstrate that within the pool of effector T cells, there is an activated T-cell subset that is more susceptible to the regulatory factors active within the granuloma. Defining the relative protective and pathological role of this activated T-cell subset (CD4+T-bet+CD69loVLA4hi) will allow for improved vaccination and immunotherapeutic intervention.

Materials and methods

Mice

All mice were bred at the Trudeau Institute and were treated according to National Institutes of Health and Trudeau Institute Animal Care and Use Committee guidelines. All animal protocols used herein were approved by the Trudeau Institute Animal Care and Use Committee. C57BL/6 and B6.129P2-nos2tm1Lau (nos2−/−) (originally purchased from JAX Mice, Maine) were used in these experiments.

Experimental infection and enumeration of viable bacteria

Mice were infected with M. avium 25291 (ATTC) at a dose of 1 × 106 cfu by lateral tail vein injection. The level of bacteria in specific organs was determined by homogenizing the organs and plating on agar and counting colonies [48, 49]. Some infected WT mice were treated with aminoguanidine hemisulfate (Sigma-Aldrich) at 2.5 g/100 mL in the drinking water for defined periods of time; control mice received water without drug.

Histology

Liver sections were placed in 10% neutral-buffered formalin, blocked in paraffin, processed for light microscopy, and stained with hematoxylin and eosin to provide cell structure. For immunofluorescence staining, liver sections were harvested into cold HBSS and 3–4 mm sections cut with a scalpel. Sections were placed in 4% low-melt agarose (Lonza Seaplaque Agarose, Fisher Scientific) in HBSS. The solidified gel containing sections of liver was then sectioned using a vibrating microtome cooled to 4°C (Leica VT1000) and 200-micron sections were collected into 12-well plates containing HBSS, FcBlock, 5% normal mouse serum. Sections were stained with fluorescently labeled antibodies, anti-CD4 PE (RM4–5), anti-CD8 PE (clone 53–6.7), Ly6G PE (clone 1A8) (all from BD Biosciences, CA, USA) and F4/80 Alexa647 (clone Cl:A3–1, AbD Serotec, NC, USA) were added to the well at a dilution of 1:200 except for F4/80 which was used at 1:75. Tissues were incubated for 2 h on ice and then washed twice with excess PBS for 15 min each. Cryosections were generated from liver tissue harvested in Tissue-Tek which were then air dried, fixed with neutral-buffered formalin, blocked with 10% normal mouse serum/1% Triton X-100/1% Tween-20 and exposed to the following fluorescently labeled antibodies–CD8 allophycocyanin (clone 53–6.7, eBioscience, CA, USA), CD4 PE (as above), polyclonal rabbit anti-p22-phox (Santa Cruz Biotechnology, CA, USA), polyclonal Rabbit anti-iNOS (BD Transduction Laboratories, CA, USA) and anti-Rabbit 488 (Invitrogen, NY, USA). Sections were also exposed to Hoechst DNA stain. All sections were exposed to appropriate laser light using the Leica SP5 confocal (Leica Microsystems, Germany) and the light emissions detected using photomultiplier tubes (PMTs) of the appropriate bandwidth. Emission spectra were collected using sequential scanning to avoid spectral bleed-through. The data were collected as Leica image files using LAS-AF version 2.2.1 software (Leica) and converted into TIFF using Fiji software (http://fiji.sc/wiki/index.php/Fiji). Sections were incubated with either CD4/CD8 and F4/80 antibodies or Ly6G and F4/80 antibodies.

Cell preparation and flow cytometry and cell sorting

Lungs of experimental mice were perfused with cold saline containing heparin and placed in cold DMEM (Mediatech-Cellgro). Livers and spleens were taken directly from experimental mice and placed in cold DMEM. All organs were then sectioned using fresh sterile razor blades and placed in DMEM containing collagenase IX (0.7 mg/mL; Sigma-Aldrich) and DNase (30 μg/mL; Sigma-Aldrich) at 37°C for 30 min [49, 50]. Digested tissue was gently dispersed by passage through a 70 μm pore size nylon tissue strainer (Falcon; BD Biosciences); the resultant single-cell suspension was treated with Gey's solution to remove any residual RBC, washed twice, and counted. The liver cells were further processed over a 40%:80% Percoll (GE Healthcare) gradient and then washed and counted. Cell suspensions were stained for surface markers, washed, processed for intracellular staining using the eBioscience “Transcription factor staining buffer set” (eBioscience) according to the manufacturer's instructions and then stained for T-bet. The antibodies were titrated for use and consisted of anti-CD3 (Clone 17A2) labeled with eFluor450, anti-CD4 (clone RM4–5) labeled with PerCP-Cy5.5, anti-CD69 (clone H1.2F3) labeled with PE-Cy7, anti-CD44 (clone IM7) labeled with allophycocyanin-eFluor780, and anti-T-bet (clone 4B10) labeled with PE (all from eBioscience). Data from stained cells were collected using Diva software on an LSRII flow cytometer (BD Biosciences) and analyzed using FlowJo software (Tristar) and the gating system is shown in Supporting Information Fig. 2A. Cells stained as above were also sorted to purity on a BD Influx, 11 color, four laser system based on live lymphocyte, CD4 positive, CD44 high, and then separated based on the expression of CD69. An example of the purity of the sorted populations is shown in Supporting Information Fig. 2B.

Preparation of nucleic acid and microarray analysis

RNA was extracted from purified populations and DNA removed with the RNeasy Plus Mini Kit (Qiagen, CA, USA) according to the manufacturer's instructions. RNA was quality tested and the yield was between 3.4 and 98 ng/uL per sample. The isolated RNA was used to generate cRNA which was then biotinylated and prepared according to the Affymetrix GeneChip 3′ IVT Express Protocol from 150 ng of total RNA. Following fragmentation, 10 μg of cRNA was hybridized for 16 h at 45°C on Mouse Genome 430 2.0 arrays. Arrays were washed and stained in the Affymetrix Fluidics Stations 450. The arrays were scanned using an Affymetrix GeneChip Scanner 3000 7G. Initial QC was performed with Affymetrix Expression Console using the RMA algorithm with quantile normalization and general background correction.

BRB-Arraytools was used for statistical analysis and results visualization [51] as described [52, 53]. Preprocessing with robust multi-array average with GC-content background correction was performed on all CEL files to provide background correction using probe sequence and GC content, quantile normalization, and a robust multichip model fit using median polish [54] (Supporting Information Table 1). GC-content background correction was selected from various other background correction algorithms because it yielded the minimum intraclass variation for a subset of experimentally relevant genes [55]. Spot filters, normalization, gene filters, and gene subsets: No spot filtering was applied. Log2 normalization was applied. The median array was used as a reference array. The default gene filters were applied which excluded genes having less than 20% of their expression values and having at least a 1.5-fold change from the median expression value or if greater than 50% of the expression values are missing. Only named genes, that is, those without “NA” as their annotated gene identifier were analyzed, thus excluding nonspecific probes and array-specific controls. Following these processes, 3366 specific probes were identified as differentially expressed with 3079 named genes identified. Gene annotation: Genes were annotated using the Affymetrix HT_MG-430B Array (mouse4302) in BRB-Arraytools. Class comparison: Class comparison was then used to identify specific genes whose expression correlated with the experimental group (i.e. WT CD69lo, WT CD69hi, nos2−/−CD69lo or nos2−/−CD69hi) using a univariate F-test at a significance threshold of p = 0.001 that yielded 911 genes.

Gene set class comparison was used to identify biologically relevant pathways by comparing the set of experimentally identified differentially expressed genes with 218 predefined BioCarta pathway gene lists (biocarta.com) representing previously investigated biological pathways. Gene set class comparison identifies biological pathways that are over-represented in the experimental data by comparing the number of differentially expressed genes for a given BioCarta pathway with that expected by random chance alone. The significance threshold for this test was p = 0.005 using a univariate F-test to define differentially expressed genes (as above) with an LS permutation test used to identify BioCarta gene sets having more genes differentially expressed among the phenotype classes than expected by chance. Of the 218 BioCarta gene lists tested, 107 gene lists contained one or more differentially expressed genes, and of these BioCarta gene lists, two were identified as significantly enriched for differentially expressed genes: “Adhesion Molecules on Lymphocytes” and “Monocyte and its Surface Molecules,” containing 11 and 12 genes, respectively. When examined, these two gene sets contained 11 of 12 identical genes.

Hierarchical clustering of genes was used to survey the differentially expressed genes to identify global patterns of expression. To perform this analysis, the genes were centered and scaled, using one-minus correlation with average linkage computed.

Statistics

Differences between the means of experimental groups were analyzed using the two-tailed Student's t-test or ANOVA as appropriate. Differences were considered significant where p ≤ 0.05. Inherently logarithmic data from bacterial growth were transformed for statistical analysis.

Acknowledgements

This work was supported by the Trudeau Institute, Inc., NIH grants AI46530 and AI069121 and an American Lung Association DeSouza Award to AMC.; PTDC/SAU-MII/099102/2008 from the FCT (Fundação para a Ciência e a Tecnologia) to RA. The Authors would like to thank Flow Cytometry Core and the Imaging Core at Trudeau Institute and Phyllis Spatrick at the Genomic Core Facility at UMASS Medical School for excellent technical support.

Conflict of interest

The authors declare no financial or commercial conflict of interest.

Abbreviation
VLA-4

very late antigen-4

Ancillary