Dispensability of Surfactant Proteins A and D in Immune Control of Mycobacterium tuberculosis Infection following Aerosol Challenge of Mice

Abstract

Surfactant proteins A and D (SP-A and -D) play a role in many acute bacterial, viral, and fungal infections and in acute allergic responses. In vitro, human SPs bind Mycobacterium tuberculosis and alter human and rat macrophage-mediated functions. Here we report the roles of SP-A and SP-D in M. tuberculosis infection following aerosol challenge of SP-A-, SP-D-, and SP-A/-D-deficient mice. These studies surprisingly identified no gross defects in uptake or immune control of M. tuberculosis in SP-A-, SP-D-, and SP-A/-D-deficient mice. While both SP-A- and SP-D-deficient mice exhibited evidence of immunopathologic defects, the CD11b^high CD11c^high dendritic cell populations and the gamma interferon (IFN-γ)-dependent CD4⁺ T cell response to M. tuberculosis were unaltered in all genotypes tested. Together, these data indicate that SP-A and SP-D are dispensable for immune control of M. tuberculosis in a low-dose, aerosol challenge, murine model of tuberculosis (TB).

Mycobacterium tuberculosis is a facultative intracellular pathogen that infects and resides in humans as its only known host and reservoir (14, 29). M. tuberculosis is most commonly transmitted by inhalation of infectious aerosols into the lung with deposition in the terminal bronchioles and alveoli (28). Early interactions of M. tuberculosis with the innate immune environment of the lung have been proposed to determine the outcome of M. tuberculosis infection (14, 29, 41), yet knowledge of such interactions remains incompletely described.

One of the first interactions M. tuberculosis encounters in the lung is the binding of pulmonary surfactant (PS) molecules. PSs are multimolecular complexes composed of lipids and proteins secreted by alveolar type II cells and Clara cells in distal bronchioles (41, 45). PSs function chiefly to reduce the surface tension of alveolar fluid and to facilitate reversible expansion. However, in addition to such mechanical functions, several protein components of pulmonary surfactant have been shown to exert immunomodulatory actions that enhance immune control of respiratory pathogens and help minimize inflammatory damage (15, 45).

These immunoregulatory properties are chiefly mediated by surfactant-associated proteins A and D (SP-A and SP-D) (41, 45). These two large hydrophilic proteins consist of a collagen-like domain in the N terminus that mediates oligomerization, a coiled-coiled area (neck), and a globular Ca⁺-dependent carbohydrate recognition domain at the C terminus (45) and have the ability to bind exposed carbohydrate residues on the surface of M. tuberculosis (9, 11).

The C-terminal lectin-binding domain of SP-A and SP-D is important in some but not all antimicrobial functions and varies in specificity for glycosylated targets (41, 45). In the case of M. tuberculosis, SP-D has been shown to bind mannosylated lipoarabinomannan (ManLam) from the Erdman strain (11), while SP-A has been shown to bind a wider range of mycobacterial targets, including ManLam (from virulent and avirulent mycobacterial strains) (18), lipomannan (39), a 60-kDa glycoprotein (30), and the M. tuberculosis surface glycoprotein Apa (33).

In addition to their carbohydrate specificities, SP-A and SP-D also differ with respect to several other structural features that determine their interactions with the innate immune system (41, 45). SP-A forms a bouquet-like 18-mer that associates with surfactant lipids and tubular myosin and can bind to the C1qR receptor, toll-like receptors TLR2 and TLR4, the CD91/calreticulin complex, the signal inhibitory regulatory protein SIRP-alpha, and the unconventional myosin XVIIIA receptor SPR210. SP-D, on the other hand, forms a cross-like dodecamer (12 chains) that resides in the aqueous phase of the alveoli and can bind microfibril-associated protein 4, CD14, defensins, decorin, C1q, TLR2, TLR4, and a 340-kDa glycoprotein of unknown function.

In vitro studies previously showed that SP-A and SP-D could modulate the antituberculosis (anti-TB) response with both negative and positive consequences for bacterial control. For example, both SP-A and SP-D can contribute to bacterial agglutination (12, 18). Accordingly, SP-A-mediated agglutination was shown to enhance M. tuberculosis association with alveolar epithelium (18) and internalization in human and murine macrophages ex vivo (4, 9, 30). This agglutination, however, was not associated with bacterial killing, since SP-A signaling has also been reported to reduce the production of reactive nitrogen (31) and oxygen intermediates (8). SP-D-mediated opsonization, by contrast, has been reported to agglutinate bacteria, delay phagocytosis, and facilitate phagolysosomal fusion, potentially contributing to the control of M. tuberculosis (10-12). Based on such studies, we report here the first studies of SP-A-, SP-D-, and SP-A/-D-deficient mice in a murine aerosol model of TB.

MATERIALS AND METHODS

Mice and bacterial strains.

SP-A KO (25), SP-D KO (6), and SP-A/-D double KO mice (21) were a kind gift of Samuel Hawgood. Mice had been backcrossed at least 10 generations against a C57BL6 background and genotyped according to previously described PCR methods (21). Age-matched wild-type C57BL/6 mice (WT) were purchased from Jackson Laboratories, Inc. All experiments were initiated with mice 6 and 12 weeks of age, and at least 4 mice of each genotype were used at each time point in the infection.

Before infection, all mice were housed at the Rockefeller University LARC immunocompromised core animal care facility in Thoren cages while receiving sterile rodent chow and water. Infected mice were housed in biosafety level 3 (BSL3) facilities. All animal procedures were approved by Institutional Animal Care and Use Committee (IACUC) protocols at Rockefeller University and Weill Cornell Medical College.

Mycobacterium tuberculosis (Erdman) was grown at 37°C in Middlebrook 7H9 (Difco) medium supplemented with 0.2% glycerol, 0.05% Tween 80, 0.5% bovine serum albumin (BSA), 0.2% dextrose, and 0.085% NaCl. Early-log-phase M. tuberculosis (optical density [OD] at 600 nm < 0.5) was used for aerosol infection using a Middlebrook inhalation exposure system (Glas-Col). The bacterial burden in lungs was determined by plating lysates from left lung lobes for CFU determination on 7H10 plates at 0 to 10⁻³ dilutions.

Quantitative reverse transcription (RT)-PCR.

At different points during infection, the right lobes of infected lungs were harvested and immediately homogenized in Tri Reagent (AB Applied Biosciences [AB]). Total RNA purification was carried out using the RiboPure kit (AB). Conversion into cDNA used the TaqMan RNA-to-CTT 2-step kit (AB). TaqMan quantitation of SP-A, SP-D, gamma interferon (IFN-γ), inducible nitric oxide synthase (iNOS), and beta-actin was carried out with inventoried primers in an AB 7900HT sequence detection system according to the manufacturer's instructions. For relative quantitation of the different mRNA species, all values were normalized to measured levels of beta-actin transcripts and expressed relative to values for uninfected WT mice using the comparative threshold cycle (C_T) method (Applied Biosystems, Guide to performing relative quantitation of gene expression using real-time quantitative PCR, part number 4371095, rev. B).

Microscopy imaging and measurement of infiltration.

Right lungs were harvested at indicated times and fixed in formalin. Formalin-fixed lungs were embedded, sectioned, and stained by hematoxylin and eosin (H&E) at the Laboratory of Comparative Pathology at Memorial Sloan Kettering Cancer Center. Slides were examined in a Zeiss bright-field microscope with a Spot Insight QE color digital camera. To determine the lung area covered by granulomas, the NIH ImageJ software program was used to calculate the area of immune infiltration at 5× resolution in 9 slices per mouse. Percent lung area covered by granulomas was calculated for each mouse by dividing the sum of granulomatous areas in these sections by the total area of the lung examined.

Flow cytometry.

At week 3 postinfection, the right lungs from infected and uninfected mice were harvested and incubated in dissociation medium as previously described (23). After an hour at 37°C, lung tissue was turned into a single-cell suspension. Cells were washed in RPMI (Cellgro) with 10% fetal bovine serum (FBS) and counted in a hematocytometer. Staining for flow cytometry was carried out in phosphate-buffered saline (PBS) with 2% BSA, 2 mM EDTA, 1 μg/ml mouse IgG (Sigma-Aldrich), 1 μg/ml rat IgG (Sigma-Aldrich), and 1 μg/ml antibodies (Abs) against FcII/III (BD Biosciences [BD]). The Abs used for staining were I-A^b-phycoerythrin (I-A^b-PE), recombinant IgG-fluorescein isiothiocyanate (rIgG-FITC), CD11c-allophycocyanin (CD11c-APC), rat IgG2a-PE, CD8α-peridinin chlorophyll protein Cy5.5 (PerCPCy5.5), CD11b-PerCPCy5.5, CD86-FITC, CD40-FITC, CD4-APC, CD45-FITC, CD45-PE, Ly6G-FITC, and CD69-FITC (BD). For intracellular cytokine staining, lung cells were incubated for 5 h at 37°C in the presence of 5 ng/ml ESAT6 peptide (a kind gift from Eric Pamer) and GolgiStop inhibitor (BD). Intracellular cytokine staining was performed using IFN-γ-FITC and rat IgG-1-FITC and a Citofix/Citoperm kit (BD). After the samples were stained, cells were fixed in 4% paraformaldehyde for 24 h before analysis. Samples were collected on a FACSort flow cytometer (BD) using the CellQuest software program and analyzed using FlowJo software.

RESULTS

To address the role of SP-A and SP-D in susceptibility to tuberculosis in vivo, we infected WT, SP-A knockout (KO), SP-D KO, and SP-A/-D KO mice (deficient in both SP-A and SP-D). Previous reports indicated that SP-A- and SP-D-deficient mice exhibited heightened susceptibility to acute viral, bacterial, and fungal pathogens (1, 16, 20, 45). We therefore examined the roles of SP proteins in the early infection with aerosolized M. tuberculosis Erdman, a strain known to bind both SPs (11, 18).

At low inoculating doses, mimicking conditions associated with natural aerosol transmission (CFU < 100), we could not identify any differences in infection load among SP-A, SP-D, and SP-A/-D KO mice and WT animals at day 1 post-aerosol challenge (Fig. 1 A). This result indicated that SPs are not essential for establishment of infection in mice and that the previously reported defects in alveolar surface tension caused by the SP-D deficiency (21) did not impair aerosol inoculation in SP-D or SP-A/-D KOs at inoculating doses up to 1,000 CFU (Fig. 1B). Control of M. tuberculosis replication following low-dose aerosol challenge was similarly unimpaired throughout the course of infection in all genotypes tested (Fig. 1C). No differences in survival were observed among the strains up to 146 days postinfection (data not shown). Thus, despite known structural and immunomodulatory defects in the lungs of SP-A- and SP-D-deficient mice, no measurable role of SP-A or SP-D in control of chronic M. tuberculosis replication could be identified.

FIG. 1.

Bacterial load. Mice were infected with either <100 CFU or >100 CFU, and lungs were harvested at day 1 and at weeks 2, 3, 4, 8, and 12 postinfection. Five independent infections were carried out, and these are representative examples. Bacteria in the lungs from 4 to 5 mice per genotype were measured by plating on 7H10 plates. Error bars represent standard errors. (A) Inoculum at day one post-aerosol challenge. No statistical differences were observed with bacterial doses < 100 CFU. (B) Inoculum of 6,000 CFU per WT lung examined at day one post-aerosol challenge. Statistical significant comparisons are represented by “*.” Analysis using Student's t test with respect to results for the WT showed P = 0.03 for SP-A KO mice, P < 0.0001 for SP-D KO mice, and P = 0.0007 for SP-A/-D (SPAD) KO mice using a high inoculum. Similar results were observed with bacterial doses of >1,000 CFU. (C) Bacterial counts during chronic infection. At different time points after inoculation with <100 CFU, the left lungs were plated at 10⁻¹ to 10⁻³ dilutions. No statistically significant differences were identified among the genotypes during chronic infection.

And yet, at inoculums higher than a CFU of >1,000, we observed more bacteria in the lungs of SP-A-, SP-D-, and SP-A/SP-D- deficient mice that in WT animals (Fig. 1B). Similar high-dose experiments have been used by others to demonstrate the acute role of SPs in other infections (1, 16, 20), and here we replicated these findings with M. tuberculosis Erdman. This early, nonredundant role suggested that interaction of murine SPs with M. tuberculosis could limit the high CFU inoculation.

To further evaluate the role of SP-A and SP-D in control of M. tuberculosis replication at physiological infection doses, we quantified SP-A and SP-D mRNA levels in WT, SP-A-, SP-D-, and SP-A/-D-deficient mice at weeks 2, 3, 4, 8, and 12 following infection with a CFU of <100. SP-A/-D double KO mice served as negative controls that, as expected, showed no increases in expression at any time point (Fig. 2 A and B). Expression of SP-A and SP-D remained constant throughout the course of M. tuberculosis infection in WT mice, indicating that these proteins are not downregulated in the face of a chronic inflammatory response. This contrasts with findings of studies in humans, where SP-A expression was shown to be regulated by IFN-γ and tumor necrosis factor alpha (TNF-α) and reduced in human lungs with TB (17, 36). It differs also from findings of acute viral infection studies in mice and acute allergic responses in rats (24, 37). These results thus indicate that chronic M. tuberculosis infection does not alter surfactant expression levels in mice (24).

FIG. 2.

SP-A and SP-D mRNA levels during M. tuberculosis infection. Mice were infected with 100 CFU, and lungs were harvested 2, 3, 4, 8, and 12 weeks postinfection. Uninfected controls for each genotype are included at week zero. mRNA was quantitated using real-time RT-PCR using primers for mouse SP-A (A) or SP-D (B). Values were calculated using the comparative C_T method and normalized using beta-actin controls. All results are expressed as a fraction of the mRNA expression of uninfected WT mice. Four mice per data point were used to calculate means, and error bars depict the lower and upper range of the results (calculated using the mean ± standard deviation at each time point).

Because the genes encoding SP-A and SP-D are closely linked and potentially affected by the insertion of neighboring deletion cassettes, we also examined SP mRNA expression in SP-A KO and SP-D KO mice. SP-A-deficient mice exhibited 40% of the WT level of SP-D expression up to week 3 postinfection (Fig. 2B) and 60% of WT levels following initiation of the adaptive immune response to M. tuberculosis after week 3. Similarly, SP-D KO mice expressed half of the SP-A mRNA levels expressed in WT mice at time points later than week 3 (Fig. 2A). These results thus establish that the apparent dispensability of SP-A and SP-D in isolation is not explained by a compensatory upregulation of the other SP in the single KO mice.

To explore potential immunomodulatory roles of SP-A and SP-D in formation of the adaptive response against M. tuberculosis, we next examined the immunopathologic responses of the lungs of SP-A, SP-D, and SP-A/-D KO mice to M. tuberculosis infection at the peak of the T cell response (weeks 3 and 4). H&E staining of lung sections showed granulomatous inflammation in all genotypes, but SP-A- and SP-A/-D-deficient mice had increased granulomatous involvement (Fig. 3A and B). As early as week 3, SP-A and SP-A/-D KO mice exhibited well-formed granulomatous lesions with discrete myeloid and lymphoid areas (Fig. 3C) that were only poorly organized in WT counterparts. SP-A- and SP-A/-D KO mice also showed evidence of increased infiltration out of most arterioles and distal bronchioles (Fig. 3B), suggesting that the negative regulatory role of SP-A protects mice from increased inflammatory responses in the lungs infected with M. tuberculosis.

FIG. 3.

Histology of M. tuberculosis infection in SP-deficient lungs at weeks 3 and 4 postinfection. (A) Quantitation of leukocyte-infiltrated area in infected lungs using ImageJ. The percentage of lung area with infiltration was calculated by measuring the area with granulomatous infiltration and dividing it by the total lung area examined. Nine sections from 4 to 6 mice per genotype were examined, and pictures were quantitated from 5 to 6 fields at 5× magnification. Statistical significance is depicted by an asterisk, indicating P = 0.06 for SP-A KO mice and P = 0.009 for SP-A/-D KO mice in Student's t test. Error bars depict the standard error for each genotype. (B) H&E lung sections of week 4 infected mice at 5× resolution. The letter A depicts infiltration around arterioles and proximal bronchioles. The letter F depicts areas with foamy macrophage accumulation. (C) Uninfected and week 3 infected lung H&E sections at 40× resolution. The letter L depicts lymphocytic areas, whereas the letter M depicts myeloid aggregates within granulomatous tissue. The letter N depicts multinucleated cells found inside granulomas, and the letter F depicts foamy macrophages visible outside granulomatous tissue.

SP-D KO mice, in contrast, showed no evidence of increased granulomatous infiltration compared to WT mice (Fig. 3A). Despite the development of emphysematous changes and increased lipid accumulation prior to challenge with M. tuberculosis(6, 42, 47), granuloma formation took place with normal kinetics (Fig. 3A and B and data not shown). Foamy macrophages were detectable at week 4 postinfection (Fig. 3C), earlier in life than what has been reported for uninfected SP-D-deficient mice (6), but this had no effect on granuloma size (Fig. 3A). While multinucleated giant cells were observed within the granulomas of M. tuberculosis-infected SP-D-deficient mice at week 4 (Fig. 3C), no evidence of impaired immune control of M. tuberculosis was identified (Fig. 1A and C). Similar macrophage infiltrates were identified in SP-A/-D KO mice (Fig. 3B and data not shown), indicating that the expression of SP-A alone does not contribute to this phenotype.

Based on previous reports describing the ability of SPs to alter epitope presentation by antigen-presenting cells and T cell proliferation (5), we also studied the impact of SPs in the generation of the T cell response against tuberculosis (10, 19). To do so, we quantified CD4⁺ T cell infiltration into the lungs 3 weeks postinoculation. As seen in Fig. 4 A, CD4⁺ T cell accumulation in all genotypes was normal, despite the increased granulomatous infiltrate observed by histology in SP-A- and SP-A/-D KO mice (Fig. 3).

FIG. 4.

Characterization of the CD4⁺ T cell responses against M. tuberculosis in SP-deficient mice. Mice were infected with 100 CFU, and lungs were harvested 2, 3, 4, 8, and 12 weeks postinfection. (A) CD4⁺ T cell counts at week 3. After isolation of total lung cells using collagenase and DNase, cells were counted in a hematocytometer and stained for flow cytometry using CD4. Total CD4⁺ T cells are depicted in gray bars. For quantitation of the M. tuberculosis-specific response, isolated total lung cells were incubated with the ESAT6 peptide in the presence of the GolgiStop inhibitor for 5 h. Then, the cells were stained for CD4, TNF-α, and IFN-γ or isotype controls. White bars depict the numbers of IFN-γ⁺ CD4⁺ T cells, which were also TNF⁺ (gates were defined using isotype control staining). A logarithmic scale was used to depict total and antigen-specific cells in the same axis. Error bars are standard deviations of results for 4 mice per genotype. Student's t test indicated no P values < 0.1. (B) Levels of IFN-γ mRNA during M. tuberculosis infection. Uninfected controls for each genotype are included at week zero. mRNA was quantitated using real-time RT-PCR and calculated using the comparative C_T method (normalized using beta-actin controls). All results are expressed as a fraction of the mRNA expression of uninfected WT mice. Four mice per data point were used to calculate means, and error bars depict the lower and upper range of the results (calculated using the mean ± standard deviation at each time point).

Using the M. tuberculosis-specific peptide ESAT6, we further examined CD4⁺ T cell-mediated IFN-γ production in the lungs of infected mice by intracellular cytokine staining. As summarized in Fig. 4A, the proportions and total numbers of T cells responding to ESAT6 by secreting IFN-γ (CD4⁺ IFN-γ+ cells) were similar in all genotypes studied. These results suggest that SP-A and SP-D do not alter the presentation of this immune-dominant peptide of M. tuberculosis.

To further confirm that the IFN-γ response against M. tuberculosis was unimpaired in SP-deficient mice, we measured IFN-γ mRNA levels by quantitative real-time PCR throughout the course of infection. As shown in Fig. 4B, IFN-γ mRNA production exhibited similar kinetics in WT and SP-A-, SP-D-, and SP-A/-D-deficient mice. The response peaked between weeks 3 and 4 at roughly 80 times the levels for uninfected WT mice, after which it remained around 60-fold throughout the chronic infection (Fig. 4B). These results thus demonstrate that M. tuberculosis-elicited IFN-γ production in the murine lung is not modulated by SP-A or SP-D.

Although the measured CD4⁺ T cell responses of SP-A, SP-D, and SP-A/-D KO mice appeared normal, histology suggested that the SP-A and or SP-D deficiency could potentially affect myeloid function (Fig. 3). In fact, effector functions such as phagocytosis, myeloid activation, nitric oxide, and oxygen reactive species are modulated by SPs in vitro (4, 9-12, 30). To address this possibility, we characterized the myeloid infiltrate on infected mice using flow cytometry (Fig. 5). Using CD45 staining, we first identified cells belonging to the hematopoietic lineage (Fig. 5A) and then gated on both CD11b and CD11c to identify distinct myeloid populations (Fig. 5A).

FIG. 5.

Characterization of the myeloid infiltrate in SP-deficient mice. Mice were infected with 100 CFU, and lungs were harvested at 3 weeks postinfection. Four mice per genotype were used in the experiments, and fluorescence-activated cell sorter (FACS) plots are representative examples. (A) Myeloid cells in SP-deficient mice. After isolation of total lung cells using collagenase and DNase, cells were counted in a hematocytometer and stained for flow cytometry using CD45, CD11b, and CD11c. As depicted, the forward scatter (FSC) and side scatter (SSC) gate was used to select live cells. The CD45⁺ population was analyzed using CD11b and CD11c. Gates depicted were used for analysis of myeloid populations in panels B to D. (B) Myeloid cell counts. The asterisk indicates statistically significant differences in the neutrophil CD11b^high CD11c^neg counts (also Ly6G⁺) compared to results for WT mice. For neutrophils in the SP-A panel, P = 0.05, and for those in the SP-D panel, P = 0.03 in Student's t test compared to WT results. Other comparisons were not statistically significant. (C) Activation markers on the CD11b^low CD11c^low macrophages. Lung cells were stained for I-A^b, CD40, CD69, and CD86, in addition to the myeloid markers described above. No statistically significant difference was identified when the geometric mean I-A^b and CD86 stainings of 4 mice per genotype were compared. However, SP-A KO mice had statistically higher geometric means for CD69 (P = 0.01) and CD40 (P = 0.02) than WT mice in a Student's t test. (D) Activation markers on the CD11b^high CD11c^high dendritic cells. Lung cells were stained for I-A^b, CD40, CD69, and CD86, in addition to the myeloid markers described above. No statistically significant difference was identified in this cell population. (E) Levels of iNOS mRNA. Mice were infected with <100 CFU, and lungs were harvested 2, 3, 4, 8, and 12 weeks postinfection. Uninfected controls for each genotype are included at week zero. mRNA was quantitated using real-time RT-PCR using the comparative C_T method (normalized using beta-actin controls). All results are expressed as a fraction of the mRNA expression of uninfected WT mice. Four mice per data point were used to calculate means, and error bars depict the lower and upper range of the results (calculated using the mean ± standard deviation at each time point).

Surprisingly, in contrast to the expected inflammatory environment of the SP KOs, SP-A- and SP-D-deficient mice had reduced neutrophil (Cd11b^high CD11c^neg Ly6G^high) infiltration within M. tuberculosis-infected lungs (Fig. 5A and B). This indicated that an absence of SPs can alter the leukocyte composition of the response to M. tuberculosis despite their not having effects in long-term bacterial control (Fig. 1C).

However, CD11b^low CD11c^low macrophages and the CD11b^high CD11c^high population exhibited similar numbers of cells in all genotypes (Fig. 5A and B). We also examined the activation level of these cells by measuring expression of the activation markers CD69, I-A^b, CD40, and CD86 (Fig. 5C and D). SP-D KO cells resembled WT expression of all the activation markers studied in both the CD11b^low CD11c^low macrophages and the CD11b^high CD11c^high dendritic cell population (Fig. 5C and D), indicating the absence of SP-D does not alter the activation status of myeloid antigen-presenting cells. Together with our finding that the IFN-γ response against M. tuberculosis is unimpaired by the absence of SPs, the normal levels of I-A^b and costimulation support the idea that SP-D does not play a dominant role in the presentation of M. tuberculosis antigens in mice.

On the other hand, compared to WT or SP-D KO mice, SP-A KO mice consistently showed slightly more CD11b^low CD11c^low activated macrophages as measured by CD69 levels (Fig. 5C). They also had high levels of CD40 but normal expression of I-A^b and CD86 (Fig. 5C). We concluded that the lack of SP-A caused a selective activated phenotype in CD11b^low CD11c^low macrophages. The SP-A deficiency did not alter CD11b^high CD11c^high cells (Fig. 5D). These findings support the idea that SP-A may play a role in the negative regulation of CD11b^low CD11c^low macrophages and does not impair the CD11b^high CD11c^high subset, which has recently been shown to contain M. tuberculosis (44) and express the highest levels of iNOS (40).

To further examine the dispensability of SPs in effector functions against M. tuberculosis, we studied the expression of iNOS during the course of the immune response against M. tuberculosis using real-time RT-PCR (Fig. 5E). In WT mice, iNOS mRNA peaked at weeks 3 and 4 and was maintained at 50 times the levels for uninfected mice up to week 12 (Fig. 5E). In the absence of the down-modulatory effects of SP-A (7, 15, 25), we expected iNOS to be elevated in SP-A- and SP-A/-D-deficient mice. In the case of SP-D, we also expected an increase in oxidative products since this has been previously found in the bronchioalveolar fluid of uninfected mice (2, 3). However, no statistically significant alterations in total lung iNOS expression were observed compared to total lung iNOS expression in infected WT mice. Furthermore, we did not detect gross differences in nitrotyrosine staining of lung sections from infected mice at week 8 postinfection (data not shown). In summary, these results thus demonstrate an unexpectedly normal IFN-γ-dependent response in the absence of either SP-A and/or SP-D (Fig. 4 and Fig. 5E).

DISCUSSION

Our results indicate that SP-A and SP-D are not essential for inoculation of M. tuberculosis in a murine aerosol model of TB at low inoculating doses. It is possible that other proteins binding mannosylated glycans can compensate for SPs, such as dendritic cell-specific ICAM3-grabbing nonintegrin (DC-SIGN), the mannose receptor MR, DC-associated C-type lectin 1 (DECTIN-1), and mannose binding lectin (MBL) (9). This level of compensation has also been observed in infected SIGNR-1 knockouts (40), suggesting that lectins efficiently overlap in the internalization of M. tuberculosis at low infection doses. The increased M. tuberculosis inoculation on SP-A and SP-D KOs at high infection doses indicates that such compensation might be saturable, but the dose needed to see an effect suggests that compensation may play a role at biologically significant infectious doses.

Our results further indicate that the lack of immunomodulation by SP-A and SP-D does not measurably impair M. tuberculosis growth, nor does it improve bacterial control. Given the susceptibility of SP-A and SP-D KO mice to many viral, bacterial, and fungal pulmonary pathogens, this finding is particularly unexpected (3, 24, 26, 27, 35). Moreover, given the increased immune infiltration in the lungs of M. tuberculosis-infected SP-A and SP-A/-D KO mice, it is especially surprising to observe a lack of difference in bacterial titers (7, 15, 25). It is possible that the lack of an SP-A-mediated transcriptional program in M. tuberculosis, which involves lipases and the synthesis of pthiocerol dimycocerosate (PDIM) (38), leads to the formation of larger granulomas and the presence of more activated macrophages expressing CD69 and high CD40. Nevertheless, the T cell responses and IFN-γ and iNOS expression follow WT levels. Potential explanations for this finding are the following: (i) that the immune-modulatory effects of SP-A and SP-D may occur too early to modify antigen presentation and the T cell response to M. tuberculosis (which are delayed until week 3 postinfection [32, 43, 44] or (ii) that SP-A might target the CD11b^low CD11c^low macrophages instead of the CD11b^high CD11c^high population, which appears to be more important in control of M. tuberculosis (40, 44). These results clearly emphasize the lack of strict correlation between inflammation and improved bacterial control.

Findings of in vitro studies of SP-D have suggested that the absence of SP-D might facilitate mycobacterial infection of human macrophages (10). Uninfected SP-D and SP-A/-D KOs also lack anti-inflammatory signals (7, 15) and have activated alveolar macrophages (6, 21, 42, 47) and increased oxides of nitrogen in bronchoalveolar fluid (2). Nevertheless, despite the pulmonary defects of SP-D-deficient mice (6, 42, 47) and the presence of foamy macrophages (2, 47), we show that such activities are dispensable in a murine aerosol challenge model of TB. This may be because alveolar macrophages are an early and transient host for the bacteria, which quickly establish infection in monocyte-derived cells recruited from the blood (34, 40) and show no differences in our experiments.

Our results also contrast with those of prior in vitro studies that suggested that both SP-A and SP-D could alter immune control of M. tuberculosis in a murine model of TB. Several differences may explain these results. First, our assays measured events 24 h to 120 days after infection, a time frame in which in vivo mechanisms of compensation may take place. These are very different from the short-term assays of previous studies in ex vivo macrophages. Our results indicate that initially, SP-D and SP-A may interact with M. tuberculosis, but their effects are minimal in the long-term chronic infection.

Surfactant purification and synthesis complicate the interpretation of in vitro assays that add bovine or human SP proteins to macrophage cultures (46). By using genetically engineered SP-A, -D, and -A/-D knockout mice, we have obviated the in vitro complexities of generating pure, properly folded, oligomerized, functional surfactant-associated proteins free of lipopolysaccharide (LPS). Moreover, while the use of these KOs has validated essential roles for surfactant-associated proteins against other pathogens, such as influenza (20), Pneumocystis carinii (1), and Pseudomonas aeruginosa (16), our studies indicate that M. tuberculosis is more resistant to surfactant immune modulation than previously thought.

In vitro assays using monocyte- or bone marrow-derived macrophages may similarly fail to adequately model the myeloid populations infected by M. tuberculosis in vivo. Our studies thus demonstrate that myeloid populations in the lungs of mice exhibit a distinct sensitivity to SP modulation that may not correspond to the monocyte-derived dendritic cells present in vivo (40).

Our results finally highlight important differences between mouse and human systems. There is an 80% amino acid similarity between murine SP-A and human SP-A1 and SP-A2 and an 86% amino acid similarity among human and mouse SP-D. Although the lectin binding domain of SPs which interacts with M. tuberculosis has high homology in the two organisms (100% for SP-D and 86% for SP-A), the amino acid differences in other regions could potentially account for distinct regulation, differential effects on macrophages and their receptors, the compensation between lectins, and the effects on the bacterium itself.

It is equally worth noting that human active tuberculosis leads to cavitation, whereas in mouse models the bacteria are mainly intracellular (22). Although not addressed by our current study, cavitation is another stage of infection where the bacteria could encounter an extracellular surfactant in the alveoli and distal bronchioles. It is possible that SP-A and SP-D agglutinate M. tuberculosis in caseous lesions and increase the infectious droplet size. In this setting, SP agglutination might generate infectious droplets too large to remain in the air and be inhaled by others, thus reducing transmissibility (28). This interpretation could potentially explain why active tuberculosis has been associated with polymorphisms in SP-A and SP-D (13, 27) but it shows no effects in the mouse model. Future epidemiological studies would be needed to further elucidate this hypothesis.