Delayed-type hypersensitivity represents high levels of protein Ag-specific adaptive immunity induced by mycobacterial infection, and can be monitored in the Ag-challenged skin. Besides protein Ags, recent evidence has suggested that a substantial immunity directed against glycolipid Ags is also elicited in response to mycobacterial infection, but skin hypersensitivity to this class of Ags has not been fully assessed. To address this issue directly, glycolipid-specific skin reactions were evaluated in guinea pigs infected with Mycobacterium avium complex (MAC). Significant skin induration was observed in MAC-infected, but not mock-infected, guinea pigs, following intradermal administration of a mixture of MAC-derived glycolipids. Surprisingly, this glycolipid-specific skin response involved up-regulated expression of IL-5 mRNA in situ and marked local infiltration of eosinophils. Challenge experiments with individual glycolipid components detected an outstanding capability for trehalose dimycolate (TDM), but not a structurally related glycolipid, glucose monomycolate, to elicit the skin response. T lymphocytes derived from the spleen of MAC-infected, but not uninfected, guinea pigs specifically responded to TDM in vitro by up-regulating IL-5 transcription, and this response was not blocked by Abs that reacted to the known guinea pig group 1 CD1 proteins. Finally, the eosinophilic skin hypersensitivity to TDM was also elicited in guinea pigs vaccinated with bacillus Calmette-Guerin, which contrasted sharply with the classical delayed-type hypersensitivity response to the purified protein derivative. Therefore, the TDM-elicited eosinophilic response defines a new form of hypersensitivity in mycobacterial infection, which may account for local infiltration of eosinophils often observed at the site of infection.
Upon infection with mycobacteria, such as Mycobacterium tuberculosis, strong Th1 responses to mycobacteria-derived protein Ags are elicited in the majority of immunocompetent individuals, and are expressed in the form of delayed-type hypersensitivity (DTH)3 (1, 2, 3). The Th1-dominant DTH response is thought to be critical for granuloma formation, a pathological process for the host to confine bacteria at the site of infection (4, 5). Failure to elicit the Th1 response often results in inefficient control of mycobacterial infection, as seen in human cases with recessive mutations in genes encoding IFN-γ and IL-12 receptors (6). DTH can be induced at the site of intradermal injection of mycobacteria-derived Ags, and thus, the tuberculin skin test with the purified protein derivative (PPD) is useful for evaluating the protein Ag-specific cellular immunity in the infected host.
Besides protein Ags, recent evidence has suggested that the immune system has evolved the ability to detect microbe-derived glycolipid Ags in innate and adaptive phases of host defense against microbial infection. C-type lectins, such as the macrophage mannose receptor, and Toll-like receptors interact with microbial glycolipids and transmit signals into innate immune cells (7). In addition, CD1-dependent pathways for T cell recognition of microbial glycolipid Ags have also been identified (8, 9, 10). The innate and adaptive immune recognition of microbial glycolipid Ags may be particularly important for controlling mycobacterial infection because the lipid-rich cell wall of mycobacteria contains unique glycolipids that are critical for their survival and virulence (11). Indeed, vaccination with M. tuberculosis-derived lipids and glycolipids confers protective immunity in the guinea pig model of human tuberculosis, implicating pathways for host defense that are distinct from, but complementary to, those directed against protein Ags (12). Despite the advances in our understanding of immune recognition of mycobacteria-derived glycolipid Ags, skin hypersensitivity reactions to this class of Ags have not been fully assessed. Because of the distinct pathways for host responses to protein and glycolipid Ags, the hypersensitivity response to glycolipids could be substantially different from that directed against protein Ags.
In the present study, we found that trehalose dimycolate (TDM), a major mycolyl glycolipid in the cell wall of mycobacteria, induced hypersensitivity reactions in the skin of M. avium complex (MAC)-infected and M. bovis bacillus Calmette-Guerin (BCG)-immunized guinea pigs. Notably, the TDM-induced hypersensitivity involved marked infiltration of eosinophils and local expression of IL-5, and thus, was distinct from the protein Ag-induced classical DTH reaction.
Materials and Methods
Purification of lipid Ags from MAC and BCG
The MAC serovar 4 strain and the BCG Tokyo 172 strain were provided by Dr. Ikuya Yano (BCG Laboratory, Tokyo, Japan) and were grown at 37°C in 7H9 medium supplemented with the Middlebrook ADC enrichment (BD Biosciences). The bacteria were harvested when the OD at 600 nm reached 1–1.5. Total lipids were extracted with chloroform/methanol and separated into chloroform, acetone, and methanol fractions using open silicic acid columns as described previously (13, 14). Preparative thin layer chromatography (TLC) was further conducted to purify the major glycolipid species included in the acetone fraction, namely TDM, glucose monomycolate (GMM), and apolar glycopeptidolipids (GPLs) (Fig. 1).
TDM and GMM were extracted with chloroform/methanol (2:1, v/v) from silica gel TLC plates (Analtech) developed in chloroform/methanol/acetone/acetic acid (90:10:10:1, v/v/v/v), and the fractions were further fractionated by TLC with a solvent system of chloroform/acetone/methanol/water (50:60:2.5:3, v/v/v/v). Finally, the TDM and the GMM fractions were extracted with chloroform/methanol (2:1, v/v), dried, and rinsed with methanol. The final preparations thus obtained provided no extra spots on analytical TLC plates, and the identity of the lipids was confirmed by mass spectrometry. Protein contamination was not detected by silver staining of SDS-PAGE gels or by the Bradford assay.
The MAC serovar 4 strain-derived GPL species were resolved as closely associated spots on TLC plates due to the subtle difference in the degree of O-methylation and acetylation on sugar residues (15). Therefore, instead of purifying individual apolar GPL species, a mixture of apolar GPL species in the upper spots (upper GPL cluster) and a mixture of those in the lower spots (lower GPL cluster) were isolated from preparative TLC plates developed in chloroform/methanol (95:5, v/v).
Animals and skin tests
Four- to six-week-old female inbred strain 2 guinea pigs were purchased from Japan SLC, and housed under specific pathogen-free conditions. Either the MAC serovar 4 strain or the BCG Tokyo 172 strain was injected intradermally (1 × 108 CFU per animal), and six weeks after infection the skin area in the left flank of guinea pigs was shaved and depilated for skin tests. Indicated amounts of each lipid preparation were dissolved in 50 μl of mineral oil and were injected intradermally. Forty-eight hours after injection, the skin response was assessed by measuring the distance across the skin induration. Experiments were repeated at least twice to confirm reproducibility of the results. All animal experiments were performed according to the institutional guidelines on animal welfare and humane treatment of laboratory animals.
Histochemistry and electron microscopy
The excised skin samples were fixed for 1 day with 4% paraformaldehyde (Nacalai Tesque), dehydrated, and embedded in paraffin. The tissue sections were stained with Giemsa’s stain solution (Merck), and observed under microscope. The number of eosinophils was determined in three randomly selected high power (×400) fields, and statistical analysis was performed. Separately, some skin samples were deep frozen in OCT compound, and the cryosections were labeled with mouse mAbs to guinea pig helper/inducer T cells (clone CT7) (Serotec), followed by incubation with FITC-conjugated donkey anti-mouse IgG Abs (Jackson ImmunoResearch Laboratories). After washing, the labeled sections were mounted with Vectashield mounting medium (Vector Laboratories), and viewed under fluorescence microscope. Positive cells were counted in three randomly selected high power fields.
For transmission electron microscopy, the skin was fixed with 2.5% glutaraldehyde (Polysciences), and postfixed with 2% osmium (VIII) oxide (Wako Pure Chemical). Ultrathin sections were observed under the Hitachi H-7000 electron microscope.
Cytokine mRNA expression in the Ag-challenged skin
The Ag-challenged skin was excised and the total RNA was extracted with the RNeasy mini kit (Qiagen) according to the manufacturer’s instructions. The first-strand cDNA was synthesized from 0.5 μg of total RNA using oligo(dT) and PrimeScript reverse transcriptase (Takara Bio). To amplify specific transcripts, the samples were subjected to PCR amplification for 35 cycles of 30 s at 94°C, 1 min at 60°C (except for IL-5 at 63°C), 1 min at 72°C, and a final cycle of 5 min at 72°C. The primers used were: 5′-GAT ATT GTA GCC ATC AAT GAT CCC T-3′ (sense) and 5′-CAT CGT ATT TGG CCG GTT TCT CCA G-3′ (anti-sense) for glyceraldehyde-3-phosphate dehydrogenase (GAPDH); 5′-CCA TGA GCA CAG AAA GCA TGA TCC G-3′ (sense) and 5′-CTC ACA GGG CAA TGA CCC CAA AGT A-3′ (anti-sense) for TNF-α; 5′-CCA TGA GGG TGC TTC TGC AGT TGG G-3′ (sense) and 5′-CTC AGC CTT CAA TTG TCC ATT CCG T-3′ (anti-sense) for IL-5. The endpoint PCR products were resolved on 1.2% agarose gels and visualized by staining with ethidium bromide and UV transillumination. The experiments were repeated at least twice to confirm reproducibility of the results.
IL-5 mRNA expression in TDM-stimulated splenocytes
Splenocytes were isolated from either MAC-infected or mock-infected guinea pigs, and RBC were removed by treatment with the ACK lysis buffer (BioWhittaker). In some experiments, CT7-reactive cells were further removed, using the MACS MicroBeads (Miltenyi Biotec) conjugated with goat Abs to mouse IgG. The isolated cells (2 × 106 per well) were placed in wells of 24-well tissue culture plates and stimulated with 10 μg/ml TDM. For inhibition assays, either the anti-guinea pig pan group 1 CD1 mAb (clone CD1F2/6B5) (16) or an isotype-matched control mAb (clone P3) was added to the culture at the concentration of 10 μg/ml. After 16 h incubation at 37°C, the cells were harvested and the total RNA was extracted using the RNeasy mini kit (Qiagen). The first-strand cDNA synthesis and RT-PCR were conducted as described above. Each experiment was repeated at least twice to confirm reproducibility of the results.
Statistical analysis was performed using Student’s t test. Values of p < 0.05 were considered statistically significant.
Skin hypersensitivity to glycolipids in MAC-infected guinea pigs
To address whether mycobacterial lipids could elicit skin hypersensitivity in mycobacteria-infected subjects, guinea pigs were either infected with MAC or mock treated. After 6 wk these animals were subjected to the skin test with the MAC-derived total lipid preparation. None of the animals showed apparent skin changes for the first several hours. At 8 h, however, MAC-infected guinea pigs began to develop detectable induration at and around the site of intradermal injection of the total lipid preparation, with the maximum response observed at 48 h (Fig. 2,A). Importantly, no significant skin reactions were observed in mock-infected guinea pigs challenged with the same Ag preparation, suggesting that the positive skin test required prior mycobacterial infection (Fig. 2 A).
Fractionation of the total lipid preparation was performed to gain insights into the lipid species towards which the observed skin hypersensitivity was directed. The total lipid preparation (Fig. 2,B, lane 1) was passed over open silicic acid columns to which all major lipid classes bound. The columns were then eluted sequentially with solvents of increasing polarity (chloroform, followed by acetone, and finally methanol), resulting in separation into three fractions, namely the chloroform fraction (Fig. 2,B, lane 2), the acetone fraction (lane 3) and the methanol fraction (lane 4). Each fraction was then individually tested for its ability to elicit skin reactions in MAC-infected and mock-infected guinea pigs. The acetone fraction induced the most prominent skin reactions in MAC-infected, but not mock-infected, guinea pigs (Fig. 2,C). Whereas the methanol fraction contained moderate bioactivity, no bioactivity was detected for the chloroform fraction (Fig. 2 C). Because of its most prominent activity among the three fractions as well as the fact that it contained a majority of glycolipids, skin reactions to the acetone fraction was further characterized at molecular and cellular levels in this study.
Marked eosinophil infiltration at the site of the glycolipid-elicited skin hypersensitivity
Histochemical analysis of the skin challenged with the acetone fraction in MAC-infected guinea pigs revealed marked infiltration of polymorphonuclear (PMN) cells in the dermis (Fig. 3,B). Infiltration of PMN cells was also detected in the epidermis, where degenerative changes or ballooning of keratinocytes were observed (Fig. 3,C). Such changes were not detected in the skin of mock-infected guinea pigs challenged with the acetone fraction (Fig. 3,A), confirming at the histological level that the skin hypersensitivity to the acetone fraction required prior infection. Cytoplasmic granules expressed in the infiltrating PMN cells showed eosinophilic staining with Giemsa (Fig. 3,C, arrows) and contained electron dense crystalloid core structure (Fig. 3 D), all of which were characteristic features of eosinophils. A small number of eosinophils and T cells appeared to begin to infiltrate around 8 h after the challenge, and there were no signs for an early wave of neutrophils (data not shown).
Among cytokines that could potentially be involved in local infiltration and accumulation of eosinophils, the cDNA sequences for guinea pig TNF-α and IL-5 have already been published, enabling us to assess their local expression by RT-PCR. Eight hours after intradermal injection of either the acetone fraction in mineral oil or mineral oil alone, the skin specimens were obtained and the total RNA was extracted, followed by RT-PCR with specific primers. As shown in Fig. 3 E, expression of TNF-α and IL-5 was up-regulated only in the acetone fraction-challenged skin of MAC-infected guinea pigs. Because both TNF-α and IL-5 are known to facilitate recruitment of eosinophils to the inflammatory foci of tissues (17), specific induction of these cytokines in the acetone fraction-challenged skin of MAC-infected animals would likely support local infiltration and accumulation of eosinophils.
Identification of TDM as a glycolipid species eliciting the eosinophilic skin hypersensitivity
To determine the lipid species in the acetone fraction capable of eliciting the eosinophilic skin hypersensitivity, major lipid species contained in the acetone fraction, namely TDM (Fig. 4,A, lane 2), GMM (lane 3), upper GPL clusters (lane 4), and lower GPL clusters (lane 5) were purified from the acetone fraction (lane 1) as described in Materials and Methods, and tested individually for their bioactivity. As shown in Fig. 4,B, significant skin reactions were elicited only in MAC-infected guinea pigs challenged with TDM, but not with other glycolipid preparations including GMM, a known CD1-presented mycolyl glycolipid (18, 19). Histochemical analysis of the TDM-challenged skin in infected animals detected prominent infiltration of eosinophils (Fig. 4,C, arrowheads), as seen in the skin challenged with the acetone fraction (Fig. 3, B and C). The number of infiltrating eosinophils in response to TDM was significantly higher in MAC-infected guinea pigs (137 ± 26.6 cells/field, p < 0.01) than in uninfected animals (7.2 ± 12.5 cells/field).
TDM-induced IL-5 induction by splenocytes derived from MAC-infected guinea pigs
As shown above, the TDM-induced, eosinophilic skin reaction was observed only in MAC-infected guinea pigs, suggesting the possibility that MAC infection might prime immune cells to respond by IL-5 secretion upon subsequent challenge with TDM. Indeed, when splenocytes derived from MAC-infected or mock-infected guinea pigs were stimulated in vitro with TDM, only those derived from MAC-infected guinea pigs up-regulated IL-5 mRNA expression in response to TDM (Fig. 5,A). Depletion of guinea pig “helper/inducer” T cells, defined by reactivity to the CT7 mAb, completely abrogated the IL-5 mRNA induction (Fig. 5,B). Considering the mycolic acid-containing glycolipid structure of TDM that was shared with the CD1-presented GMM Ag, we addressed whether the TDM-elicited IL-5 induction by CT7-reactive T cells might depend on CD1 function. As shown in Fig. 5 C, TDM-stimulated splenocytes up-regulated IL-5 mRNA expression even in the presence of saturating amounts of an anti-pan group 1 CD1 mAb (6B5), suggesting the possibility that the TDM-elicited, eosinophilic skin hypersensitivity might not depend on the function of the known guinea pig group 1 CD1 molecules.
The TDM-specific skin reaction was distinct from the classical DTH reaction to protein Ags
TDM is expressed in all mycobacteria species so far analyzed (20). To address whether the TDM-specific hypersensitivity reaction induced in MAC-infected guinea pigs might be observed also in guinea pigs infected with other mycobacteria species, and to directly compare the response with the classical DTH reactions against PPD (the tuberculin test), guinea pigs were immunized with BCG, and histochemical analysis of the skin challenged with either mineral oil alone, M. tuberculosis-derived PPD, or BCG-derived TDM was performed in parallel. Although infiltrating cells were not apparent in the mock-challenged skin (Fig. 6,A), significant infiltration of mononuclear cells was observed in the PPD-challenged skin (Fig. 6, B and E), in which few eosinophils were detected (Fig. 6,D). In sharp contrast, eosinophil infiltration was prominent in the TDM-challenged skin (Fig. 6, C and D), as seen in MAC-infected guinea pigs (Fig. 4). Furthermore, in situ up-regulation of IL-5 mRNA expression was observed only in the TDM-challenged skin, but not in PPD-challenged skin (Fig. 6 F). Taken together, the results obtained in the present study indicated that the TDM-induced response define a form of skin hypersensitivity that required prior mycobacterial infection but was distinct from the classical DTH reaction to protein Ags.
We and others (10, 21, 22) have previously detected mycobacteria-specific, group 1 CD1-restricted T responses in humans in a form analogous to that of a classical Th1 memory response to protein Ags. Therefore, the present study was initially intended to determine whether CD1-presented glycolipid Ags, such as GMM, might elicit DTH responses that were comparable to those directed against protein Ags. Similar to the classical DTH response, the TDM-elicited skin hypersensitivity identified in this study required prior mycobacterial infection, and the peak response was observed 48 h after the Ag challenge (Figs. 2 and 4). Further, specific Abs to TDM were not detected in the sera of infected guinea pigs, and administration of the sera and TDM to the skin of naive guinea pigs did not elicit any significant responses (data not shown), making it unlikely that the TDM-elicited skin response was a persistent form of immediate hypersensitivity or type I immunity. Notably, the TDM-elicited skin hypersensitivity involved prominent infiltration of eosinophils and up-regulation of IL-5 transcription, sharply contrasting with the PPD-elicited skin hypersensitivity (Fig. 6). Thus, the TDM-elicited skin response apparently defines a form of hypersensitivity that is distinct from the classical Th1-dominant DTH response to protein Ags. The identity of the TDM-specific memory T cell response detected in the spleen of mycobacteria-infected guinea pigs has not been fully determined, primarily due to the paucity of useful reagents for guinea pig studies. Interestingly, the TDM-elicited eosinophilic hypersensitivity was not induced in mice (data not shown), underscoring that the use of the guinea pig model is more appropriate. This may also suggest that immune elements, such as group 1 CD1 molecules, that exist in guinea pigs but not in mice could mediate the response. However, the blocking experiments with specific Abs indicated that none of the seven guinea pig group 1 CD1 molecules, namely four homologues of human CD1b and three homologues of human CD1c (16, 23), might not be involved directly in the TDM-elicited T cell activation (Fig. 5). Besides T cells of the adaptive immunity, those that function in the interphase between innate and adaptive immunity, such as group 2 CD1 (CD1d)-reactive NKT cells, may be relevant to the guinea pig model of the TDM-elicited hypersensitivity. TDM-induced granulomatous responses have been shown to be impaired in mice in the absence of CD1d function (24), suggesting a role for NKT cells in hypersensitive responses in mycobacterial infection. Further, a prolonged expansion of NKT cells has been noted during mycobacterial infection (25), possibly by recognition of mycobacterial phosphatidylinositol mannosides (26). A subsequent challenge with TDM may induce TLR-dependent up-regulation of IL-15 expression in CD1d+ macrophages (27), resulting in activation of a unique subset of group 2 CD1 (CD1d)-reactive NKT cells which has been shown to secrete IL-5 in the presence of CD1d+ APCs and IL-15 (28). So far, guinea pig CD1D genes have not been identified (23), but conservation of the genes across mammalian species suggests that NKT cells may also exist in guinea pigs. Alternatively, it is still possible that CD1-unrestricted T cell subsets that include γδ T cells may also contribute to the eosinophilic skin response.
The infiltration of eosinophils at the site of infection has been observed in human cases with tuberculosis (29, 30, 31, 32) and in animals infected with either M. tuberculosis, M. bovis, or M. avium (33, 34, 35, 36), but microbial components that trigger the eosinophilic response were unclear. The present study indicates that TDM and some other undefined lipid components, possibly trehalose monomycolate (TMM), in the methanol fraction (Fig. 2 C) are potent in triggering the eosinophilic response. Immune recognition of mycobacterial TDM followed by eosinophilic inflammation may function as a component of host defense because of the anti-mycobacterial effect of eosinophil peroxidase present at high levels in specific granules of eosinophils (37). The TDM-elicited eosinophilic response may highlight a new aspect of protective and pathological processes involved in mycobacterial infection.
This also raises an issue of how mycobacteria suppress or escape from the noxious attack from the host eosinophils. We recently found that mycobacteria could reduce the expression of TDM soon after their entry into the host by borrowing host-derived glucose to replace TDM with GMM. Mycobacterial mycolyltransferases are known to catalyze the final step of TDM synthesis from its precursor, TMM, by addition of a mycolyl acyl group. Upon exposure to host-derived glucose, however, competitive substrate selection of TMM and glucose by mycolyltransferases results in down-regulation of TDM accompanied with up-regulated biosynthesis of GMM (38). As shown in Fig. 4 B, GMM is inefficient in eliciting the eosinophilic response, and thus the mycolyltransferase-mediated glycolipid exchange may be an evasive maneuver that mycobacteria have evolved to avoid the eosinophilic host response. It is now intriguing to propose that GMM-specific, CD1-restricted T cell responses, detected in human cases with mycobacterial infection (10, 18, 19), come into play to control these GMM-rich mycobacteria. Cell wall mycolyl glycolipids have often proved critical for survival and virulence of mycobacteria (11). The present study suggests that host defense mechanisms directed against these critical glycolipid components can be activated at different stages of immunity, providing the host with an opportunity to efficiently monitor and control infection with mycobacteria.
We thank Dr. Chris C. Dascher (Mount Sinai School of Medicine, New York, NY) for the critical reading of the manuscript. We also thank Dr. Ikuya Yano (BCG laboratory, Tokyo, Japan) for providing valuable materials.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology (Grant-in-Aid from Scientific Research on Priority Areas) (to M.S.), and from the Japan Society for the Promotion of Science (Grant-in-Aid for Scientific Research (B) (to M.S.) and (C) (to I.M.)).
Abbreviations used in this paper: DTH, delayed-type hypersensitivity; BCG, bacillus Calmette-Guerin; MAC, Mycobacterium avium complex; PMN, polymorphonuclear cell; PPD, purified protein derivative; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GMM, glucose monomycolate; GPL, glycopeptidolipid; TDM, trehalose dimycolate; TLC, thin layer chromatography; TMM, trehalose monomycolate.