Autocrine Deactivation of Macrophages in Transgenic Mice Constitutively Overexpressing IL-10 Under Control of the Human CD68 Promoter¹

Interleukin-10 is produced by macrophages, T cells, B cells, and a variety of other cell types including mast cells, keratinocytes, and some tumor cell lines. The effects of IL-10 on immune responses are mostly inhibitory (1). In macrophages, IL-10 inhibits production of proinflammatory cytokines such as TNF-α, IL-6, and IL-12 (2, 3, 4). Inhibition of IL-12 production by IL-10 (4) may be a mechanism by which IL-10 weakens the development of Th1-type T cell responses (5). However, IL-10 can also have immunostimulatory effects by inducing the proliferation of mast cells (6) and CD8 T cells (7). As demonstrated in IL-10 gene-deficient mice, the absence of IL-10 causes overproduction of inflammatory cytokines after LPS challenge (8) and development of chronic inflammatory bowel disease (9). However, IL-10-deficient mice also show increased resistance to intracellular pathogens as diverse as Leishmania major (10), Listeria monocytogenes (11), Chlamydia trachomatis (12), Mycobacterium avium (13), and Mycobacterium bovis bacille Calmette-Guérin (BCG)³ (14, 15). Thus, IL-10 is required to prevent immunopathology, but it can also delay or impair protective responses against pathogens.

We previously found that overexpression of IL-10 from T cells in a transgenic mouse line impairs clearance of BCG (16). T cells from these mice produced abundant IFN-γ and IL-2 in response to BCG infection, indicating that excess IL-10 inhibited killing of BCG at the level of the infected macrophage without affecting T cell responses. The use of T cell IL-10-transgenic mice has also provided important information on the role of T cell-derived IL-10 in models of autoimmune diseases (17, 18). However, because IL-10 is only overexpressed in these mice when T cells become activated, effects of overproduction of IL-10 by innate immune cells cannot be studied. The phenotype of mice with a macrophage-specific disruption of STAT3 (19) closely resembles that of IL-10-deficient mice, suggesting that macrophages are the major target of IL-10 in vivo. For these reasons, we were interested in creating a mouse model to analyze the consequences of overexpression of IL-10 from macrophages themselves. To direct transgene expression to macrophages, promoter fragments of the CD11b (20), c-fms (21), scavenger receptor A (SR-A) (22), lysozyme (23), and MHC class II (MHC-II) (24) genes have been used. None of these constructs is ideal for macrophage-specific expression of transgenes, because they are either not restricted to macrophages or only active in a subset of these cells. Human CD68 and its murine homolog macrosialin are expressed in the endosomal compartment of cells of the mononuclear phagocyte lineage (25, 26, 27). To achieve macrophage-specific overexpression of murine IL-10 tagged with a Flag epitope in vivo, we have used an expression cassette that combines 2.9 kb of the CD68 5′ flanking region with the 83-bp first intron of the CD68 gene, which has been shown to contain a macrophage-specific enhancer (28). This combination of human CD68 gene sequences directed macrophage-specific expression of a type III human SR-A in transgenic mouse lines (29).

In this study, we show that human CD68 sequences can be used to direct constitutive expression of a Flag-IL-10 transgene in vivo specifically in macrophages. Transgenic IL-10 acted on macrophages in an autocrine manner, resulting in deactivated macrophages with impaired capacity to produce inflammatory cytokines when challenged with LPS. Although adaptive immune responses were largely intact, macIL-10tg mice were unable to clear mycobacteria efficiently.

The murine IL-10 cDNA was amplified by PCR to lack the endogenous sequence encoding the signal peptide. The PCR product was digested with MluI and NheI (introduced in the amplifying oligonucleotides) and cloned into the same sites in pEF-BOS/s/Flag (gift of Dr. D. Hilton, Walter and Eliza Hall Institute, Melbourne, Australia). The IL-10 coding region is fused in frame to DNA encoding the IL-3 signal sequence and the Flag epitope. To express Flag-IL-10 under control of the CD68 promoter, we made use of the CD68 promoter construct described recently (29), which combines the 2940 bp of sequence 5′ to the ATG and the 83-bp first intron of the human CD68 gene in a pcDNA3 backbone where the CMV promoter had been first deleted. The human SR-A sequence was removed by digesting the construct with XbaI. The Flag-IL-10 fragment was released from pEF-BOS/s/Flag-IL-10 with NheI and ligated into the XbaI sites of the CD68 promoter construct. To express Flag-IL-10 under control of different fragments of the CD68 promoter, the Flag-IL-10 cDNA was cloned into a version of pcDNA3.1 lacking the CMV promoter. Fragments of the CD68 promoter were amplified from the expression cassette and cloned upstream of the Flag-IL-10 cDNA using EcoRV and NotI. For injection into male pronuclei of FVB zygotes, the constructs were digested with the enzymes indicated in Fig. 1 and purified according to standard procedures, and the concentration was adjusted to 1.5 ng/μl. Offspring derived from the injections was genotyped by 3-primer PCR amplifying a 150-bp fragment of the transgene and a 200-bp fragment of the murine IL-10 gene. Southern blotting of DNA digested with EcoRI, followed by probing the membrane with the CD68-Flag-IL-10 fragment labeled with [α-³²P]dCTP, confirmed the PCR results and was used to estimate copy numbers of integrated transgene. Transgene-bearing founder mice were mated with FVB mice. As described in the results, only one founder mouse (TG445) produced viable offspring. This line was designated macIL-10tg and was bred hemizygously by mating transgene-positive males with wild-type FVB females.

Bone marrow-derived macrophages (BMDM) were generated as described previously (16, 30) by differentiation in 15% L cell-conditioned medium as a source of M-CSF for 5–7 days. To obtain peritoneal-derived macrophages (PDM), mice were injected with 3 ml of 4% Brewer’s thioglycolate and were sacrificed after 3 days, and the peritoneal cavity was flushed with 10 ml ice-cold PBS. Cell culture medium for macrophage cultures was DMEM supplemented with 10% FBS, 2-ME (50 μM), and antibiotics. Cell suspensions of spleens, lymph nodes, and thymus were prepared by straining the organs through a nylon mesh (Falcon, Mountain View, CA) and lysis of erythrocytes with NH₄Cl followed by two washes with complete medium and counting. These cells were cultured in RPMI 1640 supplemented as above stated for DMEM.

A total of 5 × 10⁴ splenocytes per well were plated in quadruplicate in 96-well U-bottom plates and stimulated with anti-CD3 mAb. After 40 h, [³H]thymidine (0.25 μCi/well) was added for 8 h, the plates were harvested onto filters, and the amount of incorporated radioactivity was determined with a scintillation counter.

The concentration of IL-10, Flag-IL-10, TNF-α, IFN-γ, and IL-12p40 in sera or culture supernatants was determined at appropriate dilutions by sandwich ELISA using Ab pairs IL-10, IL-12p40, TNF-α (all purchased from BD PharMingen, San Diego, CA), and IFN-γ (Endogen, Woburn, MA). For detection of Flag-IL-10, the plates were coated with anti-Flag M2 Ab (Sigma-Aldrich, St. Louis, MO) to capture specifically Flag-IL-10; detection was performed using biotinylated anti-IL-10 Ab.

Biotinylated anti-mouse F4/80 Ab was from Serotec (Raleigh, NC). All other Abs and reagents for flow cytometry were from BD PharMingen. For staining of cell surface proteins, cells were washed in PBS with 2% FBS and then incubated for 30 min at 4°C with the respective combinations of fluorescently labeled Abs. For analytical purposes, cells were analyzed on a FACSCalibur (BD Biosciences, San Jose, CA) and sorting was done at the St. Jude Core Facility using a MoFlo sorter (Cytomation, Fort Collins, CO). To stain for intracellular cytokines, after the staining procedure for cell surface proteins was completed cells were fixed in 2% paraformaldehyde for at least 20 min at room temperature. Cells were then washed two times in staining buffer (PBS with 0.5% saponin, 0.5% BSA, and 0.05% azide), followed by a 30-min incubation with PE-labeled anti-TNF-α or anti-IL-10 Ab in staining buffer. After washing again two times, cells were taken up in PBS and analyzed.

Total RNA from cells and mouse tissues was prepared using the Absolutely RNA kit (Stratagene, La Jolla, CA), which includes a DNase I digestion step to remove contaminating genomic DNA. One microgram of RNA was reverse-transcribed using Superscript II (Life Technologies, Rockville, MD) and a mixture of oligo(dT) and random primers. Real-time quantitative PCR was performed on a SDS7700 instrument (PE Applied Biosystems, Foster City, CA). Primers and probes were designed using PrimerExpress software and had the following sequences: β-actin (sense 5′-ACCCACACTGTGCCCATCTAC-3′; antisense 5′-AGCCAAGTCCAGACGCAGG-3′; probe 5′-AGGGCTATGCTCTCCCTCACGCCA-3′), IL-10 (antisense 5′-CCCAAGTAACCCTTAAAGTCCTGC-3′; probe 5′-CCCAAGTAACCCTTAAAGTCCTGC-3′, sense 5′-ATAACTGCACCCACTTCCCAGTC-3′), murine IL-10 (sense 5′-ATGCTGCCTGCTCTTACTGACTG-3′), Flag-IL-10 (sense 5′-CAGGACTACAAGGACGACGATGAC-3′). Fluorogenic probes were labeled with FAM at the 5′ and with TAMRA at the 3′ end and synthesized by PE Applied Biosystems. Because primers for β-actin and Flag-IL-10 amplified genomic DNA also, controls of RNA samples were included to control the efficiency of DNase I digestion. To generate standard curves, PCR products were cloned into pCR2.1 (Invitrogen, Carlsbad, CA), and miniprep DNA was digested with EcoRI and, after estimation of DNA quantity on an agarose gel, diluted serially down from 10⁶ to 1 copies/μl. cDNA samples and standard dilutions were analyzed in duplicate.

Mice were immunized with 100 μg/footpad OVA (Sigma-Aldrich) in CFA (Pierce, Rockford, IL) followed by a booster injection with 100 μg/footpad OVA in IFA 10 days later. Twenty days after the first injection, mice were sacrificed, blood was collected, and the popliteal lymph nodes were prepared. OVA-specific Ab titers were determined by ELISA. Briefly, plates were coated with OVA (2 μg/ml) and blocked with PBS containing 10% FBS. Sera were diluted down 3-fold serially and allowed to bind. After extensive washing with PBS 0.05% Tween, OVA-specific Ig isotypes were detected by incubation with AP-conjugated isotype-specific goat anti-mouse Abs. Again after extensive washing, substrate was added and the OD₄₅₀ was measured. The last serum dilution yielding an OD₄₅₀ value of 0.1 over background was recorded as the endpoint titer for each sample.

Escherichia coli LPS was purchased from Sigma-Aldrich (catalog no. L4130), dissolved at 5 mg/ml in PBS, and diluted before injection, and a volume of 200 μl was injected i.p. At the indicated time points, mice were sacrificed, blood was collected, and the serum was separated by centrifugation.

BCG Pasteur was grown in 40 ml Middlebrook 7H9 broth for 3 days, washed twice with PBS with 0.02% Tween 80, taken up in 20 ml and diluted 1/20 in PBS with 0.02% Tween 80, and sonicated in a water bath sonicator for 15 s to break up clumps. Mice were injected in a lateral tail vein with 200 μl of this solution, which corresponded to ∼10⁶ viable bacteria, as determined by plating of the inoculum. After 2, 5, and 8 wk, groups of mice were sacrificed and tail DNA was prepared for confirmatory genotyping. The left lung, upper half of the spleen, and the upper lobe of the liver were used for determination of CFU. A lobe of the right lung, a piece of the spleen, and the liver were placed in 4% formaldehyde for histology. The rest of the spleen was taken for preparation of splenocytes for restimulation in vitro.

Splenocytes from groups of six BCG-infected or control mice were prepared as described above and pooled. A total of 2.5 × 10⁸ splenocytes were resuspended in 1.8 ml of ice-cold PBS containing 2% FBS (wash buffer) followed by addition of 200 μl mouse CD4 (L3T4) MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany). Samples were incubated for 30 min at 4°C followed by magnetic separation with a LS positive selection column (Miltenyi Biotec). After washing the column with 3 ml wash buffer, the cell suspension was applied to the column and negative cells were allowed to pass through. The column was washed three times with 3 ml wash buffer. Then the column was removed from the separator and the CD4 T cells were flushed out with 6 ml wash buffer. Cells were centrifuged and resuspended in complete RPMI 1640.

BMDM were plated in 12-well plates at 2 × 10⁶/well and harvested in 150 μl RIPA buffer containing a protease inhibitor mixture (Boehringer Mannheim, Mannheim, Germany) and sodium orthovanadate to inhibit phosphatase activity. After 30 min on ice, lysates were cleared from debris and SDS sample buffer was added. A total of 40 μl lysate was used for SDS-PAGE on 4–15% gradient gels (Bio-Rad). Gels were blotted at 70 V for 2 h onto nitrocellulose membranes. After confirming even transfer by Ponceau staining, the membrane was blocked with 3% nonfat dry milk in PBS with 0.05% Tween 20 for 1 h at room temperature. Primary Abs were diluted in blocking buffer and the membrane was incubated overnight at 4°C. After washing with PBS 0.05% Tween 20, membranes were exposed to peroxidase-conjugated anti-rabbit sera (1/5000 dilution in blocking buffer) followed by three washes in PBS 0.05% Tween 20. Blots were developed using ECL substrate and exposure times between 10 s and 5 min as needed to yield the appropriate signal. Abs against STAT1 and STAT3 and phosphorylated STAT1 and STAT3 were from New England Biolabs/Cell Signaling (Beverly, MA). The anti-suppressor of cytokine signaling (SOCS)3 antiserum was a kind gift from Dr. Y. Takahashi (St. Jude Children’s Research Hospital, Memphis, TN).

Constructs for expression of Flag-IL-10 under control of three different fragments of the CD68 promoter (Fig. 1) were injected into male pronuclei of fertilized FVB oocytes. Six transgenic founder mice were obtained (Table I). Although all founder mice developed normally, only TG445 produced viable offspring. The other transgenic founder mice appeared ill after 6–10 wk of age and finally died or were euthanized for humane reasons. From two of these founder mice (TG454 and TG520) serum and tissue samples were analyzed and very high levels of IL-10 were detected in the serum, associated with high copy numbers of the transgene as determined by Southern blotting (Table I and Fig. 1). Although the cause of disease and death in these mice was unclear, histological findings were consistent with systemic infection, spreading from pyelonephritis in one mouse (TG520) and a lumbar soft tissue abscess in another (TG454) (Table I). Seminiferous tubular degeneration and aspermatogenesis in founder TG520 explain the infertility observed. The male founder TG445, harboring only one or two copies of the transgene, showed no signs of disease and transmitted the transgene in a mendelian fashion. This line was designated macIL-10tg and is described in this work.

Although IL-10 was undetectable in the sera of FVB mice, it was detected at moderate levels in macIL-10tg mice (Fig. 2,A). Injection of a biotinylated anti-IL-10 Ab greatly increased the sensitivity of detection, as described for IL-4 and IFN-γ (31), and revealed that the amount of IL-10 produced in macIL-10tg under normal conditions is similar to levels observed when FVB mice were injected with LPS (Fig. 2,A). At the mRNA level, real-time quantitative RT-PCR showed that in spleens of untreated macIL-10tg mice IL-10 mRNA is ∼100-fold more abundant than in FVB control spleen (Fig. 2,B). At the protein and at the mRNA level this difference is entirely due to transgenic Flag-IL-10 (Fig. 2, A and B). When cells from different tissues were cultured without stimulation, IL-10 could be detected in the supernatants of macIL-10tg splenocyte and bone marrow cell cultures and, to a much lesser extent, of lymph node cells but not thymocytes (Fig. 2,C). To define the source of IL-10 more directly, adherent splenocytes were cultured in the presence of brefeldin A to block secretion of cytokines and were stained for the macrophage-specific surface marker F4/80 and intracellular IL-10 (Fig. 2,D). IL-10 was detected in the F4/80⁺ macrophage population in macIL-10tg splenocytes but not in the FVB counterpart. Similarly, when splenocytes from macIL-10tg mice and FVB littermate controls were FACS sorted for T cell, B cell, granulocyte, and macrophage lineage cells, enrichment for macrophages (Mac1⁺GR1⁻) strongly increased the amount of IL-10 produced from macIL-10tg cells (Fig. 2,E). Finally, BMDM from macIL-10tg mice produced high levels of IL-10 that slightly increased after stimulation with LPS (Fig. 2 F), whereas FVB BMDM produced IL-10 only after stimulation with LPS. Taken together, these experiments show that Flag-IL-10 is expressed in a macrophage-specific and constitutive manner in macIL-10tg mice.

When compared with FVB littermates, macIL-10tg mice showed no major differences in the distribution of T cells and B lymphocytes in spleen and mesenteric lymph node, except for a slightly reduced frequency of CD8⁺ T cells (Table II). The percentage of Mac1⁺ cells was slightly but significantly increased in macIL-10tg spleens. Overall, lymphoid development and the cellular composition of the peripheral lymphoid organs appeared normal in macIL-10tg mice. When splenocytes were stimulated with anti-CD3, proliferation as measured by incorporation of [³H]thymidine was indistinguishable between FVB and macIL-10tg splenocytes (Fig. 3,A). The ability of macIL-10tg mice to mount an Ag-specific immune response was tested by immunization with OVA in CFA (Fig. 3 B). The intensity and isotype distribution of OVA-specific Ab production did not differ between FVB and macIL-10tg mice, as determined by endpoint titers. In addition to an unaltered B cell response, macIL-10tg mice developed strong T cell memory, because restimulation of popliteal lymph node cells with OVA in vitro resulted in strong production of IFN-γ (data not shown).

Consequences of overexpression of IL-10 on macrophage function were first evaluated in vitro using both PDM and BMDM. Compared with FVB macrophages, stimulation of macIL-10tg macrophages with LPS resulted in strongly reduced amounts of both TNF-α (reduced by 83.6%) and IL-12p40 (71.4%) in the culture supernatants (Fig. 4,A). As expected, addition of exogenous IL-10 strongly reduced the levels of both cytokines in FVB macrophages but had much less effect on macIL-10tg macrophages. Importantly, addition of a neutralizing anti-IL-10 Ab to the cultures restored cytokine production from macIL-10tg macrophages, demonstrating that the inhibition observed is due to transgenic Flag-IL-10. Compared with the strong inhibition of TNF-α and IL-12p40, production of NO by macrophages stimulated with LPS was less affected in macIL-10tg (30.2% reduction)- and IL-10-treated FVB macrophages (Fig. 4,A). In the case of TNF-α, the kinetics of production after LPS stimulation were analyzed by intracellular FACS staining (Fig. 4 B). In FVB macrophages, LPS treatment induced a sharp increase in the percentage of TNF-α⁺ cells (45.1% after 30 min and 63.1% after 75 min). After 150 min, TNF-α synthesis had begun to decline, returning to 16% positive cells by 300 min. The course of TNF-α staining in macIL-10tg macrophages stimulated with LPS was very similar to that of FVB macrophages exposed to IL-10 plus LPS. There was an increase in the percentage of TNF-α⁺ cells, but transgenic and exogenous IL-10 blunted the response by ∼40% at the 30- and 75-min time points. Furthermore, IL-10 induced a faster down-regulation of TNF-α production, with TNF-α staining approaching baseline levels after 150 min and a >70% inhibition relative to FVB macrophages.

We studied whether macIL-10tg mice also produce less TNF-α and IL-12p40 when challenged with LPS in vivo. After i.p. injection of 200 μg LPS, the serum levels of IL-12p40, TNF-α, and IFN-γ were measured (Fig. 5). In macIL-10tg mice the response to LPS was dampened with a significant reduction in the levels of TNF-α after 1.5 h, IFN-γ after 6 h, and both time points in the case of IL-12p40. Inhibition of cytokine production was not as strong as in the in vitro culture system (Fig. 4), which is probably due to the higher relative concentrations of Flag-IL-10 in pure macrophage cultures compared with in vivo, where the Flag-IL-10 produced will be taken up and removed continuously.

To investigate the role of macrophage IL-10 production in mycobacterial infection, macIL-10tg and FVB mice were infected i.v. with BCG. macIL-10tg mice, as well as FVB mice, controlled the infection and appeared outwardly healthy. At 2, 5, and 8 wk after infection, mice were killed and tissue homogenates of spleen, liver, and lung were plated onto 7H10 plates and the number of CFU per organ was determined (Fig. 6,A). A difference in CFU was apparent after 2 wk in all three organs but was stronger at the 5-wk time point. This higher bacterial load was maintained after 8 wk of infection in liver and spleen but not in the lung. Histopathological analysis of spleen and liver sections showed that the number of granulomas in the livers was roughly equal between macIL-10tg mice and FVB controls at all three time points. Staining for the presence of acid-fast bacilli revealed that the number of mycobacteria in the granulomas was higher in macIL-10tg mice than in FVB mice (Fig. 6,B). The development of a BCG-specific T cell response after infection was assessed by measurement of IFN-γ in culture supernatants of splenocytes from infected FVB and macIL-10tg mice restimulated in vitro with heat-killed BCG (Fig. 7). IFN-γ production from macIL-10tg splenocytes was significantly weaker than from FVB cultures 5 wk after infection, but by 8 wk after infection this difference was no longer observed and macIL-10tg splenocytes produced even more IFN-γ than controls. To dissect the contributions of the T cell and APC components of the IFN-γ response, we purified CD4 T cells from BCG-infected mice as well as control FVB and macIL-10tg mice. We used wild-type and transgenic BMDM as APC for restimulation in vitro (Fig. 8). As expected, CD4 T cells from uninfected FVB and macIL-10tg mice did not produce detectable IFN-γ when exposed to BMDM pulsed with purified protein derivative (PPD) or infected with BCG (data not shown). IFN-γ levels induced by anti-CD3/Con A in control CD4 T cells were highest for the combination of FVB CD4 T cells with FVB BMDM and lowest when both cell types came from macIL-10tg mice. Infection with BCG resulted in strongly increased IFN-γ production upon stimulation with anti-CD3/Con A for all combinations of T cells and APC. Purified CD4 T cells from BCG-infected macIL-10tg mice showed higher PPD- and BCG-specific IFN-γ production than their FVB counterparts. In contrast, macIL-10tg BMDM induced a much weaker IFN-γ response than those from FVB mice. This difference was also evident after stimulation with anti-CD3/Con A but more pronounced for PPD- or BCG-specific IFN-γ. Thus, the use of purified CD4 T cells and defined APC showed that BCG-specific Th1 T cell development is not impaired in macIL-10tg mice. Together with normal development of granulomas in macIL-10tg mice, the main reason for the impaired clearance of BCG appeared to be an inhibition of intracellular killing by infected macrophages.

Increased survival of mycobacteria in macIL-10tg macrophages may be due to a disturbed response to IFN-γ. To address this issue, we chose to analyze the phosphorylation of STAT1, an early event in IFN-γ signal transduction, in BMDM from wild-type FVB and macIL-10tg mice. Macrophages from macIL-10tg mice showed attenuated phosphorylation of STAT1 when stimulated with titrated amounts of IFN-γ (Fig. 9,A). Because inhibition of STAT1 activation by IL-10 has been correlated with expression of SOCS3 (32), protein levels of SOCS3 were determined in BMDM (Fig. 9 B). SOCS3 was absent in unstimulated FVB macrophages but was induced 3 h after addition of IFN-γ. In contrast, macIL-10tg macrophages contained constitutively expressed SOCS3, which was further increased by treatment with IFN-γ, while neutralization of IL-10 greatly reduced the level of SOCS3. Overexpression of IL-10 in macIL-10tg macrophages also resulted in constitutive phosphorylation of STAT3, which was transiently activated by IFN-γ in FVB macrophages.

This study describes a new transgenic mouse model where an epitope-tagged IL-10 is constitutively overexpressed by macrophages in vivo under the control of the CD68 promoter. Macrophage-specific overexpression of IL-10 induced deactivation of macrophages but had little impact on development of Ag-specific T and B cell responses. Self-deactivation of macrophages in macIL-10tg mice significantly affected the inflammatory response to LPS challenge and the ability to clear infection with the intracellular pathogen BCG.

Several lines of evidence indicate that expression of the Flag-IL-10 transgene by the CD68 promoter was restricted to macrophages. First, when cells from different tissues were assayed for Flag-IL-10 production, significantly more IL-10 was detected in supernatants of cultures enriched in macrophages. Second, sorting of splenocytes showed strong enrichment of Flag-IL-10 secretion in the Mac1⁺GR1⁻ population. Finally, intracellular staining detected IL-10 only in the F4/80⁺ population within adherent splenocytes of macIL-10tg mice. Overall, our findings confirm the usefulness of the CD68 promoter for macrophage-specific expression in transgenic mice (29). Expression of Flag-IL-10 in macrophages was constitutive and at least as strong as IL-10 secretion induced by treatment with LPS, at both the mRNA and protein levels. In the serum of macIL-10tg mice, IL-10 was present at concentrations comparable to what was induced by LPS in FVB mice.

Several founder mice derived from injection with different fragments of the CD68 promoter controlling Flag-IL-10 died prematurely without producing viable offspring. The copy number of integrated transgene was estimated to be between 20 and 40 in these animals, and very high levels of Flag-IL-10 were detected in the serum (Table I). Although there are no published data on toxic effects of IL-10, it cannot be ruled out that chronic exposure to excessive amounts of IL-10 can have direct devastating effects on multiple organ systems. In contrast, the pathological findings we obtained are compatible with systemic infection, in one case ascending from the genito-urinary tract, in the other spread from a large abscess. Thus, high chronic IL-10 levels in these founder mice may have caused paralysis of the immune system leading to terminal infection.

Overexpression of Flag-IL-10 in macIL-10tg mice did not obviously interfere with peripheral T and B cell homeostasis. This is consistent with other transgenic mouse lines overexpressing IL-10 either from activated T cells (16, 33) or under control of the MHC-II promoter (24). The induction of Ag-specific adaptive immunity appeared unaffected despite elevated levels of IL-10 in macIL-10tg mice (Fig. 3). This is at variance with the recently reported inhibition of Ag-specific responses by T and B cells in a transgenic mouse line expressing human IL-10 under the control of the MHC-II promoter (hIL-10 TG) (Ref. 24 and this study). This discrepancy is somewhat surprising given that IL-10 is expressed from APCs in both models, reaching similar steady-state levels in the serum of mice (Ref. 24 and this study). The cellular source of IL-10 production directed by the different promoters may be relevant here. While macIL-10tg B cells did not make Flag-IL-10, B cells from hIL-10 TG strongly secreted transgenic IL-10 upon stimulation (24). Also, the expression of IL-10 by dendritic cells can be expected to differ between the transgenic mouse lines, although Flag-IL-10 expression by dendritic cells was not tested directly in macIL-10tg. In addition to spatial differences in IL-10 production, the temporal context of constitutive or inducible overexpression of IL-10 may explain the differential effects on adaptive immune responses in macIL-10tg and hIL-10 TG mouse lines.

Stimulation of macIL-10tg macrophages in vitro with LPS resulted in similarly down-regulated levels of TNF-α and IL-12p40 as were observed in control macrophages treated with IL-10, and neutralization of IL-10 increased cytokine production by macIL-10tg macrophages. Importantly, upon challenge with LPS in vivo, a similar reduction in the output of TNF-α and IL-12p40 was observed. In addition, IFN-γ was significantly lower in macIL-10tg, probably reflecting a secondary effect of decreased IL-12 levels in the transgenic mice. These effects on the innate response to LPS are consistent with work by others using exogenous administration of IL-10 (34, 35, 36) or IL-10 gene transfer (37).

Although down-regulation of cytokine production by macrophages is desirable and beneficial in autoimmune diseases and septic shock, it may negatively influence the capacity to fight infections. Infection of macIL-10tg resulted in 10-fold-higher chronic bacterial loads in spleen and liver compared with control mice. In its effects on the clearance of infection with BCG, macrophage-specific overexpression of IL-10 is similar to what we have found previously for mice overexpressing IL-10 from activated T cells (16), thereby strengthening the argument that increased production of IL-10 during mycobacterial infection favors bacterial growth. Because IFN-γ production by T cells is essential for defense against mycobacteria (38, 39) and IL-10 can inhibit Th1 responses, one possible mechanism for increased numbers of bacteria in IL-10tg mice is impaired Th1 development. However, it appears that macIL-10tg mice mount a strong, if slightly delayed, T cell IFN-γ response to BCG. Thus, overexpression of IL-10 does not qualitatively alter IFN-γ production from T cells in BCG-infected mice. Depletion of TNF-α or the TNF type I receptor in mice by neutralization with Ab or gene targeting renders mice highly susceptible to infection with Mycobacterium tuberculosis (40, 41). Despite induction of a strong T cell response and recruitment of activated T cells to the lungs of infected TNF-α^−/− mice, the normally observed typical granuloma structure failed to develop (41). Inhibition of TNF-α levels by IL-10 might therefore result in disturbed granuloma formation. However, in BCG-infected macIL-10tg mice, granulomas developed at normal numbers and with apparently normal structural characteristics compared with FVB mice. Preliminary data suggest that TNF-α mRNA expression is reduced ∼2-fold in BCG-infected macIL-10tg mice (R. Lang, unpublished data), which may not be enough to cause effects similar to complete depletion by gene targeting. In contrast, the number of acid-fast bacilli per granuloma was increased in macIL-10tg mice. We conclude that the killing of the mycobacteria within infected macrophages is impaired.

IL-10 may inhibit macrophage killing of BCG by various mechanisms. For efficient control of mycobacteria in macrophages, expression of inducible NO synthase (iNOS) and production of NO is essential (42, 43). Treatment of murine macrophages with TNF-α and IFN-γ potentiates NO generation after mycobacterial infection and increases antimycobacterial activity (44). Production of IFN-γ by splenocytes from macIL-10tg mice was reduced only early in infection (Fig. 7), and purified CD4 T cells produced high levels of IFN-γ (Fig. 8). However, IL-10 could interfere with the signal transduction of IFN-γ and consequently inhibit the expression of iNOS. Inhibition of IFN-γ signaling by IL-10 has been described and correlated with the induction of SOCS3 (32). Indeed, when we compared the ability of IFN-γ to activate STAT1 in macIL-10tg and FVB macrophages in vitro, we found a similar inhibition in macIL-10tg macrophages as Ito et al. (32) had observed when IL-10 was added to human monocytes. In addition, constitutive activation of STAT3 and expression of SOCS3 were detected in macIL-10tg macrophages. While these data were obtained in experiments in vitro, preliminary evidence shows that untreated macIL-10tg mice have slightly (2- to 3-fold) increased levels of SOCS3 mRNA in tissues (our unpublished observations), suggesting a possible effect on IFN-γ signaling also in vivo. In addition to an iNOS-dependent pathway, TNF-α triggers growth control of BCG in macrophages in a iNOS-independent manner (45). By decreasing the levels of TNF-α, IL-10 may inhibit iNOS-dependent and -independent bactericidal macrophage functions. Macrophages from macIL-10tg mice displayed moderate inhibition of NO production after stimulation with LPS, consistent with previous reports on inhibition of NO production by IL-10 (46, 47). An important intracellular survival mechanism of pathogenic mycobacteria is their ability to prevent the maturation of the phagosome and thereby to evade destruction (48). Interestingly, in IL-10 gene-deficient mice the percentage of BCG colocalizing with a marker of acidified phagolysosomes was significantly higher than in macrophages capable of producing IL-10 (49). Likewise, overexpression of IL-10 may affect the intracellular trafficking of the bacteria and further inhibit the maturation of the phagosome. Taken together, IL-10 probably affects macrophage killing of intracellular BCG at multiple levels.

The macIL-10tg mice we have generated will provide a useful tool to analyze the effects of macrophage-specific overexpression of the immunoregulatory cytokine IL-10. This model should prove valuable not only in studying the immunology of infection but also in other fields where macrophages play a role in pathophysiology, such as tumor immunology and atherosclerosis.

We thank John Swift and John Raucci for pronuclei injection, Richard Cross for cell sorting, Mark Sangster, Yutaka Takahashi, and Doug Hilton for reagents, and Jurgen Bock and Roberto Hernan for help with microscopy and [³H]thymidine incorporation assays. Also, we are grateful to the members of the stem cell transplantation lab for assistance on how to use MACS technology.