The CXC Chemokine Murine Monokine Induced by IFN-γ (CXC Chemokine Ligand 9) Is Made by APCs, Targets Lymphocytes Including Activated B Cells, and Supports Antibody Responses to a Bacterial Pathogen In Vivo

Mice

C57BL/6 mice were purchased from the Division of Cancer Therapy of the National Cancer Institute (Frederick, MD); 129/Sv mice were obtained from the Division of Cancer Therapy of the National Cancer Institute and bred at the National Institutes of Health (Bethesda, MD); and C57BL/6J-TCRβ⁻TCRδ⁻ mice were purchased from The Jackson Laboratory (Bar Harbor, ME). The mig^−/− mice were derived from two colonies of gene-targeted embryonic stem (ES) cells and used in the 129/Sv, (129/Sv × C57BL/6)F₂, and C57BL/6 backgrounds (see below). All mice ranged between 8 and 20 wk old and were housed in sterile cages under pathogen-free conditions. Animal protocols were approved by the Animal Care and Use Committees of the National Institute of Allergy and Infectious Diseases (National Institutes of Health) and the Center for Biologics Evaluation and Research (Food and Drug Administration, Bethesda, MD).

Preparing, culturing, and activating mouse leukocytes

RPMI 1640 (Life Technologies, Gaithersburg, MD) for culturing mouse lymphocytes contained 10% FBS, 55 μM 2-ME, 1 mM pyruvate, 2 mM glutamine, 100 μM nonessential amino acids, and 100 U/ml penicillin-streptomycin. Cell staining for analyzing and sorting lymphocytes was done with the following Abs: for CD4, clone GK1.5 conjugated to allophycocyanin (Leinco Technologies, Ballwin, MO) or clone H129.19 conjugated to FITC or Cy5-PE (BD PharMingen, San Diego, CA); for CD8α, clone 53-6.7 conjugated to FITC, PE, or allophycocyanin; for CD45R/B220, clone RA3-6B2, conjugated to FITC, PE, or allophycocyanin; for CD11b, clone M1/70 conjugated to FITC or allophycocyanin; for CD11c, clone HL3 conjugated to FITC; for NK1.1, clone PK136 conjugated to FITC; for CD62L, clone MEL-14 conjugated to FITC; and for CD44, clone IM7 conjugated to FITC, PE, or Cy5-PE (all from BD PharMingen).

Assays for T cell chemotaxis and staining of T cells for flow cytometry used splenocyte preparations from C57BL/6 mice. For activation of T cells in the splenocyte population, 10⁶ splenocytes per milliliter were cultured in 24-well plates that had been coated with 10 μg/ml anti-CD3 (145-2C11; BD PharMingen). Cells were harvested on day 3 or 4. For studying CXCR3 expression on CD8⁺ T cells, splenocytes were first enriched for CD8⁺ T cells using murine T cell CD8 subset columns (R&D Systems, Minneapolis, MN) and then sorted into CD8⁺CXCR3⁻ and CD8⁺CXCR3⁺ populations that were CD4⁻CD11b⁻CD11c⁻B220⁻NK1.1⁻ using a FACStar or FACSVantage cell sorter (BD Immunocytometry Systems, San Jose, CA). These and other populations of sorted cells were always >95% pure. The CD8⁺ T cells were plated at 1 × 10⁶ cells/ml in 96-well plates coated with anti-CD3 (10 μg/ml) and containing anti-CD28 (10 μg/ml, clone 37.51; BD PharMingen) and cultured for 4 days.

For purifying T cells as sources of RNA for RT-PCR, splenocytes from C57BL/6 mice were passed over a T cell enrichment column according to the manufacturer’s protocol (R&D Systems) and CD4⁺ and CD8⁺ lymphocytes were purified by sorting on a FACStar cell sorter for cells that were CD4⁺CD8α⁻CD45R/B220⁻CD11b⁻ and CD8α⁺CD4⁻CD45R/B220⁻CD11b⁻, respectively. For activating these sorted T cells, 5 × 10⁶ cells/ml were incubated for 18 h in 96-well flat-bottom plates with anti-CD3 and anti-CD28 as above.

To isolate B cells for staining, calcium flux, chemotaxis, and preparing RNA for Northern analysis, splenocytes from C57BL/6J-TCRβ⁻TCRδ⁻ mice were plated onto polystyrene dishes at 10⁷ cells/ml in DMEM/2% FBS for 30 min at 37°C and the nonadherent cells were harvested, which were 90% B cells as determined by B220 staining. For purifying B cells as sources of RNA for RT-PCR, splenocytes were harvested from C57BL/6J-TCRβ⁻TCRδ⁻ mice and CD45R/B220⁺CD11b⁻ cells were isolated by cell sorting as above. For activation for staining for flow cytometry, calcium flux, and chemotaxis, B cells were plated at 10⁶ cells/ml onto irradiated, CD40 ligand (CD40L)-expressing fibroblasts (gift from B. Kelsall, National Institute of Allergy and Infectious Diseases, National Institutes of Health) in six-well plates in RPMI medium as above plus 5 ng/ml murine rIL-4 (BioSource International, Rockville, MD) and 10 μg/ml goat F(ab′)₂ anti-mouse IgM (Caltag Laboratories, Burlingame, CA, or ICN Pharmaceuticals, Aurora, OH). After 3 days, 50% fresh medium was added and cells were harvested at day 5. For activating sorted B cells as a source of RNA, 5 × 10⁶ cells/ml were incubated for 18 h in 96-well flat-bottom plates with 10 μg/ml goat F(ab′)₂ anti-mouse IgM (ICN Pharmaceuticals).

Enriched populations of neutrophils and macrophages were obtained at 3 or 72 h, respectively, after i.p. injection of thioglycolate. Cells were >90% neutrophils or macrophages as determined by microscopic analysis of cytospin preparations stained with Hema 3 (Biomedical Science, Swedesboro, NJ). CD8⁺ and CD8⁻ dendritic cells were isolated from spleens of 6- to 10-wk-old BALB/c mice by cell sorting as described (29).

Measurements of chemotaxis

Assays for chemotaxis of T and B cells used either ChemoTx 96-well disposable chemotaxis plates (NeuroProbe, Cabin John, MD) or Transwell plates (Costar, Cambridge, MA). For activated lymphocytes, dead cells were removed using Ficoll-Paque Plus (Amersham Pharmacia Biotech, Uppsala, Sweden) before the assay. The ChemoTx 96-well plates contained 6-mm diameter, 5-μm pore size, polyvinylpyrrolidone-free polycarbonate membranes and were used according to the manufacturer’s protocol. After incubation at 37°C for 3 h, cells from 32 lower wells were pooled, counted, stained, and analyzed as appropriate for expression of CD4, CD8α, or CD45R/B220. Assays using Transwell plates were done as described (30) using membranes with 5-μm pores and 10⁶ cells per well. Assays for chemotaxis of neutrophils and macrophages were done using polyvinylpyrrolidone-free polycarbonate membranes with 3- or 8-μm pores, respectively, in a 48-well microchemotaxis chamber (NeuroProbe). Five randomly selected fields were chosen to count cells at a ×1000 magnification. rMuMig was obtained from BD PharMingen and rHuI-TAC was obtained from PeproTech (Rocky Hill, NJ).

Detecting murine CXCR3 surface expression

To generate the anti-CXCR3 Ab, rabbits were immunized with a peptide-containing sequence from the N-terminal region, YLEVSERQVLDASDFAFOrnC, conjugated to bovine thyroglobulin. Antisera were affinity purified by BioExpress Cell Culture Services (West Lebanon, NH) using peptide coupled to Sulfo-Link gel (Pierce, Rockford, IL). For surface staining for CXCR3, 0.3 mg/ml anti-CXCR3 Ab or rabbit IgG was preincubated with or without 0.5 mg/ml immunizing peptide for 30 min at room temperature, and cells were stained with 30 μg/ml of the pretreated Abs. Following incubation of cells and primary Abs for 1 h at 4°C, biotinylated goat F(ab′)₂ anti-rabbit Ig (H and L chains; Caltag Laboratories) was added, followed by streptavidin-PE (BD PharMingen) before analysis by flow cytometry.

We isolated a mouse CXCR3 cDNA by screening a mouse splenocyte cDNA library in lambda HybriZAP (Stratagene, La Jolla, CA) kindly provided by L. Zhang and M. Lenardo (National Institute of Allergy and Infectious Diseases, National Institutes of Health), using a coding region probe obtained from a genomic clone of the human CXCR3 kindly provided by P. Murphy (National Institute of Allergy and Infectious Diseases, National Institutes of Health). cDNAs were isolated that matched the sequences subsequently published for mouse CXCR3 (31). To determine the specificity of the anti-CXCR3 Ab, Jurkat-TAg cells (derived as described (32) and kindly provided by L. Samelson, National Cancer Institute, National Institutes of Health) were transfected with 40 μg of either pCEFL (33) or pCEFL into which the mouse CXCR3 cDNA had been inserted, plus 25 μg of pEGFP-F encoding a farnesylated green fluorescent protein (GFP; Clontech Laboratories, Palo Alto, CA) as described (34) to identify cells expressing the transfected genes. The cotransfected cells expressing enhanced farnesylated GFP (but not the enhanced GFP-negative cells) stained with the anti-MuCXCR3 Ab, but not with control rabbit IgG, and the Ab staining was blocked by preincubation with the immunizing peptide. Similarly, a transfected RBL-1 cell line expressing MuCXCR3 showed staining with the anti-MuCXCR3 Ab that was blocked by the immunizing peptide, but not by an irrelevant peptide corresponding to N-terminal residues from MuCCR6 (data not shown).

Targeting of the mouse mig gene

Choices of fragments to use for homologous recombination and as probes in targeting the mig gene were based on 7.1 kb of sequence obtained from clones from an EMBL-3, C3H mouse genomic library kindly provided by R. Reeves (Johns Hopkins University) (J. M. Farber and T. M. Wright, unpublished data). To isolate DNA fragments for gene targeting, recombinant bacteriophage P1 clones containing the mig gene from a 129 mouse were obtained from Genome Systems (St. Louis, MO) after screening by PCR using primers from the gene’s fourth exon. DNA was isolated from the P1 clone for gene targeting that included a 5′ BsmI fragment of 1802 bp, containing exon 1 and 22 bp of exon 2, and a 3′ EcoRI/BsmI fragment of 2752 bp containing exons 3 and 4 (see Fig. 3⇓).

Adapters were used to ligate the 5′ BsmI fragment and the 3′ EcoRI/BsmI fragment into the unique XhoI and EcoRI sites, respectively, of pPNT (35), which contains the neomycin resistance gene (neo) and the hsv-thymidine kinase gene, each driven by the phosphoglycerate kinase (PGK) promoter. pPNT-mumig was linearized using the unique NotI site. Transfection of J1 ES cells (passage 12) and selection with G418 and ganciclovir were done as described (36). DNAs from 191 G418/ganciclovir-resistant colonies were digested with PstI and screened by Southern analysis using as probe a 450-bp fragment from the flanking region of the mig gene outside of the 5′ BsmI mig fragment used in pPNT-mumig (see Fig. 3⇓). DNAs from 12 ES cell colonies yielded a PstI fragment of ∼7 kb as predicted in the event of homologous recombination. These ES clones were analyzed with additional probes to ensure that the structure of the recombinant locus was as predicted. All but one of the ES clones were shown to have a single copy of the inserted neo gene.

Producing mig^−/− mice and breeding mice for experiments

Cells from two ES clones (numbers 17 and 63) gave chimeric progeny (as judged by coat color) after injection into C57BL/6 blastocysts as described (36), and offspring of matings between C57BL/6 females and male chimeras (lines 17 and 63) or 129/Sv females and male chimeras (line 63) showed transmission of the recombined locus as judged by Southern blotting of PstI-digested DNA as described above and/or by a PCR-based assay, the details of which are available on request.

Offspring of chimeras derived from ES clone 17 were intercrossed to produce mig^−/− and mig^+/+ (129/Sv × C57BL/6)F₂ mice. For some experiments, the mig^−/− and mig^+/+ (129/Sv × C57BL/6)F₂ mice were offspring of mig^−/− × mig^−/− or mig^+/+ × mig^+/+ matings. In these cases, all the mig homozygous breeders were offspring from (129/Sv × C57BL/6)F₁ mice, making them genetically equivalent to siblings. For some experiments, the mig^−/− and mig^+/+ (129/Sv × C57BL/6)F₂ mice were offspring of mig^+/− × mig^+/− matings. In these cases, the mig^+/− breeders were offspring from (129/Sv × C57BL/6)F₁ mice, making them genetically equivalent to siblings, or the direct offspring of these mice, and the mig^−/− and mig^+/+ mice used in experiments shared parents insofar as was practical. In later experiments, mice derived after backcrossing six generations into C57BL/6 mice were intercrossed to generate mig^−/− mice, which were used in experiments with control C57BL/6 mice from the Division of Cancer Therapy of the National Cancer Institute. Some experiments (data not shown) used mice derived from ES clone 63 that were crossed with 129/Sv mice before intercrossing to generate mig^−/− mice (from mig^−/− × mig^−/− matings) that were used with wild-type, control 129/Sv mice bred in our facility.

Analyzing mig^−/− and mig^+/+ mice for Mig protein

Analysis of production of MuMig protein in the liver by Western blot was done as described (15). MuMig protein was detected using 10 μg/ml rabbit anti-MuMig Ab 5170 that had been raised against the full-length, rMuMig (purified in our laboratory from supernatants from BTI-TN-5B1-4 Trichoplusia ni cells infected with recombinant baculovirus) and affinity purified using rMuMig (PeproTech) that had been coupled to Affigel 10 (Bio-Rad) according to the manufacturer’s protocol. Bound Ab was detected on the blot using 0.16 μg/ml HRP-conjugated goat anti-rabbit IgG (Santa Cruz Biotechnology, Santa Cruz, CA) and SuperSignal chemiluminescent substrate (Pierce) according to the manufacturer’s protocol.

Infections and Ab responses with Francisella tularensis LVS in mig^−/− and mig^+/+ mice

After obtaining preimmunization serum samples, male mig^−/− and mig^+/+ mice were infected intradermally (i.d.) at the base of the tail with 10⁴–10⁷ CFU F. tularensis live vaccine strain (LVS). Sera were obtained from weekly bleeds, stored at −20°C, and batch run for anti-F. tularensis LVS titers as described previously (37), using an ELISA with isotype-specific detection reagents from Southern Biotechnology Associates (Birmingham, AL) and plates coated with live F. tularensis LVS bacteria. End point titer was defined as the lowest dilution of serum that gave an OD₄₀₅ value >0.050 OD units and that was also 2 SDs of the mean greater than the OD value of the matched dilution of normal, pre-bleed mouse serum.

Thymus-dependent and thymus-independent Ab responses in mig^−/− and mig^+/+ mice

For determining thymus-dependent Ab responses, mig^−/− and mig^+/+ (129/Sv × C57BL/6)F₂ littermates were injected i.p. with 200 μg of (2, 4, 6)trinitrophenyl (TNP)-keyhole limpet hemocyanin (KLH) in solution without adjuvant (Biosearch Technologies, Novato, CA) and serum was collected at weekly intervals through wk 4. Ab responses were measured using ELISA with TNP-BSA as described (37). For determining thymus-independent Ab responses, mig^−/− and mig^+/+ (129/Sv × C57BL/6)F₂ littermates were injected with either 1 μg of Escherichia coli LPS serotype 055:B5 (Sigma-Aldrich, St. Louis, MO) i.p. or 0.5 μg of type III pneumococcal polysaccharide (SSS-III) s.c. (gift from P. J. Baker, National Institute of Allergy and Infectious Diseases, National Institutes of Health). Ab titers were measured using Immulon-1 plates coated with 10 μg/ml E. coli LPS in PBS (pH 7.2) or 1 μg/ml SSS-III in 0.1 M HEPES (pH 3.5).

Statistics

Differences in chemotaxis were determined using two-tailed t testing. Differences in Ab titers were determined using the two-tailed t tests on the geometric means of the reciprocals of the endpoint titers. The criterion for statistical significance was p < 0.05.

rMuMIG is a chemotactic factor for mouse T cells

To characterize the function of Mig in mice and to understand its roles in mouse models of disease, we tested its activity as a chemotactic factor for mouse leukocytes. rMuMig had no chemotactic activity for polymorphonuclear leukocytes or for macrophages isolated from thioglycolate-induced peritoneal exudates, although these cells responded well to fMLP or recombinant human macrophage-inflammatory protein (MIP)-1α (CCL3), respectively; these cells also failed to respond to another murine CXCR3 ligand, recombinant cytokine-responsive gene (CRG)-2/IP-10 (data not shown). Responses of mouse CD4⁺ and CD8⁺ T cells to rMuMig were determined using splenocytes that either were freshly isolated or had been activated in vitro for 3 or 4 days using plate-bound anti-CD3. For the freshly isolated cells, we also stained for CD44 and CD62L to quantify naive and memory subsets. As shown in Fig. 1⇓A, both CD4⁺ and CD8⁺ T cells, either freshly isolated or after activation in vitro, showed specific migration to rMuMig. Analysis of the freshly isolated cells that migrated to rMuMig using stains for CD44 and CD62L revealed that responses were found only in the memory, and not in the naive, subsets (Fig. 1⇓B). As shown in Fig. 1⇓A, the activated cells usually demonstrated more rMuMig-induced migration as compared with the freshly isolated cells, although the differences were not dramatic.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

rMuMig is a chemotactic factor for mouse T cells. A, Splenocytes from C57BL/6 mice were used for chemotaxis assays either immediately after isolation (Fresh) or after activation with anti-CD3 for 4 days (Activated). Chemotaxis was measured using Transwell plates and 10⁶ cells/well. After 3 h at 37°C, cells from multiple bottom wells were pooled, counted, and stained for expression of CD4 and CD8. Data are expressed as the percentages of input CD4⁺ or CD8⁺ T cells migrating to the lower wells containing 1000 ng/ml rMuMig minus the percentages of cells migrating to lower wells containing medium alone. Data represent means and SEM of two experiments. Differences among these populations of cells were not statistically significant. B, Cells from the starting population of freshly isolated splenocytes or cells migrating to the lower chambers in the absence or presence of 1000 ng/ml rMuMig were stained for CD4 (Cy5-PE), CD44 (PE), and CD62L (FITC) or CD8 (FITC) and CD44 (Cy5-PE). Percentages of cells are indicated within the designated regions. Three additional experiments also showed that only memory cells migrated to rMuMig.

CXCR3 is expressed on mouse T cells

Because mouse T cells responded to rMuMig we sought to analyze the expression of the mouse analog of human CXCR3, the only known human receptor for Mig, on these cells. Antiserum was raised against a carrier-conjugated peptide corresponding to aa 2–18 of the MuCXCR3 sequence. Anti-MuCXCR3 Abs were affinity-purified using the immunizing peptide and the Abs were shown to stain MuCXCR3 specifically by using transfected cells and by blocking staining with the immunizing peptide (see Materials and Methods; data not shown). Using these Abs, we found that a subset of total CD8⁺ T cells showed staining that was blocked using the MuCXCR3 peptide (Fig. 2⇓A). For freshly isolated total CD4⁺ T cells there was also a suggestion of a MuCXCR3⁺subset, but this was less clear. On 4-day activated T cells, specific staining for MuCXCR3 could be seen on both CD4⁺ and CD8⁺ T cells. For all cells, there was no staining seen with nonimmune rabbit IgG that could be blocked with the MuCXCR3 peptide (data not shown). We analyzed further the freshly isolated cells using CD44 and found that MuCXCR3 expression was limited to the CD44^bright, i.e., memory subsets (Fig. 2⇓B), of both CD4⁺ and CD8⁺ cells, although there were higher numbers of the latter as compared with the former. The restricted expression of MuCXCR3 on memory cells is consistent with the results of the chemotaxis assays. There was a suggestion (Fig. 2⇓A) that, while activation of CD8⁺ T cells led to up-regulation of MuCXCR3 on the majority of cells, the level of MuCXCR3 might in fact have been reduced on the cells that were initially MuCXCR3⁺. Freshly isolated MuCXCR3⁺CD8⁺ and MuCXCR3⁻CD8⁺ T cells were purified by cell sorting, activated in vitro, and stained for MuCXCR3. As shown in Fig. 2⇓C, MuCXCR3 was induced on the population that was initially negative, whereas levels fell, but remained positive, on the (memory) CD8⁺ T cells that were initially MuCXCR3⁺.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Mouse T cells express CXCR3. A, Expression on fresh and activated CD4⁺ and CD8⁺ T cells. Splenocytes from C57BL/6 mice, either freshly isolated or after activation with anti-CD3 for 4 days, were stained for CD4 (allophycocyanin) and CD8 (FITC) and with rabbit anti-MuCXCR3 that had been preincubated with or without the MuCXCR3 immunizing (blocking) peptide. Cell-bound rabbit IgG was detected using biotinylated goat anti-rabbit Ab and streptavidin-PE. Fresh and activated CD4⁺ and CD8⁺ T cells are displayed separately as indicated. Solid lines represent staining without blocking peptide; dashed lines represent staining with blocking peptide. No peptide-inhibitable staining was seen using control rabbit IgG (data not shown). B, Expression on CD4⁺ and CD8⁺ T cell subsets. Freshly isolated splenocytes were stained for CD4 (allophycocyanin), CXCR3 (PE), and CD44 (FITC) or CD8 (allophycocyanin), CXCR3 (PE), and CD44 (FITC). Horizontal and vertical lines were drawn based on staining with an isotype control for CD44 (rat IgG2b conjugated to FITC) and with anti-MuCXCR3 in the presence of blocking peptide, respectively. The numbers in the quadrants represent the percentage of CD4⁺ or CD8⁺ T cells within each population. C, Effect of activation on expression on CD8⁺ T cell subsets. CD8⁺ T cells were sorted into CD8⁺CXCR3⁻ and CD8⁺CXCR3⁺ populations before activation with anti-CD3 and anti-CD28 for 4 days. CXCR3 (PE) staining of freshly isolated and activated sorted cells are shown using dashed lines and solid lines, respectively. Blocking peptide eliminated CXCR3 staining (data not shown). Data for each part are from one of at least two experiments that gave similar results.

Production of mig^−/− mice

To analyze the roles for MuMig in vivo, we produced mice with targeted disruption and partial deletion of the mig gene as shown in Fig. 3⇓. Homologous recombination between the targeting DNA and the mumig gene was predicted to result in a deletion of 1458 bp of the mumig gene, including 104 bp of exon 2 and 1354 bp of the second intron, and insertion of the PGK/neo gene cassette containing multiple PstI sites. An mRNA transcribed from the deleted/interrupted gene was predicted to encode the MuMig signal peptide followed by seven amino-terminal amino acids before reaching a stop codon. Moreover, the second exon sequence deleted in the recombinant allele contains three of the four cysteines conserved among the CXC and CC chemokines. Fig. 3⇓ shows data from mice derived from two successfully targeted ES cells, clones 17 and 63. In addition to the results shown in Fig. 3⇓, analysis of targeted DNAs revealed that the structure of the targeted allele was as predicted on both the 5′ and 3′ sides of the inserted DNA. Southern blot analysis using a probe for crg-2/IP-10, a gene within 100 kb of the mumig gene (H.-H. Lee and J. M. Farber, unpublished observations), showed the single wild-type band of >12 kb, suggesting that this neighboring locus was intact in the recombinants (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Disruption/deletion of the mig gene in mice. A, Targeting strategy. Schematic representations of the mumig gene, the pPNT-mumig DNA used for gene targeting, and the predicted disrupted/deleted gene are shown in the upper, middle, and lower lines, respectively. The mumig gene is on a PstI fragment of >12 kb. The gene’s four exons are shown as filled bars. The 5′ 1.8-kb and 3′ 2.8-kb restriction fragments inserted into the targeting DNA are indicated, as are the 1.4-kb fragment containing most of the second exon and intron that are predicted to be deleted in the targeted gene, and the probe that is 5′ to the targeting fragments and that was used to screen for homologous recombination. PstI digestion of the targeted allele and hybridization to the 5′ probe was predicted to yield a band of ∼6.9 kb. As shown on the bottom line, homologous recombination was expected to result in deletion of 1458 bp, including 104 bp of exon 2 and 1354 bp of the second intron, and insertion of the PGK/neo gene at this site. B, Genotypes of ES cell clone 17 and a single litter of mice from crossing (129/Sv × C57BL/6)F₁ parents heterozygous for the disrupted/deleted mig allele. DNAs from ES cells and from tail snippings of a litter of mice were digested with PstI and hybridized with radiolabeled 5′ probe as indicated in A. The recombined mig allele yielded the predicted fragment of ∼7 kb as indicated. The mig genotypes are indicated above the lanes: +/−, heterozygous; +/+, homozygous wild-type; −/−, homozygous knockout. C, mig^−/− mice do not express Mig mRNA in spleen after injection with rMuIFN-γ. mig^+/+, mig^+/−, and mig^−/− mice from line 63 were injected with 25,000 U rMuIFN-γ at 0 and 12 h and spleens were harvested at 24 h for preparation of RNA. Twenty micrograms of total RNA was loaded and hybridized to a radiolabeled probe containing sequences from the second exon of the mumig gene. The major species of MuMig mRNAs have mobilities of ∼1.6 and 3 kb. Genotypes are indicated above the lanes. A photograph of the gel stained with ethidium bromide before transfer shows the ribosomal bands, demonstrating equal loading of total RNA per lane. An experiment using mice derived from ES cell line 17 gave similar results. D, mig^−/− mice do not express Mig protein. mig^−/− and mig^+/+ (129/Sv × C57BL/6)F₂ littermates from line 17 were injected with rMuIFN-γ or with carrier alone at 0 and 12 h before harvesting organs at 24 h. Livers were homogenized and microsomal pellets were prepared to enrich for secreted proteins. Microsomal samples containing 100 μg of protein were separated in a 15% acrylamide gel and electrotransferred to nitrocellulose. Each lane contained equal amounts of protein as visualized by staining with amido black (data not shown). The blot was probed with affinity-purified anti-MuMig rabbit Ig and rabbit Ab was detected using HRP-conjugated goat anti-rabbit IgG and a chemiluminescent substrate. Genotypes, the fractions analyzed, and injection with rMuIFN-γ (+) or carrier alone (−) are indicated above the lanes. MuMig runs as expected as multiple IFN-γ-inducible species below the 21-kDa marker. Two additional experiments analyzing mice derived from ES cell clone 17 and one experiment analyzing mice derived from ES cell clone 63 gave similar results.

The mig^−/− and mig^+/+ (129/Sv × C57BL/6)F₂ mice were analyzed for the expression of Mig mRNA and protein in tissues taken from mice that had been injected with rMuIFN-γ or from control-injected mice. As shown in Fig. 3⇑C, Northern blot analysis using a probe prepared from the second exon of the mumig gene revealed induction in spleens of mig^+/+ mice injected with rMuIFN-γ but no signal in the mig^−/− mice. Western blot analysis of liver microsomes revealed that no Mig protein was produced in response to rMuIFN-γ in the mig^−/− mice (Fig. 3⇑D). For the experiments shown below, mig^−/− mice were used in both the (129/Sv × C57BL/6)F₂ background and after backcrossing into C57BL/6 for six generations.

mig^−/− mice have an abnormality in Ab production

The mig^−/− mice were born from mig^−/+ parents with expected frequencies and had no obvious abnormalities. Microscopic examination of H&E-stained tissues, including among them lymph nodes, spleen, and Peyer’s patches, showed no differences in histology between mig^−/− and mig^+/+ mice. Likewise, the mig^−/− and mig^+/+ mice showed no differences in peripheral leukocyte and differential counts or in constituent populations of cells in thymus (CD4⁻CD8⁻, CD4⁺CD8⁺, and single-positive thymocytes), spleen (CD4⁺ and CD8⁺ T cells, B cells, macrophages, and NK cells), or peritoneum (B-1 B cells) as determined by flow cytometry (data not shown). Because the expression of mumig in vivo is highly dependent on IFN-γ, we tested the mig^−/− mice in models of host defense where IFN-γ plays a critical role. We found no significant differences in mortality between mig^−/− mice and mig^+/+ mice infected with the following pathogens: Listeria monocytogenes (EGD strain, acute primary infection or after lethal challenge of surviving mice); Salmonella typhimurium (primary infection); F. tularensis LVS (acute primary infection or after lethal challenge of mice that survived primary sublethal infections); Mycobacterium avium (primary infection); Toxoplasma gondii (acute primary infection with ME-49 strain or immunization with strain TS-4 P1 followed by challenge with strain R.H.); vaccinia virus (primary infection); ectromelia virus (primary infection); hsv type 1 (immunization with KOS strain followed by challenge with McKrae strain); Histoplasma capsulatum (primary infection); and Cryptococcus neoformans (primary infection) (data not shown). Some of these models were analyzed in more detail. For example, we performed histological examination of liver inflammation after infection with L. monocytogenes or T. gondii; counted colonies of L. monocytogenes from spleens of infected mice; counted brain cysts in mice infected with T. gondii; analyzed peritoneal exudates after infection with T. gondii or splenocytes during the course of infection with M. avium by flow cytometry; and measured footpad swelling after inoculation with ectromelia virus. These analyses also failed to reveal significant and reproducible differences between the mig^−/− and mig^+/+ mice (data not shown).

However, during analysis of responses to primary infections with the Gram-negative intracellular bacterium F. tularensis LVS we found that the mig^−/− mice produced lower titers of Abs against the bacterium as compared with the mig^+/+ mice. The titers of anti-Francisella total IgG and the major isotype produced against the organism, IgG_2a/c (37), were reduced in the mig^−/− mice (Fig. 4⇓). Diminished anti-F. tularensis LVS total IgG and/or IgG_2a/c titers were found in the mig^−/− vs mig^+/+ mice in each of six experiments using (129/Sv × C57BL/6)F₂ or backcrossed mice using a range of sublethal doses of organisms (see Fig. 4⇓). In two experiments in the 129/Sv background done with different doses of organisms, mean titers for IgG2a were also lower in the mig^−/− mice, but the differences were not statistically significant (data not shown). No reproducible differences were found for anti-Francisella LVS titers of IgM and of IgG1, IgG2b, and IgG3 for the mig^−/− vs the mig^+/+ mice (data not shown). Bacterial clearance was comparable in the mig^−/− and mig^+/+ mice in that no live organisms could be recovered from the spleens of either mig^−/− or mig^+/+ mice 3 wk after infection. In addition, at 3 wk after infection, few small germinal centers could be seen in the spleens of both mig^−/− and mig^+/+ mice by staining with peanut agglutinin (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

mig^−/− mice produce lower titers of Ab against F. tularensis LVS. Two representative experiments are shown. Strain genotypes are indicated above the plots and mig genotypes are indicated below. Each circle represents an individual mouse; • represents mig^+/+ mice and ○ represents mig^−/− mice. Horizontal bars represent antilogarithms of geometric means. A, Serum total IgG against F. tularensis LVS was measured by ELISA 6 wk after i.d. infection of (129/Sv × C57BL/6)F₂ mice with 10⁷ organisms. The range of titers was 1/3,200–1/12,800 for the mig^−/− mice and 1/6,400–1/51,200 for the mig^+/+ mice, and the mean endpoint titers were 1/4,935 for mig^−/− mice and 1/15,596 for mig^+/+ mice, which are significantly different (p < 0.04). Two additional experiments using pooled sera also showed diminished anti-F. tularensis LVS IgG in the mig^−/− mice. B, Serum IgG_2a/c against F. tularensis LVS was measured by ELISA 3 wk after i.d. infection with 10⁴ organisms of mig^−/− mice derived after backcrossing for six generations into the C57BL/6 strain vs C57BL/6 wild-type mice. The range of titers was 1/40–1/320 for the mig^−/− mice and 1/80–1/5120 for the mig^+/+ mice, and the mean endpoint titers were 1/128 for mig^−/− mice and 1/436 for mig^+/+ mice, which are significantly different (p < 0.03). A total of five experiments using (129/Sv × C57BL/6)F₂ or backcrossed mice showed diminished anti-F. tularensis LVS IgG_2a/c titers in the mig^−/− vs mig^+/+ mice.

To determine whether the mig^−/− mice responded abnormally to Ags not introduced by live infection, we analyzed responses of mig^−/− and mig^+/+ mice to immunizations with the T cell-dependent Ag TNP-KLH and with the T cell-independent Ags LPS and SSS-III. No reproducible differences were seen in titers of IgG isotypes against TNP-KLH or in titers of total IgG against LPS (data not shown). Although in a limited number of experiments we found lower titers of IgM against TNP-KLH (by approximately one-half) at wk 1 in the mig^−/− vs mig^+/+ mice, the difference was gone by wk 2, and no differences were seen in the titers of IgM against LPS or SSS-III (data not shown). Similarly, no differences were seen in pathogen-specific IgG_2a/c titers between the two groups of mice measured during the first month of infection with L. monocytogenes or T. gondii (data not shown).

Activated B cells respond to rMuMIG and express CXCR3

Although the diminished anti-Francisella LVS titers in the mig^−/− mice could reflect the effects of MuMig on T cells or other non-B cells, we investigated the possibility that MuMig might target B cells directly. Although rMuMig had no chemotactic activity on freshly isolated mouse splenic B cells, B cells activated through Ag receptor and CD40 plus IL-4 showed dose-dependent migration to rMuMig (Fig. 5⇓A). Dose-dependent responses to rMuMig of activated, but not resting, B cells were also seen using a flow cytometry-based calcium flux assay as shown in Fig. 5⇓B. A time course revealed that acquiring responses to rMuMig required at least 4 days of activation of the B cells in vitro (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Activated B cells respond to CXCR3 ligands. A, Activated B cells migrate to MuMig. Single-cell suspensions from spleens of TCRβ⁻δ⁻ C57BL/6 mice were activated for 5 days with anti-IgM, IL-4, and CD40L. Chemotaxis was measured using a 96-well chemotaxis chamber, 10⁵ cells/well, and concentrations of rMuMig in the lower wells as shown. After 3 h at 37°C, cells from multiple bottom wells were pooled, counted, stained for B220, and compared with the starting population to determine the percentages of B cells migrating. Freshly isolated B cells did not migrate toward rMuMig and no net migration of activated B cells was seen when 1000 ng/ml rMuMig was put in both upper and lower wells (data not shown). A second experiment gave similar results. B, Activated B cells flux calcium in response to CXCR3 ligands. Single-cell suspensions from spleens of TCRβ⁻δ⁻ C57BL/6 mice were activated for 5 days with anti-IgM, IL-4, and CD40L before being loaded with indo-1, stained for B220, and used for calcium flux assays as described (8 ). Ba, In these scattergrams, the y-axes show channel numbers that reflect the fluorescence ratio of emissions at 390/530 nm of the calcium probe indo-1. For each assay, buffer was added at 30 s and rMuMig or rHuI-TAC was added as indicated at 60 s to a final concentration of 2000 ng/ml. Data are shown only for B220⁺ cells. Freshly isolated B cells did not show a calcium flux in response to these ligands (data not shown). Data are from one of three experiments that gave similar results. Bb, Experiments were done as in A. A threshold value of ratio fluorescence was determined that separated the lower 95% from the highest 5% of cells following injection of buffer alone. The percentage of B cells rising above the threshold was determined at each concentration of rMuMig as described in Materials and Methods and in Ref. 8 . The percentage value with buffer alone (5%) was subtracted from the values following rMuMig to obtain the percentage of responding cells.

We presumed that mouse B cell responses to rMuMig were mediated through CXCR3. In support of this presumption, rHuI-TAC (Fig. 5⇑B) and rCRG-2/IP-10 (data not shown) also produced calcium fluxes in the activated B220⁺ cells. Consistent with the data on chemokine-induced signals, activated but not resting B cells showed specific surface staining for CXCR3 that was blocked using the immunizing peptide (Fig. 6⇓A) and Northern blotting revealed induction of CXCR3 mRNA after B cell activation (Fig. 6⇓B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Activated mouse B cells express CXCR3. A, Activation induces surface expression of MuCXCR3. Cells purified from spleens of TCRβ⁻δ⁻ C57BL/6 mice were stained for B220, and with rabbit anti-MuCXCR3 that had been preincubated with or without the MuCXCR3 immunizing peptide. Cell-bound rabbit IgG was detected using biotinylated goat anti-rabbit Ab and streptavidin-PE. Solid lines represent staining without blocking peptide; dashed lines represent staining with blocking peptide. Data are shown only for B220⁺ cells. Left panel, Freshly isolated B cells. Right panel, B cells after 5 days of activation with anti-IgM, rIL-4, and CD40L. No peptide-inhibitable staining was seen using control rabbit IgG (data not shown). Data are from one of two experiments that gave similar results. B, Activation induces the Mu CXCR3 gene. Total RNA was prepared from splenic B cells from TCRβ⁻δ⁻ C57BL/6 mice, either immediately after isolation or after 6 days of activation with anti-IgM, rIL-4, and CD40L. Twenty micrograms of total RNA from resting (R) or activated (A) B cells was loaded per lane, fractionated in a 1.2% agarose/formaldehyde gel, transferred to reinforced nitrocellulose, and hybridized to a radiolabeled MuCXCR3 cDNA probe. The image was obtained using a PhosphorImager (Molecular Devices, Sunnyvale, CA). The MuCXCR3 mRNA ran just below the 18S ribosomal RNA. A radiolabeled oligonucleotide probe for the 18S ribosomal RNA was hybridized to the blot to demonstrate equal loading as shown. Data are from one of two experiments that gave similar results.

Splenic dendritic cells express high levels of MuMig mRNA

The data on B cells, together with the other data presented above, suggested that MuMig and CXCR3 might be involved not only in mediating trafficking to sites of peripheral inflammation but also in optimizing cell recruitment or cell-cell interactions within lymphoid organs. Therefore, we analyzed the expression of MuMig mRNA by semiquantitative RT-PCR before and after treatment with rMuIFN-γ in subpopulations of mouse splenocytes sorted by flow cytometry. For comparison, we used RNA prepared from the RAW 264.7 mouse macrophage cell line where the mRNA for MuMig is known to be highly induced (38). As shown in Fig. 7⇓, significant levels of MuMig mRNA could be detected in freshly isolated CD8α⁻ and CD8α⁺ splenic dendritic cells, and these levels could be dramatically enhanced by treatment with rMuIFN-γ in vitro. Some expression and induction by rMuIFN-γ was found in B cells as well, with no effect after short-term activation through surface IgM. T cells expressed the lowest levels of MuMig mRNA, with some induction by rMuIFN-γ or by anti-CD3 plus anti-CD28. This latter induction was dependent on IFN-γ because it was not seen using T cells isolated from IFN-γ knockout mice (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

Expression of the mumig in splenic dendritic cells and lymphocytes. Semiquantitative RT-PCR analysis was performed as described (60 ) on total RNA from cells purified using a FACS and incubated for 18 h with (+) or without (−) 1000 U/ml rMuIFN-γ. CD8α⁺ and CD8α⁻ splenic dendritic cells were isolated and analyzed separately as noted. B cells were also incubated with (+) or without (−) 10 μg/ml anti-IgM and T cells with (+) or without (−) 10 μg/ml plate-bound anti-CD3 and 10 μg/ml soluble anti-CD28. Analysis was also done using RNA from the RAW 264.7 macrophage cell line incubated with (+) or without (−) 1000 U/ml rMuIFN-γ for 6 h to compare the signals in the primary cells with those in a cell line where the mumig gene is known to be highly inducible (38 ). Expression of mumig was normalized to the signals from hypoxanthine phosphoribosyltransferase and these values were then normalized to the lowest value obtained, which was from CD4⁺ T cells isolated from IFN-γ^−/− mice (data not shown), and which was set equal to 1. Data are from one of two experiments that gave similar results.