Phenotype of Mice and Macrophages Deficient in Both Phagocyte Oxidase and Inducible Nitric Oxide Synthase

Immunity, Vol. 10, 29–38, January, 1999, Copyright 1999 by Cell Press

Phenotype of Mice and Macrophages Deficient in Both Phagocyte Oxidase and Inducible Nitric Oxide Synthase Michael U. Shiloh,1 John D. MacMicking,2,7 Susan Nicholson,2 Juliet E. Brause,2 Strite Potter,1 Michael Marino,3 Ferric Fang,4 Mary Dinauer,5 and Carl Nathan1,2,6 1 Department of Microbiology and Immunology 2 Department of Medicine Weill Medical College of Cornell University New York, New York 10021 3 Ludwig Institute for Cancer Research at Memorial-Sloan Kettering Cancer Center New York, New York 10021 4 Division of Infectious Diseases University of Colorado Health Sciences Center Denver, Colorado 80262 5 Department of Pediatrics Indiana University Cancer Center Indianapolis, Indiana 46202

Summary The two genetically established antimicrobial mechanisms of macrophages are production of reactive oxygen intermediates by phagocyte oxidase (phox) and reactive nitrogen intermediates by inducible nitric oxide synthase (NOS2). Mice doubly deficient in both enzymes (gp91phox2/2/NOS22/2) formed massive abscesses containing commensal organisms, mostly enteric bacteria, even when reared under specific pathogen-free conditions with antibiotics. Neither parental strain showed such infections. Thus, phox and NOS2 appear to compensate for each other’s deficiency in providing resistance to indigenous bacteria, and no other pathway does so fully. Macrophages from gp91phox2/2/ NOS22/2mice could not kill virulent Listeria. Their killing of S. typhimurium, E. coli, and attenuated Listeria was markedly diminished but demonstrable, establishing the existence of a mechanism of macrophage antibacterial activity independent of phox and NOS2. Introduction Two antimicrobial mechanisms of tissue macrophages are production of reactive oxygen intermediates (ROI) by the phagocyte oxidase (phox) and reactive nitrogen intermediates (RNI) by inducible nitric oxide synthase (iNOS; NOS2). Although many other mechanisms have been postulated, none has been both specifically identified and genetically confirmed (Shiloh and Nathan, 1999). The importance of phox is underscored by chronic granulomatous disease (CGD), a genetic deficiency of phox components marked by recurrent infections. However, the pathogens responsible for recurrent infections 6

To whom correspondence should be addressed (e-mail: cnathan@ mail.med.cornell.edu). 7 Present address: Laboratory of Immunology, Howard Hughes Medical Institute, Rockefeller University, New York, New York 10021.

in CGD are limited in variety (Mouy et al., 1989) and monocytes from CGD patients kill some intracellular pathogens (reviewed in Nathan, 1983). Mouse knockouts of either gp91phox (Pollock et al., 1995) or gp47phox (Jackson et al., 1995) recapitulate the CGD phenotype. NOS2 knockout mice have demonstrated the enzyme’s essential contribution to host defense against a restricted set of pathogens (reviewed in Nathan, 1997), including Mycobacterium tuberculosis (MacMicking et al., 1997), Leishmania major (Wei et al., 1995; Diefenbach et al., 1998), and ectromelia virus (Karupiah et al., 1998), with a less marked phenotype upon infection with Listeria monocytogenes (MacMicking et al., 1995; North et al., 1997) and Toxoplasma gondii (Khan et al., 1997; Scharton-Kersten et al., 1997). Recent experiments with additional knockout mice have lent support to the presumption that macrophage antimicrobial mechanisms must exist other than production of ROI or RNI. Susceptibility to L. monocytogenes in vivo despite a normal capacity of macrophages to produce ROI and RNI in vitro was observed in mice deficient in IL-6 (Kopf et al., 1994), NF-IL6 (Tanaka et al., 1995), ICSBP (Fehr et al., 1997), IRF-2 (Fehr et al., 1997), RelB (Weih et al., 1997), and IFNg receptor (Huang et al., 1993; Dai et al., 1997; Fehr et al., 1997). However, disruption of these genes is expected to have pleiotropic effects. Only two of these studies demonstrated a macrophage bactericidal defect (Tanaka et al., 1995; Fehr et al., 1997), and in those, macrophage production of ROI and RNI was not tested with the infection itself as the stimulus for macrophage activation and the infecting organism as the trigger for the respiratory burst. Thus, it remains a major question in the study of innate immunity whether primary macrophages have recourse to a mechanism for killing bacteria other than production of ROI and RNI. To answer this question, we generated mice doubly deficient in gp91phox and NOS2. Their phenotype was distinct from that of either parent or any mouse reported. Their marked susceptibility to spontaneous internal infections arising from indigenous flora was especially surprising in view of the antibacterial mechanisms considered paramount in the control of commensal organisms, such as antibody, complement, and the antibacterial proteins of neutrophils. Nonetheless, their macrophages permitted the formal demonstration that a phox- and NOS2-independent antibacterial mechanism exists. Results Generation of Gp91phox2/2/NOS22/2 Mice Gp91phox2/2/NOS22/2 mice revealed only mutant alleles of NOS2 (Figure 1A) and gp91phox (Figure 1B). Activated macrophages from wild-type and gp91phox2/2 mice produced large amounts of RNI, while those from NOS22/2 and gp91phox2/2/NOS22/2 mice did not (Figure 1C). Activated macrophages from wild-type and NOS22/2 mice produced large amounts of H2O2, while gp91phox2/2 and gp91phox2/2/NOS22/2 macrophages did not (Figure 1D). The

Immunity 30

Figure 1. Generation of Gp91phox2/2/NOS22/2 Double Knockout Mice PCR genotyping of (A) NOS2 and (B) gp91phox alleles. Expected amplificands for wild-type C57BL/6 3 129Sv and knockout alleles are 413 and 268 bp for NOS2 and 240 and 195 bp for gp91phox. (C) Thioglycollate broth-elicited peritoneal macrophages were exposed for 24 hr to media alone, LPS, rMuIFNg, or both LPS and rMuIFNg and nitrite accumulation determined by the Griess assay. Values are means for triplicates 6 SEM. (D) H2O2 release from periodate-elicited peritoneal macrophages exposed for 24 hr to medium alone or rMuIFNg and then triggered with phorbol myristate acetate for 2 hr. Values are means for triplicates 6 SEM.

activation pathway was intact in gp91phox2/2/NOS22/2 macrophages, because IFNg triggered the appearance in their nuclei of proteins that bound to an IFNg response region (GRR) oligonucleotide (data not shown). Susceptibility of Gp91phox2/2/NOS22/2 Mice to Spontaneous Infections Gp91phox2/2/NOS22/2 mice were housed in autoclaved microisolator cages with filtered air under positive pressure and given sterile bedding, food, and water by workers in sterile garb. They survived weaning at Mendelian ratios and generated litters of normal size. However, early in the course of the colony, mice were found to harbor large intra-abdominal and/or intra-thoracic abscesses or were lost to pulmonary aspergillosis. We instituted prophylaxis with trimethoprim-sulfamethoxazole and itraconazole. Necropsies continued to reveal large abscesses involving the liver, spleen, intestine, diaphragm, and/or lungs (Figure 2A), with a prevalence

Figure 2. Spontaneous Infections in Gp91phox2/2/NOS22/2Mice (A) Abscess formation in the abdominal organs of a gp91phox2/2/ NOS22/2mouse. Arrows indicate abscesses. Histology of hepatic abscesses at (B) 43, (C) 1003, and (D) 4003.

ranging from 14.9% (24/161 animals randomly culled for necropsy over the course of a year) to 24.6% (20/81 mice randomly culled for necropsy on 1 day). Prevalence of infections appeared to increase with age. Mice bearing huge abscesses showed no outward signs of clinical illness. When antibiotics were withdrawn, the colony was nearly lost to spontaneous infection. Prophylaxis was resumed with chloramphenicol, amoxicillin, and/or gentamicin depending on the results of cultures. From lung and liver abscesses, we isolated E. coli (15/20 mice), non-hemolytic streptococcus (3/20), Candida guillermondii (3/20), Pseudomonas species (1/20), and Enterococcus faecalis (1/20). Anaerobic organisms were not found. Histology revealed severe pyogranulomatous inflammation with encapsulated abscesses (Figure 2B) massively infiltrated (Figure 2C) by polymorphonuclear and some mononuclear leukocytes (Figure 2D). When we generated gp91phox2/2/NOS22/2 mice on a homogeneous C57BL/6 background, all progeny succumbed to spontaneous infection at z4 weeks of age despite prophylactic antibiotics. In striking contrast, during this 3 year study, no abscesses were detected in hundreds of gp91phox2/2, NOS22/2, or wild-type mice (C57BL/6 or C57BL/6 3 129Sv) housed in the same facility and not receiving antibiotics. In the experiments reported below, every mouse was necropsied after experimental infection or peritoneal lavage. No mice used for challenge studies had large abscesses, presumably because they were ,10 weeks old, and thus none were excluded. Several gp91phox2/2/ NOS22/2 mice used for harvest of macrophages had abscesses; their macrophages were discarded. Thus, it is unlikely that advanced, spontaneous infection influenced the data below, but some of the mice may have been subclinically infected. In Vivo Challenge with S. typhimurium S. typhimurium is a facultative intracellular mouse pathogen that replicates within macrophage phagosomes, where its fate depends on the state of macrophage activation. This phenomenon contributed to the definition of immunologically mediated macrophage activation (Mackaness, 1962; Blanden et al., 1966). Mutants of S. typhimurium unable to survive within macrophages are

Bactericidal Activity of Gp91phox2/2/NOS22/2 Mice 31

Figure 3. Survival of Mice Infected with S. typhimurium (A) Percent survival of C57BL/6 3 129Sv (n 5 9), C57BL/6 (n 5 10), NOS22/2 (n 5 7), gp91phoxdeficient (n 5 10), or gp91phox2/2/NOS22/2 (n 5 8) mice injected intraperitoneally with 200– 400 CFU of S. typhimurium 14028. Differences between C57BL/6 3 129Sv and C57BL/6 (p 5 0.001), C57BL/6 3 129Sv and NOS22/2 (p 5 0.0008), C57BL/6 3 129Sv and gp91phox2/2/ NOS22/2 (p , 0.0001), C57BL/6 and gp91phoxdeficient (p 5 0.0036), and C57BL/6 and gp91phox2/2/NOS22/2 (p 5 0.0003) were statistically significant (log-rank test). (B) Percent survival of C57BL/6 3 129Sv (n 5 phox phox2/2 2 /2 2 /2 (n 5 6), gp91 -deficient (n 5 10), or gp91 /NOS2 (n 5 10) mice injected intraperitoneally with 200–400 5), C57BL/6 (n 5 11), NOS2 CFU of S. typhimurium recBC. Differences between gp91 phox-deficient or gp91phox2/2/NOS22/2mice and their wild-type controls were statistically significant (p , 0.0001) and the difference between gp91phox-deficient and gp91phox2/2/NOS22/2mice (p 5 0.001) was statistically significant. Results are pooled from two experiments.

avirulent (Fields et al., 1986). Both NOS2 (Fang, 1997) and phox (De Groote et al., 1997) are important for the control of S. typhimurium. Accordingly, we began our challenge studies of gp91phox2/2/NOS22/2 mice with S. typhimurium. A point mutation within Nramp1 encoding Asp169 renders the C57BL/6 strain susceptible to S. typhimurium, while wild-type Gly169 confers resistance on the 129Sv strain and the heterozygotes (Malo et al., 1994; Vidal et al., 1995). Thus, we haplotyped the Nramp1 locus and chose only those mice homozygous for the Asp169 (susceptible) allele for in vivo studies. S. typhimurium 14028 is a virulent, smooth strain with an LD50 in susceptible mice of ,10 organisms (Buchmeier et al., 1993). Both the gp91phox-deficient and gp91phox2/2/NOS22/2 mice succumbed early to infection with 14028 (Figure 3A). NOS22/2 mice died less quickly, although they succumbed earlier and to a greater extent than their wild-type controls (C57BL/6 3 129Sv). The difference in survival between the wild-type C57BL/6 and wild-type C57BL/6 3 129Sv mice suggests that genetic loci in addition to Nramp1 may contribute to the resistance of 129Sv mice to salmonellosis. The enhanced susceptibility of NOS22/2, gp91phox2/2, and gp91phox2/2/NOS22/2 mice to infection with S. typhimurium 14028 suggests that both phox and NOS2 are necessary for control of virulent S. typhimurium infection in mice. The S. typhimurium recBC mutant in the 14028 background is incapacitated in its ability to repair DNA damage, attenuated in mice (LD50 . 104 organisms) and sensitive to killing by respiratory burst-proficient J774 cells (Buchmeier et al., 1993). Gp91phox2/2/NOS22/2 mice infected with the recBC mutant died as rapidly as those infected with the 14028 strain (Figure 3B). Gp91phox2/2 mice survived longer than gp91phox2/2/NOS22/2 mice when infected with the recBC strain (p 5 0.001), while NOS22/2 mice were fully resistant (Figure 3B). The restored virulence of the recBC strain in the gp91phox2/2/ NOS22/2 and gp91phox2/2 mice demonstrates that ROI and to a lesser extent RNI control S. typhimurium infection in part by damaging their DNA. In Vivo Challenge with L. monocytogenes Listeria monocytogenes was another of the organisms used by Mackaness to define immunologically mediated

macrophage activation (Mackaness, 1962). Listeria has become the most widely used pathogen in gene-targeted mice, in part because successful resolution of infection requires interplay between innate and acquired immunity (Kaufmann, 1993; Unanue, 1997) involving resident and activated macrophages (Mackaness, 1962; Portnoy et al., 1989), neutrophils (Conlan and North, 1991), NK cells (Dunn and North, 1991), and T cells (Harty et al., 1992; Kaufmann and Ladel, 1994; Harty and Bevan, 1995). We infected mice with 5 3 104 viable CFU of virulent L. monocytogenes (ATCC 10403) and evaluated the splenic bacterial load 3 days later. Relative to wild-type C57BL/6 3 129Sv mice, NOS22/2 mice had z102 times more CFU (Figure 4A), confirming results with another virulent strain, EGD serotype 3B (MacMicking, et al., 1995). In contrast, the majority of gp91phox2/2 mice had spleen CFU values that equaled the values of their wildtype controls (C57BL/6). Gp91phox2/2/NOS22/2 mice showed a possibly bimodal response: about half the mice were moribund and had 103- to 104-fold more CFU in their spleens than wild-type C57BL/6 3 129Sv mice, while the remainder looked well and restricted the number of bacteria as well as wild-type mice. In comparison, TNF2/2 mice, which were extremely susceptible to Listeria in an earlier study (Pasparakis et al., 1996), were moribund and had from 103- to 106-fold more Listeria in their spleens than their wild-type controls (C57BL/6 3 129Sv). In contrast to the results with virulent Listeria, within 2 days, gp91phox2/2/NOS22/2 mice completely eradicated splenic infection by all doses of an attenuated strain of Listeria (ATCC 15313) tested up to 108 (Figure 4B). Thus, despite the remarkable susceptibility of gp91phox2/2/NOS22/2 mice to spontaneous infection with E. coli and inoculation with S. typhimurium, gp91phox2/2/ NOS22/2 mice could dispose of attenuated Listeria, and about half of them could restrict virulent Listeria in the spleen as well as wild-type mice during an early stage of infection. Infection of Macrophages with E. coli The foregoing results suggested that gp91phox2/2/NOS22/2 macrophages might have residual antibacterial activity. To test this, periodate-elicited peritoneal macrophages

Immunity 32

Figure 5. Macrophage Bactericidal Activity against E. coli

Figure 4. Bacterial Counts after Infection with L. monocytogenes (A) Relative CFU for mice (each symbol represents one mouse) injected intraperitoneally with 5 3 104 viable L. monocytogenes (ATCC strain 10403). Results shown are the aggregate of individual mice tested in three separate experiments. Within each experiment, splenic Listeria counts (CFU) in C57BL/6 3 129Sv were averaged and given the value of 1. Listeria counts for mice in the remaining groups were divided by the mean CFU of C57BL/6 3 129Sv mice to yield relative CFU. Overall significance level p , 0.0001 determined by the Kruskal-Wallis test. †p , 0.005 compared to C57BL/6 3 129Sv (Mann-Whitney U test). (B) Bacterial load in gp91phox2/2/NOS22/2 infected with L. monocytogenes (ATCC strain 15313). Mice were infected as described above with increasing starting viable CFU.

were infected with E. coli 9243 (isolated from an abscess in a gp91phox2/2/NOS22/2 mouse) (Figure 5A) or E. coli HB101 (a laboratory strain) (Figure 5B) and the degree of bacterial killing assessed 4 hr after infection, based on time course studies (Figures 5A and 5B, insets). Killing of both E. coli strains by wild-type C57BL/6 3 129Sv macrophages was profound (2.0 log). Killing by NOS22/2 and TNF2/2 macrophages was diminished, although still .1.0 log. Gp91phox2/2 macrophages were further impaired (0.5 log killing compared to 1.2 log by wild-type C57BL/6). Gp91phox2/2/NOS22/2 macrophages also managed to reduce the number of viable E. coli by z0.5 log (60%–70%). Infection of Macrophages with S. typhimurium Periodate-elicited peritoneal macrophages were infected with either virulent S. typhimurium 14028 or the attenuated S. typhimurium recBC strain. By electron microscopy, all visualized bacteria were intracellular by the end of the uptake period (15 min) designated time 0 (data not

Periodate-elicited macrophages were infected with (A) E. coli 9243 (mouse isolate) or (B) E. coli HB101. In each experiment, macrophages were pooled from 3–5 mice. Bacterial counts were determined by fluorescent microplate assay. Log killing represents killing at 4 hr normalized to the number of bacteria at time 0. Results shown are the average of multiple independent experiments for each mouse strain 6 SEM. The averaged number of experiments are indicated within each bar. (Inset) Representative time course comparing the killing of E. coli HB101 by C57BL/6 3 129Sv (squares) or gp91phox2/2/NOS22/2 (circles) macrophages (log CFU/ml 6 SEM; some error bars fall within the symbols). * p # 0.05 compared to C57BL/6; †p # 0.05 compared to C57BL/6 3 129Sv (unpaired t test).

shown). Bacterial killing was assessed at 4 hr, based on time course studies (Figure 6A, inset). Killing of strain 14028 by wild-type and NOS22/2 macrophages averaged 1.0–1.5 log10. Killing by gp91phox2/2 and gp91phox2/2/ NOS22/2 macrophages was diminished to z0.5 log. Killing by TNF2/2 macrophages was also attenuated (to z0.4 log). Nevertheless, killing by macrophages of all mutant genotypes was demonstrable. With the recBC strain, killing by wild-type macrophages was profound (z2 log) (Figure 6B). NOS22/2 macrophages maintained substantial killing and TNF2/2 macrophages improved their killing to 1.2 log. Both gp91phox2/2 and gp91phox2/2/NOS22/2 macrophages were significantly impaired in their killing of S. typhimurium recBC. Pretreatment of gp91phox2/2/NOS22/2 macrophages with IFNg, TNF, IFNa/b, LPS, or GM-CSF alone or in combination did not increase their bactericidal activity (data not shown). The comparably poor ability of gp91phox2/2 and gp91phox2/2/NOS22/2 macrophages to kill both virulent and attenuated S. typhimurium points to the vital role of the respiratory burst in macrophage-mediated killing

Bactericidal Activity of Gp91phox2/2/NOS22/2 Mice 33

Figure 6. Macrophage Bactericidal Activity against S. typhimurium Periodate-elicited macrophages were infected with (A) S. typhimurium 14028 or (B) S. typhimurium recBC. In each experiment, macrophages were pooled from 3–5 mice. Bacterial counts were determined by fluorescent microplate assay. Log killing represents killing at 4 hr normalized to the number of bacteria at time 0. Results shown are the average of multiple independent experiments for each mouse strain 6 SEM. The averaged number of experiments are indicated within each bar. (inset) Representative time course comparing the killing of S. typhimurium 14028 by C57BL/6 3 129Sv (squares) or gp91phox2/2/NOS22/2 (circles) macrophages (log CFU/ ml 6 SEM; some error bars fall within the symbols). * p # 0.05 compared to C57BL/6; †p # 0.05 compared to C57BL/6 3 129Sv (unpaired t test).

of this organism under in vitro conditions that predominantly assess early-acting mechanisms in the hostpathogen interaction. Infection of Macrophages with L. monocytogenes Virulent L. monocytogenes 10403 caused cytolysis of gp91phox2/2, gp91phox2/2/NOS22/2, and TNF2/2 macrophages by 2 hr. Intracellular bacteria were released and killed by gentamicin, giving an artifactual appearance of antilisterial activity (data not shown). Therefore, we infected periodate-activated macrophages in vitro with virulent L. monocytogenes 10403 (Figure 7A) or with attenuated L. monocytogenes 15313 (Figure 7B) and assessed killing after 2 hr in the absence of gentamicin. Wild-type and NOS22/2 macrophages killed z1.0 log (Figure 7A). TNF2/2 macrophages were listeristatic. Gp91phox2/2 and gp91phox2/2/NOS22/2 macrophages permitted nearly exponential bacterial growth. When macrophages were infected with the attenuated strain of Listeria, killing by wild-type and NOS22/2 macrophages at 2 hr amounted to z2 log (Figure 7B).

Figure 7. Macrophage Bactericidal Activity against L. monocytogenes Periodate-elicited macrophages were infected with L. monocytogenes 10403 (A) or L. monocytogenes 15313 (B). In each experiment, macrophages were pooled from 3–5 mice. Bacterial counts were determined by a fluorescent microplate assay. Log killing represents killing at 2 hr in the absence of gentamicin normalized to the number of bacteria at time 0. Results shown are the average of multiple independent experiments for each mouse strain 6 SEM. The averaged number of experiments are indicated within each bar. (Inset) Average time course (n $ 6 experiments at each time point) comparing the survival of L. monocytogenes 15313 within C57BL/6 3 129Sv (squares) or gp91phox2/2/NOS22/2 (circles) macrophages (percent survival normalized to initial inoculum 6 SEM; error bars fall within the symbols). * p # 0.05 compared to C57BL/6; †p # 0.05 compared to C57BL/6 3 129Sv (unpaired t test).

Likewise, killing by TNF2/2 macrophages was nearly 2 log greater than their killing of the virulent strain, and the killing by gp91phox2/2 and gp91phox2/2/NOS22/2 macrophages increased to 0.7 log. Nevertheless, even with attenuated Listeria, killing by gp91phox2/2 and gp91phox2/2/ NOS22/2 macrophages was diminished compared to wildtype. Killing of L. monocytogenes 15313 by gp91phox2/2/ NOS22/2 macrophages was approximately 0.7, 0.8, and 1.0 log at 2, 4, and 8 hr, respectively (inset, 7B). As for periodate-elicited macrophages, listericidal activity of resident, thioglycollate-elicited and peptoneelicited gp91phox2/2/NOS22/2 macrophages was approximately 1 log at 8 hr after infection with Listeria 15313, and this was not increased by mIFNg, mIFNa/b, mGMCSF, hTNF, or LPS (data not shown). Nramp1 impacts macrophage bactericidal activity by what may be an indirect mechanism related to metal transport (Gunshin et al., 1997). Although no association between Nramp1 alleles and L. monocytogenes susceptibility has been reported in otherwise wild-type mice,

Immunity 34

we reasoned that Nramp1 might mediate listericidal activity that is normally redundant with that effected by ROI and RNI. To test this, we segregated macrophages from gp91phox2/2/NOS22/2 mice into groups homozygous for either the Nramp1-resistant or Nramp1-susceptible allele. However, no differences were observed between Nramp1-resistant or -susceptible gp91phox2/2/NOS22/2 macrophages in the killing of Listeria (data not shown). Discussion Mutual Compensation by Phox and NOS2 in the Control of Indigenous Flora Genetic evidence for the critical roles of phox and NOS2 in control of infection comes from study of patients with CGD and of mice with engineered deficiencies in either enzyme. However, studies conducted in over 100 laboratories involving the husbandry of thousands of NOS22/2 mice have produced no reports of spontaneous infections (Nathan, 1997). Mice deficient in gp47phox developed spontaneous external infections (Jackson et al., 1995). However, gp91phox2/2 mice did not (Pollock et al., 1995), and neither strain presented with abscesses of internal organs. During this study, gp91phox2/2 and NOS22/2 mice were free from spontaneous infection when housed with gp91phox2/2/NOS22/2 mice. In contrast, with or without prophylactic antibiotics, gp91phox2/2/ NOS22/2 mice spontaneously developed massive abscesses caused by normal flora, predominantly from the gastrointestinal tract but also from the respiratory tract and skin. During antibiotic prophylaxis, the prevalence of gross abscesses ranged from 15% to 25%; more mice may have harbored microscopic abscesses. Without antibiotics, the incidence of infection rose much higher. When the double deficiency was expressed on a pure C57BL/6 background, almost all the mice died from infection despite antibiotics. It thus appears that for resistance to spontaneous infection by indigenous flora, phox and NOS2 can substantially compensate for each other’s deficiency, and no other pathway is fully able to compensate for their combined absence, which would appear to be incompatible with survival of the species in the wild. Although knockout mice have been produced with defects in nearly twenty different genes important in innate immunity, none with normal numbers of phagocytes had a phenotype comparable to that of gp91phox2/2/NOS22/2 mice. The antiinfectious role of NOS2 has been studied chiefly from the point of view of host resistance to viruses (Karupiah et al., 1993, 1998; Zaragosa et al., 1998), protozoa (Wei et al., 1995; Khan et al., 1997; SchartonKersten et al., 1997; Diefenbach et al., 1998), and facultative intracellular bacterial pathogens infecting macrophages, such as Mycobacteria (MacMicking et al., 1997) and Listeria (MacMicking et al., 1995). In these settings, phox did not appear to compensate for lack of NOS2. Except for a report that NOS2-deficient mice were more susceptible to infection with Staphylococcus aureus than wild-type mice (McInnes et al., 1998), there are no studies with NOS22/2 mice that suggest that NOS2 plays a role in control of any bacteria other than facultative intracellular pathogens of macrophages.

Some explanation for the surprising finding that NOS2 participates in control of E. coli and other commensals may be provided by the demonstration that both rat (McCall et al., 1991) and human (Evans et al., 1996; Wheeler et al., 1997) neutrophils can express NOS2. Bacteria in neutrophils stained with antibodies against nitrotyrosine (Evans et al., 1996), a footprint of peroxynitrite formation (Beckman and Koppenol, 1996) or the action of myeloperoxidase and H2O2 on nitrite (van der Vliet et al., 1997). Alternatively, the role of NOS2 in resistance to infection by indigenous flora may be indirect. RNI can act as signaling molecules (Lander, 1997), and NOS2 was essential for early production of IL-12 and IFNg, maturation of NK cells, and suppression of TGF-b production in leishmania-infected mice (Diefenbach et al., 1998). A signaling role for NOS2 has also been demonstrated in the expression of inflammatory cytokines via activation of NF-kB and STAT3 during hemorrhagic shock/resuscitation (Hierholzer et al., 1998). Defective Bactericidal Responses of Gp91phox2/2/NOS22/2 Mice and Macrophages Compared to wild-type macrophages, gp91phox2/2/NOS22/2 macrophages were markedly deficient in killing all six strains of bacteria tested in vitro. However, the degree of deficiency relative to individually phox- or NOS2-deficient macrophages varied with the organism, and results of in vitro assays differed in some respects from assays in vivo. Such discrepancies are to be expected. In vitro, we studied only macrophages. In vivo, many cells and extracellular factors are likely to be involved, including cells other than macrophages that can express iNOS and phox. In vitro assays involve large experimental inocula and a static pool of macrophages deprived of contacts with other cells and cytokines. Most important, the short duration of assays in vitro favors a role for phox at the expense of NOS2, since production of ROI by phox is consummated much more rapidly upon contact with bacteria (1–3 hr) than the production of RNI by NOS2 (1–3 days). Gp91phox2/2/NOS22/2mice were highly susceptible in vivo and in vitro to infection with virulent and attenuated S. typhimurium. Gp91phox2/2 mice and macrophages were already shown to be susceptible to infection with S. typhimurium 14028, as well as a CuZn-SOD-deficient strain (De Groote et al., 1997). Restored virulence of the recBC mutant S. typhimurium in gp91phox2/2 mice underscores the critical importance of the respiratory burst in defense against S. typhimurium, and implies that recBC mutant S. typhimurium are unable to repair oxidative damage (Buchmeier et al., 1993). However, recBC-deficient S. typhimurium are also susceptible in vitro to S-nitrosoglutathione and SIN-1 (a peroxynitrite generator) (De Groote et al., 1995), an effect potentially caused by NO-mediated DNA damage (Wink et al., 1991). Prolonged survival of gp91phox2/2 mice compared to gp91phox2/2/NOS22/2 mice infected with recBC-deficient S. typhimurium suggests that RNI are also necessary for complete host defense against S. typhimurium. Increased susceptibility of NOS22/2 mice to virulent S. typhimurium confirms the importance of RNI to the killing of S. typhimurium as demonstrated by other means (De Groote et al., 1996).

Bactericidal Activity of Gp91phox2/2/NOS22/2 Mice 35

In vitro, both gp91phox2/2 and gp91phox2/2/NOS22/2 macrophages were impaired in their ability to kill virulent and attenuated S. typhimurium. In contrast, the ability of NOS22/2 macrophages to kill as well as wild-type macrophages in vitro diverged from the fate of infected mice. In earlier work, ROI-dependent killing of Salmonella was evident at 2–4 hr (Buchmeier et al., 1993), while NO-dependent killing was greater at 24 hr (De Groote et al., 1997). Most likely, these disparities reflect that in our cultures, macrophages infected by S. typhimurium underwent apoptosis (Lindgren et al., 1996; Monack et al., 1996) before they expressed NOS2 (data not shown). The killing of virulent Listeria in vitro by gp91phox2/2 and gp91phox2/2/NOS22/2 macrophages was markedly impaired compared to wild-type. In contrast, NOS22/2 macrophages killed Listeria 10403 as well as wild-type in vitro. Conversely, gp91phox deficiency had no evident effect in vivo but was detrimental in vitro. In the short time that the macrophages interacted with Listeria in vitro, the respiratory burst may have been sufficient to kill a majority of intracellular bacteria and NOS2 may not have been induced. Alternatively, the greater susceptibility of NOS22/2 mice compared to gp91phox2/2 mice in vivo may indicate that the role of RNI in vivo is broader than direct toxicity to bacteria. RNI can affect leukocyte adhesion, migration, and accumulation (Mulligan et al., 1991; Belenky et al., 1993), as well as cytokine production. The susceptibility of NOS22/2 mice to Listeria was consistent with previous observations using a different strain of Listeria (MacMicking et al., 1995). The minimal phenotype shown by gp91phox2/2 mice infected with Listeria also mirrored previous studies in that there was no difference in CFU between wild-type and gp91phox2/2 mice on day 6 (Dinauer et al., 1997), and gp47phox2/2 mice were only more susceptible than wild-type at a very high inoculum (Endres et al., 1997). The normal resistance of about half the gp91phox2/2/ NOS22/2 mice to infection with Listeria 10403 was unexpected. Macrophages may have been preactivated in mice with subclinical infections and thus better able to withstand the challenge inoculum. As precedent, T cell–deficient mice paradoxically displayed greater resistance to infection with Listeria than wild-type mice, apparently because their immune deficiency predisposed to subclinical infections resulting in chronic macrophage activation (Cheers and Waller, 1975; Nickol and Bonventre, 1977). Consistent with this interpretation, enhanced resistance to Listeria was not evident when the T cell–deficient mice were flora defined or germ free (Czuprynski and Brown, 1985). Complement C5 and unidentified genetic factors can also affect susceptibility to Listeria (Cheers and McKenzie, 1978; Gervais et al., 1984). Existence of a Macrophage Bactericidal Mechanism Independent of both Phox and NOS2 Though their bactericidal activity was markedly impaired, gp91phox2/2/NOS22/2 macrophages were able to kill a proportion of organisms of 5/6 strains tested. These findings establish the existence of a phox-independent, NOS2-independent mechanism in primary mouse macrophages. Such a mechanism was suggested in a transformed cell line (Inoue et al., 1995) but had not been

demonstrated by genetic means nor in primary cells. Since there are other sources of ROI besides phox and other sources of RNI besides NOS2, this is not synonymous with establishing the existence of an ROI- and RNI-independent bactericidal mechanism, but the latter seems likely to be the case. Although the phox-, NOS2-independent macrophage antimicrobial activity was modest against the organisms studied, it may be more effective and perhaps essential for host resistance to other organisms against which neither ROI nor RNI provide effective resistance. Indeed, the variety of potential pathogens and their propensity to suppress or evade host defenses makes a diversity of antimicrobial mechanisms seem mandatory. Experimental Procedures Bacterial Strains and Growth Media E. coli HB101, E. coli isolated from a mouse abscess (designated 9243), S. typhimurium ATCC 14028, and S. typhimurium recBC (CL2001) (Buchmeier et al., 1993) were grown in Luria-Bertani (LB) broth or on LB plates. L. monocytogenes 15313 from ATCC and 10403 from D. Portnoy were grown in tryptose phosphate broth (TPB) or plates. Reagents Tryptone, yeast extract, tryptose phosphate, proteose peptone, and thioglycollate were from Difco. Periodate and deoxycholate were from Sigma, sodium chloride and sodium phosphate dibasic from J. T. Baker, potassium phosphate from Fischer, and dextrose from EM Science. RPMI from JRH Bioscience was supplemented with 2 mM L-glutamine and heat-inactivated fetal bovine serum from Hyclone. AlamarBlue was obtained from Sensititre/Alamar (AccuMed International Companies, Westlake, OH). Mouse IFNg was generously provided by Genentech, mIFNa/b was from Access Biomedical, mGM-CSF was from Genzyme, and LPS (E. coli O111:B4) was from Sigma. Mouse Breeding and Genotyping We derived mice lacking NOS2 (NOS22/2) and their wild-type controls (C57BL/6 3 129Sv) by interbreeding C57BL/6 3 129Sv founders heterozygous for NOS2 deficiency and maintained independent colonies of NOS22/2 and C57BL/6 3 129Sv mice by intersib matings as described (MacMicking et al., 1995). Interbreeding of mice lacking one functional gp91phox allele (Pollock et al., 1995) backcrossed six generations onto the C57BL/6 background allowed us to select and maintain a line with complete gp91phox deficiency. Wild-type C57BL/6 controls were purchased from Charles River Laboratories. Interbreeding of NOS22/2 and gp91phox2/2 mice generated gp91phox2/2/ NOS22/2mice. Intersib matings of homozygous double knockouts provided the gp91phox2/2/NOS22/2mice used in this study. TNF-deficient mice (C57BL/6 3 129Sv) were derived by intersib mating of mice heterozygous for the TNF mutation and housed as described (Marino et al., 1997). Genomic DNA was screened for the NOS2 allele by Southern analysis (MacMicking et al., 1995) or by PCR as described (MacMicking et al., 1997) or using a cocktail containing a common 59 primer flanking the NOS2 gene (59-ATCAGCCTTTCTCTGTCTCC-39) and two 39 primers, one within the NOS2 gene (59-GGCTTTCTGTCTGTT CTCTC-39) yielding a 413 bp amplificand and one specific to the PGK-neo marker in the knockout allele (59-GCCTGAAGAGACGAGA TCAGCAGCCTCTG-39) yielding a 268 bp amplificand. To screen for the gp91phox allele, a cocktail PCR was used, containing a common 59 primer (59-AAGAGAAACTCCTCTGCTGTGAA-39) and two 39 primers, one within the gp91phox gene (59-CGCACTGGAAACCCTGAGAA AGG-39) yielding a 240 bp amplificand and one specific to the PGKneo marker in the knockout allele (59-TGCCAAGTTCTAATTC CATCAGAAGC-39) yielding a 195 bp amplificand. Both PCR amplifications were performed for 30 cycles of 30 sec at 948C, 45 sec at 658C, and 1.5 min at 728C, the products resolved by electrophoresis

Immunity 36

on a 1.5% agarose gel and visualized by ethidium bromide staining. Nramp1 PCR haplotyping was as described (MacMicking et al., 1997). Mouse studies were performed according to the institutional guidelines following protocols approved by the Animal Care and Use Committee. All mice were housed in specific pathogen-free (SPF) conditions in microisolator bonnet cages with autoclaved bedding, cages, food, and water bottles. Positive ventilation pressure was maintained in the room and mice were handled in a laminar flow hood using sterile techniques. Food was autoclaved and drinking water acidified (pH 2.8). Gp91phox2/2/NOS22/2 mice were maintained on water containing itraconazole (15 mg/ml) and Bactrim (0.2 mg/ml trimethoprim and 40 mg/ml sulfamethoxazole). Antibioticfree water was provided for at least 4 days before experiments to allow elimination of the antibiotic (T1/2 ≈ 12 hr) even though giving mice access to antibiotic-containing water had no demonstrable effect on the bactericidal activity of their macrophages in vitro (data not shown).

were replenished with 100 ml complete medium plus 10 mg/ml gentamicin to kill extracellular bacteria and incubation resumed. Alternatively, where indicated, plates were replenished with 100 ml antibiotic-free complete medium. At each time point, another culture plate was lysed and aliquoted to a readout plate as above. Fractional survival was determined by dividing the CFU at each time point and condition by the starting CFU for that condition. For each experiment, the negative log10 of fractional survival was calculated to determine log killing. Log killing at specific time points was averaged for all experiments.

Macrophage NO Production and Respiratory Burst Macrophages were collected 3–4 days after i.p. injection of 4% thioglycollate broth, plated as described (Ding et al., 1988) in RPMI supplemented with 10% heat-inactivated fetal bovine serum, 1% L-glutamine (complete medium) in 96-well flat-bottom polystyrene plates (2 3 105/well), and exposed in vitro for 24 hr to medium alone, E. coli LPS (100 ng/ml), rMuIFNg (100 U/ml), or both LPS (10 ng/ ml) and rMuIFNg (10 U/ml). Nitrite accumulation in the medium in triplicate cultures was quantified by Griess’s assay (Ding et al., 1988). The respiratory burst was measured as the H2O2-dependent oxidation of fluorescent scopoletin (De la Harpe and Nathan, 1985). Macrophages were elicited by i.p. injection of 1 ml 5 mM sodium periodate and harvested 3–4 days later by peritoneal lavage. Macrophages (2 3 105/well) were allowed to adhere as above and exposed in vitro for 24 hr to medium alone or rMuIFNg (100 U/ml). After 24 hr, the monolayers were washed and production of H2O2 recorded as described (De la Harpe and Nathan, 1985).

Acknowledgments

Infection with S. typhimurium and Listeria Overnight shaking cultures of S. typhimurium 14028 or recBC were washed twice with sterile saline, resuspended in saline and diluted to a concentration of 1 3 103 CFU/ml. Mice (n 5 5 per group) housed in microisolator cages were injected i.p. with 200 organisms/mouse as confirmed by plating aliquots on LB plates and observed daily for mortality. Overnight shaking cultures of L. monocytogenes ATCC strains 10403 or 15313 were washed twice with sterile saline. Strain 10403 was diluted to yield 5 3 104 CFU/200 ml per mouse. Mice were injected i.p., and 3 days later spleens were homogenized in sterile distilled water for bacterial counts of serial 10-fold dilutions on tryptose phosphate agar plates. To normalize data from three independent experiments, spleen CFU for wild-type C57BL/6 3 129Sv mice were averaged for each experiment and used as the denominator by which all other spleen CFUs determined in that experiment were divided to give relative CFU values (fold above control). For L. monocytogenes 15313, experiments 1 and 2 were as described above, while experiments 3 and 4 were modified such that mice were sacrificed 2 days after injection.

Bactericidal Assays Bactericidal assays were as described (Shiloh et al., 1997). In brief, periodate-elicited macrophages in 96-well plates were infected with bacteria diluted from an overnight culture and opsonized in 10% normal mouse serum. Plates were centrifuged (250 3 g, 258C, 5 min), transferred to 378C, 5% CO2 for 10 min and then washed three times with sterile saline. All wells in one plate were immediately lysed by addition of 100 ml per well of sterile 0.1% Triton X-100. Lysis with 0.1% Triton X-100 had no effect on bacterial viability (data not shown). After mixing each well, 10 ml lysate was transferred from each well into 100 ml LB-AlamarBlue (or TPB-AlamarBlue for Listeria) in a readout plate in which viable bacteria were quantitated by fluorescence (Shiloh et al., 1997). Remaining cell culture plates

Statistical Analyses Using StatView 4.5 software (Abacus Concepts, Inc., Berkeley, CA), the log-rank (Mantel-Cox) test was applied to Kaplan-Meier survival data to evaluate differences between groups. The nonparametric Kruskal-Wallis test was applied to Listeria organ CFU experiments, and significance between groups was determined by the MannWhitney U test. Macrophage bactericidal assays were analyzed by one-way analysis of variance and unpaired t-test.

We thank L. Old for encouragement and support, R. North and D. Portnoy for advice and bacterial strains, N. Lipman and S. Perkins for advice on animal husbandry, Genentech for a gift of cytokines, and D. Marciano for critique of the manuscript. This work was supported by National Institutes of Health grants HL51967, AI34543, and GM53921 to C. N., HL52565 to M. D., AI39557 to F. F., and a Medical Scientist Training Program Grant GM07739 fellowship and Iris L. and Leverett S. Woodward Medical Scientist Fellowship to M. U. S. Received August 7, 1998; revised November 23, 1998. References Beckman, J.S., and Koppenol, W.H. (1996). Nitric oxide, superoxide, and peroxynitrite: the good, the bad, and ugly. Am. J. Physiol. 271, C1424–C1437. Belenky, S.N., Robbins, R.A., and Rubinstein, I. (1993). Nitric oxide synthase inhibitors attenuate human monocyte chemotaxis in vitro. J. Leukoc. Biol. 53, 498–503. Blanden, R.V., Mackaness, G.B., and Collins, F.M. (1966). Mechanisms of acquired resistance in mouse typhoid. J. Exp. Med. 124, 585–600. Buchmeier, N.A., Lipps, C.J., So, M.Y., and Heffron, F. (1993). Recombination-deficient mutants of Salmonella typhimurium are avirulent and sensitive to the oxidative burst of macrophages. Mol. Microbiol. 7, 933–936. Cheers, C., and Waller, R. (1975). Activated macrophages in congenitally athymic “nude mice” and in lethally irradiate mice. J. Immunol. 115, 844–847. Cheers, C., and McKenzie, I.F. (1978). Resistance and susceptibility of mice to bacterial infection: genetics of listeriosis. Infect. Immun. 19, 755–762. Conlan, J.W., and North, R.J. (1991). Neutrophil-mediated dissolution of infected host cells as a defense strategy against a facultative intracellular bacterium. J. Exp. Med. 174, 741–744. Czuprynski, C.J., and Brown, J.F. (1985). Phagocytes from floradefined and germfree athymic nude mice do not demonstrate enhanced antibacterial activity. Infect. Immun. 50, 425–430. Dai, W.J., Bartens, W., Kohler, G., Hufnagel, M., Kopf, M., and Brombacher, F. (1997). Impaired macrophage listericidal and cytokine activities are responsible for the rapid death of Listeria monocytogenes-infected IFN-g receptor-deficient mice. J. Immunol. 158, 5297–5304. De Groote, M.A., Granger, D., Xu, Y., Campbell, G., Prince, R., and Fang, F.C. (1995). Genetic and redox determinants of nitric oxide cytotoxicity in a Salmonella typhimurium model. Proc. Natl. Acad. Sci. USA 92, 6399–6403. De Groote, M.A., Testerman, T., Xu, Y., Stauffer, G., and Fang, F.C.

Bactericidal Activity of Gp91phox2/2/NOS22/2 Mice 37

(1996). Homocysteine antagonism of nitric oxide-related cytostasis in Salmonella typhimurium. Science 272, 414–417. De Groote, M.A., Ochsner, U.A., Shiloh, M.U., Nathan, C., McCord, J.M., Dinauer, M.C., Libby, S.J., Vazquez-Torres, A., Xu, Y., and Fang, F.C. (1997). Periplasmic superoxide dismutase protects Salmonella from products of phagocyte NADPH-oxidase and nitric oxide synthase. Proc. Natl. Acad. Sci. USA 94, 13997–14001.

Listeria monocytogenes in the J774.1 macrophage-like cell line and the lipopolysaccharide (LPS)-resistant mutant LPS1916 cell line defective in the generation of reactive oxygen intermediates after LPS treatment. Infect. Immun. 63, 1876–1886. Jackson, S.H., Gallin, J.I., and Holland, S.M. (1995). The p47phox mouse knock-out model of chronic granulomatous disease. J. Exp. Med. 182, 751–758.

De la Harpe, J., and Nathan, C.F. (1985). A semi-automated microassay for H2O2 release by human blood monocytes and mouse peritoneal macrophages. J. Immunol. Methods 78, 323–336.

Karupiah, G., Xie, Q.W., Buller, R.M., Nathan, C., Duarte, C., and MacMicking, J.D. (1993). Inhibition of viral replication by interferongamma-induced nitric oxide synthase. Science 261, 1445–1448.

Diefenbach, A., Schindler, H., Donhauser, N., Lorenz, E., Laskay, T., MacMicking, J., Ro¨llinghoff, M., Gresser, I., and Bogdan, C. (1998). Type1 interferon (IFNa/b) and type 2 nitric oxide synthase regulate the innate immune response to a protozoan parasite. Immunity 8, 77–87.

Karupiah, G., Chen, J.H., Nathan, C.F., Mahalingam, S., and MacMicking, J.D. (1998). Identification of nitric oxide synthase 2 as an innate resistance locus against ectromelia virus infection. J. Virol. 72, 7703–7706.

Dinauer, M.C., Deck, M.B., and Unanue, E.R. (1997). Mice lacking reduced nicotinamide adenine dinucleotide phosphate oxidase activity show increased susceptibility to early infection with Listeria monocytogenes. J. Immunol. 158, 5581–5583. Ding, A.H., Nathan, C.F., and Stuehr, D.J. (1988). Release of reactive nitrogen intermediates and reactive oxygen intermediates from mouse peritoneal macrophages. Comparison of activating cytokines and evidence for independent production. J. Immunol. 141, 2407– 2412. Dunn, P.L., and North, R.J. (1991). Early gamma interferon production by natural killer cells is important in defense against murine listeriosis. Infect. Immun. 59, 2892–2900. Endres, R., Luz, A., Schulze, H., Neubauer, H., Futterer, A., Holland, S.M., Wagner, H., and Pfeffer, K. (1997). Listeriosis in p47(phox2/2) and TRp552/2 mice: protection despite absence of ROI and susceptibility despite presence of RNI. Immunity 7, 419–432. Evans, T.J., Buttery, L.D., Carpenter, A., Springall, D.R., Polak, J.M., and Cohen, J. (1996). Cytokine-treated human neutrophils contain inducible nitric oxide synthase that produces nitration of ingested bacteria. Proc. Natl. Acad. Sci. USA 93, 9553–9558. Fang, F.C. (1997). Perspectives series: host/pathogen interactions. Mechanisms of nitric oxide-related antimicrobial activity. J. Clin. Invest. 99, 2818–2825. Fehr, T., Schoedon, G., Odermatt, B., Holtschke, T., Schneemann, M., Bachmann, M.F., Mak, T.W., Horak, I., and Zinkernagel, R.M. (1997). Crucial role of interferon consensus sequence binding protein, but neither of interferon regulatory factor 1 nor of nitric oxide synthesis for protection against murine listeriosis. J. Exp. Med. 185, 921–931. Fields, P.I., Swanson, R.V., Haidaris, C.G., and Heffron, F. (1986). Mutants of Salmonella typhimurium that cannot survive within the macrophage are avirulent. Proc. Natl. Acad. Sci. USA 83, 5189–5193. Gervais, F., Stevenson, M., and Skamene, E. (1984). Genetic control of resistance to Listeria monocytogenes: regulation of leukocyte inflammatory responses by the Hc locus. J. Immunol. 132, 2078– 2083. Gunshin, H., Mackenzie, B., Berger, U.V., Gunshin, Y., Romero, M.F., Boron, W.F., Nussberger, S., Gollan, J.L., and Hediger, M.A. (1997). Cloning and characterization of a mammalian proton-coupled metalion transporter. Nature 388, 482–488. Harty, J.T., and Bevan, M.J. (1995). Specific immunity to Listeria monocytogenes in the absence of IFNg. Immunity 3, 109–117. Harty, J.T., Schreiber, R.D., and Bevan, M.J. (1992). CD8 T cells can protect against an intracellular bacterium in an interferon gammaindependent fashion. Proc. Natl. Acad. Sci. USA 89, 11612–11616. Hierholzer, C., Harbrecht, B., Menezes, J.M., Kane, J., MacMicking, J., Nathan, C.F., Peitzman, A.B., Billiar, T.R., and Tweardy, D.J. (1998). Essential role of induced nitric oxide in the initiation of the inflammatory response after hemorrhagic shock. J. Exp. Med. 187, 917–928. Huang, S., Hendriks, W., Althage, A., Hemmi, S., Bluethmann, H., Kamijo, R., Vilcek, J., Zinkernagel, R.M., and Aguet, M. (1993). Immune response in mice that lack the interferon-gamma receptor. Science 259, 1742–1745. Inoue, S., Itagaki, S., and Amano, F. (1995). Intracellular killing of

Kaufmann, S.H.E. (1993). Immunity to intracellular bacteria. Annu. Rev. Immunol. 11, 129–163. Kaufmann, S.H., and Ladel, C.H. (1994). Role of T cell subsets in immunity against intracellular bacteria: experimental infections of knock-out mice with Listeria monocytogenes and Mycobacterium bovis BCG. Immunobiology 191, 509–519. Khan, I.A., Schwartzman, J.D., Matsuura, T., and Kasper, L.H. (1997). A dichotomous role for nitric oxide during acute Toxoplasma gondii infection in mice. Proc. Natl. Acad. Sci. USA 94, 13955–13960. Kopf, M., Baumann, H., Freer, G., Freudenberg, M., Lamers, M., Kishimoto, T., Zinkernagel, R., Bluethmann, H., and Kohler, G. (1994). Impaired immune and acute-phase responses in interleukin-6-deficient mice. Nature 368, 339–342. Lander, H.M. (1997). An essential role for free radicals and derived species in signal transduction. FASEB J. 11, 118–124. Lindgren, S.W., Stojiljkovic, I., and Heffron, F. (1996). Macrophage killing is an essential virulence mechanism of Salmonella typhimurium. Proc. Natl. Acad. Sci. USA 93, 4197–4201. Mackaness, G.B. (1962). Cellular resistance to infection. J. Exp. Med. 116, 381–410. MacMicking, J.D., Nathan, C., Hom, G., Chartrain, N., Fletcher, D.S., Trumbauer, M., Stevens, K., Xie, Q.W., Sokol, K., Hutchinson, N., et al. (1995). Altered responses to bacterial infection and endotoxic shock in mice lacking inducible nitric oxide synthase [published erratum appears in Cell 1995 Jun 30; 81(7):following 1170]. Cell 81, 641–650. MacMicking, J.D., North, R.J., La Course, R., Mudgett, J.S., Shah, S.K., and Nathan, C.F. (1997). Identification of nitric oxide synthase as a protective locus against tuberculosis. Proc. Natl. Acad. Sci. USA 94, 5243–5248. Malo, D., Vogan, K., Vidal, S., Hu, J., Cellier, M., Schurr, E., Fuks, A., Bumstead, N., Morgan, K., and Gros, P. (1994). Haplotype mapping and sequence analysis of the mouse Nramp gene predict susceptibility to infection with intracellular parasites. Genomics 23, 51–61. Marino, M.W., Dunn, A., Grail, D., Inglese, M., Noguchi, Y., Richards, E., Jungbluth, A., Wada, H., Moore, M., Williamson, B., Basu, S., and Old, L.J. (1997). Characterization of tumor necrosis factor-deficient mice. Proc. Natl. Acad. Sci. USA 94, 8093–8098. McCall, T.B., Palmer, R.M., and Moncada, S. (1991). Induction of nitric oxide synthase in rat peritoneal neutrophils and its inhibition by dexamethasone. Eur J. Immunol. 21, 2523–2527. McInnes, I.B., Leung, B., Wei, X.Q., Gemmell, C.C., and Liew, F.Y. (1998). Septic arthritis following Staphylococcus aureus infection in mice lacking inducible nitric oxide synthase. J. Immunol. 160, 308–315. Monack, D.M., Raupach, B., Hromockyj, A.E., and Falkow, S. (1996). Salmonella typhimurium invasion induces apoptosis in infected macrophages. Proc. Natl. Acad. Sci. USA 93, 9833–9838. Mouy, R., Fischer, A., Vilmer, E., Seger, R., and Griscelli, C. (1989). Incidence, severity and prevention of infections in chronic granulomatous disease. J. Pediatr. 114, 555–560. Mulligan, M.S., Hevel, J.M., Marletta, M.A., and Ward, P.A. (1991). Tissue injury caused by deposition of immune complexes is L-arginine dependent. Proc. Natl. Acad. Sci. USA 88, 6338–6342. Nathan, C.F. (1983). Mechanisms of macrophage antimicrobial activity. Trans. R. Soc. Trop. Med. Hyg. 77, 620–630.

Immunity 38

Nathan, C. (1997). Perspective series: nitric oxide and nitric oxide synthases. Inducible nitric oxide synthase: what difference does it make? J. Clin. Invest. 100, 2417–2423. Nickol, A.D., and Bonventre, P.F. (1977). Anomalous high native resistance to athymic mice to bacterial pathogens. Infect. Immun. 18, 636–645. North, R.J., Dunn, P.L., and Conlan, J.W. (1997). Murine listeriosis as a model of antimicrobial defense. Immunol. Rev. 158, 27–36. Pasparakis, M., Alexopoulou, L., Episkopou, V., and Kollias, G. (1996). Immune and inflammatory responses in TNF alpha-deficient mice: a critical requirement for TNF alpha in the formation of primary B cell follicles, follicular dendritic cell networks and germinal centers, and in the maturation of the humoral immune response. J. Exp. Med. 184, 1397–1411. Pollock, J.D., Williams, D.A., Gifford, M.A., Li, L.L., Du, X., Fisherman, J., Orkin, S.H., Doerschuk, C.M., and Dinauer, M.C. (1995). Mouse model of X-linked chronic granulomatous disease, an inherited defect in phagocyte superoxide production. Nat. Genet. 9, 202–209. Portnoy, D.A., Schreiber, R.D., Connelly, P., and Tilney, L.G. (1989). g interferon limits access of Listeria monocytogenes to the macrophage cytoplasm. J. Exp. Med. 170, 2141–2146. Scharton-Kersten, T.M., Yap, G., Magram, J., and Sher, A. (1997). Inducible nitric oxide is essential for host control of persistent but not acute infection with the intracellular pathogen Toxoplasma gondii. J. Exp. Med. 185, 1261–1273. Shiloh, M.U., and Nathan, C. (1999). Antimicrobial mechanisms of macrophages. In Phagocytosis and Pathogens, S. Gordon, ed. (Greenwich, CT: JAI Press, Inc.). Shiloh, M.U., Ruan, J., and Nathan, C. (1997). Evaluation of bacterial numbers and phagocyte function with a fluorescence-based microplate assay. Infect. Immun. 65, 3193–3198. Tanaka, T., Akira, S., Yoshida, K., Umemoto, M., Yoneda, Y., Shirafuji, N., Fujiwara, H., Suematsu, S., Yoshida, N., and Kishimoto, T. (1995). Targeted disruption of the NF-IL6 gene discloses its essential role in bacteria killing and tumor cytotoxicity by macrophages. Cell 80, 353–361. Unanue, E.R. (1997). Studies in listeriosis show the strong symbiosis between the innate cellular system and the T-cell response. Immunol. Rev. 158, 11–25. van der Vliet, A., Eiserich, J.P., Halliwell, B., and Cross, C.E. (1997). Formation of reactive nitrogen species during peroxidase-catalyzed oxidation of nitrite. A potential additional mechanism of nitric oxidedependent toxicity. J. Biol. Chem. 272, 7617–7625. Vidal, S., Tremblay, M.L., Govoni, G., Gauthier, S., Sebastiani, G., Malo, D., Skamene, E., Olivier, M., Jothy, S., and Gros, P. (1995). The Ity/Lsh/Bcg locus: natural resistance to infection with intracellular parasites is abrogated by disruption of the Nramp1 gene. J. Exp. Med. 182, 655–666. Wei, X.Q., Charles, I.G., Smith, A., Ure, J., Feng, G.J., Huang, F.P., Xu, D., Muller, W., Moncada, S., and Liew, F.Y. (1995). Altered immune responses in mice lacking inducible nitric oxide synthase. Nature 375, 408–411. Weih, F., Warr, G., Yang, H., and Bravo, R. (1997). Multifocal defects in immune responses in RelB-deficient mice. J. Immunol. 158, 5211– 5218. Wheeler, M.A., Smith, S.D., Garcia-Cardena, G., Nathan, C.F., Weiss, R.M., and Sessa, W.C. (1997). Bacterial infection induces nitric oxide synthase in human neutrophils. J. Clin. Invest. 99, 110–116. Wink, D.A., Kasprzak, K.S., Maragos, C.M., Elespuru, R.K., Misra, M., Dunams, T.M., Cebula, T.A., Koch, W.H., Andrews, A.W., Allen, J.S., et al. (1991). DNA deaminating ability and genotoxicity of nitric oxide and its progenitors. Science 254, 1001–1003. Zaragosa, C., Ocampo, C., Saura, M., Leppo, M., Wei, X.Q., Quick, R., Moncada, S., Liew, F.Y., and Lowenstein, C.J. (1998). The role of inducible nitric oxide synthase in the host response to Coxsackievirus myocarditis. Proc. Natl. Acad. Sci. USA 95, 2469–2474.