The Epistatic Interrelationships of IL-1, IL-1 Receptor Antagonist, and the Type I IL-1 Receptor

Mice

As described in detail previously (11, 12), the single-mutant mouse strains used in these experiments are mice that lack IL-1RI (RIko), mice that lack all forms of IL-1ra (rako), and mice that hemizygously carry a transgene that encodes the sIL-1ra mRNA under control of its endogenous promoter (ratg). The ratg line used for these experiments has been previously referred to as T14. Double mutants were generated by crossing the single-mutant strains to create founders for the various strains. Briefly, the following matings were used to generate mice for experiments: RIko from −/− × −/− or +/− × −/− or +/− × +/−; rako from +/− × +/− (because −/− are infertile; see below); ratg from T14/o × +/+; rako, RIko from +/−, +/− × +/−, +/− or −/−, −/− × −/−, −/− or −/−, −/− × +/−, −/− or −/−, −/− × −/−, +/−; rako, ratg from −/−, T14/o × +/−, o/o; and ratg, RIko from T14/o, +/− × o/o, −/−. Wild-type (wt) controls were either littermates of ratg hemizygotes, littermates of rako mice, or occasionally the offspring of wt sibling matings. Genotyping was performed by PCR as previously described (11). All mice used in these experiments were backcrossed to C57BL/6J genetic backgrounds for ≥10 generations, bred in a specific-pathogen-free facility, and handled in accordance with institutional guidelines.

Monitoring of development/homeostasis

For the monitoring of postnatal developmental abnormalities in the mutant mice, littermates were caged together (three or four mice per cage) and allowed access to food and water ad libitum. Once a week, from 3–52 wk of age, mice were weighed and scored for the presence of sickness symptoms by an investigator who was blinded to the genotypes of the mice (n = 14–36/genotype/sex). To judge sickness, a scale of 0–10 was used to access both the number and the severity of sickness symptoms (including lethargy, piloerection, hunched posture, etc.), with 0 being no abnormalities, and 10 being moribund/dead. Fertility was ascertained by mating mutant mice and littermate controls starting at 6–7 wk of age with proven breeders (n = 10/genotype/sex).

Models of septic shock and listeriosis

Endotoxic shock was induced as previously described using LPS from Salmonella typhimurium (Sigma, St. Louis MO) at 10 μg/g administered i.p. (11). For studying primary listeriosis, L. monocytogenes EGD was injected i.v. at a dose of 1.0 × 10⁶ CFU in 100 μl. Bacteria were enumerated by homogenization of organs in sterile PBS and plating serial log dilutions (14). For both challenge experiments, age-matched mice were inoculated on the same day, and survival was monitored for a total of 7 days postchallenge (n = 5–15/genotype/sex). The results reported are cumulative from three to five repeat experiments.

Analysis of blood and serum

Serum collection and blood analysis were performed as previously described (11, 14). ELISAs were performed as previously described and followed the manufacturers’ suggested protocols using the following Abs or kits: polyclonal goat anti-mIL-1ra (R&D Systems, Minneapolis, MN); DuoSet kits for IL-1β, IL-10, and TNF-α (R&D Systems); and OptEIA for IFN-γ (BD PharMingen, San Diego, CA).

Protein isolation and immunoblotting

For protein isolation, we used a modification of the TRIzol reagent protocol (Life Technologies, Gaithersburg, MD). Briefly, murine tissues were homogenized in TRIzol reagent. Following RNA isolation and removal of DNA, the proteins were isopropanol-precipitated, washed once in 0.3 M guanidine HCl/95% ethanol, and then resuspended in 6 M guanidine HCl. After dialysis overnight at 4°C with three changes of 10 mM Tris-HCl (pH 7.0) using Slide-A-Lyzer cassettes (10-kDa M_r cut-off; Pierce, Rockford, IL), the dialysate was collected and cleared by centrifugation (10,000 × g, 15 min, 4°C). The protein concentration of the resulting supernatant, which contained all the isoforms of IL-1ra (data not shown), was quantitated using the Bradford assay kit (Bio-Rad, Hercules, CA). For immunoblot analysis, 20 μg total protein was loaded on a 17.5% polyacrylamide gel and transferred to polyvinylidene difluoride membranes (Bio-Rad). Membranes were blocked for 1–2 h at room temperature in 10% nonfat dry milk in PBS/Tween, washed, incubated with 0.2 μg/ml biotinylated polyclonal goat anti-mouse IL-1ra Ab (R&D Systems) for 1 h at room temperature, washed again, and finally incubated with streptavidin-HRP (1/1,000 dilution; Amersham Pharmacia Biotech, Piscataway, NJ). Positive signal was visualized using ECL detection reagents (Amersham Pharmacia Biotech) following the manufacturer’s recommended protocol.

FACS analysis

Single-cell suspensions from thymi, spleens, and bone marrows, pooled from femur and tibia, were isolated from mutant and wt mice following standard protocols. After lysis of RBC, 10⁶ cells were stained for 30 min with the appropriate combination of mAbs. Cells were washed (3% FCS and 0.1% sodium azide/PBS), stained with secondary reagents when necessary for 30 min, washed again, and analyzed with a FACSCalibur cytometer (BD Biosciences, San Diego CA). The mAbs used for staining were anti-CD45R/B220-PE (RA3-6B2), anti-CD3-FITC (17 A2), anti-CD8a-PerCP (Ly-2/53-6.7), anti-CD4-FITC (L3T4/RM4-5), anti-CD69-PE (H1.2F3), and anti-γδ TCR-biotin (GL3; all from BD PharMingen), and anti-Mac-1-biotin (M1/70) and anti-IgM-allophycocyanin (a gift from G. Siu). For the biotin-conjugated mAbs, the secondary staining reagent used was streptavidin-PE (BD PharMingen).

Development and homeostasis

The rako mice lack all IL-1ra isoforms (11). They show abnormal development, including postnatal runting, reduced fertility, and early mortality from an unknown cause. Individual variability exists as to age of onset, symptom progression, and age of death. Nevertheless, certain generalizations can be made. The first detectable symptom, at 5–6 wk of age, is postnatal runting caused by failure to gain weight, primarily body fat (Fig. 2⇓) (15). At the same time, although otherwise asymptomatic, rako mice have an altered cellular profile, including pronounced neutrophilia and mild leukocytosis (Table I⇓). At 8–10 wk of age, rako mice have normal thymic cell populations, but alterations in some splenic cell types, specifically an increase in both γδ T cells (2.15 vs 1.56% for rako vs wt mice, respectively; n = 7 females/genotype; p = 0.03, by Student’s t test) and Mac-1⁺ cells (16.5 vs 14.2%; p = 0.02). Mac-1⁺ cells are also increased in the bone marrow of rako mice (42.5 vs 28.4%; n = 4 females/genotype; p = 0.02). Similar altered cellular profiles are present in rako males. Young adults are also subfertile; 50% of males and 30% of females were able to reproduce compared with 90–100% of wt and heterozygous littermates (n = 10/sex/genotype).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Postnatal growth curves of rako mice and demonstration of genetic rescue by both RIko and ratg. Shown are the average weekly weights of males (A) and females (B) from 14–36 mice of each genotype. The rako vs wt weights differ significantly starting at 5 wk of age; those of rako, RI +/− vs wt differ significantly starting at 7 wk of age.

As they age, rako mice become symptomatic, with piloerection, mild weight loss, and other common indications of illness (age when 50% are afflicted: females, 14 wk (n = 33); males, 17 wk (n = 36)). Mice at this stage also become 100% infertile. In a long term study, only 40% (10 of 25) of rako females and 43% (15 of 35) of rako males survived for 1 yr, compared with 100% survival of wt mice (19 of 19 females, 18 of 18 males). We have been unable to identify the ultimate cause of the early mortality of our IL-1ra knockout mice. The results of extensive histopathology, serology tests, and other analyses have not revealed any specific defect sufficient to account for the observed mortality rate.

We have taken a genetic epistasis approach to investigate the mechanism of action of the lack of IL-1ra on development and homeostasis. We constructed rako, RIko double mutants. RIko, which is developmentally normal, is epistatic to rako; deleting the type I receptor fully rescues the developmental phenotypes of rako mice (Fig. 2⇑ and Table I⇑). Further, the double-mutant IL-1ra^−/−, IL-1RI^+/− displays a partially dominant epistatic phenotype. As shown in Fig. 2⇑, removing a single RI allele restores rako halfway to wt weights, while removal of both alleles (rako, RIko) completely restores them. Removing a single RI allele also delays, rather than prevents, the onset of infertility and sickness symptoms (age of 50% affliction: females, 21 wk (n = 18); males, 27 wk (n = 16)) and improves 1-yr survival (64% (9 of 14) females and 92% (11 of 12) males; compare above). The single-mutant IL-1RI^+/− is phenotypically wt, and there is no difference in phenotype of the IL-1ra^−/−, IL-1RI^+/− mutants that correlates with parental inheritance of wt or knockout RI allele, i.e., no imprinting or other epigenetic inheritance mechanism is suggested (data not shown). Taken together, these data indicate that the developmental abnormalities seen in the rako mice are dependent upon the presence as well as the gene dosage of the IL-1RI.

The double-mutant rako, ratg was also generated by crossing the two IL-1ra mutant strains. These rako, ratg doubles show the same developmental phenotype as the ratg parental strain, which is that of wt. Thus, ratg is epistatic to rako. The developmental phenotype, including runting, neutrophilia, fertility, and earlier mortality, can be rescued by overexpression of the sIL-1ra splice variant mRNA (Fig. 2⇑, Table I⇑, and data not shown).

To investigate more thoroughly which isoform of IL-1ra was responsible for this complete dominant epistasis, we analyzed the IL-1ra protein isoforms present in the various mutant mice. Both sIL-1ra and icIL-1ra3 isoforms were detected in the livers of all mice except rako, while the icIL-1ra1 isoform was not present in rako or rako, ratg mutants (Fig. 1⇑). This demonstrates that the sIL-1ra mRNA encoded by the transgene produces both sIL-1ra and icIL-1ra3. Therefore, the genetic epistasis we observed could be caused by either one or both of these isoforms. Our results also show that the alternative splice variant, icIL-1ra1, does not play an essential role in the developmental processes affected by the deletion of IL-1ra.

Further, our results suggest that sIL-1ra mRNA, not icIL-1ra1 mRNA, is the predominant source of icIL-1ra3 in vivo. Intracellular IL-1ra3 is coexpressed with sIL-1ra in the liver following LPS administration, but icIL-1ra3 is not coexpressed with icIL-1ra1 in the skin of unstimulated mice (Fig. 1⇑). After LPS administration, wt mice express all isoforms of IL-1ra in the skin, while rako, ratg mice express only sIL-1ra and icIL-1ra3 (data not shown). Therefore, either icIL-1ra3 is expressed exclusively from sIL-1ra mRNA or the alternative translation of all IL-1ra mRNAs is induced by systemic LPS administration, which is required to induce the expression of sIL-1ra. Tissue and stimulus-specific factors or the different upstream mRNA sequences might regulate this differential expression. Similar conclusions have recently been drawn regarding the expression of icIL-1ra3 from studies of transgenic mice constitutively overproducing human IL-1ra isoforms (16).

Responses to infectious and inflammatory challenges

As described in the introduction, the differences reported for the phenotype of ratg vs that of RIko created an apparent paradox that questioned whether IL-1ra functions exclusively at the IL-1RI and whether IL-1 is essential for the host response to infection and inflammation. To investigate this, we backcrossed both ratg and RIko mutants to the C57BL/6J strain for at least 10 generations and examined their responses to models of infection and inflammation side-by-side, along with wt littermate controls. In our hands, infection with the intracellular pathogen, L. monocytogenes (10⁶ i.v., strain EGD) shows that both ratg and RIko differ from wt in survival and titers. Similarly, in a model of septic shock (i.p. administration of 10 μg/g LPS from S. typhimurium) both mutants were nearly identical in affording protection from lethality compared with wt mice (Table II⇓). Thus, on a C57BL/6J genetic background, overexpressing IL-1ra has the same effect as removing IL-1RI, supporting the view that IL-1 signaling is essential in promoting clearance of bacterial infections but is detrimental to survival after systemic inflammatory challenge.

We have also determined the survival of double-mutant mice after i.p. administration of 10 μg/g LPS and after infection with 10⁶ CFU L. monocytogenes (Table II⇑). The results show that RIko is epistatic to rako; the double mutant has the same survival rate as the RIko mutant alone. The same is true for the rako, ratg double mutant; the double mutant has the phenotype of the ratg parent. Overall, these data indicate that the sIL-1ra splice variant mRNA is sufficient to regulate inflammatory and infectious responses, and that its function is dependent upon the presence of the type I receptor. Finally, we constructed a ratg, RIko double-mutant strain. It has survival rates nearly identical with each of the parental strains that are themselves identical. Additive effects are not seen in this double mutant, re-enforcing the conclusion that the gene products encoded by the sIL-1ra transgene function only through IL-1RI and that there is no other target for IL-1ra.

Cytokine profile during septic shock

We previously reported parallel IL-1 and IL-1ra protein levels in serum following LPS administration, i.e., higher in the IL-1ra overexpressors and lower in the knockouts compared with wt mice (11). These in vivo results were counterintuitive, because earlier experiments had implicated IL-1 in an autoinduction, self-amplifying cascade (17, 18, 19, 20). We analyzed other cytokine profiles during sepsis in all single- and double-mutant strains to establish whether the parallel serum accumulation of IL-1 and IL-1ra is confined to the IL-1 system.

Sera were collected at 0, 1, 3, 8, and 16 h after administration of 10 μg/g LPS and were analyzed by ELISA to determine the levels of IL-1β, IL-1ra, TNF-α, IFN-γ, and IL-10. As presented in Table II⇑, the parallel regulation of IL-1 and IL-1ra was observed in all IL-1 system mutants, with peak levels of both cytokines elevated in the RIko and ratg mutants, and both diminished in the rako mutants, compared with wt controls. The levels of TNF-α, IFN-γ, and IL-10 were not different from those in wt throughout the inflammatory response in any mutant strain (data not shown). These results indicate that the parallel regulation of serum cytokine levels is confined to the IL-1 family of ligands. The analysis of the double mutants again shows that RIko and ratg are epistatic to rako, indicating that the observed phenotypes reflect the in vivo activity of the proteins made by the sIL-1ra splice variant acting through the type I receptor.